The Effect of Graded Cyclic Stretching on Extracellular Matrix–Related Gene Expression Profiles in Cultured Primary Human Lamina Cribrosa Cells

Abstract

Purpose.: Cyclic stretching of the glial fibrillary acidic protein (GFAP)–negative lamina cribrosa (LC) cell in vitro is associated with transcriptomic changes in genes involved in extracellular matrix (ECM) dynamics in vivo, thereby implicating this cell type in the pathophysiologic changes of the optic nerve head (ONH) in glaucoma. The purpose of the study was to determine whether exposure to different grades of mechanical stretch progressively alters the expression of ECM genes in cultured LC cells.

Methods.: Primary cultures of human LC cells from three separate donors were maintained in static culture or exposed to low-level strain (3% ± 0.5% elongation, 1 Hz) for 24 hours. A baseline comparison of the expression of 62 genes involved in ECM dynamics was performed with low-density gene arrays (LDAs). The 3% protocol was used in a 24-hour period of baseline dynamic low-level stretch, and gene expression was compared with that occurring in a further 24-hour exposure to a 12.5% or a 20% stretch. Gene expression levels were determined by qRT-PCR.

Results.: LC cells displayed a nonlinear, transcriptional response to the mechanical stretch. Ten ECM-related and growth factor genes were altered by 3% strain versus static culture (nine downregulated and one upregulated). Increasing strain from 3% to 20% resulted in a significant increase in expression of 15 ECM-elated genes. Only one gene (epidermal growth factor) was increased between the 3% and 12.5% strains.

Conclusions.: Low-level, pulsatile, cyclic strain resets a lower baseline expression of several glaucoma-associated ECM genes. The LC ECM gene response occurs above a fourfold increase in baseline strain (12.5% strain) in vitro. The study supports the use of a nonstatic baseline when studying the effect of stretch (or strain) on the activation of ONH-derived, ECM-producing cells.

Elevation in intraocular pressure (IOP) and reduced vascular perfusion have both been implicated in the pathophysiology of optic nerve head (ONH) remodeling in glaucoma. ^1,2 Lowering IOP acts to reduce mechanical stress and strain at the ONH, improving perfusion pressure and lowering the translaminar pressure gradient, and it remains the only proven method of preventing the onset and progression of glaucoma. ³  

The ONH is a significant site of stress, in an otherwise rigid corneoscleral shell. ⁴ The lamina cribrosa (LC) supports retinal ganglion cell (RGC) axons as they traverse the corneoscleral shell to the optic nerve. There, the RGC axons weave through a convoluted plexus formed by the cribriform plates. ⁵ The cribriform plates are composed of collagenous columns with integrated elastin fibrils. These columns are composed mainly of collagen types I, III, VI, and VII. ^{6 –8} Astrocyte cells rest on the columns, anchored by a basement membrane composed of collagen IV, laminin, and proteoglycans. ⁹  

In the glaucomatous ONH, elevated IOP induces compression, stretching, and rearrangement of the cribriform plates, resulting in cupping of the optic disc. ¹⁰ Chronic structural changes of the LC closely correlate with visual field loss and neuronal cell death in open-angle glaucoma. ¹⁰ Studies of glaucomatous human eyes demonstrate that the architecture of the LC and distribution of its ECM components are altered in glaucoma. ^{9,11 –13} Within the cores of the cribriform plates, there is a marked disorganization and loss of mature elastin fibers and a preponderance of collagen IV. ^11,14 The appearance of collagen VI and a decrease in the density of collagen around elastic fibers alters the mechanical properties of the tissue during the glaucomatous process. ¹⁵ Connective tissue remodeling in glaucoma alters the biomechanical response of the LC to IOP elevation in an eye-specific manner. ¹⁶ Whether this remodeling response is physiologically protective or a pathologic process has yet to be determined. 

