Biophysical validation of serotonin 5-HT2A and 5-HT2C receptor interaction

The serotonin (5-HT) 5-HT2A receptor (5-HT2AR) and 5-HT2C receptor (5-HT2CR) in the central nervous system are implicated in a range of normal behaviors (e.g., appetite, sleep) and physiological functions (e.g., endocrine secretion) while dysfunctional 5-HT2AR and/or 5-HT2CR are implicated in neuropsychiatric disorders (e.g., addiction, obesity, schizophrenia). Preclinical studies suggest that the 5-HT2AR and 5-HT2CR may act in concert to regulate the neural bases for behavior. Here, we utilize three distinct biophysical and immunocytochemistry-based approaches to identify and study this receptor complex in cultured cells. Employing a split luciferase complementation assay (LCA), we demonstrated that formation of the 5-HT2AR:5-HT2CR complex exists within 50 nm, increases proportionally to the 5-HT2CR:5-HT2AR protein expression ratio, and is specific to the receptor interaction and not due to random complementation of the luciferase fragments. Using a proximity ligation assay (PLA), we found that cells stably expressing both the 5-HT2AR and 5-HT2CR exhibit 5-HT2AR:5-HT2CR heteroreceptor complexes within 40 nm of each other. Lastly, bioluminescence resonance energy transfer (BRET) analyses indicates the formation of a specific and saturable 5-HT2AR:5-HT2CR interaction, suggesting that the 5-HT2AR and 5-HT2CR form a close interaction within 10 nm of each other in intact live cells. The bioengineered receptors generated for the LCA and the BRET exhibit 5-HT-mediated intracellular calcium signaling as seen for the native receptors. Taken together, this study validates a very close 5-HT2AR:5-HT2CR interaction in cultured cells.


Introduction
The metabotropic serotonin (5-HT) 5-HT 2 receptor (5-HT 2 R) family consists of three isoforms (5-HT 2A R, 5-HT 2B R, 5-HT 2C R) that share~50% total sequence homology and~80% sequence homology within their seven transmembrane domains [1]. The 5-HT 2A R and 5-HT 2C R localized within the central nervous system are implicated in a range of normal behaviors (e.g., appetite, sleep) and physiological functions (e.g., endocrine secretion) a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 while dysfunctional brain 5-HT 2A R and/or 5-HT 2C R are implicated in neuropsychiatric disorders (e.g., addiction, obesity, schizophrenia). Antagonism of the 5-HT 2A R is a common feature of atypical antipsychotics employed in schizophrenia [2] and is the mode of action for the recently FDA-approved pimavanserin in the treatment of psychosis in Parkinson's disease [3]. Furthermore, the 5-HT 2C R agonist lorcaserin is FDA-approved for the treatment of obesity [4]. Biochemical, behavioral and pharmacological studies indicate that the 5-HT 2A R and 5-HT 2C R interact in rodents in vivo [5][6][7][8][9][10] raising the possibility that the 5-HT 2A R and 5-HT 2C R may act in concert to regulate the neural bases for behavior (for reviews) [11,12].
The 5-HT 2A R and 5-HT 2C R transcripts co-exist in brain regions particularly associated with the limbic-corticostriatal circuitry [13][14][15][16], while the protein for these receptors colocalizes to the same neurons in the rat medial prefrontal cortex (mPFC) [17]. We recently demonstrated that the two receptors form a protein complex in the mPFC assessed by coimmunoprecipitation [5]. Expression of this complex was associated with phenotypic levels of impulsivity in the rat [5], suggesting the behavioral significance of a 5-HT 2A R:5-HT 2C R protein complex. Based upon reported evidence of a 5-HT 2A R and 5-HT 2C R association [5][6][7][8][9][10], we conducted biophysical studies to further define and validate the interaction between these homologous receptors at the single cell level.

Luciferase complementation assay (LCA)
In the LCA reporter system, two complementary N-(NLuc) and C-terminus (CLuc) components of the enzyme luciferase, which have no activity on their own, are split in half and fused to the two receptor proteins of interest [33,34,40]. The detection of two proteins in close proximity in living cells is achieved as the association of the two proteins within < 50 nm brings the inactive luciferase fragments into close proximity and the luciferase enzyme activity is reconstituted [33,34,40].

Saturation binding assay
At 24 hours post-transfection, transfection media was replaced with DMEM supplemented with 10% dialyzed FBS; 24 hours later, cells were collected by centrifugation at 4000 x g at 4˚C for 25 mins in ice cold assay buffer containing 50 mM Tris HCl, 10 mM MgCl 2 and 0.1 mM EDTA. Membranes were collected by centrifugation three times at 4500 x g at 4˚C for 20 mins and stored at -80˚C until use. Saturation binding isotherms were performed in 96-well plates using similar methods to the psychoactive drug screening program (PDSP) [41].

