Fatty-acid–binding protein 5 controls retrograde endocannabinoid signaling at central glutamate synapses

Significance

Endocannabinoids (eCBs), a family of active lipids, modulate synaptic transmission in the brain. Research conducted over the past decades has delineated the mechanisms by which eCBs control synaptic function. However, how these lipids are transported through the aqueous milieu of the synaptic cleft to engage cannabinoid receptors and ion channels remains a long-standing question in the field of eCB research. Here, we identify fatty-acid–binding protein 5 as a modulator of retrograde eCB signaling in the brain, which mediates the synaptic transport of the eCB 2-arachidonoyl glycerol. This protein may also control the synaptic transport of other eCBs and lipid-signaling molecules in the brain.

Abstract

Endocannabinoids (eCBs) are lipid-signaling molecules involved in the regulation of numerous behaviors and physiological functions. Released by postsynaptic neurons, eCBs mediate retrograde modulation of synaptic transmission and plasticity by activating presynaptic cannabinoid receptors. While the cellular mechanisms by which eCBs control synaptic function have been well characterized, the mechanisms controlling their retrograde synaptic transport remain unknown. Here, we demonstrate that fatty-acid–binding protein 5 (FABP5), a canonical intracellular carrier of eCBs, is indispensable for retrograde eCB transport in the dorsal raphe nucleus (DRn). Thus, pharmacological inhibition or genetic deletion of FABP5 abolishes both phasic and tonic eCB-mediated control of excitatory synaptic transmission in the DRn. The blockade of retrograde eCB signaling induced by FABP5 inhibition is not mediated by impaired cannabinoid receptor function or reduced eCB synthesis. These findings indicate that FABP5 is essential for retrograde eCB signaling and may serve as a synaptic carrier of eCBs at central synapses.

The endocannabinoid (eCB) system, composed of cannabinoid receptor type 1 (CB1R) and type 2 (CB2R), the transient receptor potential vanilloid type 1 receptor (TRPV1), and their endogenous lipid agonists, regulates numerous brain functions, including cognition, pain, reward, and feeding behaviors (1). Dysfunction of the eCB system is implicated in an array of psychiatric disorders, such as depression, anxiety, autism, and addiction (2–4). 2-Arachidonoylglycerol (2-AG) and anandamide (AEA), the two best-characterized eCBs (5, 6), are generally synthesized “on demand” in response to an increase in intracellular calcium induced by physiological or pathological neuronal activation (7–9) and released into the synaptic cleft by diffusion (10).

Released from postsynaptic neurons, eCBs typically act as retrograde messengers and activate presynaptic CB1Rs abundantly expressed in the brain (11). Activation of CB1Rs induces short- and/or long-term depression of glutamate and GABA release via various downstream signaling cascades, including inhibition of presynaptic voltage-dependent calcium channels, an increase in presynaptic potassium currents, and inhibition of the cAMP/PKA-signaling pathway (7, 12, 13). In addition to retrograde signaling, growing evidence indicates that eCBs can act as autocrine messengers (14, 15) and control the excitability of postsynaptic neurons by regulating several membrane currents, such as the G-protein–coupled inwardly rectified potassium current (14), the I_H current (16), and voltage-dependent potassium currents (15). Consequently, autocrine eCB signaling plays a key role in controlling dendritic excitability and the integration of synaptic inputs (16).

Synaptic eCB signaling is terminated by presynaptic reuptake and intracellular degradation of eCBs. 2-AG and AEA are hydrolyzed by monoacylglycerol lipase (MAGL) (17) and fatty acid amide hydrolase (18), respectively. While the mechanisms by which eCBs regulate synaptic transmission and neuronal excitability have been well characterized (19), the mechanism(s) underlying their synaptic transport remains unknown. In particular, it is unknown how eCBs diffuse through the synaptic cleft to engage CB1Rs. The lipophilic nature of eCBs limits their capacity to traverse the synaptic cleft, suggesting the presence of uncharacterized mechanism(s) that facilitates their synaptic transport.

