Expression of Serine/Threonine Protein-Kinases and Related Factors in Normal Monkey and Human Retinas: The Mechanistic Understanding of a CDK2 Inhibitor Induced Retinal Toxicity

Abstract

Protein-kinase inhibitors are among the most advanced compounds in development using the new drug discovery paradigm of developing small-molecule drugs against specific molecular targets in cancer. After treatment with a cyclin dependent kinase CDK2 inhibitor in monkey, histopathological analysis of the eye showed specific cellular damage in the photoreceptor layer. Since this CDK2 inhibitor showed activity also on other CDKs, in order to investigate the mechanism of toxicity of this compound, we isolated cones and rods from the retina of normal monkey and humans by Laser Capture Microdissection. Using Real-Time PCR we first measured the expression of cyclin dependent protein-kinases (CDK)1, 2, 4, 5, Glycogen synthase kinase 3β (GSK3β) and microtubule associated protein TAU. We additionally verified the presence of these proteins in monkey eye sections by immunohistochemistry and immunofluorescence analysis and afterwards quantified GSK3β, phospho-GSK3β and TAU by Reverse Phase Protein Microarrays. With this work we demonstrate how complementary gene expression and protein-based technologies constitute a powerful tool for the understanding of the molecular mechanism of a CDK2 inhibitor induced toxicity. Moreover, this investigative approach is helpful to better understand and characterize the mechanism of species-specific toxicities and further support a rational, molecular mechanism-based safety assessment in humans.

Introduction

Protein-kinases are key modulatory molecules for a variety of cellular, metabolic and physiological processes. It is well known that specific protein-kinases play a pivotal role in different types of neoplasia, ranging from involvement in cell cycle regulation to constitutive signalling pathways activation (Manning et al., 2002; Blume-Jensen et al., 2001; Cohen, 2002; Shah et al., 2005). Therefore, pharmaceutical research is showing increasingly great interest in protein-kinases as pharmacological targets and in developing molecules that selectively inhibit or multi-target protein-kinases in many patho-physiological conditions (Fabian et al., 2005; Hirai et al., 2005)

As a consequence of treatment with such molecules, however, protein-kinases are also partially and reversibly inhibited in normal tissues whereby a secondary pharmacological event could produce the onset and development of an adverse effect or a toxicological liability, either morphological or functional (or both) (Reynolds, 2005; Heijine et al., 2005).

Specifically, we were focused on an anticancer drug, a CDK2, cell-cycle inhibitor belonging to the aminothiazole class (Pevarello et al., 2005). In Cynomolgus monkey dose range finding studies performed with this compound, in order to assess its maximum tolerated dosage, a significant damage in the photoreceptor layer from histological eye sections was observed.

Since this CDK2 inhibiting chemical compound also shows enzymatic activity versus a variety of other CDKs (Table 1), an in-depth genomics and proteomics study was planned to ascertain the presence of additional similar protein-kinases in monkey and human retina samples, thus attempting to discriminate whether the occurrence of this toxicity finding may possibly be due to an on-target (CDK2) vs. putative off-target effects (inhibition of other protein-kinases). One member of the CDK proteins family (namely CDK5) plays an important role in modulation of microtubules stability via phosphorylation of TAU (Billingsley et al., 1997), a protein directly involved in microtubules stability: our working hypothesis was that even reversible and temporary inhibition of CDK5 in photoreceptors might cause a specific insult to these cell types and promote the observed damage. Using LCM technology, a molecular profiling of the expression and activation status of several protein-kinases in this target tissue was performed with the intention to ultimately monitor drug effects on kinase activity modulation specifically at the single-cell level and understand the relevance of such findings in a clinical setting.

The resulting work demonstrates both how to profile the expression and the activity of pivotal target protein-kinases in healthy tissues at the cellular level as well as eventually predict the toxicological outcome of their inhibition. Single cells (in pharmacological efficacy experiments or when they represent a target of toxicity) are first identified in the histological sections and specimens subjected to laser capture microdissection (LCM) (Emmert-Buck et al., 1996; Bonner et al., 1997). By this approach we isolated cones and rods from monkey and human retina and by Real-Time PCR we measured the expression of CDK1, 2, 4, 5, GSK3β and TAU.

