Chemically Functionalized Water-Soluble Single-Walled Carbon Nanotubes Obstruct Vesicular/Plasmalemmal Recycling in Astrocytes Down-Stream of Calcium Ions

We used single-walled carbon nanotubes chemically functionalized with polyethylene glycol (SWCNT-PEG) to assess the effects of this nanomaterial on astrocytic endocytosis and exocytosis. We observed that the SWCNT-PEG do not affect the adenosine triphosphate (ATP)-evoked Ca2+ elevations in astrocytes but significantly reduce the Ca2+-dependent glutamate release. There was a significant decrease in the endocytic load of the recycling dye during constitutive and ATP-evoked recycling. Furthermore, SWCNT-PEG hampered ATP-evoked exocytotic release of the loaded recycling dye. Thus, by functionally obstructing evoked vesicular recycling, SWCNT-PEG reduced glutamate release from astrocytes via regulated exocytosis. These effects implicate SWCNT-PEG as a modulator of Ca2+-dependent exocytosis in astrocytes downstream of Ca2+, likely at the level of vesicle fusion with/pinching off the plasma membrane.


Introduction
Single-walled carbon nanotubes (SWCNTs) have been considered candidates for applications in biotechnology and biomedicine [1,2], and in particular for neural applications [3,4]. When chemically functionalized with polyethylene glycol (PEG), SWCNTs render water solubility [5]. These colloidal solutes, SWCNT-PEG, have been used to modulate the morpho-functional properties of two main neural cell types, neurons [6] and astrocytes [7], in culture. SWCNT-PEG caused an increase in the neurite outgrowth of selected neurites in vitro and affected the neuronal Ca 2+ dynamics by reducing the depolarization-dependent influx of Ca 2+ from the extracellular space [6]. Furthermore, in neurons exposed to SWCNT-PEG, the evoked exocytotic incorporation of vesicles into the plasma membrane was not sufficiently balanced by endocytotic retrieval, a possible mechanism underlying an increase in the neurite outgrowth [8]. In vivo application of the SWCNT-PEG solute resulted in an increase in axonal regeneration and a modest functional locomotor recovery without altering reactive astrogliosis in an acute spinal cord injury rat model [9]. A detailed analysis of astrocytes in cell culture showed an Institutional Animal Care and Use Committee (Animal project number IACUC-09230 approved on 08/19/2019). Astrocytes isolated from the visual cortices of 0−2-day-old C57BL/6 mice were purified, maintained in cell culture and plated onto polyethyleneimine (PEI)-coated glass coverslips as we previously described in detail elsewhere [7,26]. SWCNT-PEG solute was synthetized and characterized as we previously described elsewhere [7]. The batch of SWCNT-PEG solute used in this study contained 72.3 weight percent (wt %) of the SWCNT backbone, 22.6 wt % of the functional group PEG (PEG 600; MW range 573-630) and 5.1 wt % of metal impurities (nickel and yttrium in~4:1 weight ratio) [11].
In the experiments using the functional group PEG alone, as a control for 5 µg/mL of SWCNT-PEG solute-treated group, PEG was added to the cells at 1 µg/mL, i.e., at the concentration corresponding to 20 wt % of the SWCNT-PEG solute.

Ca 2+ Dynamics
To study the ATP-induced Ca 2+ dynamics, time-lapse imaging was done on the astrocytes expressing RCaMP1h, 3 days post-transfection (total of 4 days in culture and SWCNT-PEG/PEG treatment where applicable), and its fluorescence was visualized and imaged using a standard tetramethylrhodamine isothiocyanate (TRITC) filter set. Images were acquired using a 60× Plan Apo objective (Nikon; numerical aperture, 1.4) and the microscope described above. After a cell of interest was identified based on its fluorescence, two subsequent time-lapse epochs were acquired to monitor the intracellular Ca 2+ levels. During the 1 st epoch (260 s, 1 frame/5 s), we replaced the external solution with external solution containing ATP (100 µM) after the 6 th frame. For a subset of the experiments, the external solution was supplemented with cyclopiazonic acid (CPA; 20 µM, Sigma) for the entire course of the experiment (Supplementary material and Figure S1). Following the completion of this epoch, the ATP-containing external solution was replaced with external solution containing 4-Bromo-A23187 (4-Br; Cells 2020, 9, 1597 4 of 18 20 µM, Molecular Probes) and the 2 nd epoch was acquired (150 s, 1 frame/15 s). The background (area of the coverslip containing no cells) subtracted RCaMP1h fluorescence intensity was expressed as dF/F 0 (%), where dF represents the change in fluorescence, while F 0 represents the baseline fluorescence of the cell before ATP stimulation (an average of the first 6 frames in the 1 st epoch). The external solution in both the epochs also contained 5 µg/mL SWCNT-PEG for all the cells treated with SWCNT-PEG.

