Targeting Mitochondria in Tumor-Associated Macrophages using a Dendrimer-Conjugated TSPO Ligand that Stimulates Antitumor Signaling in Glioblastoma

Abstract

Mitochondria mediate critical cellular processes, including proliferation, apoptosis, and immune responses; as such, their dysfunction is pathogenic in many neurodegenerative disorders and cancers. In glioblastoma, targeted delivery of mitochondria-focused anticancer therapies has failed to translate into clinical success due to the nonspecific cellular localization, heterogeneity of receptor expression across patients, poor transport across biological barriers to reach the brain, tumor, and mitochondria, and systemic side effects. Strategies that can overcome brain and solid tumor barriers and selectively target mitochondria within specific cell types may lead to improvements in glioblastoma treatment. Developments in dendrimer-mediated nanomedicines have shown promise targeting tumor-associated macrophages (TAMs) in glioblastoma, following systemic administration. Here, we present a novel dendrimer conjugated to the translocator protein (18 kDa) (TSPO) ligand 5,7-dimethylpyrazolo[1,5-α]pyrimidin-3-ylacetamide (DPA). We developed a clickable DPA for conjugation on the dendrimer surface and demonstrated in vitro that the dendrimer-DPA conjugate (D-DPA) significantly increases dendrimer colocalization with mitochondria. Compared to free TSPO ligand PK11195, D-DPA stimulates greater antitumor immune signaling. In vivo, we show that D-DPA targets mitochondria specifically within TAMs following systemic administration. Our results demonstrate that dendrimers can achieve TAM-specific targeting in glioblastoma and can be further modified to target specific intracellular compartments for organelle-specific drug delivery.

Graphical Abstract

INTRODUCTION

The mitochondrion is widely known as the powerhouse of the cell and is chiefly responsible for regulating cellular energy production and metabolism.¹ In addition, mitochondria play critical roles in regulating trafficking, proliferation, apoptosis, and immune responses.² Due to their position at the nexus of these vital physiological functions, mitochondrial dysfunction is associated with diverse diseases ranging from neurodegenerative disorders to cancers to autoimmune disorders.^3,4 Restoration of healthy mitochondrial function from the pathological activity is therefore necessary to address the progression of these diseases.⁵

Defective mitochondrial function in cell cycle regulation and apoptotic signaling is characteristic of cancer progression.⁶ Therefore, intracellular targeting anticancer therapies focused on mitochondria to slow cancer cell proliferation and induce apoptosis have been explored.⁷ In addition, due to their roles in mediating inflammatory responses, mitochondria have also been explored as targets in cancer immunotherapy approaches.⁸ For example, mitochondrial damage in T cells has been found to mediate the suppression of the anticancer immune response.^9,10 Tumor-associated macrophages (TAMs) have been found to mediate the cancer immune response, and their activation state has been shown to be regulated by mitochondrial metabolism.^11,12 Mitochondria mediate HIF-1α and HIF-2α in TAMs, and their localization to mitochondria correlates with high grade and poor prognoses of cancers.¹¹ In addition, dysfunctional mitochondria exhibit overproduction of reactive oxygen species (ROS), leading to high oxidative stress in the tumor environment that promotes tumor growth.^13,14 As a therapeutic target in TAMs, mitochondria can be leveraged to promote antitumor signaling by activating proinflammatory immune signaling.¹⁵ However, TAM-focused immunotherapies have suffered from low response rates, poor brain and tumor penetration, and nonselective activity, resulting in systemic toxicities that have hampered their clinical translation.^16,17 In addition, the failure of mitochondria-targeted therapies lies largely with their inability to preferentially accumulate in mitochondria of specific cells of interest. Therefore, a nanotechnology-mediated strategy aimed at cell-specific mitochondrial targeting may overcome these delivery challenges as an effective treatment for glioblastoma and other cancers.⁸

Translocator protein (18 kDa) (TSPO) is a transport protein located on the outer mitochondrial membrane that is responsible for transporting cholesterol into mitochondria for the synthesis of steroids.¹⁸ TSPO is minimally expressed in healthy tissues but is highly overexpressed in the context of neuroinflammation and cancer.^19,20 In the context of glioblastoma, TSPO expression correlates with glioblastoma clinical outcomes.²¹ Targeting TSPO has been explored to improve diagnostics through positron-emission tomography (PET) imaging and therapies.^22,23 TSPO is highly upregulated in anti-inflammatory macrophages consistent with the immunosuppressive TAM phenotype.²⁴ Targeting TSPO can leverage its overexpression in TAMs to manipulate immune polarization. PK11195 is a first-generation TSPO ligand with a nanomolar binding affinity that has been extensively explored as an inhibitor of tumor cell proliferation and modulator of immune signaling.^25,26 5,7-Dimethylpyrazolo[1,5-α]pyrimidin-3-ylacetamide (DPA) is a novel class of TSPO ligands, which are selective, drug-like ligands.²⁷ While these TSPO targeting compounds have been explored extensively for diagnostics in PET imaging, their utility for targeted drug delivery in the brain has been limited by poor brain penetration.^28,29 Therefore, nanomedicines that can penetrate into the brain and brain tumors may enable effective TSPO targeting for improved drug delivery.

Many nanotechnology-based strategies are being developed to target drugs to mitochondria.^30,31 However, platforms that enable mitochondria targeting within specific cell types in the brain are still being developed. Dendrimers have shown significant promise for specific gene and drug delivery to activated microglia, cancer cells, and TAMs.^31–37 In an orthotopic brain cancer models, we have previously shown that systemic administration of generation 4 hydroxyl-terminated polyamidoamine (PAMAM) dendrimers is able to fully penetrate and distribute uniformly throughout the solid brain tumor selectively in TAMs, with minimal accumulation in healthy brain and peripheral tissues.^33,38 This cell-type-specific targeting is achieved without the need for targeting ligands. This dendrimer platform has the potential for clinical applications, with a favorable safety profile and scalability of production.^39,40 Here, we present a novel dendrimer-DPA conjugate for specific targeting of mitochondria within TAMs as a unique immunotherapy approach. We describe the design, synthesis, chemical characterization, in vitro mitochondrial localization and immune repolarization, and in vivo targeting properties of this dendrimer-DPA conjugate.

MATERIALS AND METHODS

Synthesis.

Materials and Reagents.

2,4-Pentanedione, boron tribromide, ethylene glycol ditosylate (4), sodium azide, potassium carbonate, 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide methiodide (EDC), 4(dimethylamino)pyridine (DMAP), GABA_BO-C_OH, N,N-diisopropyl ethyl amine (DIPEA), copper sulfate pentahydrate, sodium ascorbate, and 5-hexynoic acid, were purchased from Sigma-Aldrich. Cy5-NHS ester was purchased from GE healthcare and used as received. All of the anhydrous solvents were purchased from Sigma and were used as received. The hydroxyl PAMAM dendrimer with the ethylenediamine core (generation 4), 64 hydroxyl terminal groups, Pharma grade, compound (7) was purchased from Dendritech as a 13 wt % methanolic solution. Methanol was evaporated using a rotary evaporator prior to use. D-Cy5 was synthesized using previously published protocols.⁴¹

Characterization.

