Engineering pH-Responsive Dendrimer-STAT3 Inhibitor Conjugates for Intracellular Delivery

Abstract

Small-molecule STAT3 inhibitors face numerous challenges in clinical translation, including poor water solubility, rapid systemic clearance, low bioavailability, poor selectivity, and high cytotoxicity. To address these limitations, we conjugated the potent STAT3 inhibitor−LLL12 to generate six (G6) hydroxyl-terminated (poly(amidoamine)) PAMAM dendrimers using pH-sensitive linkers: sulfonyl carbamate (carbamate), sulfonyl carbamoyl (amide), and hydrazone for intracellular drug delivery. Conjugation greatly enhanced LLL12 solubility (up to 10 mg/mL) and reduced cytotoxicity without altering dendrimer size or surface charge. All three G6-LLL12 conjugates remained stable under neutral pH but exhibited sustained, pH-dependent drug release correlating with in vitro potency and cytotoxicity. Notably, hydrazone-linked conjugate showed an IC₅₀ = 0.42 ± 0.035 μg/mL, comparable to free LLL12 (IC₅₀ = 0.31 ± 0.05 μg/mL) and superior to amide- and carbamate-linked conjugates. In bone marrow-derived immune suppressive myeloid cells, hydrazone-based G6-LLL12 effectively reduciii-derrive, hydrazone-based G6-LLL12 effectively reduced monocytic myeloid-derived suppressor cell expansion and promoted antigen-presenting cell maturation, highlighting a promising pH-responsive delivery system that enhances solubility and safety while retaining potency.

Graphical Abstract

INTRODUCTION

Signal transducer and activator of transcription 3 (STAT3) is a transcription factor and a key regulator of immune suppression and tumor progression. Constitutive activation of STAT3 is observed in over 70% of human cancers, making it an attractive target for cancer therapy.¹ Despite decades of research, no FDA-approved small-molecule STAT3 inhibitor is currently available. As a transcription factor, STAT3 has been difficult to target due to the absence of well-defined binding pockets suitable for small-molecule binding.² Consequently, most Janus kinase (JAK)/STAT pathway inhibitors in clinical development act upstream (e.g., the JAK inhibitor WP1066), but these compounds affect multiple signaling cascades, limiting their specificity.^2,3 Although potent direct inhibitors of STAT3 are available, many of them face major translational hurdles due to poor pharmacological properties.⁴ For example, as a novel small molecule-based direct STAT3 inhibitor, LLL12 directly binds to STAT3 SH₂ domain at the pY705 binding site of STAT3 protein, preventing subsequent phosphorylation/dimerization (pSTAT3) and activation.^5–7 LLL12 has demonstrated potent inhibition of pSTAT3 activity in various cancer cell lines, including breast, pancreatic, osteosarcoma, medulloblastoma, and glioblastoma.^5–9 However, like many other STAT3 inhibitors, LLL12 has poor water solubility, making it difficult to formulate for intravenous administration.¹⁰ The low oral bioavailability and fast systemic clearance further restrict its in vivo efficacy. Additionally, STAT3-specific toxicity, which arises from disruption of homeostatic roles of STAT3 in normal tissues, contributes to dose-limiting toxicities.^2,4,11–13 Collectively, these limitations highlight the critical need for improved delivery strategies to fully realize the therapeutic potential of STAT3 inhibitors.²

To address the translational challenges associated with STAT3 inhibitor LLL12, we formulated LLL12 drug conjugates with hydroxyl-terminated PAMAM (poly(amidoamine)) dendrimers.Dendrimers are a class of synthetic, protein-like, ultrasmall macromolecules (sub-10 nm) that have been explored as drug delivery vehicles in preclinical studies and clinical trials (NCT05395624, NCT04458298, NCT05387837, NCT03500627, NCT04321980, NCT03255343, NCT05105607, NCT05205161).^14–19 Our latest study shows that Generation six (G6) hydroxyl-terminated PAMAM dendrimer can selectively target myeloid cells within the tumor and lymphoid organs upon systemic administration.¹⁸ This in vivo selectivity is attributed to their ultrasmall size (<10 nm) and neutral surface, which allows efficient tissue penetration and access to phagocytic cells within these tissues. Inspired by these properties, hydroxyl-terminated PAMAM dendrimers have been utilized for conjugating multiple small molecule drugs and small interfering RNA (siRNA), employing linker chemistries such as ester,^20–26 disulfide,^27–35 and amide.³⁶ The choice of linker plays a critical role in determining the release mechanism and pharmacological profile of the bioconjugates.^37–41 For example, the hydrazone linker enables pH-sensitive drug release in acidic tumor microenvironment, thereby improving therapeutic efficacy, while the carbamate linker tends to limit drug release.³⁷ Disulfide linkers facilitate drug release under intracellular, glutathione-rich conditions while remaining stable in extracellular plasma.^38,39 Amide linkers offer stability, whereas ester linkers promote pH- and esterase-dependent release.⁴⁰ Overall, the linker design should balance drug potency and stability. Labile linkers enable rapid drug release and high potency but may also cause premature drug release in plasma, limiting the efficiency of targeted drug delivery. In contrast, stable linkers enhance systemic stability but can reduce drug potency due to limited release. Therefore, rational linker design is crucial for balancing stability, release kinetics, and the overall bioactivity of dendrimer-based therapeutics.

In this study, we investigated how release kinetics affect the potency and cytotoxicity of dendrimer-LLL12 conjugates in an in vitro setting. To do this, we synthesized three different dendrimer-LLL12 conjugates featuring carbamate, amide, and hydrazone linkers. All formulations enhanced the aqueous solubility of free LLL12. These conjugates remained stable under physiological conditions while showing controlled drug release at different rates under acidic pH, with hydrazone-based conjugates showing the highest release rate, followed by amide- and carbamate-based conjugates. The release rates of dendrimer-LLL12 conjugates correlated with their potency and cytotoxicity. Notably, the hydrazone-linked conjugates showed comparable activity to free LLL12 in reducing monocytic myeloid-derived suppressor cell (M-MDSC) populations and promoting antigen-presenting cell (APC) maturation ex vivo.

EXPERIMENTAL SECTION

Materials and Methods for the Synthesis of Intermediates and G6-LLL12 Conjugates.

Materials and Reagents.

G6-OH PAMAM dendrimer in methanol was purchased from Dendritech, Inc. (Midland, Michigan). The azido-PEG4−4-nitrphenyl carbonate and azidoacetic acid N-hydroxy succinimide ester were purchased from BroadPharm. Dimethyl sulfoxide (DMSO), tetrahydrofuran (THF), dichloromethane, trifluoroacetic acid (TFA), benzotriazol-1-yloxytri-pyrrolidinophosphonium hexafluorophosphate (PyBOP), and piperidine were purchased from Merck. Hydrazine monohydrate, N,N-diisopropylethylamine, ethyl acetate, and dicyclohexyl carbodiimide (DCC) were purchased from Sigma-Aldrich. Acetic acid, hexane, and acetonitrile were purchased from Fischer Chemicals. Sodium ascorbate, 4-(dimethylamino)pyridine (DMAP), and copper sulfate pentahydrate (CuSO₄.5H₂O) were purchased from Thermo Scientific. Dimethylformamide (DMF) was purchased from Thermo Fischer Scientific. 6-Heptynoic acid was purchased from Ambeed. Ethanol was purchased from Decon Laboratories Inc. Deuterated dimethyl sulfoxide (DMSO-d₆), methanol-d₄, deuterated chloroform (CDCl₃), and acetone-d6 were purchased from Merck. Dialysis membrane cutoff = 12−14 kDa was purchased from Spectrum Laboratories. 0.5 M EDTA was purchased from Gibco. All chemicals were used as they arrived.

Characterization.

Nuclear Magnetic Resonance (NMR) Spectroscopy.

¹H NMR and ¹³C spectroscopies were used to evaluate the structures of intermediates and the final dendrimer-LLL12 conjugates. The NMR spectra were recorded on a Bruker 400 and 600 MHz spectrometer with TMS as the internal standard at 25 °C. Chemical shifts are reported in ppm. The peaks of the residual protic solvent such as CDCl₃ (¹H, δ 7.20 ppm; ¹³C, δ 77.0 ppm), CD₃OD (¹H, δ 3.3 ppm; 13C, δ 49.0 ppm), acetone-d₆ (¹H, δ 2.06 ppm; ¹³C, δ 29.0 and 205.3 ppm), and DMSO-d₆ (¹H, δ 2.50 ppm; ¹³C, δ 39.55 ppm) were used for chemical shift calibration. The coupling constants (J values) are given in Hz. Electron spray ionization mass spectrometry (ESI-MS, Waters Xevo TQ-S micro) operated in positive ion mode to confirm the desired product.

Synthetic Protocols for Intermediates and G6-LLL12 Conjugates (1−20).

LLL12 1 was synthesized by employing a previously published procedure and confirmed by NMR spectroscopy (Figures S1 and S2, Supporting Information (SI)).⁷ The remaining compounds were synthesized according to the detailed experimental procedures described below, with reactions conducted either under a nitrogen atmosphere or in ambient air, as appropriate. Thin-layer chromatography (TLC) was performed on silica gel GF254 plates (Miles Scientific), and the spots were visualized with UV light. The ¹H and ¹³C NMR spectra for all intermediates are provided in the Supporting Information.