Recently, we have identified fatty-acid–binding proteins (FABPs) as intracellular carriers that deliver eCBs to their catabolic enzymes or nuclear receptors (20). The brain expresses three FABP subtypes: FABP3, FABP5, and FABP7 (21) and, of these, FABP5 and FABP7 display the highest affinities for eCBs (22). Inhibition of these FABPs reduces the cellular inactivation of AEA and increases its levels in the brain (23), highlighting their important role in intracellular eCB trafficking. In addition to this canonical role, recent studies have reported the secretion of FABPs from various cell types through a nonconventional mechanism (24, 25). This raises the possibility that FABPs may also control the extracellular trafficking of eCBs, including their synaptic transport. Here, we tested this notion by examining the impact of pharmacological and genetic inhibition of FABP5 on phasic and tonic eCB signaling in the dorsal raphe nucleus (DRn), which play a key role in the regulation of stress homeostasis and anxiety-like behaviors (4). We found that inhibition of FABP5 abolished phasic and tonic eCB-mediated control of glutamate synapses without impairing CB1R function or 2-AG and AEA synthesis. These findings indicate that FABP5 controls retrograde eCB signaling, most likely by facilitating the synaptic transport of eCBs, and uncover a possible mechanism of eCB transport at central glutamate synapses.

Results

FABP Inhibition Impairs Phasic 2-AG Signaling in the DRn.

FABP5 and FABP7 mediate the intracellular trafficking of eCBs and promote their degradation (20, 26). Consequently, inhibition of FABPs, which increases eCB levels in the brain (23), in principle should potentiate phasic retrograde eCB signaling at central synapses. To test this notion, we examined the impact of SBFI-26, an inhibitor of FABP5 and FABP7 (26), on the magnitude of the depolarization-induced suppression of excitation (DSE) in DRn neurons, a response mediated by retrograde 2-AG signaling (27, 28), as blockade of 2-AG synthesis abolished the DSE (Fig. S1). Consistent with previous reports (28, 29), in control slices, a membrane depolarization (5 s, from −70 to 0 mV) of DRn neurons elicited a robust DSE of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor-mediated excitatory postsynaptic currents (AMPAR-EPSCs) (Fig. 1A). Unexpectedly, treatment of DRn slices with SBFI-26 (10 μM), a concentration that inhibits FABPs (26), profoundly reduced the magnitude of DSE (Fig. 1A, n = 10), indicating that FABP inhibition impairs 2-AG signaling in the DRn.

Fig. 1.

Inhibition of FABP impairs phasic 2-AG signaling in the DRn. (A) Inhibition of FABPs with SBFI26 reduces the DSE. (Lower) The DSE obtained in control (●, n = 10) and in the presence of SBFI26 (10 μM, O, n = 10). (Upper) The AMPAR-EPSC traces taken at the time indicated by numbers. (B) FABP inhibition blocks the α₁-AR-LTD induced by phenylephrine (PE, 100 μM). (Lower) The α₁-AR-LTD obtained in control (●, n = 10) and in slices treated with SBFI26 (10 μM, O, n = 10). (Upper) AMPAR-EPSCs traces taken at the corresponding time points. (C) Summary of the effect of SBFI-26 (10 μM) on AMPAR-EPSC amplitude. (Upper) Sample AMPAR-EPSC traces collected at the corresponding time points. (D) FABP inhibition does not alter the function of CB1R. (Lower) Summary of the depression of AMPAR-EPSCs induced by WIN55, 212–2 (10 μM) in control (●, n = 9) and in slices treated with SBFI26 (10 μM, O, n = 9). (Upper) Sample AMPAR-EPSC traces taken at the corresponding time points. (E) Effect of FABP inhibition on 2-AG (Left) and AEA levels (Right) in the DRn. (F) Inhibition of FABP does not alter Pr of glutamate release. (Left) The average PPR of AMPAR-EPSCs obtained in control (O, n = 7) and SBFI26-treated slices (O, n = 7, P > 0.05). (Right) The corresponding AMPAR-EPSC traces. (Scale bars: 50 pA, 10 ms.)

