Cellular and Molecular Neurobiology

, Volume 31, Issue 3, pp 469–478

Effect of the PARP-1 Inhibitor PJ 34 on Excitotoxic Damage Evoked by Kainate on Rat Spinal Cord Organotypic Slices

Authors

  • Graciela L. Mazzone
    • Neurobiology SectorInternational School for Advanced Studies (SISSA)
    • Neurobiology SectorInternational School for Advanced Studies (SISSA)
    • Spinal Person Injury Neurorehabilitation Applied Laboratory (SPINAL)Istituto di Medicina Fisica e Riabilitazione
Original Research

DOI: 10.1007/s10571-010-9640-7

Cite this article as:
Mazzone, G.L. & Nistri, A. Cell Mol Neurobiol (2011) 31: 469. doi:10.1007/s10571-010-9640-7

Abstract

Excitotoxicity triggered by over-activation of glutamate receptors is thought to be an early mechanism of extensive neuronal death with consequent loss of function following lesion of spinal networks. One important process responsible for excitotoxic death is ‘parthanatos’ caused by hyperactivation of poly(ADP-ribose) polymerase (PARP) enzyme 1. Using rat organotypic spinal slices as in vitro models, the present study enquired if 2-(dimethylamino)-N-(5,6-dihydro-6-oxophenanthridin-2yl)acetamide (PJ 34), a pharmacological inhibitor of PARP-1, could counteract the excitotoxic damage evoked by transient application (1 h) of kainate, a potent analogue of glutamate. Kainate induced dose-dependent (1 μM threshold) neuronal loss (without damage to astrocytes) detected 24 h later via a PARP-1 dependent process that had peaked at 4 h after washout kainate. All spinal regions (ventral, central and dorsal) were affected, even though the largest damage was found in the dorsal area. Whereas PJ 34 did not protect against a large concentration (100 μM) of kainate, it significantly inhibited neuronal losses evoked by 10 μM kainate as long as it was co-applied with this glutamate agonist. When the application of PJ 34 was delayed to the washout time, neuroprotection was weak and regionally restricted. These data suggest that kainate-induced parthanatos developed early and was prevented by PJ 34 only when it was co-applied together with excitotoxic stimulus. Our results highlight the difficulty to arrest parthanatos as a mechanism of spinal neuron death in view of its low threshold of activation by kainate, its widespread distribution, and relatively fast development.

Keywords

Organotypic cultureKainic acidParthanatosSpinal cord injuryExcitotoxicity

Introduction

The pathophysiological mechanisms responsible for the early damage to spinal networks following acute spinal injury are incompletely understood. A number of investigations have, however, indicated that massive release of glutamate is a major contributor to trigger such a damage via a process termed excitotoxicity (Lau and Tymianski 2010). To elucidate the downstream effector systems activated by excitotoxicity, we have recently employed a large dose of the metabolically stable glutamate agonist kainate to produce extensive neuronal loss in the isolated rat spinal cord in vitro, which is a useful model to investigate the basic mechanisms of spinal injury (Taccola et al. 2008; Mazzone et al. 2010). Such a treatment evokes hyperactivation of poly(ADP-ribose) polymerase (PARP) enzyme 1 (PARP-1) that is responsible for about 90% of the cell poly(ADP-ribosylation) activity (Giansanti et al. 2010) and whose pharmacological inhibition has been successfully employed as a novel therapeutic strategy to counteract the cytotoxic effects of DNA-damaging agents (Sandhu et al. 2010). PARP-1-mediated neuronal death is believed to be a common mechanism triggered by ischaemia and injury, and can be classified as ‘parthanatos’ (Andrabi et al. 2008). This process involves generation of high concentrations of poly-ADP-ribose (PAR), which is toxic to the cell energy metabolism of spinal networks (Kuzhandaivel et al. 2010b). In this model, intracellular production of PAR leads to nuclear translocation of the apoptotic inducing factor (AIF) with subsequent DNA damage: partial neuroprotection can be observed by application of the PARP-1 inhibitor 6-5(H)-phenanthridione (PHE; Kuzhandaivel et al. 2010b).

