- Email: [email protected]
Measurement of corneal sublayer thickness and transparency in transgenic mice with altered corneal clarity using in vivo confocal microscopy
Vision Research 41 (2001) 1283– 1290 www.elsevier.com/locate/visres
Measurement of corneal sublayer thickness and transparency in transgenic mice with altered corneal clarity using in vivo confocal microscopy James V. Jester a,*, Young Ghee Lee a, Jie Li a, Shukti Chakravarti b, Jennifer Paul b, W. Matthew Petroll a, H. Dwight Cavanagh a a
Department of Ophthalmology, Uni6ersity of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Bl6d., Dallas, TX 75390 -9057, USA b Department of Medicine and Genetics, Case Western Reser6e Uni6ersity, Cle6eland, OH, USA Received 5 May 2000; received in revised form 3 August 2000
Abstract Measurement of sublayer thickness and transparency at cellular level in the living animal are critical to understanding the role of specific transgenes and transgene products in controlling corneal development and maintenance of transparency. Using two different transgenic mouse strains having altered corneal clarity, we have evaluated the ability of in vivo confocal microscopy to measure corneal haze and localize light scattering structures. Projection of 2-D and 3-D image information identified the nature and location of light scattering within the cornea and allowed correlation of unique structural differences to transgene expression. Our findings suggest that in vivo confocal microscopy can be used to identify the effects of transgene expression on mouse corneal transparency. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Cornea; Confocal microscopy; Digital imaging; Transgenic; Transparency
1. Introduction An important function of the cornea is the maintenance of an optically transparent media for the transmission and refraction of light to the neurosensory retina. Interestingly however, the molecular and supramolecular mechanisms controlling this transparent state have yet to be established clearly. Based on work by Maurice (1957, 1962), it is generally agreed that the size and uniformity of corneal collagen fibril diameter and spacing controls the light scattering properties of the normal cornea; predicting that factors controlling collagen fibril assembly play a critical role in regulating corneal transparency. More recently, a cellular basis of corneal transparency has also been identified that involves the abundant expression of
* Corresponding author. Tel.: + 1-214-6487215; fax: +1-2146484508. E-mail address: [email protected] (J.V. Jester).
unique water-soluble proteins referred to as keratocyte crystallins (Jester et al., 1999a). How these two mechanisms interplay to control the optical properties of the cornea is not yet understood. Recent developments in transgenic technology enable investigators to evaluate the role of specific genes and gene products on development, differentiation and function of the cornea. Through targeted gene disruption (knockout) and/or insertion (over-expression) transgenic animals offer the promise of greatly expanding our understanding of the molecular mechanisms controlling corneal transparency. However, to evaluate transgene effects adequately in the living animal, a quantitative approach to assessing corneal transparency is needed. While various methods such as light biomicroscopy and ultrasound pachymetry have been used to assess corneal transparency, none have yet been shown to have adequate spatial resolution to quantitatively and objectively detect and localize light scattering from the cornea, corneal cells and other intra-corneal structures.
0042-6989/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S0042-6989(00)00222-4
1284
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
In this regard, the development of confocal microscopy has lead to a dramatic improvement in both the lateral and axial resolution of optical microscopy and, has the capability to optically sectioning thick tissue specimens at different depths. Reflected-light in vivo confocal microscopy has the additional ability to image light scattering structures within the living cornea, including cells, vessels, nerves and other structures (Jester, Andrews, Petroll, Lemp, & Cavanagh, 1991; Cavanagh, Petroll, & Jester, 1993; Petroll, Jester, & Cavanagh, 1996a; Petroll, Jester, & Cavanagh, 1996b). Improvements in confocal microscope configuration, digital image acquisition, and objective lens design have enabled 3-D reconstruction of living cornea in situ, and the quantitative measurement of tissue thickness and cell density in the same cornea, non-invasively over time (Petroll, Cavanagh, & Jester, 1993; Jester et al., 1999c). These measurements provide a 4-D view of the cornea that have led to important insights into cellular responses and interactions involved in ocular irritation (Jester et al., 1998a, b), contact lens wear (Imayasu et al., 1994), wound healing (MollerPedersen, Li, Petroll, Cavanagh, & Jester, 1998a; Moller-Pedersen, Petroll, Cavanagh, & Jester, 1998b) and kidney function (Andrews, Petroll, Cavanagh, & Jester, 1991). More importantly, in vivo confocal microscopy has been used successfully to objectively and quantitatively measure the scattering of light from the cornea and to identify specific structures within the cornea that are associated with increased light scattering causing corneal haze and opacification (MollerPedersen, Cavanagh, Petroll, & Jester, 1998c; Moller-Pedersen, Cavanagh, Petroll, & Jester, 2000; Jester et al., 1999a). In this study we have used in vivo confocal microscopy to assess corneal transparency in two different transgenic mouse strains having altered corneal clarity, the PEPCK-TGFb1 over-expressing and the lumican knockout (Lum − / − ) mouse. We show that in vivo confocal microscopy can detect unique differences in the light scattering properties associated with different molecular mechanisms leading to corneal haze, opacification, and loss of transparency.
2. Materials and method
2.1. Description of transgenic mice Two strains of mice were used in this study; (1) the PEPCK-TGFb1 over-expressing mouse (kindly provided by Robert Hammer at the University of Texas Southwestern Medical Center) (Clouthier, Comerford, & Hammer, 1997) and (2) the Lumican knockout mouse (Chakravarti et al., 1998). In the first strain, constitutively active TGFb1 over-expression was
targeted to tissue specific sites using the regulatory sequences for phosphoenolpyruvate carboxykinase gene (PEPCK). The activity of PEPCK is thought to be the rate-limiting step in gluconeogenesis (Hanson & Garber, 1972) and shows a tissue-specific pattern of expression predominantly in liver, kidney, and adipose tissue in the mouse (Zimmer & Magnuson, 1990). Interestingly, over-expression of TGFb1 in these target organs results in marked fibrosis producing fibrotic liver disease, glomerulosclerosis and a lipodystrophy-like syndrome (Clouthier et al., 1997). While the ocular effects of TGFb1 over-expression in these transgenic mice have not been reported, previous immunohistochemical studies indicate that ocular tissues show expression of PEPCK, particularly in the optic nerve and lacrimal gland (Zimmer & Magnuson, 1990). TGFb1 has been shown previously to induce myofibroblast transformation of corneal keratocytes (Jester, Barry, Cavanagh, & Petroll, 1996; Jester, Barry-Lane, Petroll, Olsen, & Cavanagh, 1997; Jester et al., 1999b; Moller-Pedersen et al., 1998b). Myofibroblasts and activated keratocytes have been shown to express reduced amounts of keratocyte crystallin protein in the rabbit cornea, which is associated with the development of corneal haze after injury (Jester et al., 1999a). Furthermore, topical application of neutralizing antibodies to TGFb block keratocyte activation after injury and significantly reduce the development of corneal haze. Taken together these findings raise the possibility that over-expression of TGFb1 in the PEPCK-TGFb1 transgenic mouse may lead to changes in keratocyte differentiation, crystallin protein expression and corneal transparency. In the second strain, mice that were homozygous for a null mutation in the lumican gene were evaluated (Lum − / − ). Lumican is a major keratan sulfate containing proteoglycan of the corneal stroma and is a member of the leucine-rich proteoglycan family (Blochberger, Vergnes, Hempel, & Hassell, 1992). Lum − / − mice appear to develop bilateral corneal opacification between the ages of 5 and 34 weeks (Chakravarti et al., 1998). Based on the clinical assessment of corneal transparency, there appears to be increasing corneal clouding with age in Lum − / − mice. Ultrastructurally, the stromal collagen appears abnormal with thicker collagen fibril diameters compared to the normal cornea. These findings suggest that lumican may play an important role in regulating collagen fibril diameter and spacing within the cornea and that lumican is of critical importance to the development of corneal transparency. However, not all Lum − / − null mutants develop corneal clouding and the severest clouding is not restricted to the oldest animals based on the clinical biomicroscopic evaluation of transparency (Chakravarti et al., 1998).
