Research ArticleGPCR SIGNALING

Metalloprotease cleavage of the N terminus of the orphan G protein–coupled receptor GPR37L1 reduces its constitutive activity

See allHide authors and affiliations

Science Signaling  12 Apr 2016:
Vol. 9, Issue 423, pp. ra36
DOI: 10.1126/scisignal.aad1089

This article has been retracted. Please see:

Turning off a constitutively active GPCR

In animals, most G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptors (GPCRs) are inactive under basal conditions and require either binding of a ligand, isomerization of their bound ligand by light, or cleavage of a tethered ligand to become activated. Coleman et al. found that GPR37L1, which is a GPCR abundant in the cerebellum, signaled to Gαs and stimulated cyclic AMP signaling in the absence of any exogenously added ligand when expressed in a cultured cell line and assayed in the presence of normal growth medium. Furthermore, cleaved fragments of this receptor are detectable in human cerebral spinal fluid, and biochemical analysis revealed a long and short form of the receptor. The short form, which resulted from cleavage of the N terminus, was inactive and was the predominant form detected in rodent cerebellum. Metalloprotease inhibitors blocked the cleavage of the N terminus, suggesting that signals that regulate the activity of these proteases, rather than ligand-binding events, control the signaling mediated by this GPCR. How widespread this type of GPCR regulatory mechanism is and whether the cleaved fragments have bioactivity remain interesting questions.


Little is known about the pharmacology or physiology of GPR37L1, a G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptor that is abundant in the cerebellum. Mice deficient in this receptor exhibit precocious cerebellar development and hypertension. We showed that GPR37L1 coupled to the G protein Gαs when heterologously expressed in cultured cells in the absence of any added ligand, whereas a mutant receptor that lacked the amino terminus was inactive. Conversely, inhibition of ADAMs (a disintegrin and metalloproteases) enhanced receptor activity, indicating that the presence of the amino terminus is necessary for GPR37L1 signaling. Metalloprotease-dependent processing of GPR37L1 was evident in rodent cerebellum, where we detected predominantly the cleaved, inactive form. However, comparison of the accumulation of cAMP (adenosine 3′,5′-monophosphate) in response to phosphodiesterase inhibition in cerebellar slice preparations from wild-type and GPR37L1-null mice showed that some constitutive signaling remained in the wild-type mice. In reporter assays of Gαs or Gαi signaling, the synthetic, prosaposin-derived peptide prosaptide (TX14A) did not increase GPR37L1 activity. Our data indicate that GPR37L1 may be a constitutively active receptor, or perhaps its ligand is present under the conditions that we used for analysis, and that the activity of this receptor is instead controlled by signals that regulate metalloprotease activity in the tissue.


G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptors (GPCRs) are the largest family of integral membrane proteins encoded by the human genome, with more than 800 unique members (1). They are a major class of pharmacological targets—a function of the high degree of receptor-ligand specificity, the extracellular localization of their ligand-binding sites, and the discrete distribution of individual GPCRs in the body (2). An estimated 36% of existing, marketed pharmaceuticals act at GPCRs (3). These receptors are typically activated by binding of one or more of a diverse range of extracellular stimuli, including peptide hormones, lipids, amines, and ions, leading to activation of heterotrimeric G proteins and subsequent signal amplification (4). Depending on the receptor conformation stabilized by the ligand, multiple distinct pathways can be activated or repressed. For some GPCRs, “constitutive activity,” that is, signaling in the absence of ligand, has been reported in vitro and in vivo (5). To maintain their exquisite spatial and temporal specificity, GPCRs are tightly regulated both by receptor internalization and down-regulation (6).

The GPR37-like 1 receptor, or GPR37L1, is an orphan GPCR that modulates cerebellum development in mice (7). It was originally identified as endothelin B receptor-like protein 2 or GPCR/CNS2 by two independent groups in different species (8, 9) and, on the basis of sequence homology, belongs to the family A peptide GPCR subfamily (9). GPR37L1 is highly abundant in the central nervous system (CNS). It is also found in the heart and, to a lesser extent, in the kidneys (8). This spatial distribution suggests that it may be a promising therapeutic target for the treatment of both neurological and cardiovascular disorders, but little is known about the receptor.

Although GPR37L1 shares 32% identity and 56% similarity with the endothelin B receptor (8, 9), it does not bind endothelin or related peptides (9). Its closest relative, GPR37 (parkin-associated endothelin-like receptor), is also an orphan. Currently, little is known about GPR37L1 pharmacology (10), although one study reported that the neuropeptide agonists prosaposin (PSAP) and TX14A, a synthetic 14–amino acid neuroprotective peptide that is derived from the saposin C domain of PSAP, stimulated Gαi protein signaling downstream of GPR37L1 and GPR37 (11), but without independent validation of these data, the receptor remains classified as an orphan. Our analyses of human cerebrospinal fluid (CSF) identified peptides (1214) that are identical to three distinct regions of the GPR37L1 N terminus, suggesting a processing and activation mechanism akin to that of the protease-activated receptors (PARs) (15). Here, we tested the hypothesis that the N terminus regulates GPR37L1 function. We found that GPR37L1 is subject to metalloprotease-dependent proteolytic processing of its N terminus, when expressed both in cultured cells and in cerebellar tissue. Furthermore, unlike PARs, we found that GPR37L1 lacking the N terminus was an inactive form of an otherwise constitutively active receptor. This suggests a previously unknown mechanism of regulating GPCR activity in animals.


GPR37L1 is N-terminally processed

We stably transfected Flp-In T-REx human embryonic kidney (HEK) 293 cells with human GPR37L1 constructs in which the C terminus is fused to enhanced yellow fluorescent protein (GPR37L1-eYFP). We have previously used this doxycycline-inducible system for orphan or low-affinity GPCRs because it provides, in the absence of induction, negative expression controls that are essential when working with poorly characterized receptors (1621). We generated serial N-terminal truncations to assess the effect of removing the putative signal peptide (Δ25 GPR37L1-eYFP), the signal peptide plus the region encompassing all three reported N-terminal fragments (Δ80 GPR37L1-eYFP) found in CSF, or the entire N terminus (Δ122 GPR37L1-eYFP) on receptor production and delivery to the cell surface (Fig. 1A).

Fig. 1 GPR37L1 is expressed in HEK293 cells as both a full-length and N-terminally truncated species.

(A) Schematic of the GPR37L1 N terminus and location of serial truncations. CSF, cerebrospinal fluid–derived peptides identified by mass spectrometry (1214). Asn105 is the predicted glycosylation site. WT, wild type; TM1, start of transmembrane domain 1. (B) Immunoblot of lysates from each cell line with an antibody targeted to GFP (green fluorescent protein). Cells were induced with doxycycline (500 ng/ml) for 24 hours before harvest. WB, Western blot; αGFP, antibody recognizing GFP and also eYFP. (C) Immunoprecipitation of biotinylated lysates shown in (B). Open circle, full-length receptor; closed circle, truncated receptor. IP, immunoprecipitation. (D) Cells induced for 24 hours with doxycycline (500 ng/ml) were lysed and treated for 2 hours with either endoglycosidase H (EndoH; 100 mU) or peptide N-glycosidase F (PNGaseF; 2 U), or both, before immunoblotting with the GFP antibody. Star symbol represents deglycosylated receptor. (E) Cell lysates for full-length and Δ122 GPR37L1-eYFP were visualized by in-gel fluorescence (488 nm) for evidence of C-terminal proteolysis. All blots are representative of three independent experiments. dox, doxycycline.