Nevertheless, since PHE is not a very selective PARP-1 inhibitor, validation of these results with a more specific drug is required. 2-(Dimethylamino)-N-(5,6-dihydro-6-oxophenanthridin-2yl)acetamide (PJ 34; Abdelkarim et al. 2001) has recently been proposed as a selective blocker of PARP-1 and associated parthanatos with reportedly beneficial effects in animal models of brain and spinal cord ischemia in which excitotoxicity is thought to play a major role (Virag and Szabo 2002; Casey et al. 2005; Besson 2009; Kauppinen et al. 2009; Moroni et al. 2009).

The present report has employed organotypic slices of the rat spinal cord to investigate if PJ 34 could protect them against kainate-mediated excitotoxicity. Organotypic cultures have recently been used, as a model, to mimic early pathological processes undergoing traumatic injury (Cater et al. 2007). In keeping with this notion, a recent study from our laboratory has shown that such cultures undergo very similar processes of excitotoxicity like those observed in the spinal cord (Mazzone et al. 2010), thus offering an advantageous system to test the efficiency of neuroprotective agents on spinal networks. The present study, therefore, aimed at exploring whether PJ 34, employed at concentrations used for cell protection studies in vitro (Formentini et al. 2009), could spare neurons from either strong or threshold excitotoxic stimuli evoked by kainate. Furthermore, the present study investigated the excitotoxicity dependence on kainate concentration and its time evolution.

We adopted two protocols to apply PJ 34, namely simultaneously with kainate or after 1 h administration of kainate. Thereafter, kainate was washed out and PJ 34 continuously applied for 24 h. We then sought to find out the consequences for neuronal and glial populations investigated in three different region of the organotypic slice.

Materials and Methods

Preparation and Maintenance of Organotypic Cultures

Embryonic organotypic slice cultures of spinal cord were prepared from pregnant Wistar rat, at 13 days of gestation, in accordance with our standard procedure as recently published (Mazzone et al. 2010). The foetuses were delivered by caesarean section from timed-pregnant rats, in accordance with the National Institutes of Health guidelines and the Italian act Decreto Legislativo 27/1/92 n. 116 (implementing the European Community directives n. 86/609 and 93/88). All efforts were made to reduce the number of animals used and to minimize animal suffering.

Slices were cultured in a medium containing 82% Dulbecco’s Modified Eagle medium, 8% sterile water for tissue culture, 10% foetal bovine serum (FBS; Invitrogen, Italy), osmolarity 300 mOsm, pH 7.35, and kept in culture for 22 days in vitro (DIV) before use. During the first 5 days of incubation the medium was supplemented with 25 ng/ml nerve growth factor (NGF), subsequently reduced to 5 ng/ml (complete culture medium). Dulbecco’s modified Eagles’ medium high glucose (DME/HIGH), penicillin and streptomycin were purchased from Euroclone (UK). Foetal calf serum was obtained from Invitrogen (Carlsbad, CA, USA). NGF was from Alomone Laboratories (Jerusalem, Israel), chicken plasma was from Rockland Immunochemicals (Gilbertsville, PA, USA), thrombin was from Merck KGaA (Darmstadt, Germany), while the other reagents were purchased from Sigma-Aldrich, Milan, Italy. Kainate was purchased from Ascent Scientific (Weston-Supermare, UK).

Protocol for Excitotoxicity and its Inhibition by the PARP-1 Inhibitor PJ 34

At DIV 22 spinal cord organotypic slices were treated with kainate for 1 h (various concentrations in the range of 1 μM–1 mM) dissolved in complete culture medium. PJ 34 (Sigma-Aldrich, Milan, Italy; 30 μM) was either co-applied with kainate or administered at the time of kainate washout. In either case PJ 34 application continued for the following 24 h when cultures were fixed for immunohistochemistry.

We used PJ 34 at a standard concentration of 30 μM because previous studies on cell cultures have indicated that this concentration protects neurons against anoxia (Abdelkarim et al. 2001), NMDA neurotoxicity (Goebel and Winkler 2006) and induction of PARP-1 neuronal death (Fossati et al. 2007). Control preparations were, in each experiment, untreated sister cultures maintained in vitro for 24 h.

Immunofluorescence Organotypic Cultures

Slices were fixed in 4% paraformaldehyde for 1 h at room temperature and stored in phosphate buffer saline (PBS) until use. Cultures were processed for immunofluorescence analysis in accordance with our previous report (Mazzone et al. 2010). Briefly, slices were blocked with 3% foetal calf serum (FCS), 3% bovine serum albumin (BSA), 0.3% Triton in PBS (blocking solution) for 1 h at room temperature, followed by overnight incubation at 4°C in a blocking solution containing the mouse NeuN antibody (1:250 dilution; clone A60; Millipore, Milan, Italy), and the rabbit polyclonal anti-S100 (1:1000; Dako, Glostrup, Denmark) or the mouse monoclonal against poly ADP ribose (PAR, 1:50; Serotec, Kidlington, UK).

