Polyamidoamine dendrimer impairs mitochondrial oxidation in brain tissue

Background The potential nanocarrier polyamidoamine (PAMAM) generation 5 (G5-NH2) dendrimer has been shown to evoke lasting neuronal depolarization and cell death in a concentration-dependent manner. In this study we explored the early progression of G5-NH2 action in brain tissue on neuronal and astroglial cells. Results In order to describe early mechanisms of G5-NH2 dendrimer action in brain tissue we assessed G5-NH2 trafficking, free intracellular Ca2+ and mitochondrial membrane potential (ΨMITO) changes in the rat hippocampal slice by microfluorimetry. With the help of fluorescent dye conjugated G5-NH2, we observed predominant appearance of the dendrimer in the plasma membrane of pyramidal neurons and glial cells within 30 min. Under this condition, G5-NH2 evoked robust intracellular Ca2+ enhancements and ΨMITO depolarization both in pyramidal neurons and astroglial cells. Intracellular Ca2+ enhancements clearly preceded ΨMITO depolarization in astroglial cells. Comparing activation dynamics, neurons and glia showed prevalence of lasting and transient ΨMITO depolarization, respectively. Transient as opposed to lasting ΨMITO changes to short-term G5-NH2 application suggested better survival of astroglia, as observed in the CA3 stratum radiatum area. We also showed that direct effect of G5-NH2 on astroglial ΨMITO was significantly enhanced by neuron-astroglia interaction, subsequent to G5-NH2 evoked neuronal activation. Conclusion These findings indicate that the interaction of the PAMAM dendrimer with the plasma membrane leads to robust activation of neurons and astroglial cells, leading to mitochondrial depolarization. Distinguishable dynamics of mitochondrial depolarization in neurons and astroglia suggest that the enhanced mitochondrial depolarization followed by impaired oxidative metabolism of neurons may be the primary basis of neurotoxicity.


Background
Polyamidoamine (PAMAM) dendrimers are hyperbranched "protein-like" polymers with well-defined globular structure and monodispersed, nanoscopic particle size. PAMAM dendrimers have been reported to be able to cross the blood-brain barrier and are used as nanoparticle delivery systems to carry DNA, drugs or imaging agents to the brain [1][2][3]. Despite its wide application in the brain, only general toxic effects of dendrimers have been studied [4][5][6], much less information is available about their effects on neural cells.
In our recent study we showed that application of polycationic PAMAM generation 5 (G5-NH 2 ) dendrimers induced severe depolarization and subsequent inactivation of hippocampal pyramidal neurons in brain slices. Additionally, cell death after G5-NH 2 application was also observed in a concentration dependent manner [7]. In the present study we characterize the early intracellular processes sequential to G5-NH 2 induced neuronal depolarization and explore the effect of dendrimer application on astroglial cells.
In addition to their role in maintaining neuronal function, astroglial cells have been disclosed as active contributors in signal processing [8,9]. Furthermore, we also reported on astroglial signaling independent of neuronal activity in acute brain slices isolated from the rat nucleus accumbens [10]. In general, neurons are more susceptible to oxidative injury than astrocytes, due to their limited antioxidant capacity [11]. During oxidative stress astrocytes support neuronal function by providing antioxidant protection [11,12]. Therefore damage resulting in astrocyte dysfunction leads to increased neuronal death [11]. Oxidative damage in neural tissue can be detected by the loss of mitochondrial membrane potential (Ψ MITO ), a marker of mitochondrial dysfunction [11,13] that is sensitively coupled to neuronal and astroglial activation and survival [11,14,15]. Ψ MITO is also coupled to intracellular Ca 2+ regulation [11,13,14,16]. Changes in intracellular Ca 2+ level indicate activation of neuronal [17,18] or astroglial cells [10,19,20].
Membrane-dendrimer interactions have been studied extensively, supporting the view that cationic dendrimers interact with biological membranes [4,21,22]. Interactions are often followed by cellular internalization of cationic PAMAM dendrimers [23].
Here we report that application of G5-NH 2 dendrimer induces robust intracellular Ca 2+ signals in both neuronal and astroglial cells followed by severe Ψ MITO depolarization indicating the disruption of neuronal oxidative metabolism.

