AdipoR1 and AdipoR2 are newly discovered members of the huge family of seven-transmembrane receptors, but both receptors are structurally and functionally different from G-protein-coupled receptors. Little is known about the oligomerization of the AdipoRs. Here, we show the presence of endogenous AdipoR1 dimers in various cell lines and human muscle tissue. To directly follow and localize the dimerization, we applied bimolecular fluorescence complementation (BiFC) in combination with flow cytometry. We visualized and quantified AdipoR1 homodimers in HEK293 cells. Moreover, we identified a GxxxG dimerization motif in the fifth transmembrane domain of the AdipoR1. By mutating both glycine residues to phenylalanine or glutamic acid, we were able to modulate the dimerization of AdipoR1, implicating a role for the GxxxG motif in AdipoR1 dimerization. Furthermore, we tested whether the AdipoR1 ligand adiponectin had any influence on receptor dimerization. Interestingly, we found that adiponectin decreases the receptor dimerization in a concentration-dependent manner. This effect is mainly mediated by segments of the collagen-like domain of full-length adiponectin. Accordingly, this is the first direct read-out signal of adiponectin at the AdipoR1 receptor, which revealed the involvement of specific amino acids of both the receptor and the ligand to modulate the quaternary structure of the AdipoR1.
Adiponectin is a protein hormone that is secreted mainly by adipocytes. It is involved in a variety of biological processes, such as regulation of lipid and glucose metabolism (Berg et al., 2002; Kadowaki et al., 2006). It has also been shown to act as a vasoprotective and anti-inflammatory adipocytokine (Fasshauer et al., 2004; Tilg and Moschen, 2006). Recently, two seven-transmembrane (7TM) receptors, termed adiponectin receptor (AdipoR) 1 and 2, have been cloned and identified to bind adiponectin (Yamauchi et al., 2003). Both receptors share approximately 67% sequence identity. We, and others, have previously shown that both receptors are structurally distinct from most other 7TM proteins, because of their extracellular located C-terminus and cytosolic N-terminus (Deckert et al., 2006; Yamauchi et al., 2003). Accordingly, AdipoR1 and AdipoR2 do not use G-proteins for signaling and neither the mechanism of signal transduction nor the signaling pathways are understood (Yamauchi et al., 2003). Recent work revealed that most of the metabolic functions of adiponectin in the skeletal muscle and the liver are mediated by an increased AMP-activated protein kinase (AMPK) activity and by enhanced expression of peroxisome proliferator-activated receptor (PPAR)-α (Yamauchi et al., 2002; Yamauchi et al., 2007; Yoon et al., 2006). Very recently, we showed that the β-subunit of casein kinase (CK) II interacts with the AdipoR1 N-terminus and contributes to the intracellular-signaling cascade of the AdipoR1 (Heiker et al., 2009). Other proteins such as the adaptor protein, phosphotyrosine interaction, PH domain and leucine zipper containing (APPL) 1 and the receptor for activated protein kinase C1 (RACK1) could also be identified as adaptor proteins that interact with the AdipoR1 (Mao et al., 2006; Xu et al., 2009). Up to now, little information has been available on the quaternary structure of AdipoRs. Previous studies that used immunoprecipitation of ectopically expressed AdipoRs suggested that the receptors form oligomeric complexes (Yamauchi et al., 2003). Receptor dimerization has been described to have a role in the function of several members of the 7TM receptor family, including the GABAB receptor (Margeta-Mitrovic et al., 2000), the melanocortin receptor (Mandrika et al., 2005) or the dopamine D2 receptor (Zawarynski et al., 1998). There is a growing body of evidence suggesting that receptor dimerization has a crucial role during biosynthesis, trafficking and signaling of 7TM receptors (Milligan, 2008; Terrillon and Bouvier, 2004). Since disturbed dimerization and ineffective receptor-surface delivery has been connected with several diseases (Milligan, 2008), it is of great interest to study the oligomerization behavior of AdipoRs as potential pharmacological targets in obesity-associated disorders. In the present study, we address the question of whether AdipoRs dimerize in vitro and in vivo. Western blotting of protein lysates of various cell lines and human tissue revealed the presence of endogenous AdipoR1 homodimers. To study AdipoR1 dimerization in living cells, we used bimolecular fluorescence complementation (BiFC) in combination with fluorescence microscopy and flow cytometry. We show that AdipoR1 dimers are indeed present in human tissue, as well as in living cells. Next, we screened for hypothetical dimerization motifs and identified a glycophorin motif in transmembrane helix 5. Our study provides the first evidence that this GxxxG motif contributes to the AdipoR1 dimer formation. Moreover, we observe a ligand-induced inhibition of the receptor dimerization upon treatment with adiponectin, and we identified a segment within the collagen-like domain of adiponectin that mediates dimerization inhibition.