Two types of glial cells populate the LC: those that are positive for GFAP (the astrocytes) and those that are negative (the LC cells). ¹⁷ Results of studies in our laboratory suggest that the LC cells play an integral role in ECM remodeling. ^18,19 They produce the extracellular matrix (ECM) components of a glaucomatous phenotype when exposed to high-level mechanical strain. ¹⁸ They also adopt a profibrotic phenotype in response to treatment with transforming growth factor (TGF)-β1, ²⁰ which has been implicated as a key mediator in ONH remodeling. ²¹ LC cells are also believed to reside within or between the connective tissue plates of the LC and are therefore suspected of being important in the ECM response of this tissue to strain. ^5,22 Analogous to our findings in LC cells, Hernandez ²³ has also shown evidence that astrocytes participate in ONH remodeling. It appears that both intrinsic LC cell types play a role in the glaucomatous changes in the ONH. Although both cell types are implicated in the stretch-induced ECM response of the ONH, this article focuses solely on the LC cell. 

The ONH is continuously subject to a net retrograde displacement derived from the translaminar pressure gradients between the IOP and cerebrospinal fluid (CSF) pressures. ²⁴ This pressure differential fluctuates according to the pulsatile flow in the central retinal artery. In several studies, different levels and types of stress and strain in the eye have been examined in a variety of model systems. ^{25 –27} Downs et al. ²⁵ concluded the physiological level of strain in the sclera of rabbits and monkey eyes to be between 0% and 1%, suggesting that levels above 3.5% are pathophysiologic, as they result in detectable changes in scleral material properties. Using adult guinea pig optic nerves, Bain and Meaney ²⁸ estimated a strain threshold of 21% for morphologic axonal injury and 18% for deterioration of nerve function in the optic nerve. In a recent study of changes in ONH gene expression in response to elevations in IOP in rats, Johnson et al., ²⁹ using a baseline IOP of 28 mm Hg, demonstrated that an IOP increase of 43% results in extensive RGC axonal degeneration and induction of TGF-β1. Current models of the ONH by Roberts et al., ³⁰ taking the complex geometry of the LC into account, showed that, in certain areas of primate eyes, there is a normal strain of >2%. A major question arising from this work is whether a linear response to strain occurs or whether there is a level of strain at which induction of ECM genes begins. 

Several groups, including ours, have looked at the response of ONH cells to strain. ^18,29,31 In previous experiments, we examined the response of LC cells in stretched versus static culture. The cyclical strain application used led to large changes in strain over the course of each cycle (i.e., 0%–15%–0% 1 Hz). However, in vivo LC cells are subject to constant, small, pulsatile variations in strain above a non-0 baseline. 

In this study, we revisited LC matrix and matrix-related gene expression profiles between cells cultured in a dynamic low-level strain environment (i.e., 3% elongation) and those at two different levels of increased strain: 12.5% and 20% elongation. The straightforward objectives of these experiments were as follows: (1) to determine whether low-level strain alters the expression of ECM-related genes and (2) to investigate the pattern of ECM gene expression in LC cells exposed to different ratios of increase in strain in vitro. 

Methods

Cell Culture

Three primary cell culture lines of GFAP-negative LC cells from normal donors with no medical history of ocular disease were generated and maintained in DMEM supplemented with fetal calf serum, penicillin (100 U/mL), and streptomycin (100 μg/mL) under atmospheric conditions of 5% CO₂/21% O₂ at 37°C, as previously described. ^{18 –20} All donors gave informed consent and the acquisition of material was carried out in line with the principles of the Declaration of Helsinki. Each cell line was seeded in triplicate onto three separate wells of a stretch plate (Bioflex; Flexcell International, Hillsborough, NC). 

Application of Mechanical Strain

Confluent LC cells (passages 4–7, from three separate donors in multiple replicate cultures; n = 3) from donor eye explants were grown on laminin-coated stretch plates and exposed to 3% strain for 24 hours with a straining system (Flexercell Tension Plus FX-4000T; Flexcell International Corp.). The cells were then stretched at 3%, 12.5%, or 20% strain for a further 24 hours (Fig. 1). Triplicate controls for each donor were maintained in the same environment with no stretching for 48 hours. All cells were serum starved for 24 hours before stretching. A 0.5% stretch at 1 Hz (1 cyc/s) variation was added, to simulate the pulsatile nature of ocular blood flow. 

Figure 1.

 

The different degrees of cellular strain. Three samples from each of the three separate donors (n = 3) were initially exposed to 3% strain for 24 hours before exposure to three separate stretch conditions (3%, 12.5%, and 20% strain) for a further 24 hours. A baseline unstretched state for 48 hours was used as the control.

