Optimization of Peptide Linker-Based Fluorescent Ligands for the Histamine H1 Receptor

The histamine H1 receptor (H1R) has recently been implicated in mediating cell proliferation and cancer progression; therefore, high-affinity H1R-selective fluorescent ligands are desirable tools for further investigation of this behavior in vitro and in vivo. We previously reported a H1R fluorescent ligand, bearing a peptide-linker, based on antagonist VUF13816 and sought to further explore structure–activity relationships (SARs) around the linker, orthostere, and fluorescent moieties. Here, we report a series of high-affinity H1R fluorescent ligands varying in peptide linker composition, orthosteric targeting moiety, and fluorophore. Incorporation of a boron-dipyrromethene (BODIPY) 630/650-based fluorophore conferred high binding affinity to our H1R fluorescent ligands, remarkably overriding the linker SAR observed in corresponding unlabeled congeners. Compound 31a, both potent and subtype-selective, enabled H1R visualization using confocal microscopy at a concentration of 10 nM. Molecular docking of 31a with the human H1R predicts that the optimized peptide linker makes interactions with key residues in the receptor.


■ INTRODUCTION
The histamine H 1 receptor (H 1 R), a class A G protein-coupled receptor (GPCR), is one of four histamine receptor subtypes found in humans. The H 1 R predominantly couples to Gα q/11 proteins and upon receptor activation progresses mainly through the inositol-phospholipid-dependent pathway with a subsequent increase in intracellular Ca 2+ concentration. The H 1 R is widely distributed throughout the body, particularly in immune cells, the lungs, blood vessels, and the central nervous system. 1,2 The H 1 R has an established role in allergy and inflammation and is the therapeutic target for antihistamines (e.g., loratadine and cetirizine) in the treatment of allergic rhinitis, hayfever, and urticaria. There is however a growing body of evidence suggesting that histamine receptors, including the H 1 R, play an important role in cancer progression due to their involvement in cell proliferation and tumor growth. 3,4 Activation of histamine receptors has been shown to both promote or suppress tumor growth depending on the tumor microenvironment, histamine metabolism, the receptor subtypes involved, and the local histamine receptor balance. 3−6 Given the complexity of histamine cancer pharmacology, a better understanding of H 1 R pharmacology in various tumor microenvironments could facilitate the development of H 1 Rtargeted cancer therapeutics. 7−9 Fluorescent ligands for GPCRs are synthesized by attaching a fluorophore to a receptor-targeting moiety or "orthostere" (an agonist or antagonist ligand) for the receptor of interest. A linker is usually necessary to separate the fluorophore from the orthostere to avoid disruption of the key ligand−target interactions that occur in the parent ligand binding pocket. 10−13 Fluorescent ligands have proven themselves to be valuable tools in the study of GPCR pharmacology as they can provide insight into ligand−receptor binding and kinetics, 13−25 receptor localization and internalization, 20,21,26 organization and oligomerization 27−34 at the single-cell level using primarily confocal microscopy, and resonance-energytransfer techniques, both in vitro and in vivo. 35,36 In addition, it has been recently demonstrated that fluorescent ligands or small molecule ligands in general could be utilized to fluorescently label endogenous GPCRs expressed in living tissues using ligand-directed chemistry. 37−39 As such, fluorescent ligands bear the potential to probe receptor behavior, expression, density, and organization in primary cell cultures derived from specific cancers. To date, several fluorescent ligands for the H 1 R have been reported, many of which presented several issues (e.g., high levels of non-specific binding and cellular uptake) that affected their utility for pharmacological studies. 25,28,40 Previously, work within our group led to the discovery of the fluorescent ligand 1 25 ( Figure  1), which is based on the antagonist H 1 R ligand VUF13816 41 consisting of a BODIPY 630/650-X fluorophore and an Ala-Ala-Ala tripeptide linker. Compound 1 demonstrated improved physicochemical and imaging properties compared to a previously reported mepyramine-based fluorescent ligand, 28 which were attributed to the incorporation of a peptide linker. 25 Other work within our group has demonstrated the peptide linker strategy to be successful in converting a nonselective adenosine receptor antagonist into a high-affinity adenosine A 3 receptor-selective fluorescent probe. 42 Recent advances in GPCR structural biology have provided a platform for more rational design of drugs targeting this protein class, and we reasoned that this could also be extended to the rational design of fluorescent ligands for studying GPCR pharmacology. 43 The crystal structure of the antagonist doxepin bound to the human H 1 R (PDB code: 3RZE) was published in 2011 and revealed several basic residues at the entrance of the binding pocket, namely, K179 ECL2 (Ballesteros−Weinstein nomenclature 44 ), K191 5.39 , and H450 7. 35 . It was proposed that these residues played a role in interacting with the carboxylic acid group present in second-generation zwitterionic antihistamines (e.g., cetirizine and acrivastine). 45 We were attracted to the possibility of synthesizing highaffinity H 1 R-selective fluorescent ligands through targeting these residues, in addition to exploring additional receptor− ligand interactions by probing the outer vestibule of the receptor binding pocket. We envisioned that this could be achieved by optimizing the composition of the peptide-linker moiety to engage in binding interactions with residues lining the ligand entry route when the fluorescent ligand is bound to the receptor. The detection sensitivity offered using high affinity fluorescent ligands opens up the potential for their use in physiological and diseased tissue (which exhibit much lower cell surface receptor expression compared to engineered cell lines in both in vitro and in vivo paradigms). Furthermore, this could be achieved with very low concentrations of fluorescent ligand, thus reducing non-specific binding and improving the signal-to-noise ratio.
In this study, we report the design, synthesis, and pharmacological characterization of a series of high-affinity H 1 R fluorescent ligands comprising a peptide linker of varying composition and explore two orthosteres. The fluorophores were attached to the congeners, utilizing either coppercatalyzed alkyne-azide cycloaddition (CuAAC) or amide coupling. Our efforts led to the discovery of 31a, which exhibits nanomolar affinity at the H 1 R, is >100-fold selective for H 1 R over the H 3 R and H 4 R subtypes, and provides the capability to visualize cells at a low ligand concentration (10 nM). Although the linker SAR was established in the unlabeled congeners, further investigation revealed that the nature of the fluorophore itself plays a significant role in determining the overall affinity of our series of fluorescent ligands, which appears to override the linker SAR observed in our library of congeners.

■ RESULTS AND DISCUSSION
Chemistry. Our primary objective was to synthesize highaffinity peptide linker-based H 1 R fluorescent ligands by optimizing the linker moiety, using CuAAC for fluorophore conjugation ( Figure 2). The use of CuAAC further broadens the scope of this approach by offering orthogonal reaction conditions, allowing incorporation of amino acids into the linker whose side chains (e.g., amine-and carboxylic acidcontaining side groups) might otherwise be reactive toward electrophilic fluorophore labeling reagents or interfere with coupling chemistries. This strategy allows us to fully explore peptide linker SARs with minimum side-chain cross-reactivities at the fluorophore conjugation step and achieve high efficiency in overall fluorescent ligand synthesis.