Induction of wild-type GPR37L1-eYFP with doxycycline resulted in robust accumulation of the receptor (Fig. 1B). Of note, GPR37L1-eYFP resolved as two molecular weight species by SDS–polyacrylamide gel electrophoresis (SDS-PAGE): a larger band of Mr ~65 to 70 kD (Fig. 1B, open circle) and a smaller band of ~40 to 45 kD (Fig. 1B, closed circle). The smaller–molecular weight species was equivalent in size to Δ122 GPR37L1-eYFP, which we detected as a single species. Cell surface biotinylation labeled both GPR37L1-eYFP species, and the relative ratio of larger–molecular weight to smaller–molecular weight species appeared similar between whole-cell lysates and biotinylated samples (Fig. 1, B versus C). We also detected Δ122 GPR37L1-eYFP at the cell surface; thus, we reasoned that the smaller species of GPR37L1 might represent an N-terminally processed version of the receptor. Deglycosylation of the single predicted glycosylation site, Asn105 (8), failed to reduce the larger band to the same size as the smaller band (Fig. 1D), indicating that the smaller species is unlikely to be a precursor of full-length GPR37L1, which is consistent with the biotinylation results. Neither the smaller band observed in cells expressing GPR37L1-eYFP nor the band in cells expressing Δ122 GPR37L1-eYFP shifted in response to deglycosylation, consistent with cleavage of the N terminus occurring after Asn105. Furthermore, we could exclude C-terminal truncation because both the full-length and truncated versions of GPR37L1-eYFP were visible by in-gel fluorescence (Fig. 1E), indicating that the C-terminal eYFP moiety was still present.

Unlike the Δ122 GPR37L1-eYFP construct, which presumably reaches the cell surface as a result of a cryptic signal sequence within its first transmembrane domain (22), we could not detect either the Δ25 or Δ80 truncated forms even when we exposed the cells to the highest concentrations of doxycycline (1 μg/ml), as confirmed by confocal microscopy (fig. S1A), immunoblotting (Fig. 1, B and C), and whole-cell eYFP fluorescence (fig. S1B). To determine whether this was due to the rapid degradation of improperly folded receptor, we exposed the cells for 6 hours with the proteasomal degradation inhibitor MG132 (Z-Leu-Leu-Leu-H). However, the proteins were undetectable under these conditions (fig. S1C), indicating that these truncated versions could not be properly folded or synthesized.

The N terminus of GPR37L1 is truncated by a metalloprotease

Because some GPCRs undergo N-terminal cleavage catalyzed by metalloproteases (2325), we hypothesized that MMPs (matrix metalloproteases) or ADAMs (a disintegrin and metalloproteases) cleaved GPR37L1. Induction of GPR37L1 expression in the presence of the nonspecific MMP and ADAM inhibitor BB94 (batimastat) almost completely prevented generation of the smaller receptor species in cells expressing GPR37L1-eYFP (Fig. 2A). Furthermore, we observed a concomitant concentration-dependent increase in full-length GPR37L1-eYFP and reduction in the smaller species (Fig. 2B), consistent with a product-precursor relationship between the two proteins. The presence of BB94 did not alter the abundance of Δ122 GPR37L1-eYFP, which is expected because this form already lacks the N terminus.

Fig. 2 GPR37L1 is processed at the N terminus by an ADAM. GPR37L1-eYFP cells were induced with doxycycline (500 ng/ml).

(A) Cells were induced to express either WT or Δ122 GPR37L1-eYFP in the presence of BB94 (20 μM) for 24 hours and visualized by in-gel fluorescence. (B) GPR37L1-eYFP cells were induced with doxycycline (500 ng/ml) and increasing concentrations of BB94 (quantified in the right panel). (C to F) GPR37L1-eYFP cells were induced in the presence of increasing concentrations of human recombinant TIMP-1, TIMP-2, TIMP-3, or TIMP-4 for 24 hours and then immunoblotted with the GFP antibody. Biological activity of recombinant TIMPs was validated using activated MMP-2 and a fluorogenic substrate. AFU, arbitrary fluorescence units; MMP-2 act., MMP-2 activity. (G and H) GPR37L1-eYFP cells were induced in the presence of the indicated concentrations of TAPI-1 or TAPI-2 for 24 hours, and the amount of full-length and truncated species was quantified compared to the control cells. Left: In-gel fluorescence of cell lysates. Right: Quantification of full-length (~65 to 70 kD, open circle) and truncated (~40 to 45 kD, closed circle) GPR37L1 fluorescence. n = 4. Each image is representative of at least four individual experiments.

Tissue inhibitors of metalloproteases (TIMPs) are endogenous inhibitors of MMPs and ADAMs, and TIMPs have overlapping, but not identical, inhibition profiles for their substrates (26). TIMP-1 is an inhibitor of ADAM10 and shows weak inhibition of MMP-14, MMP-16, MMP-19, and MMP-24. TIMP-2 inhibits all MMPs and also ADAM12. TIMP-3 similarly inhibits all MMPs and also ADAM10, ADAM12, ADAM17, ADAM28, and ADAM33. TIMP-4 also displays a wide MMP inhibition profile with further inhibition at ADAM17, ADAM28, and ADAM33. GPR37L1-eYFP induced in the presence of four different TIMPs still produced both the full-length and short forms (Fig. 2, C to F). The lack of an effect was not due to the inactivity of recombinant TIMPs, because the same batch of each recombinant TIMP inhibited recombinant MMP-2 (Fig. 2, C to F). These data indicated that all MMPs, as well as ADAM10, ADAM12, ADAM17, ADAM28, and ADAM33, were unlikely mediators of GPR37L1 N-terminal cleavage.

ADAM and MMPs with a Zn2+-containing catalytic domain can be inhibited by nonspecific inhibitors that interact with the catalytic domain. These inhibitors include BB94 and tumor necrosis factor–α protease inhibitors 1 (TAPI-1) and 2 (TAPI-2) (27). Similar to the results with BB94 (Fig. 2B), the presence of TAPI-1 or TAPI-2 prevented GPR37L1-eYFP cleavage in a concentration-dependent manner (Fig. 2, G and H). From these data and those with TIMPs, we concluded that one or more of the ADAM family members that are not inhibited by TIMPs but have a Zn2+ catalytic domain, that is, ADAM8, ADAM9, ADAM15, ADAM19, ADAM20, ADAM21, and ADAM30 (28), may be responsible for GPR37L1 N-terminal proteolysis.