RNA Extraction

Total RNA (n = 3 for each level of strain) was extracted from each well (Trizol Reagent; Invitrogen, Dublin, Ireland), chloroform phase separation, and isopropanol precipitation. cDNA was generated by reverse transcription of 0.5 μg of DNase-treated total RNA, using the random primer method (Invitrogen). 

Low-Density Array

Transcript levels were determined by real-time PCR using primer and probe sets formatted as predeveloped assays and run on 384-well, low-density array plates (TaqMan; Applied Biosystems, Inc. [ABI], Warrington, UK; see Table 1 for 62 target genes). Probes were labeled with 5′-FAM and with 3′-TAMRA as a quencher. PCR was performed on a thermal cycler (model 7700; PerkinElmer, Wellesley, MA), with the following steps: (1) 2 minutes at 50°C; (2) 10 minutes at 95°C; (3) 15 seconds at 95°C; (4) 1 minute at 60°C; and (5) repeat from step 3 for an additional 39 cycles. 

Table 1.

 

The Response of 62 ECM-Related Genes to Strain

	3% Strain	12.5% Strain	20% Strain
Matrix Genes and Matrix Assembly Genes
Elastin (ELN)	NS	NS	NS
Biglycan (BGN)	↓ −0.45 ± 0.02-fold, P = 0.033	NS	↑ 3.27 ± 1.4-fold, P = 0.045
Collagen type I, alpha 1 (COL1A1)	↓ −0.59 ± 0.16-fold, P = 0.033	NS	↑ 2.80 ± 0.04-fold, P = 0.001
Collagen type I alpha 2 (COL1A2)	NS	NS	NS
Collagen type III alpha 1 (COL3A1)	NS	NS	NS
Collagen, type IV, alpha 2 (COLAA2)	NS	NS	NS
Collagen, type IV, alpha 3 (COLAA3)	↓ −0.62 ± 0.10-fold, P = 0.05	NS	NS
Collagen type V alpha 1 (COL5A1)	↓ −0.51 ± 0.12-fold, P = 0.006	NS	↑ 2.71 ± 0.24-fold, P = 0.038
Collagen type V alpha 2 (COL5A2)	NS	NS	NS
Collagen type VI alpha 3 (COL6A3)	NS	NS	↑ 2.32 ± 0.756-fold, P = 0.047
Collagen, type VIII, alpha 2 (COL8A2)	↓ −0.28 ± 0.06-fold, P = 0.008	NS	↑ 4.63 ± 1.57-fold, P = 0.021
Collagen type XI alpha 1 (COL11A1)	NS	NS	NS
Collagen type XIII alpha 1 (COL13A1)	NS	NS	NS
Collagen, type XIV, alpha 1 (COL14A1)	↓ −0.42 ± 0.03-fold, P = 0.034	NS	NS
Collagen type XVI alpha 1 (COL16A1)	NS	NS	NS
Collagen type XVIII alpha 1 (COL18A1)	↓ −0.61 ± 0.11-fold, P = 0.011	NS	↑ 3.42 ± 0.64-fold, P = 0.002
Versican (chondroitin sulfate proteoglycan 2)	NS	NS	↑ 3.55 ± 0.62-fold, P = 0.008
Perlecan (heparan sulfate proteoglycan)	NS	NS	NS
Tenascin XB (TXB)	NS	NS	NS
Fibulin 1 (FBLN1)	NS	NS	NS
Fibronectin (FN1)	NS	NS	NS
Laminin alpha 3 (LAMA3)	NS	NS	NS
Dystrophin (DMD)	↓ −0.57 ± 0.08-fold, P = 0.008	NS	NS
CRTL-1 (HAPLN1)	NS	NS	NS
SPARC	↓ −0.63 ± 0.11-fold, P = 0.001	NS	NS
Nidogen (NID1)	NS	NS	NS
Microfibril-associated glycoprotein-2 (MFAP2)	NS	NS	NS
Enzymatic/Chaperone Inducers of Collagen Assembly
Heat shock protein 47 (HSP47)	NE	NE	NE
Lysyl oxidase (LOX)	NS	NS	NS
ADAM TS1	NS	NS	NS
ADAM TS3	NS	NS	NS
ADAM2	NS	NS	NS
ADAM7	NS	NS	NS
ADAM8	NS	NS	NS
ADAM9	NS	NS	NS
MMP-1	NS	NS	NS
MMP-2	NS	NS	↑ 4.35 ± 1.6-fold, P = 0.031
MMP-9	NE	NE	NE
MMP-14	↑ 1.52 ± 0.25-fold, P = 0.033	NS	↑ 2.5 ± 0.65-fold, P = 0.017
TIMP-3	NS	NS	↑ 3.99 ± 1.47-fold, P = 0.02
TGF-b (Ligands, Receptors and Binders)
Transforming growth factor 1 (TGFB1)	NS	NS	↑ 4.21 ± 1.38-fold, P = 0.016
Transforming growth factor 2 (TGFB2)	NS	NS	NS
Latent TGF-β binding protein 4 (LTBP1)	NS	NS	NS
TGF-β type 1 receptor (TGFBR1)	NS	NS	↑ 2.80 ± 0.49-fold, P = 0.008
TGF-β type 2 receptor (TGFBR2)	NS	NS	NS
Betaglycan receptor (TGFBR3)	NS	NS	NS
Endoglin (ENG)	NS	NS	↑ 2.47 ± 0.68-fold, P = 0.035
Thrombospondin 1 (THBS1)	NS	NS	NS
Decorin (DCN)	NS	NS	↑ 2.28 ± 0.92-fold, P = 0.049
Other Growth Factors (Fibrosis)
Connective tissue growth factor (CTGF)	NS	NS	NS
Bone morphogenetic protein 7 (BMP-7)	NE	NE	NE
Fibroblast growth factor 2 (bFGF)	NS	NS	NS
Epidermal growth factor (EGF)	NS	↑ 2.53 ± 0.87-fold, P = 0.033	NS
Vascular endothelial growth factor A (VEGF)	NS	NS	↑ 3.16 ± 1.06-fold, P = 0.02
PreproEndothelin-1 (EDN1)	NS	NS	NS
Angiotensinogen (AGT)	NS	NS	NS
Angiotensin I converting enzyme (ACE)	NS	NS	NS
Renin (REN)	NE	NE	NE
Fibroblastic Activation Markers
S100 calcium binding protein A4 (S100A4)	NS	NS	NS
Vimentin (VIM)	NS	NS	NS
Alpha smooth muscle actin (ACTA1)	NS	NS	NS
Periostin (POSTN)	NS	NS	NS

