Activation of adenosine A2B receptors enhances ciliary beat frequency in mouse lateral ventricle ependymal cells

Background Ependymal cells form a protective monolayer between the brain parenchyma and cerebrospinal fluid (CSF). They possess motile cilia important for directing the flow of CSF through the ventricular system. While ciliary beat frequency in airway epithelia has been extensively studied, fewer reports have looked at the mechanisms involved in regulating ciliary beat frequency in ependyma. Prior studies have demonstrated that ependymal cells express at least one purinergic receptor (P2X7). An understanding of the full range of purinergic receptors expressed by ependymal cells, however, is not yet complete. The objective of this study was to identify purinergic receptors which may be involved in regulating ciliary beat frequency in lateral ventricle ependymal cells. Methods High-speed video analysis of ciliary movement in the presence and absence of purinergic agents was performed using differential interference contrast microscopy in slices of mouse brain (total number of animals = 67). Receptor identification by this pharmacological approach was corroborated by immunocytochemistry, calcium imaging experiments, and the use of two separate lines of knockout mice. Results Ciliary beat frequency was enhanced by application of a commonly used P2X7 agonist. Subsequent experiments, however, demonstrated that this enhancement was observed in both P2X7+/+ and P2X7-/- mice and was reduced by pre-incubation with an ecto-5'-nucleotidase inhibitor. This suggested that enhancement was primarily due to a metabolic breakdown product acting on another purinergic receptor subtype. Further studies revealed that ciliary beat frequency enhancement was also induced by adenosine receptor agonists, and pharmacological studies revealed that ciliary beat frequency enhancement was primarily due to A2B receptor activation. A2B expression by ependymal cells was subsequently confirmed using A2B-/-/β-galactosidase reporter gene knock-in mice. Conclusion This study demonstrates that A2B receptor activation enhances ciliary beat frequency in lateral ventricle ependymal cells. Ependymal cell ciliary beat frequency regulation may play an important role in cerebral fluid balance and cerebrospinal fluid dynamics.


Background
The cerebral ventricles are lined by a layer of ciliated ependymal cells that play an important role in cerebral fluid balance [1]. It has been estimated that each ependymal cell possesses 20-30 motile cilia, which are 8-20 μm in length with a 9 + 2 microtubule structure. Their ciliary tufts are organized in a manner consistent with the direction of cerebrospinal fluid (CSF) flow [2]. Abnormalities in ciliogenesis or ciliary function are frequently associated with hydrocephalus [3][4][5][6][7][8][9][10][11], and ependymal denudation can be observed in cases of communicating hydrocephalus [12]. Despite the increased recognition that ependymal cells are important for regulating CSF dynamics, only a few reports have specifically looked at the extracellular signaling mechanisms involved ependymal cell ciliary beat frequency modulation. Nelson and Wright (1974) noted enhancement of frog brain ependymal ciliary beat frequency by ATP (adenosine 5'-triphosphate), cAMP (adenosine 3',5'-cyclic monophosphate), theophylline, and acetylcholine, as well as decreases in ciliary beat frequency by a number of other agents, using an in vitro preparation [13]. A later study by Nguyen et al. (2001) observed an ATP-mediated decrease in ciliary beat frequency, as well as a serotonin-mediated increase, in 4 th ventricle ependymal cells in cultured rat brain slices and acutely isolated ependymal cells [14]. Finally, reports from O'Callaghan et al. have demonstrated that both hydrogen peroxide and bacterial pneumolysin inhibit ciliary beat frequency in rat brain ependymal cells [15,16].
Recent work from our laboratory demonstrated that the purinergic P2X 7 receptor is widely expressed on ependymal cells [17]. Furthermore, receptor activation leads to increases in intracellular calcium ([Ca 2+ ] i ) both in the soma and cilia. Working under the hypothesis that the P2X 7 receptor may be involved in regulating ciliary beat frequency, we have conducted experiments using highspeed video capture and differential interference contrast (DIC) microscopy to investigate potential modulation of ciliary beat frequency by purinergic agonists. These experiments have demonstrated, however, that the adenosine A 2B receptor is primarily responsible for ciliary beat frequency enhancement by these agents. Further experiments using A 2B -/-/β-galactosidase reporter gene knock-in mice confirmed this observation and also demonstrated a residual P2X 7 -mediated component to ciliary beat frequency enhancement.

