Expression properties of NKs and their homo-/hetero-dimerization

Individual NKs may play distinct roles in responses to their cognate neuropeptides. However, due to sequence similarity of peptides, these receptors are expected to affect each other, potentially provoking diverse cellular responses through ligand binding modes and receptor complex formation. As a first step to prove this idea, we investigated expression levels of these receptors in several cell lines. RT-PCR revealed that all three receptors were expressed in most cells with some exceptions, implying that they might have biological relevance to each other in cellular responses upon binding to ligands. According to their band intensities, NK1 might be the dominant receptor in A549 and DLD1 cells, whereas NK2 and NK3 might not be expressed in T98G and A549 cells, respectively (Fig. 1A). Next, their membrane localization was determined in cells expressing HiBiT-tagged forms since GPCRs function through neuropeptide binding on the cell surface. HiBiT-mediated luminescence increased depending on the plasmid amount transfected into HEK293 cells. Signals were quite strong in cells expressing NK1 and NK2, but relatively weak in NK3, implying that either protein expression or membrane localization of NK3 was relatively low (Fig. 1B). To determine protein amounts, cells expressing C-terminal HA-tagged receptors driven by CMV promoter were lysed with RIPA buffer and used for western blotting with anti-HA antibodies. As shown in Fig. 1C, the main band of NK1 was detected between 50 and 70 kDa. NK2 was detected as multiple bands, with the main signal located between 35 and 50 kDa. A clear band of NK3 was detected at 50 kDa, although there were weak bands at different sizes. Comparing total amounts of proteins, NK3 expression was prominently lower than those of others, indicating that NK3 was expressed with a low efficiency.

Fig. 1

Expression properties and molecular interaction of neurokinin receptors (NKs). A RT-PCR analysis of mRNA expression. Total RNAs from cell lines were analyzed with isotype-specific primers and PCR products after 35 cycles for the receptors or 25 cycles for β-actin gene were visualized on a 1.5% agarose gel. A 1 kb-plus ladder was used as a size marker. B N-terminal HiBiT-tagged receptors were expressed in HEK293 cells with varying amounts of plasmids. HiBiT activity was assessed by adding Nano-Glo® HiBiT extracellular detection reagent. **: p < 0.01 vs. vector C Expression of C-terminal HA-tagged NKs. HEK293 cells transfected with plasmids containing C-terminal HA-tagged NKs were lysed with RIPA buffer and subjected to western blotting with anti-HA antibodies. D Luminescence induced by receptor dimerization. HEK293 cells co-expressing NKs of C-terminal SmBiT- and LgBiT-tagged forms were analyzed by NanoBiT assay. **: p < 0.01 vs. vector (V). E Co-immunoprecipitation. HEK293 cells co-expressing FLAG- and HA-tagged receptors were lysed and subjected to immunoprecipitation with anti-FLAG agarose, followed by western blotting with anti-HA antibody

Next, complex formation of these receptors was investigated using NanoBiT. When C-terminal LgBiT-tagged NKs were expressed with C-terminal SmBiT-tagged NKs in HEK293 cells, luciferase activities were enhanced, indicating that LgBiT/SmBiT could bind through homo- or hetero-dimerization of NK1 and NK2. However, luciferase activities in cells expressing NK3 with either NK1 or NK2 were not increased, possibly due to a low expression of NK3 by a relatively weak Ubiquitin C promoter or structural incompatibility of NanoBiT-tagged form (Fig. 1D). Direct interaction between NK1 and NK2 was verified by co-immunoprecipitation. Due to migration properties of GPCRs on SDS-PAGE, precipitates were subjected to SDS-PAGE without boiling. Interestingly, co-precipitated NK1 was not detected at the size where directly loaded proteins were located. Instead, main precipitates were detected at the start line of the running gel. On the other hand, co-precipitated NK2 was detected at the same size as directly loaded ones, suggesting that NK2 could be easily dissociated from the protein complex by the sample buffer, whereas NK1 could not (Fig. 1E). We expected that NK3 might also bind to NK1 or NK2 since they belong to the same receptor group and share the same ligands. However, immunoprecipitation for NK3 revealed that NK3 itself was co-precipitated, whereas neither NK1 nor NK2 was precipitated, indicating that NK3 could homodimerize without heterodimerize with other members (see right blot in Fig. 1E). Based on these data, NK3 is expected to function independently without any influence from other receptors.