Taking compound 1 as a starting point, we sought to further optimize the composition of the tripeptide linker in the context of a triazole linkage to the fluorophore. We approached this by conducting an initial screening study using unlabeled peptide congeners performed in a systematic manner to determine a more optimal peptide linker. For the purposes of more facile SAR exploration, we chose to emulate the presence of the triazole linking moiety with a phenylalanine residue, serving as a surrogate aryl system uncomplicated by functional reactivity. At the time of the study, the 3-azido-alanine that would ultimately be installed as the terminal linker residue was not commercially available, while the enantiomer 3-azido-D-alanine was. With this in mind, we opted to use D-Phe as a surrogate for the corresponding triazole that would feature in the final fluorescent ligands.
During the study, the required N-Fmoc-3-azido-D-alanine, which was intended for use in the synthesis of the final fluorescent ligands, ceased to be commercially available; however, the enantiomer N-Fmoc-3-azido-alanine had returned to commercial availability. As we had carried out the congener SAR study using a D-Phe surrogate, we deemed it prudent to also synthesize the corresponding L-Phe bearing epimers of the best analogues, to evaluate whether the stereochemistry at this distal point of the linker was likely to play a role in determining affinity. With this in mind, 11a,b (Scheme 2) were synthesized as the L-Phe epimers of 10d,e (the highest affinity congeners, Table 1).
Synthesis of CuAAC-Coupled H 1 R Fluorescent Ligands. We proceeded with synthesis of fluorescent ligands using CuAAC, incorporating the optimized peptide linker identified from the above congener screening study. To achieve this, we synthesized a novel, alkyne-functionalized fluorophore based on BODIPY 630/650 (20) via a twelve-step synthesis route as outlined in Scheme 3. 2-Formylpyrrole (12) was sequentially reacted with Boc-anhydride and trimethylorthoformate to protect the pyrrole N−H and aldehyde groups, respectively, to give 13. Boronation at C5 via triisopropyl borate in the presence of lithium diisopropylamide and subsequent acidmediated deprotection using sodium hydrogen sulfate yielded the boronic acid product 14, which was then subjected to a Suzuki reaction with 2-bromothiophene to afford 15.
Tributylphosphine was added to para-methoxybenzyl chloride (16) and the mixture subsequently stirred in anhydrous toluene to give the phosphonium salt (17), which was then treated with NaOH and subjected to a Wittig reaction with 2-formylpyrrole at 100°C under microwave conditions to afford 18. Demethylation of the methoxy group was achieved in the presence of NaSEt to yield the corresponding phenol, which was subsequently alkylated with propargyl bromide, affording 19. Compound 15 and 19 were then stirred in anhydrous dichloromethane (DCM) in the presence of POCl 3 overnight and subsequently treated with an excess of DIPEA and BF 3 ·Et 2 O to afford the BODIPY fluorophore 20 as a red iridescent solid.
As described earlier, inspection of the crystal structure of doxepin bound to the human H 1 R (PDB code: 3RZE) reveals basic residues (K179 ECL2 , K191 5.39 , and H450 7.35 ) lining the entrance to the orthosteric binding site. Preliminary molecular docking studies (Supporting information Figure S1) suggested that incorporating an acidic amino acid residue at AA 2 might permit salt bridge formation with K179 ECL2 and K191 5.39 ; however, improvement in affinity was not achieved with 9i (AA 1 = Val, AA 2 = Asp, AA 3 = Ala) in our congener screen (Table 1). We decided to pursue this hypothesis further by incorporating glutamate at AA 2 to probe for the desired salt bridge interaction, which was potentially achievable given the flexibility associated with the additional methylene spacer within the glutamate side group compared to the aspartate side group. Three analogues (21a−c) bearing either the Val-Ser-Asn (21a), linker which was determined to be the optimum peptide linker from the congener screen, Ala-Ala-Ala (21b) linker as a control, or Val-Glu-Asn (21c) linker were synthesized using SPPS and conjugated to 20 using CuAAC to afford their corresponding fluorescent ligands (22a-c) (Scheme 4).
The three main components of a fluorescent ligand, i.e., orthostere, linker, and fluorophore, may exert an effect on fluorescent ligand binding affinity. 47 To this end, we conducted a series of investigations to understand the effects of each of these components on the binding affinity of our H 1 R fluorescent ligands. We hypothesized that doxepin, which possessed subnanomolar affinity for H 1 R (radioligand binding assay pK i = 9.75), 48 when incorporated into fluorescent ligands as the orthostere could improve overall ligand affinity at H 1 R relative to VUF13816 (radioligand binding assay pK i = 8.20). 48 The initial series of fluorescent ligands (22a−c) were considerably shorter in overall length in comparison to our previously reported probe (1). As such, we sought to synthesize analogues, which were more comparable in size to 1, through addition of a spacer between AA 3 and the triazole link to the fluorophore. Congeners (27a−d) were synthesized using SPPS on Rink amide resin, initiated by resin loading with Fmoc-5-azido-L-norvaline-OH (Fmoc-Orn(N 3 )-OH) and the  subsequent incorporation of β-alanine, the required tripeptide, and final capping with the orthostere. The propargyl-bearing fluorophore (20) was conjugated to congeners 27a−d by CuAAC in the presence of 2,6-lutidine and a Cu(I) catalyst generated in situ through the reduction of Cu(II) by sodium ascorbate, to afford four fluorescent ligands (28a−d) with a linker length similar to that of 1 (Scheme 6). Compounds 28c,d were pharmacologically characterized as a mixture of E/ Z isomers (79:21), arising from the doxepin-based orthostere (26).
Synthesis of Amide-Coupled H 1 R Fluorescent Ligands. Next, we compared our triazole-linked approach to fluorophore installation with a selection of corresponding amidelinked analogues (Scheme 7). The peptide linker selection was based upon the amino acids previously observed to provide the greatest improvement in congener affinity, namely, AA 1 = Val, AA 2 = Ser or Ala, and AA 3 = Asn or Tyr. We opted to explore these in all possible combinations as well as AA 1 = Phe, AA 2 = Ala, and AA 3 = Ala (corresponding to 8d) as a control. As significant improvement in congener affinity was only observed with the linker optimization at AA, 1 we sought to explore the SAR at AA 1 in further detail, through the incorporation of a hydrophilic Thr and the sterically bulkier Ile and Leu, which had not been previously explored at this position. The congeners (29a−k) were assembled on Rink amide resin via SPPS, starting with Fmoc-Lys(Boc)-OH for resin loading (Nε-Boc-protection masking the fluorophore attachment point), followed by β-alanine, the required tripeptide, and capping with 7. Compounds 29a−k were conjugated to a carboxylic acid-functionalized BODIPY 630/650-based fluorophore 49 (30) in the presence of the coupling reagent 1-[bis-(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium-3-oxide hexafluorophosphate (HATU) and DIPEA to afford fluorescent ligands 31a−k.