The presence or absence of the corresponding proteins was further assessed by immunohistochemistry (IHC), immunofluorescence (IF) and reverse-phase protein microarrays (RPPM) (Pawlak et al., 2002); phosphorylated forms of CDK5 and TAU were detected on frozen monkey eye sections. An analoguous translational study in human biological specimens also allowed prediction of the relevance and occurrence of this toxicity in patients

Materials and Methods

Toxicology Study and Experimental Design

Two males and one female Cynomolgus monkeys were treated with the vehicle (50% PEG 400 in 5% dextrose) as control animals and 2 males and 2 females were treated with a potent CDK2 inhibitor. Treatment consisted of 5 mg/Kg/dose; animals were dosed daily over a 6-hours infusion (2 mL/Kg/day) period on three consecutive days mimicking the proposed clinical regimen, followed by a 3-week observation period

Necropsy, Tissue Collection and Examination

The animals were sacrificed by exsanguinations from the femoral vein after sodium thiopental anesthesia at Day 23 (end observation period). Eyes were collected and fixed 24 hrs in Davidson’ fixative, embedded in paraffin, sectioned at 5 μm and stained with hematoxylin and eosin.

Tissue Collection for Gene and Protein Analysis

The eyes from two nontreated Cynomolgus monkeys were removed, and frozen in liquid nitrogen. Frozen human eyes were obtained from ABS (Analytical Biological Services GmbH) and treated as described below for LCM and RNA extraction followed by gene expression evaluation.

From frozen specimens (both monkey and human), 8 to 12 μm sections were cut with a cryostate (Shandon Cryotome) and mounted on Superfrost Plus Gold slides (DiaPath) or RNase-free glass slides (Arcturus Engineering) and stored at −80°C until use.

Immunohistochemistry

Immunohistochemical analysis was performed using the streptavidin-peroxidase technique and DAKO EnVision+ System, Peroxidase (DAKO) according to the manufacturer’s protocol, amplified with TSA (Perkin Elmer) and visualized with 3-amino-9-ethylcarbazole (AEC DAKO). Briefly, 12 μm frozen tissue sections were dehydrated and fixed in acetone. Slides were incubated for 10 min in 3% hydrogen peroxide to block endogenous peroxidase activity. After blocking with 10% Normal Growth Serum, 30 min at room temperature, the primary antibody was incubated for 1 hour at room temperature. A panel of primary antibodies was used for the detection of six proteins (Table 2). Secondary anti-mouse or anti-rabbit peroxidase-labeled polymer was used. Sections were then counterstained with Mayer’s hematoxylin solution.

Immunofluorescence

Sections of 12 μm thickness were used for immunofluorescence analysis. Sections were fixed with 4% paraformaldehyde for 15 min; after washing with PBS-Triton 0.1%, sections were incubated with the primary antibody for 1 hour. After a washing step, sections were incubated with the labeled secondary antibodies (1:400) for 1 hour. Both primary and secondary antibodies were diluted in PBS-TRITON 0.1%. Antibodies are listed in Table 3. Fluorescence signals were detected with a confocal microscope (Olympus). The position of the layers is usually identified from images obtained through a Nomarski differential interference contrast (Nomarski et al., 1955).

Laser Capture Microdissection (LCM)

The 8 μm frozen sections were thawed for 30s and fixed in 75% ethanol for 30s followed by a wash in RNase-free water for 30s. The sections were stained with HistGene (Arcturus Engineering) staining solutions for 15s followed by a wash with RNase-free water for 30s. The sections were dehydrated in graded ethanol solutions (75%, 30s, 95%, 30s, 100%, 30s) and cleared in xylene 5% for 1 minute. After air-drying for 30 minutes, the slides were kept in vacuum desiccators for a minimum of 2 hours

The retinal portion of some of the sections was scraped off with a clean razor blade for RNA and protein analysis as described next.

