Vanilloid receptor 1 regulates multiple calcium compartments and contributes to Ca2+-induced Ca2+ release in sensory neurons.

Vanilloid receptor 1 belongs to the transient receptor potential ion channel family and transduces sensations of noxious heat and inflammatory hyperalgesia in nociceptive neurons. These neurons contain two vanilloid receptor pools, one in the plasma membrane and the other in the endoplasmic reticulum. The present experiments characterize these two pools and their functional significance using calcium imaging and 45Ca uptake in stably transfected cells or dorsal root ganglion neurons. The plasma membrane localized receptor is directly activated by vanilloids. The endoplasmic reticulum pool was demonstrated to be independently activated with 20 microm capsaicin or 1.6 microm resiniferatoxin using a bathing solution containing 10 microm Ruthenium Red (to selectively block plasma membrane-localized receptors) and 100 microm EGTA. We also demonstrate an overlap between the endoplasmic reticulum-localized vanilloid receptor regulated stores and thapsigargin-sensitive stores. Direct depletion of calcium via activation of endoplasmic reticulum-localized vanilloid receptor 1 triggered store operated calcium entry. Furthermore, we found that, in the presence of low extracellular calcium (10(-5) m), either 2 microm capsaicin or 0.1 nm-1.6 microm resiniferatoxin caused a pronounced calcium-induced calcium release in either vanilloid receptor-expressing neurons or heterologous expression systems. This phenomenon may allow new insight into how nociceptive neuron function in response to a variety of nociceptive stimuli both acutely and during prolonged nociceptive signaling.

The sensation of pain can be as short as a pin-prick or in chronic disorders can last for many years. The vanilloid receptor 1 (TRPV1/VR1), 1 a sodium/calcium-permeable cation channel, is responsible for nerve terminal depolarization and generation of action potentials upon activation by ligand binding or thermal nociceptive stimuli (1). It is known that overactivation of TRPV1 (e.g. by intradermal administration of the vanilloid ligands capsaicin (CAP) or resiniferatoxin (RTX)) can damage neurons or nerve terminals through a calcium-dependent process, producing desensitization to painful stimuli (2)(3)(4)(5)(6). However, in most naturally occurring painful conditions such as inflammation or cancer-related pain, the nociceptive nerve ending maintains its sensitivity for prolonged periods (7,8). These observations suggest that nociceptive neurons have sophisticated mechanisms for intraneuronal Ca 2ϩ regulation to maintain both sensory transduction/transmission and structural integrity.
Similar to other excitable and non-excitable cells, nociceptive neurons have elaborated a multitude of mechanisms to maintain a low resting concentration of free calcium in the cytosol ([Ca 2ϩ ] i ) (9 -11). Negatively charged phospholipids, Ca 2ϩ -binding proteins, and Ca 2ϩ -transport proteins at the plasma membrane (PM), the endoplasmic reticulum (ER), and mitochondria (12)(13)(14)(15)(16) ensure that [Ca 2ϩ ] i is maintained at Ͻ100 nM. In addition to calcium elevation due to nociceptive stimuli, in tumor-associated pain or inflammatory conditions, TRPV1 can be sensitized by PKC or PKA phosphorylation (17)(18)(19)(20)(21), leading to a further elevation of the intracellular calcium burden. Although these data suggest that the integrity of the neuron or nerve ending can be jeopardized by chronic activation (4,5,22,23), other observations suggest that just the opposite occurs. Evidence from experimental models of arthritis and tumorinduced pain indicate that nerve fibers remain functional and even hypertrophy and proliferate in the vicinity of the neoplasm in conjunction with the development of hyperalgesia (8,24,25).
Recently, we identified two pools of TRPV1, one in the plasma membrane (TRPV1 PM ) and the other in the ER (TRPV1 ER ), and showed that activation of either population was capable of elevating intracellular calcium (5). Subsequently, other groups explored the possibility of vanilloid-sensitive intracellular calcium stores or used TRPV1 as a probe for intracellular Ca 2ϩ release in heterologous expression systems (26 -28). In this paper, we investigate 1) the role of TRPV1 ER in regulation of intracellular calcium and 2) the interaction between TRPV1 PM and the ER calcium stores in the context of a possible role in peripheral signaling mechanisms such as sensitization and neurogenic inflammation. Furthermore, we provide evidence that vanilloids or endovanilloids may use calcium-induced calcium release (CICR) to amplify Ca 2ϩ signals entering through TRPV1 PM or calcium increases, generated within the nerve ending through activation of TRV1 ER . The ability of the nociceptive neuron to access multiple calcium pools may be used to modulate sensitivity and possibly maintain the integrity of primary afferent endings and thereby nociceptive signaling in prolonged pain states.