Our final area of SAR exploration was to investigate how the structure of fluorophore and linker moieties affected overall ligand binding affinity at the H 1 R by conjugating 29a bearing the optimized linker (AA 1 = Val, AA 2 = Ser, AA 3 = Asn) and 29k bearing an unoptimized linker (AA 1 = Phe, AA 2 = AA 3 = Ala) as controls, to six different fluorophores. We sought to explore the SAR relating to the BODIPY scaffold and also to compare lipophilic and hydrophilic fluorophores with similar absorption/emission profiles. The panel of commercially available N-reactive (as the N-hydroxysuccinimide (NHS) ester) fluorophores included those incorporating a flexible hexanoyl linker: BODIPY FL-X (green-emitting, lipophilic), BODIPY 630/650-X (red-emitting, lipophilic), and sulfo-Cyanine 5 (sulfo-Cy5, water soluble, red-emitting) and the corresponding BODIPY-FL (green-emitting, lipophilic), which lacks the hexanoyl linker. In order to compare the structures of BODIPY 630/650 and BODIPY-FL more effectively, we additionally designed and synthesized hybridized versions of these fluorophores: BODIPY A (35) and BODIPY B (37) from 32 and 15, respectively (Schemes 8 and 9).
Compound 18 was subjected to demethylation with NaSEt before treating it with K 2 CO 3 and alkylation with methyl bromoacetate to give the ester 32. Condensation of 32 and 3,5dimethyl-1H-pyrrole-2-carbaldehyde (33) in the presence of POCl 3 and subsequent treatment with BF 3 ·Et 2 O afforded the BODIPY ester 34, which upon acid hydrolysis in the presence of H 3 PO 4 afforded the free acid BODIPY A (35).
BODIPY B (37) was synthesized in a similar manner with the condensation of 15 and methyl 3-(1H-pyrrol-2-yl)-  propanoate (36, followed by mild acidic hydrolysis of the resulting ester in the presence of H 3 PO 4 to afford the free acid 37. Finally, 29a and 29k were individually coupled to each fluorophore to afford twelve fluorescent ligands 38a−f and 39a−f (Scheme 9). The BODIPY A-based (38b, 39b) and BODIPY B-based (38c, 39c) fluorescent ligands were spectrally characterized on a FlexStation3 plate reader (Molecular Devices, Sunnyvale, CA) and their absorption/ emission (λ max , ex /λ max , em ) spectral profiles were determined to be 570 nm/586 nm and 560 nm/570 nm, respectively ( Figures  S2−S5). A figurative summary of all synthesized fluorescent compounds is included in the Supporting Information ( Figure  S6).
Pharmacology. Binding Affinity Evaluation of Peptide Congeners (8a−j, 9a−i, 10a−i, 11a,b). It has been previously reported that NanoBRET-based binding assays performed on Nanoluciferase-tagged human H 1 R (Nluc-H 1 R)expressing HEK293T cells produced binding affinity values (pK i ) comparable to those obtained from radioligand binding assays as well as pK B values obtained from Ca 2+ mobilization assays. 25 As such, NanoBRET-based assays performed on Nluc-H 1 R-expressing HEK293T cells were selected as the primary assay for the pharmacological characterization of the unlabeled peptide congeners and the final fluorescent conjugates in this study. NanoBRET assays have several advantages over traditional radioligand binding assays includ-  ing ease of performing the assay and the lack of safety issues associated with disposal of radioactive material. 13−17 The binding affinity of 8a−j, 9a−i, and 10a−i at the H 1 R was determined by competition with 1 at Nluc-H 1 R expressed on HEK293T cells in a whole cell NanoBRET equilibrium binding assay ( Table 1). The affinities (K i ) ranged from 95 nM (10d/10f) to 2399 nM (9h), indicating that the composition of the peptide linker does affect affinity. The analogues with the highest affinity 10d (AA 1 = Val, AA 2 = Ser, AA 3 = Tyr; K i = 95 nM) and 10f (AA 1 = Val, AA 2 = Ser, AA 3 = Asn; K i = 95 nM) conferred a 10-fold improvement in affinity compared to 8a (AA 1 = AA 2 = AA 3 = Ala; K i = 912 nM), which served as the starting point for our optimization effort. These observations provided confidence that the synthesis of a fluorescent ligand with a higher binding affinity at H 1 R compared to 1 could be achieved through peptide linker optimization. Although we observed a stepwise, gradual improvement in affinity progressing along the three series, the greatest improvement in affinity from the linker optimization effort was attributed to the incorporation of Val at AA 1 (8c, AA 1 = Val, AA 2 = AA 3 = Ala; K i = 158 nM), which provided a 6-fold improvement in affinity over 8a. Optimization efforts at AA 2 (from 8c to 9e) and AA 3 (from 9e to 10d and 10f) did not significantly improve affinity (less than 2-fold) as indicated by the lack of significant leftward shifts in the position of displacement curves ( Figure 3). In addition, we observed that the configuration of the terminal phenylalanine in our set of congeners had a context-dependent effect on overall affinity. In the case of 11b (AA 1 = Val, AA 2 = Ser, AA 3 = Tyr; K i = 437 nM), H 1 R affinity was 5-fold lower than that of the D-Phe epimer 10d (K i = 95 nM). In contrast, 11a (AA 1 = Val, AA 2 = Ser, AA 3 = Asn; K i = 89 nM) and its epimer 10f (K i = 95 nM) had comparable affinity. The lower affinity of 11b compared to its epimer, 10d may be explained by unfavorable intramolecular interactions that might occur in one epimer due to the adjacent aromatic groups present in AA 3  . Baseline-corrected displacement curves for 8a, 8c, 9e, 10d, and 10f determined by a NanoBRET competition assay on live Nluc-H 1 R expressing HEK293T cells treated with 25 nM 1 as the fluorophore-labeled competitive ligand with increasing concentrations of the unlabeled congeners. A gradual leftward shift of the displacement curve relative to that of 8a was observed upon linker optimization at AA 1 (8c), AA 2 (9e), and AA 3 (10d and 10f). Linker optimization at AA 1 (8a to 8c) provided the greatest increase in affinity, whereas linker optimization at AA 2 (8c to 9e) and AA 3 (9e to 10d and 10f) did not significantly improve affinity. Data were normalized to maximal BRET signal obtained in the absence of labeled competitive ligand (1) and data shown represent the combined mean ± SEM of three experiments performed in triplicate.
(Tyr) and Phe. Given the later incorporation of a lipophilic fluorophore, we opted to proceed with only the Val-Ser-Asn linker being incorporated into CuAAC-conjugated H 1 R fluorescent ligands. The presence of Asn at AA 3 rather than Tyr confers improved overall physicochemical properties to the molecule in the context of the significant lipophilicity ultimately added by the fluorophore.
Binding Affinity Evaluation of CuAAC-Coupled Fluorescent Ligands (22a−c, 28a−d). The binding affinity of fluorescent ligands 22a−c and 28a−d was determined by a NanoBRET saturation binding assay performed on Nluc-H 1 R expressed in HEK293T cells, whereby all seven fluorescent ligands displayed concentration-dependent, saturable and reversible binding to Nluc-H 1 R ( Figures S7 and S8). It was interesting to observe that the SAR determined in the linker optimization study did not translate through in the context of the full CuAAC-coupled fluorescent ligand as compounds 22a, 28a, and 28c (all with AA 1 = Val, AA 2 = Ser, AA 3 = Asn) did not exhibit a significant improvement in binding affinity compared to their corresponding control compounds 22b, 28b, and 28d, respectively (AA 1 = AA 2 = AA 3 = Ala) ( Table  2). Moreover, doxepin-based fluorescent ligand 28d (AA 1 = AA 2 = AA 3 = Ala; K D = 5 nM) exhibited an affinity that was 4fold higher than 28c (AA 1 = Val, AA 2 = Ser, AA 3 = Asn; K D = 19 nM) at the H 1 R.