Slice preparation
Research protocols were approved by the Yale University Institutional Animal Care and Use Committee (approval #A3230-01). C57BL/6 mice (n = 48; Jackson Laboratories, Bar Harbor, ME, USA), CD1 mice (n = 7; Charles River Laboratories, Wilmington, MA, USA), P2X 7 knockout mice (n = 5; P2rx7 tm1Gab , Jackson Laboratories, [18]), and A 2B knockout mice (n = 7, [19]) were used for the present experiments. Mean age of animals was 24.3 ± 1.0 days (range . Animals were anesthetized with pentobarbital, 50 mg/kg, intraperitoneal (IP); after craniotomy and dissection, horizontal brain slices (250-300 μm) were prepared in chilled (4°C)  and a 2× teleconverter. Ciliated ependymal cells were visually identified along the subventricular zone (SVZ) border lining the lateral surface of the lateral ventricles (e.g. Fig. 1A). Agonists and antagonists were bath applied, and only one exposure or experimental condition was permitted per slice. After preliminary time course experiments (see Fig. 1B), ciliary beat frequency measurements were analyzed at baseline then five min after agonist application unless otherwise indicated. Antagonists and inhibitors were always pre-applied (range 4-15 min), depending on the site of action (extracellular versus intracellular), and our prior experience using these agents in patch clamp experiments [17]; they were also present during agonist applications (for antagonist experiments only) to decrease the possibility of antagonist washout.

Ciliary Beat Frequency Analysis
Ciliary beat frequency on lateral ventricle ependymal cells was analyzed using modifications of a previously published approach [15]. High-speed video acquisition of ciliary beat frequency was performed using a Pioneer A640-210 gm GigE camera (Basler Vision Technologies, Exton, PA, USA) with StreamPix3 software (Norpix Inc., Montreal, Quebec, Canada). One-sec videos along the ependymal wall were digitally acquired to a Dell Computer (Round Rock, TX, USA) at 200 frames per sec (fps). Files were converted to multi-TIFF stacks of 200 images and imported into ImageJ (NIH, Bethesda, MD, USA), where the stacks were re-sliced along a line placed across the ciliary tuft, thus creating pseudo-line scans. Ciliary beat frequency was calculated by measuring peak-to-peak intervals of periodicity evident in the pseudo-line scan and derived from the following equation, with each pixel representing 1/200 of a second.
Ten periods were measured for each video, representing cilia from 3-6 ependymal cells on average. Analysis was conducted blinded to experimental conditions and with randomized file names and chronology, thus decreasing potential bias. Ciliary beat frequency data from pharmacological studies are presented using the following two equations: Comparison of our methodology to separate manual counting of ciliary beat frequency in the 1 sec video playback, as well as repeat analysis of duplicate files (with randomized names and chronology), yielded a strong correlation as determined by linear regression (r 2 = 0.905 and 0.95 respectively; data not shown).

β-galactosidase (β-gal) expression analysis in A 2B reporter mice β-gal expression in A 2B
-/-/β-gal reporter gene knock-in mice was studied according to a previously published protocol [19]. Mice were anesthetized with isoflurane, perfused with 20 ml PBS through the left heart ventricle, and perfusion fixed with 30 ml 2% paraformaldehyde in PBS. Brains were removed, cut into 2 mm coronal sections containing intact lateral ventricular walls, and stained for βgal activity using X-gal staining solution: 5 mM K 3 Fe(CN) 6 , 5 mM K 4 Fe(CN) 6 ·2 mM MgCl 2 in PBS, with a final concentration of 1 mg/ml 5-bromo-4-chloro-3indolyl-β-D-galactopyranoside (X-gal, American Bioanalytical; Nantick, MA, USA), then incubated at 37° for 6-12 h, rinsed in PBS, and stored in 4% paraformaldehyde. Sections were embedded in low melting point agarose (American Bioanalytical), resectioned to 100 μm, and mounted directly onto slides or used for subsequent immunocytochemistry as previously described.