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
2.2. In 6i6o confocal microscopy of transgenic mice All mice were treated according to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research (ARVO, 1994). Backscattering of light and corneal stromal thickness in the mouse eye was measured using a modification of previously described in vivo confocal microscopic techniques (Jester et al., 1999c). Briefly, in vivo examinations were performed using a Tandem Scanning Confocal Microscope (Tandem Scanning Corporation, Reston, VA) with a 24× surface-contact objective (numerical aperture, 0.6 mm; working distance, 1.5 mm). The system design and use has been described previously in detail (Petroll et al., 1996a; Petroll et al., 1996b). Prior to confocal examination mice were anesthetized with 100 mg/kg body weight ketamine HCl, 10 mg/kg body weight xylazine, and 2 mg/kg body weight acepromazine by intraperitoneal injection. During scanning and observation, both eyes were kept moist using topically applied tear solution (Celluvisc, Allergan Inc., Irvine, CA) to avoid corneal desiccation. A drop of 2.5% hydroxypropyl methylcellulose (Goniosol, IOLAB Pharmaceuticals, Claremont, CA) was then placed on the tip of objective to serve as an immersion fluid. With video camera gain, kilovolts, and black level kept in automatic mode, high quality images were obtained from all corneal sublayers. Then the video camera gain, kilovolts, and black level were switched to manual settings and kept constant throughout the remainder of the evaluation on each eye to allow direct comparison of scans between mouse eyes. A confocal microscopy through focusing (CMTF) data set was obtained using techniques previously described (Jester et al., 1999c). CMTF analysis was performed as a continuous, z-axis scan through the entire cornea starting in front of the epithelium and ending below the endothelium. The focal plane position was controlled using an Oriel 18011 Encoder Mike Controller (Oriel Corp., Stratford, CT) interfaced to a personal computer. The illumination source was a 100 W mercury arc lamp, and realtime image detection was performed using Dage MTI VE-1000 camera (Dage MTI, Michigan City, IN). The continuous z-scan image was saved to the hard disk of a personal computer and analyzed on-line using a recently developed WINDOWS-based application program (Lie, Jester, Cavanagh, & Petroll, 2000). With the current objective and camera, the system has an approximate field-of-view of 475×350 mm and an optical slice thickness (z-axis resolution) of 9 mm. Four consecutive forward CMTF z-scans were performed in the corneal center at each examination. During CMTF scanning, the focal plane was moved at a speed of approximately 64 mm/s: thus, individual images that were digitized at 30 frame/s were separated in
1285
the z-axis by approximately 2.13 mm. CMTF images obtained from the camera were digitized on-line and analyzed automatically. To generate depth intensity profiles, the average pixel intensity in the central 285× 285-mm region (180×180 pixels) of each consecutive image in the CMTF sequence was determined and plotted as a function of z-depth. Using an interactive cursor, points along the profile were selected and the corresponding image simultaneously displayed to identify structures of interest including surface epithelium, epithelial basal lamina, endothelium, etc. Points of interest were marked with the interactive cursor and distances between structures calculated to determine corneal, epithelial and stromal thickness. The CMTF haze estimate was determined by calculating the area under the CMTF pixel intensity curve using techniques described below. This integrated pixel intensity area was used as the objective estimate of corneal light scattering or corneal haze. Before each examination, all lens surfaces were cleaned. Both the light source and microscope alignment were checked to ensure no major changes in alignment occurred during the study to influence scattering measurements. The video camera sensitivity was kept constant by using the same manual settings for gain, kilovolts, and black level in all scans. We monitor the day-to-day variation by scanning through a drop of 2.5% hydroxypropyl methylcellulose placed on the objective tip. Past studies have shown that the coefficient of variation (S.D./mean) for these test measurements are less than 6% with no significant increase or decrease over time. These procedures are identical to those already established for the rabbit and human studies (Moller-Pedersen et al., 1998c; Moller-Pedersen et al., 2000).
2.3. Measurement of tissue thickness and light scattering Using the 2-D depth-intensity profile, tissue thickness was objectively determined by measuring the distance between the marked regions within the cornea using the interactive cursor or by measuring the distance between intensity peaks, without operator bias. Corneal thickness (CT), epithelial thickness (ET) and stromal thickness (ST) were calculated using the following equations: CT = ZEndo − ZEpi
(1)
ET = ZBL − ZEpi
(2)
ST= ZEndo − ZBL
(3)
where ZEpi, ZBL, ZEndo are the axial depths for the superficial epithelial, basal lamina and endothelial intensity peak (Jester et al., 1999c). The amount of backscattered light from the mouse cornea was also quantified. Backscattering was esti-
1286
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
mated by integrating the area under the pixel intensity curve, using the following equation: Area = %{(Ii +Ii + 1)/2 − IAC}% zi ;
%zi =zi −zi − 1 (4)
where Ii is the intensity value of image i; IAC is the background intensity of the anterior chamber; zi and zi − 1 are the focal plane positions of image i and i− 1 (Jester, Petroll, & Cavanagh, 1999c). The unit of measurement (U) is defined as mm ×pixel intensity. A close correlation between the corneal transparency, as measured by the clinical examination of the cornea, and the CMTF estimate of light scattering has been previously shown (Moller-Pedersen et al., 1997; Moller-Pedersen et al., 2000).
Fig. 1. In vivo confocal microscopy of control (left) and TGFb over-expressing (right) mouse corneas. 3-D images with superficial epithelial cells on top and endothelium at the bottom were reconstructed in control (A) and TGFb over-expressing mice (B). Note the increased stromal reflectivity appearing to originate from discrete structures within the corneal stroma in TGFb over-expressing mice (arrow). Superficial epithelial cells in the PEPCK-TGFb over-expressing mouse cornea (D) appeared smaller and dystrophic compared to control (C). Keratocytes were barely detectable throughout the stroma in control mice (E). However, keratocytes with marked increased cellular reflectivity were noted in TGFb over-expressing mice (arrows) (F).
3. In vivo confocal microscopy of transgenic mice Thus far we have evaluated a small series of PEPCKTGFb1 over-expressing mice and their normal littermates. As shown in Fig. 1, the normal mouse cornea (A) appears similar to that of rabbit corneas as viewed by 3-D reconstruction of confocal microscopic images in that there are three major regions that significantly backscatter light, i.e. the surface epithelium, the basal lamina and corneal endothelium. The amount of light scattering from the normal mouse cornea is better illustrated by the pixel intensity profile shown in Fig. 2A%, where specific peaks, a, b, and d, correspond to light scattering from the surface epithelium, basal lamina and endothelium respectively. Importantly, corneal keratocytes in the normal stroma show very little backscattering of light (Fig. 1E and Fig. 2A). On the other hand, TGFb1 over-expressing mice (Fig. 1B), show several notable changes. First, the surface epithelium appears dystrophic with smaller and more irregular cells (D) compared to normal surface corneal epithelial cells (C). This was associated with a slight thickening of the corneal epithelium and a more marked scattering of light from the suprabasal epithelial cells (Fig. 2A% and B%, peak a). Secondly, within the corneal stroma, there was considerable scattering of light from what appeared to be corneal keratocytes (Fig. 1F, arrows). This increased backscattering resulted in the elevation of the CMTF curve over the stromal region (Fig. 2B%, region c) and a slight but not significant increase in the stromal thickness. Overall, eight eyes from eight transgenic mice showed a significant increase in the entire corneal thickness compared to eight littermate controls (126.29 11.0 vs. 112.997.0 mm, PB 0.01, respectively). Quantitative measurement of the light scattering in the eight littermates and eight TGFb1 over-expressing mice also showed a significantly elevated stromal haze associated with the changes in the stromal keratocytes (18039 237 vs. 13549 206 U, PB 0.005). Clearly, more detailed studies are needed to determine whether the increased backscattering of light in the PEPCK-TGFb1 over-expressing mouse is due to the differentiation of corneal keratocytes to a myofibroblast phenotype or the deposition of abnormal extracellular matrix surrounding these cells. Nevertheless, these findings are consistent with the earlier observation that corneal keratocytes may contribute to the loss of transparency in response to corneal injury or the release of cytokines associated with the wound healing response. Light scattering from the Lum − / − knockout mouse cornea appeared distinctly different from that of the PEPCK-TGFb1 over-expressing mouse cornea. 3-D reconstructions of the cornea from Lum − / − knockout mice showed diffuse light scattering from predominantly the posterior region of the corneal stroma (Fig.