Synthesis of G6-LLL12C Conjugate (7).

Synthesis of Azido-Terminated PEG4-LLL12 (3).

LLL12 1 (0.2 mmol, 1.10 eq, 60 mg) and azido-PEG4−4-nitrophenyl carbonate 2 (0.178 mmol, 1.0 eq, 69 mg) were dissolved in DMF (2 mL), and DMAP (0.356 mmol, 2.0 eq, 43.5 mg) was added to the solution at room temperature. The resulting reaction mixture was stirred for 40 h and diluted with ethyl acetate (15 mL), washed with brine (10 mL), dried over Na₂SO_4, and concentrated in vacuo. The crude was purified by silica gel column chromatography (by gradient elution of dichloromethane and methanol, 95/05 to 85/15) to afford compound 3. Yellow solid; yield 54 mg, 55%; R_f = 0.2 (10% methanol in dichloromethane); ¹H NMR (400 MHz, methanol-d₄) δ 8.62 (d, J = 8.0 Hz, 1H), 8.56 (d, J = 8.0 Hz, 1H), 7.96 (t, J = 8.0 Hz, 1H), 7.66 (overlapped t, J = 8.0 Hz, 1H), 7.65 (overlapped d, J = 8.0 Hz, 1H), 7.21 (d, J = 8.0 Hz, 1H), 4.08 (dt, J = 12.0, 4.0 Hz, 2H), 3.40−3.52 (unresolved m, 12H), 3.22 (dt, J = 12.0, 4.0 Hz, 2H), exchangeable NH and OH protons are not appearing due to deuterium exchange; ¹³C NMR (100 MHz, methanol-d₄) 186.7, 181.0, 161.7, 139.5, 138.5, 137.1, 135.5, 134.2, 133.3, 132.5, 131.8, 123.6, 119.5, 115.0, 70.2 (x 2), 70.1, 70.0, 69.7, 68.4, 65.2, 50.3. ESI-MS m/z calculated for C₂₃H₂₄N₄O₁₀SNa [M + Na]⁺: 571.11; found 571.26.

Synthesis of Heptyne-Derived G6 Bifunctional Dendrimer (6).

To the solution of 6-heptynoic acid 5 (0.9176 mmol, 50 eq, 116 mg) in DMF (3 mL) was added the solution of PyBOP (1.375 mmol, 75 eq, 715 mg) and DIEA (1.832 mmol, 100 eq, 237 mg, 320 uL) in DMF (4 mL) at 0 °C. After 30 min, PAMAM G6 hydroxyl dendrimer 4 (0.0183 mmol, 1.0 eq, 1070 mg) in DMF (4 mL) was added to the reaction mixture. The reaction mixture was brought to room temperature and stirred for 48 h. Then the solvent was evaporated to dryness under high vacuum. The crude was redissolved in DMF and dialyzed against DMF for 12 h by changing the solvent at least three times. The collected solvent was evaporated under high vacuum to dryness, and the resulting product was dissolved in water (2 mL), further subjected to DI-water dialysis for 12 h by changing the solvent four times. The collected water was directly lyophilized to get heptyne-derived dendrimer 6. White solid; yield 1.05 g, 95%; ¹H NMR (600 MHz, DMSO-d₆) δ 8.06 (s, modified internal amide NH protons after linker attachment), 7.80−7.94 (m, internal amide NH protons of G6-OH), 4.73 (bs, unmodified OH protons of G6-OH), 4.01 (t, J = 6.0 Hz, shifted CH₂ protons of G6-OH at the site of linker attachment), 3.40−3.42 (m, CH₂ protons of G6-OH), 3.28 (t, J = 4.0 Hz, shifted CH₂ protons of G6-OH after linker attachment), 3.12−3.13 (m, CH₂ protons of G6-OH), 2.67 (s, CH₂ protons of G6-OH), 2.46 (s, CH protons of G6-OH), 2.31 (t, J = 6.0 Hz, CH₂ protons of linker), 2.22 (s, CH₂ protons of G6-OH), 2.16 (t, J = 6.0 Hz, CH₂ protons of linker), 1.60 (dt, J = 12.0, 6.0 Hz, CH₂ protons of linker), 1.45 (dt, J = 12.0, 6.0 Hz, CH₂ protons of linker).

Synthesis of G6-LLL12C Conjugate (7).

In a 20 mL microwave vial, heptyne-derived dendrimer 6 (0.00248 mmol, 1.0 eq, 150 mg) was dissolved in water (5 mL), CuSO₄ (0.060 mmol, 15 mg) in water (1.25 mL), and sodium ascorbate (0.060 mmol, 12 mg) in water (1.25 mL) were added successively. To this solution, azido-terminated PEG4-LLL12 3 (0.055 mmol, 22 eq, 30 mg) in THF (7.5 mL) and DMF (150 μL) were added at room temperature. The resulting reaction mixture was irradiated in a microwave (Biotage) at 45 °C for 8.5 h. Upon completion (monitored by TLC), the reaction mixture was evaporated in vacuo and then diluted with DMF (2 mL). The resulting mixture was dialyzed against DMF for 12 h by changing the solvent after every 4 h. The collected solvent was evaporated to dryness, and the residual product was dissolved in DI-water (2 mL), treated with EDTA (1 mL, 0.5M, pH = 8.0). The mixture was stirred for 30 min at room temperature and then subjected to dialysis against water for 12 h by changing the water every 3 h. The collected water was lyophilized, obtained product was further purified by column chromatography (Sephadex G-25) using DI water to get pure G6-LLL12C conjugate 7. Orange solid; yield 160 mg, 93%; ¹H NMR (600 MHz, DMSO-d₆) δ 8.49 (d, J = 9.0 Hz, drug aromatic CH protons), 8.34 (d, J = 9.0 Hz, drug aromatic CH protons), 8.06 (s, modified amide NH protons), 7.81−7.94 (merged m, internal amide NH protons of G6-OH), 7.88−7.91 (merged t, J = 9.0 Hz, drug aromatic CH protons), 7.58 (d, J = 9.0 Hz, drug aromatic CH protons), 7.32 (d, J = 9.0 Hz, drug aromatic CH protons), 4.75 (bs, unmodified OH protons of G6-OH), 4.45 (s, PEG linker CH₂ protons), 3.99 (s, shifted OCH₂ protons of G6-OH at the site of linker attachment), 3.78 (d, J = 6.0 Hz, PEG linker terminal OCH₂ protons), 3.44−3.49 (PEG CH₂ protons) 3.40 (CH₂ protons of G6-OH) 3.12 (s, CH₂ protons of G6-OH), 2.59−2.70 (m, CH₂ protons of G6-OH), 2.39−2.44 (bs, CH protons of G6-OH), 2.30 (t, J = 6.0 Hz, dendrimer attached heptyne linker CH₂ protons), 2.21 (s, CH₂ protons of G6-OH), 1.57 (m, CH₂ protons of heptyne linker).

Synthesis of G6-LLL12S Conjugate (16).

Synthesis of tert-Butyl Ester-Terminated LLL12 (9).

To a solution of 5-(tert-butoxy)-5-oxopentanoic acid 8 (0.275 mmol, 1.1 eq, 52 mg) in dichloromethane (3 mL) was added N,N’-dicyclohexylcarbodiimide (0.375 mmol, 1.5 eq, 77 mg) and 4-dimethylaminopyridine (0.05 mmol, 0.2 eq, 6 mg) at 0 °C. To the resulting solution, LLL12 1 (0.25 mmol, 1.0 eq, 76 mg) was added and continued stirring for 10 min at the same temperature. Then the reaction mixture was brought to room temperature and continued stirring until completion. After completion of the reaction (monitored by TLC), the precipitated dicyclohexylurea (DCU) was filtered off and washed with 5% aqueous acetic acid (5 mL) and water (3 × 5 mL), dried over Na₂SO_4, and concentrated under reduced pressure. The resulting crude was purified by silica gel column chromatography (by gradient-elution of EtOAc and hexane, 10/90−30/70) to give 9. Yellow solid; yield 78 mg, 65%; R_f = 0.5 (40% EtOAc in hexane); ¹H NMR (400 MHz, DMSO-d₆) δ 12.10 (s, 1H), 9.43 (s, 1H), 8.72 (d, J = 8.0 Hz, 1H), 8.60 (d, J = 8.0 Hz, 1H), 7.93 (t, J = 8.0 Hz, 1H), 7.69 (d, J = 8.0 Hz, 1H), 7.64 (t, J = 8.0 Hz, 1H), 7.27 (d, J = 8.0 Hz, 1H), 2.38 (t, J = 8.0 Hz, 2H), 2.14 (t, J = 8.0 Hz, 2H), 1.78 (quint, J = 8.0 Hz, 2H), 1.31 (s, 9H); ¹³C NMR (100 MHz, DMSO-d₆) 185.4, 181.2, 171.1, 169.7, 161.2, 138.3, 137.5, 136.5, 134.5, 132.9, 131.5, 123.7, 119.4, 113.9, 79.6, 34.6, 33.0, 28.7, 27.0, 18.5.

Synthesis of Acid-Terminated LLL12 (10).

