Ординатура / Офтальмология / Английские материалы / Mechanisms of the Glaucomas_Shields, Tombran-Tink, Barnstable_2008

Fig. 1. Evolving technologies to evaluate gene expression. (A) Use of a single-gene probe.

(B) Use of a probe representive the entire transcriptome.

to be isolated, cloned, and sequenced, and it is therefore a “known gene.” The gene probe can extend the whole gene cDNA or a small specific fragment, such as an oligonucleotide. The detection methods comprise molecular hybridization, linear DNA amplification (polymerase chain reaction, PCR), or a combination of both. Based on the genes’ function, genetics, and/or physiology of the tissue, this approach follows the assumption that a certain gene “could” be involved in the development of the disease. The evolving technologies to investigate the differential expression using a single gene are depicted in Fig. 1A and are as follows.

Hybridization to RNA Blotted Filters (Northern Blots)

This procedure relies on the hybridization of electrophoretically separated total RNA from normal and treated specimens to a labeled gene probe. The total RNA is run on a denaturing gel to open its secondary structure and blotted to a filter membrane by capillary action. The filter is then immersed into a salt solution with the labeled probe, incubated at the annealing temperature, washed, and exposed to an X-ray film to detect the specifically hybridized band Fig. 1A, left). In parallel, or in the same reaction, the filter is hybridized to a housekeeping gene, which is accepted as being expressed equally under the compared conditions under study. This procedure, still considered by many as the golden standard, allows to determine the real size of the transcript of the gene in question. It has therefore the ability to detect splice variants of the gene and whether or not such variants are affected differently by the disease. It is however a difficult technique that requires considerable amounts of very good quality, full length RNA and high technical skills.

Exponential Amplification of the Gene’s cDNA (PCR)

This procedure, which caused a revolution in current molecular biology back in 1985 (13), allows the sequential amplification of very small amounts of reverse-transcribed RNA (cDNA) to the point that the amplified product becomes detectable. A parallel amplification of a control housekeeping gene is necessary for proper normalization.

The experiment uses a pair of specific short DNA fragments (primers) flanking a small region of the gene’s cDNA, and a heat-resistant DNA polymerase. If the amplification is not carried to saturation but rather just to the number of cycles where amplification of both untreated and treated samples is linear, the levels of the resulting product would be an accurate representation of the differential expression of the given gene. Detection of the amplified product was originally determined by gel electrophoresis but it has now been replaced with a much higher sensitivity fluorescence method. The fluorescent label is provided either on the primers or on the nucleotide building blocks. The software of the instrument allows following the course of amplification and rationally selects the cycle at which all amplifications are exponential. Other than the need of caring for potential cross-contamination, these experiments are simple and because of the kits available, quite reliable. The drawback of this technique is that there is no guarantee that the amplified fragment truly corresponds to the gene of interest. Amplification conditions can cause primers to anneal to a different cDNA region of the transcriptome and produce artifacts and false positives. But once the proper conditions for a given set of primers have been set, comparison of the hybridization kinetics between treated and untreated samples would be rightly representative of the differential expression of the gene between treatments.

Exponential Amplification of cDNA Combined with Hybridization Probes (TaqMan PCR)

TaqMan PCR is the latest technology for the analysis of single gene expression. It includes the control for specificity that was missing in the above conventional real-time PCR. To accomplish this, in addition to using flanking primers to amplify the cDNA of the gene of interest, this upgraded procedure includes a gene-specific hybridizing probe that anneals to the amplified product as it is being made. The fluorescent label is incorporated in the probe rather than in the primers or nucleotides, thus assuring that the amplified detected product is gene specific (Fig. 1A, right). Probes and primers are available commercially in an easy single tube, and reactions are conducted in the same automated instrument as that of the standard PCR. Results obtained with this technology are very accurate and highly reproducible.