 LC cells were exposed to a low-level dynamic strain (3% ± 0.5% at 1 Hz) for a period of 24 hours followed by a 24-hour stretch at 3%, 12.5%, or 20% strain. When genes were not expressed to a detectable level, they are denoted NE (not expressed). Genes with no significant alteration in expression in response to the different levels of strain are denoted NS (no significant difference).

Gene expression rates were compared by using 18s rRNA–normalized threshold cycle (CT) values. The replicates for each gene were examined to establish whether a reliable group mean could be obtained. When the group mean lay within 1.5 cycles (2.5-fold difference) of the range of individual NLC (nonlinear calibration) replicates, the group mean was considered reliable and used as a calibrator. Subtracting the NLC mean calibrator CT from each individual sample yields ΔCT. The equation 2−ΔCT was used to derive a ratio of change in gene expression. When original CT values for the target gene were above 40 cycles, it was designated as no detectable expression. Where no reliable average could be obtained for the NLC calibrator, comparative analysis was not pursued. 

Statistical Analyses

Gene expression data are presented as the mean ± SD. Data were analyzed using one-way ANOVA with post hoc Bonferroni testing to account for comparison of the expression of individual genes at various levels of strain. Statistical significance was set at P < 0.05. 

Results

Of the different experimental conditions, 3% and 20% strain provoked the greatest transcriptional alteration. Ten extracellular matrix–related and growth factor genes evaluated in the LDA were altered by 3% strain versus static culture (nine downregulated and one upregulated). Fifteen showed a significant upregulation between 3% and 20% strain. Only one gene (EGF) was increased at 12.5% strain. 