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
1287
Fig. 2. In vivo CMTF stromal images (A and B) and CMTF profiles (A% and B%) of control mice (A and A%) and TGFb over-expressing mice (B and B%). Two peaks corresponding to superficial epithelium (a) and endothelium (d) were noted in both groups (A% and B%). The epithelial –stromal interface (arrow, b) was defined by the first in-focus image of stromal tissue. It should be noted that entire stromal reflectivity was increased in TGFb over-expressing mice cornea (B%, region c), which correlated with the prominent appearance of stromal keratocytes (B, arrow). The pixel intensity curves have been offset in the y-axis and aligned to the corneal epithelial peak for demonstration purposes.
3A, arrow) rather than evenly throughout the cornea as detected in the PEPCK-TGFb1 over-expressing mouse. This localization was further confirmed by CMTF analysis (Fig. 3C, c-arrow), which showed a marked increase in the intensity of scattered light originating from the posterior stromal region closer to the corneal endothelium. 2-D images of light scattering also showed a more diffuse, uniform and indistinct pattern throughout the posterior stroma (Fig. 3B) rather than the discrete pattern suggestive of stromal keratocytes detected in the PEPCK-TGFb1 over-expressing mouse cornea. The diffuse nature of the light scattering was consistent with random light scattering that would be associated with the disorganized collagen fibril organization noted to occur in the Lum − / − knockout mouse cornea (Chakravarti et al., 1998). The finding that light scattering appears predominantly in the posterior cornea suggests that the posterior cornea may be more affected in these mice than the anterior corneal stroma. Alternatively, lumican may play a more critical role in the organization of collagen fibrils in the posterior cornea than the anterior cornea.
4. Discussion Measurement of light scattering by in vivo confocal microscopy has been previously used to evaluate the
development of corneal haze and clouding following photorefractive keratectomy (PRK) in patients (MollerPedersen et al., 1997; Moller-Pedersen et al., 2000) and rabbits (Moller-Pedersen et al., 1998a; Moller-Pedersen et al., 1998c), assess the effects of neutralizing antibodies to TGFb on development of haze following PRK in rabbits (Moller-Pedersen et al., 1998b), assess corneal haze after LASIK eye surgery in patients (Vesaluoma et al., 2000), and to monitor corneal responses following ocular irritation in the rabbit eye (Jester et al., 1998b; Maurer et al., 1999). These studies have not only provided important insights into the wound healing and irritancy response of the cornea, but have also lead to the discovery of a new and important class of keratocyte crystallin proteins that may play an important role in regulating corneal transparency at the cellular level (Jester et al., 1999a). The present study is the first to use in vivo confocal microscopy specifically to evaluate light scattering in the mouse cornea and to determine the tissue localization and pattern of light scattering from mice genetically altered to specifically explore the basic molecular and cellular mechanisms of corneal transparency. As expected in the Lum − / − knockout mouse, a diffuse pattern of light scattering was detected consistent with the disorganization of collagen fibril diameter and spacing previously shown to occur in these eyes (Chakravarti et al., 1998). Based on our current under-
1288
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
standing of the structural basis of corneal transparency established by Maurice (1957), light transmission through the cornea is dependent in part on the latticelike organization of small diameter collagen fibrils such that scattered light between collagen fibrils is destructively interfered. When spacing between collagen fibers is increased as occurs in corneal edema, there is a loss of this destructive interference and a random scattering of light throughout the swollen cornea. Likewise, increased scattering occurs when collagen fibril size increases and inter-fibrillar spacing becomes more irregular. Since collagen fibers (32 nm in diameter) are not resolved by confocal microscopy, which has a lateral resolution of approximately 0.2– 1.2 mm, random scattering of light would appear as a diffuse increase in the amount of light detected in the backward direction as observed in the Lum − / − knockout mouse. A similar diffuse increase in light intensity from the cornea is detected by in vivo confocal microscopy in irritated eyes that are markedly swollen (Jester et al., 1998b). Interestingly, however, light scattering in the Lum − / − knockout was preferentially localized to the posterior region of the cornea. This is consistent with recent observations that the posterior stroma is more affected in the Lum − / − knockout mouse, with both collagen fibril size and inter-fibril spacing markedly altered
(Chakravarti et al., 2000). The greater susceptibility of the posterior stroma is also consistent with the higher concentration of keratan sulfated proteoglycans in this region (Castoro, Bettelheim, & Bettelheim, 1988). While more detailed study is necessary, this finding suggests that lumican may play a more important role in controlling fibril diameter and spacing in the posterior cornea. Unlike the Lum − / − knockout mouse, the PEPCKTGFb1 over-expressing mouse showed a pattern of light scattering that appeared to be localized to corneal keratocytes. TGFb is known to increase extracellular matrix synthesis and is thought to control the development of fibrosis following corneal injury (Ohji, SundarRaj, & Thoft, 1993; Jester et al., 1996). In the PEPCK-TGFb1 over-expressing mouse, there is notable increase in the fibrotic response of major organs including the liver, kidney, fat and skin. It is possible that the increased light scattering seen in these corneas may be due to the deposition of irregularly arranged collagen fibrils immediately adjacent to the mouse keratocytes. Alternatively, TGFb1 is known to control myofibroblast differentiation in rabbit corneal keratocytes. Myofibroblasts show a unique pattern of light scattering localized to the cell body and extended cell processes, unlike the totally transparent nature of the
Fig. 3. In vivo CMTF images from the Lum − / − knockout mouse. 3-D reconstruction of a 5-month-old Lum − / − knockout mouse cornea (A) showed a marked increase in light scattering from the posterior region of the corneal stroma (arrow). 2-D images from the posterior region (B) showed a diffuse pattern of light scattering suggestive of random scattered light due to changes in the collagen fibril organization. CMTF analysis of the cornea (C) further supported the localization of the light scattering to the posterior stroma as indicated by the rise in light intensity (c, arrow) extending back toward the endothelium (peak d). a, epithelial peak; b, basal lamina peak; c, stromal region; and d, endothelial peak.
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
normal keratocyte. Myofibroblasts and activated keratocytes from the rabbit also show a greatly reduced expression of two water-soluble, crystallin proteins, transketolase and aldehyde dehydrogenase class 1. Although the pattern of light scattering in the PEPCKTGFb1 over-expressing mouse is not exactly similar to that observed for the rabbit corneal myofibroblasts, the cellular localization of light scattering is consistent with a cellular mechanism based on crystallin protein expression. Clearly, further work is required to determine more precisely the cause of light scattering in this model. Overall, in vivo confocal microscopy appears to be a sensitive and quantitative method for evaluating light scattering from the cornea in transgenic mice. Not only can it detect corneal clouding that is difficult to appreciate by macroscopic examination, in vivo confocal microscopy is also capable of localizing scattered light to specific structures within the cornea. In the current study, two different patterns of light scattering were detected, a diffuse scattering localized to the posterior cornea and a discrete, cellular based light scattering detected throughout the cornea. It is tempting to postulate that these differences are related to the distinct differences in the mechanisms by which the corneal transparency was altered, the first being a structural alteration of the collagen organization and the second being a cellular alteration in the cellular protein expression. While further work is needed to more precisely define these transgenic models, the application of in vivo confocal microscopy to studying transgenically altered mice will clearly provide new and important insights into the molecular and cellular basis of corneal transparency.
Acknowledgements This work was supported, in part, by grants EY07348 (JVJ), EY11654 (SC) and EY13215 (JVJ) from the National Eye Institute, Bethesda, MD, and Research to Prevent Blindness, Inc., New York, NY.