To compound 9 (0.165 mmol, 1.0 eq, 78 mg) in dichloromethane (3 mL) was added trifluoroacetic acid (4.125 mmol, 25.0 eq, 470 mg) at room temperature. The resulting reaction mixture was stirred until completion. After completion (monitored by TLC), the reaction mixture was diluted with dichloromethane (15 mL) and washed with water (3 × 5 mL), dried over Na₂SO_4, and concentrated under reduced pressure. The resulting crude was purified by triturating with hexane to offer 10. Yellow solid; yield 66 mg, 96%; R_f = 0.3 (10% hexane in EtOAc); ¹H NMR (400 MHz, Acetone-d₆) δ 12.05 (s, 1H), 10.65 (s, 1H), 8.64 (d, J = 8.0 Hz, 1H), 8.57 (d, J = 8.0 Hz, 1H), 8.04 (t, J = 8.0 Hz, 1H), 7.72 (t, J = 8.0 Hz, 1H), 7.58 (d, J = 8.0 Hz, 1H), 7.25 (d, J = 8.0 Hz, 1H), 2.43 (t, J = 8.0 Hz, 2H), 2.13 (t, J = 8.0 Hz, 2H), 1.67 (quint, J = 8.0 Hz, 2H); ¹³C NMR (100 MHz, Acetone-d₆) 187.2, 181.4, 173.4, 173.1, 162.0, 139.9, 138.9, 137.6, 135.6, 134.5, 133.7, 132.7, 131.9, 123.8, 119.5, 115.3, 34.9, 32.1, 19.5.

Synthesis of LLL12-NHS Ester (12).

To compound 10 (0.158 mmol, 1.0 eq, 66 mg) in dry THF (3 mL) was added N,N’-dicyclohexylcarbodiimide (0.174 mmol, 1.1 eq, 36 mg) and N-hydroxy succinimide 11 (0.174 mmol, 1.1 eq, 20 mg) at 0 °C. The ice bath was removed after 2 h, and the reaction was continuously stirred for 24 h. Upon completion of the reaction (monitored by TLC), the precipitated white solid DCU was filtered off, and the solvent was concentrated under reduced pressure. The resulting residue was purified by silica gel column chromatography (by gradient elution of methanol and DCM, 05/95−10/90) to give LLL12-NHS ester 12. Yellow solid; yield 63 mg, 78%; R_f = 0.5 (5% methanol in dichloromethane); ¹H NMR (400 MHz, CDCl₃) δ 12.13 (s, 1H), 9.39 (s, 1H), 8.74 (d, J = 8.0 Hz, 1H), 8.63 (d, J = 8.0 Hz, 1H), 7.95 (t, J = 8.0 Hz, 1H), 7.71 (d, J = 8.0 Hz, 1H), 7.66 (t, J = 8.0 Hz, 1H), 7.29 (d, J = 8.0 Hz, 1H), 2.73 (s, 4H), 2.55 (t, J = 8.0 Hz, 2H), 2.50 (t, J = 8.0 Hz, 2H), 1.95 (quint, J = 8.0 Hz, 2H). ¹³C NMR (150 MHz, CDCl₃) 185.4, 181.3, 169.1, 168.0, 166.9, 161.3, 138.2, 137.4, 136.4, 134.6, 133.0, 131.6, 131.5, 123.8, 119.4, 114.0, 33.5, 28.6, 24.5, 18.3. ESI-MS m/z calculated for C₂₃H₁₈N₂O₁₀SNa [M + Na]⁺: 537.06; found 537.08.

Synthesis of Amine-Derived G6 Bifunctional Dendrimer (15).⁴²

Step 1: To a solution of Fmoc-GABA−OH 13 (0.144 mmol, 60 eq, 47 mg) in DMF (3 mL) in a 50 mL round-bottom flask under nitrogen atmosphere was added PyBOP (0.216 mmol, 90 eq, 112 mg) in DMF (3 mL) and DIEA (0.288 mmol, 120 eq, 37 mg, 50 uL). The resulting mixture was allowed to stir for an hour in an ice bath. Then, PAMAM G6-OH 4 (0.00240 mmol, 1.0 eq, 140 mg) was dissolved in DMF (10 mL) and added to the reaction mixture. The reaction mixture was allowed to reach room temperature and stirred for 48 h. The solvent was then evaporated at 40 °C under reduced pressure. The crude was redissolved in DMF (2 mL) and dialyzed against DMF for 12 h by changing the DMF every 4 h. The collected solvent was evaporated and subjected to high vacuum overnight to offer pure off-white semisolid Fmoc functionalized bifunctional dendrimer 14. Yield 142 mg, 89%; ¹H NMR (400 MHz, DMSO-d₆) δ 8.05 (s, internal amide NH protons of G6-OH after linker attachment), 7.85−7.97 (m, internal amide NH protons of G6-OH), 7.80 (bs, Fmoc aromatic protons), 7.67 (d, J = 9.0 Hz, Fmoc aromatic protons), 7.39 (t, J = 9.0 Hz, Fmoc aromatic protons), 7.31 (t, J = 9.0 Hz, Fmoc aromatic protons), 4.74 (bs, OH protons of G6-OH), 4.29−4.31 (m, Fmoc OCH₂ protons), 4.19−4.21 (m, Fmoc benzylic CH₂ protons), 4.01 (t, J = 9.0 Hz, OCH₂ protons of G6-OH), 3.39−3.42 (m, CH₂ protons of G6-OH), 3.03−3.13 (m, CH₂ protons of G6-OH), 3.00−3.02 (m, N−CH₂ protons of linker), 2.67 (bs, CH₂ protons of G6-OH), 2.45 (bs, CH₂ protons of G6-OH), 2.29 (overlapped s, CH₂ protons of linker), 2.22 (overlapped s, CH₂ protons of G6-OH), 1.65 (t, J = 9.0 Hz, CH₂ protons of linker).

Step 2: The whole batch of the above Fmoc-functionalized bifunctional dendrimer 14 was dissolved in DMF (5 mL), and 5 mL of piperidine: DMF (1:4) was added under a nitrogen atmosphere. The reaction mixture was stirred for 30 min at 0 °C, and then the solvents were evaporated under vacuum. The crude product was coevaporated with DMF 5 mL under high vacuum and subjected to dialysis for 18 h by changing the DMF after every 6 h. The collected solvent was evaporated and dialyzed against DI water for 4 h by changing the solvent every 2 h. The collected water was lyophilized to get bifunctional dendrimer 15. White solid; Yield 130 mg, 90%; ¹H NMR (400 MHz, DMSO-d₆) δ 8.14 (s, internal amide NH protons of G6-OH after linker attachment), 7.86−7.97 (m, internal amide NH protons of G6-OH), 4.64 (bs, OH protons of G6-OH), 4.02 (t, J = 4.0 Hz, OCH₂ protons of G6OH), 3.38−3.51 (m, CH₂ protons of G6-OH), 3.11−3.12 (m, CH₂ protons of G6-OH), 2.65 (s, CH₂ protons of G6-OH), 2.43 (s, CH₂ protons of G6-OH), 2.21 (s, CH₂ protons of G6-OH), 1.79 (t, J = 6.0 Hz, linker CH₂ protons).

Synthesis of G6-LLL12S Conjugate (16).

To the bifunctional dendrimer 15 (0.002187 mmol, 1.0 eq, 130 mg) and DIEA (0.1640 mmol, 75.0 eq, 28.5 uL) in DMSO (4 mL) was added LLL12-NHS ester 12 (0.0546 mmol, 25.0 eq, 28.5 mg) in DMSO (1 mL). The resulting reaction mixture was stirred for 12 h at room temperature. After completion (monitored by TLC), the reaction mixture was subjected to dialysis against DMF for 18 h by changing the solvent after every 6 h. The obtained solution was evaporated to dryness under reduced pressure at room temperature, and the final product was dissolved in water and subjected to further dialysis against DI-water for 6 h by changing the water after every 2 h. The solution obtained was lyophilized to obtain the G6-LLL12S conjugates. Yellow solid; yield 133 mg, 94%; ¹H NMR (400 MHz, DMSO-d₆) δ 8.54 (bs, LLL12 aromatic CH proton), 8.43 (bs, LLL12 aromatic CH proton), 8.16−7.98 (m, internal amide NH protons of G6-OH), 7.80 (s, LLL12 aromatic CH proton), 7.64 (s, LLL12 aromatic CH proton), 7.33 (s, LLL12 aromatic CH proton), 4.09 (linker CH₂ protons), 3.51 (bs, CH₂ protons of G6-OH), 3.36 (s, linker CH₂ protons), 3.22 (bs, CH₂ protons of G6-OH), 2.76 (bs, CH₂ protons of G6-OH), 2.57 (s, CH₂ protons of G6-OH), 2.34 (s, CH₂ protons of G6-OH), 2.11 (s, linker CH₂ protons), 1.70 (s, linker CH₂ protons).

Synthesis of G6-LLL12H Conjugate (20).

Synthesis of LLL12-Hydrazone (17).