Comparison of NK subtypes-mediated cellular responses using NanoBiT constructs

NKs have been extensively studied, especially with respect to neuropeptide-stimulated Ca2+ responses [1]. Recently, we developed a real-time calcium assay based on NanoBiT and the Ca2+-dependent interaction of calmodulin and its target proteins to analyze NK-mediated cellular responses [28]. First, neuropeptide-stimulated luciferase activation was measured in cells expressing NKs and calcium probes to determine the suitability of our assay method. Neuropeptides induced luciferase activation with slightly different patterns at 1 μM, indicating that all receptors could mediate cellular responses to neuropeptides at high concentration (Additional file 1: Fig. S1). Next, the activation efficiency of receptors by each peptide was investigated on Ca2+ responses based on NanoBiT assay. Dose–response curves revealed that all tested peptides acted as full agonists for the three NKs with different potency: SP > NKA = NKB for NK1; NKA >  > NKB > SP for NK2; NKB >  > NKA > SP for NK3. These preference results of tachykinin for each receptor were in agreement with previous reports showing that SP, NKA, and NKB were dominant ligands for NK1, NK2, and NK3, respectively (Additional file 1: Fig. S2A) [2, 4, 29].

Ligand-stimulated β-arrestin recruitment to receptors was also examined using the NanoBiT system. The NanoBiT system was used to create construct combinations of receptors and β-arrestin, which were tagged with either SmBiT or LgBiT at N- or C-terminal. These constructs were then expressed in HEK293 cells. The combination for each receptor showing the highest luciferase activity by ligand treatment was selected. Interaction of NK1-SmBiT with β-arrestin1-LgBiT was induced by all peptides in a dose-dependent manner, with SP being the dominant stimulator of this binding. The interaction of NK2-LgBiT with β-arrestin1-SmBiT was more sensitive to NKA, whereas it was barely induced by SP in terms of potency and efficacy (Additional file 1: Fig. S2B). Unfortunately, the interaction of NK3 with β-arrestin1 was difficult to determine with NanoBiT constructs due to a low expression level of the receptor. Overall, despite different sensitivity, both calcium assay and β-arrestin recruitment assay are feasible for analyzing ligand stimulation properties of GPCRs since they measure separated downstream signaling pathways.

NK1 enhances SP-stimulated β-arrestin recruitment to NK2.

Recruitment of β-arrestin1 to NKs clearly demonstrated that SP acted as a full agonist to NK1 but partial agonist to NK2 with relatively low affinity. Apart from properties of individual receptors, co-expression of multiple receptors in the same cells can result in complex formation and potentially lead to distinct signaling events compared to those mediated by individual receptors. Thus, dimerization effect of NK1 and NK2 was examined in SP-stimulated β-arrestin1 recruitment to receptors. We transfected HEK293 cells with three plasmids. Luciferase activities of NK2-LgBiT and β-arrestin1-SmBiT by SP were similar to those obtained from the system with two plasmids (459 nM vs. 409 nM in EC50) (Fig. 2A, left graph). Luciferase activities were slightly enhanced in the presence of intact NK1 (164 nM in EC50) (Fig. 2A, middle graph). However, luciferase activities were not changed by intact NK2 (278 nM in EC50) (Fig. 2A, right graph). The effect of NK1 on the interaction of NK2-LgBiT with β-arrestin1-SmBiT was obvious at a low concentration of SP (see red box). Figure 2B shows real-time luciferase activities by low concentrations of SP. The lowest concentration of SP to stimulate the luciferase activities through β-arrrestin1 interaction to NK2 was 78 nM, while activities were observed even at 1.2 nM SP in the presence of NK1, indicating that SP could stimulate NK1 and that ensuing GRK-dependent phosphorylation of both receptors in NK1/NK2-LgBiT complex was enough to recruit β-arrestin1-SmBiT to NK2-LgBiT (Fig. 2B). Basal luciferase activities of NK2-LgBiT and β-arrestin1-SmBiT were increased by NK1 compared to NK2, although the difference seemed to have no effect on ligand-stimulated activities (Fig. 2C). Dose-dependent curves of β-arrestin1 recruitment showed that NK1 was much more sensitive to SP than NK2. To explore the effect of NK2 on NK1/β-arrestin1 interaction, HEK293 cells expressing NK1-SmBiT and β-arrestin1-LgBiT with intact NKs were stimulated by 0.1 nM SP, a concentration that could induce β-arrestin1 recruitment to NK1-SmBiT but not to NK2-LgBiT. Co-expression of NK1 lowered luciferase activities of NK1-SmBiT/β-arrestin1-LgBiT by SP, implying that intact NK1 might compete with NK1-SmBiT for limited amount of ligands. In contrast, co-expression of NK2 enhanced luciferase activities, suggesting that NK1-SmBiT/NK2 complex might increase the chance to bind β-arrestin1-LgBiT, although NK2 itself could hardly bind to SP (Fig. 2D).