AdipoR1 forms dimers in different cell lines and human tissue
In contrast to other cell-surface receptors, for which dimerization has been proven to have an important role in receptor expression and/or signaling, little is known about the oligomerization of AdipoR1. Accordingly, we examined the oligomerization of endogenously expressed AdipoR1 in a panel of human cell lines (HEK293, MCF-7, SK-N-MC, HUVEC) as well as in human femoral muscle tissue. Following cell lysis and reducing SDS-PAGE of total protein extracts, we could identify AdipoR1 by western blotting in all analyzed cell lines (Fig. 1A, lower panel), which suggested that the AdipoR1 protein is widely expressed among human cells from a range of different tissues. Additionally, we showed that the AdipoR1 protein is strongly expressed in human femoral muscle tissue (Fig. 1A). We confirmed the specificity of the primary antibody by using a blocking-epitope peptide, which almost eliminated the AdipoR1 protein band (Fig. 1A). However, in addition to the monomeric receptor, which migrated in SDS-PAGE with a molecular mass of about 45-50 kDa, we found a second faint but distinct protein band at 80-85 kDa, which also disappeared upon treatment with blocking-epitope peptide (Fig. 1A, upper panel). This indicates that even under reducing conditions, at least part of the cellular AdipoR1 protein exists as dimer or possibly also higher molecular mass forms. As we found receptor monomers and dimers in human cells and tissue, we were interested to know whether they were also present in a cell line from a different species. Thus, we analyzed total cell extracts of C2C12 (murine myoblast) cells for AdipoR1 expression. Concomitant with our results described above, we found the AdipoR1 in its monomeric and dimeric form (Fig. 1B). Interestingly, western blotting of murine AdipoR1 showed a slightly lower molecular mass (ca. 42 kDa) compared with human AdipoR1 (ca. 45-50 kDa), which might reflect different post-translational modifications. Furthermore, we investigated the dimerization capacity of the ectopically expressed AdipoR1 by carrying out western blot analysis of MCF-7 cells transiently transfected with an HA-tagged AdipoR1 and a receptor fused to eYFP, respectively (Fig. 1C). For both transiently expressed receptor forms, we identified monomeric and dimeric species (Fig. 1C). Compared with the HA-tagged AdipoR1, receptors fused to eYFP showed a higher molecular weight of about 65 kDa, which was due to the mass shift of the eYFP fusion. Dimer bands could be observed at 130 kDa. Since we observed only protein bands resembling the predicted molecular mass of the ectopically expressed AdipoR1, it is rather unlikely that a protein of similar size as the endogenous AdipoR1 interacts with the receptor imitating an AdipoR1 dimer. Taken together, our results suggest that the endogenous and the transiently transfected AdipoR1 are present in monomers as well as dimers in cell lines of different species and in human tissue.
AdipoR1 dimerization is found in living cells
To exclude the possibility that AdipoR1 dimers were formed during protein isolation, we studied dimerization in living cells. We used HEK293 cells as a reliable system, because these cells possess a large number of endogenous AdipoR1 dimers (Fig. 1A). BiFC provides an advanced and sophisticated tool to study protein-protein interactions, such as receptor dimerization, directly in living cellular systems (Hu et al., 2002). This approach is based on the complementation of two non-fluorescent reporter molecule fragments, which are able to reconstitute a native and fluorescent reporter molecule. Here, we used the non-fluorescent N-terminal (VN) and C-terminal (VC) fragments of the Venus protein, an improved mutant of YFP (Nagai et al., 2002), which we fused N-terminally to the AdipoR1 (AdipoR1-VN and AdipoR1-VC, respectively). If both Venus fragments come in close spatial proximity triggered by receptor dimerization, the native Venus protein will be formed, leading to a detectable fluorescence (Fig. 2A). Fluorescence microscopy analysis of HEK293 cells co-transfected with plasmids carrying the cDNA of the respective fusion proteins, revealed a strong fluorescent signal. This suggests that AdipoR1s also dimerize in living cells (Fig. 2B, upper panel), confirming our results from the western blot system. We observed fluorescent signals for the receptor dimerization at the cell surface, as well as in the cytoplasm. As expression controls, we immunostained fixed cells with antibodies specific for HA- and FLAG-tag epitopes.