The remaining sections were used for microdissection using a PixCell instrument and CapSure LCM Caps (Arcturus Engineering) (Emmert-Buck, 1996). Cones and rods of retina were selected and captured from the sections. The PixCell II LCM system was set to the following parameters: 7.5 μm laser spot size, 100 mW power, and 1–2 ms duration

RNA Extraction and Retro-Transcription

The microdissected cells were lysed in a cap containing 20 μl of extraction buffer for 30 minutes at room temperature; each cap containing about 10,000 cells; 8 cap samples in total were collected. Total RNA was extracted from both microdissected cells as well as scraped tissue sections with the RNeasy mini Kit (Qiagen) according to the manufacturer’s protocol. DNase treatment was performed during the extraction procedure. In addition RNA was also extracted from 30 mg of the entire eye tissue.

RNAs from 4 caps were pooled and RNA concentration was measured using the NanoDrop spectrophotometer (Nan-oDrop). Scraped sections derived RNA was analyzed on an Agilent Bioanalyzer using the RNA PicoChip (Agilent Technologies)

50 ng of RNA were reverse transcribed in a total volume of 25 μl including random hexamers and 50 U of Superscript II RT (Invitrogen). After a first step of denaturation at 65°C for 5 minutes of the RNA and primers, the RT mix was incubated at 25°C for 10 minutes, at 42°C for 1 hour and at 72°C for 15 minutes for RT inactivation

Each cDNA was diluted 1:10 for further Real-Time PCR analysis.

Real-Time PC

Quantitative Real-Time PCR was performed using the 7900 Sequence Detector (Applied Biosystems) and both Taq-Man and SybrGreen systems. Then, 12.5 μl reaction was set up using the 2X Universal Master Mix, both TaqMan and SybrGreen (Applied Biosystems), 5 μl template cDNA and adequate concentrations of primers and probes. All samples were run in triplicate and each assay was repeated twice; data were analyzed using the standard curve method (User Bulletin #2: Rev B, Applied Biosystems) with serial dilution of cDNA made of a mix of 40 normal tissues RNA. Expression levels were normalized to the expression of three housekeeping genes, 18S ribosomal RNA (18S rRNA) (Applied Biosystems), ribosomal protein large subunit 19 (RPL-19) and tubulin, as internal controls.

Table 4 shows the primers and probes list; assays were designed using Primer3 (http://janus.itner.eu.pnu.com/cgi-bin/primer3-www.cgi) or Primer Express (Applied Biosystems) software, and where it was possible, the assays were designed across the exon-exon junction. Most of the sequences used were from human (high percent of homology with monkey) since monkey sequences were not published when the experiments were done.

Statistical analysis was performed using the two-tailed student t-test, differences with the p value less then 0.05 were considered statistically significant.

Protein Array Printing

Five samples of retinal portions scraped from frozen sections, as described above, were incubated at room temperature with 100 μl of lysis buffer, pooled and disrupted by shacking using the Mixer Mill MM 300 (Retsch), frequency 1/30s for 20s. Total protein content of the sample was determined by a modified Bradford test (Coomasie Protein Kit, Pierce).

The micro-dissected cells were incubated with 30 μl of lysis buffer per cap for 30 m at room temperature while shacking, after a brief spin vials were incubated at 70°C for 10 m and at 95°C for 2 m. Each cap contained about 30000 cells; three caps were prepared under comparable conditions to address the reproducibility of the sample preparation process. Recombinant GSK3β kinase was prepared according to protocol previously described (Bertrand et al., 2003) and used as standards on the chip. Standard solutions contained the pure kinases in a concentration range of 0 to 3000 ng/ml, containing additions of 0.1 mg/ml albumin. Protein solutions providing constant fluorescence contained 0.5 nM Cy5-labeled bovine serum albumin and were used as a reference for assay signals. All samples, LCM preparations, standards and references, were prepared in cell lysis buffer CLB2 (Zeptosens–a Division of Bayer (Schweiz) AG).