GPR37L1 signals constitutively through Gαs

To understand the biological relevance of the N-terminal processing of GPR37L1, we had to identify the signal transduction pathways activated by this receptor. We used a G protein–coupling screen using Saccharomyces cerevisiae with β-galactosidase as the output, which we have previously used to identify the G protein coupling of another orphan receptor, GPR35 (16). In this system, the candidate receptor is transformed into an engineered yeast strain with a modified pheromone-response pathway. Receptor activity resulting in G protein coupling leads to dissociation of the Gβγ protein, transducing the signal to a mitogen-activated protein kinase cascade, ultimately resulting in transcription of a β-galactosidase reporter gene controlled by the FUS1 promoter (29, 30).

In this system, GPR37L1-eYFP coupled to the chimeric G proteins Gpa1/Gαs (Fig. 3A) and Gpa1/Gα16, but not Gαq/11 or Gα14 chimeras (Fig. 3A). Moreover, contrary to another study (11), we did not observe any constitutive coupling to members of the Gαi/o family (Fig. 3A), nor did the receptor couple to either Gα12 or Gα13 (Fig. 3A). In contrast to the full-length receptor, Δ122 GPR37L1-eYFP did not result in significant coupling of the receptor to any of the tested G protein chimeras (Fig. 3A). Given that Gα16 (now known as Gα15 in humans) couples promiscuously to GPCRs that signal through the Gαs, Gαi/o, or Gαq/11, but not Gα12/13, G protein families (31, 32), these findings strongly suggest that GPR37L1 is a Gαs-coupled and, thus, adenylyl cyclase–coupled receptor.

Fig. 3 GPR37L1 N terminus is necessary for Gαs constitutive activity.

(A) WT GPR37L1-eYFP, Δ122 GPR37L1-eYFP, or empty vector was transformed into yeast strains bearing individual G protein chimeras, and β-galactosidase (β-gal) activity was measured in AFU after 24 hours. n = 4 to 6 individual transformants. Dotted line represents vector-transformed Gpa1 β-galactosidase activity for comparison. (B) Flp-In HEK293 cells were left either uninduced (−) or induced (+) with doxycycline (500 ng/ml) for 16 to 20 hours before being incubated for 30 min with cAMP AlphaScreen beads to assess activity. **P < 0.01 according to an unpaired t test with Welch’s correction for unequal variances. Forskolin (FSK) (1 μM) was used as a positive control. n = 6. (C) Transiently transfected HEK293 cells expressing either pcDNA3 (−) or GPR37L1 (+) and a CRE-luc reporter were incubated for 48 hours (2 hours for FSK) before analysis of luciferase relative light units (RLU). n = 4. (D) Effect of N-terminal truncation (Δ122) on GPR37L1 signaling. n = 7. (E) Increasing amounts of either WT or Δ122 GPR37L1 DNA were titrated to compare the effect on CRE-luc accumulation. n = 3. *P < 0.05 according to two-way analysis of variance (ANOVA) with post hoc analysis between rows (receptor). (F) Effect of inhibition of MMPs and ADAMs on WT GPR37L1 activity in the CRE-luc assay; 24 hours after transfection with WT GPR37L1 and CRE-luc, cells were incubated for a further 24 hours in the presence or absence of TAPI-1 or TAPI-2. n = 3 to 4. (G) Concentration-response curve for TAPI-2 effect on CRE-luc accumulation. n = 3. *P < 0.05, **P < 0.01, ***P < 0.001 according to one-way ANOVA with Bonferroni’s post hoc analysis, unless otherwise stated.

To confirm that GPR37L1 also couples to full-length mammalian Gαs, we examined the ability of GPR37L1 to stimulate cAMP (adenosine 3′,5′-monophosphate) production in stable Flp-In T-REx HEK293 cells expressing GPR37L1-eYFP. Comparison of the cAMP signal produced within 30 min in the presence of the phosphodiesterase inhibitor IBMX, in uninduced cells and GPR37L1-eYFP–expressing cells, showed that GPR37L1-eYFP–expressing cells produced more cAMP, although the amount of cAMP was much less than the maximum cAMP signal detected in uninduced cells exposed to the adenylyl cyclase activator forskolin (Fig. 3B). We selected 30 min to avoid toxicity associated with IBMX exposure. To examine GPR37L1-mediated signaling over longer time periods, we used HEK293 cells transiently transfected with both untagged GPR37L1 and a Gαs signaling pathway reporter construct comprising a cAMP response element fused to luciferase (CRE-luc) (30, 33) and measured luciferase activity 48 hours after transfection. In this assay, we found that GPR37L1 generated equivalent levels of luciferase to that observed with a single 2-hour exposure of vector-transfected cells to forskolin (Fig. 3C), indicating that GPR37L1 coupled to mammalian Gαs in the presence of Dulbecco’s modified Eagle medium supplemented with fetal calf serum. Furthermore, by using an eYFP-tagged version of GPR37L1 in the yeast assay and an untagged version in the CRE-luc experiments, we demonstrated that the eYFP fusion did not impair receptor signal transduction.

Because the detection of N-terminal peptides derived from GPR37L1 in CSF indicated that, like PARs, GPR37L1 is proteolytically processed, we examined whether short peptides corresponding to regions identified as proteolytic products of GPR37L1 in human CSF (1214) could activate the receptor. We call the three peptides CSF1 (residues 21 to 33), CSF2 (residues 47 to 65), and CSF3 (residues 67 to 74) (Fig. 1A and fig. S2A). Cleavage of the N terminus to release CSF1 (residues 21 to 33) would leave the N-terminal domain from residue 34 to the first transmembrane domain at residue 126. Within this remaining N-terminal section would be sequences that produce CSF2 (residues 47 to 65) and CSF3 (residues 67 to 74). Cleavage to release CSF2 would eliminate the N-terminal region to residue 68, just at the position where CSF3 commences. Finally, proteolytic cleavage of CSF3 would eliminate all three peptides from the N-terminal region. We also tested whether addition of the peptides corresponding to residues 34 to 46 (“tethered-1,” the N-terminal end up to the CSF2 sequence that would be produced by cleavage and release of CSF1) or residues 75 to 86 (“tethered-2,” an N-terminal peptide sequence of arbitrary length that would be revealed upon cleavage and release of CSF3) altered receptor activity. However, we found no effect of the CSF peptides or tethered peptides in either the yeast chimeric G protein assay (fig. S2, B and C) or the mammalian system when tested with wild-type GPR37L1-eYFP or Δ122 GPR37L1-eYFP (fig. S2).