To assess BiFC assay specificity, we co-transfected HEK293 cells with AdipoR1 and the human NPY Y2 (hY2) receptor, each construct bearing a complementary Venus fragment. Because the hY2 receptor possesses an inverse orientation in the plasma membrane compared with the AdipoR1, we fused the complementary Venus fragment C-terminally to the hY2 receptor, which ensures that the Venus fragment is located within the cytoplasm. Cell surface expression of the hY2 receptor in the transfected cells was confirmed by NPY ligand binding assays (data not shown). As depicted in Fig. 2B, co-expression of AdipoR1 and the hY2 receptor led only to a low background fluorescence, which proved that the BiFC systems will only provide significant fluorescent signals in case of a genuine interaction. Similar results were obtained for transfections with single BiFC plasmids (AdipoR1-VN and AdipoR1-VC; Fig. 2B). As we also observed BiFC fluorescence in the cytoplasm of transfected cells, it could be that these were artefact signals caused by the fixation and permeabilization procedure during immunostaining or irrelevant fluorescent signals. We therefore performed live-cell microscopy with HEK293 cells transfected with the AdipoR1-VN/AdipoR1-VC constructs and the AdipoR1-eYFP, as a control without immunostaining. Importantly, we could observe prominent membrane localization for both the AdipoR1 dimer as well as the AdipoR1-eYFP, and only a small amount of fluorescent proteins inside the cell (Fig. 2C). Accordingly, the intracellular fluorescence observed during immunostaining was concluded to be due to artefact signals. Next, we analyzed the dimerization of AdipoR1 by flow cytometry. This method allows a fast and real-time quantification of a large number of cells rather than single-cell analysis, as performed by BiFC microscopy. Transfections were carried out as described above for fluorescence microscopy. As expected, a strong fluorescent signal was detected for the interaction of AdipoR1-VN/AdipoR1-VC, whereas only very low background fluorescence was observed for the combination of AdipoR1-VN/hY2-VC, as well as for sole AdipoR1-VN and AdipoR1-VC, respectively (Fig. 2D). By using BiFC analysis and flow cytometry, we provide strong evidence that AdipoR1 forms receptor dimers in the membrane of living cells.
AdipoR1 dimerization depends on a glycophorine-A-like dimerization motif in TMD5
As we found AdipoR1 dimers in living cells, we were interested in how these unusual receptors formed dimers. For that reason, we screened the AdipoR1 amino acid sequence for common transmembrane dimerization motifs. In fact, we found a sequence motif (269GVFLG273, Fig. 3A) similar to the GxxxG motif that promotes dimerization of glycophorin A (GpA) (Lemmon et al., 1992; Rath et al., 2007) in the predicted fifth transmembrane domain. Bulkier amino acids such as valine or leucine are located adjacent to each glycine residue, which is typical for the sequence context of the GxxxG motif (Lemmon et al., 1992). Interestingly, this motif appears to be unique in the receptor, because it occurs only once in the AdipoR1 sequence. Amino acid sequence alignment also revealed that this consensus motif is conserved in the AdipoR1 among higher mammals, and even in the zebrafish (Fig. 3A). We mutated Gly269 and Gly273 in the wild-type AdipoR1-VN and AdipoR1-VC BiFC constructs to a larger and hydrophobic phenylalanine residue, because small-to-large mutations tend to be more effective in mapping interaction interfaces than isosteric substitutions (Senes et al., 2004). Surprisingly, HEK293 cells transfected with the respective AdipoR1-G269F,G273F mutant had significantly more homodimers than the wild type when assessed by BiFC and quantitative flow cytometry (Fig. 3B). Concomitantly, heterodimers consisting of the wild-type AdipoR1 and the mutant receptor showed BiFC signals approximately at wild-type level. However, exchange of Gly263 and Gly273 to the polar glutamic acid, led to a decrease of 85% in the BiFC dimerization signal compared with the wild-type homodimer (Fig. 3B). Even the heterodimer consisting of the wild-type and mutant receptor showed a 70% reduction in signal. We confirmed these differences in dimerization behavior between the wild type and the AdipoR1 mutants by western blotting of protein lysates from HEK293 cells transfected with BiFC constructs of the wt-AdipoR1 or the AdipoR1-G269F,G273F and AdipoR1-G269E,G273E mutant receptor, respectively (Fig. 3C). For the wild-type receptor and the G269F,G273F mutant, we observed protein bands corresponding to the receptor monomer and dimer, whereas for the AdipoR1-G269E,G273E mutant hardly any dimeric receptor was detectable. To exclude aberrant expression of the mutant receptors, we performed immunostaining for both mutant AdipoR1s in HEK293 cells (Fig. 4). All mutant receptors were expressed, as shown by immunostaining of their respective HA-tag or FLAG-tag. Moreover, by using fluorescence microscopy, we also observed a weaker BiFC signal for the AdipoR1-G269E,G273E mutants than for the G269F,G273F mutant, which is consistent with the results outlined above.