Proton nuclear magnetic resonance spectra (NMR) were recorded using a Bruker spectrometer (500 MHz). The data is reported as a chemical shift (δ ppm). The peaks are relative to the residual protonated solvent resonance and are reported in terms of multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet), coupling constants (J, Hz), and assignment. High-resolution mass spectra (HRMS) were recorded using electrospray ionization (ESI) recorded on a Bruker microTOF-II mass spectrometer in the positive mode. The samples were introduced via direct flow using the CH₃CN/H₂O (9:1) solvent system. The empirical formula confirmation is reported as protonated molecular ions [M + nH]ⁿ⁺ or adducts [M + nX]ⁿ⁺ (X = Na). High-pressure liquid chromatography (HPLC) was performed on a Waters system with a 2998 photodiode array detector and a 2475 multi λ fluorescence detector, an in-line degasser, a 1525 binary pump, and Waters Empower 2 Software. The separation was achieved on a Waters column (C18 symmetry 300, 5 μm, 4.6 × 250 mm²). The flow rate was maintained at 1.0 mL/min over a gradient from 100:0 (A/B) gradually increasing to 50:50 (A/B) at 25 min, 10:90 (A/B) at 35 min, and finally returning to 100:0 (A/B) at 40 min maintaining a flow rate of 1 mL/min (A: 0.1% trifluoroacetic acid (TFA) and 5% acetonitrile (ACN) in water and buffer B: 0.1% TFA in ACN) and the eluent was monitored at wavelengths of 210 and 650 nm. Flash chromatography was performed using the Teledyne Combiflash system. Size and ζ-potential measurements were carried out via dynamic light scattering (DLS) using Malvern Zetasizer Nano ZS as previously reported by our group.⁴²

Procedures for the Synthesis of Intermediates and Dendrimer Conjugates. Synthesis of Compound 2.

A solution of compound 1 (500 mg, 1.65 mmoles) and 2,4-pentanedione (0.17 mL, 1.65 mmoles) in anhydrous ethanol (15 mL) was heated at 60 °C in a microwave reactor for 20 h. Upon completion, the solvent was evaporated and the residue was dissolved in dichloromethane (DCM). The organic layer was washed with water and brine and dried over sodium sulfate. The dried organic layer was then evaporated and the crude product was purified using Combiflash. The pure fractions were obtained in 3% methanol in dichloromethane and were evaporated to afford an off-white solid product. Yield: 85%.

¹H NMR (500 MHz, CDCl₃) δ 7.86–7.66 (m, 2H), 7.08–6.87 (m, 2H), 6.51 (d, J = 0.7 Hz, 1H), 3.92 (s, 2H), 3.86 (s, 3H), 3.51 (q, J = 7.1 Hz, 2H), 3.42 (q, J = 7.1 Hz, 2H), 2.75 (s, 3H), 2.55 (s, 3H), 1.21 (t, J = 7.1 Hz, 3H), 1.12 (t, J = 7.1 Hz, 3H) (Figure S1).

¹³C NMR (126 MHz, CDCl₃) δ 170.14, 159.81, 157.44, 155.05, 147.69, 144.70, 130.00, 126.39, 113.98, 108.15, 100.81, 77.31, 77.05, 76.80, 55.34, 42.33, 40.62, 28.21, 24.68, 16.94, 14.36, 13.11 (Figure S2).

ESI-MS:

Theoretical for C₂₁H₂₆N₄O₂ is 366.21, observed is 367.21 [M + 1]⁺ (Figure S3).

Synthesis of Compound 3.

To a stirring solution of compound 2 (450 mg, 1.23 mmoles) in anhydrous DCM (5 mL), 1 M boron tribromide in tetrahydrofuran (THF; 6.13 mL, 5 equiv) was added dropwise at −60 °C. The stirring was continued for 2 h. Upon completion, the reaction mixture was quenched by pouring onto ice. The reaction mixture was diluted with water, and the product was extracted out in DCM. The DCM layer was evaporated to afford compound 3 as a yellow solid in quantitative yield.

¹H NMR (500 MHz, MeOD) δ 7.53 (d, J = 8.3 Hz, 2H), 6.95–6.75 (m, 3H), 3.98 (s, 2H), 3.54 (q, J = 7.1 Hz, 2H), 3.42 (q, J = 7.1 Hz, 2H), 2.83 (d, J = 7.6 Hz, 3H), 2.64 (s, 3H), 1.32–1.18 (m, 3H), 1.13 (t, J = 7.1 Hz, 3H) (Figure S4).

¹³C NMR (126 MHz, MeOD) δ 172.5, 160.27, 156.99, 129.79, 115.12, 100.35–98.18, 42.24, 40.70, 27.43, 21.56, 15.86, 12.88, 11.85 (Figure S5).

ESI-MS:

Theoretical for C₂₀H₂₄N₄O₂ is 352.19, observed is 353.19 [M + 1]⁺ (Figure S6).

Synthesis of Compound 5.

To a stirring solution of compound 4 (3 g, 8.1 mmoles) in DMF, sodium azide (790 mg, 12.15 mmoles) was added and the reaction mixture was heated at 40 °C for 24 h. Upon cooling to ambient temperature, the reaction mixture was diluted with ethyl acetate and washed with water. The organic layer was dried over sodium sulfate and evaporated. The crude product was purified via Combiflash using 30% ethyl acetate in hexane. The product was obtained as a transparent liquid. Yield: 60%.

¹H NMR (500 MHz, CDCl₃) δ 7.82 (d, J = 8.3 Hz, 2H), 7.37 (d, J = 8.1 Hz, 2H), 4.24–4.06 (m, 2H), 3.49 (t, J = 5.1 Hz, 2H), 2.46 (s, 3H) (Figure S7).

¹³C NMR (126 MHz, CDCl₃) δ 145.26, 132.63, 130.00, 128.00, 68.04, 49.62, 21.71, 0.02 (Figure S8).

Synthesis of Compound 6 (DPA-azide).

To a stirring solution of compound 3 (360 mg, 1.02 mmoles) in anhydrous tetrahydrofuran (THF, 5 mL), potassium carbonate (705 mg, 5.11 mmoles) was added and the stirring was continued for 30 min, followed by the addition of compound 5 (344.5 mg, 1.42 mmoles) in DMF (2 mL). The reaction mixture was stirred at 50 °C overnight. Upon completion, the solvent was evaporated. The residue was dissolved in ethyl acetate and washed with water. Column purification was performed to afford the pure product. Yield: 82%

¹H NMR (500 MHz, CDCl₃) δ 7.85–7.74 (m, 2H), 7.08–6.93 (m, 2H), 6.51 (d, J = 0.5 Hz, 1H), 4.26–4.15 (m, 2H), 3.93 (s, 2H), 3.62 (t, J = 5.0 Hz, 2H), 3.51 (q, J = 7.1 Hz, 2H), 3.41 (q, J = 7.1 Hz, 2H), 2.75 (s, 3H), 2.55 (s, 3H), 1.21 (t, J = 7.1 Hz, 3H), 1.11 (t, J = 7.1 Hz, 3H) (Figure S9).