To further test the role of FABPs in controlling phasic 2-AG signaling, we examined the effect of FABP inhibition on G_q/G₁₁-coupled receptor-driven 2-AG signaling in the DRn. Activation of postsynaptic G_q/G₁₁-coupled receptors, including α₁-adrenergic receptors (α₁-ARs), triggers the release of 2-AG (27, 30), which mediates long-term depression (LTD) of glutamate synapses in the DRn (30). Therefore, we assessed the impact of FABP inhibition on the α₁-AR-LTD. In control slices, administration of the α₁-AR agonist phenylephrine (PE, 100 µM) readily induced an LTD of AMPAR-EPSCs (Fig. 1B). Remarkably, inhibition of FABPs with SBFI-26 (10 µM) abolished the α₁-AR-LTD (Fig. 1B, n = 10). To test whether the blockade of α₁-AR-LTD was due to a dysfunction of α₁-ARs, we examined the effect of SBFI-26 on the amplitude of α₁-AR–induced inward current (I_PHE) (31). We found that inhibition of FABPs did not alter the amplitude of I_PHE (control = 55.5 ± 5.2 pA; SBFI-26 = 52.5 ± 6.5 pA, P = 0.125, n = 10), indicating that the blockade of α₁-AR-LTD was not mediated by altered α₁-AR function. Collectively, these findings indicate that inhibition of FABPs does not facilitate, but rather impairs, phasic 2-AG signaling at DRn glutamate synapses.

Although previous study has shown that SBFI-26 exhibits no detectable affinity for CB1Rs (26), we examined whether the blockade of phasic 2-AG signaling was mediated by an off-target effect of SBFI-26 on CB1Rs. We first tested whether SBFI-26 (10 μM) can activate CB1Rs and occlude the DSE and α₁-AR-LTD. Stimulation of CB1Rs is known to inhibit AMPAR-EPSCs in the DRn (28). Therefore, we tested the effect of SBFI-26 on AMPAR-EPSC amplitude and found that administration of SBFI-26 (10 μM) did not inhibit, but rather induced, small nonsignificant potentiation of AMPAR-EPSCs (Fig. 1C, n = 7), indicating that SBFI-26 does not activate CB1Rs in this essay. Consequently, the SBFI-26–induced blockade of DSE and α₁-AR-LTD could not be attributed to an occlusion effect.

Next, we examined whether the SBFI-26–induced blockade of phasic 2-AG signaling was mediated by an altered CB1R function. This was accomplished by examining the effects of SBFI-26 on the depression of AMPAR-EPSCs induced by the CB1R agonist WIN55, 212–2 (10 µM). We found that treatment of DRn slices with SBFI-26 (10 μM) did not affect the magnitude and the duration of WIN55, 212–2–induced depression of AMPAR-EPSCs (Fig. 1D, n = 9). These results indicate that SBFI-26 does not affect CB1R function and exclude the possibility that the SBFI-26–induced blockade of phasic 2-AG signaling is mediated by altered CB1R function.

Because inhibition of FABPs increases eCB levels in the brain (23), it is possible that the SBFI-26–induced blockade of DSE and α₁-AR-LTD could be mediated by an occlusion effect caused by persistent CB1R-induced inhibition of glutamate release. We tested this possibility using two distinct strategies. First, we assessed the impact of SBFI-26 on 2-AG levels and found that inhibition of FABPs enhanced AEA (Fig. 1E, n = 5), but did not significantly alter 2-AG levels in the DRn (Fig. 1E, n = 5). Second, we reasoned that, if inhibition of FABPs were to occlude the DSE and α₁-AR-LTD, treatment of DRn slices with SBFI-26 (10 μM) should induce a persistent inhibition of glutamate release. We tested this notion by examining the impact of FABP inhibition on the paired-pulse ratio (PPR) and coefficient of variation (CV) of AMPAR-EPSCs, two measures of the probability (Pr) of glutamate release. The results showed that pretreatment of slices with SBFI-26 (10 μM) did not affect the PPR (Fig. 1F) and CV (control = 0.35 ± 0.04; SBFI-26 = 0.37 ± 0.05, n = 7) of AMPAR-EPSCs. Such findings demonstrate that inhibition of FABPs does not affect glutamate release and that the blockade of DSE and α₁-AR-LTD is unlikely to be mediated by an occlusion effect.

FABP Inhibition Abolishes Tonic 2-AG Signaling in the DRn.