Glutamate Release
To study the ATP-induced glutamate release and plasma membrane dynamics, time-lapse imaging was done on astrocytes transfected with plasmids encoding iGluSnFR or Lck 1-26 -EGFP (Supplementary material and Figure S2), respectively, 3 days post-transfection (total of 4 days in culture and SWCNT-PEG/PEG treatment where applicable), and their fluorescence was visualized and imaged using a standard fluorescein isothiocyanate (FITC) filter set. To monitor the extracellular glutamate levels, imaging was done as described above except that the external solution in the 2 nd epoch contained exogenous glutamate (100 µM) instead of 4-Br. For a subset of the experiments, the external solution in the 1 st epoch was replaced with external solution lacking ATP (Supplementary material and Figure S3).

Plasma Membrane Recycling
To assess plasma membrane recycling, cultured astrocytes were loaded with N-(3-triethylammoniumpropyl)-4-(6-(4-(diethylamino)phenyl)hexatrienyl)pyridinium dibromide (FM4-64; 10 µM, 5 min, Molecular Probes) without and with the addition of ATP (100 µM) for constitutive and ATP-stimulated recycling, respectively, 4 days post-plating. FM4-64 was visualized using a standard TRITC filter set and time-lapse imaging was done using a 20× Plan Fluor objective (Nikon; numerical aperture, 0.5) in two epochs. During the 1 st epoch the baseline fluorescence was acquired in external solution (50 s, 1 frame/10 s), which was replaced with external solution containing FM4-64 + ATP. After washing off the excess FM4-64, the 2 nd epoch was acquired in external solution (500 s, 1 frame/10 s). The data was reported as the background (area of the coverslip containing no cells) subtracted FM4-64 fluorescence intensity, dF. The baseline auto-fluorescence of astrocytes before the addition of FM4-64 was used as F 0 . To obtain the outlines of the cells, prior to FM4-64 loading, astrocytes were loaded with β-Ala-Lys-Nε-AMCA (20 µM at 37 • C for 2 h in cell culture medium; AMCA, 7-amino-4-methylcoumarin-3-acetic acid) and imaged using a standard 4 ,6-diamidino-2-phenylindole (DAPI) filter set. To study the ATP-stimulated exocytosis of FM4-64, the external solution in the 2 nd epoch was replaced with external solution containing ATP (100 µM) after the 50 th frame and the imaging was continued (300 s, 1 frame/10 s). The external solution in all the experiments also contained 5 µg/mL SWCNT-PEG or 1 µg/mL PEG for the cells with respective treatments.

Statistics
Statistical analysis was done using GraphPad Prism 8 Statistical Software (GraphPad Software, San Diego, CA) and SAS Software, version 9.4, of the SAS software for Windows (SAS Institute Inc., Cary, NC). The number of subjects required for individual set of experiments was estimated using power analysis (set at 80% and α = 0.05). The summary graphs are reported as means ± standard errors of means (SEMs) or medians with the interquartile range (IQR). Traces in Figures 1 and 2 are reported as means ± SEM, while the traces in Figures 3 and 4 are reported as medians without IQRs, for simplicity. All the experiments contain cells originating from at least three independent experimental runs/culture preparations. Student's t-test (pooled variances) was used for the experiments comparing the two independent groups conforming to normality based on Shapiro-Wilk test for normality (Figures 1 and 2, Figures S1 and S3). For data sets containing groups that deviated from normality, nonparametric statistics were used, with multiple independent groups analyzed using Kruskal−Wallis one-way ANOVA (KWA) followed by Dunn's test (Figures 3 and 4). The significance was established at p < 0.05. PEG and after 1 day in culture, transfected the cells with one of the two plasmids, RCaMP1h or iGluSnFR. Three days post-transfection and the continued treatment of a subset of cells with SWCNT-PEG (total of 4 days of SWCNT-PEG), the cells were visualized and imaged using a fluorescence microscope and standard TRITC (for RCaMP1h) or FITC (for iGuSnFR) filter sets. Time-lapse imaging was done to study astrocytic changes in intracellular Ca 2+ levels, [Ca 2+ ]i ( Figure 1A) and extracellular glutamate levels, [Glut]e ( Figure 2A).  reduction in the peak ( Figure 2E) and cumulative ( Figure 2F) normalized dF compared to the control cells implying that the SWCNT-PEG solute hampers the ATP-induced glutamate release from astrocytes without affecting Ca 2+ dynamics in astrocytes. We showed previously that the functional group PEG itself does not affect glutamate homeostasis while SWCNT-PEG increased the glutamate uptake in astrocytes ( Figure 1B of Reference [11]). Hence, the effect that we observed here is likely due to the SWCNT backbone and not the PEG functionalization group.  