¹³C NMR (126 MHz, CDCl₃) δ 170.03, 158.46, 157.49, 130.11, 127.07, 114.68, 108.18, 100.91, 77.29, 77.04, 76.78, 67.00, 50.21, 42.34, 40.63, 28.19, 24.55, 16.97, 14.38, 13.12, 0.02 (Figure S10).

ESI-MS:

Theoretical for C₂₂H₂₇N₇O₂ is 421.22, observed is 422.23 [M + 1]⁺ (Figure S11).

Synthesis of Compound 8.

To a stirring solution of compound 7 (800 mg, 0.056 mmoles) in anhydrous DMF (10 mL), pentynoic acid (82.44 mg, 0.840 mmoles) was added, followed by the addition of DMAP (108.12 mg, 0.885 mmoles) and EDC (169.65 mg, 0.885 mmoles). The reaction mixture was stirred at room temperature for 24 h. This was followed by the dialysis against DMF and then water. The aqueous solution was lyophilized to yield compound 8 as a hygroscopic white solid. Yield: 82%

¹H NMR (500 MHz, DMSO) δ 8.10–7.62 (D-NH-CO), 4.71 (s, D-OH), 4.03 (t, D-CH₂-COO), 3.49–3.22 (m, D-CH₂), 3.18–2.94 (m, D-CH₂), 2.78 (s, linker −C≡CH), 2.82–2.55 (m, D-CH₂), 2.45–2.30 (m, D-CH₂), 2.29–2.18 (m, D-CH₂) (Figure S12).

Synthesis of Compound 9.

To a stirring solution of compound 8 (500 mg, 0.033 mmoles) in anhydrous DMF (10 mL), BOC-GABA-OH (33 mg, 0.166 mmoles) was added, followed by the addition of DMAP (24.1 mg, 0.198 mmoles) and EDC (37.8 mg, 0.198 mmoles). The reaction mixture was stirred at room temperature for 24 h. This was followed by the dialysis against DMF and then water. The aqueous solution was lyophilized to yield compound 9 as a hygroscopic white solid. Yield: 82%

¹H NMR (500 MHz, DMSO) δ 8.19–7.68 (D-NH-CO), 4.72 (s, D-OH), 4.09–4.34 (m, D-CH₂-COO), 3.53–3.00 (m, D-CH₂), 2.80–2.57 (m, D-CH₂), 2.47–1.92 (m, D-CH₂), 1.62 (t, linker −CH₂), and 1.37 (BOC H) (Figure S13).

Synthesis of Compound 10.

A solution of compound 9 (400 mg) in 30% trifluoroacetic acid (TFA) in DCM was stirred vigorously for 12 h. The solution was coevaporated with methanol to remove TFA to afford compound 10 as a hygroscopic solid in quantitative yield, which was used directly for the next step.

¹H NMR (500 MHz, DMSO) δ 8.19–7.64 (D-NH-CO), 4.71 (s, D-OH), 4.09 (m, D-CH₂-COO), 4.02 (m, D-CH₂-COO), 3.53–3.00 (m, D-CH₂), 2.78 (s, linker −C≡CH), 2.85–2.53 (m, D-CH₂), 2.46–2.07 (m, D-CH₂), 1.72 (t, linker −CH₂) (Figure S14).

Synthesis of Compound 11.

To a stirring solution of compound 10 (403 mg, 0.026 mmoles) and compound 6 (165.45 mg, 0.392 mmoles) in the DMF/THF mixture, a solution of CuSO₄·5H₂O (1 mg, 0.004 mmoles) in water (1 mL) was added. This was followed by the addition of sodium ascorbate (1.5 mg, 0.008 mmoles) in water (1 mL). The reaction mixture was left to stir overnight at 40 °C. The aqueous solution was lyophilized to afford compound 11 as a white solid. Yield: 83%.

¹H NMR (500 MHz, DMSO) δ 8.05–7.67 (D-NH-CO), 7.65–7.56 (m, DPA ArH), 7.00–6.90 (m, DPA ArH), 6.77–6.71 (m, DPA ArH), 4.66 (DPA H), 4.35 (DPA H), 4.06–3.94 (m, D-CH₂-COO), 3.79–3.72 (m, linker-CH₂), 3.50–3.11 (m, D-CH₂), 3.11–2.89 (m, D-CH₂), 2.38–2.25 (m, D-CH₂), 2.22–1.99 (m, D-CH₂), 1.10 (t, DPA −CH₃), 0.93 (t, DPA −CH₃) (Figure S15).

Synthesis of Compound 12 (DPA–D-Cy5).

To a stirring solution of compound 11 (193 mg, 0.009 mmoles) in DMF, Cy5-NHS ester (12.13 mg, 0.019 mmoles) was added at pH 7.5. The reaction mixture was left to stir overnight at room temperature. The solution was then dialyzed against DMF followed by water. The aqueous solution was lyophilized to afford compound 12 as a blue solid. Yield: 72%.

¹H NMR (500 MHz, DMSO) δ 8.39–8.29 (m, Cy5H), 8.13–7.74 (D-NH-CO), 7.72–7.63 (m, DPA ArH), 7.34–7.27 (m, Cy5H), 7.04–6.96 (m, DPA ArH), 6.82–6.75 (m, DPA ArH), 6.62–6.52 (m, Cy5H), 6.32–6.21 (m, Cy5H), 4.9–4.56 (m, DPA H), 4.48–4.28 (m, DPA H), 4.05–3.94 (m, D-CH₂-COO), 3.87–3.75 (m, linker-CH₂), 3.50–3.19 (m, D-CH₂), 3.18–3.00 (m, D-CH₂), 2.72–2.57 (m, D-CH₂), 2.31–2.011 (m, D-CH₂). 76–1.62 (Cy5H), 1.44–0.89 (DPA −CH₃, linker H and Cy5H) (Figure S16).

HPLC:

Retention time: 23.1 min, Purity: 99.8% (Figure S17).

Synthesis of Compound 13 (D-DPA).

To a stirring solution of compound 8 (500 mg, 0.033 mmoles) and compound 6 (165.45 mg, 0.392 mmoles) in the DMF/THF mixture, a solution of CuSO₄·5H₂O (1 mg, 0.004 mmoles) in water (1 mL) was added. This was followed by the addition of sodium ascorbate (1.5 mg, 0.008 mmoles) in water (1 mL). The reaction mixture was left to stir overnight at 40 °C. The aqueous solution was lyophilized to afford compound 13 as a white solid. Yield: 87%.

¹H NMR (500 MHz, DMSO) δ 8.09–7.73 (D-NH-CO), 7.72–7.62 (m, DPA ArH), 7.04–6.98 (m, DPA ArH), 6.85–6.77 (m, DPA ArH), 4.73 (t, DPA H), 4.43 (DPA H), 4.00 (m, D-CH₂-COO), 4.82 (t, DPA H), 3.50 (m, DPA H), 3.48–2.97 (m, linker-CH₂), 2.73–2.57 (m, D-CH₂), 2.28–2.07 (m, D-CH₂), 1.84 (t, linker H), 1.17 (t, DPA −CH₃), 1.01 (t, DPA −CH₃) (Figure S18).