In addition to phasic 2-AG signaling, previous studies have reported tonic 2-AG release in the brain (32, 33), which fine-tunes basal synaptic transmission (32, 33). Therefore, to examine the role of FABPs in controlling tonic 2-AG signaling, we first tested whether glutamatergic transmission in the DRn is under the control of tonic 2-AG signaling by assessing the effects of the CB1R blockade on AMPAR-EPSC amplitude. We found that AM251 (3 µM), a CB1R antagonist/inverse agonist, significantly increased the AMPAR-EPSC amplitude (Fig. 2A, n = 8) and reduced the PPR (Fig. 2B, n = 8), indicating an increase in glutamate release. Similar potentiation of AMPR-EPSCs (135.65 ± 4.9%, P = 0.002, n = 8) was induced by NESS0327 (1 µM), a neutral CB1R antagonist, demonstrating the presence of tonic eCB signaling that controls glutamatergic transmission in the DRn. To identify the eCB species involved, we tested the effect of inhibiting diacylglycerol lipase alpha (DAGLα), which mediates 2-AG synthesis (33), on the AM251-induced potentiation of AMPAR-EPSCs. Our results showed that treatment of DRn slices with the DAGLα inhibitor tetrahydrolipstatin (THL, 10 µM) abolished the AM251-induced potentiation of AMPR-EPSCs (Fig. 2C, n = 10), indicating that tonic eCB signaling is mediated mainly by 2-AG. Next, we tested the effect of FABP inhibition on the AM251-induced potentiation of AMPAR-EPSCs, a measure of tonic 2-AG signaling. We found that inhibition of FABPs with SBFI-26 (10 μM), which did not alter CB1R function (Fig. 1D), blunted the AM251-induced potentiation of AMPAR-EPSCs (Fig. 2D, n = 10). Collectively, these results indicate that FABPs also control tonic 2-AG signaling in the DRn.

Fig. 2.

Inhibition of FABP blocks tonic 2-AG signaling in the DRn. (A) Tonic eCB controls baseline glutamatergic transmission in the DRn. (Left) The potentiation of AMPAR-EPSC amplitude induced by AM251 (3 μM, n = 8). (B) Summary of the effect of AM 251 on the PPR of AMPAR-EPSC. (C) Tonic eCB signaling is mediated by 2-AG. (Lower) The AM251-induced potentiation of AMPAR-EPSCs obtained in control (●, n = 10) and in slices treated with the DAGLα inhibitor THL (10 µM, O, n = 10). (Upper) The corresponding AMPAR-EPSC traces. (D) Inhibition of FABPs abolishes tonic 2-AG–mediated control of AMPAR-EPSCs. (Lower) The AM251-induced potentiation of AMPR-EPSCs obtained in control (●, n = 10) and in SBFI26-treated slices (10 µM, O, n = 10). Treatment with SBFI26 prevented the AM251-induced potentiation of AMPAR-EPSCs. (Upper) AMPAR-EPSC traces taken at the indicated time points. (Scale bars: 100 pA, 10 ms.)

FABP5 Is Expressed and Localized in the Synapses of the DRn.

Of the two FABP subtypes inhibited by SBFI-26, FABP5 exhibits a wider distribution in the brain (21) and constitutes a likely candidate for controlling tonic and phasic 2-AG signaling. Therefore, using an immunohistochemical approach, we assessed the expression of FABP5 in the DRn. The results revealed significant FABP5 labeling in all DRn subdivisions of wild-type, but not FABP5 KO mice, thereby confirming the selectivity of the antibody used and the expression of FABP5 in the DRn (Fig. 3A). To serve as synaptic 2-AG transport, FABP5 should be localized at the synapses. We tested this notion using postembedding immunogold electron microscopy (EM) and determined the subcellular distribution of FABP5 in the DRn. FABP5 distribution in DRn synapses was quantified using the relative labeling index (RLI) approach that compares gold particle distribution frequency within synapses with distribution in the axon terminals, postsynaptic density, and whole cells (34). Our results showed the expected cytoplasmic FABP5 distribution. In addition, we routinely observed deposition of gold particles at synapses (Fig. 3B). RLI analysis revealed an enrichment of gold particles at synapses and postsynaptic densities compared with a predicted random distribution pattern (Fig. 3 B and C and Table S1). The specificity of our approach was confirmed in sections from FABP5 KO mice that displayed significantly lower tissue labeling (Fig. 3B) and a lack of synaptic enrichment, indicating random gold particle distribution (Table S2). Chi-squared (χ2) analysis confirmed the nonrandom enrichment of gold particles in synapses and postsynaptic density (Table S2).

Fig. 3.