To LLL12 1 (0.6596 mmol, 1.0 eq, 200 mg) in ethanol (200 mL) was added acetic acid (1 drop), followed by hydrazine hydrate (6.596 mmol, 10.0 eq, 320 μL) at room temperature. The resulting reaction mixture was stirred at room temperature for 30 min and then heated to 84 °C (on an oil bath), continued stirring until completion. After completion (monitored by TLC), the reaction mixture was brought to room temperature and stored at 4 °C overnight. The precipitated LLL12-hydrazone 17 was filtered off and washed with cold ethanol (2 × 2 mL) and dried in vacuo. Green solid; yield 166 mg, 79%; R_f = 0.5 (25% hexane in EtOAc); ¹H NMR (600 MHz, DMSO-d₆) δ 12.73 (s, 1H), 8.50 (s, 2H), 8.39 (t, J = 9.0 Hz, 2H), 7.78 (t, J = 9.0 Hz, 1H), 7.68 (t, J = 9.0 Hz, 2H), 7.24 (s, 2H), 7.10 (d, J = 9.0 Hz, 1H); 13C NMR (150 MHz, DMSO-d₆) 187.4, 162.6, 140.3, 138.6, 137.4, 135.2, 131.4, 131.0, 130.4, 128.7, 127.1, 118.3, 117.4, 115.1.

Synthesis of Azidoacetyl-LLL12-hydrazone (19).

To LLL12-hydrazone 17 (0.4415 mmol, 1.0 eq, 140 mg) in DMSO (3 mL) was added azido acetic acid NHS ester (0.4415 mmol, 1.0 eq, 87.5 mg) and DIEA (1.3247 mmol, 3.0 eq, 231 μL) at room temperature. The resulting reaction mixture was stirred at room temperature until completion. After completion (monitored by TLC), the reaction mixture was diluted with DMF (9 mL) and evaporated under high vacuum. The resulting crude residue was purified by silica gel column chromatography (by gradient-elution of EtOAc and hexane, 40/60− 80/20) to afford 19. Orange solid; yield 148 mg, 84%; R_f = 0.5 (15% hexane in EtOAc); ¹H NMR (600 MHz, DMSO-d₆) δ 11.94 (s, 1H), 11.62 (s, 1H), 8.38 (t, J = 9.0 Hz, 2H), 8.02 (s, 2H), 7.84 (t, J = 9.0 Hz, 1H), 7.80 (t, J = 9.0 Hz, 1H), 7.67 (d, J = 9.0 Hz, 1H), 7.27 (t, J = 9.0 Hz, 1H), 4.22 (s, 2H); ¹³C NMR (150 MHz, DMSO-d₆) 187.0, 169.1, 161.7, 142.7, 141.7, 137.5, 137.0, 135.0, 133.4, 130.3, 130.2, 120.7, 120.1, 115.8, 50.3. ESI-MS m/z calculated for C₁₆H₁₂N₆O₅SNa [M + Na]⁺: 423.05; found 423.06.

Synthesis of G6-LLL12H Conjugate (20).

Bifunctional dendrimer 6 (0.01232 mmol, 1.0 eq, 740 mg) was dissolved in water (3 mL), CuSO₄ (0.4163 mmol, 104 mg) in water (1.5 mL) and sodium ascorbate (0.4163 mmol, 82.5 mg) in water (1.5 mL) were added successively. To this solution, azido-terminated LLL12-hydrazone 19 (0.3203 mmol, 26 eq, 128 mg) in THF (1.5 mL) and DMF (6 mL) were added at room temperature. The resulting mixture was irradiated in a microwave (Biotage) at 45 °C for 15 h. After completion (monitored by TLC), the reaction mixture was evaporated in vacuo and then diluted with DMF (4 mL). The resulting mixture was dialyzed against DMF for 24 h by changing the solvent after every 6 h. The collected solvent evaporated to dryness, and the residual product was dissolved in DI-water (2 mL), treated with EDTA (2 mL, 0.5M, pH = 8.0). The mixture was stirred for 30 min at room temperature and then subjected to dialysis against water for 12 h by changing the water every 3 h. The collected water was lyophilized, obtained product was further purified by column chromatography (Sephadex G-25) using DI-water to get pure G6-LLL12H conjugate. Dark brown solid; yield 776 mg, 95%; ¹H NMR (600 MHz, DMSO-d₆) δ 13.16 (s, LLL12 OH proton), 8.78 (s, sulfonamide NH₂ protons), 8.40 (d, J = 9.0 Hz, drug aromatic CH protons), 8.06 (s, modified amide NH protons of G6-OH after drug conjugation), 7.79−7.94 (m, internal amide NH protons of G6-OH), 7.69 (s, hydrazone linker NH protons), 7.62 (t, J = 9.0 Hz, drug aromatic CH protons), 7.57 (t, J = 9.0 Hz, drug aromatic CH protons), 6.95 (d, J = 9.0 Hz, drug aromatic CH protons), 5.14 (s, COCH₂ protons of hydrazone), 4.72 (bs, OH protons of G6-OH), 4.01 (s, shifted OCH₂ protons of G6-OH at the site of linker attachment), 3.40−3.52 (internal CH₂ protons of G6-OH), 3.11−3.23 (internal CH₂ protons of G6-OH) 2.65 (internal CH₂ protons of G6-OH) 2.43 (internal CH₂ protons of G6-OH), 2.33−2.36 (m, CH₂ protons of propargyl linker), 2.21 (internal CH₂ protons of G6-OH), 1.60−1.61 (m, CH₂ protons of propargyl linker).

Size and Zeta (ζ) Potential.

The particle size and ζ-potential of G6-OH and conjugates (G6-LLL12C, G6-LLL12S, and G6-LLL12H) were analyzed using Litesizer 500 (Anton Paar Instruments) at 25 °C. To study physical properties, 10 mM NaCl was first filtered using a 0.2 μm, 13 mm polyether sulfone (PES) syringe filter (Cytiva). For hydrodynamic radius measurements via dynamic light scattering (DLS), G6-OH and conjugates were dissolved in the prefiltered 10 mM NaCl solution at a concentration of 1 mg/mL. The size measurements were conducted using a 1 mL cuvette (Sarstedt). For ζ-potential measurements, the dendrimers and conjugates were prepared similarly and measured in an omega cuvette (Anton Paar).

High Performance Liquid Chromatography (HPLC).

The conjugates were analyzed by HPLC (Vanquish System, Thermo Fischer Scientific) equipped with a binary pump, dual UV detector, and autosampler interfaced with Chromeleon software. The HPLC chromatogram was monitored at 210 and 250 nm simultaneously using a dual UV absorbance detector. The water/acetonitrile was freshly prepared, filtered, degassed, and used as a mobile phase. Symmetry C8 and C18 reverse phase columns with 5 μm particle size, 25 cm length, and 4.6 mm internal diameter were used. A gradient flow was used with the initial condition 90:10 to 5:95 (H₂O/ACN) in 8 min and returning to 90:10 (H₂O/ACN) in 25 min with a flow rate of 1 mL/min.

Drug Release Study.

The release of LLL12 from G6-LLL12C, G6-LLL12S, and G6-LLL12H was evaluated at 37 °C in physiological (pH 7.4, phosphate buffer and nonheat inactivated Fetal Bovine Serum) and acidic (pH 4.5, citrate buffer) conditions, using 10 mg of G6-LLL12C, 8 mg of G6-LLL12S, and 10 mg of G6-LLL12H in 1.25 mL of each buffer under continuous mixing, individually. At predetermined time intervals, 50 μL aliquots were withdrawn from the incubation mixture and directly analyzed by HPLC. Acetonitrile and water were used as mobile phase, and the release of LLL12 was monitored at 250 nm and dendrimer at 210 nm. The percentage of LLL12 released from conjugates was quantified by a calibration curve. Additionally, LLL12 release from G6-LLL12H was evaluated under deionized water and saline conditions.

Materials for In Vitro Studies.

RPMI 1640 media, 1 mM HEPES, GlutaMax, and 2-mercaptoethanol were purchased from Gibco. Heat-inactivated Fetal Bovine Serum (HI-FBS), Dulbecco modified Eagle medium (DMEM), and cell scrapers were purchased from Thermo-fisher. Penicillin/streptomycin (P/S) and 100 mM sodium pyruvate (Corning), puromycin (Santa Cruz Biotechnology, Dallas, TX, USA, sc-108071A), STAT3 luciferase reporter lentivirus (BPS Biosciences, San Diego, CA), Polybrene (Santa Cruz Biotechnology; no. sc-134220), IL-6 (Acro Biosystems, Newark, USA), GM-CSF (R&D 415-ML), D-Luciferin (PerkinElmer, Waltham, USA), and CellTiter-Blue Reagent from Promega were purchased.

Cell Culture.

The THP-1 cell line was gifted by Dr. Mei He at the University of Florida College of Pharmacy. THP-1 cells were grown in RPMI 1640, 10% HI-FBS, 1% (1X P/S), 25 mM HEPES, 0.05 mM 2-mercaptoethanol in a humidified incubator at 37 °C. The THP-1_STAT3‑Luc cells were cultured in RPMI 1640 with 10% HI-FBS, 1% (1X P/S), 25 mM HEPES, 1.0 mM sodium pyruvate, 1.0 μg/mL puromycin, and 0.05 mM 2-mercaptoethanol in an incubator at 37 °C (see SI, Figure S27). The KR158 glioma cells were cultured in Dulbecco modified Eagle medium (DMEM), supplemented with 1% (1X P/S) and 10% heat-inactivated fetal bovine albumin (HI-FBA) in an incubator at 37 °C.

