Synthesis and Structure-Activity Relationships of 5,6,7-substituted Pyrazolopyrimidines: Discovery of a novel TSPO PET Ligand for Cancer Imaging

Abstract

Focused library synthesis and structure-activity relationship development of 5,6,7-substituted pyrazolopyrimidines led to the discovery of 2-(5,7-diethyl-2-(4-(2-fluoroethoxy)phenyl)pyrazolo[1,5-a]pyrimidin-3-yl)-N,N-diethylacetamide (6b), a novel translocator protein (TSPO) ligand exhibiting a 36-fold enhancement in affinity compared to another pyrazolopyrimidine-based TSPO ligand, 6a (DPA-714). Radiolabeling with fluorine-18 (¹⁸F) facilitated production of 2-(5,7-diethyl-2-(4-(2-[¹⁸F]fluoroethoxy)phenyl)pyrazolo[1,5-a]pyrimidin-3-yl)-N,N-diethylacetamide (¹⁸F-6b) in high radiochemical yield and specific activity. In vivo studies of ¹⁸F-6b were performed which illuminated this agent as an improved probe for molecular imaging of TSPO-expressing cancers.

Introduction

There remains a critical need to develop and rigorously validate molecular imaging biomarkers that aid tumor diagnosis, predict clinical outcome, and quantify response to therapeutic interventions. Imaging techniques routinely used in clinical oncology include magnetic resonance imaging (MRI), X-ray computed tomography (CT), ultrasound imaging (US), and positron emission tomography (PET). Of these, the sensitivity and quantitative nature of PET, coupled with the ability to readily produce biologically active compounds bearing positron-emitting isotopes (e.g., ¹¹C, ¹⁸F), renders PET imaging as one the most attractive techniques for detecting tumors and profiling their molecular features. By far, the most widely used PET tracer in clinical oncology is 2-deoxy-2-[¹⁸F]fluoro-D-glucose (¹⁸F-FDG), a probe that accumulates in tissue as a function of glucose utilization. PET using ¹⁸F-FDG is a powerful approach for tumor detection in many organ sites. However, not all tumors exhibit elevated glucose avidity, and ¹⁸F-FDG uptake can be affected by a plethora of normal metabolic processes. Furthermore, tumor imaging can be confounded by ¹⁸F-FDG uptake in normal tissues such as healthy brain. These issues highlight an unmet need to explore and validate additional molecular targets for cancer imaging.

Our laboratory has explored translocator protein (TSPO) expression as a target for molecular imaging of cancer.^1–5 Formerly referred to as peripheral benzodiazepine receptor (PBR), TSPO is an 18-kDa protein typically localized to the outer mitochondria membrane. TSPO participates in the regulation of numerous cellular processes, including steroid biosynthesis, cholesterol metabolism, apoptosis, and cellular proliferation.⁶ In normal tissues, TSPO tends to be expressed in those that produce steroids and those that are mitochondrial-enriched such as myocardium, skeletal muscle, and renal tissue. Tissues such as liver and brain exhibit comparatively modest expression.⁶ While classically exploited as a target in neuroscience, elevated TSPO expression is also observed in many cancers.¹³ In oncology, TSPO expression is typically linked with disease progression and diminished survival and is a hallmark of aggressive and potentially metastatic tumors.¹³ For this reason, our laboratory has explored the use of TSPO imaging ligands within the context of colon cancer,¹ breast cancer, ² and glioma,^{4, 5} as these agents could potentially serve as useful cancer imaging biomarkers. We recently reported the first utilizations of the PET agents N-[¹⁸F]fluoroacetyl-N-(2,5-dimethoxybenzyl)-2-phenoxyaniline (¹⁸F-PBR06)⁵ and N,N-diethyl-2-(2-(4-(2-[¹⁸F]fluoroethoxy)phenyl)-5,7-dimethylpyrazolo[1,5-a]pyrimidin-3-yl)acetamide (¹⁸F-DPA-714)⁴ for quantitative assessment of TSPO expression in preclinical glioma. In these proof-of-principle PET imaging studies, tumors were detectable among surrounding normal brain and, importantly, TSPO levels could be quantitatively assayed in tumors using compartmental analysis of the PET data.^{4, 5} However, drawbacks were observed with both agents in this context, including tracer accumulation in normal brain that reached levels potentially sufficient to prevent detection of gliomas with modest TSPO expression. Both tracers also exhibited significant metabolism in vivo, which required correction of plasma input functions for quantitative analysis. While illustrating the potential of TSPO PET to detect tumors in brain, these studies prompted our desire to develop novel TSPO PET ligands with improved properties for cancer imaging.

