Single cell imaging of Bruton's Tyrosine Kinase using an irreversible inhibitor

A number of Bruton's tyrosine kinase (BTK) inhibitors are currently in development, yet it has been difficult to visualize BTK expression and pharmacological inhibition in vivo in real time. We synthesized a fluorescent, irreversible BTK binder based on the drug Ibrutinib and characterized its behavior in cells and in vivo. We show a 200 nM affinity of the imaging agent, high selectivity, and irreversible binding to its target following initial washout, resulting in surprisingly high target-to-background ratios. In vivo, the imaging agent rapidly distributed to BTK expressing tumor cells, but also to BTK-positive tumor-associated host cells.

B ruton's tyrosine kinase (BTK) is a non-receptor tyrosine kinase with restricted cellular expression largely limited to B lymphocytes, macrophages/monocytes, and certain cancer cells 1-4 . As a critical component of the B cell receptor (BCR) signaling network, BTK is crucial for B cell development 5,6 and acts in multiple antiapoptotic signaling pathways, including the PI3K-AKT 7 , STAT5 8 and NF-kB 9,10 pathways. BTK is thus intimately involved in regulating cell survival, proliferation, and differentiation. In human haematological malignancies, BTK is abundantly expressed and activated in malignant cells from patients with B-cell multiple myeloma 11 , acute myeloid leukemia (AML) 12 , chronic lymphocytic leukemia (CLL) 13 , and non-Hodgkin's lymphoma (NHL) 14,15 . It is thus estimated that there are about 80,000 new BTK-positive haematologic malignancies in the US per year.
Several BTK inhibitors are under development and have shown remarkable efficacy in early clinical trials [16][17][18][19][20] . Ibrutinib (PCI-32765) is one example of a selective, irreversible BTK inhibitor, whose covalent binding results in long-lasting target occupancy, retaining inhibitory effect until new protein is synthesized 21,22 . The irreversible inhibitory effect of Ibrutinib is attributed to an electrophilic group on the molecule binding covalently to Cys 481 in the active site of BTK 23 . Most clinical trials to date have relied on insensitive standardized Response Evaluation Criteria approaches, such as computed tomography (CT), to image drug effects, while a denaturing gel electrophoresis assay has been used when tissue is available in Ibrutinib trials 21,24 . In the latter assay, a fluorescent probe binds any unoccupied BTK in tissue biopsy or blood to produce a fluorescent band; the lighter the band, the more BTK is occupied by drug. Even in co-clinical trials using mouse models, drug efficacy is largely tested by volumetrics or cell counts, while little is known about the kinetics of drug distribution in vivo, accumulation across cell types, and their respective heterogeneities or drug effects. There is therefore a need for imaging tools to study BTK inhibitor distribution at the single cell level in vivo. Such tools could be valuable to better understand kinetics, selectivity, drug action, inform on dose ranging studies, and allow in vitro testing of harvested cells from patients. Furthermore, imaging would be especially useful in the development of next generation BTK inhibitors 19,25,26 .
We hypothesized that an Ibrutinib-like scaffold could be converted into a companion diagnostic imaging agent by modification with a fluorescent tag while preserving irreversible target binding. The goal of the current study was to explore whether terminal modification of Ibrutinib could generate a BTK-selective imaging agent for in vivo use. Given the irreversible nature of target binding, one would expect improved target-to-background ratios following the clearance of unbound fractions. We indeed show remarkable target localization, specificity, and the ability to measure drug distribution and target inhibition in vivo. As more attention is paid to cell-to-cell heterogeneity in drug response and its impact on efficacy, we believe this will be a useful tool to study BTK expression and inhibition 27 .