Proximity ligation assay (PLA)
The PLA is a flexible and informative technology that expands upon traditional immunocytochemistry to include direct detection of low levels of individual proteins, the existence of protein:protein interactions ( 40 nm) as well as the subcellular localization of the protein:protein interaction with high specificity and sensitivity. The PLA was performed using the recommended manufacturer's protocol with minor modification, similar to those previously reported [42]. . Four non-overlapping fields of view per well (20X magnification) were identified and photomicrographs acquired under each experimental condition. Images were acquired with Leica LASX Software and processed with NIH ImageJ [43,44]. To discriminate PLA puncta from the background fluorescence, identical for all conditions, the manually selected threshold (70) was applied to all images. The number of nuclei (DAPI+;~170 per field of view per condition) and total puncta (red spots) were counted using the Duolink 1 Image Tool Software (Olink Bioscience) from each of the four field of views and averaged for each experimental condition for statistical comparison, with a total of five biological replicates.

Bioluminescence resonance energy transfer (BRET) assay
BRET is a biophysical technique that utilizes energy transfer between Renilla luciferase (RLuc) and a fluorescent protein, such as yellow fluorescent protein (YFP), that enables the spatial resolution of protein:protein interactions to 10 nm in living cells. The BRET assay is commonly used to characterize protein:protein interactions with GPCRs and was adapted from previous publications with minor modifications [8,45].

Data analysis and statistics
All data were analyzed with GraphPad Prism 7.0 software (La Jolla, CA). Maximal luminescence values (± SEM) from the LCA and MTT assays were compared using a one-way analysis of variance (ANOVA) followed by a priori comparisons conducted with a Tukey's test. We assessed the potency (pEC 50 ) and efficacy (E MAX ) of 5-HT to activate Ca i 2+ release for all novel receptor fusion proteins to ensure similar signaling properties relative to the wildtype receptors. Data from Ca i 2+ release assays are presented as half maximum (pEC 50 ) and maximum (E MAX ) values (mean ± SEM), representing potency and efficacy, respectively, as computed by GraphPad using a four parameter nonlinear regression curve-fitting algorithm. To assess the effect of different transfection ratios of receptor plasmids on the total receptor protein level, we compared B max values obtained from radioligand binding studies. We also determined the potential effect of different transfection ratios of receptor plasmids on receptor ligand affinity, by statistically comparing receptor affinity values (K d ). Data from saturation binding isotherms are presented as B MAX and K d values (mean ± SEM), as computed by GraphPad using one site-specific binding nonlinear regression curve-fitting algorithm; a one-way ANOVA was conducted followed by a priori comparisons with Dunnett's test. The total number of PLA puncta (mean ± SEM) from four individual fields of view under each experimental condition were averaged for each separate condition; a one-way ANOVA followed by a priori comparisons with a Tukey's test or a Student's t-test was conducted, as appropriate. Data from BRET assays are presented as mBRET 50 and mBRET MAX values (mean ± SEM) as computed by GraphPad using specific binding with Hill slope nonlinear regression curve-fitting algorithm. The experiment wise error rate for all analyses was set at α = 0.05.
Ideally, a 1:1 protein ratio would be achieved for these studies, however, the 5-HT 2A R-CLuc consistently expresses at a higher efficiency and level than the 5-HT 2C R-NLuc in HEK293 cells, as observed in Table 1. Further conditions to attempt to achieve a 1:1 protein ratio (i.e., above the 3:1 plasmid ratio) negatively impacted the health of the cells using the MTT assay (data not shown). Thus, we were unable to obtain a higher expression of the 5-HT 2C R without increasing the overall amount of transfected plasmid DNA and negatively impacting cell health. Taken together, these data suggest that a complex formation exists within 50 nm (limit of detection for the LCA) and that complex formation increases proportionally to the 5-HT 2C R:5-HT 2A R expression ratio. We next determined if the LCA interaction could be inhibited by co-transfection of the 5-HT 2C R-NLuc:5-HT 2A R-CLuc (3:1 transfection ratio, Fig 2, "control") with increasing amounts of cDNA (0.25-2 μg) for untagged WT 5-HT 2C R (Fig 2A), WT 5-HT 2A R (Fig 2B) or empty vector (Fig 2C). We predicted that transfection of the untagged WT 5-HT 2 R would competitively interact with its respective 5-HT 2 R LCA construct for the heteromeric complex, thereby reducing the ability to reconstitute luciferase, resulting in a reduction in the luminescence signal. A main effect of the WT 5-HT 2C R transfection condition on maximum light counts/sec was detected [F 4, 12 = 21.13; p < 0.0001; Fig 2A]; the native WT 5-HT 2C R at all amounts of cDNA co-transfected decreased the luminescence (mean ± SEM) of the 5-HT 2C R-NLuc:5-HT 2A R-CLuc (p < 0.05 vs. control; Fig 2A). A main effect of the WT 5-HT 2A R transfection condition on maximum light counts/sec was detected [F 4, 12 = 68.83; p < 0.0001; Fig 2B]; the native WT 5-HT 2A R at all amounts of cDNA co-transfected decreased the luminescence of the 5-HT 2C R-NLuc:5-HT 2A R-CLuc (p < 0.05 vs. control; Fig 2B) A main effect of the empty vector transfection condition on maximum light counts/sec was detected [F 4, 11 = 4.034; p < 0.05; Fig 2C]; the empty vector at 2 μg decreased the luminescence of the 5-HT 2C R-NLuc:5-HT 2A R-CLuc (p < 0.05 vs. control; Fig 2C). Taken together, these data indicate that the luciferase complementation between the 5-HT 2A R:5-HT 2C R is not due to a random complementation of the luciferase fragments.
We next employed an MTT cell viability assay to determine if cellular metabolic activity was affected following these various plasmid co-transfections (S1 Fig). A one-way ANOVA indicated a main effect of experimental condition on absorbance (F 13,28 = 12.34; p < 0.0001); the positive control H 2 O 2 , which increases reactive oxygen species causing cell death, significantly suppressed absorbance, indicating a decrease in cell viability (S1A Fig