References Andrews, P. M., Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1991). Tandem scanning confocal microscopy (TSCM) of normal and ischemic living kidneys. American Journal of Anatomy, 191, 95 – 102. ARVO (1994). Statement for the use of animals in ophthalmic and vision research. In6estigati6e Ophthalmology and Visual Science, 35, v– vi. Blochberger, T. C., Vergnes, J.-P., Hempel, J., & Hassell, J. R. (1992). cDNA to chick lumican (corneal keratan sulfate proteoglycan) reveals homology to the small interstitial groteoglycan gene family and expression in muscle and intestine. Journal of Biological Chemistry, 267, 347–352.
1289
Castoro, J. A., Bettelheim, A. A., & Bettelheim, F. A. (1988). Water gradients across bovine cornea. In6estigati6e Ophthalmology and Visual Science, 29, 963 – 968. Cavanagh, H. D., Petroll, W. M., & Jester, J. V. (1993). The application of confocal microscopy to the study of living systems. Neuroscience and Biobeha6ioral Re6iews, 17, 483 – 498. Chakravarti, S., Magnuson, T., Lass, J. H., Jepsen, K. J., La Mantia, C., & Carroll, H. (1998). Lumican regulates collagen fibril assembly: skin fragility and corneal opacity in the absence of lumican. Journal Cell Biology, 141, 1277 – 1286. Chakravarti, S., Petroll, W. M., Hassell, J., Jester, J., Lass, J., Paul, J., & Birk, D. E. (2000). Corneal opacity in lumican-deficient mice due to defects in collagen fibril structure and packing in the posterior stroma. In6estigati6e Ophthalmology and Visual Science, in press Clouthier, D. E., Comerford, S. A., & Hammer, R. E. (1997). Hepatic fibrosis, glomerulosclerosis, and a lipohystrophy-like syndrome in PEPCK-TGF-b1 transgenic mice. Journal of Clinical In6estigation, 100, 2697 – 2713. Hanson, R. W., & Garber, A. J. (1972). Phosphoenolpyruvate carboxykinase, 1. Its role in gluconeogenesis. American Journal of Clinical Nutrition, 25, 1010. Imayasu, M., Petroll, W. M., Jester, J. V., Patel, S. K., Ohashi, J., & Cavanagh, H. D. (1994). The relation between contact lens oxygen transmissibility and binding of Pseudomonas aeruginosa to the cornea after overnight wear. Ophthalmology, 101, 371 – 388. Jester, J. V., Andrews, P. M., Petroll, W. M., Lemp, M. A., & Cavanagh, H. D. (1991). In vivo, real-time confocal imaging. J. Elect. Microsc. Tech., 18, 50 – 60. Jester, J. V., Barry, P. A., Cavanagh, H. D., & Petroll, W. M. (1996). Induction of a-smooth muscle actin (a-SM) expression and myofibroblast transformation in cultured keratocytes. Cornea, 15, 505 – 516. Jester, J. V., Barry-Lane, P. A., Petroll, W. M., Olsen, D. R., & Cavanagh, H. D. (1997). Inhibition of corneal fibrosis by topical application of blocking antibodies to TGF beta in the rabbit. Cornea, 16, 177 – 187. Jester, J. V., Huang, J., Barry-Lane, P. A., Kao, W. W., Petroll, W. M., & Cavanagh, H. D. (1999b). TGFb-mediated myofibroblast differentiation and a smooth-muscle actin expression in corneal fibroblasts requires actin re-organization and focal adhesion assembly. In6estigati6e Ophthalmology and Visual Science, 40, 1959– 1967. Jester, J. V., Li, H.-F., Petroll, W. M., Parker, R. D., Cavanagh, H. D., Carr, G. J., Smith, B., & Maurer, J. K. (1998a). Area and depth of surfactant-induced corneal irritancy correlates with cell death. In6estigati6e Ophthalmology and Visual Science, 39, 922– 936. Jester, J. V., Moller-Pedersen, T., Huang, J., Sax, C. M., Petroll, W. M., Cavanagh, H. D., & Piatigorsky, J. (1999a). The cellular basis of corneal transparency: evidence for ‘corneal crystallins’. Journal of Cell Science, 112, 613 – 622. Jester, J. V., Petroll, W. M., & Cavanagh, H. D. (1999c). Measurement of tissue thickness using confocal microscopy. In P. M. Conn, Confocal microscopy, vol. 307 (pp. 230 – 245). San Diego, CA: Academic Press. Jester, J. V., Petroll, W. M., Bean, J., Parker, R. D., Carr, G. J., Cavanagh, H. D., & Maurer, J. K. (1998b). Area and depth of surfactant-induced corneal injury predicts extent of subsequent ocular responses. In6estigati6e Ophthalmology and Visual Science, 39, 2610 – 2625. Lie, J., Jester, J. V., Cavanagh, H. D., & Petroll, W. M. (2000). On-line 3-dimensional confocal imaging in vivo. In6estigati6e Ophthalmology and Visual Science, 41, 2945 – 2953. Maurer, J. K., Parker, R. D., Petroll, W. M., Carr, G. J., Cavanagh, H. D., & Jester, J. V. (1999). Quantitative measurement of acute corneal injury in rabbits with surfactants of different type and irritancy. Toxicology and Applied Pharmacology, 158, 61 –70.
1290
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
Maurice, D. M. (1957). The structure and transparency of the cornea. Journal of Physiology, 136, 263–286. Maurice, D. M. (1962). The cornea and sclera. In H. Davson, The eye, vol. 1 (pp. 289 – 361). New York: Academic Press. Moller-Pedersen, T., Cavanagh, H. D., Petroll, W. M., & Jester, J. V. (1998c). Corneal haze development after PRK is regulated by volume of stromal tissue removal. Cornea, 17, 627–639. Moller-Pedersen, T., Cavanagh, H. D., Petroll, W. M., & Jester, J. V. (2000). Stromal wound healing explains refractive instability and haze development after photorefractive keratectomy. A one-year confocal microscopic study. Ophthalmology, 107, 1235 – 1245. Moller-Pedersen, T., Li, H.-F., Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1998a). Confocal microscopic characterization of wound repair following photorefractive keratectomy. In6estigati6e Ophthalmology and Visual Science, 39, 487–501. Moller-Pedersen, T., Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1998b). Neutralizing antibody to TGFb modulates stromal fibrosis but not regression of photoablative effect following PRK. Current Eye Research, 17, 736–747. Moller-Pedersen, T., Vogel, M., Li, H.-F., Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1997). Quantification of stromal thinning, epithelial thickness, and corneal haze after photorefractive kerate-
.
ctomy using in vivo confocal microscopy. Ophthalmology, 104, 360 – 368. Ohji, M., SundarRaj, N., & Thoft, R. A. (1993). Transforming growth factor-beta stimulates collagen and fibronectin synthesis by human corneal stromal fibroblasts in vitro. Current Eye Research, 12, 703 – 709. Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1993). Three-dimensional imaging of corneal cells using in vivo confocal microscopy. Journal of Microscopy, 170, 213 – 219. Petroll, W. M., Jester, J. V., & Cavanagh, H. D. (1996a). Quantitative three-dimensional confocal imaging of the corneal in situ and in vivo: system design and calibration. Scanning, 18, 45 –49. Petroll, W. M., Jester, J. V., & Cavanagh, H. D. (1996b). In vivo confocal imaging. International Re6iew of Experimental Pathology, 36, 93 – 129. Vesaluoma, M., Perez-Santonja, J., Petroll, W. M., Linna, T., Alio, J., & Tervo, T. (2000). Corneal stromal changes induced by myopic LASIK. In6estigati6e Ophthalmology and Visual Science, 41, 369 – 376. Zimmer, D. B., & Magnuson, M. A. (1990). Immunohistochemical localization of phosphoenolpyruvate carboxykinase in adult and developing mouse tissues. Journal of Histochemistry Cytochemistry, 38, 171 – 178.