When comparing ligands with a short spacer between the tripeptide sequence and the fluorophore (22a and 22b) with ligands bearing the same tripeptide sequence, but a longer spacer (28a and 28b respectively), the difference in spacer length had no discernible effect on overall ligand affinity. Similarly, comparing ligands with a constant tripeptide linker and spacer length, but varying the orthostere from VUF13816 (28a-b) to doxepin (28c-d), had no significant effect on overall H 1 R affinity.
This indicates that the fluorescent ligands, which differ from their corresponding unlabeled peptide congeners in terms of their overall structure, size, and physicochemical properties, can present divergent SAR from the congener screen. This is in agreement to previous work by ourselves and others, whereby fluorescent ligands or their congeners must be treated as new pharmacological entities in their own rights rather than simple extensions of the original orthostere. 13,42,47,50−53 Our efforts to engage the three basic residues at the entrance of the binding pocket (K179 45.49 , K191 5.39 , and H450 7.35 ) via salt bridge interactions using acidic side-chain bearing peptide linkers did not confer the anticipated improvement to congener affinity as seen in the case of 8j (AA 1 = Asp; AA 2 = AA 3 = Ala; K i = 1023 nM), 9i (AA 1 = Val, AA 2 = Asp, AA 3 = Ala; K i = 550 nM), and 10i (AA 1 = Val, AA 2 = Ser, AA 3 = Asp; K i = 174 nM). This was also mirrored in the fluorescent ligands, as switching AA 2 from the polar, neutral serine in 22a (AA 1 = Val, AA 2 = Ser, AA 3 = Asn; K D = 28 nM) to the acidic glutamate in 22c (AA 1 = Val, AA 2 = Glu, AA 3 = Asn; K D = 162 nM) led to a 6-fold loss in affinity. We therefore concluded that it was not possible to improve fluorescent ligand affinity through the acidic linker approach.
Although there were initial concerns regarding the CuAACderived triazole interfering with fluorescent ligand binding, our results have demonstrated that CuAAC is a feasible approach towards fluorophore conjugation with the synthesis of high affinity H 1 R fluorescent ligands (K D values ranging from 6 to 31 nM). Therefore, CuAAC should be considered as an Table 2. Binding Affinity of 22a-c and 28a-d at Nluc-H 1 R Expressed in HEK293T Cells a a alternative fluorophore conjugation strategy in the synthesis of fluorescent ligands, especially where the orthostere or linker moieties bear nucleophilic moieties liable to partake in undesirable side reactions when using a classical N-reactive fluorophore approach to generate an amide bond linkage. Furthermore, while protective group strategies could be considered in this situation, their removal following fluorophore conjugation is not always feasible without detriment to fluorophore stability.
Binding Affinity Evaluation of Amide-Coupled Fluorescent Ligands (31a−k, 38a−f, 39a−f). The binding affinity of 31a−k, 38a−f, and 39a−f was determined by a NanoBRET saturation binding assay performed on Nluc-H 1 R expressed in HEK293T cells, whereby all twenty-three fluorescent ligands displayed concentration-dependent, saturable and reversible Binding affinity (pK D /K D ) determined from saturation binding curves in A NanoBRET saturation binding assay conducted in Nluc-H 1 R expressing HEK293T cells. The data shown are the means ± SEM from four separate experiments.  Binding affinity (pK D /K D ) determined from saturation binding curves in NanoBRET saturation binding assay conducted in Nluc-H 1 R expressing HEK293T cells. The data shown are the means ± SEM from four separate experiments. b Binding affinity (pK B /K B ) calculated using the Gaddum equation from the shift in EC 25 obtained from the histamine stimulated dose-response curve in the presence of 1 μM of the antagonist obtained from a Ca 2+ mobilization assay performed on human H 1 R-expressing CHO cells. The data shown are the means ± SEM from four separate experiments. '-' = not determined. c cLogP values of the fluorophore moiety up to and including the amide bond which serves as the point of attachment, determined from ChemBioDraw 19.1. binding to Nluc-H 1 R ( Figures S9-S12). For 31a−k, K D values ranged from 4−16 nM, representing a 4-fold difference overall (Table 3). Compound 31a (AA 1 = Val, AA 2 = Ser, AA 3 = Asn; K D = 4 nM) showed a 3-fold improvement in binding affinity over 31k (AA 1 = Phe, AA 2 = AA 3 = Ala; K D = 12 nM), considerably lower than the 14-fold increase in binding affinity observed between the corresponding congeners 10f (AA 1 = Val, AA 2 = Ser, AA 3 = Asn; K i = 95 nM) and 8d (AA 1 = Phe, AA 2 = AA 3 = Ala; K i = 1318 nM). This provides further evidence that the addition of a fluorophore to the congener seems to override the SAR observed in the context of the congeners alone.
Previous work on fluorescent ligands has shown that the linker and the fluorophore could affect the pharmacology of the final fluorescent conjugate. 47 In this study, we observed that the affinities of our peptide congeners series (pK i = 5.62− 7.05) were at least 10-fold lower when compared to VUF14454 (radioligand binding assay pK i at H 1 R = 8.2), 54 an N-methylated VUF13816 analogue, which retains a tertiary amine structure as seen in our VUF13816-based orthostere. This suggests that linker incorporation has a detrimental effect to overall ligand affinity, in contrast to our initial intention to improve ligand−receptor interaction and thus affinity via an optimized linker. Indeed, affinity could not be improved solely based on ligand-receptor interaction without the consideration of potentially unfavorable ligand binding orientation within the receptor and/or desolvation energy, which might be the case for our peptide congeners. In addition, we also observed a large increase in the binding affinity of fluorescent ligands compared to their corresponding congeners. For example, compound 31a (AA 1 = Val, AA 2 = Ser, AA 3 = Asn; K D = 4 nM) and 31k (AA 1 = Phe, AA 2 = AA 3 = Ala; K D = 12 nM) had significantly higher binding affinities at H 1 R compared to their corresponding congeners 10f (AA 1 = Val, AA 2 = Ser, AA 3 = Asn; K i = 95 nM) and 8d (AA 1 = Phe, AA 2 = AA 3 = Ala; K i = 1318 nM), respectively, with a difference of more than 100-fold seen between 31k and 8d. This observation suggests the fluorophore itself has a key role to play in determining overall ligand binding affinity and is not merely a passive bystander moiety in this regard. In addition, in our previous development of BODIPY 630/650-based fluorescent ligands at a range of receptors, we have observed that this particular fluorophore appears to confer a high overall ligand binding affinity. 42,47,52,55 A clear trend was observed for the binding affinity of fluorescent ligands 38a−f and 39a−f determined through a NanoBRET binding assay (Table 4). Interestingly, fluorescent ligand binding affinity decreased gradually in varying magnitude with decreasing fluorophore lipophilicity (cLogP, Table 4) from BODIPY 630/650-X to Sulfo-Cy5 and this was consistent across 38a−f (AA 1 = Val, AA 2 = Ser, AA 3 = Asn) and 39a−f (AA 1 = Phe, AA 2 = AA 3 = Ala). Once again in the context of the whole fluorescent ligand, where each fluorophore was kept the same, the congener composition had little effect on overall affinity (i.e., 31a compared to 31k and each of 38a−f compared to 39a−f). The fluorescent ligands incorporating the additional hexanoyl linker, namely, BODIPY 630/650-X-containing 38a (K D = 18 nM) and 39a (K D = 10 nM) and BODIPY FL-X-containing 38e (K D = 5248 nM) and 39e (K D = 2692 nM) broadly retained comparable binding affinity to their shorter counterparts without the hexanoyl linker 31a (K D = 4 nM), 31k (K D = 12 nM), 38d (K D = 1288 nM), and 39d (K D = 1445 nM) respectively. However, the relatively small difference between the shorter and longer ligands was more pronounced where AA 1 = Val, AA 2 = Ser, and AA 3 = Asn.