The goal of this study was to determine whether optimization of the pyrazolopyrimidine scaffold, specifically at the 5-, 6-, and 7-positions, would yield TSPO ligands with greater affinity and potentially serve as more robust PET imaging ligands in vivo. These experiments led to the discovery of 2-(5,7-diethyl-2-(4-(2-fluoroethoxy)phenyl)pyrazolo[1,5-a]pyrimidin-3-yl)-N,N-diethylacetamide (6b), a novel TSPO-selective ligand that exhibits a surprising 36-fold enhancement in affinity compared to another pyrazolopyrimidine, 6a (DPA-714). An appropriate analog of 6b, compound 7, could be radiolabeled with fluorine-18 (¹⁸F) to yield the novel TSPO PET ligand 2-(5,7-diethyl-2-(4-(2-[¹⁸F]fluoroethoxy)phenyl)pyrazolo[1,5-a]pyrimidin-3-yl)-N,N-diethylacetamide (¹⁸F-6b), which was subsequently evaluated in vivo in healthy rats and a preclinical model of glioma. ¹⁸F-6b exhibited negligible accumulation in normal brain, yet robust accumulation in tumor tissue, which facilitated excellent imaging contrast. Overall, these studies illuminate ¹⁸F-6b as a promising, novel PET ligand for evaluating TSPO expression in tumors and potentially other diseases.

Results and Discussion

CHEMISTRY

MAOS of 5,6,7-substituted Pyrazolopyrimidine Library

A library of structurally diverse pyrazolopyrimidines (Table 1) was assembled using a microwave-assisted organic synthesis (MAOS) method that prioritized points of divergence at the 5-R₃, 6-R₂, and 7-R₁ positions on the core pyrazolopyrimidine scaffold. Members of this library were formed through condensation of a pyrazole core (4) with various substituted diones.¹¹ Entries in Table 1 fall into two distinct series (5 and 6), with seven compounds (a–g) per series. Each compound within the series results from one of seven diones utilized, while each series is differentiated by the installation of a unique surrogate imaging handle at the 4-postion (R₄) of the 2-phenyl group pendant to the core scaffold. Series 5 entries (5a–5g) feature a methoxy group at R₄ and include a previously reported pyrazolopyrimidine, 5a (DPA-713),^7–10 while series 6 entries (6a–6g) feature a 2-fluoroethoxy group at R₄ and include 6a.^{7–9, 12} Given our ultimate goal of developing novel PET imaging ligands, coupled with an interest in expanding structure-activity relationships (SAR) around the 5,6,7-substituted pyrazolopyrimidine core, we rationalized that maintaining the surrogate imaging handles of future representative imaging probes would accelerate the development of novel agents.

Table 1.

Affinity, liphopilicity, and molecular weight of pyraxolopyrimidines

Compd	R1	R2	R3	R4	MW	LogP7.5	Ki (nM)δ
5aα	-Me	-H	-Me	-OMe	366.21	2.40	12.23
5b	-Et	-H	-Et	-OMe	394.51	2.84	0.18
5cβ	-iPr	-H	-iPr	-OMe	422.56	2.98	>500
5d	-Me	-Me	-Me	-OMe	380.48	2.45	93.75
5e	-Me	-Et	-Me	-OMe	394.51	2.78	157.14
5f	-Me	-CH2C(O)CH3	-Me	-OMe	408.49	2.57	200.89
5g	-Me	-Cl	-Me	-OMe	400.9	2.99	55.36
6aγ	-Me	-H	-Me	-OCH2CH2F	398.47	2.12	9.73
6b	-Et	-H	-Et	-OCH2CH2F	426.53	2.50	0.27ε
6c	-iPr	-H	-iPr	-OCH2CH2F	454.58	2.73	>500
6d	-Me	-Me	-Me	-OCH2CH2F	412.5	2.47	83.04
6e	-Me	-Et	-Me	-OCH2CH2F	426.53	2.53	276.79
6f	-Me	-CH2C(O)CH3	-Me	-OCH2CH2F	440.51	2.07	251.79
6g	-Me	-Cl	-Me	-OCH2CH2F	432.92	2.55	67.86

^αDPA-713, see references.^7–10

^βSee reference.¹¹

^γDPA-714, see references.^{7–9, 12}

^δCompetitive binding against ³H-PK 11195 in C6 glioma cell lysate.

^εKi versus ³H-flunitrazepam > 10,000 nM.