(PDB : 3GEN, Fig. 1a). A BODIPY-FL modified Ibrutinib (Ibrutinib-BFL) was designed and synthesized de novo in seven steps (Fig. 1b). Briefly, iodination of commercially available pyrazolopyrimidine compound with N-Iodosuccinimide, followed by Suzuki coupling of the product with 4-phenyloxybenzene boronic acid, resulted in compound 2. Mitsunobu reaction of compound 2 with N-Boc-3hydroxypiperidine resulted in compound 3. After deprotection of the Boc protecting group in acidic conditions, the crude product was coupled with the linker (compound 5) to introduce a Michael acceptor for the irreversible binding affinity. Coupling of the crude Boc-deprotected compound 6 with BODIPY-FL-NHS finalized the synthetic steps to produce Ibrutinib-BFL (7) at an overall yield of ,11%.
To confirm the effect of BFL modification on the inhibition efficacy of the drug, half-maximal inhibitory concentration (IC 50 ) of Ibrutinib and Ibrutinib-BFL were determined against purified BTK enzyme. Ibrutinib-BFL had an IC 50 of ,200 nM, which is less potent than the parent drug (,2 nM IC 50 ; data not shown). Although it may be possible to further optimize the affinity of Ibrutinib-BFL by testing various linkers, we found the current generation probe to be quite acceptable for imaging, as shown in subsequent experiments. We next determined whether Ibrutinib-BFL would bind to purified BTK in vitro, endogenous BTK in live cells, and ultimately in vivo. Purified BTK was incubated with varying concentrations of the imaging probe for one hour at room temperature, denatured at 70uC for 10 minutes and then processed for SDS-PAGE gel analysis. There was a clear dose-response increase of the fluorescent signal around 80 kDa (BTK molecular weight is 76 kDa), as well as at the bottom of the gel (unbound fraction of Ibrutinib-BFL) (Fig. 2a). Additionally, binding could be blocked by pre-incubation with the parent compound and silver staining of the gel showed equal loading of BTK protein ( Supplementary Fig. S1). These results clearly confirmed the covalent binding property of Ibrutinib-BFL toward purified BTK.   (Supplementary Fig. S1). As expected, T cells did not express BTK. We found high BTK expression in Daudi and Toledo cell lines, and henceforth used Toledo as model BTK-positive cells and Jurkat as negative control cells. Toledo and Jurkat cells were incubated with different doses of Ibrutinib-BFL, and cell lysates were processed for SDS-PAGE and analyzed by fluorescent gel scanning. The imaging probe showed remarkable specificity, with binding observed only at a single band (Fig. 2b). The specificity was further confirmed by the absence of a band in BTK-negative Jurkat cells, even at the highest concentration of probe (Fig. 2b), as well as by silver staining of the gel ( Supplementary Fig. S1).
We next performed live cell imaging experiments using an imaging flow cytometry system. To prepare Toledo and Jurkat cells, we incubated them with 100 nM Ibrutinib-BFL for two hours, followed by washing. Figure 3 and Supplementary Fig. S2 summarize some of the results confirming target binding, specificity via blocking, and the ability to perform live cell imaging. To quantify colocalization between the imaging probe and BTK at the subcellular level, we created a stable transgenic cell line expressing a BTK-mCherry fusion protein in HT1080 human fibrosarcoma cells. In vitro cell experiments showed excellent co-localization and blocking (r 2 5 0.9851; Fig. 4).
We next performed in vivo experiments using three-color (blue: vasculature, green: Ibrutinib-BFL, red: BTK-mCherry-HT1080 cells) time-lapse intravital imaging. The intravascular half-life of Ibrutinib-BFL was ,10 minutes ( Supplementary Fig. S3). Within an hour after systemic administration, there was extensive leakage of the compound into the tumor interstitium. At later time points, cellular uptake became apparent, presumably due to interstitial washout and/or intracellular accumulation. The ability to image in multiple channels allowed us to ask whether Ibrutinib specifically localized in tumor cells. We show that greater than 99% of all BTK-mCherry-HT1080 cells had achieved therapeutic drug concentrations within one hour. This effective intracellular dose persisted for prolonged periods of time and the compound was still detectable inside cancer cells 24 hours after administration (Fig. 5). Interestingly, there was also accumula-tion of Ibrutinib-BFL in non-tumor cells even at late time points. Given the exquisite specificity of the drug (see Fig. 2), we hypothesized that these non-target cells also contain BTK. We thus performed correlative immunohistochemistry using anti-BTK antibody. Our data indicates that Ibrutinib-BTK also accumulates in tumor-associated macrophages and lymphocytes (Fig. 6).