Adiponectin stimulation reduces AdipoR1 dimerization of wild-type and mutant receptors
The only natural ligand for AdipoRs known today is adiponectin (Yamauchi et al., 2007). Accordingly, we were interested to determine the adiponectin-mediated effect on receptor dimerization. We transfected HEK293 cells with AdipoR1-VN and AdipoR1-VC, and stimulated the cells directly after transfection with recombinantly expressed full-length adiponectin for 17, 20 and 24 hours at concentrations of 25 μg/ml and 50 μg/ml. We analyzed the changes in fluorescence intensity by flow cytometry. Interestingly, BiFC fluorescence for AdipoR1 dimerization determined by flow cytometry decreased in a concentration-dependent manner upon stimulation with adiponectin (Fig. 5A). We observed the most significant signal reduction (50%) in dimerization for stimulation with 50 μg/ml full-length adiponectin over 17 and 20 hours (see Fig. 5A; P<0.05). Prolonged incubation periods for 24 hours with 50 μg/ml full-length adiponectin decreased the BiFC signal to a lesser extent (25% reduction). For treatment with the lower concentration of adiponectin (25 μg/ml), we observed a 30% reduction in the BiFC signal at all time points compared with the control. Stimulation of globular adiponectin showed no detectable effect on AdipoR1 dimerization (Fig. 5B). To exclude the possibility that the decrease in the fluorescence signal results from an endocytotically degraded receptor, we performed immunostaining for HA- and FLAG-tagged AdipoR1 upon stimulation with 25 μg/ml and 50 μg/ml full-length adiponectin for 24 hours (supplementary material Fig. S1A). Additionally, we performed flow-cytometry analysis of HEK293 cells transfected with AdipoR1-eYFP and treated for 17 and 24 hours with 50 μg/ml full-length adiponectin, to assess receptor stability (supplementary material Fig. S1B). However, we found no significant differences in receptor stability upon ligand treatment using either immunofluorescence or flow cytometry. Furthermore, we used western blotting to determine whether adiponectin treatment induced an upregulation of the endogenous receptor. This could potentially compete with the Venus-fused receptor, and would lead to a decrease of the BiFC signal. As we could not observe any significant differences in endogenous receptor expression, this could be ruled out (data not shown). Next, we investigated the influence of full-length adiponectin treatment on the endogenous receptor dimerization (Fig. 5C). HEK293 cells were treated with 25 μg/ml and 50 μg/ml adiponectin, and after 17 hours of stimulation, the cells were harvested and total cellular protein was subjected to western blot. In agreement with the results described above, the AdipoR1 dimer disappeared upon treatment with higher concentrations of adiponectin (Fig. 5C, upper panel). The expression of the AdipoR1 monomer was not affected by adiponectin (Fig. 5C, lower panel). To investigate the modulatory effect of adiponectin on the dimerization of the mutant receptors, we stimulated transfected HEK293 cells for 17 hours with 50 μg/ml full-length adiponectin and analyzed the changes in fluorescence intensity by flow cytometry. Adiponectin treatment of the AdipoR1-G269F,G273F homodimer revealed a signal decrease of only 50% when compared with that observed for the wild-type dimer (70% decrease compared with untreated control cells, Fig. 5D). Remarkably, we observed a signal reduction upon ligand stimulation for the heterodimer AdipoR1-G269F,G273F and wt-AdipoR1 that was similar to the wild-type receptor dimer. Only minor effects were detected for the homo- and heterodimer of the AdipoR1-G269E,G273E mutant (Fig. 5D) after adiponectin treatment. Since we did not observe any effect for the globular form of adiponectin, we speculated whether the collagen-like domain (residues 42-107) might be responsible for the dimer dissociation. To test this hypothesis, we synthesized three overlapping peptide segments of the collagen-like domain Ad(42-74), Ad(75-107) and Ad(60-89) (Fig. 6A). We stimulated transfected HEK293 cells with all three peptides individually (1 μM) or in combination with full-length adiponectin (25 μg/ml) for 17 hours as described above. Treatment with the peptide Ad(60-89) resulted in the most significant reduction in fluorescent signal intensity (Fig. 6B), which was comparable with the effect of full-length adiponectin. Interestingly, co-stimulation with equimolar concentrations of full-length adiponectin reduced the fluorescence signal intensity for all peptides being tested, but did not lead to a further substantial signal reduction for Ad(60-89). Finally, we pretreated MCF-7 cells with 50 μg/ml full-length adiponectin for 17 hours to inhibit dimer formation, and investigated globular adiponectin-induced ACC phosphorylation. Untreated and pretreated cells not stimulated with globular adiponectin served as a control. Pretreated cells led to a 50% increase in ACC phosphorylation compared with untreated cells (supplementary material Fig. S3), suggesting increased receptor activity. However, because pretreated cells without globular adiponectin stimulation also produced a mild increase, further experiments are required to fully understand the functional consequence of AdipoR dimerization.