The overall synthetic methodology is presented in Scheme 1.¹¹ Here, compound 4, bearing a 3-amino-1H-pyrazole core, was accessible in two steps using MAOS from commercially available phenylpropanenitrile 1 and chloroacetamide 2. From 4, the synthesis diverges at the condensation step with a series of unique diones to yield the “five” series of pyrazolopyrimidines (5a–5g) that mimic analogous ¹¹C-based PET probes with an R₄-OMe. Subsequently, compounds 6a–6g were achieved by cleavage of methoxy group followed by microwave–assisted ether synthesis with 2-fluoroethyl-4-methylbenzenesulfonate which yielded an R₄-2-fluoroethoxy surrogate imaging handle to mimic ¹⁸F-labeled PET probes (Table 1). The library developed primarily explored the effects of R₁, R₂, and R₃ substituents that varied in steric bulk. Generally, substituents at R₁ and R₃ mirrored one another and consisted of linear and branched alkyl substituents such as methyl, ethyl, and isopropyl moieties. These substituents were cross-matched with similar groups at the R₂ position, which included hydrogen, methyl, ethyl, chloro, and 2-propanone.

Scheme 1. Synthetic methodology utilized to generate a focused library of 5,6,7-substituted pyrazolopyrimidines and ¹⁸F-6b^α.

^αReagents and conditions: (a) NaOH, NaI, microwave, 80 °C, 40 min, 80% EtOH in H₂O, 70%; (b) hydrazine, AcOH, microwave, 90 °C, 40 min, EtOH, 42%; (c) microwave, 140–190 °C, 45min, EtOH, 10–90%; (d) HTPB, microwave, 110 °C, 40 min, aqueous HBr; (e) NaH, 2-fluoroethyl-4-methylbenzenesulfonate, microwave, 120 °C, 30 min, dry THF. (f) ethylene di(p-toluenesulfonate), NaH, microwave, 120 °C, 30 min, dry THF, 45%. (g) Fluoride-18 ion, K⁺-K_2.2.2/K₂CO₃, 99 °C, 20 min, dry DMSO.

Lipophilicity

Lipophilicity of series 5 and 6 were evaluated at pH = 7.5 (log P_7.5) using reversed phase HPLC¹⁴ and is tabulated in Table 1. The values for library entries varied from 2.0 to 3.0 for both series and correlated predictably with dione structure, where polarity and hydrocarbon content of the R₁–R₃ substituents appeared to be key determinants. In both series, analogs a, d, and f, with R₁–R₃ combinations of hydrogen, methyl, and 2-propanone, proved the least lipophilic. Introduction of ethyl groups at R₁, R₂, or R₃ increased log P_7.5 values accordingly, as with analogs b and e, which ranged from 2.50–2.84. The most lipophilic analogs proved to be c and g (log P_7.5 = 2.55–2.99), with isopropyl groups at the R₁ and R₃ positions or chlorine at R₂. Interestingly, of the two series, series 5 (R₄ = OMe), tended to be more lipophilic than series 6 (R₄ = OCH₂CH₂F). Overall, the library entries exhibited lipophilicities (2.07–2.99) suitable for in vivo imaging,¹⁵ where potential agents must be sufficiently membrane penetrant to bind intracellular targets such as TSPO.