Calcium imaging
Acute mouse brain slices were loaded with the Ca 2+ -sensitive dye Fluo-4 AM (Invitrogen; 4 μM in dimethyl sulfoxide (DMSO) with 20% Pluronic F-127) using ependymadirected applications by a Picospritzer II (1-2 psi; Parker Instrumentation, Cleveland, OH, USA). Slices were washed for a minimum of 10 min before recording. The Ca 2+ imaging system consisted of a confocal laser scan-ning microscope (Olympus) with a 60× water objective (NA 0.9) and Fluoview software (Olympus). Agonists in Ca 2+ imaging experiments were focally applied using a Picospritzer II (as above). Calcium data were analyzed using the Calsignal program [20].

Statistics
Data were analyzed and presented in SigmaPlot 8.0 (SPSS, Chicago, IL, USA). Statistical significance was determined using the Student's t-test (P < 0.05). Data are presented as mean ± standard error of the mean (SEM) unless otherwise indicated. Reported n values refer to the number of slices tested (with each slice including 10 ciliary beat frequency measurements; see above).

Purinergic enhancement of ciliary beat frequency is present in P2X7 +/+ and P2X7 -/mice and is induced by nonselective adenosine receptor agonist
Given our prior demonstration of ciliary (and somatic) localization of P2X 7 receptors on lateral ventricle ependymal cells [17], we first sought to determine whether BzATP (a commonly used P2X 7 agonist) was also able to induce changes in ciliary beat frequency. Ciliated ependymal cells were visualized in horizontal mouse brain slices using high-speed DIC microscopy ( Fig. 1A; see Methods for ciliary beat frequency calculations). Average baseline ciliary beat frequency was 11.4 ± 0.2 Hz (n = 160) in wildtype mice. While 300 μM BzATP (Fig. 1B) was able to increase ciliary beat frequency in C57BL/6 wild-type (P2X 7 +/+ ) mice, a similar increase was also observed in P2X 7 -/animals. Fig. 1C shows the % increase after a 5 min application of 300 μM BzATP in P2X 7 +/+ mice (58.9 ± 3.4%) and in P2X 7 -/mice (58.6 ± 4.1%). These responses were not significantly different.

Histochemical and functional evidence for A 2B expression: immunocytochemistry and A 2B -/-/β-gal reporter gene knock-in mice
We next sought to confirm A 2B expression by ependymal cells using immunocytochemical methods. While distinct A 2B immunoreactivity was observed in ependymal cells (Fig. 3A), non-selective nuclear staining was also observed throughout the central nervous system (CNS) and therefore precluded definitive interpretation. Two additional A 2B antibodies did not show any CNS labeling (data not shown). A 2A immunoreactivity was evident in the striatum and in a scattered distribution along the SVZ but not in ependymal cells (Fig. 3B). An alternative approach was therefore used as a verification of the presence of A 2B receptors.

Previously characterized A 2B
-/-/β-gal reporter gene knockin mice [19] were surveyed for A 2B gene promoter-driven expression of β-galactosidase in ependymal cells along the lateral ventricle. Strong X-gal reaction product was observed in the lateral septal nucleus, and clear intracellular labeling was also visible in ependymal cells and scattered throughout the cortex and striatum (Fig. 3C). This pattern was observed in A 2B -/-/β-gal mice but not in wildtype controls. β-gal immunoreactivity was also observed in the A 2B -/-/β-gal mice in S100β-positive ependymal cells (Fig. 3D-F), providing an additional layer of evidence for A 2B expression by ependyma. Ependymal X-gal reaction product is also visible in the corresponding Fig. 3G.