EXPERIMENTAL PROCEDURES
Determination of [Ca 2ϩ ] i -DRG cultures were prepared from E16 embryonic rats as described previously (5). The culture conditions (poly-D-lysine/collagen-coated slides) yield neurons with both axons and dendrites (29). The dendrites are large processes branching from the cell body that taper in diameter further from the cell body. The axons are a constant diameter and project over long distances from the cell body. Before [Ca 2ϩ ] i determination, cells were washed three times in imaging medium (IM) composed of HEPES-buffered (pH 7.4) Hanks' balanced salt solution supplemented with 0.01% bovine serum albumin, 1.25 mM CaCl 2 , 0.8 mM MgCl 2 , 1 mM ascorbic acid, and 1 mM pyruvate and then loaded with 5 M Fluo-4 AM dye (Molecular Probes, Eugene, OR) for 30 min at 20°C. To remove excess dye, cells were washed three times in IM and kept in the dark for at least 20 min before being imaged in a closed chamber (PH-2, Warner Instrument Corp.) perfused with a pump (Miniplus 3, Gilson, Middleton, WI). In case of "zero" calcium experiments if not otherwise specified, before the addition of drugs, cells were washed for 1 min in IM containing 10 M Ruthenium Red (RR) and 100 M EGTA without the addition of any Ca 2ϩ . For determination of [Ca 2ϩ ] i , an upright microscope (Olympus BX60) illuminated with an arc lamp and equipped with a ϫ20 Uapo/340 water immersion objective was used. A shutter (Uniblitz, Rochester, NY), an excitation filter (495 nm), appropriate dichroic mirrors, and a long pass filter (515 nm) were positioned in the light path. The fluorescent signal was intensified (Model KS-1380, Videoscope International, Washington, D. C.) before capture by a CCD camera (Pulnix, Sunnyvale, CA). Images were digitized and integrated (2 frames/image) on a Macintosh computer running Synapse 3.7, an image acquisition and analysis program (Synergy Research, Silver Spring, MD). Neurons were outlined, and fluorescent intensities were analyzed. Fluorescence intensity values in the non-zero pixels within each slice were averaged (F) and plotted as normalized fluorescence intensities (⌬F/F 0 ) against time. Analysis of the traces and area measurements was performed by the Kaleidograph software (Synergy Software, Reading, PA).
Determination of Ca 2ϩ Release-TRPV1⑀-NIH3T3 cells (10 5 ), permanently expressing a C-terminally ⑀-epitope-tagged TRPV1 (5), were seeded in 24-well plates and then, on the next day, loaded with 10 M 45 Ca 2ϩ for 2 h. Ca 2ϩ release was determined during a 10-min incubation with RTX, CAP, and drugs as indicated in either 10 Ϫ5 M or zero [Ca 2ϩ ] o . To stop release, after incubation, cells were washed three times (ϳ2 min each) in 10 Ϫ5 M Ca 2ϩ -containing medium and the intracellularly retained 45 Ca 2ϩ was determined by liquid scintillation counting (Top-Count, Packard/PerkinElmer Life Sciences) after lysis for 30 min as described previously (5). Lower than normal extracellular (10 Ϫ5 M) Ca 2ϩ was necessary for washing, because 1.25 mM Ca 2ϩ causes passive calcium exchange by itself (data not shown).
Fluorescent Confocal Microscopy-COS7 cells or E15 rat primary dorsal root ganglion neurons were seeded on 25-mm coverslips and transfected with 1 mg of the TRPV1eGFP plasmid (5). Transfection of the COS7 cells was performed by SuperFect reagent (Qiagen, Valencia, CA), and DRG neurons were transfected with the Nupherin reagent (Biomol, Plymouth Meeting, PA) cultured for 24 h post-transfection at 35°C and then mounted in a 1-ml open chamber. To label the ER fluorescently, the ER-Tracker Blue-White® probe was employed (Molecular Probes) and loading of the cells was performed according to the recommendations of the manufacturer. The TRPV1eGFP/ER-Trackerlabeled cells were imaged with an MRC-1024 Bio-Rad confocal system with the appropriate GFP and UV excitation and emission filters.