SWCNT-PEG cause a decrease in constitutive endocytosis which could represent an underlying mechanism for our previously reported increase in astrocytic size [7,10] and for an increase of glutamate uptake by astrocytes [11]; the latter mediated by an increase in the amount of Lglutamate/L-aspartate transporter (GLAST/ EAAT1) at the plasma membrane. These findings also imply an imbalance between constitutive exocytosis and endocytosis processes.   To study the amount and rate of exocytosis, we normalized the decay in FM4-64 fluorescence to the steady state level (t = 890 s) just prior to the application of ATP ( Figure 4B). Following ATP stimulation and as expected, we found that the control cells showed an expected exponential decrease (R 2 = 0.9044) in FM4-64 fluorescence as the ATP application caused exocytosis of the FM4-64 loaded in vesicles. Astrocytes show a very slow time course of regulated exocytosis [25,40,41]. Indeed, we calculated a decay time constant of 888 s based on the fitted equation. The SWCNT-PEG-treated cells, on the other hand, showed an unexpected linear decrease (R 2 = 0.9484) in FM4-64 fluorescence following ATP stimulation. Because of this lack of an exponential fit and to avoid possible errors related to the asymptotic portion of the FM4-64 curves, we assessed the time domain relevant to the rising phase of the ATP-evoked glutamate ( Figure 2C) and reported on the time to destain 10% of the FM4-64 load, i.e., t90. The t90 in the control astrocytes was 20 s, which was grossly extended (5 times)  Figure 3 (pulse), followed by second ATP (100 µM) application to stimulate the exocytosis of FM4-64 loaded in the vesicles (chase), the time of addition of which is indicated by the horizontal double-headed arrow. Other annotations as in Figure 3. Number of astrocytes studied in each condition is shown in parentheses and represents a fraction of the astrocytes already reported in Figure 3 but here stimulated again with ATP for the second time. The schematics show FM4-64 labeling and membrane recycling in astrocytes after dye application, at steady state after the dye washout, ATP-stimulated exocytosis and at steady state after ATP-stimulated dye washout, respectively (left to right), the time points of which are marked by the bold arrows. (B) FM4-64 fluorescence normalized to the steady-state level just prior to the second application of ATP. Traces in (A) and (B) show medians. Asterisks and pound signs indicate a statistical difference between the groups indicated on the right at the specific time points. KWA followed by Dunn's test; * p < 0.05, # p < 0.01.

Results
In the present study, we assessed the effect of SWCNT-PEG colloidal solute on the Ca 2+ dynamics and the consequential glutamate release from cortical astrocytes in response to ATP stimulation. We monitored the levels of intracellular Ca 2+ and extracellular glutamate using the genetically encoded intracellular Ca 2+ indicator, RCaMP1h [27] and the genetically encoded plasma membrane anchored extracellular glutamate sensor, iGluSnFR [29], respectively. We plated astrocytes onto polyethyleneimine (PEI)-coated glass coverslips in the absence and the presence of 5 µg/mL SWCNT-PEG and after 1 day in culture, transfected the cells with one of the two plasmids, RCaMP1h or iGluSnFR. Three days post-transfection and the continued treatment of a subset of cells with SWCNT-PEG (total of 4 days of SWCNT-PEG), the cells were visualized and imaged using a fluorescence Cells 2020, 9, 1597 9 of 18 microscope and standard TRITC (for RCaMP1h) or FITC (for iGuSnFR) filter sets. Time-lapse imaging was done to study astrocytic changes in intracellular Ca 2+ levels, [Ca 2+ ] i ( Figure 1A) and extracellular glutamate levels, [Glut] e (Figure 2A).

SWCNT-PEG does not Affect ATP-evoked Ca 2+ Dynamics in Astrocytes
Astrocytes were exposed to ATP (100 µM) to induce [Ca 2+ ] i elevations in these cells via the ER Ca 2+ stores (Supplementary material and Figure S1). This was followed by the application of the Ca 2+ ionophore 4-Bromo-A23187 (4-Br; 20 µM) that, upon incorporation into the plasma membrane, presents a conduit for Ca 2+ entry from the extracellular space to elicit a maximal [Ca 2+ ] i response [32,33] ( Figure 1A Figure 1C). We found that the SWCNT-PEG solute didn't cause any significant differences in the peak ( Figure 1D) or the cumulative ( Figure 1E) normalized dF compared to the control implying that the SWCNT-PEG solute doesn't affect the ATP-induced Ca 2+ dynamics in astrocytes.