Establishment of THP-1_STAT3‑Luc Reporter Cell Line.

To establish a stable THP-1_STAT3‑Luc reporter cell line, THP-1 cells were seeded at a density of 5 × 10³ to 10 × 10³ cells per well in a white opaque 96-well microplate. The THP-1 cells were transduced by STAT3 Luciferase Reporter Lentivirus (BPS Bioscience, Catalog #79744) following the manufacturer’s protocol. After transduction, cells were exposed to puromycin in culture media to ensure a homogeneous population of cells with stable STAT3 luciferase expression. To validate that the reporter signal responds to STAT3 activation, THP-1_STAT3‑Luc cells were stimulated with increasing concentrations of IL-6 to activate pSTAT3. After 5−6 h of incubation, STAT3 activity within the THP-1_STAT3‑Luc cells was evaluated by adding D-luciferin to each well and measuring luminescence using a luminometer.

pSTAT3 Inhibition IC₅₀ and Cytotoxicity Evaluation.

THP-1_STAT3‑Luc cells were seeded at 6 × 10⁴ cells per well in 100 μL RPMI 1640 medium (without phenol red), 10% FBS (Heat-inactivated), 25 mM HEPES, 100 U/mL penicillin, 100 μg/mL streptomycin (1 x P/S). Cells were treated with LLL12, G6-LLL12C, G6-LLL12S, and G6-LLL12H at concentrations of 300, 100, 33.3, 11.1, 3.7, 1.23, 0.4111, 0.137, and 0.0456 μg/mL, each in triplicate, and incubated for 72 h at 37 °C.

For efficacy studies, after 72 h of treatment, 10 ng of IL-6 was added to each well, and the plate was incubated for another 5 h to induce STAT3 activation. Then, 15 μL of D-luciferin (150 μg/mL) was added to each well, and the plate was returned to the incubator for 15 min before measuring luminescence using a (SpectraMax iD3) plate reader. The experimental setup included three control conditions: (1) cells alone (background control), (2) cells treated with D-luciferin only (to assess endogenous pSTAT3 expression), and (3) cells treated with both IL-6 and D-luciferin (to represent maximum pSTAT3 activation in THP-1_STAT3‑Luc cells).

To assess cytotoxicity, 20 μL of CellTiter-Blue reagent was added to each treatment well after 72 h of incubation, followed by an additional 3-h incubation at 37 °C. Fluorescence was measured using a (SpectraMax iD3) plate reader at an excitation/emission of 560/590 nm. Control conditions included CellTiter-Blue reagent in cell-free wells (background), untreated cells (negative control), and cells treated with 1% Triton X-100 (positive control). Half maximal inhibitory concentrations (IC₅₀) and cell viability analyses were performed using GraphPad Prism software.

Functional Study on Bone Marrow-Derived Cells.

Animals.

Wildtype C57BL/6 mice were purchased from the Jackson Laboratory. All procedures involving animal housing and handling were conducted in accordance with the guidelines of the University of Florida Institutional Animal Care and Use Committee.

M-MDSC Induction and Culture.

The induction of M-MDSC was adapted from reported literature.⁴³ Bone marrow cells from wildtype C57BL/6 mice were isolated and plated at a density of 4 × 10⁵ cells/cm² and concentration of 1 × 10³ cells/μL in KR158 cell conditioned media, consisting of 50% KR158 conditioned media and 50% complete RPMI (RPMI 1640 + 10% FBS + 1% (1X P/S) + 1% GlutaMax). Additionally, the media was supplemented with 40 ng/mL GM-CSF and 40 ng/mL IL-6. Then the cells were treated with LLL12 and G6-LLL12H with a range of concentrations from 0.004 to 0.4 μg/mL. After 3 days of incubation at 37 °C (in a humidified atmosphere with 5% CO₂), suspended cells were collected, and adherent cells were scraped with a cell scraper and washed in PBS. Cells were collected by centrifugation at 500 × g for 5 min at 4 °C and counted using the Trypan blue exclusion method. Cells were then analyzed by flow cytometry.

Flow Cytometry.

Single-cell suspensions were centrifuged at 500 × g for 5 min at 4 °C and stained for markers of interest (Table S1, see SI) for 30 min at 4 °C. Cells were then washed twice with PBS and stained with a viability dye for 15 min at room temperature, protected from light. After viability staining, cells were resuspended in FACS buffer and kept on ice until flow cytometry analysis. Samples were analyzed in a single flow cytometry tube using a Sony Spectral Analyzer (SP6800). Raw data was quantified using FlowJo v10.10 (BD Biosciences). The flow cytometry gating strategy and antibody details can be found in Supporting Information (Figure S28, see SI).

RESULTS AND DISCUSSION

To achieve targeted intracellular drug delivery, an ideal formulation of dendrimer-drug conjugate should: (1) remain stable under physiological conditions, (2) enable sustained drug release in response to intracellular chemical cues (e.g., the intracellular acidic pH), (3) maintain the overall physiochemical properties of the dendrimer (e.g., small sizes, neutral surface charge, and high aqueous solubility). Guided by these principles, we designed G6-LLL12 conjugates using three acid-labile, hydrolyzable linkers based on carbamate, amide, and hydrazone bonds. Drug linkers based on these chemical bonds have been widely used in the design of polymer-drug conjugates and antibody-drug conjugates (ADCs).^44–47 Notably, hydrazone-based chemical linkers have been used in two clinically approved ADCs, i.e, Mylotarg and Besponsa.

Synthesis of Carbamate Linker-Based G6-LLL12 Conjugate (G6-LLL12C, 7).

To facilitate the effective conjugation of LLL12 to the surface of hydroxyl dendrimers, we first synthesized intermediate compounds azido-PEG4-LLL12 3 and heptyne-derived G6 bifunctional dendrimer 6. The LLL12 1 contains a heterofunctional reactive sulfonamide at C1 and a hydroxyl group at C5. Our recurring attempts to introduce linkers to the C5 hydroxyl group were unsuccessful. Consequently, we shifted our approach to introducing linkers at the C1 sulfonamide position. Preparation of azido-PEG4-LLL12 3 is illustrated in Figure 1Ai. Briefly, LLL12 1 upon reaction with azido-PEG4−4-nitrophenyl carbonate 2 in the presence of DMAP in DMF provided azido-PEG4-LLL12 3. Azido-PEG4 incorporation into LLL12 was evidenced by the characteristic peaks at 4.1−3.2 ppm in the ¹H NMR spectrum and 70−50 ppm in the ¹³C NMR spectrum (Figures S3−S4, see SI). The structure of the resulting drug−linker was further verified by mass spectrometry (Figure S5, see SI).

Figure 1.

Construction of G6 hydroxyl PAMAM dendrimer-LLL12 conjugate through a sulfonyl carbamate linkage (G6-LLL12C). (A) Schematic illustration showing the stepwise synthesis of G6-LLL12C: (i) Preparation of azido-terminated PEG3-LLL12 by the reaction of LLL12 with azido-terminated PEG3-nitrophenyl carbonate; (ii) Preparation of heptyne-derived G6 bifunctional dendrimer via PyBOP coupling; (iii) Synthesis of final conjugate (G6-LLL12C) via CuAAC click chemistry with 7.25% (w/w) LLL12 payload. (B) ¹H NMR of G6-LLL12C in DMSO-d₆ showing characteristic peaks of LLL12, linker, and dendrimer with color-coded assignments corresponding to the conjugate structure.

We then synthesized heptyne-derived G6 bifunctional dendrimer 6 by partly modifying the dendrimer surface hydroxyl groups with heptyne units. This modification aimed to reduce steric hindrance and facilitate the subsequent reaction with compound 3. As shown in Figure 1Aii, 6-heptynoic acid 5 was coupled to dendrimer 4 under PyBOP-mediated coupling conditions to yield compound 6. The structure was confirmed by ¹H NMR spectroscopy, where new peaks at 1.6 and 1.5 ppm correspond to methylene protons of heptyne units. The shift in OH attached methylene protons to 4.0 ppm and the shift in amide protons to 8.1 ppm indicate successful modification. Proton integration analysis confirmed the attachment of approximately 16.43 heptyne units per dendrimer (Figure S6, see SI). Given that our goal is to maintain the overall physicochemical properties of the hydroxyl dendrimers, we purposely kept a low ratio of heptyne units per dendrimer by only modifying 16.43 out of 256 surface hydroxyl functional groups with heptyne.