It has been shown for many members of the cell-surface receptor families that they can form oligomers, which are important for biological function and intracellular trafficking (Dinger et al., 2003; Herrick-Davis et al., 2006; Salahpour et al., 2004). Since the discovery of AdipoR1 and AdipoR2 as new members of the 7TM receptor family, little is known about their oligomerization behavior. In the present study, we provide strong evidence that the AdipoR1 dimerizes in vitro and in vivo, and that dimerization is driven by a GxxxG motif in the fifth transmembrane receptor domain. We observed AdipoR1 dimers in a large panel of cell lines and in human muscle tissue where the AdipoR1 is predominantly expressed and exerts major functions in regulating fatty acid oxidation and glucose uptake (Tsuchida et al., 2004; Yamauchi et al., 2003). Concordantly, muscle tissue showed the highest amount of AdipoR1 dimers compared with all investigated human cell lines. We detected AdipoR1 dimers also in C2C12 cells, which suggests that AdipoR1 dimerization occurs across different species. Our results are in agreement with previous western blot data (Yamauchi et al., 2003).
Other receptor systems that are also involved in the regulation of the carbohydrate and fat metabolism are known to form dimers. For example, the insulin receptor, a receptor tyrosine kinase, exists as a homodimer where both receptor monomers are covalently linked by disulfide bonds (Ward et al., 2007). A potential homodimer formation has also been described for the obesity-related leptin receptor, where the dimerization is suggested to trigger downstream signaling events (Nakashima et al., 1997; White et al., 1997). By using BiFC in combination with fluorescence microscopy, we extended our investigations on AdipoR1 dimerization in living cells. BiFC supplies a native and physiological environment for the detection of intracellular protein interactions, by maintaining the conditions for correct protein folding and function (Kerppola, 2006). Using this technique, we were able to visualize the AdipoR1 homodimers localized in intact and living cells. Accordingly, the AdipoR1 dimers observed in western blots originate from dimers that already are formed in the cell membrane. Detection of receptor-protein interactions using BiFC has been successfully shown for the dimerization of several other 7TM receptors, such as the calcitonin receptor-like receptor (Heroux et al., 2007) and the α1b-adrenoreceptor (Lopez-Gimenez et al., 2007) and a variety of other transmembrane proteins (Chen et al., 2006; Li et al., 2007).
For a more detailed understanding of AdipoR1 dimer formation and its function in the cellular context, it is necessary to analyze the underlying mechanism that triggers receptor dimerization. Only a few mechanisms by which heptahelical transmembrane receptors are able to form dimers are known. One possibility is the formation of covalent disulfide bonds between receptor monomers (Berthouze et al., 2007; Zanna et al., 2008). However, the contribution of disulfide bonds to AdipoR1 dimerization is rather unlikely, because we also observed the dimers under reducing conditions in the presence of β-mercaptoethanol (Fig. 1A,B). Other models of receptor dimerization involve a swap of transmembrane domain between the receptor monomers, or the direct contact of individual residues in transmembrane regions (Milligan, 2008). Inspection of the AdipoR1 protein sequence revealed a GxxxG sequence in the N-terminal part of the predicted fifth transmembrane domain from amino acids 269 to 273. This motif provides a flat interaction surface for receptor dimerization, as described for several transmembrane proteins such as GpA or the α-factor G-protein-coupled receptor (Lemmon et al., 1992; Overton et al., 2003). Basically, it is possible to investigate the role of such dimerization motifs by either substituting the glycine residues to a more bulky, hydrophobic side chain or to introduce polar amino acids to influence the interaction surface. Interestingly, mutation of both glycines to phenylalanine led to a substantial increase in AdipoR1 dimerization. However, stabilization of π-π interactions of the aromatic ring systems could account for the higher tendency to form receptor dimers, as we observed for the AdipoR1-G269F,G273F mutant. Such effects of aromatic side-chain interactions have been investigated within single peptides (McGaughey et al., 1998), but recently it has been shown that introduction of aromatic residues in artificial peptide sequences significantly enhanced their dimerization in E. coli membranes (Johnson et al., 2007). Interestingly, the introduction of amino acids with smaller side chains, unable to promote π-π interactions, disrupts the dimerization surface and abolishes the protein-protein interaction (Horschitz et al., 2008; Kienlen-Campard et al., 2008). However, by mutating the glycine residues to glutamic acid, we showed that the dimerization capacity of the AdipoR1 is nearly eliminated. Here, electrostatic repulsion of the charged carboxylic acid moieties, which are brought in close proximity as a result of the dimerization, impressively demonstrate the contribution of the GxxxG motif in AdipoR1 dimer formation. Thus, the direct contact of amino acids is likely to be the major mechanism for AdipoR dimerization, although we cannot exclude other mechanisms, such as domain swapping.