Generally, single droplets of 400 pL volume of each sample were arrayed onto ZeptoMARK Hydrophobic Chips by means of a commercial inkjet spotter NP1.2 (GeSiM). Each sample at each condition was printed in duplicate spots. Six identical arrays were printed on each chip. For the LCM preparations, the starting material for printing was 15 μl (~15 × 000 cells) per tip (two tips operated in parallel), thus ~0.4 cell equivalents of each preparation were deposited into each of the two duplicate spots.

Hence, 1 μl of dissected material is sufficient to produce several hundred arrays and to complete large and comprehensive analysis studies. In addition, printed chips and source plates for printing can be archived in the freezer and applied for further use upon need. After printing, chips were blocked with 10 mg/ml 3% BSA for 30 m, thoroughly washed with dd_H2O dried under nitrogen and stored at 4°C inthe dark until further use.

Assay Preparation, Array Readout and Data Analysis

For the experiments, chips were mounted onto fluidic structures, each one allowing the 6 replica arrays of a chip to be contacted individually with the respective assay solutions (15 μl chamber volume). On each array, a direct immunoassay was performed, incubating a single, different antibody, array-by-array. Arrays were incubated over-night, at room temperature, with a 1:500 dilution of anti-GSK3β, anti-phosphoGSK3b(Ser9) (#9332, #9336, both from Cell Signaling Technology Inc.) and anti-TAU (#sc-1995, Santa Cruz Biotechnology Inc.). After a thorough wash, a 1:500 dilution of fluorescence-labeled secondary anti-species antibody (Zenon Alexa Fluor 647 anti-rabbit Fab fragments, Molecular Probes) was applied for 1h at room temperature to generate the signals. After washing, the arrays were imaged in solution with the ZeptoREADER instrument (Zeptosens–a Division of Bayer (Schweiz) AG) at excitation/emission wavelengths of 635/670 nm and typical exposure times of 1–4s. For each antibody, two arrays were measured respectively, one in the presence and one in the absence (blank) of specific antibody.

Array images were analyzed with ZeptoVIEW PRO array analysis software (Zeptosens–a Division of Bayer (Schweiz) AG). For each array spot, background-corrected mean fluorescence signals were determined. All sample spots were locally referenced calculating the ratio of the mean sample and adjacent reference spot intensities (in RFI = Referenced Fluorescence Intensity units). Data points in the figures represent the mean, blank-corrected RFI signals of the duplicate spots, error bars indicate the standard deviations.

Results

Histopathology

Histopathological alterations in the eye retinal layers of photoreceptors and retinal pigmented epithelium were consistently observed in monkeys upon repeated treatment with the inhibitor. They were histologically characterized by diffuse loss of photoreceptors, with reduction of cones and rods nuclei in the outer nuclear layer, reduction of the density of photoreceptor inner segment and uniform reduction of the outer segment length, resulting in a less compacted photoreceptor layer. In addition, deeply pigmented cells formed intraretinal aggregates that indented the photoreceptor layer, or lined up along the retinal pigmented epithelium (Figure 1). Similar cells were also present throughout the sub-retinal space.

Photoreceptor Isolation by LCM from Monkey and Human Frozen Eye Sections

Histological slides were prepared from the whole eye and the quality of these specimens-specifically with respect to the retina layer-was investigated using standard staining techniques. The obtained slides showed excellent morphology and were indeed suitable for LCM experiments aimed at enriching the cones and rods layer (Figures 2 and 3A). Cones and rods were then dissected by LCM starting from frozen sections of the whole eye derived from nontreated monkey.

Figure 3 illustrates the monkey eye section before (panel A) and after shot (panel B) in panel C is shown the micro-dissected photoreceptor layer. Gene expression levels were evaluated by Real-Time PCR (Fink et al.,1998; Luo et al., 1999; Betsuyaku et al., 2001; Vincent et al., 2002; Fend et al., 1999), while protein expression was assessed in the first instance by immunohistochemistry and immunofluorescence and ultimately by Reverse Phase Protein Microarray (RPPM).