0% versus 3%

Collagen IAI (−0.59 ± 0.16-fold, P = 0.033), collagen IVA3 (−0.62 ± 0.10-fold, P = 0.05), collagen VAI (−0.51 ± 0.12-fold, P = 0.006), collagen VIII A2 (−0.28 ± 0.06-fold, P = 0.008), collagen XIVA1 (−0.42 ± 0.03-fold, P = 0.034), collagen XVIIIA1 (−0.61 ± 0.11-fold, P = 0.011), SPARC (−0.63 ± 0.11-fold, P = 0.001), dystrophin (−0.57 ± 0.08-fold, P = 0.008), and biglycan (−0.45 ± 0.02-fold, P = 0.033) were all significantly downregulated by 3% strain. MMP14 was the only gene to show a significant increase in mRNA levels at 3% stain (+1.52 ± 0.25-fold, P = 0.033; Fig. 2). 

Figure 2.

 

Transcripts showing alterations in expression levels between static (0%) and low-level (3%) dynamic strain conditions. LC cells were maintained in static culture condition or exposed to a low-level (3% ± 0.5% at 1 Hz) dynamic strain for a period of 48 hours. Total RNA was then harvested, and the expression levels of 62 glaucoma-related target genes were assessed by real-time RT-PCR. The graph shows the results for 10 genes, the expression of which was significantly altered by low-grade dynamic strain. *P < 0.05, 0% vs. 3% strain, n = 3 per group.

3% versus 12.5%

Only one gene showed significantly elevated mRNA levels at 12.5% versus 3% strain. EGF mRNA was increased at 12.5% strain (2.53 ± 0.87-fold, P = 0.033), compared with a baseline at 3% strain (Fig. 3). 

Figure 3.

 

Genes involved in ECM dynamics mRNA induction after stretching. LC cells were exposed to a low-level dynamic strain (3% ± 0.5% at 1 Hz) for a period of 24 hours, followed by a further 24 hours at a 3%, 12.5%, or 20% strain. Total RNA was then harvested, and the expression levels of 62 glaucoma-related target genes were assessed by real-time RT-PCR. The graph shows the results for five gene mRNAs involved in ECM dynamics, the expression of which was significantly altered by a 20% strain. *P < 0.05, 3% vs. 20% strain, n = 3 per group.

3% versus 20%

The following collagen α-chain genes were significantly upregulated at a 20% versus 3% strain: collagen IA1 (2.80 ± 0.046-fold, P = 0.001), collagen VA1 (2.71 ± 0.24-fold, P = 0.038), collagen VIA3 (2.32 ± 0.75-fold, P = 0.047), collagen VIIIA2 (4.63 ± 1.57-fold, P = 0.021), and collagen XVIIIA1 (3.42 ± 0.64-fold, P = 0.002; Fig. 4). The proteoglycans biglycan (3.27 ± 1.4-fold, P = 0.045) versican (3.55 ± 0.62-fold, P = 0.008), and decorin (2.28 ± 0.92-fold, P = 0.05) also showed increased expression with a 20% strain (Fig. 5). Vascular endothelial growth factor (VEGF; 3.16 ± 1.06-fold, P = 0.02) was upregulated threefold by 20% strain (Fig. 3). Certain matrix metallopeptidases (MMPs) similarly showed an increase in mRNA levels at a 20% strain. Expression of MMP2 and MMP14 was increased 4-fold (4.35 ± 1.6, P = 0.031) and 2.5-fold (2.50 ± 0.65-fold, P = 0.017), respectively (Fig. 3). 

Figure 4.

 

Collagen α-chain gene mRNA induction after stretching. LC cells were exposed to a low-level dynamic strain (3% ± 0.5% at 1 Hz) for a period of 24 hours, followed by a further 24 hours at a 3%, 12.5%, or 20% strain. Total RNA was then harvested, and the expression levels of 62 glaucoma-related genes were assessed by real-time RT-PCR. The graph shows the results for four collagen α-chain gene mRNAs, the expression of which was significantly altered by a 20% strain. *P < 0.05, 3% vs. 20% strain, n = 3 per group.

Figure 5.

 

Proteoglycan gene mRNA induction after stretching. LC cells were exposed to a low-level dynamic strain (3% ± 0.5% at 1 Hz) for a period of 24 hours followed by a further 24 hours at a 3%, 12.5%, or 20% strain. Total RNA was then harvested, and the expression levels of 62 glaucoma-related target genes were assessed by real-time RT-PCR. The graph shows the results for three proteoglycan gene mRNAs, the expression of which was significantly altered by a 20% strain. *P < 0.05, 3% vs. 20% strain, n = 3 per group.