The N terminus is necessary for GPR37L1 constitutive activity

Because the N-terminal truncation mutant Δ122 GPR37L1-eYFP was inactive in the yeast chimeric G protein assay, we examined the effect of N-terminal truncation on receptor signaling in human cells using the CRE-luc assay. In contrast to wild-type GPR37L1, Δ122 GPR37L1 did not stimulate expression of the cAMP-dependent reporter (Fig. 3D), supporting the hypothesis that the N terminus is necessary for constitutive GPR37L1 signal transduction. Furthermore, increasing amounts of only the wild-type GPR37L1–encoding plasmid DNA increased the CRE-luc signal, indicating that the absence of detectable Δ122 signaling was unlikely the result of differences in receptor expression (Fig. 2E). Finally, exposing the cells expressing GPR37L1 to either TAPI-1 or TAPI-2, ADAM and MMP inhibitors that prevented the N-terminal cleavage of the receptor (Fig. 2), significantly enhanced GPR37L1 activity (Fig. 3F), and the enhancing effect of TAPI-2 was concentration-dependent (Fig. 3G), confirming that GPR37L1 cleavage by a metalloprotease is sufficient to inactivate its constitutive activity. Together, these results suggested that GPR37L1 signaling is stimulated posttranslationally by its own N terminus.

GPR37L1 signaling is independent of TX14A

A study identified the peptide TX14A, which is also known as prosaptide and is a synthetic 14–amino acid neuroprotective peptide that is derived from PSAP, as a dual GPR37 and GPR37L1 agonist that induced receptor internalization and Gαi-mediated signal transduction (11). We tested the effect of TX14A on GPR37L1-mediated G protein coupling in the yeast chimeric G protein assay for Gαs, which we found coupled to GPR37L1 (Fig. 3A), and for Gαi1 and Gαi3, the G proteins reportedly stimulated by TX14A (11). GPR37L1-eYFP coupling to the Gαs chimeric protein was the same in the presence or absence of TX14A, and TX14A did not stimulate coupling to either of the Gαi chimeric subunits (Fig. 4A). Because the yeast assay uses chimeric G proteins and the eYFP fusion protein, we also examined the effect of TX14A on GPR37L1-mediated, Gαs-dependent activation of the CRE-luc reporter and on GPR37L1 coupling to Gαi signaling using the serum response element luciferase reporter (SRE-luc), both of which rely on the endogenous G proteins in HEK293 cells. Consistent with the yeast results, TX14A did not affect GPR37L1 coupling to Gαs to stimulate CRE-luc (Fig. 4B). Both in the presence or absence of TX14A, we detected a small but statistically insignificant induction of SRE-luc when coexpressed with GPR37L1, but not with Δ122 GPR37L1 (Fig. 4C), indicating that constitutive coupling may also occur with Gαi. To rule out the possibility that the purchased TX14A was spoiled or inactivated, we tested multiple batches of TX14A with the same outcome. These data indicated that the activity of wild-type GPR37L1 was not responsive to TX14A and that the absence of an effect of TX14A was not due to the constitutive nature of GPR37L1 obscuring the response.

Fig. 4 TX14A does not stimulate GPR37L1 signaling or alter protein abundance.

(A) β-Galactosidase activity in Gpa1 or Gαs, Gαi1, or Gαi3 yeast chimeric strains transformed to express either p426GPD vector or WT GPR37L1-eYFP and stimulated with 10 μM TX14A for 24 hours. n = 12 individual transformants per strain. (B and C) Effect of TX14A on GPR37L1 activity in the HEK293 cell CRE-luc or SRE-luc reporter assays for Gαs to increased cAMP signaling or Gαi to decreased cAMP signaling. Cells were transfected with pcDNA3 vector, WT or Δ122 GPR37L1, and the indicated reporter, and 24 hours later, exposed to TX14A for 24 hours. Data are normalized to positive controls: 1 μM FSK for CRE-luc or 20% fetal bovine serum supplemented with 1 μM phorbol myristate acetate for SRE-luc. n = 4. n.s., not significant. (D) The effect of TX14A or metalloprotease inhibition on the ratio of full-length to truncated GPR37L1. Cells were induced with doxycycline (500 ng/ml) for 24 hours to express WT GPR37L1-eYFP and then exposed to 10 μM TX14A, 10 μM TAPI-2, or 20 μM BB94 for 24 hours. Lysates were obtained and proteins were immunoblotted with the GFP antibody. A representative blot is shown on the left, and on the right is the densitometry of the relative GPR37L1 species as a percentage of total GPR37L1 (both species). n = 4. MW, molecular weight. (E) Analysis of TX14A binding to membranes from cells expressing WT or Δ122 GPR37L1-eYFP. Membrane preparations were immunoprecipitated by incubation with bTX14A conjugated to streptavidin beads. Input and immunoprecipitated samples were probed with an antibody for GFP. Representative blot, n = 3. (F) Analysis of TX14A binding to cerebellar membranes from WT or GPR37L1-null mice. Samples were immunoprecipitated by incubation with bTX14A conjugated to streptavidin beads and probed with an antibody raised against GPR37L1. *P < 0.05, ***P < 0.001 according to one-way ANOVA with Bonferroni’s post hoc analysis.

Given that the observed acute signaling effects of TX14A in the original paper were modest for a Gαi-coupled receptor (11), we reasoned that chronic treatment may reveal an effect at the level of protein abundance. Both TAPI-2 and BB94 significantly shifted the ratio of full-length to truncated GPR37L1-eYFP without altering the overall amount of the receptor, whereas TX14A treatment did not significantly change the ratio (Fig. 4D). Finally, in the absence of evidence for a signaling or regulatory role for TX14A, we aimed to confirm direct binding of TX14A to GPR37L1 using a biotinylated version of the peptide (bTX14A), as described in the original paper (11). Membranes from cells expressing either wild-type or Δ122 GPR37L1-eYFP were incubated with bTX14A before isolation with streptavidin-coated beads. However, no GPR37L1 was evident after pull-down (Fig. 4E) in these cells or in mouse cerebellum (Fig. 4F). Thus, we find no evidence that TX14A binds to GPR37L1.

Because PSAP, the endogenous ligand proposed for GPR37L1, is present in cell culture medium (11), we used an antibody that recognizes the saposin C domain of PSAP to deplete the protein from the assay medium as described previously (34). However, despite successfully immunoprecipitating PSAP from the culture medium (Fig. 5A), there was no specific effect of either PSAP depletion or TX14A stimulation on CRE-luc (Fig. 5B) or SRE-luc (Fig. 5C) reporter activity. Instead, we noted that the addition of the saposin C antibody had a nonspecific inhibitory effect on reporter activity that was evident even in the vector-transfected control cells. Finally, we titrated the GPR37L1-encoding plasmid DNA in the presence of constant CRE-luc and removed the serum from the medium 24 hours before performing the assay to limit the amount of PSAP present in the media. Even in the absence of serum, GPR37L1 coupled to CRE-luc, and Δ122 was inactive (Fig. 5D). Although it is possible that PSAP in the culture medium or produced by the cells may remain bound to the receptor, we think it is unlikely that endogenous PSAP masked the effect of TX14A or directly activated GPR37L1 in this system.

Fig. 5 Effect of PSAP depletion on GPR37L1 activity in HEK293 cells.