Real-Time PCR on Photoreceptors, Histological Sections and Entire Eye Sample

Initially, Real-Time PCR was performed on the cDNA obtained from the micro-dissected sample (LCM), the scraped section and total eye tissue. Total RNA from scraped sections is shown in Figure 4; 18S and 28S bands were intact and their corresponding picks well defined. RNA was not degraded and its integrity was preserved during cutting and staining sections.

After normalization against 18S, RPL-19 and tubulin genes, CDK1, 2, 4, 5, GSK3β and TAU expression levels were evaluated and compared among samples (Figures 5 and 6). While CDK1 and 4 were low abundant or not detectable, CDK2 and 5, GSK3β and TAU showed medium to high expression levels in each sample, both for human and monkey samples.

In monkey derived preparations (Figure 5) some differences were observed among the three different types of samples: CDK4 was not expressed in the LCM sample while in extract of total eye its expression at the gene level was low-medium. On the contrary GSK3β, CDK2 expression was significantly higher (p = 0.039 and p = 0.024 respectively) in cones and rods than in the total eye sample. Also CDK5 level in the LCM sample was slightly higher than in the total eye sample (p >0.05). TAU mRNA was similarly expressed in each sample and no significant variations were observed. The section sample showed an expression profile similar to that of LCM sample with the exception of CDK5 that showed a higher expression in cones and rods, albeit non-statistically significant (p >0.05)

Human gene expression profile (Figure 6) was similar to monkey for CDK2 and GSK3β with respect to the trend of expression levels among the 3 different samples. TAU was significantly (p = 0.005) higher expressed in photoreceptors than in the whole eye. The expression of CDK5 in human LCM sample was lower than in monkey.

Protein Detection by Immunohistochemical Analysis

CDK1, CDK2, CDK4, CDK5, GSK3β and TAU protein levels were further evaluated by immunohistochemistry (IHC) on the frozen sections of the monkey eye, to verify previous observations at the transcript level. These analyses highlighted a differential expression and tissue distribution both among the protein-kinases themselves as well as among different layers in the retina. As seen in Figure 7, focusing on the photoreceptor layer, CDK1 and CDK4 were expressed at very low levels or absent, thus confirming the gene expression data; CDK2 was low to medium expressed while CDK5, GSK3β and TAU were strongly expressed.

Except for CDK4, all proteins were present in the outer plexiform layer. Moreover CDK5, GSK3β and TAU were variably expressed almost in each retina layer. Table 5 provides a comprehensive summary of gene expression results and IHC findings.

CDK5 and TAU Phosphorylated Status in Cones and Rods: Immunofluorescence Staining

To investigate the functional role for some of the observed protein-kinases in these cells, their phosphorylation status within the sections was analysed in an immunofluorescence staining experiment using a confocal microscope (Olympus). We focused on total GSK3β protein, CDK5 and TAU in both total and phosphorylated forms (Tyr15 and Ser202, respectively). As shown in Figure 8, panel A, B and C each of these proteins is present in the photoreceptor layer; p-CDK5 (Tyr15) and p-TAU (Ser202) are also present in that layer (Figure 8, panels D and E, respectively). The total CDK5 was strongly recognized by the antibody in the lower part of the layer whilst its phosphorylated form in Tyr15 is preferentially localized in the upper part, suggesting a translocation of the protein from the cytoplasm to the nucleus as a consequence of protein activation (Figure 8, panel D).

Interestingly, p-TAU Ser202 is very abundant in the outler nuclear layer part of the layer and it appears to co-localize with the active form of p-CDK5 (Tyr15) (Figure 8 panel E).

Reverse Phase Protein Microarrays (RPPM): GSK3β, phospho-GSK3β and TAU Quantification

Reverse Phase Protein Microarrays (RPPM), generated from microdissected photoreceptor layer extracts, were probed to determine GSK phospho-GSK3β and TAU protein levels. Images of a typical protein arrays probed with GSK3β and TAU are depicted in Figure 9. Array signals for the three analytes were quantified and results are summarized in Figures 10 and 11.