FABP5 is expressed and localized in DRn synapses. (A) Immunohistochemical distribution of FABP5 in the DRn of WT (Left) and FABP5 KO mice (Right). Note the expression of FABP5 (red) in all DRn subdivisions of WT but not KO mice. Nuclei (blue) are labeled with DAPI. (B) EM images of the cellular distribution of FABP5 in the DRn of WT (Left panels) and FABP5KO (Right panels). Arrows indicate a synapse location. (C) Specific RLI were calculated from immunogold distribution in the DRn of wild-type and FABP5 KO mice.

FABP5 Is Required for Phasic and Tonic 2-AG Signaling in the DRn.

The findings that inhibition of FABP5 blocks 2-AG signaling combined with its synaptic localization in the DRn suggest that this protein carrier is necessary for retrograde synaptic 2-AG signaling. To further confirm this notion, we examined the impact of genetic deletion of FABP5 on both phasic and tonic 2-AG signaling in the DRn. Phasic 2-AG signaling was assessed by the magnitude of the DSE obtained in the DRn of wild-type and FABP5 KO mice. We found that genetic deletion of FABP5 abolished the DSE (Fig. 4A, n = 8). To test the impact of FABP5 deletion on tonic 2-AG signaling, we used two different approaches. First, we measured the potentiation of AMPAR-EPSCs induced by blockade of CB1R in the DRn of wild-type and FABP5 KO mice. As expected, in wild-type mice, administration of AM251 (3 μM) elicited a potentiation of AMPAR-EPSCs similar in magnitude to that obtained in rat DRn slices (Fig. 4B, n = 10). However, in FABP5 KO, administration of AM 251 failed to potentiate the amplitude of AMPA-EPSCs (Fig. 4B, n = 12). Second, we examined the impact of FABP5 deletion on the depression of AMPAR-EPSC induced by inhibition of 2-AG degradation. We found that administration of the MAGL inhibitor JZL184 (1 μM), which presumably increased synaptic 2-AG levels in the DRn, significantly depressed the amplitude of AMPAR-EPSCs in wild type (Fig. 4C, P = 0.015, n = 9), but not in FABP5 KO mice (Fig. 4C, P = 0.552, n = 9), confirming that FABP5 is indispensable for tonic 2-AG signaling. Importantly, because of the presynaptic localization of MAGL, where 2-AG is degraded, the present finding further supports that FABP5 is required for synaptic transport of 2-AG.

Fig. 4.

Genetic deletion of FABP5 abolishes phasic and tonic 2-AG signaling in the DRn. (A) FABP5 deletion blocks the DSE. (Left) DSE in WT (●) and FABP5 KO mice (O, n = 8). (Right) The corresponding AMPAR-EPSC traces collected at the indicated time points. (B) FABP5 deletion abolishes the AM251-induced potentiation of AMPAR-EPSCs. (Lower) A summary of AM251-induced potentiation of AMPAR-EPSCs in WT (●, n = 12) and FABP5 KO mice (O, n = 12). (Upper) AMPAR-EPSC traces collected at the corresponding time points. (C) Genetic deletion of FABP5 blocks the JZL184-induced depression of AMPAR-EPSCs. (Lower) The depression of AMPAR-EPSCs induced by JZL184 (1 μM) in WT (●, n = 9) and FABP5 KO mice (O, n = 9). (Upper) AMPAR-EPSCs collected before and during JZL184. (Scale bars: 100 pA, 20 ms.)

The blockade of phasic and tonic 2-AG signaling in FABP5 KO mice was not mediated by reduced expression and function of CB1Rs. Indeed, Western blot analysis revealed that the expression levels of CB1R in the DRn of FABP5 KO mice were comparable to wild type (Fig. S2A). Importantly, genetic deletion of FABP5 did not affect the magnitude of the depression of AMPAR-EPSCs induced by the CB1R agonist WIN55, 212–2 (10 μM) (Fig. S2B, P > 0.05 vs. wild type, n = 9), further confirming that CB1R function is not altered in FABP5 KO mice.