(A) Confirmation of depletion of PSAP from HEK293 cell–conditioned medium. Western blot shows the concentrated medium with PSAP pulled down with the saposin C antibody (αSapC) or the amount of cell-associated PSAP pulled down with the saposin C antibody. PSAP was detected with an antibody that recognizes the full-protein αPSAP (n = 3). IgG, immunoglobulin G. (B and C) Effect of PSAP depletion and TX14A on GPR37L1 activity in HEK293 cells as detected by the CRE-luc or SRE-luc reporter assays. Cells were transfected with GPR37L1 or empty vector and either CRE-luc or SRE-luc. After 24 hours, cells were treated for 24 hours with the saposin C antibody in the presence or absence of 10 μM TX14A. n = 4. *P < 0.05, ***P < 0.001 according to one-way ANOVA with Dunn’s correction for multiple comparisons. (D) Effect of serum starvation on GPR37L1 activity in HEK293 cells using the CRE-luc reporter assay. Cells were transfected with increasing amounts of either WT GPR37L1 or Δ122 GPR37L1 and serum-starved for 24 hours before assessment of CRE-luc activity. n = 3. *P < 0.05, **P < 0.01 according to two-way ANOVA with Sidak’s multiple comparison test.

GPR37L1 is proteolytically processed in vivo

Thus far, our evidence for the proteolytic processing of the N terminus has come from heterologously expressed GPR37L1-eYFP in Flp-In HEK293 cells. To confirm that this processing occurs in vivo and is physiologically relevant, we examined GPR37L1 abundance in the cerebellum from C57BL/6J wild-type and GPR37L1-null mice (Fig. 6A) (35). We chose cerebellum because GPR37L1 mRNA and protein are most abundant in this tissue and global GPR37L1-knockout mice have a cerebellar phenotype (79). Furthermore, there is little difference in sequence between orthologs (fig. S3A), with mouse and rat sharing 91% identity and 100% similarity with human GPR37L1. In the cerebellum of 12-week-old male mice, we found that GPR37L1 existed almost entirely as the smaller–molecular weight species in this tissue, with very little full-length species detected (Fig. 6A). No GPR37L1 was identified in membranes prepared from the GPR37L1-null mice, confirming that the antibody used to detect GPR37L1 was specific for the receptor. Although the tissue was snap-frozen upon harvest and processed on ice in the presence of a protease inhibitor cocktail, it was possible that the marked proteolytic processing observed was the result of proteases that were released during tissue lysis. Thus, we prepared a separate set of cerebellar tissues in the constant presence of BB94 but found no difference in GPR37L1 processing (fig. S3B). The band detected in the cerebellar lysates from the cerebella of the null mice (fig. S3B) was not present in the cerebellar plasma membrane samples (Fig. 6A), indicating that this is a nonspecific band. Thus, these data indicated that, in the mouse cerebellum, GPR37L1 exists almost entirely in the cleaved form, suggesting that receptor signaling is largely silenced in this tissue.

Fig. 6 GPR37L1 is N-terminally proteolytically processed in vivo.

(A) Plasma membrane was isolated from the cerebella from adult GPR37L1-null or C57BL/6J mice, and samples were probed for endogenous GPR37L1. n = 3 twelve-week-old male mice per genotype. (B) Mouse organotypic cerebellar slice preparations from either WT C57BL/6J or GPR37L1-null mice were immediately exposed to 20 or 40 μM BB94 for 24 hours and then harvested for membrane preparation and Western blotting. Two representative samples of four biological replicates are shown. Equal protein was loaded as determined by protein concentrations calculated with the Merck Direct Detect. (C) cAMP accumulation in cerebellar slices from C57BL/6J (WT) or GPR37L1-null mice cultured for 1 hour in the presence or absence of 1 mM IBMX. n = 6, where each sample contains tissue from two to four separate mice. *P < 0.05.

To assess cAMP generation and ADAM- or MMP-dependent cleavage in an endogenous setting, we established mouse cerebellar organotypic slice preparations. We found that the abundance of GPR37L1 was markedly reduced in the long-term cultures (fig. S3C; note the reduction in protein abundance within 24 hours). Immediately after isolation, the cultured tissue had both the full-length and the shorter form of GPR37L1 (Fig. 6B). We placed the cerebellar preparations immediately into BB94-containing medium for 24 hours and detected full-length and processed GPR37L1 by Western blotting. After 24 hours in the absence of BB94, full-length GPR37L1 was almost undetectable; however, the full-length form was detectable when the cultures were maintained in BB94 (Fig. 6B). Because of the loss of GPR37L1 within the 24 hours (fig. S3C) needed for the BB94 treatment, the Western blot in Fig. 6B was exposed for a relatively long amount of time, which makes the full-length species more apparent and overrepresents the amount of the full-length form at time zero, because the signal of the truncated form saturates. We observed a similar pattern in rat organotypic cerebellar slices at 48 hours, although the amount of GPR37L1 was very low and at the limit of detection (fig. S3D). Together, these results showed that a metalloprotease inhibitor reduced the processing of GPR37L1 in both rat and mouse cerebellum.

Finally, we examined the functional consequence of GPR37L1 deletion in cerebellar slices by measuring cAMP concentrations either in the absence or in the presence of phosphodiesterase inhibition with IBMX. Although baseline cAMP accumulation was the same between C57BL/6J and GPR37L1-null mice, incubation of the slices with IBMX for 1 hour revealed that deletion of GPR37L1 resulted in significantly less cAMP production in the cerebellum (Fig. 6C). Thus, despite the observation that GPR37L1 exists predominantly as a processed, and presumably inactive, species, mice lacking GPR37L1 have less cAMP production in the cerebellum. Thus, the remaining full-length GPR37L1 appears sufficient to account for the difference that we observe in cAMP accumulation in vivo.


The key findings of our study are that the GPCR GPR37L1 is a constitutively active Gαs-coupled receptor when expressed in cell culture and that without its N terminus, the protein is inactive. Furthermore, we found that N-terminal proteolysis by a metalloprotease occurred both in cultured cells and in rodent cerebellum. Intrigued by the observation that three different fragments of the GPR37L1 N terminus have been identified in human CSF (1214), we posited that GPR37L1 may function in a manner akin to the PARs, which have a tethered agonist that is only revealed upon proteolytic cleavage of the receptor (36). Instead, we found the opposite—GPR37L1 expressed in HEK293 cells or in a yeast assay was constitutively active in the presence of its N terminus, and proteolytic processing of the receptor leads to its inactivation. Thus, we have identified a previously unknown mechanism of signaling for this receptor, which is critical for cerebellar development (7) and implicated in cardiovascular disease (35).