Figure 9, panel A shows the spot signals of a calibration curve, with recombinant GSK3β protein deposited in a concentration range of 1–3000 ng/ml, and probed with the different antibodies. Only the anti-GSK3β antibody revealed prominent specific signals, as expected. No signals were observed for anti-phospho-GSK3β, and anti-TAU, which is indicative of the absence of phosphorylation of the recombinant protein, and of the absence of any cross-reactivity. Panel B shows the spot signals of three different LCM cap samples and the eye section extract at 4 different protein concentrations. Panel C shows the negative control signals obtained with buffer only.

Quantitative signals are depicted in Figure 10 and 11. Clear LCM signals were detectable for GSK3β, including its phosphorylated form on serine 9, and the TAU protein. GSK3β concentrations were calculated from the calibration curve and were equal to 19.0 ± 2.7, 8.0 ± 3.0 and 9.0 ± 1.2 ng/ml. Relative degree of GSK3β phosphorylation was estimated from maximum binding responses of two corresponding antibodies (Table 6). The maximum response signals were determined from additional experiments, applying different antibody dilutions to the arrays. These corresponded to 1:100 dilution for GSKβ and 1:500 dilution for the phospho-GSK3β antibody. Relative GSK3β phosphorylations in the three LCM samples were in a range of 12–26% (mean = 21%). Expression signals of the eye section extract were comparable to LCM sample signals. Unfortunately no CDK5 specific antibody reagent did work in the RPPM experiments, thus preventing us from assessing in a quantitative manner its expression level, both as total as well as phosphorylated form in monkey retinal layers.

Discussion

Because of the high degree of complexity and heterogeneity of animal tissues—derived either from biopsies or histological section—the gene and protein expression profiles specific for a given target cell are often masked and diluted by the presence of non-target cells, which often constitute the majority of the biological specimen. To overcome this problem, in a pivotal in vivo study with a protein-kinase inhibitor we used LCM to investigate its mechanism of toxicity; this technique permits to specifically select cells of interest in different tissues. LCM is a promising tool for the selective analysis of cell types within tissues: when an injury occurs after treatment with a compound, in order to apply genomic and proteomic technologies it is ideally necessary to analyse specifically the subset of cells that are involved in the pathological finding, while possibly correlating this to additional and more descriptive findings, like histopathological examinations, immunohistochemistry, clinical chemistry analysis, or electron microscopy evaluations (Guweidhi et al., 2004; Simone et al., 2000; Bonaventure et al., 2002).

The work described here focused on gene and protein profiling of five serine/threonine protein-kinases (namely CDK1, CDK2, CDK4, CDK5, GSK3β and one kinase-substrate TAU) in monkey and human retina samples. These investigations were aimed at understanding whether pharmacological inhibition of these kinases could produce a direct molecular insult to cones and rods and therefore alter the vision in treated animals.

Our data show that a CDK2 inhibitor, which also cross-reacts with a number of other, cell cycle related protein-kinases, possibly induces toxicological effects in the retina via modulation of the activity of one (or several) closely related protein-kinases family members. The same type of retinal toxicity with a non-selective ‘pan’-CDK inhibitor was recently observed in mice (Illanes et al., 2006). Also in this case after systemic exposure, the injury was localized at the photoreceptors layer and they supposed the inhibition of CDK5 as a possible cause of the observed damage.

Though the compound shows prominent activity on the CDK1 enzyme, CDK1 is not expressed in photoreceptors, thus leading to the reasonable hypothesis that the toxic effect could be due to either CDK2 or CDK5 inhibition (or both). Based on previous observations with other CDK2 inhibiting compounds and their pharmacokinetic properties (data not shown), we could exclude the on-target effect on CDK2 as the cause of toxicity and focused on CDK5 because the damage we observed may be related to a disassembly and subsequent derangement of cytoskeleton components, probably indicating that TAU and other microtubule associated proteins could be involved.