Discussion
Inhibition of BTK is emerging as a promising target for B-cell malignancies, other cancers with BTK over-expression, and certain autoimmune diseases where BTK is involved. Ibrutinib, an irreversible inhibitor, is approved for treatment of mantle cell lymphoma and CLL, and is currently undergoing late-stage efficacy studies in patients with various B-cell malignancies. Based on its covalent target binding, we hypothesized that the molecule could serve as a companion imaging agent. Here we show that this is indeed the case. Ibrutinib-BFL co-localized with BTK in BTK-positive malignant cells and had low background accumulation in non-BTK cells, including those expressing structurally related interleukin-2-inducible T-cell kinase (ITK), which is expressed in T cells and Jurkat cells (see Supplementary Fig. S1). The companion imaging drug, Ibrutinib-BFL, also showed a predictable dose response curve, could be competitively inhibited, allowed drug concentrations to be quantitated in vivo, and enabled mapping of drug distributions at the single cell level. As such, we believe that Ibrutinib-BFL could have several applications, including use as a companion diagnostic for flow cytometry in haematologic malignancies, as an imaging agent to localize and map BTK-positive tumors, as a method to track subcellular localization of endogenous BTK, and as a tool to measure pharmacokinetics and pharmacodynamics in experimental settings during development of novel BTK-pathway inhibitors.
BTK is a cytoplasmic tyrosine kinase belonging to the Tec family. It is expressed in the B-cell lineage, plays a pivotal role in signaling and development, and is highly active in several haematological malignancies 28,29 . Some previous BTK imaging has been done with fluorescent protein tags (BTK-GFP and BTK-mCherry) to understand its activation and nucleocytoplasmic shuttling [30][31][32] , and its roles in myeloid cell chemotaxis 33 and infection 34,35 . Alternative research methods have primarily involved fluorescently labeled antibodies for immunohistochemistry and flow cytometry applications. The www.nature.com/scientificreports former is limited to experimental models and requires protein overexpression, and the latter requires cell permeabilization and fixation. The approach developed here, utilizing a small molecule affinity ligand, is compatible with live cells, can be used in vivo, and has potential clinical applicability. Not only does Ibrutinib-BFL specifically bind to BTK, but also it remains bound until protein turnover due to the virtually nonexistent off-rate of covalent inhibitors. This feature will allow for long-term study of endogenous BTK in live cells, providing a window into drug pharmacodynamics, as well as innate heterogeneity in responses to drugs targeting the BCR signaling pathway 24,27 .
Beyond utilizing Ibrutinib-BFL in pharmacologic studies of next generation inhibitors, there are future diagnostic opportunities in which BTK-expressing lymphomas could be imaged in the clinic. While the current work focused on single cell imaging in vivo, we also anticipate whole body imaging applications. For example, the fluorine in BODIPY-FL could be exchanged for 18 F for positron emission tomography (PET) imaging, or entirely replaced via bioorthogonal ligands or direct 18 F attachment [36][37][38][39] . Alternatively, longer-lived isotopes such as Zirconium-89 could also be utilized in order to take full advantage of the probe's irreversible binding kinetics [40][41][42][43][44] . Such molecules may be useful in clinical imaging-based tests for whole body distribution and inhibition of BTK. Other areas of interest are to use these molecules for imaging BTK in macrophages during infection, or to use them as a readout during gene therapy for the immunodeficiency disorder X-linked agammaglobulinemia, which results from loss of functional BTK 45 . Irrespective of the contemplated use, we believe that the developed agent should be useful in a number of different applications. As covalent inhibitors have gained interest, we anticipate covalent imaging agents to follow, and Ibrutinib-BFL can provide a roadmap for such development.