Immunohistochemistry-Rat dorsal root ganglia were fixed in S.T.F. (Streck Laboratories, Inc., Omaha, NE) and paraffin-processed. Sections (5-m thick) were immunostained by a rat TRPV1-specific rabbit polyclonal antibody (Affinity Bioreagents Inc., Golden, CO). Following deparaffinization, sections were blocked with 10% normal goat serum (S-1000, Vector Laboratories, Inc., Burlingame, CA). Epitope unmasking was performed with Target Retrieval Solution (S1700, Dako, Carpinteria, CA) at 95°C for 20 min. Visualization of the immunocomplexes was done by the Vectastain Elite rabbit IgG and the peroxidase substrate kits (PK-6101 and SK-4100, Vector Laboratories, Inc.). Control sections for assessment of nonspecific binding were processed in an identical way with the exception of the omission of the primary or secondary antibodies. The stained sections were visualized with an Olympus BX 60 microscope equipped with a CCD camera (RT Slider, Diagnostic Instruments Inc. Burroughs, MI).

RESULTS
Morphological Evidence for ER Localization of TRPV1-The localization of TRPV1 to the ER membranes is an intriguing observation; however, most of the morphological evidence was obtained in fixed specimens. To better define the intracellular location of TRPV1 in vivo, live cell dual-label confocal imaging was employed. TRPV1eGFP, a previously constructed and functionally active green fluoroprotein-tagged chimera (5) was transiently transfected in COS7 cells and primary dorsal root ganglion cultures. Confocal microscopy of TRPV1eGFP fluorescence in live cells revealed the typical ER network. Simultaneous labeling with ER-Tracker Blue-White, a vital ER-labeling fluorescent dye, confirmed co-localization of TRPV1 to the ER (Fig. 1, a and b). Likewise, the ER lattice was also readily apparent when TRPV1eGFP was transiently transfected into DRG neurons, although the size and spherical geometry in sensory neurons makes it difficult to achieve resolution similar to the COS7 cells ( Fig. 1, a-c). These results are consistent with the dense neuronal intracellular staining seen in sections of trigeminal ganglion after fixation and immunohistological staining of tissue sections for TRPV1 (Fig. 1d).
Functional Characterization of TRPV1 ER -Calcium-imaging experiments were performed on native non-transfected rat DRG neurons. Because the DRG culture is a heterogeneous population of neurons, a scouting or screening pulse of capsaicin (1 M CAP in 1.25 mM Ca 2ϩ for 15 s) was administered to identify TRPV1-positive neurons and the short application time allowed us to minimize any desensitizing effects of CAP (Fig. 1e). Resiniferatoxin produced a remarkably prolonged response when applied for the same time even in 1000-fold lower concentration (Fig. 1f). To functionally characterize the ER-dependent elevations of [Ca 2ϩ ] i in isolation, experiments on TRPV1 ER were carried out in a closed flow-through chamber using buffer containing 10 M RR, a PM-impermeable TRPV1 channel blocker, and 100 M EGTA to ensure extracellular Ca 2ϩ levels of Յ10 Ϫ7 M (zero [Ca 2ϩ ] o ). RR effectively blocks Ca 2ϩ entry through TRPV1 in radioactive calcium transport experiments (Fig. 1g). Once the CAP-sensitive neurons were identified, the perfusion medium was changed to zero [Ca 2ϩ ] o containing 10 M RR. In the absence of extracellular calcium and with the TRPV1 PM channel blocker, 1 M CAP for 1 min did not evoke any change in [Ca 2ϩ ] i (data not shown). However, further elevation of the CAP concentration to 20 M induced a transient [Ca 2ϩ ] i increase and, upon repetition of the same CAP dose, the amplitude of the [Ca 2ϩ ] i increase was reduced to ϳ50% of the initial elevation (Fig. 2a).