p-CDK5 on tyrosine 15 residue, which we have successfully identified and measured, is indeed one of the activated forms of CDK5 (Zukerberg et al., 2000; Cheng et al., 2003) while TAU is specifically phosphorylated on serine 202 by CDK5 and, to a lower extent, by GSKβ (Billingsley et al., 1997). Our results cannot rule out so far the possibility that TAU is indeed phosphorylated and active in the eye retina as a consequence of either CDK5 or GSKβ activity. Interestingly, however, limited studies were done to investigate CDK5 specific physiological function in the photoreceptor layer (Hayashi et al., 2000; Nakayama et al., 1999), it is well known that this protein has a fundamental role in post-mitotic nerve maturation, cytoskeleton regulation and synaptic transport (Dhavan et al., 2001).

CDK5 is also clearly different from its closely related kinase analogues (such as CDK2 and CDK4) since it is not related to either signalling or cell cycle regulation processes. It is tempting to speculate that CDK5 could act in a manner similar to a “transport chaperone” or be involved in the correct transport and maintenance of the key elements that maintain integrity of the photoreceptor membranes. Further experiments will be needed to test these hypotheses and best characterize the protein’s functional role in such a highly specialized tissue. In addition, the obtained results have lead directly to a comparison, at the molecular level, between Cynomolgus monkey and man, thus supporting the process of safety assessment for this compound to enter clinical trials. By means of gene expression profiling we discovered that CDK5 levels in human photoreceptors is lower than in monkey ones whilst TAU protein is strongly expressed, and presumably activated, in both species. This fact may underline that likely inhibition of CDK5 in humans can lead to a different effect from what we observed in monkey. Whether a different in vivo inhibition profile because of varying enzymatic expression levels will lead to more or less severe toxicological outcomes remains to be proven but the knowledge generated by this and similar approaches can significantly impact our understanding of developing CDK-targeted therapies as well as the biology and processes of vision in different animal species.

Since immunohistochemical and immunofluorescence analysis are hard to reproduce, difficult to quantify and are not applicable to a large number of samples such as genomic and proteomic approaches, protein arrays are now becoming the most appropriate tool to detect protein in LCM samples (Roberts et al., 2004). Due to the limited amount of starting material (about 30000 cells), the protein quantification in LCM samples is difficult to achieve, especially and more significantly for low abundant protein as protein-kinases in normal tissues (Martinet et al., 2004). Kinases are labile proteins, which execute fundamental roles within a cell but generally with a very short half-life and subjected to tight regulation and turn-over, both at the transcriptional as well as at the post-translational levels (Senderowicz, 2000). The most promising results for protein detection in laser-captured samples were obtained by global profiling by two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) (Jones et al., 2002) and surface-enhanced laser desorption/ionisation mass spectrometry (SELDI) (Kwapiszewska et al., 2004).

However, it is important to note that these techniques are not sensitive enough to detect and characterize low levels of proteins; moreover they are time consuming, generally low throughput and relatively expensive in terms of equipment. The RPPM technology that we illustrate in this work has the advantage of providing a solid relative quantitation of the selected protein-kinases after LCM; this is achieved by an ideal combination of surface chemistry, highly sensitive read-outs and application of robust immunological assays, developed by adopting similar or identical antibodies to the ones used for the IHC and IF experiments.

The final purpose is that the developed technique, combining Real-Time PCR and RPPM, can now be applied to treated versus untreated samples from in vivo experimental studies, in order to evaluate drug related effects and explore the molecular mechanism(s) at the basis of the observed histological injury in specific tissues which are reported to be targets of toxicity.

This study illustrates how gene expression and protein-based technologies can efficiently complement each other and be utilized to address biological questions linked to quantity and activity of rare and low abundant protein-kinases, as they act in a physiological and nonpathologically altered environment (Johnson et al., 2005; Kesavapany et al., 2004). These findings have also prompted us to specifically verify the expression and putative functional roles of CDK5 and GSKβ in the cones and rods layer from retina sections, paving the way for a better understanding of the complex biology of the eye and its relationship to functional protein-kinases (Nakayama et al., 1999; Lefevre et al., 2002).