Using chimeric yeast Gpa1 fused to the last five amino acids of each human Gα subunit, we provided evidence for GPR37L1 constitutive activity and showed that this receptor coupled to Gαs. This may explain the preliminary report of a deleterious effect of adenoviral-mediated GPR37L1 overexpression in cardiomyocytes (35), because excessive stimulation of the Gαs/adenylyl cyclase/cAMP pathway by β-adrenergic receptors decreases cardiomyocyte viability, potentially as a result of cAMP-mediated calcium overload (37). Although we also found that GPR37L1 coupled to the Gα16 chimera (now recognized as Gα15), this is most likely due to the propensity of Gα15/16 to promiscuously couple to GPCRs linked to Gαs, Gαi/o, or Gαq/11, but not Gα12/13 (31, 32). Thus, we interpret the Gα15/16 signal as further support of Gαs coupling, because we did not observe GPR37L1 signaling through either Gαi/o or Gαq/11 family members (and to our knowledge, there are no examples of GPCRs that couple exclusively to Gα15/16). Furthermore, GPR37L1 abundance does not correlate with that of Gα15/16; the former is abundant in the CNS (8), and the latter is abundant in stem cells and thymic epithelial cells (31) and has a low abundance in the brain (38).

Our finding of Gαs, but not Gαi/o, coupling to GPR37L1 contrasts with a report that identified the neuropeptides TX14A and PSAP as agonists for both GPR37 and GPR37L1 (11). Using HEK293T cells transiently transfected with Flag-tagged GPR37L1, the authors reported a 5% increase of Gαi/o-mediated [35S]guanosine 5′-O-(3′-thio)triphosphate accumulation in response to TX14A [our previous studies with Gαi/o-coupled receptors using this assay typically resulted in a 100% increase above baseline with ligand stimulation (17, 19, 20)] and 50% amplification of phosphorylated extracellular signal–regulated kinases 1 and 2 in primary astrocytes (11). In our study, we failed to substantiate functional coupling of either full-length or Δ122 GPR37L1-eYFP to any Gαi/o family yeast chimeras (Gαi1/2, Gαi3, or Gαo), although this assay works with other Gαi/o-coupled receptors (16, 39). Additionally, GPCRs that fail to couple to Gα15/16 tend to be Gαi/o-selective (32), yet we also observed constitutive signaling through the Gα15/16 chimera. In the SRE-luc reporter assay, we found a statistically insignificant coupling of untagged GPR37L1, but not Δ122 GPR37L1, to Gαi/o signal transduction; however, we failed to demonstrate TX14A signaling in either yeast or mammalian assays. Our efforts to demonstrate TX14A or PSAP activity at GPR37L1 have been extensive: we have tested multiple batches of TX14A; we used several isolation conditions to test bTX14A binding; we have tested the effects of PSAP-depleting antibodies; and we have grown cells in low serum conditions to reduce any PSAP secreted into the culture medium. PSAP depletion reduced the signal in the CRE-luc assay; however, this effect occurred in the vector-transfected conditions and so did not involve GPR37L1. Similarly, PSAP depletion of vector-transfected cells in the SRE-luc assay significantly reduced the signal, indicating that PSAP depletion decreased the reporter signal independently of GPR37L1. If TX14A had functioned as the ligand in the absence of PSAP, then the GPR37L1-dependent signal should have been restored by adding TX14A, and it was not. Thus, we did not find any evidence for PSAP or TX14A as agonists for GPR37L1.

In a large β-arrestin recruitment screen for endogenous and surrogate ligands at orphan GPCRs, including GPR37L1, PSAP was not detected as a GPR37 or GPR37L1 “hit” (40). The discrepancy between Meyer et al.’s and our results (and the β-arrestin screen) is difficult to reconcile, but one potential explanation is that we are measuring constitutive coupling (that is, signaling in the absence of ligand), whereas modest Gαi/o activity was only reported after TX14A or PSAP stimulation (11), suggesting that GPR37L1 may display signal bias whereby constitutive activity and ligand-mediated signaling occur through two separate pathways. This is not without precedent (41).

GPR37L1 is not alone in having its N terminus processed by a metalloprotease; other examples include the endothelin-B (24) and β1-adrenergic (25) receptors, for which N-terminal cleavage influences cell surface abundance but not activity. In contrast, PARs exemplify proteolytic processing that leads to receptor activation. PARs are cleaved at their N terminus by serine proteases, revealing a tethered ligand sequence on the newly truncated N terminus that serves to activate the receptor. In some cases, PARs are cleaved by MMPs, which results in aberrant receptor activation in the case of PAR1 being cleaved by MMP-1, promoting tumorigenesis in breast carcinoma cells (23). Moreover, multiple members of the adhesion GPCR family contain cryptic tethered agonists within their proximal N-terminal domains (4244). For this reason, we tested whether GPR37L1 was activated by a tethered ligand as observed for the PARs, but we did not find any evidence for such a mechanism of activation by N-terminal peptides for either the full-length or truncated versions of GPR37L1. Instead, the truncated version lacked activity, indicating that proteolysis inactivated the receptor. This is perhaps unsurprising, because the metalloprotease-mediated processing of GPR37L1 that occurred in cultured cells corresponds to the removal of the entire N terminus, whereas the N-terminal peptides found in human CSF are fragments upstream of this cleavage site (1214). Instead, the human CSF peptides may be indicative of proteolytic breakdown of the released GPR37L1 N terminus, after its metalloprotease-mediated cleavage. This is supported by the activity of prolyl carboxypeptidase on what we have termed CSF3—for prolyl carboxypeptidase to have biological activity at CSF3, a peptide one residue longer than CSF3 must have existed (14). The biological function of the GPR37L1 peptides found in CSF, therefore, remains unclear—it is possible they represent processing to form active ligands for other receptors or are merely a by-product of a degradation process.

In contrast to the PARs, we found that GPR37L1 is constitutively active and that N-terminal proteolysis therefore inactivated receptor signaling. Such a role for the N terminus is reminiscent of the melanocortin-4 receptor, a unique case of constitutive activity resulting from autoagonism by the receptor’s own N terminus (45). In a fascinating example of negative feedback, an endogenous inverse agonist regulates this constitutive activity by antagonizing the N-terminal autoagonism (4547). Thus, it is tempting to speculate that GPR37L1 has both an endogenous agonist and an inverse agonist that remain to be identified, or perhaps the metalloprotease shedding of the GPR37L1 N terminus provides sufficient negative feedback that an inverse agonist is unnecessary. Of course, it is also possible that GPR37L1 simply does not have an endogenous ligand; such a concept will be difficult to prove, however, because absence of evidence is not itself an evidence of absence.