The primary antibody was visualized using the corresponding secondary fluorescent antibody (Alexa Fluor 488 and 546, 1:500 dilution, Invitrogen, Carlsbad, CA, USA). To visualize cell nuclei, slices were incubated in 1 μg/ml solution of 4′,6-diamidino-2-phenylindole (DAPI) for 1 h and mounted using Vectastain mounting medium (Vector Laboratories, Burlingame, CA, USA). DAPI staining results were analyzed using a Zeiss Axioskop2 microscope. NeuN-positive cells were analyzed (at 488 nm) using a confocal microscope (Leica DMIR2) equipped with an Ar/ArKr laser. The S100 mean fluorescence intensity of three regions of interest was quantified by densitometry analysis of a 500 × 500 μm area, using Zeiss Axioskop2 microscope and MetaVue software (Molecular Devices, Sunnyvale, CA, USA). PAR-positive cells were analyzed with confocal microscopy and their number was quantified with ‘eCELLence’ software (Glance Vision Tech, Trieste, Italy).

The identification and quantification of cell death, using DAPI nuclear staining in the three different regions of interest, namely dorsal, central and ventral, were analyzed in each slice as previously described (Mazzone et al. 2010). In particular, the relative size of each area was arbitrarily chosen as it represented about 1/3 of the total size of the slice. As shown in Fig. 3b (black columns), the number of NeuN positive cells was similar in all three regions. Since organotypic cultures were used after 22 days in culture, the number of cells had stabilized and dead elements had been eliminated by the use of the roller system (Spenger et al. 1991; Streit et al. 1991).

Data Analysis

The database of the present study comprises at least 15 cultures, each one yielding, on average, 30 Petri dishes. Thus, n = number of slices unless otherwise stated. Statistical analysis was carried out using SigmaStat (SigmaStat 3.1, Systat Software, Chicago, IL, USA). After distinguishing between parametric and non-parametric data, parametric values were analyzed with the one-way ANOVA for multiple comparisons (with Tukey–Kramer post-hoc test). For non-parametric values, the Kruskal–Wallis test was used. The accepted level of significance was always P < 0.05.

Results

Time and Concentration Dependence of Kainate Excitotoxicity on Organotypic Cultures

Our previous data demonstrated that, in the rat isolated spinal cord, neurons showed PAR-immunopositivity and pyknotic nuclei 24 h after a large concentration (1 mM) of kainate (Kuzhandaivel et al. 2010b). Thus, we first investigated whether kainate induced strong expression of PAR also in organotypic slices and what timecourse this phenomenon might follow. Figure 1a shows an example of enhanced PAR positivity at the washout after 1 h of kainate treatment, a phenomenon that became more intense 6 h later. Figure 1b plots the average increment in the number of PAR positive elements following 0.1 mM kainate application and subsequent washout: the largest value was detected at 4 h and then became very low 24 h later when extensive neuronal degeneration is known to occur with appearance of condensed nuclear chromatin (pyknosis; Mazzone et al. 2010) and unavoidable loss of PAR signal.
https://static-content.springer.com/image/art%3A10.1007%2Fs10571-010-9640-7/MediaObjects/10571_2010_9640_Fig1_HTML.gif
Fig. 1

Time-dependence of PAR immunofluorescence in organotypic slices after treatment with kainate. a Examples of PAR staining in control condition (left) and after 1 or 6 h after washing out kainate (0.1 mM for 1 h). Immunopositivity signals are almost absent in control condition. The inset to the middle panel shows an example of the PAR immunoreactivity detected as a cytoplasmic rim around the nucleus. All calibration bars = 75 μm. b Histograms showing the increase (with respect to control) in the number of PAR positive nuclei at various times after washing out kainate. The data are from three experiments with six slices; *P < 0.05, **P < 0.01 vs. control. c Plots of percentage of cells (in the central region of the slice) with condensed chromatin nucleus (pyknosis; left; filled squares) or percent of NeuN positive cells (neurons; right; filled diamonds) for various concentrations of kainate (log scale) in the range from 1 μM to 1 mM (applied for 1 h). Data were collected after 24 h washout from at least three different experiments, n = 4–12; **P < 0.01, ***P < 0.001 vs. control