These data are remarkable and indicate that the fluorophore moiety itself has a significant impact on the binding affinity of our H 1 R fluorescent ligands irrespective of the peptide linker composition, with the BODIPY 630/650 fluorophore significantly contributing toward the high binding affinity of our series of fluorescent ligands.
BODIPY FL-based fluorescent ligands (38d-e and 39d-e) and their corresponding Sulfo-Cy5-based fluorescent ligands (38f and 39f) had similarly low binding affinity at Nluc-H 1 R (K D < 1 μM), despite large differences in terms of the physicochemical properties and size of their respective fluorophores in which the BODIPY FL fluorophore is small and lipophilic, whereas the Sulfo-Cy5 fluorophore is relatively larger and hydrophilic. This might suggest that the lipophilic contribution of the fluorophore alone does not sufficiently account for observed increases in affinity. Fluorophore size and possible shape also appear to be important factors; thus, it seems less likely that non-specific interactions are governing the observed fluorophore SAR (e.g., simple insertion of the fluorophore into the lipophilic membrane environment). It may be that the fluorophore is able to make specific binding interactions with parts of the receptor − a concept that has not been extensively explored previously. Accordingly, we sought to rationalize the observed SAR through molecular docking studies (see section below and Figures 8−10). It is interesting to note that the observed effects of the fluorophore on the fluorescent ligand binding affinity was not observed in a study involving six adenosine A 2A receptor-selective fluorescent ligands, which exhibited affinities (K D ) ranging from 20 to 83 nM despite significant differences in their fluorophores (BODIPY 630/650, BODIPY 630/650-X, BODIPY FL, BODIPY FL-X, Sulfo-Cy5, Alexa Fluor 647). 56 This suggests that the effects of the fluorophore on fluorescent ligand binding affinity cannot be assumed and are not translatable to ligands targeting all GPCRs.
We conducted a secondary assay to determine the binding affinity of ten selected fluorescent ligands (31a, 31k, 38b−d, 38f, 39b−d, 39f) in a Ca 2+ mobilization assay performed on H 1 R-expressing Chinese hamster ovary (CHO) cells, in order to confirm our observation in a functional paradigm. All 10 fluorescent ligands shifted the histamine-mediated concentration−response curves rightward, indicating receptor antagonism ( Figure 4). Maximum Ca 2+ response (E max ) was not achieved in the presence of all fluorescent ligands due to the transient nature of the agonist response and the consequent non-equilibrium kinetics of the calcium mobilization assay. 57 As such, EC 25 values were used instead of EC 50 to estimate the binding affinity (K B ) of the fluorescent compounds. 25,58,59 K B values were in close agreement to K D values obtained from NanoBRET saturation assay (K D ), with the exception of BODIPY-FL containing fluorescent ligands 38d and 39d. This provides convincing evidence that the BODIPY 630/650 fluorophore is a significant contributing factor toward the synthesis of high binding affinity H 1 R fluorescent ligands (Table 4).

H 1 R Selectivity against the H 3 R And H 4 R Subtypes and
Binding Kinetics Profile at H 1 R of 31a and 31k. Target selectivity is one of many desirable traits fluorescent ligands must possess for their wider pharmacological applications in both in vitro and in vivo paradigms. To this end, we sought to define the H 1 R selectivity of 31a and 31k against the H 3 R and H 4 R subtypes. As ligand binding is often driven by the orthostere, understanding the selectivity of the parent orthostere is crucial to fully defining the H 1 R selectivity of our fluorescent ligands. As such, the H 1 R selectivity profile of the VUF13816-based orthostere within our series of fluorescent ligands was determined by NanoBRET competition binding assay at H 3 R and H 4 R using VUF14454 (radioligand binding assay pK i at H 1 R = 8.2). 54 NanoBRET competition binding assays involving the displacement of the commercially available clobenpropit-based BODIPY 630/ 650 TM H 3 R/H 4 R fluorescent ligand CA200843 (HelloBio, Bristol UK) separately from Nluc-H 3 R and Nluc-H 4 R expressing HEK293T cell homogenates by increasing concentrations of VUF14454 showed that VUF14454 did not bind to both receptors at concentrations of up to 10 μM (Supporting Information Figure S13), indicating high H 1 R selectivity (> 1000-fold over H 3 R and H 4 R). Subsequently, fluorescent ligand selectivity for the H 1 R over the H 3 R and H 4 R subtypes was determined by a NanoBRET saturation binding assay performed using 31a (AA 1 = Val, AA 2 = Ser, AA 3 = Asn) and 31k (AA 1 = Phe, AA 2 = AA 3 = Ala) on cell homogenates prepared from Nluc-H 3 R and Nluc-H 4 R expressing HEK293T cells. CA200843 was included in the assay on the same assay plate as a positive control for both H 3 R and H 4 R binding experiments. Compound 31a and 31k showed little to no binding at both H 3 R and H 4 R at concentrations of up to 1 μM, whereas CA200843 displayed concentration-dependent, saturable and reversible binding to Nluc-H 3 R and Nluc-H 4 R ( Figure 5). The determined binding affinity of CA200843 was in close agreement (difference of 3-fold) to values reported in existing literature 23 (Table 5). This confirms that both 31a and 31k bind selectively to H 1 R over the H 3 R/H 4 R subtypes with more than or approximately 100-fold selectivity respectively and that the VUF13816-based orthostere serve as the main contributing factor for the H 1 R selectivity of 31a and 31k, with limited influence from the peptide linker.