By combining multiple processes of elimination, we have narrowed down both the likely mediators of GPR37L1 proteolysis and the N-terminal cleavage site. Few specific ADAM or MMP inhibitors exist; thus, we took advantage of the overlapping, but not identical, inhibition profiles of various small-molecule antagonists and TIMPs to identify ADAM8, ADAM9, ADAM15, ADAM19, ADAM20, ADAM21, and ADAM30 as possible mediators of GPR37L1 N-terminal truncation. Of these proteases, ADAM19 is the most attractive candidate, because, like GPR37L1, it too is present in the heart (48, 49), and although far more severe than the GPR37L1−/− cardiovascular phenotype (35), ADAM19−/− mice develop cardiac defects (50, 51). Furthermore, Wei et al. report that ADAM19 mRNA in the human brain is highest in the cerebellum (48), and precocious development of the cerebellum occurs in GPR37L1−/− mice (7). Given the lack of specific pharmacological inhibitors, it is likely that genetic tools will be required to confirm the involvement of individual ADAMs in GPR37L1 N-terminal proteolysis. Likewise, further study is required to identify the specific cleavage site within the N terminus of GPR37L1, an undertaking complicated by the lack of consensus sequence for MMP or ADAM cleavage (52). Metalloproteases have evolved to have their substrate specificity regulated at a number of levels, with amino acid substitutions within the region of cleavage tolerated in many instances and at a variable proximity to the cleaved residues (53). This broad specificity at the level of amino acid sequence is compensated for by the existence of exosites—regions outside the catalytic domain that provide the metalloprotease with critical substrate recognition (54). At the very least, the N-terminal deglycosylation data point to proteolysis occurring between Asn105 and the start of transmembrane domain 1. Most likely, more than one ADAM is responsible for N-terminal processing of GPR37L1.

An exciting finding of this study is that N-terminal processing occurs in vivo and in cerebellar cultures. In heterologous expression systems, we observed the larger–molecular weight GPR37L1 species to be greater or equal in proportion to the smaller, cleaved species. However, our data suggested that, in vivo, the proteolytically processed GPR37L1 predominates. Given our finding that GPR37L1 lacking the N terminus was inactive, the abundance of the proteolyzed form suggests that GPR37L1 exists largely in an inactivated form. However, some signaling-competent receptor must be present because GPR37L1-null mice exhibited reduced cAMP accumulation compared to their wild-type counterparts. Furthermore, changes to extracellular metalloprotease activity would be expected to “tune” basal GPR37L1 signaling by posttranslational modification. Future studies will need to examine the role of the individual GPR37L1 protein species in the entire animal to understand the physiological importance of such proteolysis. For example, it would be interesting to see whether transgenic mice expressing only Δ122 GPR37L1 recapitulate the precocious cerebellar developmental and motor phenotype reported for GPR37L1-knockout mice (7).

In summary, we provide evidence that GPR37L1 is subject to N-terminal proteolysis and that this blunts the constitutive Gαs coupling of the receptor. For many years, orphan GPCRs have been considered an exciting, albeit untapped, resource for new therapeutic targets (55), yet an understanding of the physiology of these receptors or their involvement in disease has been an intractable problem in the absence of a ligand. Here, we show that knowing the ligand or possibly even the presence of a ligand is not always necessary for investigating the basic function of a GPCR. Instead, our data and methods revealed that proteolysis may be a key mechanism to inactivate a GPCR that may have constitutive activity in vivo, because either it does not require a ligand or its ligand is always present and bound to the receptor.


Reagents and constructs

All reagents were from Sigma-Aldrich or Life Technologies (including Invitrogen), unless otherwise stated. A previously reported (11, 56) human GPR37L1 was purchased from Multispan; however, this construct lacked the endogenous GPR37L1 signal peptide and instead had an optimized signal peptide and N-terminal Flag tag, so we reintroduced the original sequence with a Hind III restriction site and Kozak sequence by polymerase chain reaction. All constructs in pcDNA5/FRT/TO or pcDNA3 contain human GPR37L1 with or without eYFP fused to its C terminus; human GPR37L1-eYFP (wild type and Δ122) was cloned into p426GPD for S. cerevisiae experiments. Constructs were verified by DNA sequencing. Epitope tagging of GPR37L1 at the N terminus and expression in Flp-In HEK293 cells resulted in the appearance of large vacuoles and subsequent cell death. N-terminal CSF and “tethered” peptides were synthesized by GenScript.

Cell culture and transfection

Stable Flp-In T-REx cells were generated as previously described (18). For all Flp-In assays, cells were cultured in 10% fetal calf serum on poly-d-lysine–coated (0.1 mg/ml) tissue culture dishes for 24 hours before induction with doxycycline. Transient HEK293 cell reporter assays for either CRE- or SRE-luc were performed as previously described (30). Unless otherwise stated, all HEK293 experiments were performed in the presence of 10% fetal calf serum except the SRE-luc assays, where serum was withdrawn 24 hours before assay termination.

Rodent cerebellar tissue harvest and organotypic brain slice cultures

All animal tissue harvests were performed according to the Australian Code for the Care and Use of Animals for Scientific Purposes, Eighth Edition (2013), and were approved by either the Garvan Institute of Medical Research/St Vincent’s Hospital Animal Ethics Committee project 13/30 (mouse) or the University of New South Wales Animal Care and Ethics Committee project 15/12B (rat). C57BL/6J wild-type, GPR37L1-null, or cardiac-specific overexpressing transgenic (cTg) male mice between 4 and 6 months of age were euthanized by cervical dislocation for cerebellar tissue extraction. Cerebella were isolated and snap-frozen in liquid nitrogen for −80°C storage until required. Organotypic cerebellar slice cultures were prepared essentially as described for the hippocampus (57, 58). Briefly, 15-day-old mixed-sex, C57BL/6J wild-type or GPR37L1-null mice or 8- to 9-day-old mixed-sex Sprague Dawley rats were euthanized by decapitation, and brains were removed and placed in ice-cold Hanks’ balanced salt solution (HBSS) supplemented with 30 mM d-glucose and penicillin/streptomycin (100 U/ml) and saturated with carbogen (95% oxygen, 5% carbon dioxide). The cerebella were removed and sliced into 400-μm sagittal sections using a McIlwain tissue chopper and then plated evenly across two Millicell cell culture inserts (Merck) in six-well plates (two inserts per cerebellum, six to eight slices per insert). All rodent cultures were maintained in culture media containing 50% minimum essential medium, 25% HBSS, 25% heat-inactivated horse serum, 36 mM d-glucose, 25 mM Hepes, 4 mM NaHCO3, 2 mM GlutaMAX, 1% Fungizone, and penicillin/streptomycin (100 U/ml). Rat cerebellum cultures were maintained at 37°C in an atmosphere of 5% CO2, and culture media were changed every 3 days. After 10 days in culture, a single well of each matched pair was treated by an addition of 20 μM BB94 (batimastat) to culture media (the other well in each pair was left untreated). Forty-eight hours after treatment, slices were collected and homogenized in ice-cold RIPA (radioimmunoprecipitation assay) buffer, as described below. Mouse cerebellar cultures were grown in identical conditions but were harvested after 1 day in culture, with or without 20 or 40 μM BB94. Data were obtained from at least three independent litters of animals.