HPLC:

Retention time: 24.2 min, Purity: 99% (Figure S19).

Synthesis of DPA-triazole.

The synthetic protocol for the synthesis of DPA-triazole is described in the Supporting Information. The ¹H and ¹³C NMR spectra for DPA-triazole are presented in Figures S20 and S21.

Biology Studies.

Materials.

Dulbecco’s modified Eagle’s medium (DMEM), Roswell Park Memorial Institute medium (RPMI), fetal bovine serum (FBS), penicillin–streptomycin (P/S), 0.25% trypsin-ethylenediaminetetraacetic acid (EDTA), NucBlue fixed cell stained, goat anti-rabbit Alexa Fluor 488, TRIzol, and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) reagents were purchased from Invitrogen (Carlsbad, CA). Lysine-coated glass-bottom dishes were purchased from MatTek Inc. (Ashland, MA). Anti-AIF antibody was purchased from Abcam (Cambridge, U.K.). Methanol, 4% formalin solution, Triton-X, bovine serum albumin (BSA), and lipopolysaccharide (LPS) were purchased from Sigma-Aldrich (St. Louis, MO). Mitochondria isolation for the mammalian cell kit and normal goat serum (NGS) were purchased from ThermoFisher (Waltham, MA). Lectin Dylight 594 was purchased from Vector Laboratories Inc. (Burlingame, CA). Phosphate-buffered saline (PBS) and Tris-buffered saline (TBS) were obtained from Corning Inc. (Corning, NY).

Mitochondrial Isolation.

To assess comparative binding affinities, mitochondria were isolated from kidneys and adrenal glands harvested from neonatal rabbits. These tissues were chosen because TSPO has been shown to have high expression in steroid producing tissues such as adrenal glands.⁶⁸ Mitochondrial isolation was performed using a mitochondria extraction from mammalian tissue kit as per the manufacturer’s procedure (ThermoFisher).

TSPO Binding Affinity of D-DPA.

Binding of compounds to the TSPO receptor was evaluated using a radiolabeled membrane binding assay in tandem with size exclusion chromatography. [³H]-PK11195 (Cat# NET885250UC, Perkin Elmer, Boston, MA) was used as the radioligand and mitochondria, as isolated above, used as the receptor source. Briefly, mitochondria were incubated with a range of concentrations of the tested compounds (1 pM–100 μM) and [³H]-PK11195 (1.2 nM) in 50 mM Tris-HCl buffer at pH 7.4 and in a total volume of 100 μL for 30 min and at room temperature (RT). Total binding was determined in the presence of [³H]-PK11195 alone, and nonspecific binding was determined in the presence of both [³H]-PK11195 and 100 nM unlabeled PK11195. At the end of the incubation period, 90 μL of the reaction mixture was transferred to pre-equilibrated, 7 K molecular weight cut off, 96-well, Zeba Spin Desalting plates (Cat# 89807, Thermo Scientific, Rockford, IL). [³H]-PK11195-bound mitochondria were eluted upon centrifugation at 1000g for 2 min (Beckman GS-6R centrifuge with PTS-200 swinging bucket rotor), and the radioactivity in 80 μL of the flow was detected using solid scintillator-coated 96-well LumaPlates (Perkin Elmer, Cat# 6005630) in conjunction with Perkin Elmer’s TopCount instrument. Finally, IC₅₀ values were determined from CPM results of the specific binding using Microsoft Office Excel, IDBS’ XLfit, and GraphPad Prism programs.

Cell Culture.

HMC3 human macrophages were acquired from ATCC (Manassas, VA). BV2 murine microglia were obtained from the Children’s Hospital of Michigan’s cell culture facility. GL261 murine glioblastoma cells were obtained from the DTP/DCTD/NCI Tumor Repository (Frederick, MA). HMC3 and BV2 cells were maintained in DMEM supplemented with 10% FBS and 1% P/S. The GL261 cells were maintained in low glutamine RPMI supplemented with 10% FBS, 1% P/S, and 1% L-glutamine. The cells were kept at 37 °C in a 5% CO₂ atmosphere. Treatments were performed in half-serum media (5% FBS).

Mitochondrial Colocalization and Quantification.

Imaging and colocalization of mitochondria with dendrimers were performed according to previously published protocols.⁴¹ Briefly, HMC3 macrophages were seeded on glass-bottom culture dishes. The HMC3 cells were used to image mitochondrial targeting due to their large cell bodies. The cells were treated with a fluorescently labeled dendrimer (D-Cy5) and a DPA-conjugated dendrimer (Cy5-D-DPA) for 48 h at 50 μg/mL; fixed in 4% formalin solution; and stained with rabbit anti-mouse AIF primary antibody (1:200) and anti-rabbit Alexa Fluor 488 (1:200) to label mitochondria, Lectin Dylight 594 to label cell membranes, and DAPI to label cell nuclei.

For quantification of the dendrimer content in isolated mitochondria, BV2 murine microglia were used due to greater yield of isolated mitochondria. Dendrimer quantification from cell extracts was performed as described previously with modifications for mitochondrial isolation.⁶⁹ The cells were treated with 50 μg/mL D-Cy5 or Cy5-D-DPA for 24 h and were freshly fractionated to yield the isolated mitochondria and cytosolic fractions. The isolated mitochondria were then resuspended in isolation buffer, and dendrimers were extracted via three freeze/thaw cycles with liquid nitrogen. Dendrimer fluorescence in each fraction was measured using a Shimadzu RF-3501PC spectrofluorophotometer (Shimadzu Corporation, Columbia, MD). Fluorescence intensities were converted into masses using calibration curves of known dendrimer concentrations.

Analysis of Inflammatory Expression.

BV2 murine microglia were stimulated with LPS at 300 EU/mL for 3 h, followed by cotreatment with PK11195 or the DPA-conjugated dendrimer (D-DPA) with LPS for 24 h. Then, the cells were exposed to fresh media for 24 h and collected for analysis of cellular and extracellular signals. LPS was chosen as a stimulant despite the induction of proinflammatory immune response due to its role in mediating cancer cell metastasis, proliferation, and immune recruitment.^41,42 The cells were collected in TRIzol, and mRNA was extracted according to the manufacturer’s procedure. cDNA conversion was then performed using a cDNA conversion kit (ThermoFisher) and a thermocycler (Bio-Rad Laboratories, Hercules, CA). The cells were then analyzed for immune expression using the rt-qPCR StepOne Plus system (Applied Biosystems). Primer sequences used were as follows: TNFα (F: CCA GTG TGG GAA GCT GTC TT; R: GTG TAA TTA AGC CTC CGA CTT G), IL1β (F: AGC TTC AAA TCT CGA AGC AG; R: TGT CCT CAT CCT GGA AGG TC), arginase-1 (F: TCA TGG AAG TGA ACC CAA CTC TTG; R: TCA GTC CCT GGC TTA TGG TTA CC), and GAPDH (F: TGT CGT GGA GTC TAC TGG TGT CTT C; R: CGT GGT TCA CAC CCA TCA CAA). The cell supernatants were assessed for TNFα and nitrite secretion using the mouse TNFα Quantikine ELISA kit (R&D Systems, Minneapolis, MN) and the Griess reagent assay kit (Promega, Madison, WI) according to the manufacturer’s procedure, respectively. The samples were read on a Synergy Mx microplate reader (BioTek, Winooski, VT). Cytotoxicity of D-DPA compared to PK11195 was assessed using the MTT assay.