Fluorescence microscopy studies of mutant receptors indicate that the G269F,G273F mutants are predominantly located in the cell membrane and that substitution of both glycines to glutamic acid leads to an intracellular accumulation of the receptor. Impaired intracellular trafficking has also been described for GxxxG mutants of the α-factor receptor in yeast (Overton et al., 2003). Whether the dimerization of the AdipoR1 is crucial for correct cell-membrane delivery remains elusive and is an objective for further studies. Co-expression of mutant and wild-type receptors did not modify cell-surface localization (supplementary material Fig. S2). BiFC analysis of AdipoR1 interaction is not only restricted to fluorescence microscopy, it can also be used for quantitative analysis in combination with flow cytometry, as previously shown for cytosolic (Morell et al., 2007) and transmembrane proteins (Szczesna-Skorupa and Kemper, 2006) in mammalian cells. This powerful coupling of techniques allowed us to analyze a large number of cells, which statistically verified the specificity of the AdipoR1 receptor interaction. Furthermore, it implies that a combination of BiFC and flow cytometry can be applied to determine changes in the oligomerization behavior caused by environmental influences such as ligand binding.
Dynamics in protein oligomerization detected by BiFC have been shown for a variety of signaling protein complexes, but this issue is still controversial (Guo et al., 2005; Schmidt et al., 2003). In addition, BiFC is advantageous over other fluorescence-based techniques such as Förster resonance energy transfer (FRET) or bioluminescence resonance energy transfer (BRET), because BiFC signals are much less affected by changes in the cellular environment that can alter the intrinsic properties of fluorescent proteins, e.g. the lifetime and fluorescence intensity (Kerppola, 2006). Furthermore, unlike FRET, BiFC requires only a small amount of interacting fusion protein to provide a detectable fluorescent signal (Kerppola, 2006). To gain insight in the influence of the native ligand adiponectin on AdipoR1 dimerization, we applied the BiFC flow-cytometry approach to quantify changes in receptor oligomerization upon ligand stimulation. Treatment with human full-length adiponectin was performed directly after the transfection of HEK293 cells to observe any ligand-induced effects as soon as the receptor reached the plasma membrane, since once the BiFC complex is formed, it is regarded as being irreversible. Adiponectin led to a significant decrease in the BiFC signal for AdipoR1 dimerization after 17 hours of stimulation. Prolonged incubation slightly increased the BiFC signal, which might reflect a degradation or inactivation of adiponectin, or receptor desensitization. We confirmed these data also for the endogenous AdipoR1 dimer, which likewise disappeared upon ligand treatment as studied by western blotting. One possible explanation for the decrease in receptor dimerization is a ligand-induced dissociation of AdipoR1 dimers. However, it is very likely that adiponectin prevents the initial association between receptor monomers into dimers rather than dissociation of dimeric receptors. Such agonist-dependent influences on the oligomerization state of receptors have been described for the human somatostatin receptor 2 (Grant et al., 2004) and the thyrotropin receptor (Latif et al., 2002), which are in both cases physiologically required to induce receptor internalization after ligand binding. We observed a similar inhibitory effect on receptor dimers for the AdipoR1-G269F,G273F mutants, but here the stabilizing effect of the phenylalanine residues in the mutant receptor partially counteracted the dissociative effect of adiponectin. For AdipoR1, these data support the contribution of the GxxxG motif to receptor oligomerization and to quaternary receptor structure.