When PARP-1 over-activation occurs, the PAR polymer is synthesized in the nucleus and released into the cytoplasm (Yu et al. 2006). In line with this notion, the inset to Fig. 1a (middle) shows that the PAR polymer was present as a thin cytoplasmatic rim around the nucleus (identified by DAPI nuclear staining). It was, however, difficult to routinely use PAR immunoreactivity to assess PARP-1-mediated neuronal damage because this antibody cross-reacts with the neuronal marker NeuN, and because it would be uncertain how to select a threshold of PAR signal intensity for deciding irreversible neuronal loss. Hence, for sake of simplicity, we compared the emergence of pyknosis with the loss of cells positive to NeuN, a nuclear biomarker of neurons. The plot of Fig. 1c shows that, for the central region which is endowed with rhythmic activity (Sibilla and Ballerini 2009), the concentration/effect curve for kainate-induced pyknosis (measured 24 h after 1 h application) was very similar to the curve indicating the simultaneous loss of neurons. Analogous data were observed for the ventral and dorsal regions (not shown). Thus, most pyknotic cells were neurons and kainate concentrations as low as 1 μM could already induce significant damage that reached its maximum at 0.1 mM.

The Timing of Application PJ 34 Determined its Protective Action Against Kainate

In order to block the activity of PARP-1 induced by kainate, we treated the cultures with the PARP-1 selective inhibitor PJ 34 (30 μM) for 24 h starting its application either together with kainate or at kainate washout. Figure 2a shows examples of the DAPI nuclear staining in the ventral zone following either protocol of PJ 34 administration that produced comparable outcome in terms of cell protection. Figure 2b shows that the excitotoxic effect of 10 μM kainate was significantly counteracted by PJ 34 (30 μM) with the only exception of the dorsal region when this PARP-1 inhibitor was given upon kainate washout. Conversely, when a much larger dose (0.1 mM) of kainate was used (Fig. 2c), PJ 34 induced a clearly limited protection of spinal cultures that was observed, just in the ventral and central regions, only upon its co-application with kainate.
https://static-content.springer.com/image/art%3A10.1007%2Fs10571-010-9640-7/MediaObjects/10571_2010_9640_Fig2_HTML.gif
Fig. 2

The PARP-1 inhibitor PJ 34 could counteract the toxic effect of kainate. a Example of the ventral region with cells manifesting condensed chromatin 24 h after 1 h application of kainate (10 μM). Fewer pyknotic nuclei were apparent when PJ 34 was co-applied with kainate or applied at the washout time. b Histograms showing average percent of pyknotic cells in the three regions of interests after 10 μM kainate alone, or kainate together with PJ 34, or kainate plus PJ 34 applied 1 h later at washout. c Histograms showing average percent of pyknotic cells in the three regions of interests after 0.1 mM kainate alone, or kainate together with PJ 34, or kainate plus PJ 34 applied 1 h later at washout. Average data are from three experiments in which 7–12 slices were used. #P < 0.05, ##P < 0.01, ###P < 0.001 vs. kainate treatment

These results suggested that a strong excitotoxic stimulus could not be efficiently suppressed by PJ 34, especially when its administration was delayed. This observation prompted us to investigate, in more detail, the cell types which could be saved by PJ 34 following a milder excitotoxic stimulus (10 μM).

Neuronal and Glial Sensitivity to Kainate and PJ 34 Application

Figure 3a, b shows the topography and degree of neuronal damage elicited by 10 μM kainate plus PJ 34. Whenever PJ 34 was applied on kainate washout (see light grey columns), neuronal numbers (expressed as absolute values of NeuN positive elements) were similar to those observed after kainate alone and no significant neuroprotection could be detected. Conversely, when PJ 34 was given together with kainate (darker grey columns), it could significantly improve the number of surviving neurons in all three regions versus the data obtained with kainate alone (values of NeuN positive cells were incremented by 48%, 50% and 39% for the ventral, central and dorsal regions, respectively).
https://static-content.springer.com/image/art%3A10.1007%2Fs10571-010-9640-7/MediaObjects/10571_2010_9640_Fig3_HTML.gif
Fig. 3