SWCNT-PEG Hampers ATP-evoked Glutamate Release from Astrocytes
Next, we assessed the effect of SWCNT-PEG solute on the ATP-induced glutamate release from astrocytes. After baseline acquisition of iGluSnFR fluorescence, the cells were stimulated with ATP (bath application, 100 µM) and the resultant release of glutamate into the extracellular space was recorded as the change in iGluSnFR fluorescence (Figure 2A,B). Following ATP stimulation, the cells were also challenged with exogenously added glutamate (Glut; bath application, 100 µM) to obtain the maximum/saturation fluorescence of iGluSnFR [29] (Figure 2A,B). All the data were background subtracted and expressed as the percentage change (dF/F 0 ) in iGluSnFR fluorescence compared to the baseline iGluSnFR fluorescence before ATP stimulation (F 0 ; Figure 2B). We found that all the control cells (n = 13) studied showed a rapid transient decrease, a trough, in iGluSnFR fluorescence after the addition of ATP, indicative of a decrease in extracellular glutamate level ( Figure 2B), followed by a slower transient increase in fluorescence, indicative of an increase in extracellular glutamate level. While the trough represents a temporal dilution of glutamate in the extracellular milieu as a result of the bath exchange that took place when ATP was applied (Supplementary Material and Figures S2 and S3), the remaining signal represents a typical time course of evoked astrocytic glutamate release [24,25]. The cells treated with and imaged in presence of SWCNT-PEG (n = 12) also showed a similar time course in the change of iGluSnFR fluorescence ( Figure 2B). However, the magnitude of the increase in fluorescence was much lower than that in the control cells. Similar to the 4-Br evoked maximal [Ca 2+ ] i response, the saturated iGluSnFR fluorescence was also significantly different between the CNT-treated and untreated groups (Student's t-test, p < 0.01). Thus, we normalized the changes in the extracellular glutamate levels for each of the cells studied to their respective maximum iGluSnFR fluorescence ( Figure 2C). We found that the SWCNT-PEG solute did not cause any significant difference in the trough observed immediately upon ATP application ( Figure 2D), as expected from a perfusion artifact. Unexpectedly, however, SWCNT-PEG caused a significant reduction in the peak ( Figure 2E) and cumulative ( Figure 2F) normalized dF compared to the control cells implying that the SWCNT-PEG solute hampers the ATP-induced glutamate release from astrocytes without affecting Ca 2+ dynamics in astrocytes. We showed previously that the functional group PEG itself does not affect glutamate homeostasis while SWCNT-PEG increased the glutamate uptake in astrocytes ( Figure  1B of Reference [11]). Hence, the effect that we observed here is likely due to the SWCNT backbone and not the PEG functionalization group.
There is a strict relationship between cytosolic Ca 2+ increase and the amount of exocytotically released glutamate from astrocytes [32]. Under certain conditions, this relationship can be modulated down stream of Ca 2+ signal [17,24,34,35]. Thus, the disparity between the Ca 2+ and glutamate dynamics in the presence of SWCNT-PEG solute we observed in the present work (compare Figure 1C-E to Figures 2C and 2E,F, respectively) represents a negative modulation down stream of Ca 2+ signal. In part, this effect could be explained by the fact that SWCNT-PEG augmented glutamate uptake by astrocytes, as we reported elsewhere [11]. However, the reported 26% increase in glutamate uptake ( Figure 1B of Reference [11]) may only explain about half of the 53% diminution of glutamate release seen here ( Figure 2F). As SWCNT-PEG can tamper with vesicular recycling in neurons [8], it is possible that the negative modulation of the relationship between astrocytic cytosolic Ca 2+ and released glutamate could be due to an aberration in astrocytic vesicular trafficking and/or membrane recycling, which we explored next.

SWCNT-PEG Inhibits Constitutive and ATP-stimulated Membrane Recycling in Astrocytes
To assess if SWCNT-PEG solute interferes with plasma membrane-vesicular recycling in astrocytes, we used a recycling dye, FM4-64 [36,37]. Since FM4-64 does not passively diffuse across cell membranes, it is taken up by endocytosis. After 4 days in culture, astrocytes in the absence and the presence of 5 µg/mL SWCNT-PEG were preloaded with a dipeptide β-Ala-Lys conjugated to 7-amino-4-methylcoumarin-3-acetic acid (AMCA) (20 µM) [24]; this peptide is specifically taken up by the astrocytes via pepT2 peptide transporter [38] into their cytosol. We imaged dipeptide loaded astrocytes using a standard DAPI filter set to get the proper outlines of the cells ( Figure 3A). This outline aided cell autofluorescence assessment in the subsequent time-lapse imaging, which was done to study the changes in FM4-64 fluorescence using a fluorescence microscope and a standard TRITC filter set ( Figure 3A). After baseline acquisition to measure the autofluorescence level of astrocytes (used as F 0 after background subtraction) ( Figure 3A, t = 10 s and Figure 3B, left), the cells were exposed to FM4-64 (10 µM, 5 min) to monitor constitutive vesicular recycling which occurs in unstimulated cells. The dynamics in the FM4-64 fluorescence overtime was recorded and reported as the background subtracted FM4-64 fluorescence intensity, dF ( Figure 3C). The peak FM4-64 fluorescence measured 5 min after FM4-64 exposure reported on the extent of total FM4-64 cellular load (plasma membrane and vesicles) ( Figure 3A, t = 400 s and Figure 3B, middle), while the steady state FM4-64 fluorescence after extensive rinsing reported on the vesicular load since plasma membrane bound FM4-64 re-partitioned in the external solution and washed away ( Figure 3A, t = 850 s and Figure 3B, right). In unstimulated control cells, there was an expected exponential decrease (R 2 = 0.8774) in FM4-64 fluorescence with decay time constant, i.e., time to reach dF at maximum/exponent e (~37% maximum fluorescence), of 700 s. The unstimulated cells treated with SWCNT-PEG solute also showed an exponential decrease (R 2 = 0.9128) in FM4-64 fluorescence but substantially shorter decay time constant of 600 s, along with a significant decrease in the peak ( Figure 3D) and steady state dF ( Figure 3E) of FM4-64 fluorescence. When combined, these data imply that SWCNT-PEG cause a decrease in constitutive endocytosis which could represent an underlying mechanism for our previously reported increase in astrocytic size [7,10] and for an increase of glutamate uptake by astrocytes [11]; the latter mediated by an increase in the amount of L-glutamate/L-aspartate transporter (GLAST/ EAAT1) at the plasma membrane. These findings also imply an imbalance between constitutive exocytosis and endocytosis processes.