Genes involved in the TGF-β signaling pathway were also significantly stimulated by a 20% strain. These were TGF-β1 (4.21 ± 1.38-fold, P = 0.016), TGF-β receptor 1 (2.80 ± 0.49-fold, P = 0.008), endoglin (2.47 ± 0.68, P = 0.035) (Fig. 6). 

Figure 6.

 

Expression of TGF-β signaling pathway ligands and receptors after stretching. LC cells were exposed to a low-level dynamic strain (3% ± 0.5% at 1 Hz) for a period of 24 hours followed by a further 24 hours at 3%, 12.5%, or 20% strain. Total RNA was then harvested, and the expression levels of 62 glaucoma-related target genes were assessed by real-time RT-PCR. The graph shows the results for three components of this pathway that were significantly increased by a 20% strain. *P < 0.05 3% vs. 20% strain, n = 3 per group.

Only 4 of the 62 genes (heat shock protein 47, MMP9, BMP7, and renin) were insufficiently expressed or absent, such that no signal was detected. 

Discussion

This study presents the results of a refined in vitro model of mechanical strain–induced profibrotic activation in LC cells. The approach allows for delineation of sequential changes in gene expression that occur with increasing strain above a nonstatic baseline. 

The model protocol was based on the fact that LC cells in the ONH are in a dynamic environment and therefore are continuously exposed to mechanical strain. The nature of the strain applied in this study differed from that used in our previous report in three major features. First, the frequency of cyclic strain remained at 1 Hz (1 cyc/s), but the amplitude varied to only a minor degree (0.5%), to mimic small oscillations resulting from pulsatile pressure. Second, we subjected the LC cells to a low-level baseline strain, whereas our previous study had used static conditions. Finally, in our previous work, we used 15% elongation as a pathologic level of strain; in this study, we used a level above (20%) and below (12.5%) the previously tested 15% strain level to ascertain whether there is a graded ECM response. 

As mechanical strain has previously been used to examine profibrotic activation of LC cells, ¹⁸ the results for 62 genes implicated in glaucomatous ECM changes were compared and contrasted with those found in our previous studies and in other relevant publications in the literature. 

Of the 62 genes chosen, we had previously shown increased expression in 19 of these genes in response to 15% mechanical strain, by using gene chip technology (Affymetrix, Santa Clara, CA). ¹⁸ Of these 19 genes, we found only 6 that were upregulated in the revised 20% strain model (collagen VIA1, collagen VIIIA3, biglycan, VEGF, versican, and TGFB1). The remaining 13 not found to be upregulated in our model were collagen XIV, collagen IVA1, elastin, tenascin XB, EMMPRIN, thrombomodulin, BMP7, TGFB2, perlican, cartilage-related linking protein, lysl oxidase, ADAMTS1, and latent TGF-β-binding protein. 

There remained a further subset of nine genes that we identified as being upregulated in our model (TGF-β receptor 1, endoglin, collagen IA1, collagen XVIIIA1, collagen VA1, MMP2, MMP14, TIMP3, and decorin), but which were not identified by microarray analyses in our previous study. ¹⁸ It is interesting to note that the increased expression of four of these genes has been documented in glaucomatous versus normal donor LC cell cultures (collagen 1A1, collagen VA1, decorin and TIMP3). ¹⁹ Five of the nine genes were also shown to be increased during the onset stage in a pressure-induced model of glaucoma in rats (TGF-β receptor 1, collagen IA1, collagen VA1, MMP2, and decorin). ²⁹  

We demonstrate that significant alteration in the expression levels of several ECM-related genes occurs when cells are moved from a static culture into an environment of low-level dynamic strain. We also show that only 1 of 62 genes studied was significantly altered between cells in low-level strain (3%) versus those placed under a fourfold higher level of strain (12.5%). Of all the genes studied, 15 were significantly increased by high-grade strain (17% increase in strain). 

Preconditioning the LC cells to a 3% strain for 24 hours before stretching significantly reduced mRNA levels in 9 of the 62 genes studied. Only (MMP14) showed an upregulation from 0% to 3% strain. 