As compounds with similar binding affinities may have different binding kinetics at the target receptor, we were interested to understand the effect of peptide linker variation on the binding kinetics of 31a and 31k at H 1 R. This was achieved through measurement of NanoBRET association kinetics performed on Nluc-H 1 R expressed in HEK293T cells which involved real-time measurement of fluorescence/ luminescence signal upon fluorescent ligand addition to the assay plate. Compounds 31a and 31k displayed similar binding kinetics in terms of on-rate (k on ), off-rate (k off ) and both possess mean residence times (MRT) of more than 1 hour, thus the composition of the peptide linker seems to have little effect on the overall binding affinity, subtype selectivity profile or binding kinetics of our peptide linker-based H 1 R fluorescent probes (Table 6, Figure 6). The long residence time of 31a and 31k also explains their susceptibility (due to hemi-equilibrium issues) to produce submaximal responses in the calcium mobilization assays.
Confocal Microscopy Studies. Compound 31a (AA 1 = Val, AA 2 = Ser, AA 3 = Asn) was subsequently used for visualizing H 1 R at the single cell level using confocal microscopy to study the nature of the ligand binding and to determine whether cellular uptake occurred. Varying concentrations of 31a (1 nM, 5 nM, 10 nM and 25 nM) were incubated for 30 min with CHO cells expressing yellow fluorescence protein (YFP)-tagged H 1 R at 37°C in the presence and absence of 10 μM mepyramine pre-treatment to determine non-specific binding. Confocal imaging revealed that 31a was localized to the plasma membrane of the cells with almost no intracellular uptake across all concentrations tested. These was attributed to the antagonist nature of 31a (thus not promoting receptor internalization), as well as the peptide linker which reduced ligand lipophilicity and limits the ability of the membrane permeability of the fluorescent ligand. Receptor binding was specific as shown in the overlay of the green and red channels as well as the lack of signal in the red channel in the wells containing a high concentration of mepyramine ( Figure 7). Compound 31a allowed visualization of receptors on CHO cells expressing H 1 R at a concentration of 10 nM on a confocal microscope with limited non-specific signals to produce confocal images with good signal-to-noise ratio.
Molecular Docking Studies. In order to understand ligand-receptor interaction at the structural level, molecular docking studies were performed with 31a (AA 1 = Val, AA 2 = Ser, AA 3 = Asn) and the energy-minimized H 1 R crystal structure (PDB code: 3RZE, with the ligand doxepin bound) 45 using Glide within the Schrodinger Molecular Modelling Suite. The only binding pose which formed the canonical salt bridge interaction between the protonated piperidine nitrogen present in the orthostere and D107 3.32 as well as having good overlay between the orthostere and the co-crystallized ligand doxepin is shown in Figure 8.
The VUF13816 motif was predicted to occupy the same binding pocket as doxepin (i.e., the orthosteric site), whereby the aromatic moiety of the VUF13816 motif was located in close proximity to F424 6.44 , W428 6.48 and 432 6.52 as well as the protonated piperidine nitrogen forming a salt bridge interaction with D107 3.32 (Figure 9a). The isopropyl side group of Val at AA 1 within the peptide linker was predicted to occupy the space between I454 7.39 and H450 7.35 (Figure 9b), in which the same space was linked to the combined affinity/ kinetics profile of rupatadine and its analogues. 60 The hydroxymethyl group of Ser at AA 2 within the peptide linker was in close proximity to D178 ECL2 (2.6 Å), K179 ECL2 (3.0 Å) and K191 5.39 (2.3 Å) suggesting a network of charge-reinforced hydrogen bonding interactions (Figure 9c). The presence of the acidic D178 ECL2 within the space predicted to be occupied by the side chain of AA 2 on the peptide linker also provides an explanation for the lack of improved affinity when AA 2 = Asp, due to electrostatic repulsion. The carboxamide moiety of Asn at AA 3 was predicted to form hydrogen bond interactions with N443 ECL3 (2.3 Å) and E447 7.32 (1.7 Å) (Figure 9d). Nluc-H 3 R pK D (K D (nM)) Nluc-H 4 R pK D (K D (nM)) 31a b < 6 (< 1000) < 6 (< 1000) 31k b < 6 (< 1000) < 6 (< 1000) CA200843 c 7.36 ± 0.05 (44)   a The k on , k off , T r and pK D of 31a and 31k were determined from realtime NanoBRET binding kinetic experiments conducted in Nluc-H 1 R expressing HEK293T cells. All values represent mean ± SEM from n separate experiments performed in triplicates. b MRT, mean residence time (1/k off ). c pK D determined from k on and k off (k on /k off ).
Despite offering insight into the potential ligand-receptor interactions at AA 2 (Ser) and AA 3 (Asn) of the peptide linker, it is important to consider that the proposed receptor residues interacting with these side groups (D178 ECL2 , K179 ECL2 , K191 5.39 , N443 ECL3 and E447 7.32 ) are hydrophilic and located in or near the extracellular loop region of the receptor. As such, we propose that interactions between the peptide linker side groups and the receptor in these regions are likely formed at the expense of disrupting existing water hydrogen bond networks, thus the free energy gained from hydrogen bonding between the interaction offsets the energy required to disrupt these networks. The flexible nature of the extracellular loop region would also suggest that a fixed conformation is not likely to be adopted, making the extracellular vestibule of the receptor especially at AA 3 hard to target for improving ligandreceptor interaction. This explains why linker optimization at AA 2 and AA 3 did not offer significant improvement to congener affinity, beyond those conferred by the optimization of AA. 1 Phosphatidylcholine-based phospholipid bilayer was modeled around the T4 lysozyme-truncated H 1 R crystal structure using ProBLM 61 and aligned with the predicted binding pose of 31a using PyMOL 2.1.1 ( Figure 10). The lipophilic BODIPY 630/650 TM fluorophore was predicted to be positioned in a space between TM5 and TM6 at the receptor outer surface, surrounded by the lipid bilayer and likely driven by the hydrophobic effect. 61 Indeed, the gain in free energy from the hydrophobic effect attributed to the fluorophore may increase the overall ligand affinity as well as allowing the receptor to adopt non-energy minimum conformations to accommodate the various linker side groups, especially at AA 1 where the Val side group significantly contributes to the observed linker SAR which is not seen in the corresponding fluorescent ligands. However, our data shows that H 1 R fluorescent ligands bearing   As such, we propose that although the hydrophobic effect is likely to have contributed to ligand binding, additional binding architecture surrounding the outer surface of the receptor specific to the BODIPY 630/650 TM fluorophore may be present. Interestingly, F440 6.60 was predicted to be positioned   Journal of Medicinal Chemistry pubs.acs.org/jmc Article directly above the BODIPY core of 31a (Figure 10), potentially forming a 'lid' above the fluorophore moiety. This 'lid' and the space between TM5 and TM6 could well be the proposed binding architecture. This hypothesis, however, could not be investigated further without structural information determined experimentally via X-ray crystallography or cryogenic electron microscopy on the binding of our fluorescent ligands to H 1 R.