Tumor Inoculations.

All animals were housed at the Johns Hopkins University animal facilities and were given free access to food and water. All animal experiments were conducted in accordance with protocols approved by the Johns Hopkins University institutional animal care and use committee.

To establish the GL261 orthotopic immunocompetent model of glioblastoma, male and female C57BL/6 (Jackson Laboratory, Bar Harbor, ME) mice of 6–8 weeks of age were intracranially implanted with the GL261 cells. The cells were brought to a concentration of 100 000 cells per 2 μL. Mice were anesthetized using a ketamine/xylazine for survival surgeries. A midline scalp incision was created, and a burr hole was drilled 1 mm posterior to the bregma and 2 mm lateral to the midline to inject cells into the striatum. A 2 μL Hamilton syringe (Hamilton Company, Reno, NV) was lowered to a depth of 2.5 and injected 2 μL of cell solution over 10 min with an automated syringe pump (Stoelting Co., Wood Dale, IL), and the syringe was withdrawn slowly. The incision was sutured together, and an antibiotic cream was applied to the wound.

To determine TSPO overexpression in this model of glioblastoma, tumors and contralateral hemisphere tissues were dissected from brains collected 14 days after tumor inoculation. Tissues were processed for mRNA extraction and rt-qPCR, as described above, with additional agitation initially to dissociate tissues. Premade TSPO primers were obtained from Bio-Rad.

To assess systemic biodistribution of Cy5-D-DPA compared to D-Cy5, glioblastoma brain tumor-bearing mice were intravenously injected with dendrimers at 55 mg/kg on day 14 postinoculation. Clearance organs (kidney, liver, spleen) were collected 24 h later after perfusion. Organs were dissected into equivalent mass samples and homogenized with stainless steel beads in a bullet blender homogenizer (Next Advance Inc., Troy, NY) to extract dendrimers into the methanolic solution. Fluorescence was measured using the spectrofluorophotometer and converted to quantities using calibration curves of known dendrimer concentrations.

Tissue Processing and Immunohistochemistry.

Glioblastoma brain tumor-bearing mice were injected intravenously with D-Cy5 or Cy5-D-DPA on day 14 postinoculation. Mice were perfused, and brains were collected 24 h after injection. Brains were fixed in 4% formalin solution, followed by sucrose gradient to remove residual formalin (10%, 20%, and then 30% sucrose in PBS overnight for each). Brains were sectioned axially into 30 μm using a Leica CM 1905 cryostat (Wetzlar, Germany) and stained with DAPI to label nuclei, Iba1 (1:200) to label TAMs, and AIF (1:200) to label mitochondria. Slices were blocked in 1× TBS + 0.1% Triton-X + 1% BSA + 5% NGS for 4 h at room temperature. Slices were then incubated with primary antibodies diluted in 1× TBS + 0.1% Triton-X + 1% BSA overnight at 4 °C. Then, the slices were washed and incubated with secondary antibodies (goat anti-rabbit 488) for 2 h at room temperature. Finally, the slices were incubated with DAPI for 15 min, mounted, and sealed.

Confocal Imaging.

Images were acquired using a Zeiss LSM710 confocal microscope (Hertfordshire, U.K.). Background fluorescence in cells and tissues were set with untreated samples. Microscope settings such as laser intensity, gain, and offset were kept constant across compared images. Zen Lite 2011 software was used to process the obtained images, and any adjustments to brightness and contrast were kept constant across all compared images. Colocalization and fluorescence profiles were obtained with Zen Lite 2011 software.

Statistical Analyses.

Graphs and statistical analyses were performed using GraphPad Prism v8.0 software (San Diego, CA). Error bars presented represent mean ± standard errors. Statistical significance between colocalization coefficients was determined with Student’s t-test. Significances in comparisons between PK11195 and D-DPA were performed with two-way ANOVAs.

RESULTS

Synthesis of the Dendrimer-DPA Conjugate and Intermediates.

To enable mitochondria targeting within TAMs in glioblastoma, DPA was conjugated to generation 4 hydroxyl-terminated polyamidoamine dendrimers via non-cleavable linkages. Chemical conjugation of DPA on the surface of the dendrimer is challenging due to the absence of functional groups that can be easily modified to attach a linker. To address this issue, we first synthesized a conjugable form of DPA (Figure 1A) with an azide terminal group separated by a two carbon linker. This served two purposes, (i) the presence of an azide terminal group participated in the copper (I)-catalyzed alkyne-azide click (CuAAC) reaction on the dendrimer surface, and (ii) the resulting aromatic triazole ring with two carbon spacers for enhanced TSPO binding affinity, as modifications with phenethyl ether side-chains have resulted in DPA analogues with highest TSPO binding affinities.²⁷ The synthesis of DPA-azide began with the condensation of compound 1 with 2,4-pentanedione to generate pyrazolopyrimidine derivative 2, which on subsequent cleavage of the methoxy group with boron tribromide afforded phenolic derivative 3. The two carbon orthogonal linker was synthesized by monoazidation of ethylene glycol ditosylate 4 to afford linker 5. The phenolic intermediate 3 was subsequently reacted with 5 using potassium carbonate and heated mildly heating to afford a clickable version of DPA as DPA-azide (6). To enable fluorescence imaging of DPA-conjugated dendrimers, the dendrimer surface was functionalized with two orthogonal functional groups to obtain a trifunctional dendrimer for the simultaneous conjugation of a cyanine 5 (Cy5) fluorescent probe and DPA (Figure 1B). Using simple esterification reactions, dendrimers (7) were first modified with pentynoic acid to afford alkyne groups to enable the click reaction with azide-terminated DPA (6). The alkyne-terminating dendrimer 8 was then modified in two steps by first reacting with GABA–BOC-OH, followed by BOC deprotection in mild acidic conditions, to afford amine groups to enable conjugation to Cy5-NHS ester, resulting in a trifunctional dendrimer with two orthogonal functional groups for further attachment (10). The alkyne groups in trifunctional dendrimer 10 were further reacted with DPA-azide (6) via the CuAAC click reaction to afford dendrimer 11, which was subsequently reacted with Cy5-NHS via the activated acid-amine coupling reaction to obtain the final fluorescently labeled dendrimer-DPA conjugate (12). For binding and efficacy studies, a nonfluorescently labeled dendrimer-DPA conjugate was synthesized by simply reacting alkyne-terminating dendrimer 8 with DPA-azide (6) to afford D-DPA (13, Figure 1C). We conjugated only ~9 molecules of DPA (19% w/w) on the surface of the hydroxyl dendrimer to keep the inherent biodistribution and targeting potential of the parent dendrimer intact. In addition, for comparison purposes, we also synthesized a DPA-triazole analogue by reacting DPA-azide with propargyl alcohol (Scheme S1).

Figure 1.