Discussion
The present experiments demonstrated that activation of the adenosine A 2B receptor enhanced ciliary beat frequency in mouse lateral ventricle ependymal cells -a conclusion supported by pharmacological experiments using selective adenosine receptor agonists and antagonists, as well as experiments using the A 2B -/-/β-gal mice. The fact that BzATP application onto mouse brain slices can lead to activation of a non-P2X 7 -mediated pathway is not surprising. For example, prior studies in the hippocampus have demonstrated that BzATP can induce non-P2X 7mediated effects through the action of ecto-nucleotidases, nucleoside transporters, and subsequent adenosine receptor activation [27]. Ependymal cells have been shown to express ecto-nucleotide pryrophasphatase/phosphodiesterase 1 (NPP1) and ecto-5'-nucleotidase [28,29], and the decrease in BzATP-mediated effects after pre-incubation with αβmADP (an ecto-5'-nucleotidase inhibitor; Fig.  1C) suggests that ciliary beat frequency enhancement is Ependymal localization of A 2B : evidence from immunocytochemistry and X-gal staining Figure 3 Ependymal localization of A 2B : evidence from immunocytochemistry and X-gal staining. (A) Cytoplasmic A 2Bimmunoreactivity was evident in ependymal cells (see inset) of wild type mice, although non-specific nuclear labeling was also evident throughout the brain and confounds interpretation of ependymal immunoreactivity. (B) No labeling of ependymal cells was observed using an antibody to A 2A receptors in wild type mice, although strong immunoreactivity was evident in the striatum and in a scattered distribution along the SVZ. (C) DIC image from an A 2B -/-/β-gal reporter gene knock-in mouse showing darkening of cells due to X-gal precipitate in regions surrounding the ependymal layer (e). Strong X-gal labeling was observed in the lateral septal nucleus (ls), while scattered labeling was observed in the striatum (st) and cortex (ctx) but not in the corpus callosum (cc). The septum mechanically separated from the corpus callosum during the staining procedure, thus obliterating the dorso-medial boundary of the lateral ventricle (lv) in this slice. (Bar = 500 μm). (D-F) Immunocytochemistry from an A 2B -/-/ β-gal reporter gene knock-in mouse demonstrating that β-galactosidase (D; green, Bar = 25 μm) and S100β (E, red) are colocalized in ependymal cells (F). Nuclei are stained with DAPI (blue). (G) Corresponding DIC image with darkening of the ependyma due to X-gal precipitate.
largely dependent on a metabolic breakdown product rather than BzATP itself.
It should be noted that BBG was used as the sole P2X 7 antagonist in these studies, as the more commonly used adenosine 5'triphosphate-2',3'-dialdehyde (oATP) induced toxicity in prior experiments (unpublished observations) and KN-62 has demonstrated a weaker activity at mouse versus human P2X 7 receptors [30]. Our recent whole-cell patch clamp experiments, however, showed clear antagonism of ependymal cell BzATP-induced currents by low concentrations of BBG [17]. Furthermore, no additional P2X receptor subtypes were detected during patch clamp recordings of P2X 7 -/mice [17]. While data in Fig. 1C, Fig. 2C, and Fig. 4B argue that a minor P2X 7mediated component to BzATP-induced ciliary beat frequency enhancement is present, it is most easily observed in the absence of the adenosine A 2B receptor (Fig. 4B).
The lack of ciliary beat frequency enhancement with 100 μM UTP (Fig. 1D), and the absence of a suramin-mediated antagonism of ATP-induced changes in beat frequency (Fig. 1D), strongly argue against a P2Y-mediated modula-tion of frequency in the present experiments. These data do not altogether eliminate the possibility, however, that another subtype of P2Y-receptor may play a role in beat frequency modulation. A more extensive pharmacological analysis (with inclusion of appropriate ecto-nucleotidase inhibitors to prevent breakdown of purinergic drugs into adenosine receptor agonists) is clearly desirable and should be the focus of future investigation.