The final G6-LLL12C conjugate 7 was constructed via highly efficient copper(I)-catalyzed azide−alkyne cycloaddition (CuAAC) click chemistry between compound 3 and bifunctional dendrimer 6 using a microwave reactor (Biotage) at 45 °C (Figure 1Aiii). CuAAC, known for its exceptional efficiency, selectivity, and biocompatibility, has become a foundation reaction in nanomedicine and materials science.⁴⁸ Structure analysis through ¹H NMR revealed the presence of four doublets corresponding to aromatic protons of LLL12 at 8.50, 8.34, 7.58, and 7.33 ppm along with dendrimer peaks at 3.49−2.21 ppm, confirming the formation of conjugates. The LLL12 payload was quantified by comparing amidic protons of the dendrimer (7.94−7.79 ppm) with aromatic protons of LLL12 in the ¹H NMR spectrum. This analysis revealed an average of 16.27 LLL12 molecules were conjugated per dendrimer, corresponding to a weight percentage (w/w%) of 7.25% (Figure 1B). The purity of the resulting conjugate was confirmed by Reverse-Phase HPLC (RP-HPLC). A distinct new peak corresponding to G6-LLL12C appeared at 6.05 min in the HPLC chromatogram at 250 nm, clearly differentiating it from peaks of free LLL12 (8.2 min). Compared to the peak of unmodified G6-OH (5.09 min), the peak of G6-LLL12C shifted to 6.05 min and appeared to be broader, indicating G6-LLL12C is slightly more hydrophobic and poly disperse, potentially caused by the conjugation of LLL12 to the dendrimer surface (Figure S7, see SI).

Synthesis of Amide Linker-Based G6-LLL12 Conjugate (G6-LLL12S, 16).

To construct the G6-LLL12 conjugates with an amide-based linker, amidation was employed to conjugate LLL12 onto the dendrimer (Figure 2A). Initially, intermediate compounds LLL12 NHS ester and amine-derived G6 bifunctional dendrimer were synthesized. To generate LLL12 NHS ester, we modified the sulfonamide functional group on the LLL12 following the steps illustrated in Figure 2Ai. In brief, LLL12 1 was first reacted with 5-(tert-butoxy)-5-oxopentanoic acid 8 under DCC coupling conditions to give compound 9. Subsequent deprotection of the tert-butyl ester using excess trifluoroacetic acid resulted in the formation of acid compound 10 in excellent yield (96%). Next, acid compound 10 was activated with N-hydroxy succinimide 11 using DCC to produce LLL12 NHS ester 12 in 78% yield. The chemical structures of compounds 9, 10, and 12 were confirmed by ¹H and ¹³C NMR spectroscopy (Figure S8-15, see SI). The structural integrity of the desired drug−linker 12 was further verified by mass spectrometry (Figure S16, see SI)

Figure 2.

Construction of G6 hydroxyl PAMAM dendrimer-LLL12 conjugate through sulfonyl carbamoyl linkage (G6-LLL12S). (A) Stepwise synthesis of G6-LLL12S: (i) Preparation of LLL12-NHS ester via DCC coupling, deprotection, and NHS activation; (ii) Preparation of G6 bifunctional dendrimer using Fmoc-GABA−OH; (iii) Final conjugation of bifunctional dendrimer with LLL12-NHS ester via amidation, with 5.9% (w/w) LLL12 payload. (B) ¹H NMR spectrum of G6-LLL12S in DMSO-d₆ showing characteristic peaks of LLL12, linker, and dendrimer with color-coded assignments.

In parallel, we synthesized amine-derived G6 bifunctional dendrimer, following previously established procedures.⁴² We first functionalized 27.14 out of the 256 terminal hydroxyl groups on dendrimer 4 by introducing Fmoc-protected amine moieties via Fmoc-GABA−OH 13 (Figure S15, see SI). Without further purification, the Fmoc group was removed using a piperidine:DMF (1:4) solution, which afforded bifunctional dendrimer 15 (Figure 2Aii). ¹H NMR analysis confirmed the successful attachment of 13.13 amine groups to the dendrimer by proton integration method, while 14.0 amine units were lost during the Fmoc deprotection reaction. The characteristic peaks of GABA have been observed at 1.8 ppm, along with a shift in OH-attached methylene protons at 4.0 ppm and amide NH at 8.2 ppm, supporting structural modifications (Figures S17−S18, see SI).

Final conjugation involved the amide bond formation between bifunctional dendrimer 15 and activated LLL12 NHS ester 12 in the presence of di-isopropyl ethylamine (DIEA). The resulting reaction afforded G6-LLL12S 16 in 94% yield (Figure 2Aiii). Purity was confirmed by RP-HPLC, where the appearance of a new peak at 5.33 min in the HPLC chromatogram at 250 nm is free of unconjugated LLL12 (8.2 min) or dendrimer (5.09 min). Compared to unmodified dendrimers, the peak of G6-LLL12S has a higher retention time and is more broadly distributed (Figure S19, see SI). The ¹H NMR analysis displayed four doublets corresponding to aromatic protons of LLL12 at 8.57, 8.48, 7.61, 7.38 ppm, its linker methylene protons at 1.98 ppm, and dendrimer peaks at 2.23−3.41 ppm region, confirming the formation of G6-LLL12S conjugate. Quantification of LLL12 loading was performed by integrating the ¹H NMR peaks of the dendrimer and comparing them with the aromatic proton peaks of LLL12. This analysis revealed that 12.89 LLL12 molecules were conjugated per dendrimer, corresponding to an approximate weight percentage of 5.9% (Figure 2B).

Synthesis of Hydrazone Linker-Based G6-LLL12 Conjugate (G6-LLL12H, 20).

To construct G6-LLL12H, we utilized C10 ketone in LLL12 to introduce a hydrazone for the synthesis of intermediate compound azidoacetyl-hydrazone-LLL12 19. First, LLL12 reacted with hydrazine monohydrate in refluxing ethanol in the presence of a catalytic amount of acetic acid, affording LLL12-hydrazone intermediate 17. This intermediate was subsequently reacted with azidoacetic acid NHS ester 18 to afford azidoacetyl-hydrazone-LLL12 19 in 84% yield (Figure 3Ai). Compounds 17 and 19 were thoroughly characterized by ¹H and ¹³C NMR spectroscopy (Figure S20−S23, see SI). Notable key peaks observed at 4.22 ppm (linker methylene protons), 11.62 ppm (hydrazone NH), 11.94 ppm (OH), and 8.02 ppm (sulfonamide NH₂) confirmed successful modification of LLL12 (Figure S22, see SI). Additionally, the disappearance of one of the ketone peaks in the ¹³C NMR spectrum (previously observed between 181 and 183 ppm) further validated the incorporation of the hydrazone linker at the C10 ketone position (Figures S21 and S23, see SI). The appearance of the linker methylene carbon at 50.3 ppm, together with mass spectrometry data, confirmed the successful formation of LLL12−hydrazone linker (Figure S23-24, see SI).

Figure 3.

Construction of G6-LLL12 conjugate via hydrazone linkage (G6-LLL12H). (A) Schematic representation of the stepwise synthesis of G6-LLL12H: (i) Azido-terminated LLL12-hydrazone was prepared by treating LLL12 with hydrazine monohydrate, followed by coupling with azidoacetic acid NHS ester; (ii) The final G6-LLL12H conjugate was obtained via CuAAC click chemistry, with a LLL12 loading of 7.25% (w/w). (B) ¹H NMR spectrum of G6-LLL12H in DMSO-d₆, showing characteristic peaks corresponding to LLL12, linker, and dendrimer moieties, with color-coded assignments matching the conjugate structure.

To construct G6-LLL12H 20, we next used the same heptyne-derived bifunctional dendrimer employed in the synthesis of G6-LLL12C. In this instance, we carried out CuAAC click chemistry between compound 19 and dendrimer 6 using a microwave reactor at 45 °C. This reaction provided 0.77 g of purified G6-LLL12H 20 with a yield of 95% (Figure 3Aii). A similar yield (91%) was obtained when the reaction was scaled up to 2.25 g (0.03696 mmol of 6 and 0.96 mmol of 19). ¹H NMR analysis revealed two doublets at 8.41 and 6.95 ppm, and two triplets at 7.62 and 7.57 ppm, which correspond to the aromatic protons of LLL12 (Figure 3B). The successful conjugation was further validated by the presence of a singlet at 7.69 ppm for the triazole CH proton, and linker methylene protons at 5.14 and 1.61 ppm, sulfonamide NH₂ protons at 8.78 ppm, and an OH proton at 13.10 ppm, along with dendrimer peaks in between 4.72 and 2.21 ppm range (Figure 3B). The LLL12 payload was quantified by ¹H NMR spectroscopy through comparative integration of the dendrimer amide proton peaks (7.94−7.79 ppm) and the aromatic proton signals of LLL12. This analysis revealed an average of 16.44 LLL12 molecules covalently attached per dendrimer, corresponding to a weight load of 7.25% (Figure 3B). Collectively, these results confirmed the successful conjugation of LLL12 to dendrimers while preserving their structural integrity. The RP-HPLC chromatogram of G6-LLL12H at 250 nm displayed a single broadly distributed peak at 5.44 min, indicating a high-purity conjugate with no detectable free LLL12 (8.20 min) or LLL12 derivatives. Similar to the chromatogram of other dendrimer conjugates, the peak of G6-LLL12H shifted toward a higher retention time and appeared to be broader, indicating G6-LLL12H is slightly more hydrophobic and poly disperse than G6-OH (5.09 min), potentially caused by the conjugation of LLL12 to the dendrimer surface (Figure S25, see SI).

Physiochemical Property Characterization of G6-LLL12 Conjugates.

The size and surface properties are critical parameters in determining the unique in vivo behaviors of dendrimer-based therapeutics.^19,49,50 To determine how conjugation of LLL12 affected these properties, we measured the particle hydrodynamic diameter and their ζ-potential in 10 mM NaCl solution.