Measurement of corneal sublayer thickness and transparency in transgenic mice with altered corneal clarity using in vivo confocal microscopy James V. Jester a,*, Young Ghee Lee a, Jie Li a, Shukti Chakravarti b, Jennifer Paul b, W. Matthew Petroll a, H. Dwight Cavanagh a a
Department of Ophthalmology, Uni6ersity of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Bl6d., Dallas, TX 75390 -9057, USA b Department of Medicine and Genetics, Case Western Reser6e Uni6ersity, Cle6eland, OH, USA Received 5 May 2000; received in revised form 3 August 2000
Abstract Measurement of sublayer thickness and transparency at cellular level in the living animal are critical to understanding the role of specific transgenes and transgene products in controlling corneal development and maintenance of transparency. Using two different transgenic mouse strains having altered corneal clarity, we have evaluated the ability of in vivo confocal microscopy to measure corneal haze and localize light scattering structures. Projection of 2-D and 3-D image information identified the nature and location of light scattering within the cornea and allowed correlation of unique structural differences to transgene expression. Our findings suggest that in vivo confocal microscopy can be used to identify the effects of transgene expression on mouse corneal transparency. © 2001 Elsevier Science Ltd. All rights reserved. Keywords: Cornea; Confocal microscopy; Digital imaging; Transgenic; Transparency
1. Introduction An important function of the cornea is the maintenance of an optically transparent media for the transmission and refraction of light to the neurosensory retina. Interestingly however, the molecular and supramolecular mechanisms controlling this transparent state have yet to be established clearly. Based on work by Maurice (1957, 1962), it is generally agreed that the size and uniformity of corneal collagen fibril diameter and spacing controls the light scattering properties of the normal cornea; predicting that factors controlling collagen fibril assembly play a critical role in regulating corneal transparency. More recently, a cellular basis of corneal transparency has also been identified that involves the abundant expression of
* Corresponding author. Tel.: + 1-214-6487215; fax: +1-2146484508. E-mail address: [email protected] (J.V. Jester).
unique water-soluble proteins referred to as keratocyte crystallins (Jester et al., 1999a). How these two mechanisms interplay to control the optical properties of the cornea is not yet understood. Recent developments in transgenic technology enable investigators to evaluate the role of specific genes and gene products on development, differentiation and function of the cornea. Through targeted gene disruption (knockout) and/or insertion (over-expression) transgenic animals offer the promise of greatly expanding our understanding of the molecular mechanisms controlling corneal transparency. However, to evaluate transgene effects adequately in the living animal, a quantitative approach to assessing corneal transparency is needed. While various methods such as light biomicroscopy and ultrasound pachymetry have been used to assess corneal transparency, none have yet been shown to have adequate spatial resolution to quantitatively and objectively detect and localize light scattering from the cornea, corneal cells and other intra-corneal structures.
0042-6989/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S0042-6989(00)00222-4
1284
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
In this regard, the development of confocal microscopy has lead to a dramatic improvement in both the lateral and axial resolution of optical microscopy and, has the capability to optically sectioning thick tissue specimens at different depths. Reflected-light in vivo confocal microscopy has the additional ability to image light scattering structures within the living cornea, including cells, vessels, nerves and other structures (Jester, Andrews, Petroll, Lemp, & Cavanagh, 1991; Cavanagh, Petroll, & Jester, 1993; Petroll, Jester, & Cavanagh, 1996a; Petroll, Jester, & Cavanagh, 1996b). Improvements in confocal microscope configuration, digital image acquisition, and objective lens design have enabled 3-D reconstruction of living cornea in situ, and the quantitative measurement of tissue thickness and cell density in the same cornea, non-invasively over time (Petroll, Cavanagh, & Jester, 1993; Jester et al., 1999c). These measurements provide a 4-D view of the cornea that have led to important insights into cellular responses and interactions involved in ocular irritation (Jester et al., 1998a, b), contact lens wear (Imayasu et al., 1994), wound healing (MollerPedersen, Li, Petroll, Cavanagh, & Jester, 1998a; Moller-Pedersen, Petroll, Cavanagh, & Jester, 1998b) and kidney function (Andrews, Petroll, Cavanagh, & Jester, 1991). More importantly, in vivo confocal microscopy has been used successfully to objectively and quantitatively measure the scattering of light from the cornea and to identify specific structures within the cornea that are associated with increased light scattering causing corneal haze and opacification (MollerPedersen, Cavanagh, Petroll, & Jester, 1998c; Moller-Pedersen, Cavanagh, Petroll, & Jester, 2000; Jester et al., 1999a). In this study we have used in vivo confocal microscopy to assess corneal transparency in two different transgenic mouse strains having altered corneal clarity, the PEPCK-TGFb1 over-expressing and the lumican knockout (Lum − / − ) mouse. We show that in vivo confocal microscopy can detect unique differences in the light scattering properties associated with different molecular mechanisms leading to corneal haze, opacification, and loss of transparency.
2. Materials and method
2.1. Description of transgenic mice Two strains of mice were used in this study; (1) the PEPCK-TGFb1 over-expressing mouse (kindly provided by Robert Hammer at the University of Texas Southwestern Medical Center) (Clouthier, Comerford, & Hammer, 1997) and (2) the Lumican knockout mouse (Chakravarti et al., 1998). In the first strain, constitutively active TGFb1 over-expression was
targeted to tissue specific sites using the regulatory sequences for phosphoenolpyruvate carboxykinase gene (PEPCK). The activity of PEPCK is thought to be the rate-limiting step in gluconeogenesis (Hanson & Garber, 1972) and shows a tissue-specific pattern of expression predominantly in liver, kidney, and adipose tissue in the mouse (Zimmer & Magnuson, 1990). Interestingly, over-expression of TGFb1 in these target organs results in marked fibrosis producing fibrotic liver disease, glomerulosclerosis and a lipodystrophy-like syndrome (Clouthier et al., 1997). While the ocular effects of TGFb1 over-expression in these transgenic mice have not been reported, previous immunohistochemical studies indicate that ocular tissues show expression of PEPCK, particularly in the optic nerve and lacrimal gland (Zimmer & Magnuson, 1990). TGFb1 has been shown previously to induce myofibroblast transformation of corneal keratocytes (Jester, Barry, Cavanagh, & Petroll, 1996; Jester, Barry-Lane, Petroll, Olsen, & Cavanagh, 1997; Jester et al., 1999b; Moller-Pedersen et al., 1998b). Myofibroblasts and activated keratocytes have been shown to express reduced amounts of keratocyte crystallin protein in the rabbit cornea, which is associated with the development of corneal haze after injury (Jester et al., 1999a). Furthermore, topical application of neutralizing antibodies to TGFb block keratocyte activation after injury and significantly reduce the development of corneal haze. Taken together these findings raise the possibility that over-expression of TGFb1 in the PEPCK-TGFb1 transgenic mouse may lead to changes in keratocyte differentiation, crystallin protein expression and corneal transparency. In the second strain, mice that were homozygous for a null mutation in the lumican gene were evaluated (Lum − / − ). Lumican is a major keratan sulfate containing proteoglycan of the corneal stroma and is a member of the leucine-rich proteoglycan family (Blochberger, Vergnes, Hempel, & Hassell, 1992). Lum − / − mice appear to develop bilateral corneal opacification between the ages of 5 and 34 weeks (Chakravarti et al., 1998). Based on the clinical assessment of corneal transparency, there appears to be increasing corneal clouding with age in Lum − / − mice. Ultrastructurally, the stromal collagen appears abnormal with thicker collagen fibril diameters compared to the normal cornea. These findings suggest that lumican may play an important role in regulating collagen fibril diameter and spacing within the cornea and that lumican is of critical importance to the development of corneal transparency. However, not all Lum − / − null mutants develop corneal clouding and the severest clouding is not restricted to the oldest animals based on the clinical biomicroscopic evaluation of transparency (Chakravarti et al., 1998).