Only neurons positive to the screening pulse of CAP responded to 20 M CAP in zero [Ca 2ϩ ] o conditions. This demonstrates that the [Ca 2ϩ ] i transients are specific for neurons with TRPV1 PM and those neurons contain TRPV1 ER . The decrease in the amplitude of the [Ca 2ϩ ] i transients observed after the second CAP application is consistent with the idea that CAP depletes an intracellular store of Ca 2ϩ of finite capacity, which is gated by TRPV1 ER . The increase in Fluo-4 AM fluorescence, indicative of increased [Ca 2ϩ ] i , was not confined to the cell body but also was observed along all of the visible neuronal processes ( Fig. 2, b and c), thus revealing vanilloid-inducible TRPV1 ER -operated Ca 2ϩ store throughout the entire extent of the neuronal dendritic and axonal specializations. After a 3-min re-perfusion with 1.25 mM Ca 2ϩ , 1 M CAP induced a [Ca 2ϩ ] i transient comparable to control levels.
To demonstrate that the CAP-gated store can be reversibly depleted and replenished, the cells were cycled between zero [Ca 2ϩ ] o perfusion medium supplemented with RR and medium containing a physiological calcium concentration (without RR) with periodic CAP challenges (Fig. 2, d-f). The first CAP challenge in zero [Ca 2ϩ ] o (Fig. 2d) was followed by a wash in normal calcium medium for 15 min after which cells were switched back to zero [Ca 2ϩ ] o . A second capsaicin application yielded a response of similar amplitude as the first one, and the response decreased when it was repeated without a preceding wash in normal calcium (Fig. 2e). Reperfusion of the cells with normal calcium-containing solution was sufficient to replenish the CAP-gated store and return the response obtained in zero [Ca 2ϩ ] o to control levels (Fig. 2f). These data indicate that the CAP-sensitive store can be depleted and replenished in a reversible fashion. To obtain total store depletion in these neurons and to characterize the relationship of the vanilloid-depletable pool with other known intracellular stores, we used the more potent vanilloid, RTX. Resiniferatoxin is approximately 1000 -10,000-fold more potent than CAP (30), and we also found that its off-rate from TRPV1 is much slower compared with CAP ( Fig. 1f) (see below) (31). Resiniferatoxin (1.6 M) produces almost complete depletion of the TRPV1-sensitive pool (Fig. 3a) (Fig. 3a), supporting the idea that calcium either leaves the cell or is sequestered in intracellular stores such as the mitochondria.
Vanilloid-induced Capacitative Ca 2ϩ Entry-One consequence of ER Ca 2ϩ depletion is activation of store-operated Ca 2ϩ entry at the PM (32). To test whether the RTX-induced increase in [Ca 2ϩ ] i comes from depletion of ER stores, we first completely depleted the TRPV1 ER Ca 2ϩ pool with RTX in EGTA and RR-containing extracellular fluid and then cells were switched back to normal calcium concentration in the presence of RR (Fig. 3b) to block TRPV1 PM . This resulted in a slowly rising but long duration elevation of [Ca 2ϩ ] i . The observed slow increase in [Ca 2ϩ ] i shows similar characteristics to classical store-operated Ca 2ϩ entry. The process was restricted to the vanilloid-sensitive subpopulation of neurons because cells in the same field of view that were negative to initial CAP or RTX stimulation did not show any change after the extracellular medium was returned to the normal Ca 2ϩ concentration (data not shown). These data are consistent with the idea that TRPV1-induced store depletion stimulates store-operated Ca 2ϩ entry. Parenthetically, it is important to note that without RR blocking when switching back to normal calcium concentrations, there is a massive calcium entry via TRPV1 PM because RTX maintains TRPV1 in the open state for prolonged period (data not shown). TRPV1 ER Overlaps with a Thapsigargin-Sensitive Pool-We further characterized whether the TRPV1 ER compartment overlapped with the thapsigargin (TG)-sensitive Ca 2ϩ pool. After identifying TRPV1-positive neurons with the CAPscreening pulse, 1.6 M RTX was applied two times in zero extracellular Ca 2ϩ to completely deplete TRPV1 ER . Neurons were then perfused with 2 M TG for an additional 10 min. Two main neuronal populations were distinguished by their characteristic responses. The first group and the main one studied in this paper is composed of CAP-inducible TRPV1-positive neurons in which RTX elicits a strong [Ca 2ϩ ] i transient. In these neurons, subsequent TG exposure minimally or did not provoke further Ca 2ϩ release (Fig. 3c). Measurement of the areas below the traces (n ϭ 39) showed that in vanilloidsensitive neurons, TG application produced a small calcium transient, which was 10 -14% on average compared with the RTX response. A second group of neurons did not respond to either CAP or RTX but only to TG (Fig. 3d) and was apparently TRPV1-negative. These findings indicate that vanilloids via TRPV1 ER can deplete an intracellular Ca 2ϩ store, which is replenished by SERCA-mediated Ca 2ϩ uptake (Fig. 3, c versus  d). Vanilloid-insensitive neurons and glia responded only to TG, indicating that RTX specifically affects a subpopulation of neurons, which are presumed to be nociceptors (Fig. 3, e versus f).