To our knowledge, no previous work on LC cells has looked at the effects of different degrees of cyclical mechanical strain. The initial 3% strain resets the baseline transcriptional levels for a significant proportion of ECM genes, demonstrating a possible acclimatization. This result is significant, given that LC cells and the LC as a whole is under some degree of strain at normal IOP. ^{4,26,27,30,32,33}  

Although all cells and tissues of the body are exposed to some mechanical stress, LC cells populate a tissue that is exquisitely designed to act as a buffer of a translaminar pressure gradient (i.e., the interface of the intraocular and cerebrospinal fluid pressures). Therefore, studies of this cell type may be improved if a low level of strain in vitro is used as a baseline against which to gauge the effects of alterations in pressure. Relevant to this point, it has been noted that aortic valve interstitial cells similarly reduce specific gene profile expression when exposed to an in vitro approximation of normal environmental strain parameters. ³⁴ Smith et al. ³⁴ exposed aortic interstitial cells to various degrees of strain that resulted in reduced gene expression of VCAM-1, MCP-1, and GM-CSF. Similar to our results, Smith et al. showed that the lowest transcriptional expression of genes implicated in disease progression was at a level of strain representing normal conditions (namely, 15% strain for aortic valves). 

Of the 62 genes in this study, only 1 (EGF) was significantly altered at 12.5% strain. EGF has been implicated in the initial response of ONH cells to strain. ³⁵ Activation of EGF signaling pathways is responsible for altered cell shape, adhesion, and migration in response to pressure in ONH cells. ^36,37  

Johnson et al. ²⁹ recently published a rodent glaucoma model in which induction of certain ECM genes occurred only above a certain level of IOP elevation. In this glaucoma model, rodent eyes were subjected to various elevations in IOP by sclerosing the aqueous outflow pathways. Interestingly, no significant elevation in qPCR-based assessment of ECM gene expression was noted until an optic nerve injury grade of 3.5 (a subjective measure of axonal damage) was reached, equating to IOPs of approximately 36 mm Hg. The authors also showed biphasic responses of tenascin C, TIMP1, and fibulin 2. These three genes showed peak levels of expression at ∼36 mm Hg. 

Johnson et al. ²⁹ also showed a significant upregulation of TGFB1 at only the highest level IOP, but no elevation in TGFB2 gene expression was noted. This TGFB profile parallels our current findings. In contrast, Kirwan et al. ¹⁸ showed elevated TGFB1 and TGFB2 in a cyclical stretch model. Previous studies have shown increased expression in TGFB1 mRNA in response to 12 hours of cyclical mechanical strain. ³¹ Subsequent work involving LC cells treated with TGF-β1 showed an increase in most of the ECM and ECM-related genes (collagen I, collagen V, collagen VI, collagen VII, collagen VIII, VEGF, biglycan, decorin, and TGFB1) upregulated by 20% strain in the present study. The implication is that TGFB1 is a stretch-induced gene that in turn promotes other ECM gene alterations. Further work is necessary to deduce whether ECM genes are stretch responsive in the absence of TGF-β1. The concomitant involvement of TGF-β1 and mechanical stretching in ECM changes is well documented in several other tissues. ³⁸  

On the basis of the work presented here, we propose that the induction of ECM genes involved in ECM dynamics by LC cells in vitro occurs at levels of strain exceeding baseline strain by fourfold. Preconditioning LC cells, with a low level of cyclic strain, results in increased levels of strain being necessary to induce a detectable ECM response. While the study presents some further clarity on the dynamics of the in vitro strain model, this information cannot be extrapolated in terms of absolute units to the in vivo setting in the ONH, in which pressure thresholds are likely to be the net effect of a variety of factors, including translaminar pressure gradients and local tissue density and composition-dependent differences in compliance. 

However, this study clearly describes a model for studying strain-related responses in LC cells in a mechanically active in vitro model system. The study emphasizes the use of a nonstatic baseline and differential grades of strain as important refinements to studying ECM gene dynamics in LC cells in vitro. 

Footnotes

 Supported by grants from The American Health Assistance Foundation and Mater College.

Footnotes

 Disclosure: B. Quill, None; N.G. Docherty, None; A.F. Clark, None; C.J. O'Brien None