In-gel fluorescence and immunoblot studies

After GPR37L1 induction by doxycycline for 24 hours, cells were placed on ice, washed with ice-cold phosphate-buffered saline, and harvested with ice-cold RIPA buffer [50 mM Hepes, 150 mM NaCl, 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS, 10 mM NaF, 5 mM EDTA, 10 mM NaH2PO4, 5% ethylene glycol, and cOmplete EDTA-free Protease Inhibitor Cocktail Tablet (pH 7.4) (Roche Diagnostics)]. Cell lysates were cleared of insoluble debris by centrifugation, and the supernatant was mixed with Laemmli buffer [final concentration was as follows: 63 mM tris, 50 mM dithiothreitol, 80 mM SDS, 10% glycerol, 0.004% bromophenol blue (pH 6.8)], followed by 15-min incubation in a 37°C water bath. For deglycosylation experiments, cleared lysates were treated with either 2 U of peptide N-glycosidase F or 100 mU of endoglycosidase H, or both, for 2 hours in a 37°C water bath before addition of Laemmli buffer. Cell surface biotinylation was performed as previously described (59). Equal amounts of protein were then separated by 4 to 12% bis-tris SDS-PAGE in NuPAGE MOPS SDS Running Buffer (Life Technologies) either at room temperature (for Western blot) or at 4°C in ice-cold buffer [for in-gel fluorescence, imaged with a FLA-5100 (Fujifilm)]. For immunoblotting, proteins were transferred to a 0.45-μm pore size polyvinylidene fluoride transfer membrane (Merck Millipore), and nonspecific binding was blocked in tris-buffered saline/0.1% Tween 20 with 5% skim milk powder. GPR37L1-eYFP was detected with an antibody against GFP (1:10,000 dilution; Abcam) and IgG anti-rabbit horseradish peroxidase–conjugated secondary antibody (1:10,000; GE Healthcare UK). For whole cerebellum and cerebellar slices, snap-frozen tissue was prepared using a POLYTRON homogenizer in the presence of RIPA buffer, and insoluble debris was removed as described above. Crude membranes were isolated as previously described (19). Endogenous GPR37L1 was detected by goat anti-GPR37L1 (C-12) antibody (1:1000; sc-164532, Santa Cruz Biotechnology). Chemiluminescence was detected with Western Lightning ECL reagent (PerkinElmer).

Inhibitor studies and verification of recombinant TIMP activity

Metalloprotease inhibitors, including BB94 (batimastat), TAPI-1 and TAPI-2 (Santa Cruz Biotechnology), and TIMP1, TIMP2, TIMP3, and TIMP4 (R&D Systems), were added at the time of GPR37L1 induction or 24 hours after transfection, and cells were then incubated for 24 hours before lysis. For MG132 proteasome inhibition, cells were induced with doxycycline overnight and then treated with MG132 for 6 hours before harvest. Verification of TIMP activity in vitro using activated MMP-2 and the fluorogenic peptide substrate MCA-Pro-Leu-Gly-Leu-DPA-Ala-Arg-NH2 was performed as per the protocol provided by R&D Systems.

Assessment of G protein coupling using S. cerevisiae G protein chimeras

p426GPD (high-copy vector), p426GPD–GPR37L1-eYFP, or p426GPD–Δ122 GPR37L1-eYFP was transformed into 11 individual FUS1-regulated β-galactosidase yeast reporter strains that had been modified to express either the endogenous Gα protein Gpa1 or chimeras comprising Gpa1/Gα fusions containing the last five amino acids of each human G protein (29). Individual S. cerevisiae transformants were then screened for constitutive coupling through yeast chimeras using the same protocol as previously described (30). Fluorescence was detected using a BMG PHERAstar FS (excitation/emission, 485/520, in AFU) (BMG).

cAMP, cAMP response element, and serum response element reporter assays

The accumulation of cAMP was measured in Flp-In HEK293 cells using the cAMP AlphaScreen assay according to the manufacturer’s instructions (PerkinElmer). CRE-luc and SRE-luc assays were performed as previously described (30) using the Dual-Luciferase Reporter Assay System (Promega). For DNA titration experiments, a total of 0.1 μg of DNA was added per 96 wells, with 0.02 μg of CRE-luc constant and empty pcDNA vector varied to balance total DNA. Under serum-free conditions (SRE-luc or serum-free CRE-luc assay), media were changed on cells the day after transfection, and cells were harvested 24 hours later. Samples were processed as per manufacturer’s instructions, and luciferase activity was detected using a BMG PHERAstar FS. To assess the cAMP levels in mouse cerebellar slices, tissue was harvested as described above, except that slices were incubated for exactly 1 hour in the presence or absence of the phosphodiesterase inhibitor IBMX (1 mM). To ensure that there was sufficient tissue for the assay, slices from two to four separate mixed-sex pups were pooled and considered to represent n = 1. cAMP was measured using the DetectX Direct cAMP ELISA (enzyme-linked immunosorbent assay) kit (Arbor Assays), as per manufacturer’s instructions. For PSAP depletion studies, medium was changed to serum-free medium containing either antibody (40 μg/ml) recognizing saposin C (sc-32875, Santa Cruz) or rabbit IgG (Santa Cruz) for 2.5 hours and incubated with protein A to pull down the antibody-associated PSAP. Then, the cells were lysed, and both the lysates and the protein A–concentrated medium sample (supernatant) were analyzed by Western blotting with an antibody recognizing full-length PSAP (1:1000; ab68466, Abcam).

Data analysis

All data analysis was performed with GraphPad Prism 6 (GraphPad Software Inc.). Graphs show means ± SEM, unless otherwise stated. Where variances were deemed unequal according to a Brown-Forsythe test, and therefore violating the assumptions of ANOVA, data were transformed to logarithmic values before one-way ANOVA analysis. In-gel fluorescence densitometry was performed using ImageJ software (National Institutes of Health) within the linear range, and immunoblots were processed without altering relative intensities of the bands.


Fig. S1. Partial truncation of the GPR37L1 N terminus prevents receptor expression in HEK293 cells.

Fig. S2. GPR37L1 screen in S. cerevisiae with chimeric G proteins and in HEK293 cells by CRE-luc assay.

Fig. S3. Comparison of mouse and rat GPR37L1 proteolysis in cerebellar organotypic cultures.


Acknowledgments: We thank R. Hall for helpful discussions regarding prosaptide/prosaposin experiments, A. Durrell for technical assistance with rodent brain slices, and A. Finch for advice. Yeast chimeras were provided by S. Dowell from GlaxoSmithKline under a material transfer agreement (MTA). Funding: This work was funded in part by National Health and Medical Research Council of Australia (NHMRC) Program Grants 573732 and 1074386 (R.M.G.), an NHMRC and National Heart Foundation of Australia C.J. Martin Fellowship (N.J.S.), Australian Postgraduate Awards (J.L.J.C. and T.N.), and a Simon and Michal Wilkenfeld Scholarship (J.L.J.C.). Author contributions: N.J.S. designed the study, supervised the project, performed the experiments, and wrote the manuscript. J.L.J.C. and T.N. performed the experiments and wrote the manuscript. J.S., N.M., C.K.L., and N.M.J. performed the experiments or generated the constructs. R.M.G. co-supervised the project and wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Yeast chimeras are available from GlaxoSmithKline under an MTA. All other materials are available from the authors upon request.
View Abstract

Stay Connected to Science Signaling

Navigate This Article