To further confirm that thapsigargin and TRPV1 ER agonists act on the same ER compartment, the order of store depletion was reversed. After the short (15 s) CAP-screening pulse, cultures were perfused first with 2 M TG, which induced a prolonged [Ca 2ϩ ] i transient (Fig. 4a). When [Ca 2ϩ ] i returned to base line, a second 15-s pulse of CAP was applied, which doubled the response compared with the first pulse (area under curve from the average trace of 24 neurons is 53.43 and 102.14, respectively). This increase suggests that normally the ER     (Fig. 4b). To visualize activation of both populations of TRPV1, RTX was applied in buffer containing low (10 Ϫ5 M) extracellular Ca 2ϩ . The low extracellular Ca 2ϩ resulted in a biphasic elevation of [Ca 2ϩ ] i , an initial rapid onset (within 1-2 s indicated by the asterisk) with an amplitude substantially smaller than in 1.25 mM extracellular Ca 2ϩ followed by a larger second peak (Fig. 4c). The first increase is attributed to opening of TRPV1 PM , whereas the second elevation represents Ca 2ϩ release from intracellular stores. Subsequently, the intracellular calcium level subsides to base line. Upon re-introduction of normal calcium concentration and without the addition of RTX, an abrupt long-lasting rise of [ To determine whether CICR can be eliminated by blocking TRPV1 PM , 10 M RR was added to the 10 M Ca 2ϩ -containing imaging medium (Fig. 4d). The resultant increase in intracellular calcium was comparable in amplitude and area with the activation of TRPV1 ER only, and it exhibited the same 20-s lag period (Fig. 4, b versus d). We attribute the delay to the time necessary to open TRPV1 ER , whereas no lag time is present in case of the CICR response. We next examined the ability of RR to block CICR in normal calcium conditions (Fig. 4e). The increase in intracellular calcium began with an initial sudden rise (no lag period) of [Ca 2ϩ ] i followed by a slower higher peak.

Activation of CICR in DRG Neurons-Activation
This finding is consistent with some leakage through TRPV1 PM prior to the full RR block.
Radioactive 45 Ca 2ϩ release experiments in the stably expressing TRPV1⑀-NIH3T3 cells provided further evidence for CICR. In these assays, cells were incubated with radioactive 45 Ca 2ϩ , allowing an exchange and replacement of the internal Ca 2ϩ pools with the radioactive isotope. This radioactively labeled pool was depleted with different vanilloid agonists, and the retained amount of 45 Ca 2ϩ was measured after the lysis of the cells. In 10 Ϫ5 M [Ca 2ϩ ] o , activation of CICR caused an augmentation of release and therefore lower counts were retained by the cells in comparison to the zero [Ca 2ϩ ] o condition where no CICR is evoked. Placing the cells into 10 Ϫ5 M or zero [Ca 2ϩ ] o did not change significantly the amount of the releasable 45 Ca 2ϩ ; however, the addition of vanilloid agonist produced a concentration-dependent release (Fig. 5a). RTX released more 45 Ca 2ϩ (i.e. less 45 Ca 2ϩ is retained) in the presence of 10 Ϫ5 M extracellular Ca 2ϩ than in zero [Ca 2ϩ ] o , which is consistent with CICR as well as with the calcium imaging data in Fig. 4, panels b and c. In separate experiments, CICR and the TRPV1 ER response was further characterized pharmacologically (Fig. 5b). RTX produced CICR and TRPV1 ER release even in low (2 nM) concentration (Fig. 5b, bar 1 versus 3 and bar 2 versus 4), which could be blocked by capsazepine (CPZ) (bar 5). At 2 M concentration, capsaicin also produced CICR (bar 6) but this concentration could not evoke direct Ca 2ϩ release via TRPV1 ER in zero [Ca 2ϩ ] o conditions (bar 7). Calcium-induced calcium release by CAP was also blocked by administration of 20 M CPZ (bars 6 versus 8), indicating the role of both TRPV1 PM and TRPV1 ER in CICR. Co-incubation of CAP with 100 M ryanodine, an inhibitor of ER Ca 2ϩ release (33), could not prevent the loss of Ca 2ϩ from intracellular sources in the presence of 10 Ϫ5 M extracellular Ca 2ϩ , suggesting that in the TRPV1⑀-NIH3T3 cell line, the vanilloid-induced CICR may use channels other than the ryanodine receptor (bars 6 versus 10). . CPZ specifically blocked the TRPV1 ER response elicited by RTX (bar 5) and inhibited CAP-induced CICR (bar 8). Co-incubation of CAP with a high concentration of RY (100 M) was not able to inhibit release of Ca 2ϩ from intracellular sources in 10 M Ca 2ϩ , suggesting that, in TRPV1⑀-NIH3T3 cells, vanilloid-induced CICR may use a channel(s) different from the RY receptor (bar 6 versus 10).