■ CONCLUSIONS
In this study, we report the synthesis of a library of novel H 1 R fluorescent antagonist ligands with variations in the orthostere, peptide linker, and fluorophore moiety. Further exploration of compound 31a reveals that it is a high-affinity H 1 R-selective fluorescent antagonist, with a K D of 4 nM and a receptor residence time of more than 1 h at H 1 R, exhibiting at least 100fold selectivity for H 1 R over H 3 R and H 4 R. Our initial optimization of the peptide linker through the synthesis of a library of non-fluorescent congeners established clear SAR relating to the specific amino acid residues incorporated into this region. Surprisingly, this SAR does not translate to the context of the full fluorescent ligand, having no significant impact on the potency, selectivity, or binding kinetics of our series of H 1 R fluorescent ligands. Almost all BODIPY 630/ 650-based fluorescent ligands (all but 22c) synthesized and pharmacologically characterized in this study exhibited binding affinities of 4−28 nM at the H 1 R, regardless of differences in linker composition, length, and nature of the orthostere. These observations suggest that the fluorophore is a major determinant of overall ligand binding affinity in our series of H 1 R fluorescent ligands. Our subsequent systematic exploration of the effect of different fluorophores on overall fluorescent ligand affinity, using a pairwise matched approach demonstrates that SAR relating to the fluorophore itself is an important consideration when designing fluorescent ligands. To our knowledge, this exploration of fluorophore SAR in relation to target binding affinity is the first of its kind. Molecular docking studies with 31a provide insight into predicted ligand−receptor interactions and provide a rationale for the optimized linker employed, whereby the main improvement in congener affinity through linker optimization might be attributed to Val at AA 1 , which could effectively position itself into a space between H450 7.35 and I454 7.39 . The fluorophore is predicted to embed itself between the H 1 R receptor outer surface and the extracellular face of phospholipid bilayer, which could significantly improve the binding of the fluorescent ligands through a combination of the hydrophobic effect and potentially fluorophore-driven ligandreceptor interaction. Experimentally determined structural information on how our fluorescent ligands bind to the H 1 R is desirable to further elucidate the role of the fluorophore in overall ligand binding to the receptor. Finally, we have demonstrated that 31a is a valuable tool for cell imaging studies as it is capable of producing high quality images at concentrations as low as 10 nM in cell visualization experiments on the confocal microscope and thus could be a useful tool for the study of H 1 R pharmacology and receptor expression in endogenous expressing systems as well as to assess drug-target engagement in vivo.

■ EXPERIMENTAL SECTION
Chemistry General Information. All chemicals and solvents were obtained from Fischer Scientific UK, Acros Organics, Sigma-Aldrich, Merck Millipore, Fluorochem or Tocris Bioscience and used without further purification. BODIPY 630/650-X-NHS (D10000), BODIPY FL-NHS (D2184), and BODIPY FL-X-NHS (D6102) were purchased from Molecular Probes (Eugene, OR), whereas Sulfo-Cy5-NHS (23320) was purchased from Lumiprobe (Hunt Valley, MD). TLC was performed using Merck TLC Silica gel 60 Å F 254 plates. TLC plate visualizations were conducted under UV light (256 and 366 nm). Column chromatography was carried out using Sigma Aldrich silica gel (pore size 60 Å, 230−440 mesh, 0.040−0.063 mm). LC-MS was recorded on a Shimadzu UFLCXR HPLC system combined with an Applied Biosystems MDS SCIEX API2000 electrospray ionization mass spectrometer. The column used was a Gemini 3 μm C18 110 Å, LC Column 50 × 2 mm, and the solvent system was an increasing gradient (from 5 to 95% over 5 min) of acetonitrile in water containing 0.1% formic acid, flowing at 0.5 mL/ min. NMR spectra were recorded on a Bruker-AV 400 equipped with a 5 mm dual 1 H/ 13 C helium-cooled cryoprobe, which recorded the 1 H and 13 C NMR at 400.13 MHz and 101.62 MHz, respectively. The data was processed using iNMR (version 5.5.7), which referenced the spectra to those of the residual solvents. Chemical shifts (δ) were quoted in parts per million (ppm) and coupling constants (J) were reported to the nearest 0.1 Hz along with peak multiplicities using the following abbreviations: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet and br, broad. HRMS was conducted with a Bruker microTOF II mass spectrometer using electrospray ionization (ESI-TOF). Adducts within errors of ±6 ppm were reported. Analytical, semi-preparative and preparative RP HPLC were performed on a Waters 2767 sample manager coupled to Waters 2525 binary-gradient module and a Waters 2457 dual-wavelength absorbance detector or manually on a Rheodyne 7725i injector coupled to Waters 515 HPLC Pump and a Waters 996 photometric diode array detector. System 1 (for Analytical HPLC): Phenomenex Gemini 5μm reverse phase C18 column (250 × 4.6mm), a flow rate of 1.00 mL/min, and UV detection at 220, 254, and 330 nm. Linear gradient 5% to 95% solvent B over 25 min. Solvent A: 0.1% formic acid in water; solvent B: 0.1% formic acid in acetonitrile. Semi-preparative or preparative HPLC was conducted to purify the congeners and final fluorescent compound. Phenomenex Gemini 5μm reverse phase C18 column (250 x 21.2 mm), a flow rate of 20.00 mL/min, UV detection at 254 nm, and at various gradients of Solvent A:0.05% TFA in water; solvent B: 0.05% TFA in acetonitrile. System 2: 5% to 65% solvent B over 15 min. System 3: 20% to 50% solvent B over 10 min, System 4: 40% to 70% solvent B over 10 min. Phenomenex Gemini 5 μm reverse phase C18 column (250 × 10 mm), a flow rate of 5.00 mL/min, UV detection at 254 nm and at various gradients of Solvent A: 0.1% formic acid in water; solvent B: 0.1% formic acid in acetonitrile. System 5: 5% to 50% solvent B over 15 min. System 6: 30% to 60% solvent B over 15 min. System 7: 20% to 50% solvent B over 15 min. All pharmacologically tested compounds are >95% pure by HPLC.
Spectral Characterization of Fluorescent Ligands. 1 mM standard solution of the fluorescent ligand in DMSO was diluted to 0.1 mM with Hank's balanced salt solution (HBSS; 145 mmol/L NaCl, 5 mmol/L KCl, 1.7 mmol/L CaCl 2 , 1 mmol/L MgSO 4 , 10 mmol/L HEPES, 2 mmol/L sodium pyruvate, 1.5 mmol/L NaHCO 3 , 10 mmol/L D-glucose, pH7.4), and 100 μL of the dilution was transferred to a black-walled, clear-bottom 96-well plate. The plate was then placed in multi-well fluorometric imaging plate reader FlexStation3 (Molecular Devices, Sunnyvale, CA). The excitation spectrum was determined by measuring the intensity of emission at 620 nm after excitation with light of increasing wavelengths ranging from 450−580 nm. The emission spectrum was determined by measuring the intensity of light emitted across a range of wavelengths 550−700 nm under 500 nm light emission.
Transfection and Preparation of Cell Homogenates. HEK293T cells were transiently transfected with 2.5 μg of cDNA encoding Nluc-H 3 R or Nluc-H 4 R and 2.5 μg of empty pcDEF3 vector using the polyethylenimine method as described previously in Mocking et al. 23 Two days after transfection, cells were collected in phosphate-buffered saline (PBS) and centrifuged at 1900g for 10 min and the cell pellet was stored at −20°C until the day of experiment. The pellet was resuspended in HBSS buffer, and cells were disrupted using a Branson sonifier 250 (Boom bv., Meppel, the Netherlands) prior to use in a NanoBRET binding assay.