Neuronal loss evoked by kainate was inhibited by PJ 34 co-treatment. a Example of how neuronal loss (in the central region) evoked by kainate (10 μM; 1 h) was counteracted by co-application of PJ 34 (30 μM). Images were collected after 24 h washout. b Histograms showing the average number of NeuN positive cells in the three regions analyzed. Each bar represents mean data from three experiments in which 6–11 slices were used. #P < 0.05 vs. kainate treatment

In particular, taking as representative example the data (Fig. 3b) from the central region (that is also depicted in Fig. 1c for controls) shows that co-application of kainate (10 μM) with PJ 34 allowed detection of 163 ± 20 NeuN positive elements that correspond to 83% of the control value without any treatment (193 ± 30). This number of surviving cells is close to the number of non-pyknotic cells (83%) obtained after subtracting the DAPI-positive pyknosis values (17 ± 7%) from total number of cells observable in the same region after the same pharmacological treatment (Fig. 2b). These values, therefore, show strong correspondence between DAPI and NeuN results after co-application of PJ 34 and kainate (10 μM), confirming that this protocol could produce adequate protection against kainate and that this phenomenon was primarily directed to neurons.

Figure 4a, b indicates that, even at the concentration of 0.1 mM, kainate did not induce significant damage to S100-positive cells that are typically protoplasmic astrocytes found in the grey matter of the spinal cord (Donato 2003; Kuzhandaivel et al. 2010a). Likewise, application of PJ 34 on kainate washout evoked no change in S100 immunoreactive cells in all three regions examined (Fig. 4a, b).
https://static-content.springer.com/image/art%3A10.1007%2Fs10571-010-9640-7/MediaObjects/10571_2010_9640_Fig4_HTML.gif
Fig. 4

Characterization of kainate effects on S100 positive astrocytes with or without application of PJ 34. a Examples of S100 immunopositive astrocytes in the ventral region after kainate (0.1 mM; 1 h; middle) and 24 h washout with PJ 34 (right). Note good expression of S100 immunoreactivity in the three panels. b On the same preparations shown in Fig. 3, S100 immunostaining was quantified (with densitometry analysis of a 500 × 500 μm area) to provide mean data for at least three experiments in which 4–12 slices were used. Note no significant change versus control

Discussion

The principal findings of the present report are the novel demonstration that the PARP-1 inhibitor PJ 34 could significantly protect spinal neurons in an in vitro model of kainate-evoked excitotoxicity, and that such a neuroprotection was detected only when PJ 34 was co-applied with kainate. These new data provide the interesting information that PARP-1-mediated neuronal death (‘parthanatos’) was actually a relatively rapid process of neurotoxicity even when the excitotoxic stimulus was far from being maximally effective.

Topography and Dose Dependence of Kainate-Mediated Excitotoxicity

Even if organotypic slice cultures are known to retain the cytoarchitecture and rhythmic electrical activity of the brain or spinal regions from which they originate (Spenger et al. 1991; Streit et al. 1991; Streit 1993; Sibilla and Ballerini 2009), the present results showed that they were strongly damaged by kainate concentrations below those normally tolerated by the isolated spinal cord (Mazzone et al. 2010). While the mechanism of heightened sensitivity of these cultures remains a subject for future investigation, it is clear that cell death was chiefly affecting neurons rather than glia via a process that involved hyperproduction of PAR, and was therefore considered to be expression of parthanatos. Apart from the enhanced vulnerability of such cultures to excitotoxicity, other features of this process (rate, extent of neuronal losses, sparing of glial cells) were essentially the same as those recently reported for the rat isolated spinal cord (Kuzhandaivel et al. 2010a, b).

Kainate is a broad spectrum glutamate receptor agonist (Traynelis et al. 2010) that evokes multiple effects due to activation of specific kainate receptors (at pre and postsynaptic level), activation of AMPA receptors (with weak desensitization) and, indirectly, to the depolarization-dependent release of endogenous neurotransmitters: all these factors contribute to confer regional specificity of cell damage, for example to organotypic hippocampal cultures (Zimmer et al. 2000; Kristensen et al. 2001). It seems likely that any large release of endogenous glutamate caused by the kainate-induced depolarization (Mazzone et al. 2010; Taccola et al. 2008) would find a large number of effector systems to amplify and widen the direct action of kainate itself. For these reasons, the consequences of the action of kainate on spinal cultures are complex. In the present study, the dorsal region was the most sensitive to kainate damage, a result closely similar to the one observed with the rat isolated spinal cord (Mazzone et al. 2010). In addition, dorsal afferent fibres of the neonatal rat spinal cord express kainate receptors that mediate large depolarizations and, thus, strong release of glutamate within the dorsal horn area (Agrawal and Evans 1986). In keeping with this notion, previous studies have also shown extensive cell loss not only from the ventral horn but also from the dorsal horn area of organotypic spinal slices treated with 50 μM kainate (Calderó et al. 2010) or by block of the glutamate uptake systems (Corse et al. 1999; Maragakis et al. 2005). Thus, at least from the point of view of glutamatergic network development, the organotypic slices were similar to postnatal spinal networks (Sibilla and Ballerini 2009).