With regards to the ligand, we could show that the inhibitory effect on dimer formation is induced by the collagen-like domain of adiponectin. In particular, the peptide segment Ad(60-89) revealed similar modulatory effects as observed for the full-length form of adiponectin. Additional treatment of cells with the peptide and the full-length ligand did not substantially alter the effect of Ad(60-89). Hence, these data imply that full-length adiponectin, rather than its globular form, prevents formation of AdipoR1 dimers, and that the inhibitory effect is mainly mediated by the collagen-like domain of the ligand. We cannot, however, exclude the possibility that upon ligand binding, the receptor conformation changes, and hence both Venus fragments are no longer able to interact with each other to reconstitute the mature fluorophore. But this will be the subject of further detailed investigation.
As we observed ligand-induced influence of AdipoR1 dimerization, it would be of interest to understand the physiological importance of this receptor modulation. Preliminary results with MCF-7 cells that have been preincubated with full-length adiponectin show an increased signal for globular adiponectin in ACC phosphorylation (supplementary material Fig. S3). Whether this is due to an increased sensitivity of the AdipoR1 monomers compared with the dimers, or is simply an additive effect, cannot yet be determined. Furthermore, it might be possible that AdipoR1 dimerization also influences other signaling pathways. Since AdipoR1 is expressed in nearly every cell line, it is difficult to study the impact of dimerization with dimerization-defective AdipoR1 mutants because the endogenous receptor interferes substantially with any established signal transduction assays for AdipoRs; such studies will have to be performed with receptor-knockout cells.
In conclusion, in the present study, we use western blotting and BiFC flow cytometry to show that endogenous and transiently transfected AdipoR1 forms receptor dimers. Dimerization of the receptor is at least partly triggered by a GxxxG motif. Moreover, we observed a concentration-dependent inhibitory effect of adiponectin on the AdipoR1 homodimerization, which is mediated by the amino acids 60 to 89 of the collagen-like domain of the ligand. To our knowledge, this is the first contribution towards an understanding how these receptors interact with each other and their natural ligand. Concomitantly, this study sheds more light on the question of how the signaling capabilities of adiponectin receptors are regulated by adiponectin, which is important for a more-detailed view of signal modulation of adiponectin in tissues from normal and obese individuals.
Materials and Methods
Tissue culture material was purchased from PAA Laboratories (Pasching, Austria). The pEYFP-N1, pCFP-N1, pCDNA3 and pET-15b vector were from Clontech (Heidelberg, Germany), Invitrogen (Karlsruhe, Germany) and Novagen (Darmstadt, Germany), respectively. Antibodies for BiFC immunostaining were obtained from Sigma (Munich, Germany) or Dianova (Hamburg, Germany). Anti-AdipoR1 antibodies and blocking peptide against anti-AdipoR1 were purchased from SantaCruz (Heidelberg, Germany) or Acris Antibodies (Hiddenhausen, Germany). Antibodies against phosphorylated ACC and total ACC were obtained from Cell Signaling Technologies (Danver, MA).
Plasmid construction and AdipoR1 mutants
Cloning of the human ADIPOR1 cDNA into pEYFP-N1 and pCDNA3 expression vectors was essentially done as described previously (Deckert et al., 2006). BiFC plasmids encoding the N-terminal (VN, amino acids 1-154) or C-terminal (VC, amino acids 155-239) fragment of the Venus protein were kindly provided by Stefan Hüttelmaier and Mechthild Hatzfeld (Martin-Luther University, Halle/Saale, Germany) (Wolf et al., 2006). The plasmids were constructed as such that the VN fragment carries a C-terminal FLAG-tag and the VC fragment a hemagglutinin (HA)-tag at its C-terminus (Wolf et al., 2006). Human ADIPOR1 cDNA was cloned C-terminally of the respective Venus fragment. Human NPY2R (hY2 receptor) cDNA was cloned N-terminally of the VC segment. AdipoR1-G269F,G273F mutants and AdipoR1-G269E,G273E mutants of AdipoR1-VN and AdipoR1-VC BiFC constructs were generated by QuikChange site-directed mutagenesis (Stratagene). Correct mutations were confirmed by DNA sequencing.
Cell culture and transfection
Culture of HEK293, MCF-7, SK-N-MC, HUVEC and C2C12 cells was done as recommended by the supplier (DSMZ, Braunschweig, Germany). For transfection, 0.25 μg plasmid was transfected into HEK293 cells using Lipofectamine™ 2000 (Invitrogen, Germany) according to the manufacturer's instructions. Transfection of MCF-7 cells was done similarly with 1 μg plasmid.