Since vesicular recycling in astrocytes can be evoked in a Ca 2+ -dependent manner, we studied the effect of SWCNT-PEG on ATP-stimulated vesicular recycling by supplementing the FM4-64 containing external solution with ATP (100 µM) and repeating the experiment described above. As expected, the control cells showed a significant increase in the peak and steady state FM4-64 fluorescence in the presence of ATP when compared to unstimulated astrocytes ( Figure 3D,E); there was an exponential decrease (R 2 = 0.8711) in FM4-64 fluorescence with a decay time constant of 620 s. The cells treated with SWCNT-PEG solute and stimulated with ATP also showed an exponential decrease (R 2 = 0.927) in FM4-64 fluorescence with a comparable decay time constant of 600 s; there was a significant increase only in the peak but not the steady state FM4-64 fluorescence when compared to unstimulated SWCNT-PEG-treated cells ( Figure 3D,E). In addition, the ATP-stimulated peak and steady state FM4-64 fluorescence were significantly lower in the cells treated with SWCNT-PEG solute compared to the corresponding control cells (Figure 3D,E).
We further addressed the possibility that the observed effects might be due to the PEG functional group by treating a subset of the cells with PEG (1 µg/mL; the concentration corresponding to 20 wt % of SWCNT-PEG) and repeating the experiments described above. We found that the unstimulated cells treated with PEG solute showed a significant decrease in the peak ( Figure 3D) and steady state ( Figure 3E) FM4-64 fluorescence at rest when compared to controls cells, but these values were significantly higher than those observed in the SWCNT-PEG-treated astrocytes. These findings imply that PEG itself could partially affect constitutive recycling in astrocytes. On the other hand, in the presence of ATP, PEG-treated astrocytes showed no significant differences in the peak ( Figure 3D) and steady state ( Figure 3E) FM4-64 fluorescence compared to the control stimulated astrocytes. These fluorescence values were also significantly higher than the values observed with SWCNT-PEG-treated cells implying that PEG does not affect the ATP-stimulated recycling in astrocytes. Both PEG-treated groups showed exponential decay of FM-64 fluorescence and comparable decay time constants to control groups (for unstimulated cells R 2 = 0.8343 and 700 s; for ATP-stimulated cells R 2 = 0.8764 and 620 s, respectively). Taken together, these results imply that the SWCNT-PEG solute affects both the constitutive and ATP-stimulated membrane recycling in astrocytes, with PEG playing a partial role in the constitutive membrane recycling.

SWCNT-PEG Obstructs ATP-evoked Exocytosis in Astrocytes
The total cellular and vesicular load of the recycling dye is the result of a complex interplay between exocytotic and endocytotic processes. To assess if SWCNT-PEG might have an additional effect (to that on endocytosis) on regulated exocytosis, we executed pulse-chase experiments, also referred to as "paired-pulse" experiments [8,39]. The above described loading of astrocytes with FM4-64, utilizing ATP as a stimulus, followed by extensive rinsing ( Figure 3C, and replotted in Figure 4A, t = 0-890 s) represent the pulse portion of the experiments. At the steady state level of FM4-64 fluorescence, we executed the "chase" component of the experiments, i.e., we stimulated the astrocytes a second time with external solution containing ATP (100 µM, 5 min) alone and lacking FM4-64 ( Figure 4A, t = 900-1190 s). This chase/second stimulus would lead to evoked exocytosis of FM4-64 previously loaded into the vesicles.