Fig. 2

Effect of NK1 on SP-stimulated β-arrestin1 recruitment to NK2. A Cells expressing NK2-LgBiT and β-arrestin1-SmBiT with intact NKs were treated with different concentrations of SP and changes of luciferase activities were measured in real time. Maximal activities from each concentration were plotted on the graph. B Time-dependent responses at designated concentration which was boxed area in graphs of (A). C Basal luciferase activities in cells expressing NanoBiT constructs with intact receptors. D Effect of NKs on SP-stimulated β-arrestin1 recruitment to NK1. Upper graph: dose-dependent curves of SP-stimulated β-arrestin1 recruitment to NKs. Lower graph: luciferase activities of NK1-SmBiT and β-arrestin1-LgBiT in the presence of intact NKs in cells treated with 0.1 nM SP, a concentration that stimulated NK1 but not NK2 (indicated by red arrow in the upper graph)

NK2 negatively regulates NK1-mediated cellular responses to SP

Recruitment of β-arrestin to the cognate receptor upon ligand stimulation is an important step for downregulation of cellular responses to ligands by inducing receptor internalization. However, this process can also induce other signaling events apart from the ones mediated by heterotrimeric G proteins [30,31,32]. As previously mentioned, NK2 could enhance SP-stimulated β-arrestin interaction with NK1, suggesting a possible role of NK2 as a modulator of NK1 signaling. While NK1 was 100 folds more sensitive to SP than NK2 in terms of Ca2+ response (0.013 nM vs 2.2 nM in EC50, see Fig. S2A), both receptors had similar sensitivity to NKA (0.027 nM vs. 0.012 nM in EC50). To explore the effect of NK2 on NK1-mediated Ca2+ influx, cells expressing receptors with NanoBiT-based Ca2+ probes were treated with 0.1 nM SP, a concentration that induced nearly maximal activation of NK1 but no response of NK2. Transfection with different amounts of NK1 plasmids did not change the Ca2+ influx pattern possibly because this concentration of the ligand was almost at a saturation level. On the other hand, Ca2+ responses were significantly decreased in cells expressing both NK1 and NK2, although cells expressing NK2 alone did not show any response. This result suggests that NK2 might modulate NK1 activation through receptor complex formation (Fig. 3A, left graphs). Ca2+ responses to NKA were quite similar in cells expressing a single receptor or co-expressing both receptors since these receptors had similar sensitivity to NKA (Fig. 3A, right graph).

Fig. 3

Effects of NK2 on NK1-mediated Ca2+ influx and ERK phosphorylation. A Optimal concentrations of tachykinin-stimulated Ca2+ responses were determined from dose–response curves. HEK293 cells expressing Ca2+ probes with NKs were treated with 0.1 nM SP or NKA. The concentration of SP at 0.1 nM was the concentration that fully stimulated NK1 but not NK2. The concentration of NK at 0.1 nM was the concentration that fully activated both receptors. B Effect of SP pretreatment on NKA-stimulated Ca2+ responses. Cells expressing Ca2+ probes with NKs were treated with 0.1 nM SP for 30 min. Medium was then replaced and 0.1 nM NKA was added. C HEK293 cells and exogenous NK2-expressing cells were serum-starved for 24 h and treated with 0.1 nM peptides. Cells were harvested at designated time point and subjected to western blotting with anti-ERK or phospho-ERK antibodies

To further confirm the dimerization of NK1 and NK2, cells expressing receptors were pretreated with 0.1 nM SP for 30 min. After the medium was removed, NanoBiT substrate was added. Treatment with 0.1 nM NKA induced weak Ca2+ responses through a small number of NK1 proteins that remained unresponsive to SP. Cells expressing NK2 showed a strong Ca2+ response to NKA regardless of SP pretreatment. However, such response was significantly decreased in cells expressing both NK1 and NK2, suggesting that some NK2 proteins were dimerized with NK1 and internalized with their binding partners by SP stimulation. Consequently, the remaining NK2 attended to NKA binding and mediated Ca2+ response (Fig. 3B).