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
2.2. In 6i6o confocal microscopy of transgenic mice All mice were treated according to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research (ARVO, 1994). Backscattering of light and corneal stromal thickness in the mouse eye was measured using a modification of previously described in vivo confocal microscopic techniques (Jester et al., 1999c). Briefly, in vivo examinations were performed using a Tandem Scanning Confocal Microscope (Tandem Scanning Corporation, Reston, VA) with a 24× surface-contact objective (numerical aperture, 0.6 mm; working distance, 1.5 mm). The system design and use has been described previously in detail (Petroll et al., 1996a; Petroll et al., 1996b). Prior to confocal examination mice were anesthetized with 100 mg/kg body weight ketamine HCl, 10 mg/kg body weight xylazine, and 2 mg/kg body weight acepromazine by intraperitoneal injection. During scanning and observation, both eyes were kept moist using topically applied tear solution (Celluvisc, Allergan Inc., Irvine, CA) to avoid corneal desiccation. A drop of 2.5% hydroxypropyl methylcellulose (Goniosol, IOLAB Pharmaceuticals, Claremont, CA) was then placed on the tip of objective to serve as an immersion fluid. With video camera gain, kilovolts, and black level kept in automatic mode, high quality images were obtained from all corneal sublayers. Then the video camera gain, kilovolts, and black level were switched to manual settings and kept constant throughout the remainder of the evaluation on each eye to allow direct comparison of scans between mouse eyes. A confocal microscopy through focusing (CMTF) data set was obtained using techniques previously described (Jester et al., 1999c). CMTF analysis was performed as a continuous, z-axis scan through the entire cornea starting in front of the epithelium and ending below the endothelium. The focal plane position was controlled using an Oriel 18011 Encoder Mike Controller (Oriel Corp., Stratford, CT) interfaced to a personal computer. The illumination source was a 100 W mercury arc lamp, and realtime image detection was performed using Dage MTI VE-1000 camera (Dage MTI, Michigan City, IN). The continuous z-scan image was saved to the hard disk of a personal computer and analyzed on-line using a recently developed WINDOWS-based application program (Lie, Jester, Cavanagh, & Petroll, 2000). With the current objective and camera, the system has an approximate field-of-view of 475×350 mm and an optical slice thickness (z-axis resolution) of 9 mm. Four consecutive forward CMTF z-scans were performed in the corneal center at each examination. During CMTF scanning, the focal plane was moved at a speed of approximately 64 mm/s: thus, individual images that were digitized at 30 frame/s were separated in
1285
the z-axis by approximately 2.13 mm. CMTF images obtained from the camera were digitized on-line and analyzed automatically. To generate depth intensity profiles, the average pixel intensity in the central 285× 285-mm region (180×180 pixels) of each consecutive image in the CMTF sequence was determined and plotted as a function of z-depth. Using an interactive cursor, points along the profile were selected and the corresponding image simultaneously displayed to identify structures of interest including surface epithelium, epithelial basal lamina, endothelium, etc. Points of interest were marked with the interactive cursor and distances between structures calculated to determine corneal, epithelial and stromal thickness. The CMTF haze estimate was determined by calculating the area under the CMTF pixel intensity curve using techniques described below. This integrated pixel intensity area was used as the objective estimate of corneal light scattering or corneal haze. Before each examination, all lens surfaces were cleaned. Both the light source and microscope alignment were checked to ensure no major changes in alignment occurred during the study to influence scattering measurements. The video camera sensitivity was kept constant by using the same manual settings for gain, kilovolts, and black level in all scans. We monitor the day-to-day variation by scanning through a drop of 2.5% hydroxypropyl methylcellulose placed on the objective tip. Past studies have shown that the coefficient of variation (S.D./mean) for these test measurements are less than 6% with no significant increase or decrease over time. These procedures are identical to those already established for the rabbit and human studies (Moller-Pedersen et al., 1998c; Moller-Pedersen et al., 2000).
2.3. Measurement of tissue thickness and light scattering Using the 2-D depth-intensity profile, tissue thickness was objectively determined by measuring the distance between the marked regions within the cornea using the interactive cursor or by measuring the distance between intensity peaks, without operator bias. Corneal thickness (CT), epithelial thickness (ET) and stromal thickness (ST) were calculated using the following equations: CT = ZEndo − ZEpi
(1)
ET = ZBL − ZEpi
(2)
ST= ZEndo − ZBL
(3)
where ZEpi, ZBL, ZEndo are the axial depths for the superficial epithelial, basal lamina and endothelial intensity peak (Jester et al., 1999c). The amount of backscattered light from the mouse cornea was also quantified. Backscattering was esti-
1286
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
mated by integrating the area under the pixel intensity curve, using the following equation: Area = %{(Ii +Ii + 1)/2 − IAC}% zi ;
%zi =zi −zi − 1 (4)
where Ii is the intensity value of image i; IAC is the background intensity of the anterior chamber; zi and zi − 1 are the focal plane positions of image i and i− 1 (Jester, Petroll, & Cavanagh, 1999c). The unit of measurement (U) is defined as mm ×pixel intensity. A close correlation between the corneal transparency, as measured by the clinical examination of the cornea, and the CMTF estimate of light scattering has been previously shown (Moller-Pedersen et al., 1997; Moller-Pedersen et al., 2000).
Fig. 1. In vivo confocal microscopy of control (left) and TGFb over-expressing (right) mouse corneas. 3-D images with superficial epithelial cells on top and endothelium at the bottom were reconstructed in control (A) and TGFb over-expressing mice (B). Note the increased stromal reflectivity appearing to originate from discrete structures within the corneal stroma in TGFb over-expressing mice (arrow). Superficial epithelial cells in the PEPCK-TGFb over-expressing mouse cornea (D) appeared smaller and dystrophic compared to control (C). Keratocytes were barely detectable throughout the stroma in control mice (E). However, keratocytes with marked increased cellular reflectivity were noted in TGFb over-expressing mice (arrows) (F).
3. In vivo confocal microscopy of transgenic mice Thus far we have evaluated a small series of PEPCKTGFb1 over-expressing mice and their normal littermates. As shown in Fig. 1, the normal mouse cornea (A) appears similar to that of rabbit corneas as viewed by 3-D reconstruction of confocal microscopic images in that there are three major regions that significantly backscatter light, i.e. the surface epithelium, the basal lamina and corneal endothelium. The amount of light scattering from the normal mouse cornea is better illustrated by the pixel intensity profile shown in Fig. 2A%, where specific peaks, a, b, and d, correspond to light scattering from the surface epithelium, basal lamina and endothelium respectively. Importantly, corneal keratocytes in the normal stroma show very little backscattering of light (Fig. 1E and Fig. 2A). On the other hand, TGFb1 over-expressing mice (Fig. 1B), show several notable changes. First, the surface epithelium appears dystrophic with smaller and more irregular cells (D) compared to normal surface corneal epithelial cells (C). This was associated with a slight thickening of the corneal epithelium and a more marked scattering of light from the suprabasal epithelial cells (Fig. 2A% and B%, peak a). Secondly, within the corneal stroma, there was considerable scattering of light from what appeared to be corneal keratocytes (Fig. 1F, arrows). This increased backscattering resulted in the elevation of the CMTF curve over the stromal region (Fig. 2B%, region c) and a slight but not significant increase in the stromal thickness. Overall, eight eyes from eight transgenic mice showed a significant increase in the entire corneal thickness compared to eight littermate controls (126.29 11.0 vs. 112.997.0 mm, PB 0.01, respectively). Quantitative measurement of the light scattering in the eight littermates and eight TGFb1 over-expressing mice also showed a significantly elevated stromal haze associated with the changes in the stromal keratocytes (18039 237 vs. 13549 206 U, PB 0.005). Clearly, more detailed studies are needed to determine whether the increased backscattering of light in the PEPCK-TGFb1 over-expressing mouse is due to the differentiation of corneal keratocytes to a myofibroblast phenotype or the deposition of abnormal extracellular matrix surrounding these cells. Nevertheless, these findings are consistent with the earlier observation that corneal keratocytes may contribute to the loss of transparency in response to corneal injury or the release of cytokines associated with the wound healing response. Light scattering from the Lum − / − knockout mouse cornea appeared distinctly different from that of the PEPCK-TGFb1 over-expressing mouse cornea. 3-D reconstructions of the cornea from Lum − / − knockout mice showed diffuse light scattering from predominantly the posterior region of the corneal stroma (Fig.