To test whether the blockade of synaptic 2-AG signaling in FABP5 KO mice was mediated by an overall deficit of 2-AG synthesis and or an increase of its degradation, we assessed the expression of DAGLα and quantified the levels of 2-AG in DRn slices of wild-type and FABP5 KO mice. The results of Western blot analysis revealed that the level of DAGLα expression in FABP5 KO mice was similar to that obtained in wild-type mice (Fig. S2 C and D). In addition, deletion of FABP5 significantly increased 2-AG (Fig. S2E, P = 0.013, n = 5), but not AEA, levels (Fig. S2F, P = 0.160, n = 5) in the DRn. We also examined the expression of MAGL in the DRn and found that its expression level in FABP5 KO mice was comparable to wild type (Fig. S2 G and H). Collectively, these results indicate that the impaired synaptic 2-AG signaling induced by genetic deletion of FABP5 was not mediated by a dysfunction of CB1R or a deficit in 2-AG synthesis and metabolism, but most likely by a blockade of synaptic transport of 2-AG.

To serve as synaptic transport of 2-AG, FABP5 should be secreted in the extracellular milieu by neurons, astrocytes, or microglia. In the brain, FABP5 is expressed in neurons and astrocytes, but not microglia (21). Therefore, we examined FABP5 secretion using a primary culture of neurons and astrocytes that express FABP5 (Fig. S3A). Our results revealed time-dependent secretion of FABP5 by astrocytes, but not by neurons (Fig. S3B). In addition to FABP5, astrocytes also express FABP7, which was not secreted (Fig. S3C), indicating a preferential secretion of FABP5. Such results are consistent with early reports of the presence of FABP5 in the cerebrospinal fluid (35, 36) and further support that FABP5 may serve as synaptic carrier of 2-AG.

Discussion

Retrograde synaptic eCB signaling is governed by the synthesis of various eCB species in postsynaptic neurons, their transport to presynaptic CB1Rs, and their prompt degradation (19). Unlike classical hydrophilic neurotransmitters, which are released by exocytosis and diffuse across the synaptic cleft, eCBs and other lipids are thought to be released by passive diffusion (10). However, once released, their hydrophobicity limits their diffusion through the synaptic cleft, suggesting the existence of molecular mechanisms that facilitate their synaptic transport. Here we demonstrate that FABP5, an intracellular carrier of eCBs (20, 22), controls retrograde 2-AG signaling, most likely by facilitating its synaptic transport.

Several lines of evidence support this conclusion. First, pharmacological inhibition of FABP5 blocks the DSE and the PE-induced LTD, which are mediated by phasic 2-AG signaling (28, 30). Inhibition of FABP5 also prevents the potentiation of AMPAR-EPSCs induced by CB1R antagonists, which is mediated by tonic 2-AG signaling. Second, genetic deletion of FABP5 impairs both phasic and tonic 2-AG–mediated control of glutamatergic transmission in the DRn. Third, FABP5 is highly expressed in all of the subdivisions of the DRn and is secreted and clustered at synapses, thereby positioning this protein carrier as a key regulator of synaptic 2-AG signaling in the DRn.

Examination of the mechanisms by which FABP5 inhibition impairs synaptic 2-AG signaling revealed that this effect was not mediated by a dysfunction of CB1Rs. Indeed, we showed that inhibition of FABP5 did not alter the expression of CB1R nor their ability to inhibit AMPAR-EPSC. In addition, because FABP5 inhibition did not affect glutamate release, we concluded that the blockade of tonic and phasic 2-AG signaling was not caused by a persistent CB1R-induced depression of glutamate release, which could occlude the DSE and α₁-AR-LTD. Furthermore, although deletion of FABP5 increased 2-AG levels in the DRn, it did not enhance, but rather blocked, the DSE. Deletion of FABP5 also blunted the potentiation and depression of AMPAR-EPSC amplitude induced by blockade of CB1R and inhibition of MAGL, respectively, two independent measures of tonic 2-AG signaling. A parsimonious interpretation of these results is that FABP5 inhibition impairs tonic and phasic 2-AG signaling in the DRn, most likely by blocking its synaptic transport. This conclusion is further substantiated by the finding that inhibition of FABP5 prevents the depression of AMPAR-EPSCs induced by inhibition of MAGL, a presynaptically located enzyme that metabolizes 2-AG. Thus, in addition to the intracellular transport of eCBs, FABP5 may also serve as a synaptic carrier for 2-AG.