Proximity ligation analysis indicates WT 5-HT 2A R:5-HT 2C R interaction within 40 nm
We tested the hypothesis that the 5-HT 2A R and 5-HT 2C R interact using a proximity ligation assay (PLA) conducted in a cell line stably expressing both receptors (5-HT 2A+2C R-CHO cells). In the PLA, when proteins of interest are in complex within close spatial proximity ( 40 nm), short DNA strands complement, ligate and allow rolling circle amplification to produce fluorescent signals [28,42,46,47] (Fig 3A). The PLA was employed to identify and confirm expression of the 5-HT 2A R (S2A Fig, top left) and 5-HT 2C R (S2A Fig, top right) protomers in the dual-expressing cell line. Reverse transcription of RNA followed by qRT-PCR confirmed that 5-HT 2A+2C R-CHO cells expressed both 5-HT 2A R mRNA (ΔC t = 7.67 ± 0.10) and 5-HT 2C R mRNA (ΔC t = 14.18 ± 0.13), but did not express 5-HT 2B R mRNA (crossing threshold not determined). Similar to the transiently transfected HEK293 cells employed herein, the 5-HT 2A R transcript was in greater abundance relative to the 5-HT 2C R transcript in the stably expressing 5-HT 2A+2C R-CHO cells. Experimental controls in which the appropriate primary antibody was omitted and the PLA probe included (e.g., rabbit+/rabbit-; mouse+/mouse-) demonstrated no puncta, as expected (S2A Fig, bottom). The total number of puncta specific to 5-HT 2A R (p < 0.05; S2B Fig) or 5-HT 2C R (p < 0.05; S2B Fig) was significantly higher relative to controls. We then quantified 5-HT 2A R and 5-HT 2C R co-localized in the native environment of the dual expressing 5-HT 2A+2C R-CHO cells. We observed a distinct positive signal (red puncta spot) indicating that the 5-HT 2A R and 5-HT 2C R are in close proximity in the dual expressing cell line (Fig 3B, top). To provide sufficient controls and to ensure that observed fluorescent signal was caused by a receptor:receptor interaction, and not due to nonspecific binding of the antibodies, cells in separate wells were labeled no primary antibody, the anti-5-HT 2A R antibody or the anti-5-HT 2C R antibody alone (Fig 3B). A main effect of experimental conditions for the total number of puncta in the 5-HT 2A+2C R-CHO cells was detected [F(3, 16) = 22,59; p < 0.05; Fig 3C]; the total number of puncta was significantly higher vs. all provided control conditions (p < 0.05; Fig 3C), indicating specificity of the antibody labeling. These data support the conclusion that WT 5-HT 2A R and 5-HT 2C R interact in cells within 40 nm of each other (limit of detection for PLA) [28,42,46,47].