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
1287
Fig. 2. In vivo CMTF stromal images (A and B) and CMTF profiles (A% and B%) of control mice (A and A%) and TGFb over-expressing mice (B and B%). Two peaks corresponding to superficial epithelium (a) and endothelium (d) were noted in both groups (A% and B%). The epithelial –stromal interface (arrow, b) was defined by the first in-focus image of stromal tissue. It should be noted that entire stromal reflectivity was increased in TGFb over-expressing mice cornea (B%, region c), which correlated with the prominent appearance of stromal keratocytes (B, arrow). The pixel intensity curves have been offset in the y-axis and aligned to the corneal epithelial peak for demonstration purposes.
3A, arrow) rather than evenly throughout the cornea as detected in the PEPCK-TGFb1 over-expressing mouse. This localization was further confirmed by CMTF analysis (Fig. 3C, c-arrow), which showed a marked increase in the intensity of scattered light originating from the posterior stromal region closer to the corneal endothelium. 2-D images of light scattering also showed a more diffuse, uniform and indistinct pattern throughout the posterior stroma (Fig. 3B) rather than the discrete pattern suggestive of stromal keratocytes detected in the PEPCK-TGFb1 over-expressing mouse cornea. The diffuse nature of the light scattering was consistent with random light scattering that would be associated with the disorganized collagen fibril organization noted to occur in the Lum − / − knockout mouse cornea (Chakravarti et al., 1998). The finding that light scattering appears predominantly in the posterior cornea suggests that the posterior cornea may be more affected in these mice than the anterior corneal stroma. Alternatively, lumican may play a more critical role in the organization of collagen fibrils in the posterior cornea than the anterior cornea.
4. Discussion Measurement of light scattering by in vivo confocal microscopy has been previously used to evaluate the
development of corneal haze and clouding following photorefractive keratectomy (PRK) in patients (MollerPedersen et al., 1997; Moller-Pedersen et al., 2000) and rabbits (Moller-Pedersen et al., 1998a; Moller-Pedersen et al., 1998c), assess the effects of neutralizing antibodies to TGFb on development of haze following PRK in rabbits (Moller-Pedersen et al., 1998b), assess corneal haze after LASIK eye surgery in patients (Vesaluoma et al., 2000), and to monitor corneal responses following ocular irritation in the rabbit eye (Jester et al., 1998b; Maurer et al., 1999). These studies have not only provided important insights into the wound healing and irritancy response of the cornea, but have also lead to the discovery of a new and important class of keratocyte crystallin proteins that may play an important role in regulating corneal transparency at the cellular level (Jester et al., 1999a). The present study is the first to use in vivo confocal microscopy specifically to evaluate light scattering in the mouse cornea and to determine the tissue localization and pattern of light scattering from mice genetically altered to specifically explore the basic molecular and cellular mechanisms of corneal transparency. As expected in the Lum − / − knockout mouse, a diffuse pattern of light scattering was detected consistent with the disorganization of collagen fibril diameter and spacing previously shown to occur in these eyes (Chakravarti et al., 1998). Based on our current under-
1288
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
standing of the structural basis of corneal transparency established by Maurice (1957), light transmission through the cornea is dependent in part on the latticelike organization of small diameter collagen fibrils such that scattered light between collagen fibrils is destructively interfered. When spacing between collagen fibers is increased as occurs in corneal edema, there is a loss of this destructive interference and a random scattering of light throughout the swollen cornea. Likewise, increased scattering occurs when collagen fibril size increases and inter-fibrillar spacing becomes more irregular. Since collagen fibers (32 nm in diameter) are not resolved by confocal microscopy, which has a lateral resolution of approximately 0.2– 1.2 mm, random scattering of light would appear as a diffuse increase in the amount of light detected in the backward direction as observed in the Lum − / − knockout mouse. A similar diffuse increase in light intensity from the cornea is detected by in vivo confocal microscopy in irritated eyes that are markedly swollen (Jester et al., 1998b). Interestingly, however, light scattering in the Lum − / − knockout was preferentially localized to the posterior region of the cornea. This is consistent with recent observations that the posterior stroma is more affected in the Lum − / − knockout mouse, with both collagen fibril size and inter-fibril spacing markedly altered
(Chakravarti et al., 2000). The greater susceptibility of the posterior stroma is also consistent with the higher concentration of keratan sulfated proteoglycans in this region (Castoro, Bettelheim, & Bettelheim, 1988). While more detailed study is necessary, this finding suggests that lumican may play a more important role in controlling fibril diameter and spacing in the posterior cornea. Unlike the Lum − / − knockout mouse, the PEPCKTGFb1 over-expressing mouse showed a pattern of light scattering that appeared to be localized to corneal keratocytes. TGFb is known to increase extracellular matrix synthesis and is thought to control the development of fibrosis following corneal injury (Ohji, SundarRaj, & Thoft, 1993; Jester et al., 1996). In the PEPCK-TGFb1 over-expressing mouse, there is notable increase in the fibrotic response of major organs including the liver, kidney, fat and skin. It is possible that the increased light scattering seen in these corneas may be due to the deposition of irregularly arranged collagen fibrils immediately adjacent to the mouse keratocytes. Alternatively, TGFb1 is known to control myofibroblast differentiation in rabbit corneal keratocytes. Myofibroblasts show a unique pattern of light scattering localized to the cell body and extended cell processes, unlike the totally transparent nature of the
Fig. 3. In vivo CMTF images from the Lum − / − knockout mouse. 3-D reconstruction of a 5-month-old Lum − / − knockout mouse cornea (A) showed a marked increase in light scattering from the posterior region of the corneal stroma (arrow). 2-D images from the posterior region (B) showed a diffuse pattern of light scattering suggestive of random scattered light due to changes in the collagen fibril organization. CMTF analysis of the cornea (C) further supported the localization of the light scattering to the posterior stroma as indicated by the rise in light intensity (c, arrow) extending back toward the endothelium (peak d). a, epithelial peak; b, basal lamina peak; c, stromal region; and d, endothelial peak.
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
normal keratocyte. Myofibroblasts and activated keratocytes from the rabbit also show a greatly reduced expression of two water-soluble, crystallin proteins, transketolase and aldehyde dehydrogenase class 1. Although the pattern of light scattering in the PEPCKTGFb1 over-expressing mouse is not exactly similar to that observed for the rabbit corneal myofibroblasts, the cellular localization of light scattering is consistent with a cellular mechanism based on crystallin protein expression. Clearly, further work is required to determine more precisely the cause of light scattering in this model. Overall, in vivo confocal microscopy appears to be a sensitive and quantitative method for evaluating light scattering from the cornea in transgenic mice. Not only can it detect corneal clouding that is difficult to appreciate by macroscopic examination, in vivo confocal microscopy is also capable of localizing scattered light to specific structures within the cornea. In the current study, two different patterns of light scattering were detected, a diffuse scattering localized to the posterior cornea and a discrete, cellular based light scattering detected throughout the cornea. It is tempting to postulate that these differences are related to the distinct differences in the mechanisms by which the corneal transparency was altered, the first being a structural alteration of the collagen organization and the second being a cellular alteration in the cellular protein expression. While further work is needed to more precisely define these transgenic models, the application of in vivo confocal microscopy to studying transgenically altered mice will clearly provide new and important insights into the molecular and cellular basis of corneal transparency.
Acknowledgements This work was supported, in part, by grants EY07348 (JVJ), EY11654 (SC) and EY13215 (JVJ) from the National Eye Institute, Bethesda, MD, and Research to Prevent Blindness, Inc., New York, NY.