Generation of recombinant adiponectin and western blot analysis
Human recombinant full-length and globular adiponectin was cloned into pET-15b bacterial expression vector and expressed as His-tagged protein in E. coli BL21(DE3) pLysS_Rare. His-tagged adiponectin was purified by Ni-NTA affinity chromatography and purity was confirmed by SDS-PAGE and mass spectrometry. For western blot analysis, cells were lysed in lysis buffer (25 mM Tris-HCl, pH 7.6, 150 mM NaCl, 2 mM EDTA, 1% NP-40, 2 mM PMSF, 4 μg/ml aprotinin, 10 μg/ml leupeptin, 1 μg/ml pepstatin, 0.05 mM Pefabloc) and briefly sonicated. Samples were separated by SDS-PAGE, and analyzed by western blot using specific antibodies and visualized with ECL detection (ECL western blotting substrate; Thermo Scientific, Germany). To verify the specificity of the AdipoR1 protein bands, anti-AdipoR1 antibody was incubated with a fivefold excess of a specific blocking peptide before immunodetection. Muscle biopsies of normal glucose-tolerant healthy volunteers were taken, as previously described (Bergstrom, 1975; Oberbach et al., 2006). In brief, participants were admitted into the study at 9.00 am after 14 hours of fasting. They had been asked to avoid vigorous exercise for at least 3 days before the muscle biopsy. After administration of local anesthesia, muscle biopsy specimens were obtained from the musculus vastus lateralis using a biopsy device (Somatex 2.1, Teltow, Germany). Samples were treated with talcum powder and frozen immediately in liquid nitrogen.
BiFC analysis by fluorescence microscopy and flow cytometry
For BiFC analysis by fluorescence microscopy, HEK293 cells were grown on coverslips in 24-well plates. 17 hours after transfection, the cells were fixed in 4% paraformaldehyde. For immunostaining of HA- and FLAG-tagged receptors, cells were blocked and permeabilized with 0.5% Triton X-100, 10% BSA in PBS for 1 hour (37°C) and incubated for 2 hours (37°C) with the primary antibody followed by a 1.5 hour (37°C) incubation with the appropriate secondary antibody. HA- and FLAG-tags were detected by using rabbit anti-HA and mouse mAb anti-FLAG antibody, respectively followed by tetramethyl rhodamine iso-thiocyanate (TRITC)-conjugated goat anti-rabbit IgG and Cyanin 5 (Cy5)-conjugated donkey anti-mouse IgG. Cells were imaged with a Zeiss Axio Observer.Z1 fluorescence microscope (Carl Zeiss AG, Germany) using the appropriate filter settings for yellow fluorescent protein (YFP) (excitation: BP500/20; emission: BP535/30), Cy5 (excitation: BP640/30; emission: BP690/50) and TRITC (excitation: BP565/30; emission: BP620/60). Live-cell microscopy was performed in eight-well μ-Slides (ibidi, Martinsried, Germany) at 37°C. Transfection of HEK293 cells in μ-Slides was carried out as described above. For flow cytometry, HEK293 cells were grown in 24-well plates and 17-24 hours after transfection, cells were gently trypsinized, washed with DMEM, Ham's-F12 and resuspended in 4% paraformaldehyde. The cells were analyzed with a CyFlow™ ML flow cytometer and FlowMax™ software (Partec, Germany) using an excitation wavelength of 488 nm. The control cell population was set to <0.5% above the threshold. The percentage and intensity of cells above the threshold fluorescence was determined for all samples. Values are presented as mean values ± s.e.m. Statistical significance was analyzed by one-way ANOVA.
Peptides were synthesized by solid-phase synthesis on a Syro II robot (Syro, MultiSynTech, Bochum, Germany) using fluorenylmethoxycarbonyl/tert-butyl (Fmoc/t-Bu) strategy as previously described (Cabrele et al., 2000). Peptides were analyzed by RP-HPLC (Vydac RP18-column, 4.6×250 mm; 5 μm/300Å, Merck Hitachi, Darmstadt, Germany) and MALDI-TOF-TOF-MS (UltraflexIII, Bruker) and purified by RP-HPLC (Shimadzu RP18-column, 12.5×250 mm; 5 μm/300Å). To obtain free C-termini all peptides were synthesized on a Wang resin (Novabiochem). For analytical data of the synthesized peptides, see supplementary material Table S1.
We gratefully thank Regina Reppich-Sacher for mass spectrometry analysis, Janet Schwesinger for DNA sequencing, Kristin Löbner for cell culture and Stefanie Nagel for excellent assistance with western blotting. This work was supported by grants from the Deutsche Forschungsgemeinschaft (KFO152, Be 1264/10-1) and the Europäische Fonds für regionale Entwicklung EFRE-13401. J.T.H. is grateful to the Studienstiftung des deutschen Volkes for a scholarship.