To study the amount and rate of exocytosis, we normalized the decay in FM4-64 fluorescence to the steady state level (t = 890 s) just prior to the application of ATP ( Figure 4B). Following ATP stimulation and as expected, we found that the control cells showed an expected exponential decrease (R 2 = 0.9044) in FM4-64 fluorescence as the ATP application caused exocytosis of the FM4-64 loaded in vesicles. Astrocytes show a very slow time course of regulated exocytosis [25,40,41]. Indeed, we calculated a decay time constant of 888 s based on the fitted equation. The SWCNT-PEG-treated cells, on the other hand, showed an unexpected linear decrease (R 2 = 0.9484) in FM4-64 fluorescence following ATP stimulation. Because of this lack of an exponential fit and to avoid possible errors related to the asymptotic portion of the FM4-64 curves, we assessed the time domain relevant to the rising phase of the ATP-evoked glutamate ( Figure 2C) and reported on the time to destain 10% of the FM4-64 load, i.e., t 90 . The t 90 in the control astrocytes was 20 s, which was grossly extended (5 times) to 100 s in the SWCNT-PEG group, implying a severe deficit in the evoked exocytosis of the loaded FM4-64 in the astrocytes treated with SWCNT-PEG solute. Interestingly, PEG had some effect on the destaining time course. The PEG-treated cells showed an exponential decrease (R 2 = 0.9045) in FM4-64 fluorescence following ATP stimulation with initial destaining being only~50% longer (t 90 = 30 s) than that of control ( Figure 4B), while the late portion of the curve resembled more that of SWCNT-PEG-treated astrocytes ( Figure 4B). Taken together, our results show a significant decrease in ATP-evoked (pulse) vesicular load in astrocytes treated with SWCNT-PEG, but not with PEG itself. The ATP-evoked exocytosis (chase) of the FM4-64 loaded in vesicles was obstructed by SWCNT-PEG with both of its components, CNT backbone and PEG, having partial effects on ATP-evoked exocytosis.

Discussion
The ATP-evoked elevations in astrocytic intracellular Ca 2+ were unaffected in the presence of SWCNT-PEG (Figure 1). This finding is seemingly at odds with our previous report showing that SWCNT-PEG reduced depolarization-dependent Ca 2+ influx via plasmalemmal voltage-gated channels from the extracellular space [6]. However, ATP-induced Ca 2+ dynamics in astrocytes mainly result from the activity of the smooth ER Ca 2+ store (Supplementary material, Figure S1) and TRPC1-containing channels [20]. This would imply that SWCNT-PEG might have some selective antagonistic effect on a subset of plasmalemmal Ca 2+ channels. Indeed, it has been reported that SWCNTs blocked a variety of phylogenetically distant voltage-gated K + channels, but not Clchannels implying selectivity to cation channels. Although it is tempting to speculate that electronic properties of SWCNTs might underlie this selectivity, the work on K + channels suggests that it is the shape (tube) and size (~1 nm in diameter) of these materials, but not electrochemical interactions, that governed the inhibition of ion channels [42]. Our water-soluble SWCNTs are of similar diameters [6][7][8]11,26] (also see below).
Unlike ATP-induced Ca 2+ dynamics that were preserved when astrocytes were exposed to SWCNT-PEG (Figure 1), the consequential/ Ca 2+ -dependent glutamate release was severely reduced (Figure 2), which implicated the negative modulation of the relationship between cytosolic Ca 2+ increase and the amount of exocytotically released glutamate from astrocytes. Interestingly, such negative modulation has been initially reported when tampering with astrocytic vesicular proteins [24]. Namely, the exocytotic release of glutamate from astrocytes critically relies on the presence of synaptobrevin 2/VAMP2 that docks vesicles to the plasma membrane, while the storage of glutamate in astrocytic vesicles requires the action of vacuolar-type H + -ATPase (V-ATPase) that creates H + gradient used by vesicular glutamate transporters (VGLUTs) delivering glutamate into the vesicular lumen [13]. Pre-treatment of astrocytes with a holoprotein of Tetanus toxin, that cleaves synaptobrevin-2, Bafilomycin A1, a specific inhibitor of V-ATPase, or Rose Bengal, an inhibitor of VGLUTs, all reduced mechanically-induced glutamate release, without affecting Ca 2+ responses. In addition, Rose Bengal reduced ATP-induced glutamate release, without affecting Ca 2+ responses [24]. Thus, it is possible that SWCNTs would interact with the lumenal portion of these vesicular proteins (and/or other vesicular proteins, lipids, and/or their glycosylated forms) to affect vesicular recycling and ATP-evoked glutamate release. If so, the type of vesicular fusion, size of vesicles and their fusion pores should matter and so would the geometry of SWCNT-PEG.