References Andrews, P. M., Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1991). Tandem scanning confocal microscopy (TSCM) of normal and ischemic living kidneys. American Journal of Anatomy, 191, 95 – 102. ARVO (1994). Statement for the use of animals in ophthalmic and vision research. In6estigati6e Ophthalmology and Visual Science, 35, v– vi. Blochberger, T. C., Vergnes, J.-P., Hempel, J., & Hassell, J. R. (1992). cDNA to chick lumican (corneal keratan sulfate proteoglycan) reveals homology to the small interstitial groteoglycan gene family and expression in muscle and intestine. Journal of Biological Chemistry, 267, 347–352.
1289
Castoro, J. A., Bettelheim, A. A., & Bettelheim, F. A. (1988). Water gradients across bovine cornea. In6estigati6e Ophthalmology and Visual Science, 29, 963 – 968. Cavanagh, H. D., Petroll, W. M., & Jester, J. V. (1993). The application of confocal microscopy to the study of living systems. Neuroscience and Biobeha6ioral Re6iews, 17, 483 – 498. Chakravarti, S., Magnuson, T., Lass, J. H., Jepsen, K. J., La Mantia, C., & Carroll, H. (1998). Lumican regulates collagen fibril assembly: skin fragility and corneal opacity in the absence of lumican. Journal Cell Biology, 141, 1277 – 1286. Chakravarti, S., Petroll, W. M., Hassell, J., Jester, J., Lass, J., Paul, J., & Birk, D. E. (2000). Corneal opacity in lumican-deficient mice due to defects in collagen fibril structure and packing in the posterior stroma. In6estigati6e Ophthalmology and Visual Science, in press Clouthier, D. E., Comerford, S. A., & Hammer, R. E. (1997). Hepatic fibrosis, glomerulosclerosis, and a lipohystrophy-like syndrome in PEPCK-TGF-b1 transgenic mice. Journal of Clinical In6estigation, 100, 2697 – 2713. Hanson, R. W., & Garber, A. J. (1972). Phosphoenolpyruvate carboxykinase, 1. Its role in gluconeogenesis. American Journal of Clinical Nutrition, 25, 1010. Imayasu, M., Petroll, W. M., Jester, J. V., Patel, S. K., Ohashi, J., & Cavanagh, H. D. (1994). The relation between contact lens oxygen transmissibility and binding of Pseudomonas aeruginosa to the cornea after overnight wear. Ophthalmology, 101, 371 – 388. Jester, J. V., Andrews, P. M., Petroll, W. M., Lemp, M. A., & Cavanagh, H. D. (1991). In vivo, real-time confocal imaging. J. Elect. Microsc. Tech., 18, 50 – 60. Jester, J. V., Barry, P. A., Cavanagh, H. D., & Petroll, W. M. (1996). Induction of a-smooth muscle actin (a-SM) expression and myofibroblast transformation in cultured keratocytes. Cornea, 15, 505 – 516. Jester, J. V., Barry-Lane, P. A., Petroll, W. M., Olsen, D. R., & Cavanagh, H. D. (1997). Inhibition of corneal fibrosis by topical application of blocking antibodies to TGF beta in the rabbit. Cornea, 16, 177 – 187. Jester, J. V., Huang, J., Barry-Lane, P. A., Kao, W. W., Petroll, W. M., & Cavanagh, H. D. (1999b). TGFb-mediated myofibroblast differentiation and a smooth-muscle actin expression in corneal fibroblasts requires actin re-organization and focal adhesion assembly. In6estigati6e Ophthalmology and Visual Science, 40, 1959– 1967. Jester, J. V., Li, H.-F., Petroll, W. M., Parker, R. D., Cavanagh, H. D., Carr, G. J., Smith, B., & Maurer, J. K. (1998a). Area and depth of surfactant-induced corneal irritancy correlates with cell death. In6estigati6e Ophthalmology and Visual Science, 39, 922– 936. Jester, J. V., Moller-Pedersen, T., Huang, J., Sax, C. M., Petroll, W. M., Cavanagh, H. D., & Piatigorsky, J. (1999a). The cellular basis of corneal transparency: evidence for ‘corneal crystallins’. Journal of Cell Science, 112, 613 – 622. Jester, J. V., Petroll, W. M., & Cavanagh, H. D. (1999c). Measurement of tissue thickness using confocal microscopy. In P. M. Conn, Confocal microscopy, vol. 307 (pp. 230 – 245). San Diego, CA: Academic Press. Jester, J. V., Petroll, W. M., Bean, J., Parker, R. D., Carr, G. J., Cavanagh, H. D., & Maurer, J. K. (1998b). Area and depth of surfactant-induced corneal injury predicts extent of subsequent ocular responses. In6estigati6e Ophthalmology and Visual Science, 39, 2610 – 2625. Lie, J., Jester, J. V., Cavanagh, H. D., & Petroll, W. M. (2000). On-line 3-dimensional confocal imaging in vivo. In6estigati6e Ophthalmology and Visual Science, 41, 2945 – 2953. Maurer, J. K., Parker, R. D., Petroll, W. M., Carr, G. J., Cavanagh, H. D., & Jester, J. V. (1999). Quantitative measurement of acute corneal injury in rabbits with surfactants of different type and irritancy. Toxicology and Applied Pharmacology, 158, 61 –70.
1290
J.V. Jester et al. / Vision Research 41 (2001) 1283–1290
Maurice, D. M. (1957). The structure and transparency of the cornea. Journal of Physiology, 136, 263–286. Maurice, D. M. (1962). The cornea and sclera. In H. Davson, The eye, vol. 1 (pp. 289 – 361). New York: Academic Press. Moller-Pedersen, T., Cavanagh, H. D., Petroll, W. M., & Jester, J. V. (1998c). Corneal haze development after PRK is regulated by volume of stromal tissue removal. Cornea, 17, 627–639. Moller-Pedersen, T., Cavanagh, H. D., Petroll, W. M., & Jester, J. V. (2000). Stromal wound healing explains refractive instability and haze development after photorefractive keratectomy. A one-year confocal microscopic study. Ophthalmology, 107, 1235 – 1245. Moller-Pedersen, T., Li, H.-F., Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1998a). Confocal microscopic characterization of wound repair following photorefractive keratectomy. In6estigati6e Ophthalmology and Visual Science, 39, 487–501. Moller-Pedersen, T., Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1998b). Neutralizing antibody to TGFb modulates stromal fibrosis but not regression of photoablative effect following PRK. Current Eye Research, 17, 736–747. Moller-Pedersen, T., Vogel, M., Li, H.-F., Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1997). Quantification of stromal thinning, epithelial thickness, and corneal haze after photorefractive kerate-
.
ctomy using in vivo confocal microscopy. Ophthalmology, 104, 360 – 368. Ohji, M., SundarRaj, N., & Thoft, R. A. (1993). Transforming growth factor-beta stimulates collagen and fibronectin synthesis by human corneal stromal fibroblasts in vitro. Current Eye Research, 12, 703 – 709. Petroll, W. M., Cavanagh, H. D., & Jester, J. V. (1993). Three-dimensional imaging of corneal cells using in vivo confocal microscopy. Journal of Microscopy, 170, 213 – 219. Petroll, W. M., Jester, J. V., & Cavanagh, H. D. (1996a). Quantitative three-dimensional confocal imaging of the corneal in situ and in vivo: system design and calibration. Scanning, 18, 45 –49. Petroll, W. M., Jester, J. V., & Cavanagh, H. D. (1996b). In vivo confocal imaging. International Re6iew of Experimental Pathology, 36, 93 – 129. Vesaluoma, M., Perez-Santonja, J., Petroll, W. M., Linna, T., Alio, J., & Tervo, T. (2000). Corneal stromal changes induced by myopic LASIK. In6estigati6e Ophthalmology and Visual Science, 41, 369 – 376. Zimmer, D. B., & Magnuson, M. A. (1990). Immunohistochemical localization of phosphoenolpyruvate carboxykinase in adult and developing mouse tissues. Journal of Histochemistry Cytochemistry, 38, 171 – 178.