DISCUSSION
We present morphological and functional evidence for operation of TRPV1 in the ER of nociceptive neurons and cells expressing TRPV1 ectopically. Transient expression of TRPV1eGFP in DRG neurons and dual vital staining of TRPV1eGFP with the ER-Tracker Blue-White dye provides high resolution evidence of co-localization in living cells, confirming previous observations using immunohistochemistry in fixed preparations as well as demonstrating that this is not a fixation artifact (5,27,28,34).
The data presented here show that at least four sources can contribute to the pool of free intracellular Ca 2ϩ upon stimulation with vanilloid agonists: direct increases from opening TRPV1 in the PM or in the ER, store operated Ca 2ϩ entry, and CICR (summarized diagrammatically in Fig. 6). TRPV1 ER can be opened by vanilloid ligands, most of which are lipid-soluble and can cross the plasma membrane and thereby gain access to the intracellularly located TRPV1 (Fig. 6a). The TRPV1 ERgated compartment is finite, can be depleted and replenished, displays store-operated features, and overlaps with a TG-sensitive pool of Ca 2ϩ . The depletion by either agent appears to leave some residual calcium (7-15%) in the ER, which can be released by further treatment with either agent, although we cannot fully rule out the presence of small separate insensitive pools.
The selective demonstration of TRPV1 ER requires the use of RR as well as calcium-free extracellular fluid. The activation of TRPV1 ER probably extends to endovanilloids, which may either cross the PM (35) or be produced from intracellular membranes because of inflammation and nerve injury (36). Experiments using restricted extracellular Ca 2ϩ (10 Ϫ5 M Ca 2ϩ ) revealed the capacity for activation of CICR (Fig. 6b). Even a small amount of vanilloid-induced transmembrane calcium flux yields a large increase in [Ca 2ϩ ] i . The first phase of elevation of [Ca 2ϩ ] i is attributed to a flux via TRPV1 PM (Figs. 4c, asterisks, and 6b, inset) followed by a second much larger release of Ca 2ϩ from intracellular stores (37). The elevation in [Ca 2ϩ ] i triggered by CICR differs in several respects from that elicited by TRPV1 ER . First, with trace amounts of [Ca 2ϩ ] o , there is no temporal delay in the onset of [Ca 2ϩ ] i elevation, whereas TRPV1 ER exhibits a ϳ20-s delay. Second, the absolute elevation in [Ca 2ϩ ] i is approximately 3-fold greater than that obtained in the complete absence of [Ca 2ϩ ] o by measuring the area under the corresponding traces (Fig. 4, b versus c). Third, the elevation in [Ca 2ϩ ] i via TRPV1 ER is not preceded by the characteristic initial peak or shoulder. We think that this part of the trace can derive from calcium ions in close spatial association with the pore loop part of the channel as shown diagrammatically in Fig.  6b. These experiments clearly distinguish between TRPV1 ER and TRPV1 PM /CICR mechanisms for elevation of [Ca 2ϩ ] i . This same pool may also play a role in the initial transient seen when RR is present in normal calcium (see Fig. 6d, asterisk). Further experiments in which the preincubation period in low calcium was more prolonged (i.e. Ͼ150 s) (data not shown) suggest that the pool is loosely bound and can eventually equilibrate with the surrounding extracellular fluid. Nonetheless, even when the "peri-receptor" pool is equilibrated, an extracellular concentration of 10 Ϫ5 M Ca 2ϩ is sufficient to trigger CICR. In normal calcium and without RR present (Fig. 6c), we hypothesize that the vanilloid-stimulated increase in intracellular Ca 2ϩ is a composite involving TRPV1 PM , CICR, TRPV1 ER , and possibly store-operated entry. The deconvolution of the system requires careful isolation of the various components and compartments, and models of the conditions evaluated are summarized in Fig. 6.