Most GPCRs can activate ERK phosphorylation, which is the most sensitive cellular response. When HEK293 cells were treated with 0.1 nM SP, ERK phosphorylation was transiently induced probably through NK1, implying that NK1 dominated in response to the ligand, although other NKs might be expressed according to RT-PCR. Interestingly, such ligand-stimulated phosphorylation was significantly reduced by exogenous expression of NK2. This might be another evidence of NK2’s ability to modulate SP-stimulated NK1 activation (Fig. 3C, left panel). NKA-dependent ERK phosphorylation was weakly induced in HEK293 cells. It was further increased in the presence of exogenous NK2, indicating that NK2 was a positive mediator of NKA (Fig. 3C, right panel).

NK1-mediated NK2/β-arrestin1 interaction is inhibited by an NK1-specific inhibitor

The interaction between NK1 and NK2 was further confirmed using aprepitant, an NK1-specific inhibitor. SP-stimulated luciferase activities of NK1-SmBiT and β-arrestin1-LgBiT were completely inhibited by aprepitant even at a high concentration (1 μM), while luciferase activities of NK2-LgBiT and β-arrestin1-SmBiT were not affected by this inhibitor (Fig. 4A, left and middle graphs). However, when NanoBiT constructs of NK1 and β-arrestin1 were expressed with intact NK2, luciferase activities were slightly increased by SP even in the presence of the inhibitor (Fig. 4A, right graph). This result suggests that NK2 activated by high concentration of SP might recruit β-arrestin1-LgBiT to the complex of NK2 and NK1-SmBiT. Subsequently, β-arrestin1-LgBiT and inactive NK1-SmBiT caused by aprepitant were getting closer to each other enough to induce luciferase activation.

Fig. 4

Inhibition of NK1-mediated NK2/β-arrestin1 interaction by NK1-specific inhibitor. A The effect of aprepitant on 1 μM SP-stimulated β-arrestin1 recruitment to NKs was examined in cells expressing NanoBiT constructs of NKs and β-arrestin1 with or without NK2. Cells were pretreated with 1 μM aprepitant for 30 min before the addition of SP. B Aprepitant inhibited 10 nM SP-stimulated interaction of NK2-LgBiT and β-arrestin1-SmBiT in the presence of NK1. C Aprepitant inhibited NKA-stimulated β-arrestin1 recruitment to NK1 but not NK2. Cells expressing NanoBiT constructs of NKs and β-arrestin1 with or without another NKs were pretreated with aprepitant for 30 min and then stimulated with 10 nM NKA. Luciferase activities were measured with a luminometer

As another approach, cells expressing NK2-LgBiT and β-arrestin1-SmBiT with intact NK1 were treated with 10 nM SP, a non-effective concentration to activate NK2. Ligand-dependent luciferase activities were likely from NK1-dependent interaction of NK2-LgBiT and β-arrestin1-SmBiT, as aprepitant inhibited luciferase activities (Fig. 4B).

NKA-stimulated luciferase activities were not influenced by aprepitant in cells expressing both NK2-LgBiT and β-arrestin1-SmBiT (Fig. 4C, upper left). Luciferase activities were not downregulated by the inhibitor in cells expressing NanoBiT constructs with intact NK1 either, implying that inactive NK1 in the receptor complex had no effect on NKA-stimulated interaction of NK2-LgBiT with β-arrestin1-SmBiT (Fig. 4C, upper right). However, NKA-stimulated luciferase activities were inhibited by aprepitant in cells expressing both NK1-SmBiT and β-arrestin1-LgBiT, confirming that aprepitant was a specific inhibitor of NK1 (Fig. 4C, lower left). When intact NK2 was co-expressed with these NanoBiT constructs, NKA-stimulated luciferase activities were slightly increased by NKA in the presence of aprepitant, similar to the response to SP in cells expressing the same plasmids. This result indicates NK2-dependent interaction of inactive NK1-SmBiT and β-arrestin1-LgBiT (Fig. 4C, lower right).