Discussion
The present study used a tiered strategy employing three complementary biophysical techniques with increasing spatial resolution to confirm that the 5-HT 2A R and 5-HT 2C R form a cellular level protein-protein complex interaction. In the LCA assay, we demonstrated that formation of the 5-HT 2A R:5-HT 2C R complex exists within 50 nm and increases proportionally to the 5-HT 2C R:5-HT 2A R protein expression ratio. Using the PLA, we found that cells stably expressing both the 5-HT 2A R and 5-HT 2C R exhibit 5-HT 2A R:5-HT 2C R heteroreceptor complexes within 40 nm of each other. Lastly, BRET analyses signify a specific and saturable 5-HT 2A R:5-HT 2C R interaction, indicating that the 5-HT 2A R and 5-HT 2C R form a close interaction within 10 nm of each other in intact live cells. Importantly, the bioengineered receptors generated for the LCA and the BRET exhibit 5-HT-mediated intracellular calcium signaling as seen for the native receptors, assuring that the tagged constructs do not impede receptor function. Thus, these studies support the conclusion that a very close 5-HT 2A R:5-HT 2C R heteromeric receptor interaction occurs in cultured cells. Energy transfer-based technologies, such as BRET, have been fundamental in the development of a database of knowledge concerning the homo-and heterodimerization of GPCRs [45,49,50]. Considerable research has implicated that the energy transfer from the donor to the acceptor is below 10 nm [45,49,50]. We demonstrated a specific BRET saturation curve in cells expressing a constant amount of the 5-HT 2A R-RLuc donor construct and increasing amounts of the 5-HT 2C R-eYFP acceptor construct. A lower, more linear BRET saturation curve was seen in cells expressing a constant amount of donor β 2 -AR-RLuc construct and increasing amounts of acceptor 5-HT 2C R-eYFP construct, a receptor pair previously suggested to minimally associate in live cells [48]. Because the diameter of the seven transmembrane helical core is estimated at~5 nm, the observed, positive BRET signal strongly suggests that oligomerization has occurred between the 5-HT 2A R and 5-HT 2C R [45,49,50]. Furthermore, a reciprocal interaction between these two receptors is corroborated by previous analyses of molecular and pharmacological properties of the 5-HT 2A R:5-HT 2C R heterocomplex in vitro [8]. Intriguingly, this formation of the 5-HT 2A R:5-HT 2C R complex did not modify the Gα q/11 coupling of the protomers, but rather the 5-HT 2C R exerts dominance when in complex with the 5-HT 2A R, such that only the 5-HT 2C R couples with the G protein to generate intracellular signaling; the 5-HT 2A R signaling is 'masked' [8]. Thus, the 5-HT 2A R:5-HT 2C R protein complex appears to be a distinct molecular species that contributes to the control of cellular signaling, triggering unique intracellular signaling properties when co-expressed in vitro [8].
The detection of a 5-HT 2A R:5-HT 2C R protein complex in transfected cells [present results, [8] and rodent brain [5] extends observations of the co-localization of these receptors in the single neurons of the rat mPFC [15][16][17]. Our previous studies indicate that the lowest expression levels of the 5-HT 2A R:5-HT 2C R complex assessed by co-immunoprecipitation in the mPFC associates with the highest level of phenotypic impulsive action in the rat and that the ratio of 5-HT 2A R to 5-HT 2C R protein expression in the rat mPFC predicts the inherent level of motor impulsivity in individual rats [5]. Engineered 5-HT 2C R knockdown in mPFC resulted in increased motor impulsivity [5,51] concomitant with elevated 5-HT 2A R expression and greater potency of the 5-HT 2A R antagonist M100907 to suppress motor impulsivity [5]. Intriguingly, rats exposed to a low-protein diet in utero exhibited reduced 5-HT 2C R and elevated 5-HT 2A R expression in the hypothalamus concomitant with impaired sensitivity to 5-HT-mediated appetite in adulthood [52].
Although not yet linked to the formation or action of heteromers, these previous studies support a potential role for the 5-HT 2A R:5-HT 2C R interaction in vivo [11,12].
The oligomerization of GPCRs has emerged as a vital property of receptor structure and function with identified macromolecular complexes comprising of at least two different protomers with biochemical properties that are discernibly distinct from those of the individual protomers [8,45,53]. The main functional GPCR unit for the 5-HT 2A R [20,21] and 5-HT 2C R [22][23][24][25] is proposed to be their homomeric form, and recent studies suggest that different interacting homomers may constitute heteromeric complexes [53]. Future studies are needed to clarify the degree to which downstream signaling pathways are recruited by the individual receptors vs. heteroreceptor complexes, especially in terms of the level of ligand-directed signaling [54] as well as the functional effects of the 5-HT 2 R and their underlying role in behavior. Furthermore, drug discovery efforts selectively targeting either receptor should take into account the formation of a heteromeric complex when analyzing physiological responses [8].