The ability of TRPV1 to activate CICR may serve several functions. One potential function might be to act as a signal amplifier for weak endogenous agonists generated during inflammatory states. In many, if not most instances of chronic pain, the nerve endings containing TRPV1 are not exposed to potent vanilloids or 46 -51°C heat. Rather, in deep tissues, the nerve ending is exposed to a variety of algesic peptides, hydrogen ions, prostaglandins, and other lipid-based endovanilloid substances (e.g. in an inflamed knee joint). Many of these substances either sensitize the nerve ending or act as weak TRPV1 agonists. Thus, the amount of calcium influx these weak agonists can stimulate is generally and substantially less than that obtained with capsaicin or RTX. Therefore, the coupling of TRPV1 to CICR may be an important mechanism for increasing intracellular calcium evoked by weak agonists and modulating calcium-dependent processes such as vesicle release. Release studies of neuropeptides stored by primary afferents suggest that calcium elevations from either TRPV1 PM or TRPV1 ER can mediate release (38).
At the same time, this mechanism for calcium elevation may represent a more controlled source than high level flux through the plasma membrane (39,40). Upon vanilloid stimulation of TRPV1 PM , elevated [Ca 2ϩ /Na ϩ ] i , especially the calcium component (41), can be detrimental to the integrity of the primary afferent nerve ending or neuronal perikarya. This has been demonstrated in many studies using a wide variety of xenobiotic vanilloids and endovanilloid agonists and routes of administration from topical application to intraganglionic microinjections (2-6, 22, 41-44). However, functional integrity of the nerve ending is maintained over a wide range of nociceptive stimulus intensities and time periods (8,25). Intracellular calcium levels are known to be involved in regulation of TRPV1 PM and most probably TRPV1 ER via calcium-dependent desensitization and binding of calmodulin (45,46). Thus, during nociceptive stimulation, a very dynamic process may occur to maintain signaling capability that can involve CICR as well as calcium derived from some of the other sources identified in this report.
In addition to CICR, another major finding is that TRPV1 ER activation leads to store-operated Ca 2ϩ influx (32). The ER store can be depleted via activation of TRPV1 ER in zero Ca 2ϩ and in the presence of RR. The store-operated replenishment appears to occur via another ion channel, which is resistant to RR (Fig. 3b). The functional significance of the TG-sensitive ER compartment is underscored by the effective buffering of vanilloid-stimulated Ca 2ϩ influx mediated by this compartment in nociceptive neurons (Fig. 4a). This buffering has been noted previously in other neurons and non-neuronal cells (47,48).
The use of RTX revealed a robust functional TRPV1 ER pool, not only in the neuronal perikarya, but also in the extended axonal and dendritic processes (see Fig. 2, b and c). This is one of the first functional and visual demonstrations of an extensive ligand-activated ER network that is capable of Ca 2ϩ signaling. These data are consistent with ER-like membrane structures in the axoplasm (49 -52), which was proposed earlier as an alternative/parallel mechanism for signaling hypothesized in the "neuron within neuron" model (14). It is conceivable that the long unmyelinated axons of nociceptive neurons can employ Ca 2ϩ signaling mechanisms as an alternative or supplement to the normal regenerative sodium action potentials. At the nerve terminal, such locally generated Ca 2ϩ may contribute to axon reflexes and neurogenic inflammation by releasing tachykinins. More distantly spreading calcium waves may be involved in calcium-regulated transcriptional processes in the perikarya and alterations in expression profiles after nerve injury and during regeneration (7, 53-58). There is a quick onset of the signal followed by a shoulder (asterisk) and a slower rising second peak. The first part of the response is probably due to the Ca 2ϩ loosely associated with the external face of the TRPV1 pore, whereas the following response is a complex phenomenon containing Ca 2ϩ directly released from TRPV1 ER and CICR.