B. Use of the Entire Gene Expression Population as a Probe

An alternative approach for studying gene expression is that of searching for expression of all genes of the human transcriptome at once rather than for a single gene at a time. The technology uses complete human genome DNA sets (most often oligonucleotides) embedded in a solid surface, such as a filter membrane or a microchip. The collection of small DNA fragments in the solid surface is representative of all genes known to have been transcribed in the human genome at any given time. The probe used to hybridize to the solid surface is made on the whole population of RNAs expressed under the medical, pharmacological, or biological condition of interest. By hybridizing replicas of the same surface to probes obtained from the RNA of two parallel conditions (e.g. presence and absence of insult, normal versus diseased tissue, young vs. old, etc.), genes that are differentially transcribed can be identified. The functions associated with those differentially expressed genes are most likely to be

relevant for the particular condition studied. This technology is commonly referred to as a high-throughput study (see Fig. 1B).

The solid surface started by being a filter membrane (Fig. 1B, left) and has now almost entirely been substituted by that of a glass slide, DNA chip (Fig. 1B, right). DNA chips are commercially available and may contain either short oligonucleotides or somewhat longer cDNA fragments. They may also have spots for the entire genome or subsets of genes related with given pathways, such as oxidative stress, growth factors, ECM proteins, and so on and can also be custom made. Because of the “open-minded” approach of this technology, differential expression studies performed in this manner give the unique opportunity of uncovering potential new mechanisms and/or candidate genes. What was once belittled as a “lack of hypothesis/fishing expedition” approach may now hold the potential of a true breakthrough.

Macroarray Filters

This technology uses nylon filters that had been robotically spotted with fragments of cDNA. The cDNAs, including representative genes and internal controls, are gridded in a double-spotted pattern, holding eight genes in a small square (see Fig. 1B, left). The technology, though key for the development of the proof of concept of the system, was cumbersome and pretty much short lived. In a period of 3–4 years, it was replaced by the higher capacity, automated processed, fluorescence detection-based microarray. The macroarrays (membrane filters) varied in size and consequently in the number of genes each contained. They could contain either a random subset of eukaryotic genes (up to 18,000) (14,15) or a subset of genes associated with a given functional category (16). For each experiment, two identical membranes were hybridized to probes obtained from the two conditions of interest (e.g. elevated versus normal IOP). Membranes with the radioactively labeled spots were exposed to either an X-ray film or captured with a phosphoimager. Analysis between the identical membranes was conducted after normalization for overall intensity and expression of housekeeping genes.

Microarrays/GeneChips

The GeneChip systems, whose upgrade is almost constant, consist of highdensity oligonucleotide-based DNA arrays that are microfabricated using state of the art optical lithography technology (http://www.affymetrix.com/support/technical/ brochures/genechip_system_brochure.pdf).

The number of microarrays available has exploded. Leading companies in the field are now offering chips containing the entire transcriptomes not only from human but from a collection of eukaryotic species ranging from the cereal plant barley to the nematode C. elegans. Each chip provides multiple probes of each transcript (different regions of the same gene) (probe set) and numerous internal controls, which in combination with sophisticated software analysis packages result in a robust quantitation of gene expression. Hybridization to these chips is very involved and requires expensive and sophisticated instrumentation. Fortunately, most universities count with core facilities with specialized personnel that perform the whole procedure upon providing them with good quality RNA. The technology has also evolved as to need each time smaller quantities of RNA for the hybridization step. Newly developed small protocols allow

the use of amounts as low as 50 ng of RNA per chip, which for tissues of the eye does allow the processing of samples from a single individual.

Evaluation of the Microarray Data

The wealth of data generated even by a single comparison can be overwhelming and investigators need to focus on what they are looking for. Two kinds of output files are obtained. “Basic data” files contain a list and description of all genes in the chip (e.g., 54,676 in the U133 plus2 Affymetrix GeneChip) together with a number of parameters which include signal intensities and detection p-values. For this, expression analysis algorithms used by the instrument software averages signal intensity values of the probe set and determines whether their corresponding gene is present, absent, or marginally present in the sample. Comparative analyses between treated and untreated samples are performed by the same software after the normalization of the two chips for their different intensities. The “comparative” output files are similar to the basic ones but also include signal log ratios and change p-values that will be used to convert the expression differences to “fold changes” and their significance. Further analyses and sorting of the comparative data (“data mining”) is accomplished with the help of additional specific software, which also provides a more comprehensive description of the gene. Guided by the common sense approach that highly expressed genes are bound to have a relevant role in the physiology of a particular tissue, a number of original reports focused on identifying the most upor down-regulated genes. This type of analysis can be done with a simple excel program (and some patience) by eliminating absent–absent calls, sorting data for fold change, p-values and implementing filters for the selection of relevant genes (e.g. changes greater that twofold with p-values lower than 0.01, consistently upor down-regulated in three repeated biological samples, etc.). Alternatively, the data can be imported into one of a number of available programs to obtain a more refined analysis that includes re-normalizing of the intensities, increased choice of statistical tests, filtering for genes of interest, filtering flags, and excellent graphic representations.