Results and discussion
Plasma membrane appearance of fluorescently labeled G5-NH 2 in neuronal and astroglial cells To determine the localization of G5-NH 2 in the rat hippocampal slices we covalently conjugated the fluorescent Rhodamine Green dye to G5-NH 2 and studied the localization of the fluorescently labeled dendrimer with confocal microscopy after 30 minutes of incubation. We found robust plasma membrane appearance of 0.1 mg/ ml G5-NH 2 in the pyramidal layer of acute hippocampal slices ( Figure 1A, n = 3 slices). Fluorescent G5-NH 2 appeared predominantly on the cell membrane of pyramidal neurons, although weak internalization of the dendrimer was also observed ( Figure 1A). On astroglial cells a patchy membrane distribution of the green fluorescent G5-NH 2 was detected as colocalization with the astroglia-specific red fluorescent marker sulforhodamine-101, SR101, Figure 1B, n = 5 slices), suggesting direct interaction between the astroglial cell membrane and the dendrimer ( Figure 1B).

Intracellular Ca 2+ responses of astroglial cells and neurons
To investigate whether interaction of G5-NH 2 with the plasma membrane of neurons and astroglia affects their function, we monitored intracellular Ca 2+ signals that sensitively reflect the activity of both cell types. Astroglial Ca 2+ signals were monitored in the astroglia-rich stratum radiatum area in the hippocampal CA3 region after bulk loading of the rat hippocampal slice with the Ca 2+ sensitive fluorescent dye Fluo-4 [10,24] ( To quantify the effect of G5-NH 2 on the astroglial Ca 2+ dynamics we determined the average number of responding cells per slice and the average frequency of Ca 2+ transients (calculated from the intervals measured between subsequent peaks). Both parameters increased significantly during the 30 minute application period ( Figure 2D). Number of responding cells and frequency of transients fell back to the control level during the washout period ( Figure 2D) indicating reversible effect of G5-NH 2 on astrocytes. In some cells, however, intracellular Ca 2+ level remained slightly elevated ( Figure 2C, Left, red and orange lines). In contrast, intracellular Ca 2+ enhancements in pyramidal neurons were characterized by quickly developing, lasting increase in dendritic processes and almost linear increase of intrasomal Ca 2+ level after 0.1 mg/ml G5-NH 2 application ( Figure 2C, Right).

Distinguishable Ψ MITO depolarization in neurons and astroglial cells
The observed transient or lasting enhancements of intracellular Ca 2+ level after dendrimer application may result in impaired oxidative metabolism through Ca 2+ influx-induced depolarization of Ψ MITO [13,14,16]. Mitochondrial cell death pathways have been suggested to contribute to the cytotoxic character of cationic PAMAM dendrimers in human lung cells [25]. In the brain, the neuronal activity and the mitochondrial function are highly correlated [15]. In addition, neuronal function and survival are very sensitive to mitochondrial dysfunction which can be monitored by measurement of the mitochondrial membrane potential [13,14,26]. To further explore this issue, we studied Ψ MITO changes in astroglia and neurons using the Ψ MITO sensitive dye rhodamine-123 [14].
Application of G5-NH 2 (0.1 mg/ml, 30 min) significantly increased the fluorescence of the Ψ MITO sensitive dye rhodamine-123 in both pyramidal neurons and astroglia ( Figure 3A-C, astroglial cells: n = 6 slices, neurons: n = 4 slices), indicating Ψ MITO depolarization and impaired oxidative metabolism in both cell types. The dynamics of neuronal and astroglial response, however, showed distinctive features. Similarly to the Ca 2+ responses, Ψ MITO increase was found to be transient in most astroglial cells ( Figure 3C and D) Responses were considered to be transient if the fluorescence intensity returned to ±20% of the baseline value within the application of G5-NH 2 . In contrast to astroglia, Ψ MITO remained elevated in the majority of neurons until the end of the experiment ( Figure 3C and D, lasting response), suggesting irreversible Ψ MITO depolarization. In addition, the duration of the response was shorter ( Figure 3D). Since astroglia is morphologically similar to interneurons, we completed colocalization experiments to identify astroglial cells. The observed colocalization of the astroglia specific fluorescent dye SR101 with Ψ MITO depolarization monitored by the red fluorescent rhodamine-123 confirmed that, indeed, astroglial cells were probed (c.f. yellow color in the merged image Figure 3A).
To quantitatively compare the onset dynamics of Ca 2+ enhancements and Ψ MITO depolarization we determined the temporal distribution of Ca 2+ transients (1783 Ca 2+ transients were identified in 303 cells in 6 slices) and Ψ MITO responses (122 Ψ MITO peaks were measured in 144 cells in 7 slices) in all responding astroglialcells after G5-NH 2 application (Figure 4). The appearance of intracellular Ca 2+ transients was immediate and remarkably preceded Ψ MITO depolarization (Figure 4) despite of the fact that the average number of responding astroglial cells per slice showing Ca 2+ enhancements and Ψ MITO depolarization was not significantly different (39 ± 7 vs. 24 ± 5, respectively; p = 0.117, one-way Anova). These data suggest a causal link between the two processes. It is to note that dynamics of intracellular Ca 2+ transients and Ψ MITO depolarization in pyramidal neurons has been shown to be coupled during seizure-like events [14].