The multi-compartmental localization of TRPV1 and mobilization of several sources of Ca 2ϩ may provide new insights into some previous in vivo observations. RR, although a potent antagonist of Ca 2ϩ entry via TRPV1 PM in vitro (Fig. 1g), is not a very effective analgesic in vivo (59). CAP-induced signaling cannot be blocked by RR or by removing extracellular Ca 2ϩ from the washing medium (60,61). In our experiments, the activation of TRPV1 ER was possible in both conditions, implying a role of TRPV1 ER in nociceptive signaling as summarized in Fig. 6, a and d. In fact, capsazepine, a cell-permeable antagonist of both TRPV1 PM and TRPV1 ER , appears to reduce CAPinduced nociception and/or desensitization more efficiently than RR in vivo (see also Fig. 5b, bar 5) (62, 63).
The concurrent presence of CICR and agonist-activated TRPV1 ER may clarify several inconsistencies in the literature. In contrast to Liu et al. (27), we and others (28) found that the CAP-sensitive intracellular Ca 2ϩ store is co-extensive with the TG-depletable one. Although their experimental design was different, Wisnoskey et al. (28) did not observe store-operated Ca 2ϩ entry using TRPV1-transfected insect cells. In experiments on rat DRG, release from intracellular stores was attributed to current through ER-localized TRPV1, but the characteristic biphasic CICR pattern while evident (Figs. 2a and 4a) (see Ref. 27) was not commented on and might be attributed to the presence of trace amounts of [Ca 2ϩ ] o sufficient to elicit CICR. Furthermore, the reported EC 50 of 133 Ϯ 20 nM for CAP-induced Ca 2ϩ release by TRPV1 ER is similar to that obtained in normal Ca 2ϩ conditions by Cholewinski et al. (64). Our data are in concert with data showing an EC 50 for CAP of 13.5 Ϯ 0.22 M and 0.645 M for RTX (65,66). The difference in sensitivity of TRPV1 PM compared with TRPV1 ER may be related to differences in accessibility of the two pools to agonist. One possibility for the higher concentration needed to activate TRPV1 ER is that lipophylic substances such as capsaicin may be partitioned into the plasma membrane and other membrane-bound intracellular organelles. The free concentration in the cytoplasm may therefore be reduced compared with the extracellular concentration of the agonist. A second factor could be that the receptors are assembled into larger macromolecular entities or rafts in the PM versus the ER or differential localization of accessory proteins. This may change the local microenvironment in such a way that the TRPV1 PM is more accessible than TRPV1 ER to vanilloid agonists.
Earlier, the CAP-induced increase in [Ca 2ϩ ] i was attributed to direct activation of RY receptor rather than TRPV1 ER (65). Our studies suggest that release from intracellular sites does not employ the RY receptor-sensitive stores but this needs to be further characterized in DRG neurons. An overlap or interaction with an IP 3 -sensitive store cannot be ruled out (67). In fact, experiments with phospholipase C (PLC)-generated IP 3 and release of Ca 2ϩ from IP 3 -receptor sensitive stores in TRPV1transfected human embryonic kidney-293 cells indeed suggest an overlap between the RTX-and IP 3 -sensitive stores (66), but the direct participation of TRPV1 ER was not taken into consideration. A partial overlap between IP 3 and TRPV1-regulated stores has been reported in mast cells transiently transfected with TRPV1 (26). Although the pools may overlap, our results with 45 Ca 2ϩ and CPZ, a vanilloid antagonist, suggest that no additional intermediaries or second messenger(s) are required and that ligand binding to TRPV1 ER alone is necessary and sufficient for Ca 2ϩ release from the ER. Thus, it appears that TRPV1 ER is a common denominator in both ectopically expressing cells and endogenously expressing neurons and is a probable contributor to some of the differential observations.
In summary, the multi-compartmental localization of TRPV1 reveals an intricate and functionally important regulation of intracellular calcium in nociceptive neurons and warrants further research to elucidate its potential role in pain signaling.