Although FABP5 lacks the signal peptide that directs its entry into the secretory pathway, and consequently attains a predominantly intracellular localization, recent reports indicate that FABP5 is secreted by various cell types (24, 25) and is present in the cerebrospinal fluid (35, 36). Consistent with these reports, our data revealed that astrocytes secrete FABP5, further supporting the role of FABP5 as an extracellular carrier of eCBs. Thus, it is tempting to speculate that secreted FABP5 controls the synaptic transport of 2-AG. However, the involvement of intracellular FABP5 in controlling 2-AG release and/or transport cannot be excluded at the present time. Future studies are required to determine the precise contribution of intracellular and secreted FABP5 in mediating the synaptic transport of 2-AG.

We have previously shown that inhibition of FABP5 elevates AEA levels and produces CB1R-mediated antinociceptive effects (23, 26, 37), which appear incongruent with the present findings that FABP5 inhibition blocks synaptic eCB signaling. However, it is worth noting that the CB1R-mediated antinociceptive effects are mediated by inhibition of peripheral, but not central, FABP5 (37). Although speculative, this may suggest different mechanisms that control peripheral eCB signaling. Importantly, the observation that inhibition of central FABP5 does not elicit CB1R-dependent antinociceptive effects (37) is consistent with a blockade of eCB signaling. Interestingly, inhibition of FABP5 has also been shown to block AEA signaling at intracellular receptors that regulate cognition (38), which further supports our conclusion that FABP5 may serve as a synaptic carrier of eCBs.

The concept that FABP5 could serve as a synaptic 2-AG carrier may address an unexplained feature of eCB signaling. Indeed, eCBs including 2-AG are lipophilic molecules that do not require membrane transporters for their cellular release and uptake (22). However, previous studies have suggested the presence of a putative membrane eCB transporter. This notion is based on the inhibition of eCB uptake by structural eCB analogs (39, 40) and the reduction of retrograde eCB signaling, which has been attributed to inhibition of eCB release (41, 42). Interestingly, we have previously shown that the putative eCB membrane transporter inhibitors exert their effects through inhibition of intracellular FABP5 function (22). Based on these results and the present finding that FABP5 is indispensable for retrograde 2-AG signaling, we propose that FABP5 inhibition may also underlie the previously reported effects of inhibitors of the putative eCB membrane transporter.

In addition to eCBs, FABP5 promiscuously binds and transports long-chain fatty acids (43). Thus, it is tempting to speculate that, in addition to eCBs, FABP5 may likewise control retrograde synaptic signaling mediated by other neuroactive lipids such as arachidonic acid, which has been shown to fine-tune synaptic transmission and plasticity in the brain (44). Collectively, our study ascribes a function of FABP5 in synaptic eCB transport and positions this protein as a regulator of eCB signaling in the brain.

Materials and Methods

Animals.

Male C57BL/6 mice, FABP5 KO mice on a C57BL/6 background, and Sprague–Dawley rats (6–10 wk old) were used for all experiments, which were approved by the University at Buffalo and Stony Brook University Animal Care and Use Committees in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (45).

In Vitro Electrophysiological Recordings.

Whole-cell recordings were obtained from DRn neurons with patch electrodes (3–5 mΩ) using standard procedures (27). AMPAR-EPSCs were evoked with local electrical stimulation delivered using patch pipettes filled with artificial cerebrospinal fluid (ACSF). More details on procedure and data analysis are provided in SI Materials and Methods.

Immunofluorescence and Electron Microscopy.

DRn sections (30–50 μm) from wild-type and FABP5 KO mice were processed with rabbit anti-FABP5 at 1:500 dilution for immunofluorescence and electron microscopy analysis. Details on the procedure and analysis are provided in SI Materials and Methods.

Western Blotting.

DRn slice lysates were separated by gel electrophoresis using 4–12% gradient SDS-polyacrylamide gels. The blots were processed using polyclonal anti-FABP5 or anti-GAPDH as primary antibodies and horseradish peroxidase-conjugated goat anti-rabbit as a secondary antibody. A detailed procedure is provided in SI Materials and Methods.

Primary Cell Culture and Secretion Studies.

For secretion studies, primary neuronal cultures were maintained in 24-well culture plates. Every 48 h 50% of the media was replaced with fresh B27-supplemented neurobasal media. The aspirated neuronal culture media was pooled and stored at −20 °C until further Western blot analysis. A detailed procedure is provided in SI Materials and Methods.

Statistics.

All data are represented as mean ± SEM, and P < 0.05 derived from the standard Student t test is considered statically significant.