Astrocytic vesicles can exhibit full fusions, where vesicular membrane collapses into the plasma membrane with subsequent emergence of endocytic vesicles by pinching off the plasma membrane, or transient fusions, so called kiss-and-run fusions, where a fusion pore is transiently formed between the vesicular lumen and the extracellular space, vesicle undocks and then locally recycles. When stimulated with ATP, astrocytes displayed more full (on average 64%) than transient (on average 36%) vesicle fusion events [25] with vesicles in live astrocytes measuring~300 nm in diameter [25,43]. When astrocytes are bathed in a saline containing Ca 2+ at physiological extracellular levels (2 mM; same as here), spontaneous transient fusion pores can be, due to technical limitations, classified into two populations (50% each), one with vesicles with a wider, over 2.6 nm in diameter, pore size and the other with vesicles that have a narrower, on average 0.75 nm in diameter, fusion pore size [44]. As per the geometry of our SWCNT-PEG with lengths ranging from 0.1 to 1.8 µm (mode~0.3 µm) and diameters between 1 and 7 nm (mode~3 nm) [26], the vesicular lumen of most of the fusing vesicles in astrocytes would be accessible to SWCNT-PEG. This suggests that only~18% of vesicles (50% of 35% fusion events) would be inaccessible to SWCNTs in astrocytes due to a mismatch between the fusion pore size and diameter of the SWCNTs. The proportion of inaccessible astrocytic vesicles are likely even smaller, however, as ATP stimulation can dilate the fusion pore [45]. Be that as it may, if the physical insertion of SWCNT-PEG into the pore and their interaction with the vesicular lumen drive a reduction in glutamate release, there would be plenty (82% or more) of vesicles available to interact with SWCNTs and produce a 53% diminution of the glutamate release as we report here. However, if the SWCNT-PEG need to be endocytosed via full and/or kiss-and-run fusions in order to exert the effect on recycling, it is their length that would obstruct such a process. Furthermore, several groups have conjugated CNTs with biological molecules to get them into cells [46][47][48][49] albeit the mechanism of translocation into cells is elusive. Anecdotally, we find very rarely that astrocytes (less than 1 in a thousand) accumulate SWCNT-PEG in the cytoplasm ( Figure S4), which would implicate that endocytosis and/or CNT translocation into the cytosol are unlikely to mediate the inhibitory effect of evoked exocytosis. Rather, SWCNTs could be inserting into/ plugging the fusion pore to produce such an effect. This decrease in astrocytic glutamate release in the presence of SWCNT-PEG could be beneficial in conditions like epilepsy where an increase in glutamate release from astrocytes has been linked to epileptic activity [14]. However, appropriate safety guidance and methods to deliver these nanomaterials into the brain parenchyma for therapeutic use are not defined at present.
The decrease in the total and vesicular recycling dye load that we observed in the presence of SWCNT-PEG ( Figure 3) is qualitatively similar to that we previously observed in neurons [8]. As we used two different recycling dyes, FM4-64 here and FM1-43 in our previous work with neurons, known to have different properties [50], we stay away from temptation to make additional quantitative comparisons. However, it has to be stated as a well-accepted fact that the endocytosis/exocytosis process in astrocytes is much slower than that in neurons [24,25,40].
The process of constitutive exocytosis leads to the incorporation of new membrane and transporter-laden vesicles into the plasma membrane of astrocytes and this membrane can be retrieved by the process of endocytosis. Similar to many other membrane transporters, the glutamate transporters are also trafficked to and from the plasma membrane in an activity-dependent manner [51,52]. We show here that SWCNT-PEG blocks the process of constitutive endocytosis in astrocytes thereby causing an increase in the retention of new membrane leading to an increase in the area of astrocytes, as we reported previously [7,10,11]. This could also cause a decrease in the glutamate transporters being endocytosed once they are trafficked to the plasma membrane, hence causing an increase in the surface presence of the glutamate transporters and the resultant increase in glutamate uptake, as we have shown previously [11]. Though untested, this phenomenon could be explored in the future by transfecting the astrocytes with the plasmid encoding sodium-dependent excitatory amino acid transporters tagged with enhanced green fluorescence protein and record their vesicular trafficking [53] as well as plasma membrane dynamics while challenging astrocytes with SWCNT-PEG.