MODELS TO STUDY GENE RESPONSE TO MECHANICAL STRESS: ELEVATED IOP AND STRETCH

Mechanical forces, including fluid mechanical stimuli, are known to influence cellular pathways and the expression of cellular genes. Genes responding to mechanical insults could be encoding proteins that maintain the physiological state, respond to stress, proteins that try to counteract an insult (homeostasis), or those that can cause direct damage to cells and tissue. Because the trabecular meshwork is subjected to the continuous flow of aqueous humor, expression of the genes of this tissue under pressure insults holds the capability of revealing functions that would be key to its physiology. To identify genes encoding these functions, the appropriate mechanical stimulation has to be applied. A unique, important characteristic to take into consideration is the fact that the pressure in the outflow pathway decreases after crossing the inner wall of the SC, that is, the pressure in the trabecular meshwork proximal to the canal is higher than the pressure in the canal’s lumen (17,18). These mechanical forces are for instance, not present in traditional cell cultures, where the bottom of the dish

TaqMan PCR or DNA genechips. An advantage of this model is that it allows the use of post-mortem human eyes at close to physiological conditions. Because of the well-established fact that outflow pathways of humans are quite different than those of vertebrate animals (29), the capability of studying this type of gene expression in humans is an important advantage. Further, because this model compares eyes from the same individual, it provides the additional assurance that differential gene expression studies will not be confounded by individual genetic differences. The drawback of this system though is that the equipment and software are not commercially available and that the human material is scarce and has an elevated cost.

Animal Models

There are a number of animal models of elevated IOP. Just a few of them are genetic and have resulted from an IOP increase in a transgenic animal. A mouse model, the DBA/2J, is a secondary angle-closure glaucoma due to iris atrophy and pigment dispersion (30). With age, the mouse develops elevated IOP due to pigment release from the iris and obstruction of the outflow pathway. The obstruction does not damage the trabecular meshwork tissue and could be considered a quite natural model of glaucoma. However, the period when DBA/2J mice develop IOP goes from 6 to 16 months and is not synchronous in all eyes (31), bringing in difficult evaluation problems. In addition, the RGC cell loss and optic nerve damage in the DBA/2J mouse can be rescued with high-dose radiation and bone marrow transfer without reducing IOP (32). This is an indication that intrinsic retinal defects other than IOP might be causing the glaucoma observed in this model and reduces its applicability to study the relevance of a dysfunctional trabecular meshwork. Another genetic model has recently been identified on a retinal degeneration rat model RCS-rdy−strain that spontaneously develops elevated IOP (33). It is not known though whether this elevated IOP is caused by a dysfunction of the trabecular meshwork or as a consequence of other retinal defects.

Most other animal models of chronic elevated IOP available have been obtained by sealing the outflow pathway or by inducing sclerosis in the trabecular meshwork and thus damaging the tissue. Such models include laser treatment of the angle of the eye (34–37), episcleral venous occlusion in rats or porcine (38–40), and injection of a diluted concentration of saline into the rats’ episcleral veins (41). Although useful to determine effects of pressure on gene expression of the RGC, these models are of no use to study trabecular meshwork function.

Currently, new approaches for the development of elevated pressure animal models with an intact trabecular meshwork are being pursued by our laboratory using gene transfer methods. These models are attempted to be created by delivering detrimental genes to the trabecular meshwork tissue with the aid of viral vectors. With the use of tissue preferred and/or inducible promoters, if successful, these transient transgenic animals could be a unique source for elevated pressure gene studies.