PAMAM dendrimer evokes astroglial Ψ MITO depolarization directly and via neuron-astroglia interaction
Since neuronal activation results in the release of major excitatory and inhibitory neurotransmitters Glu and γ-aminobutyric acid (GABA), respectively, and glutamatergic activation can lead to Ψ MITO changes [13,14]), we explored whether neuronal activation modifies astroglial responses. To examine whether G5-NH 2 directly affects astroglial mitochondrial function or it is the consequence of the preceding neuronal depolarization, we measured G5-NH 2 evoked Ψ MITO depolarization in the presence of the following inhibitors: blocker of voltage-gated Na + channels tetrodotoxin (TTX, 1 μM), antagonists of Glu receptors (N-methyl-D-aspartate type: DL-2amino-5-phosphonopentanoic acid APV, 100 μM; AMPA/ kainate type: 6-cyano-7-nitroquinoxaline-2,3-dion CNQX, 10 μM) and the GABA A receptor antagonist picrotoxin (100 μM). In the presence of the antagonists, the number of astrocytes showing Ψ MITO depolarization did not change, while the number of responding neurons significantly decreased ( Figure 5A, astroglia n = 7 slices, neurons n = 3 slices). However, the blockade of neuronal activity decreased both the duration of the astroglial response (10.2 ± 0.7 min vs. 7.8 ± 0.8 min; p = 0.049, one-way Anova) and the percentage of lasting astroglial (but not the neuronal) Ψ MITO depolarization ( Figure 5B). The average intensity of ΔF/F 0 changes in neurons and astrocytes were also significantly decreased ( Figure 5C).
Neurons and astroglial cells are functionally interconnected within the brain. Increased neuronal activation could led to astroglial Ψ MITO depolarization [13,14]. If astroglial Ψ MITO depolarization found in our experiments is only the consequence of the G5-NH 2 -evoked neuronal activation then inhibition of neuronal activity should prevent Ψ MITO depolarization in astroglia. Therefore the unchanged number of responding glial cells ( Figure 5A) indicates that G5-NH 2 directly evoked mitochondrial depolarization in astroglia, while the decreased duration ( Figure 5B) and intensity ( Figure 5C) in astroglial cells suggests that neuronal activation by G5-NH 2 intensified the astroglial responses.
Astrocytes are more resistant to PAMAM dendrimer neurotoxicity than neurons Lasting Ψ MITO depolarization of neuronal and some astroglial cells might indicate irreversible disturbances of cellular metabolism [13][14][15]27]. Predominantly shorter astroglial responses, however, suggest that G5-NH 2 application might be less harmful to astrocytes probably because astroglial Ψ MITO can be recovered after several minutes of depolarization [26]. To assess the consequence of G5-NH 2 induced Ψ MITO depolarization we measured the viability of astrocytes and neurons by labeling the live cells with calcein after 30 min exposure to G5-NH 2 . The SR101 positive astroglial cells showed robust calcein fluorescence in the stratum radiatum after 30 min of G5-NH 2 application indicating the presence of functional, viable astroglial cells [28] (Figure 6A Top, n = 3 slices), although viability of astrocytes in the stratum lucidum region may also be compromised. Contrary, in accordance with our previous observations [13], a large proportion of hippocampal pyramidal neurons lost their viability after 30 min application of G5-NH 2 despite the survival of astrocytes in the same region ( Figure 6 Middle and Bottom). These findings are in accordance with the neuronal activity-dependent lasting Ψ MITO (c.f. Figure 4) and plasma membrane [7] depolarization. Transient as opposed to lasting Ψ MITO depolarization in astroglia and neurons, respectively, indicates that the neurotoxicity [7] of G5-NH 2 may predominantly be restricted to neurons over astroglia.