BIOLOGICAL TESTING

Binding Affinity in Cancer Cell Lysate and SAR Analysis

To evaluate binding of the library entries to TSPO, radioligand displacement was carried out in rat glioma cell lysate (C6) using N-(sec-butyl)-1-(2-chlorophenyl)-N-methyl-³H-isoquinoline-3-carboxamide (³H-PK 11195)^{4, 5}. Affinities of library members are expressed as Ki (nM) values in Table 1. Encompassing modifications at R₁–R₄, both series exhibited a spectrum of Ki values that spanned micromolar to sub-nanomolar affinity. As expected, replacing the 4-methoxy with 4-(2-fluoroethoxy) at the R₄ position had minimal impact upon TSPO affinity, as similar SAR trends emerged within series 5 and 6. In contrast, R₁, R₂, and R₃ modification led to major impacts upon affinity. Though the R₁ and R₃ positions showed intolerance to steric bulk beyond ethyl, a surprising level of improvement in affinity could be realized by replacing methyl with ethyl at the R₁ and R₃ positions. For example, subnanomolar affinities (Ki) were observed for analogs 5b (0.18 nM) and 6b (0.27 nM), which represent major improvements over their respective leads, 5a (12.23 nM) and 6a (9.73 nM) (Table 1), as well as another high-affinity TSPO ligand, N-(2-methoxybenzyl)-N-(4-phenoxypyridin-3-yl)acetamide (PBR28) (4.0 nM)¹⁶. Increasing the R₁ and R₃ to a bulkier isopropyl group proved detrimental, with Ki values for both 5c and 6c greater than 500 nM. Overall, the R₂ position was less tolerant to steric or electron density deviations from hydrogen in these studies. Substitution of the nascent R₂ hydrogen (5a, 6a) with methyl (5d, 93.75 nM; 6d, 83.04 nM) decreased affinity approximately 10-fold with respect to their parent compounds, 5a and 6b, respectively; this trend was more dramatic with ethyl substitution (5e, 157.14 nM; 6e, 276.79 nM). Introduction of polarity at the R₂ position with 2-propanone (5f, 200.89 nM; 6f, 251.79 nM) bore similar adverse effects upon affinity. Somewhat less dramatic decreases in affinity were observed upon halogen substitution at R₂ with chlorine (5g, 55.36 nM; 6g, 67.86 nM). Given the size of chlorine atom, lying between methyl and ethyl, it is reasonable that the Ki of entries 5g and 6g are only slightly worse than those with alkyl substituents at the R₂ position (Table 1). Overall, these experiments illuminated 5,7-diethyl, 6-hydrogen substitution as the optimal pattern in this series to impart improved affinity. Additionally, the studies also suggested that an analog could be produced for PET imaging using either fluorine-18 (6b) or carbon-11 (5a) substitution on the R₄ position with minimal effect on TSPO affinity and only minor effect on lipophilicity. Since fluorine-18 has a significantly longer half-life than carbon-11 (109.8 min vs. 20.4 min), we elected to carry compound 6b forward for further characterization. The selectivity of compound 6b for TSPO over the central benzodiazepine receptor (CBR) was evaluated using radioligand displacement of the CBR ligand ³H-flunitrazepam in rat cerebral cortex membranes (Table 1). Unlike the displacement studies with ³H-PK 11195, compound 6b demonstrated poor affinity for CBR (Ki > 10,000 nM), indicating TSPO selectivity.

RADIOCHEMISTRY

Radioligand Precursor Preparation and Radiosynthesis

To produce ¹⁸F-6b, we synthesized precursor (7), a tosylated intermediate prepared from 5b in two steps with an overall yield of approximately 45% (Scheme 1). Nucleophilic fluorination of 7 with fluorine-18 was then performed (Scheme 1) in anhydrous dimethyl sulfoxide at 165 °C for 5 minutes. Purification of ¹⁸F-6b was carried out with preparative HPLC using 10 mM sodium phosphate buffer (pH 6.7) in ethanol (47.5/52.5, v/v). The retention time of ¹⁸F-6b was 12 minutes according to gamma detection and corresponded to the UV retention time of non-radioactive 6b. Radiochemical purity was consistently greater than 99%, with specific activity consistently greater than 4203 Ci/mmol (156 TBq/mmol) (n = 33).

IMAGING STUDIES

Uptake of ¹⁸F-6b in C6 Glioma

The in vivo performance of ¹⁸F-6b was evaluated in glioma-bearing Wistar rats using microPET imaging, with a typical study shown in Figure 1. MRI (T₂-weighted) was used to localize tumors and for registration of anatomical features with PET^{4, 5} (Figure 1A). Dynamic PET imaging with ¹⁸F-6b illustrated that the majority of tracer uptake in the brain was localized to the tumor, with very modest accumulation in the adjacent, normal brain (Figure 1B and 1C). The tumor-selective uptake characteristics of ¹⁸F-6b afforded excellent imaging contrast between tumor and normal tissue. In addition to the tumor, accumulation of ¹⁸F-6b was observed outside of the brain in olfactory epithelium and the Harderian glands, as well as the tongue, where TSPO expression is elevated.¹³ Figure 1D illustrates time-activity curves (TACs) that were typical of eight representative studies for tumor, normal brain, and plasma over a 90-minute dynamic acquisition. We found that ¹⁸F-6b was rapidly delivered to tumor and normal brain, but cleared tumor tissues at a much slower rate than normal brain. After an initial spike in radioactivity consistent with tracer injection and rapid distribution, ¹⁸F-6b quickly cleared from the plasma. To validate the PET, imaging-matched brains were processed for post mortem staining and immunohistochemistry. We observed close agreement between tumor tissue (H&E, Figure 1E), elevated TSPO expression by immunohistochemistry (Figure 1F), and tumor accumulation of ¹⁸F-6b accumulation (Figure 1C). To more quantitatively compare the performance of ¹⁸F-6b to ¹⁸F-6a, we conducted dynamic PET using ¹⁸F-6b in a cohort of tumor-bearing rats (n = 5) and carried out pharmacokinetic modeling of the PET data, similar to our previous studies^{4, 5}. We found that ¹⁸F-6b demonstrated a higher ratio of total distribution volume (V_T) between tumor and normal brain (6.0, n = 5), when compared to ¹⁸F-6a (3.9, n =11),⁴ which resulted in greater signal-to-noise between tumor and surround normal brain.