Synthesis of DPA derivative and dendrimer conjugates. Schematic representation of the synthesis pathways for (A) conjugable azide-terminated DPA derivative (6, DPA-N₃), (B) Cy5 fluorescently labeled dendrimer-DPA conjugate (12, Cy5-D-DPA), and (C) dendrimer-DPA (13, D-DPA).

Chemical Characterization of Intermediates and Final Dendrimer-DPA Conjugates.

The structures of intermediates, DPA-triazole analogue, and the final dendrimer-DPA conjugates were characterized using ¹H NMR, mass spectroscopy, and HPLC (Figures S1–S21). The ¹H NMR spectra of trifunctional dendrimer 10 (Figure 2A, green spectrum) exhibited characteristic ester methylene protons from two linkers at δ 4.90 and 4.02 ppm. The comparative integration of protons from dendrimer internal amides at δ 8.19–7.64 ppm with ester methylene protons suggested the attachment of on an average ~9 alkyne linkers and ~2 amine terminating linkers. The success of the click reaction between the DPA-azide and trifunctional dendrimers was analyzed via ¹H NMR by comparing the spectra of the resulting product (blue spectrum) with the starting materials (red and green spectra), showing the characteristic dendrimer peaks along with the DPA aromatic peaks in between δ 7.65–7.56, 7.00–6.90, and 6.77–6.71 ppm and aliphatic methyl protons at δ 1.10 and 0.93 ppm. Finally, the success of cy5 conjugation was evident from the Cy5 protons in the aromatic region (magenta spectrum). The purities of the final conjugates (D-DPA and Cy5-D-DPA) were >99%, as analyzed by HPLC (Figures 2B and S17 and S19). The conjugation of DPA to the dendrimer surface resulted in a shift in retention time from 19.9 min for the trifunctional dendrimer to 23.1 min for Cy5-D-DPA (Figure 2B). DLS and ζ-potential measurements revealed a hydrodynamic radius of 4.8 nm (Figure 2C,D) and ζ-potential of +3.6 mV (Figure 2D,E).

Figure 2.

Chemical characterization of intermediate products and final conjugates. (A) Comparative ¹H NMR spectra of the fluorescently labeled dendrimer-DPA conjugate (Cy5-D-DPA), dendrimer-DPA (D-DPA), DPA-azide, and trifunctional dendrimer. Characteristic proton signals are labeled. (B) Comparative HPLC chromatogram of intermediates and final conjugate. (C) Size distribution of D-DPA measured using dynamic light scattering (DLS). (D) Table showing physicochemical characterization of D-DPA. (E) ζ-Potential distribution of D-DPA measured by DLS.

TSPO Binding Affinity of Dendrimer-DPA Conjugates.

To evaluate whether D-DPA was active as a TSPO ligand, TSPO binding affinity in mitochondria was measured for D-DPA and was compared to free PK11195, DPA-triazole, and an empty dendrimer (D-OH) (Table 1). The half-maximal inhibitory concentration (IC₅₀) of D-DPA was 70 nM ± 4, approximately 14-fold less potent relative to free PK11195 (5 nM ± 0.5) and ~1.4-fold more potent than the DPA-triazole analogue (100 nM ± 20). The negative control empty dendrimer (D-OH) displayed no binding affinity.

Table 1.

Comparative Half-Maximal Inhibitory Concentration (IC₅₀) of PK11195, DPA-Triazole Monomer, D-DPA, and PAMAM-D-OH Affinity to TSPO Receptor

Compound	Structure	Average IC50
PK11195 (F wt. 353 Da)
DPA-triazole (F. wt. 478 Da)
D-DPA (F. wt. 18.7 kDa)
D-OH (F. wt. 14.3 kDa)

DPA Conjugation to Dendrimer Enables Mitochondrial Targeting.

Before proceeding with in vitro experiments, we first assessed the potential cytotoxicity of PK11195 and D-DPA via cell viability measurements. BV2 murine microglia were exposed to PK11195 or D-DPA over a concentration range of 1–1000 μg/mL for 24 h. Consistent with previous in vitro studies,^43,44 PK11195 demonstrated significant dose-dependent toxicity, with the 1000 μg/mL dose reducing cell viability to ~5% of controls (Figure S22). In contrast, D-DPA exhibited significantly less cytotoxicity (p < 0.0001 PK11195 vs D-DPA), with the 1000 μg/mL dose exhibiting ~60% cell viability. PK11195 has been shown to induce apoptosis via interference with mitochondrial permeability pores; therefore, the slightly weaker binding affinity by D-DPA may contribute to the significantly reduced cytotoxicity.⁴⁵ This indicates that the dendrimer delivery of a TSPO ligand may ameliorate cytotoxic effects for more effective and safer therapies.

A previous in vivo dendrimer study has shown that dendrimers selectively target TAMs in gliomas.³³ To explore the mitochondria targeting capabilities of DPA-conjugated dendrimers, HMC3 human macrophages were exposed to Cy5-D-DPA for 48 h at 50 μg/mL. The cells were then fixed and stained to label nuclei, cell membranes, and mitochondria. Cy5-D-DPA exhibited a highly punctated signal corresponding to the mitochondria, as shown by the yellow signal, indicating an overlap between the dendrimer (red) and mitochondrial (green) signals (Figure 3A). In contrast, the unmodified dendrimer (D-Cy5) exhibits a diffuse perinuclear signal consistent with what we have observed previously in vitro and in vivo.⁴⁶ While some overlap between the dendrimer and mitochondrial signals is observed with D-Cy5, this appears to arise from the broad, cytosolic signal pattern of D-Cy5 rather than the specific interactions with mitochondria. Representative fluorescence line profiles through cells show that the dendrimer and mitochondrial signals correlate closely with Cy5-D-DPA, whereas D-Cy5 exhibits regions of the mitochondrial signal without the corresponding dendrimer signal and vice versa (Figure 3B). Semiquantitative analysis shows that the conjugation of DPA to the dendrimer increases its colocalization coefficient by ~2-fold (Figure 3C, p < 0.0001).

Figure 3.

Conjugation of DPA to dendrimers enables targeting of mitochondria. HMC3 human macrophages were treated with a fluorescently labeled dendrimer (red) with (Cy5-D-DPA) or without (D-Cy5) DPA conjugation. Following 48 h of exposure, the cells were fixed and stained with DAPI to label cell nuclei (blue) and AIF to label mitochondria (green) for confocal imaging. (A) Conjugation of DPA to dendrimers improves mitochondrial localization, as indicated by the yellow regions, signifying colocalization of the green mitochondrial and red dendrimer signals. D-Cy5 exhibits a diffuse perinuclear signal. (B) Representative fluorescence line profiles through cells demonstrate the close association between the Cy5-D-DPA and mitochondrial signals, whereas the D-Cy5 signal does not exhibit such correspondence. (C) Cy5-D-DPA exhibits a significantly greater colocalization coefficient with the mitochondrial signal than D-Cy5. ***p < 0.001.