Size.

The hydrodynamic diameter of the parent dendrimer was determined to be 5.85 ± 0.11 nm. Conjugation of LLL12 via a PEG4-carbamate linker or hydrazone linker did not significantly alter the size of the final G6-LLL12 conjugates (G6-LLL12C: 6.42 ± 0.46 nm, G6-LLL12H: 7.08 ± 0.88 nm). In contrast, the amid linker-based conjugate G6-LLL12S showed a slightly larger size: 8.18 ± 0.65 nm (Figure 4A and Table 1). Previous studies have shown that covalent attachment of bulky hydrophobic molecules to the dendrimer surface could yield complexes with variable sizes due to the hydrophobic interactions induced by the payload, even at a 1:1 payload to dendrimer ratio.^51,52 We did not observe an extensive size increase after LLL12 attachment in our system. Higher generation dendrimers have large capacities to entrap guest molecules in their hydrophobic interior cavities.⁵³ It is possible that the linker might have facilitated the folding of the hydrophobic LLL12 into the hydrophobic interior core, preventing the aggregation of the dendrimers through hydrophobic interaction. To further investigate aggregation behavior, we measured the polydispersity index (PDI) of the conjugates. All conjugates exhibited acceptable PDI values (≤0.3), indicating that they were well dispersed and nonaggregated (Figure 4B and Figure S26). It should be noted that based on previous studies, the distribution of drug molecules on dendrimer surface is inherently random and often follow Poisson distribution,^54–56 therefore the molecular payload of 16 drug per dendrimer is an average value.

Figure 4.

Size, zeta potential, and release kinetics of G6-LLL12 conjugates.(A) Size distribution profiles of G6-LLL12C, G6-LLL12S, and G6-LLL12H conjugates measured by dynamic light scattering (DLS) at 1 mg/mL concentration in 10 mM NaCl solution. For each sample group, three biological replicates were assessed. Statistically significant differences were observed between the sizes of G6-OH and G6-LLL12S (**p < 0.01). (B) The Polydispersity index of G6-OH and G6-LLL12 conjugates was measured in a 10 mM NaCl solution. For each sample group, three biological replicates were assessed, and no statistically significant differences were found between them. (C) The zeta (ζ) potential of G6-LLL12C, G6-LLL12S, and G6-LLL12H conjugates was measured at 1 mg/mL concentration in 10 mM NaCl solution, demonstrating near-neutral surface charge. For each sample group, three biological replicates were assessed, and no statistically significant differences were found between them. (D) Comparison of water solubility between free LLL12 and its dendrimer conjugates (G6-LLL12C, G6-LLL12S, and G6-LLL12H) at a concentration equivalent to 10 mg/mL of LLL12. (E) Comparison of LLL12 release percentage through HPLC chromatogram in citrate (pH 4.5) buffer after 30 min vs after 30 days at 250 nm. (F) The release profiles of LLL12 from G6-LLL12C, G6-LLL12S, and G6-LLL12H were assessed in a pH 4.5 citrate buffer by dissolving 10 mg of G6-LLL12C, 8 mg of G6-LLL12S, and 10 mg of G6-LLL12H, which were individually incubated in 1.25 mL of citrate buffer at 37 °C under continuous mixing. At predetermined time points, 50 μL aliquots were withdrawn and directly analyzed by HPLC at 250 nm to quantify the released LLL12. Release data was collected over time to compare the release efficiency of each formulation. (G) Following the pH 4.5 release study, the release of LLL12 from all conjugates was assessed at 250 nm in pH 7.4 (PBS), with the release percentage compared at different time points. (H) Following the previous release study, the release of LLL12 from all conjugates was assessed at 250 nm in fetal bovine serum (FBS), with the release percentage compared at different time points. (I) Comparison of LLL12 release percentage from G6-LLL12H in pH 4.5 vs pH 7.4 vs saline vs water vs serum at 250 nm.

Table 1.

Particle Size and ζ-Potential of G6-LLL12 Nanoformulations

Compound	Size (nm)	ζ-potential (mV)
G6-OH (4)	5.85 ± 0.11	0.4 ± 0.4
G6-LLL12C (7)	6.42 ± 0.46	−1.1 ± 0.9
G6-LLL12S (16)	8.18 ± 0.65	0.57 ± 0.27
G6-LLL12H (20)	7.08 ± 0.88	0.2 ± 1.0

Surface Charge.

Overall, the ζ-potential of all conjugates remains neutral. Specifically, the ζ-potentials of G6-LLL12C (−1.1 ± 0.9 mV), G6-LLL12S (0.57 ± 0.27 mV), and G6-LLL12H (0.2 ± 1.0 mV) were comparable to that of the parent G6-OH dendrimer (0.4 ± 0.4 mV) (Figure 4C and Table 1). This suggests that conjugating LLL12 to dendrimers did not significantly affect its surface charge.

Water Solubility.

LLL12 has a relatively high LogP value of 2.596.⁵⁷ The lipophilic nature of this molecule poses challenges for drug formulation. G3−G5 PAMAM dendrimers have been widely used to enhance the solubility of hydrophobic molecules.¹⁶ Next, we sought to determine how dendrimer-based formulation can improve the aqueous solubility of the hydrophobic LLL12. At an equivalent concentration of 10 mg/mL LLL12, all three G6-LLL12 conjugates were completely dissolved in water. In contrast, free LLL12 was poorly soluble, forming a visible precipitate (Figure 4D). In previous studies, LLL12 has been dissolved in DMSO for in vitro and in vivo experiments.^7,8 Our results showed that dendrimer-based formulation improved the aqueous solubility of LLL12, enabling the administration of LLL12 through a systemic route.

Hydrolyzable Linkers Enable pH-sensitive Drug Release at Different Rates.

An ideal dendrimer-drug conjugate needs to remain stable under physiological conditions while efficiently releasing drug molecules in response to intracellular chemical cues, such as the intracellular acidic pH. To evaluate the in vitro release characteristics of G6-LLL12 conjugates, we incubated G6-LLL12 conjugates separately in 1X PBS (pH 7.4) or serum (nonheat inactivated Fetal Bovine Serum) or 0.5 M citrate buffer (pH 4.5) at 37 °C. On the designated dates, release study samples were collected and analyzed using RP-HPLC (Figure 4E). For example, after 30 min of incubation in citrate buffer (pH = 4.5), a broad peak at RT = 5.44 min was observed, corresponding to the peak of G6-LLL12H (Figure S25, see SI), along with a small sharp peak at RT = 8.20 min, corresponding to the peak of LLL12. This indicates the LLL12 was released from G6-LLL12H in an intact form. After 38 days of incubation, the size of the G6-LLL12H peak decreased, while the size of LLL12’s peak increased substantially, indicating the successful release of LLL12. Based on the RP-HPLC, we analyzed the release rates for all G6-LLL12 conjugates in 1X PBS (pH 7.4), serum (physiological condition) and 0.5 M citrate buffer (pH 4.5). Our results showed that all three G6-LLL12 conjugates showed a faster release rate at pH 4.5 than pH 7.4 and serum (Figure 4F−H), indicating amide, carbamate, and hydrazone linker-based linkers are more labile under acidic conditions. Under physiological serum conditions, no notable influence of proteins or enzymes on drug release was observed, and the release behavior was comparable to that in pH 7.4 PBS.

In pH 4.5 (Figure 4F), all three G6-LLL12 conjugates showed a sustained release with no initial burst release. The release rates for all conjugates follow G6-LLL12H (hydrazone) > G6-LLL12S (amide) > G6-LLL12C (carbamate), with hydrazone-based conjugates releasing 98% payload by day 87 (Figure 4F). This trend is attributed to the pH sensitivity of hydrazone linkage, which is more prone to hydrolysis even under mild acidic conditions when compared to the counterparts’ carbamate and amide functionalities. We further evaluated the release profiles of G6-LLL12H in water, saline, 1X PBS, and serum, all of which showed minimal drug release (Figure 4I), indicating pH-sensitivity of the hydrazone-based conjugate G6-LLL12H was triggered by the lower pH instead of the salts, proteins, or enzymes in the medium.

G6-LLL12H Exhibits Highest Inhibition Potency and Cytotoxicity.

To determine whether the G6-LLL12 conjugates can successfully inhibit the STAT3 activation, we determined the IC₅₀ for all three conjugates using a THP-1 human monocytic leukemia cell line, which expresses activated STAT3 in response to human interleukin 6 (rhIL-6).⁵⁸ To enable the monitoring of STAT3 activation/inhibition, we generated a stable THP-1_STAT3‑Luc reporter cell line containing a firefly luciferase reporter driven by STAT3 response elements located upstream of the minimal TATA promoter. Upon escalating doses of rhIL-6 stimulation, the THP-1_STAT3‑Luc cells showed luciferase activity in a dose-dependent manner, while the unstimulated control displayed baseline luciferase activity, validating that the observed increase was specific to rhIL-6-induced STAT3 activation (Figure S27, see SI).