Figure 1.

PET imaging of preclinical glioma using ¹⁸F-6b.

Characterization of ¹⁸F-6b Radiometabolites in Plasma

Analogous to our previous work,⁴ radio-HPLC was used to evaluate metabolism of ¹⁸F-6b in plasma at multiple time points during the PET scanning period (2, 12, 30, 60, 90 min) (Supplemental Figure 1). Parent compound (¹⁸F-6b) was the only radioactive species detected in plasma at all time points, suggesting a lack of circulating ¹⁸F-6b metabolites observed over a 90-minute period. This behavior contrasted the in vivo performance of ¹⁸F-6a, where we previously observed up to 30% metabolism within the first 12 minutes post-administration.⁴ Potentially, the apparent improved stability of ¹⁸F-6b over ¹⁸F-6a may stem from the rapid clearance of ¹⁸F-6b from plasma, rendering detection of low levels of metabolites challenging with HPLC.

Biodistribution and Specific Binding of ¹⁸F-6b in Rats

The biodistribution and TSPO-specificity of ¹⁸F-6b was evaluated in normal tissues. Tissue samples were harvested 60 minutes after infusion of the ¹⁸F-6b. The radioactivity of harvested tissue was measured using standard analytical methods and recorded as percentage of the injected dose per gram of tissue (%ID/g) (Supplemental Figure 2A). Consistent with known TSPO densities,¹³ we observed elevated accumulation of ¹⁸F-6b in spleen, heart, kidney, and lung.¹⁷ Moderate uptake was also observed in liver, colon, small bowel, and stomach. Very minor accumulation was observed in the brain, testes, skull, and skeletal muscle. At 60 minutes post-administration of tracer, very little radio-activity was observed in biological fluids, such as whole blood or urine.

To evaluate the specific binding of ¹⁸F-6b in normal tissue, we carried out a similar 60-minute biodistribution assay that included a bolus infusion of non-radioactive 6b (10 mg/kg) at 30 minutes (Supplemental Figure 2B). With displacement, we observed that activity in the organs exhibiting the greatest accumulation of ¹⁸F-6b, such as spleen, heart, kidney, and lung, was reduced to background levels, indicating near complete displacement and reversible binding in healthy tissue. Organs that exhibited low to moderate accumulation of ¹⁸F-6b without displacement, such as muscle or liver, were essentially unchanged with displacement, suggesting that the low to moderate tracer accumulation in these tissues may reflect non-specificity. Interestingly, activity found in the urine did not follow this trend and was elevated with displacement, suggesting that blocking TSPO with non-radioactive 6b altered the excretion profile of the agent. Overall, the results indicate a high degree of specific binding and reversibility of ¹⁸F-6b to TSPO in healthy tissues.

CRYSTALLOGRAPHY

6b was recrystallized from hot acetonitrile and ethanol (95%) by slow evaporation at room temperature and characterized by diffractometry. 6b demonstrated two ethyl branches at the 5- and 7-positions of the pyrazolopyrimidine ring, corresponding to the structure elucidated from NMR spectroscopy. The crystal structure also revealed three distinct planar entities, the pyrazolopyrimidine ring, the phenyl ring, and the amide group (Supplemental Figure 3).

Conclusion

In this study, we found that the 5,6,7-substitution pattern is a critical determinant of TSPO affinity among pyrazolopyrimidines. Experiments shown here led to the discovery of 6b, a TSPO ligand exhibiting significantly increased affinity compared to 6a and TSPO ligands derived from other chemical entities.¹⁸ Furthermore, in addition to improved TSPO affinity, compound 6b was TSPO-selective, exhibiting negligible binding affinity to CBR. Analog 7 of 6b could be radiolabeled with fluorine-18 to yield ¹⁸F-6b. Suggesting its potential as a probe for cancer imaging, preclinical imaging studies demonstrated robust accumulation of ¹⁸F-6b in tumor tissue and negligible accumulation in normal brain. Overall, these studies illuminate ¹⁸F-6b as a promising, novel PET ligand for evaluating TSPO expression in tumors and potentially other diseases.