To further explore the mitochondria-targeting properties of D-DPA, BV2 murine microglia were treated with unmodified or DPA-conjugated dendrimers for 24 h. BV2 microglia were chosen due to their higher yield of isolated mitochondria compared to HMC3 macrophages. The cells were then fractionated to yield mitochondrial and cytosolic fractions. Cy5-D-DPA exhibited significantly greater partitioning to the isolated mitochondrial fraction compared to D-Cy5 (Figure S23A, p = 0.0021). However, total cellular internalization was not changed (Figure S23B, p = 0.859), indicating that the dendrimer uptake properties were preserved with DPA conjugation, consistent with previous studies, where up to 20 wt % loading of conjugated therapies did not interfere with dendrimer interactions with cells.⁴⁷

Dendrimer-DPA Promotes Antitumor Immune Signaling.

To evaluate the impact of PK11195 or D-DPA treatment on modulating TAM-like phenotype, markers of immune activation were assessed after treatment in LPS stimulated BV2 murine microglia. LPS was chosen due to its role in promoting tumor immune recruitment, metastasis, and proliferation.⁴⁸ D-DPA treatment (100 μg/mL on a DPA loading basis) significantly increased secretions of TNFα by activated microglia ~2-fold compared to PK11195 treatment (100 μg/mL) (Figure 4A). D-DPA treatment similarly upregulated the expression of antitumor, proinflammatory cytokines TNFα and IL1β ~3-fold compared to PK11195 treatment (Figure 4C,D). Notably, both PK11195 and D-DPA treatment did not upregulate anti-inflammatory signal Arg-1 (Figure 4E). Both PK11195 and D-DPA did decrease the secretion of reactive oxygen species (Figure 4B), which are implicated in tumor recruitment of TAMs, tumorigenesis via inducing DNA damage, and tumor proliferation.^49,50 These results indicate that in the context of TAM-like immune activation, D-DPA exerts TSPO agonism activity to promote proinflammatory, antitumor immune signaling, while reducing oxidative stress.

Figure 4.

D-DPA treatment increases antitumor inflammatory signals. BV2 murine microglia were stimulated with LPS at 300 EU/mL for 3 h, followed by cotreatment with PK11195 or the DPA-conjugated dendrimer (D-DPA) for 24 h. Then, the cells were exposed to fresh media for 24 h and collected for analyses of cellular and extracellular signals. (A) Treatment with D-DPA significantly increases the secretion of TNFα, a tumor-killing signal, from stimulated BV2 microglia compared to PK11195 treatment. **p < 0.01. (B) Both treatments reduced secretion of nitrite, a reactive oxygen species indicative of oxidative stress. Analyses of mRNA expression via rt-qPCR demonstrates that D-DPA significantly increases the expression of antitumor signals (C) TNFα and (D) IL1β while limiting the expression of (E) protumor cytokine arginase-1 (Arg-1). ***p < 0.001.

Systemically Administered Dendrimer-DPA Targets Mitochondria Specifically within TAMs in vivo.

Based on these promising in vitro results, we proceeded to evaluate the in vivo targeting properties of the dendrimer-DPA conjugate in an orthotopic, immunocompetent model of glioblastoma. First, to validate the rationale for TSPO targeting in this model of glioblastoma, we compared the expression of TSPO within and outside the brain tumor. TSPO was significantly upregulated in the tumor ~5-fold compared to the contralateral hemisphere (Figure S24, p < 0.0001). In addition, we also examined the systemic biodistribution of Cy5-D-DPA compared to unmodified D-Cy5. TSPO has been shown to exhibit high expression in kidneys, livers, and spleens,^20,51 creating the potential for off-target Cy5-D-DPA accumulation. Organs were collected 24 h after systemic injection, and dendrimers were extracted and quantified using fluorescence spectrometry. Cy5-D-DPA exhibited similar levels in kidney, liver, and spleen compared to D-Cy5 (Figure S25). This indicates that Cy5-D-DPA is not interacting with TSPO expressed in peripheral organs for increased accumulation. Therefore, the dendrimer transport properties dominate, and DPA interactions with TSPO arise only once dendrimers have carried DPA into the intracellular space.

Cy5-D-DPA was intravenously injected into mice with orthotopic glioblastoma brain tumors 14 days after tumor inoculation. Brains were collected 24 h after injection, fixed, and stained for confocal imaging. Imaging in the tumor core indicated that Cy5-D-DPA fully penetrated the solid brain tumor upon systemic administration and targeted mitochondria within the tumor (Figure 5, white arrows). This signal contrasts with our previously observed diffuse, cytosolic signal pattern in vivo in TAMs of unmodified dendrimers, indicating that the addition of TSPO ligand DPA confers mitochondrial targeting.³⁸ Unlike in the in vitro images, nuclear targeting of Cy5-D-DPA was not observed. Notably, the green signals corresponding to mitochondria were observed throughout the tumor, but Cy5-D-DPA only exhibited cellular signals in specific cells.

Figure 5.

D-DPA targets mitochondria within the glioblastoma tumor. Glioblastoma brain tumor-bearing mice were injected intravenously with fluorescently labeled dendrimer-DPA conjugates (Cy5-D-DPA, red) 14 days after tumor inoculation. 24 h after injection, brains were collected, fixed, and stained with DAPI to label cell nuclei (blue) and AIF to label mitochondria (Mito, green). Cy5-D-DPA targets mitochondria within the tumor (white arrows) in tumor-associated macrophages upon systemic administration. Localization with the nucleus, as seen in the in vitro images, was not observed in vivo.

To examine the cell-type localization, we then stained brains with Iba1 to label TAMs. We observed that Cy5-D-DPA penetrated the solid tumor and distributed uniformly throughout, with high fidelity for the tumor border (Figure 6A). The minimal Cy5-D-DPA signal was observed in the surrounding healthy tissue, indicating highly specific tumor targeting. Imaging at higher magnification revealed that the dendrimer signal was within TAMs (Figure 6B, white arrows). Therefore, DPA-conjugated dendrimers localize specifically to TAMs and target their mitochondria.

Figure 6.

D-DPA specifically localizes to tumor-associated macrophages in the glioblastoma tumor upon systemic administration. Glioblastoma brain tumor-bearing mice were injected intravenously with fluorescently labeled dendrimer-DPA conjugates (Cy5-D-DPA, red) 14 days after tumor inoculation. 24 h after injection, brains were collected, fixed, and stained with DAPI to label cell nuclei (blue) and Iba1 to label tumor-associated macrophages (TAMs, green) for confocal imaging. (A) Whole tumor images (top panels) demonstrate that Cy5-D-DPA penetrates throughout the solid glioblastoma tumor after systemic administration. Imaging of the tumor border (bottom panels) demonstrates that Cy5-D-DPA localizes specifically within the tumor with high fidelity for the tumor border. (B) Cy5-D-DPA signal colocalizes with Iba1 (white arrows), indicating specific targeting of TAMs within the glioblastoma tumor.

DISCUSSION

In this study, we present a novel TSPO targeting generation 4 hydroxyl-terminated PAMAM dendrimer by conjugating DPA targeting ligands to the dendrimer surface. Upon systemic administration, the dendrimer conjugate penetrates and distributes homogeneously within the solid GBM tumor with high specificity compared to healthy brain tissue. The surface decoration with DPA enables specific targeting of mitochondria within TAMs. In vitro, we demonstrate that using the dendrimer conjugate to target TSPO, we can induce antitumor immune signaling to potentially promote the tumor-killing immune response.