Conclusion
G5-NH 2 activates both astrocytes and neurons in acute hippocampal slices as reflected by intracellular Ca 2+ enhancement and Ψ MITO depolarization. We showed that the interaction of PAMAM dendrimer with the plasmamembrane evokes Ψ MITO depolarization most probably via the enhancement of intracellular Ca 2+ level. Vast majority of astrocytes shows transient response and remains viable. In contrast, lasting activation of neurons by G5-NH 2 provokes fatal consequences in accordance with the predominantly irreversible early depolarization of neurons [7]. Due to the connection between elevated Ca 2+ signal and Ψ MITO depolarization, as well as formation of reactive oxygen species [11,13,16,27], we can also infer the early disturbance of oxidative metabolism as the primary cause of PAMAM dendrimer evoked neuronal toxicity.

Imaging
In order to monitor changes in intracellular Ca 2+ , rat brain hippocampal slices were incubated with 5 μM Fluo-4 AM in ACSF for one hour at 35°C in the dark under humidified carbogen atmosphere after preincubation in 2% pluronic acid containing ACSF for 2 minutes [10]. To allow the cleavage of the AM ester group of Fluo-4, slices were transferred to dye-free ACSF at least 30 minutes before the start of the experiment. In order to monitor changes in Ψ MITO rat brain hippocampal slices were loaded with the fluorescent Ψ MITO indicator rhodamine-123 (15 μg/ml in ACSF) for 20 minutes at 25°C [14]. To identify astrocytes slices were loaded with sulforodamine-101 immediately after slicing (1 μM, 20 minutes, 35°C, [29]) before rhodamine-123 loading.
Astroglial cell viability was measured using Calcein-AM fluorescent dye (λ ex = 488 nm, λ em = 510-530 nm). The intracellular esterase activity could be used as a probe of viability and plasma membrane competence and as an indicator of the cellular functionality [28]. Calcein-AM is a membrane-permeable nonfluorescent molecule that enters intact living cells, then it is cleaved by endogenous esterases to produce the highly fluorescent, membrane impermeable molecule, calcein.

Data evaluation
Images recorded by the FluoView300 software were processed using the free ImageJ 1.41 image analysis software (http://rsbweb.nih.gov/ij/). Matlab 6.1 was used to evaluate fluorescence changes and the number of responding cells and frequency of fluorescent transients. To avoid differences between slices G5-NH 2 evoked changes in fluorescence intensity (ΔF/F 0 ) were normalized to the average response of the cells to 10 μM CCCP, a well known mitochondrial inhibitor applied at the end of the experiments (ΔF/F 0 after CCCP application was 2.2 ± 0.4 for astroglial cells and 1.5 ± 0.16 for neurons). Data presented are mean ± S.E.M. Statistical analysis was performed using one-way Anova (OriginLab Co., Northampton, UK) and p < 0.05 was considered statistically significant.

Synthesis of fluorescently labeled G5-NH 2
Rhodamine Green was covalently bound to G5-NH 2 by reacting aqueous solution of G5-NH 2 (1400 μl, 1.97 μmol, 4.05 w/w%) with Rhodamine Green carboxylic acid succinimidyl ester hydrochloride mixed isomers (5(6)-CR 110, SE; Molecular Probes, Eugene, OR, USA) (1 mg, 1.97 μmol) dissolved in N,N-dimethylformamide (100 μl) in 0.1 M NaHCO 3 buffer (1.4 ml, pH 8.5) at room temperature for 2 h in dark. The unreacted dye was removed from the solution by ultrafiltration (3 000 MWCO) in Amicon Ultracel -3K centrifugal filter units. Amine reactive form of the Rhodamine Green dye was coupled covalently to the PAMAM dendrimer forming amide bonds. The unreacted dye was then removed by ultrafiltration and the conjugate containing no dye was used throughout the experiments. The conjugate is hydrolitically stable under physiological conditions therefore the localization of the dendrimer can be interpreted by the detected fluorescence.