NanoBRET Binding Assay. For whole cell NanoBRET assays, HEK293T cells stably expressing Nluc-tagged H 1 R were seeded in white 96-well microplates and grown for 24 h prior to experimentation in normal growth medium. The media was replaced with HBSS right before experimentation. For whole cell saturation binding experiments, the required concentrations of fluorescent ligand and the competing unlabeled ligand mepyramine (final concentration of 10 μM) were added simultaneously to the plate and incubated for 2 h at 37°C in the absence of CO 2 . For whole cell competition binding experiments, the required concentrations of unlabeled ligand and the competing fluorescent ligand 1 (final concentration of 25 nM) were added simultaneously and incubated for 2 h at 37°C in the absence of CO 2 .
For cell membrane NanoBRET saturation binding experiments, cell homogenates expressing the Nluc-H 3 R or Nluc-H 4 R were incubated with the required concentrations of fluorescent ligand in the absence and presence of 10 μM clobenpropit in HBSS for 2 h at 25°C. For all NanoBRET binding experiments, 10 μM NanoGlo substrate (Promega) was added to each well after the incubation period and fluorescence and luminescence emissions were measured after 5 min.
For whole cell NanoBRET association kinetic experiments, 10 μM furimazine was added to each well and incubated at room temperature in the dark for 15 min to allow for stabilization of the luminescence signal. The required concentration of fluorescent ligand in the presence or absence of 10 μM of doxepin was added simultaneously, and the plates were read immediately with each well being read once per min for 90 min.
Intracellular Calcium Mobilization Assay. CHO cells expressing H 1 R were grown to confluence in black-walled, clear-bottom 96well plates. On the day of the experiment, media were replaced with 100 μL of HBSS containing 2.5 nM probenecid, 2.3 μM Fluo-4AM (Invitrogen), 0.023% Pluronic F-127, 0.5 mM Brilliant Black BN, and 1 μM required fluorescent ligand and the plates incubated for 1 h at 37°C in the dark. The plates were then placed in the multi-well fluorometric imaging plate reader FlexStation3 (Molecular Devices, Sunnyvale, CA) and HBSS only or HBSS containing the required concentration of histamine was added after 15 s. Fluo-4 fluorescence was measured at 520 nm (excitation at 485 nm) every 1.52 s for 200 s. For each well, the peak maximum−minimum−maximum change in fluorescence value was taken and plotted against concentration of histamine to give a dose−response curve.
Confocal Imaging. H 1 R-YFP expressing CHO cells were grown to approximately 80% confluency on an 8-well Labtek chambered coverglass (Nunc Nalgene). On the day of experiment, the cells were washed two times with 200 μL of HBSS and subsequently incubated in the presence or absence of 10 μM mepyramine for 30 min at 37°C in humidified air with 0% CO 2 . The required concentration of 31a was added to the wells and incubated for another 30 min at 37°C. Images were acquired using a Zeiss LSM710 confocal microscope (Carl Zeiss GmbH, Hena, Germany) fitted with a 63× plan-Apochromat NA1.3 Ph3 oil-immersion lens. For YFP, a 488 nm argon laser was used for excitation and emission was detected using a BP505-30 filter. For fluorescent ligand 31a, a 633 nm helium-neon laser was used for excitation and emission was detected using a 650 nm long pass filter. A pinhole of 1 Airy Unit was used, and fixed laser power, gain, and offset for the BODIPY 630/650 containing compounds were kept constant across all experiments.
Data Analysis. All data were analyzed and presented using Prism 7 (GraphPad Software, San Diego, CA).
Analysis of NanoBRET Binding Experiments. Total and nonspecific saturation binding curves from NanoBRET saturation binding assay were fitted simultaneously using eq 1: where B max is the maximal specific binding, [B] is the concentration of fluorescent ligand in nM, K D is the equilibrium dissociation constant in nM, M is the slope of the non-specific component, and C is the intercept with the Y axis. NanoBRET competition binding curves were fitted using the Cheng-Prusoff equation (2) For the NanoBRET association kinetic experiments, the BRET ratio was determined for each concentration of fluorescent ligand at each time point in the presence or absence of 10 μM of doxepin. Specific binding was calculated by subtracting non-specific binding from total binding. The k on , k off , and K D values were obtained from the data using eq 4: where K D is the equilibrium dissociation constant and k off is the dissociation rate constant of the ligand in min −1 . k on is the association rate constant in M −1 min −1 and is calculated as follows in eq 5: where [L] is the ligand concentration in M and k obs is calculated from global fitting of the data to the following equation, which expresses monoexponential association function (6): k t max obs (6) where Y max equals levels of binding at infinite time (t) and k obs is the rate constant for the observed rate of association.
Analysis of Ca 2+ Mobilization Experiment. Estimated affinity values (pK B ) were calculated from the shift in agonist concentration− response curves in the presence of the fluorescent antagonists using the Gaddum equation (7): where DR (dose ratio) is the ratio of the agonist concentration required to stimulate an identical response in the presence and absence of antagonist, [B]. DR was determined from the EC 25 value as there was a decrease in observed maximal efficacy of histamine in the presence of all 10 fluorescent antagonists. The EC 25 value was determined from the nonlinear regression fit of the normalized maximum−minimum fluorescence values using eq 8: min max min log 50 (8) where Y min and Y max are the lowest and highest normalized maximum−minimum fluorescence value and EC 50 is the concentration of the histamine at 50% response.
Molecular Docking Simulation. The H 1 R crystal structure (PDB code: 3RZE) was obtained from the protein data bank and prepared by using Protein Preparation Wizard within Maestro of the Schrodinger modeling suite. Missing atoms (in K442, R481) and missing residues (F168−V174 in extracellular loop 2) were modeled, phosphate ions, water molecules, and the Z-isomer of doxepin (D7V) were removed retaining only the E-isomer (5EH), hydrogen bondings were optimized using PROPKA 62,63 at pH 7, and the protein structure was minimized using the OPLS3 force field. 64 Ligands for docking were prepared and minimized using LigPrep 65 within the modeling suite and docked into the receptor using Glide under XP (extra precision) 66 mode without imposing any restraints using the OPLS3 force field and flexible ligand sampling. The sampling space encompassed the orthosteric binding domain and the entire space above it. The binding poses that formed the crucial salt bridge interaction with Asp107 were selected and examined using Maestro within the Schrodinger modeling suite. This was modeled using PyMOL 2.1.1 67 to include amino acid labels and distance measurements between the fluorescent ligand and nearby receptor residues.
Membrane Bilayer Modeling. The T4-lysozyme of the H 1 R crystal structure (PDB: 3RZE) was truncated on PyMOL 2.1.1 and saved as a PDB file. The membrane bilayer was modeled using the ProBLM web server 50 by uploading the PDB file as input and selecting POPC-75x75 (phospholiphase C) as the membrane file on the server. The membrane bilayer model was subsequently aligned with the docking pose of 31a using PyMOL 2.