The results on destaining recycling dye from astrocytes treated with SWCNT-PEG during second stimulus ( Figure 4) are in stark contrast to the data obtained from neurons [8]. Such destaining was accelerated and more effective in neurons treated with SWCNT-PEG when compared to the untreated control neurons, both groups showing expected exponential rate of decay [8]. When this finding was combined with the reduction of endocytotic loading of the recycling dye during the first stimulation, we concluded that SWCNT-PEG preferentially inhibited regulated endocytosis [8]. Prior work showing that SWCNT-PEG dampens depolarization-induced cytoplasmic Ca 2+ elevation in neurons [6] diminished a possibility that the faster rate of destaining was due to an increase in exocytosis. In the present work, destaining during the second stimulus was slower and less efficient in astrocytes treated with SWCNT-PEG than in untreated control astrocytes. In fact, this deficiency was so severe that the normal exponential rate of destaining decay seen in control astrocytes was not present in SWCNT-PEG-treated astrocytes and assumed a linear rate of decay ( Figure 4). When these findings are combined with the reduction of endocytotic loading of the recycling dye during the first stimulation (Figure 3), we can only conclude that in astrocytes both arms of the regulated/evoked recycling pathway, endocytosis and exocytosis, are impaired by SWCNT-PEG. This gross difference between astrocytic and neuronal destainings could not be of technical nature, i.e., due to the use of different recycling dyes. The underlying molecular underpinnings of the dichotomy in destaining needs to be investigated, however. It is an attractive hypothesis that the difference in the average size of small/clear vesicles,~40 nm in neurons [54,55] vs.~300 nm in astrocytes [25,43] as well as in the microanatomy of vesicles and protein secretory machinery [13,25,41,43,55,56] might underlie the observed dichotomy in destaining. Another tempting speculation is that the nature of the stimulus, depolarization in neurons and ATP in astrocytes, could result in such a dichotomy. This notion is routed in our previous study showing that in astrocytes the proportion of two distinct types of events, transient and full fusions, was stimulus dependent, and so was the stability of the vesicle fusion pore [25].
Finally, we studied the effects of the PEG functional group on vesicular recycling in astrocytes. We previously demonstrated that the functional group PEG itself did not affect glutamate homeostasis unlike SWCNT-PEG that increased glutamate uptake (Figure 1B of ref. [11]) In the present work, PEG itself marginally inhibited constitutive endocytic load in astrocytes, but had no effect on ATP-evoked (pulse) endocytosis. Similarly, PEG also had some effect on the destaining curve during the second ATP stimulus (chase), showing slower and lesser destaining than the control. In all the conditions, PEG-treated groups showed a typical exponential decay of FM4-64 fluorescence. Taken together, we concluded that PEG itself does not mediate the major effects we observed that were exerted by SWCNT-PEG. However, data we obtained using PEG 600 (molecular weight range 573-630) might be considered unexpected. Namely, treatment of the drought-susceptible barrel clover (Medicago truncatula) accession Jemalong A17 for 5 days with 15% PEG (higher molecular weight of 8000) stimulated endocytosis in rhizodermal cells of the upper growth differentiation zone 4 of the roots. This effect, qualitatively opposite to what we observed in astrocytes using the same recycling dye FM4-64, was absent when the drought-resistant succession HM298 was used [57]. Undeniably, comparison between astrocytes isolated from the visual cortex of a rodent and rhizodermal cells in the roots of a legume may look awkward at the first sight. However, in legumes the PEG can simulate water deficit [57,58]. Given the well-recognized astrocytic function in water balance [59], it is tempting to speculate that the effects of PEG on vesicular recycling we observed in astrocytes may be related to the activity of aquaporin-4, a water channel expressed on the plasma membrane of astrocytes in the mammalian cerebrum [60].

Conclusions
Here, we investigated the mechanism of action of SWCNT-PEG solute on astrocytes by studying their intracellular Ca 2+ dynamics and the associated glutamate release along with their plasma membrane recycling. We show that the SWCNT-PEG solute does not affect the intracellular Ca 2+ elevations in astrocytes but causes a reduction in the ATP-induced glutamate release. We also show that the SWCNT-PEG solute causes a decrease in the total and vesicular load of the recycling dye during ATP-stimulated recycling with obstructed release of the loaded recycling dye following ATP stimulation ( Figure 5). Supplementary Materials: The following are available online at www.mdpi.com/2073-4409/9/7/1597/s1, Figure  S1: CPA blocks the ATP-induced intracellular Ca 2+ elevations in cultured mouse cortical astrocytes indicating that the ER Ca 2+ store is the primary source of this increase in Ca 2+ , Figure S2: The decrease/trough in the fluorescence of iGluSnFR at the astrocytic plasma membrane observed after the addition of ATP is not caused by the plasma membrane dynamics as evidenced by the lack of change in the fluorescence of Lck1-26-EGFP, Figure  S3: The decrease in the fluorescence of iGluSnFR observed after replacing the bath media is not ATP dependent, Figure S4: An example of rarely seen intercellular accumulation of SWCNT-PEG in an astrocyte.   Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4409/9/7/1597/s1, Figure S1: CPA blocks the ATP-induced intracellular Ca 2+ elevations in cultured mouse cortical astrocytes indicating that the ER Ca 2+ store is the primary source of this increase in Ca 2+ , Figure S2: The decrease/trough in the fluorescence of iGluSnFR at the astrocytic plasma membrane observed after the addition of ATP is not caused by the plasma membrane dynamics as evidenced by the lack of change in the fluorescence of Lck 1-26 -EGFP, Figure  S3: The decrease in the fluorescence of iGluSnFR observed after replacing the bath media is not ATP dependent, Figure S4: An example of rarely seen intercellular accumulation of SWCNT-PEG in an astrocyte.