Gel Electrophoresis. To test the covalent binding of Ibrutinib-BFL to BTK, 0.1 mg (1 mL) purified BTK was combined with 0.4 mL Ibrutinib-BFL (prepared in advance in 2-fold dilutions ranging from 200 mM to 0.19 mM, 33% DMSO in PBS) and 18.6 mL PBS, and incubated in the dark at room temperature for one hour. In the second experiment, Toledo and Jurkat cells (2.2 3 10 6 per well in culture media) were incubated in growth media containing 5-fold serial dilutions of Ibrutinib-BFL ranging from 6 mM to 9.6 nM in final 2% DMSO at 37uC for two hours. Control samples were incubated in growth media containing 2% DMSO. Cells were washed once with ice cold PBS, then lysed in 150 mL 1X RIPA buffer (Cell Signaling Technology, Beverly, MA, USA) containing protease inhibitors. To the purified enzyme samples or cell lysates, NuPAGE LDS sample buffer and NuPAGE reducing agent (Invitrogen) were added for final 25% and 10% concentrations, respectively, and samples were heated to 70uC for 10-12 minutes in a Mastercycler thermal cycler (Eppendorf, Hamburg, Germany). 25 mL per lane was loaded into 12-well NuPAGE Novex 4-12% Bis-Tris gels (Invitrogen). Using 10 mL of Novex Sharp Pre-stained Protein Standard (Invitrogen) as a size marker, the gels were run in NuPAGE MES SDS running buffer (Invitrogen) at 200 V for 35 minutes in the XCell SureLock Mini-Electrophoresis system (Invitrogen). The gels were removed from the cassette and imaged using a Typhoon 9410 fluorescence scanner (GE Healthcare, Pittsburgh, PA, USA) using 488 nm excitation and a 520 nm emission filter. To show total protein loading, gels were silver-stained using the Pierce Silver Stain for Mass Spectrometry kit (Thermo Fisher Scientific, Rockford, IL, USA).
Imaging of adherent cells by microscopy. HT1080-BTK-mCherry cells were seeded into a 96-well plate at 20,000 cells per well and allowed to grow to confluence overnight. Cells were incubated in growth media containing 1 mM Ibrutinib in final 0.1% DMSO, or control 0.1% DMSO, at 37uC for 1.5 hours. Without washout, a 503 stock of Ibrutinib-BFL in 5% DMSO was added for a final concentration of 500 nM. Control wells contained equivalent DMSO without Ibrutinib-BFL. Cells were incubated for one hour at 37uC and then washed once with media for five minutes. The media was then replaced and cells were incubated overnight at 37uC. The live cells were subsequently imaged on the DeltaVision imaging system (Applied Precision, a GE Healthcare Company). Images were processed with Fiji software, an open-source version of ImageJ.
In vivo tumor imaging. Nu/nu mice were implanted with 2 3 10 6 HT1080-BTK-mCherry cells into a dorsal skinfold window chamber (APJ Trading Company, Ventura, CA, USA) according to established protocols 47 and according with guidelines from the Institutional Subcommittee on Research Animal Care. Tumors were allowed to grow and vascularize for two weeks. 75 nmol Ibrutinib-BFL in 150 mL solution containing DMAc and solutol was injected via tail vein as reported previously 48 . Mice were anesthetized with 2% isoflurane in 2 L/min oxygen. Timelapse microscopy was performed for two hours using a customized Olympus FV1000 confocal/multiphoton microscope equipped with a 203 objective (both Olympus America, Chelmsford, MA, USA). In addition, tumors were imaged before injection, and at 2, 5, and 24 hours post-injection. Images were processed with Fiji software.