Human Genes Expressed During Pressure-Induced Homeostatic Response

Elucidating the molecular events that take place during the homeostatic response is of great importance. It would lead to the understanding of how individuals respond to pressure and would open the door to figuring out which mechanisms the trabecular meshwork uses to regulate and manage pressure spikes. Initial experiments using northern blots, individual probes, and the human perfusion model provided us with the proof of concept that differential gene expression under an IOP insult was occurring and could be measured. Probes for myocilin and stromelysin (MMP3) showed that the expression of these two relevant genes was differently regulated. Stromelysin was induced after 6 h of elevated IOP, whereas induction of myocilin occurred only after a prolonged insult. In 2004, studies were conducted in our laboratory to elucidate genes whose expression was altered during the homeostatic period (2–4 days) (15). Because of the relevance of using samples from single individuals, those experiments used macroarray membranes rather than gene chips, which at the time required an amount of RNA higher than the one obtained from a single human trabecular meshwork.

The study identified 40 upand 14 down-regulated genes whose expression is consistently altered in three individuals (68–77 years old). An overall finding was that functions of the proteins encoded by the altered genes were representative of several mechanisms. This indicates that the response to pressure appears to be mediated through more than one pathway. Some of the identified proteins have been previously and independently associated with regulating outflow facility, but unexpectedly, among those switched on/off were genes known from the literature to be involved in the physiology of unrelated tissues, such as bone and cartilage, and in vascular calcification. Those genes had never been associated before with the trabecular meshwork.

A summary of selected genes upand down-regulated during the homeostatic response is included in Fig. 5 (21,61–75). A few of these genes deserve special attention. Three of them have been associated with processes of bone metabolism and vascular calcification. The Matrix Gla (MGP) gene encodes a vitamin K-dependent protein that binds calcium and is expressed in tissues that produce an uncalcified ECM. MGP was initially discovered in demineralized extracts of bone (76) but was found to be highly expressed in a variety of tissues like cartilage and vascular smooth muscle cells (77,78). MGP was proven to be an inhibitor of calcification when mice, either treated with warfarin or having their MGP gene knockout, developed extensive calcification of the arteries and cartilaginous tissues (79,80). Today, it is well established that protection of soft tissue calcification is partly because of the complexes of calcium to inhibitors of calcification such as MGP (63,81). The first hint of the potential relevance of MGP in the trabecular meshwork came when independent human libraries from this tissue showed MGP among the ten most abundant genes of the trabecular meshwork (12,82,83). It was then speculated that the MGP presence in the trabecular meshwork was to avoid a detrimental hardening of this soft tissue. Later, in the homeostatic pressure study, MGP was highly up-regulated in the three individuals analyzed (15). The combined findings place MGP as an important player in trabecular meshwork physiology and suggest that ECM softness by preventing calcification could be an important mechanism to influence outflow facility. Such a mechanism could be an active part of the homeostatic processes that tend to adjust the elevated pressure insult.

From data on Vittitow and Borrás, J of Cellular Physiology 201, 126 (2004).

Fig. 5. Selected Human Genes Modified During Pressure-Induced Homeostasis Encoding Potentially Relevant Functions to TM Physiology.

In the same line of thought on trying to maintain softness of the trabecular meshwork tissue during the elevated pressure, another osteogenic gene osteomodulin is down-regulated. Osteomodulin, also known as osteoadherin, is expressed in mature osteoblasts and has been proposed to have a role in mineralization of the ECM (84). Osteomodulin has been identified as part of the mechanosensitive genes in osteoblasts (85). A down-regulation of an ossification protein during the adaptive mechanism would appear as a good choice, as supposedly, it could help to insure the softness of the tissue and contribute to better outflow. The existence of a calcification process in the human trabecular meshwork has recently been proposed (81,86).

The osteoblast-specific factor 2, also known as or periostin, was originally identified as a protein secreted by osteoblasts that supported adhesion and increased its expression by TGF 1 treatment (87). Recently, it has been reported that periostin can regulate collagen type I fibrillogenesis and thereby serves as a mediator of the biomechanical properties of connective tissues (88). Whether influencing adhesion or collagen type I fiber formation, an increase in the expression of periostin in the trabecular meshwork