NK1-mediated responses to SP are downregulated by NK2 in A549 cells

Based on results of RT-PCR (Fig. 1A), A549 cells are expected to express both NK1 and NK2. To confirm the expression of these two receptors, we measured ligand-stimulated ERK phosphorylation. Although basal levels of phospho-ERK were high in 24 h serum-starved A549 cells, treatment with 0.1 nM SP induced ERK phosphorylation, indicating expression of NK1 proteins in cells. Furthermore, exogenous expression of NK2 led to a decrease in SP-stimulated ERK phosphorylation, while 0.1 nM NKA slightly increased ERK phosphorylation and significantly increased it in the presence of exogenous NK2 (Additional file 1: Fig. S3).

To further investigate the effect of NK2 on SP-stimulated NK1 signaling, we generated A549 cells lacking either NK1 or NK2 using CRISPR/Cas9. These cells did not respond to low concentrations of tachykinins with respect to Ca2+ responses. When NK1 was exogenously expressed in A549 cells, 0.1 nM SP increased Ca2+ influx, which was significantly decreased by co-expression of NK1 and NK2 (Fig. 5A, upper left). However, the Ca2+ influx patterns by 0.1 nM NKA were quite the same for cells expressing NK1 with or without NK2 (Fig. 5A, upper right). When NK2 was exogenously expressed in NK1 knock-out A549 cells, NKA induced Ca2+ influx, but SP did not (Fig. 5A, lower left). The SP-stimulated maximal Ca2+ increase through NK1 was significantly downregulated by NK2 expression, although maximal responses to NKA were similar under all conditions (Fig. 5A, lower right). In NK2 knock-out A549 cells expressing exogenous NK1, treatment with 0.1 nM SP resulted in prominent and sustained ERK phosphorylation, while phosphorylation intensity was decreased in cells co-expressing NK1 and NK2, indicating negative modulation of NK1 by NK2. However, 0.1 nM NKA-stimulated ERK phosphorylation was quite similar under all expression conditions (Fig. 5B).

Fig. 5

NK1-mediated responses to SP were downregulated by NK2 in A549 cells. A A549 cells lacking either NK were reconstituted with exogenous NKs. After incubation with calcium 6 reagent, cells were treated with 0.1 nM SP or NKA and the calcium responses were measured with a FlexStation® 3 microplate reader. Maximal increases are shown in the graph. **: p < 0.01 (NK1 alone vs. NK1 + NK2). B A549 cells lacking NK2 and reconstituted with NKs were starved for 24 h and treated with 0.1 nM SP or NKA. Cells were lysed at the designated time point and cell lysates were subjected to western blotting with anti-ERK or phospho-ERK antibodies. The graph shows the intensity of maximal phosphorylation normalized with ERK. *: p < 0.05, ***: p < 0.001 (NK1 alone vs. NK1 + NK2). C Migration of A549 cells toward 0.1 nM SP. Cells lacking NKs and reconstituted with NKs in pictures were added into upper wells of transwell migration chambers and 0.1 nM SP was added to the lower well. After 18 h of incubation, non-migrated cells in upper well were removed with a cotton swab and cells migrated to the lower surface of the transwell were stained with Diff-quick staining solution and counted under a microscope. **: p < 0.01 vs. not treated. #: p < 0.05 (wild type vs. NK2KO or NK2KO + NK1)

Furthermore, since SP has been reported to enhance cellular migration [8, 25, 33, 34], we performed a chemotactic assay with NKs-deleted and reconstituted A549 cells in the presence of the ligand. A549 cells were active in movement. Their migration was enhanced by 0.1 nM SP. However, migration was not increased in NK1 knock-out cells with or without exogenous NK2. Interestingly, NK2 knock-out cells migrated well regardless of exogenous NK1, while NK2 reconstitution reduced migration, suggesting that NK2 could negatively modulate NK1-mediated cellular migration toward SP.