DPA was conjugated to the surface of the dendrimer through a copper-catalyzed click reaction. This method of copper-catalyzed click reaction was utilized as the conjugation chemistry due to its robust and highly efficient chemistry. In addition, previous reports have shown that the modification of the side chain of DPA with a phenethyl ring achieved the most effective binding affinity.²⁷ By utilizing this click chemistry approach, a triazole ring was introduced with an ethyl spacer. We hypothesized that by mimicking this aromatic ring shown to enhance binding affinity to TSPO, the optimal binding affinity of DPA on the dendrimer could be achieved. To assess binding affinity to TSPO, mitochondria were isolated from adrenal tissues. PK11195 exhibited 5 nM binding affinity to TSPO, consistent with previous reports (Table 1).²⁷ As expected, the unmodified dendrimer exhibited no binding affinity for TSPO, as shown by >100 000 nM IC₅₀. Dendrimer-DPA conjugates (D-DPA) exhibited an average IC₅₀ value of 70 nM. While we had expected the binding affinity to improve compared to PK11195 or the DPA-triazole monomer due to multivalency effects,⁵² D-DPA only showed a slight improvement to the DPA-triazole monomer. DPA moieties may be scattered across the dendrimer surface to prevent the simultaneous presentation of multiple ligands to the TSPO receptor. However, the ~1.4-fold enhanced binding affinity exhibited by D-DPA compared to DPA-triazole is consistent with some reports where conjugation of ligands yields unchanged or decreased binding affinity for their target due to attachment to bulky macromolecules, which can interfere with receptor interactions.^53,54 However, the compromised binding affinity as compared to the free TSPO ligand (PK11195) can be overcome by inherent TAM targeting potential of dendrimers in vivo.

While nanoparticle-mediated strategies have extensively explored cell-type-specific targeting, the development of enhanced nanoparticles displaying targeting ligands to achieve intracellular targeting in addition to cell-specific targeting remains an incredibly promising and relatively little explored space. In particular, mitochondria within activated immune cells, such as TAMs in the context of cancers, are attractive therapeutic targets due to their roles in regulating the tumor immune response. The mitochondrial colocalization of Cy5-D-DPA compares favorably to other nanoparticle platforms targeting mitochondria, which exhibit colocalization of 40–60%.^55,56 Notably, the colocalization of Cy5-D-DPA also compared favorably to TPP-conjugated PAMAM dendrimers, which exhibit colocalization with mitochondria of ~65%.⁴¹ This indicates that receptor-mediated mitochondrial targeting may yield more effective targeting compared to electrostatic-mediated strategies. These results demonstrate that these hydroxyl PAMAM dendrimers are promising vehicles for intracellular delivery, and they can be further modified to effectively target intracellular compartments such as the mitochondria. Notably, while previous studies have not explored this mitochondrial targeting in vivo, we demonstrate that Cy5-D-DPA specifically targets mitochondria within TAMs following systemic administration.

Interestingly, D-DPA exhibited robust localization with the nucleus in vitro, which has not been observed previously with these dendrimers. This phenomenon may arise from the role of TSPO in mediating mitochondria-nuclear signaling.⁵⁷ TSPO upregulation has also been shown to occur in the context of inflammation on other membrane-bound organelles,⁵⁸ so D-DPA may be targeting TSPO expressed on the nuclear membrane. However, we did not observe nuclear targeting with D-DPA in the GBM tumor in vivo. This nuclear targeting property is beyond the scope of this study but warrants further exploration and may indicate the delivery of genetic material or therapies that target nuclear-localized factors.

TSPO targeting has been shown to influence immune activation, although its effects have shown conflicting results. Some studies have reported anti-inflammatory, neuroprotective activity of PK11195,^59,60 while others have shown that PK11195 exacerbates neurodegeneration and promotes apoptosis.^61,62 This differential activity appears to depend on PK11195 activity as an agonist or antagonist of TSPO in specific disease contexts. Here, we show that D-DPA significantly upregulates antitumor signaling of TNFα and IL1β, as well as limits the expression of protumor signal Arg-1 (Figure 4). This is consistent with previous reports where TSPO ligands can induce the expression of TNFα and other proinflammatory cytokines.^63,64 Due to the slightly weaker binding affinity of D-DPA for TSPO compared to PK11195, this improvement in promoting antitumor signaling is likely attributable to the high levels of cellular internalization exhibited by these dendrimers.⁶⁵ This has promising implications for cancer therapy, since the delivery of gene vectors encoding TNFα or the direct administration of TNFα have been shown to inhibit tumor progression.⁶⁶ Interestingly, PK11195 has been shown to exacerbate TNFα-induced apoptosis, which may contribute to its dose-dependent cytotoxicity (Figure S22). However, despite significantly increasing the induced expression levels of TNFα, D-DPA does not demonstrate the associated TNFα-induced cell death. This may be due to weaker interactions with TSPO due to its binding affinity, although the precise mechanism of how D-DPA seems to decouple the promotion of TNFα expression from TNFα-induced apoptosis warrants further investigation. Both PK11195 and D-DPA reduced secretion of nitric oxide, consistent with previous reports where PK11195 treatment ameliorates ROS production.⁶⁷ These findings have exciting implications for glioblastoma treatment, as the dendrimer presentation of the targeting ligand alone elicits antitumor immune activation in vitro and can be coupled with dendrimer delivery of mitochondria-targeted therapeutics for multimodal synergistic effects.

Interestingly, we observed a small amount of dendrimer signal outside TAMs, and the morphology of these signals suggest that they reside within the cells. Recent studies in interactions between the tumor cells and immune cells have revealed that the whole mitochondria may be transferred between the cells.⁶⁸ Metabolic shifts in cancer cells lead them to generate energy via glycolysis rather than mitochondrial respiration, and highly glycolytic cancer cells correlate with poor patient outcomes.⁶⁹ However, cancer cells exhibiting respiratory incompetence are also deficient in other tumor progression functions and communicate with the surrounding cells such as stromal cells and TAMs to acquire functional mitochondria to compensate.⁶⁹ Therefore, the small amount of signal observed outside TAMs may arise from Cy5-D-DPA being transferred to cancer cells along with mitochondria. The reasons and mechanisms behind this emerging concept require further exploration but may have significant implications for mitochondrial-targeted immunotherapies.

CONCLUSIONS

In this study, we present the design and synthesis of a DPA-conjugated dendrimer for targeting of mitochondria in TAMs. The dendrimer surface decorated with DPA moieties exhibited strong binding to TSPO, favorable in vivo TAM targeting, and cellular internalization. We demonstrate that DPA-conjugated dendrimers exhibit significantly greater colocalization with mitochondria compared to unmodified dendrimers. By targeting TSPO, the DPA-conjugated dendrimers stimulate antitumor immune signaling superior to PK11195, a classical TSPO ligand. In vivo, systemically administered dendrimer-DPA penetrates the solid brain tumor with high specificity and localizes within TAMs to target mitochondria on a cell-type-specific level. These results suggest that dendrimer-DPA may significantly improve targeted delivery of immunotherapies to mitochondria specifically within TAMs for the treatment of glioblastoma and other cancers.