Interestingly, ATP has previously been shown to decrease ciliary beat frequency in rat 4 th ventricle ependymal cells [14]. It is reasonable to assume that species and regionspecific differences may exist in ependymal cell response to ATP, which is obviously dependent on the subtypes of purinergic receptors expressed. For example, in our mouse lateral ventricle ependymal cells, [Ca 2+ ] i increases rapidly after BzATP application ( Fig. 2D and [17]); this is in sharp contrast to ATP's lack of [Ca 2+ ] i effect in the previously mentioned rat experiments [14]. Other proteins expressed by ependyma during development -such as glial fibrillary acidic protein (GFAP) and vimentin -vary markedly between species, developmental stage, and location along the ventricular system [31]. Future work on anatomic as Ciliary beat frequency analysis in A 2B -/-/β-gal reporter gene knock-in mice Figure 4 Ciliary beat frequency analysis in A 2B -/-/β-gal reporter gene knock-in mice. (A) Histogram showing the absence of ciliary beat frequency enhancement due to 1 μM NECA (n = 10) and 30 μM adenosine (n = 10) in the A 2B -/mice. (B) Enhancement due to 300 μM BzATP application was reduced in the A 2B -/mice (n = 8) versus wild-type P2X 7 +/+ mice (n = 5). BzATPinduced enhancement was eliminated in the A 2B -/mice after pre-incubation of the slices with 100 nM BBG (n = 8). *: P < 0.05 for all panels, data are means ± SEM. The n value indicates the number of slices tested. (C) Summary diagram showing enzymatic breakdown of BzATP and subsequent receptor activation. ATP (an endogenous signaling molecule analogous to BzATP) is shown in grey.
well as species-specific differences in ependymal cell ciliary beat frequency regulation is clearly warranted.
A 2B can be coupled to multiple G-protein cascades, including the adenylate cyclase (Gs; cAMP) pathway and the phospholipase C (Gq11) signaling pathways [32][33][34][35]. Furthermore, activation of the phospholipase C -mediated pathway can lead to [Ca 2+ ] i increases after A 2B activation [32]. In the present experiments, however, neither NECA (1 μM) nor adenosine (30 μM) were able to induce [Ca 2+ ] i increases in ependymal cells (Fig. 2D), arguing against a Ca 2+ -mediated mechanism for A 2B -induced enhancement of ciliary beat frequency. While additional pathways involved in A 2B -mediated signaling were not explored in the present experiments, a complete understanding of these pathways may prove critical for determining the importance of receptor signaling cascades in CSF dynamics. For example, a recent study by Mönkkönen et al. (2007) has demonstrated that knockdown of G αi2 can lead to ciliary stasis and ventricular dilation [11].
Nucleotide signaling and purinergic receptor expression in the developing brain has been the subject of intense investigation (for review, see [36]). For example, the developmental precursors of ependyma -radial glia [37] can propagate ATP-mediated Ca 2+ waves that are dependent on P2Y 1 receptor expression [38]. Immature ependyma are born between embryonic days E14 and E16 in the mouse, although cell maturation and cilia formation typically occur during the first postnatal week [37]. Little is known regarding the functional role of purinergic receptors on these cells during this time. It should also be noted that neuroblast migration from the SVZ to the rostral migratory stream depends on the normal flow of CSF, and ciliary motility is required for maintaining a diffusional gradient of inhibitory guidance molecules in the CSF [39]. Whether receptor-mediated changes in ciliary beat frequency play a role in this phenomenon is not known. Purinergic receptor expression on CSF secreting cells of the choroid plexus has also been the subject of recent investigations [40,41].
Additional questions clearly remain to be answered. Is the source of endogenous ATP or adenosine autocrine or paracrine? Does ciliary beat frequency correlate with the metabolic requirements in the CNS, and might ciliary beat frequency dysregulation be associated with hydrocephalus? While answers to these questions are beyond the scope of the present experiments, much remains to be learned about the role of purinergic receptors and ciliary beat frequency in cerebral fluid dynamics.

Conclusion
While abnormal ciliary structure and function has been associated with hydrocephalus in several experimental models, the signaling mechanisms responsible for the normal regulation of ependymal cell ciliary beat frequency are not well understood. The present experiments demonstrate that activation of the adenosine A 2B receptor enhances ciliary beat frequency in lateral ventricle ependymal cells. A residual contribution of purinergic P2X 7 receptors to frequency regulation is also supported. Purinergic modulation of ependymal cell beat frequency may play an important role in maintaining normal fluid balance in the CNS. Future experiments should focus on understanding whether purinergic dysregulation contributes to pathologic conditions such as hydrocephalus.