In the rat spinal cord, kainate receptors are strongly expressed in the most superficial laminae of the dorsal horn, where AMPA and NMDA receptor subunits are also abundantly found (Tölle et al. 1993). Spinal neurons of the dorsal horn area endowed with limited Ca2+ buffering capacity (because of different expression of Ca2+ binding proteins; Sibilla et al. 2009) might be very sensitive to kainate-dependent damage.

Neuroprotection by PJ 34

PJ 34 is regarded as a selective blocker of PARP-1-mediated neuronal death (Abdelkarim et al. 2001). We previously used a less selective antagonist of PARP-1, namely PHE, and observed analogous results on the rat isolated spinal cord (Kuzhandaivel et al. 2010b). Since PJ 34 is reported to be also an inhibitor of metalloproteinases (Nicolescu et al. 2009), we cannot exclude the possibility that this effect contributed to the current data. However, metalloproteinases comprise a large family with contrasting roles during the post-traumatic phase of spinal cord injury (Agrawal et al. 2008), making it difficult to identify if anyone of them might have been involved in the neuroprotective action exerted by PJ 34. Future studies with other PARP-1 blockers may shed further light on this issue.

The present report showed that excitotoxicity of spinal slice cultures was associated with PAR generation and pyknosis, thus displaying canonical signs of PARP-1 dependent cell death. However, this deleterious phenomenon developed rapidly and could be efficiently inhibited by PJ 34 only when a number of conditions were met: (1) the spinal region examined was not the dorsal one; (2) PJ 34 was co-applied with kainate; (3) the concentration of kainate did not elicit a maximum deadly effect. Previous studies have shown that the concentration of PJ 34 employed in the present report are largely in excess to those necessary for full block of PARP-1 hyperactivity (Virag and Szabo 2002). Hence, it seems unlikely that the limited ability by PJ 34 to neuroprotect spinal cultures was due to partial inhibition of this important enzymatic activity. It is more probable that, in addition (or alongside) to PARP-1 hyper-activation, other cell death mechanisms (for instance PARP-2; Moroni 2008; Moroni et al. 2009) had been triggered by kainate. This suggestion obviously makes more complex any attempt to obtain substantial neuroprotection of spinal networks even in vitro.

Parthanatos: An Obstinate Process of Spinal Network Damage

In analogy with brain stroke, it is important to treat spinal lesions as soon as possible to maximize the chances of recovery (Park et al. 2004; Rowland et al. 2008). Nonetheless, this strategy can be successful only with drugs targeted to suppress well-identified pathological processes. As excitotoxicity is a major, rapid-onset component of this phenomenon (Doble 1999; Park et al. 2004), it would be desirable to explore the usefulness of agents blocking certain lethal biochemical derangements activated by it. Our data are, of course, limited to a simple in vitro model, yet they suggest that the time window to combat excitotoxicity even at a rather downstream step like PARP-1 hyperactivity is rather short. Future electrophysiological studies with the rat isolated spinal cord will be necessary to find out if the number of surviving neurons after application of PJ 34 and kainate is compatible with the minimal network membership required for the production of locomotor network activity (Nistri et al. 2010).

Acknowledgments

We thank to Dr. Miranda Mladinic for her helpful advice throughout the project, Dr. Beatrice Pastore for her assistance with organotypic cultures, and Dr. Micaela Grandolfo for her microscopy support. We are most grateful to Dr. Walter Vanzella (Glance Vision Technologies, Trieste) for his software support for image analysis. This study was supported by grants from the Friuli Venezia Giulia government and from the Italian Ministry of Education and Research (MIUR) with their PRIN program.

Copyright information

© Springer Science+Business Media, LLC 2010