To evaluate the STAT3 inhibition efficiency, we preincubated THP-1_STAT3‑Luc cells with escalating doses of G6-LLL12 conjugates and vehicle controls for 72 h. On the day of evaluation, THP-1_STAT3‑Luc cells were stimulated with IL-6 for 5 h to allow maximum STAT3 activation before analysis of inhibition efficiency (Figure 5A). Based on the dose−response curve of STAT3 inhibition (Figure 5Bi), we calculated the IC₅₀ of STAT3 inhibition for each treatment group (Figure 5Bii). The hydrazone-based formulation G6-LLL12H showed the highest potency, with IC₅₀ = 0.42 ± 0.035 μg/mL, similar to the free drug (IC₅₀ = 0.31 ± 0.05 μg/mL) (*p = 0.0163), followed by amide-based formulation G6-LLL12S (IC₅₀ = 6.97 ± 3.69 μg/mL) and carbamate-based formulation G6-LLL12C (IC₅₀ = 28.82 ± 15.30 μg/mL). Of note, the hydroxyl dendrimer G6-OH per se showed nonspecific STAT3 inhibition, with an IC₅₀ = 38.52 ± 6.27 μg/mL, which is not statistically significant compared to the IC₅₀ of carbamate-based formulation G6-LLL12C (*p-value of 0.0198), indicating the STAT3 inhibition activity of G6-LLL12C might not be a direct effect of LLL12.

Figure 5.

G6-LLL12 conjugates exhibit comparable efficacy to LLL12 while offering a safer profile with an enhanced therapeutic window. (A) Genetically engineered human monocytic leukemia STAT3 reporter cell THP-1_STAT3-Luc were treated with increasing concentrations of G6-OH, LLL12, and G6-LLL12 conjugates for 72 h. STAT3 inhibition was evaluated by rhIL-6 stimulation for 5 h, followed by measuring luminescence using D-Luciferin. Cytotoxicity was assessed using CellTiter-Blue^ⓇReagent, with fluorescence measured at 560/590 nm. All treatments were conducted in triplicate, with appropriate controls. (B) (i) Dose−response curves depicting the STAT3 inhibitory effects of LLL12, G6-LLL12C, G6-LLL12S, G6-LLL12H. Data is presented as mean ± SD (n = 3). IC₅₀ curve was plotted using the Standard Sigmoidal, 4PL, X is log(concentration) G6-OH: 38.52 μg/mL, LLL12: 0.307 μg/mL; G6-LLL12H: 0.42 μg/mL, G6-LLL12C: 28.82 μg/mL and G6-LLL12S: 6.97 μg/mL. (ii) G6-LLL12H has high STAT3 inhibitory activity followed by G6-LLL12S. G6-OH treatment indicates minimal or no STAT3 inhibitory effect, *p = 0.0198; LLL12 v.s. G6-LLL12H shows a marginal decrease in LLL12 potency, *p = 0.016. (C) G6-LLL12 conjugates have reduced toxicity due to controlled release of LLL12. Data is presented as mean ± SD (n = 3). The statistical significance between the treatment compounds at specific concentrations was calculated by two-way ANOVA using Šidak multiple comparisons test. (D) The therapeutic window of the small molecule STAT3 inhibitors was evaluated by overlaying the dose−response relationship (black curve) representing efficacy and cell viability (red curve). The broader therapeutic window observed for G6-LLL12 conjugates compared to LLL12 suggests a potentially improved safety and efficacy profile.

We next compared the cytotoxicity of all G6-LLL12 formulations with the free LLL12, following a similar treatment scheme as the efficacy study (Figure 5A). Previous studies have shown many high potency STAT3 inhibitors can induce off-target toxicities at high intracellular concentration, due to nonspecific binding with proteins beyond STAT3.^59–62 Our results showed that free LLL12 was toxic, with almost 100% cell death observed at a concentration beyond 1.23 μg/mL, while G6-OH per se showed minimal to low cytotoxicity within the entire dosing range (Figure 5C). All three G6-LLL12 conjugates showed lower cytotoxicity compared to free LLL12. Specifically, at the equivalent dose range, G6-LLL12H showed the highest cytotoxicity, with over 50% cells remaining viable up to 0.4 μg/mL. G6-LLL12S-treated cells maintained over 50% viability up to 11.1 μg/mL, whereas G6-LLL12C showed the lowest cytotoxicity; over 50% cells remained viable up to 33.3 μg/mL treatment.

Both the IC₅₀ and cytotoxicity analyses showed similar trends for all three G6-LLL12 conjugates, with the hydrazone-based formulation G6-LLL12H being the most potent and toxic candidate, followed by the amide-based formulation G6-LLL12S. The carbamate-based formulation G6-LLL12C is the least potent and toxic candidate. This trend also correlates with the release rate of these formulations, with the release rate of G6-LLL12H > G6-LLL12S > G6-LLL12C. It is well-established that the drug release rate/linker stability is a critical attribute of dendrimer-drug conjugates that determines their potency and cytotoxicity.^37–41 Stable linkers such as carbamate reduced premature release and off-target cytotoxicity associated with the LLL12, but also showed minimal STAT3 inhibition, while unstable linkers such as hydrazone ensured efficient drug release and potency but also increased the cytotoxicity. The balance between linker stability and payload release is crucial in enhancing the therapeutic index of dendrimer-drug conjugates.

G6-LLL12 Formulations Demonstrate Enhanced Safety and Therapeutic Window.

To determine the effect of release rate/linker stability on the therapeutic window of each formulation, we overlaid the IC₅₀ curve for pSTAT3 inhibition (black) with the corresponding cell viability curve (red) on a single plot (Figure 5D). In this context, we defined the in vitro therapeutic window as the dose range between the onset of biological efficacy and the concentration at which 50% of the cells remain viable. The shaded area between the two curves represents this window, offering a visual and comparative measure of each formulation’s safety and efficacy profile. LLL12 demonstrated a significantly narrower therapeutic window compared to G6-LLL12 conjugates, indicating a limited margin between achieving STAT3 pathway suppression and compromising overall cell viability. This relationship highlights a key limitation of LLL12, although it potently inhibits the activated STAT3 and its associated signaling pathway, its high cytotoxicity at concentrations close to its IC₅₀ indicates limited selectivity and a narrow therapeutic window. All three G6-LLL12 conjugates demonstrated a broader therapeutic index, with 4.3-fold, 6-fold, and 2.8-fold increases observed for G6-LLL12C, G6-LLL12S, and G6-LLL12H, respectively, compared to free LLL12. It is possible that the wider therapeutic window was caused by the sustained release of LLL12 from the G6-LLL12 conjugates, which minimized the nonspecific cytotoxicity associated with the presence of extremely high concentrations of intracellular LLL12. Further studies should focus on investigating whether G6-LLL12 conjugates can extend the therapeutic window of the free drug in vivo, which requires a balance between efficient intracellular payload release and premature payload release during systemic circulation.

G6-LLL12H Exhibits Similar Activity to LLL12 in Reducing M-MDSC Populations and Promoting APC Maturation.

To determine if the G6-LLL12 conjugates could alter the functional outcomes of LLL12-induced STAT3 inhibition, we selected the most potent formulation, G6-LLL12H, for evaluation. As a master regulator of immunosuppression, aberrant STAT3 activation in cancer can initiate emergency myelopoiesis, resulting in the expansion of immature myeloid cells such as myeloid-derived suppressor cells (MDSCs) and reduction of mature functional dendritic cells (DCs). Previous studies have shown that targeting the JAK2/STAT3 pathway blocks the transcription of STAT3-regulated genes, such as Bcl-2, Cyclin D1, and c-Myc, thereby suppressing the monocytic MDSC (M-MDSC) expansion and promoting the maturation of DCs.^2,7,11,62 Here, we evaluated the effects of LLL12 and G6-LLL12H on MDSC expansion and dendritic cell maturation (Figure 6) using a previously established ex vivo assay.^45,63

Figure 6.

G6-LLL12H and LLL12 reduce M-MDSCs population and enhance dendritic cell maturation in bone marrow cultures. (A) Schematic illustration shows the generation of M-MDSCs from the bone marrow of wild-type C57BL/6 mice. Generation of M-MDSCs was done by isolating bone marrow cells and culturing them in KR158 glioma cell-conditioned media for 3 days. Treatment (LLL12, G6-LLL12H, or vehicle) was added at day 1 of the culture, and flow cytometry was performed at day 3 to assess for dendritic cell maturation and the total number of M-MDSCs. (B) Representative flow cytometry plot denoting G6-LLL12H has a similar effect compared to LLL12 in reducing M-MDSCs population and enrichment of dendritic maturation and activation after 3 days in culture. (C−E) Quantitative graphs from previous panels. Graphs show the total number of (C) M-MDSCs (Ly6-C^High/Ly6-G^Low), (D) Dendritic cells (CD11c), (E) Costimulatory activation marker (CD86), under 0.4 μg/mL G6-LLL12H, 0.4 μg/mL LLL12, and water (vehicle). (n = 3 mice). One-way ANOVA statistical test was performed (Dunnett’s multiple comparison test). Differences are compared to the vehicle (control condition). P-values: ****p < 0.0001; ***p < 0.0008; **p < 0.0016.

Briefly, bone marrow cells from C57BL/6 mice were plated on day 0, to model the effect of cancer on myelopoiesis, bone marrow cells were cultured in KR158 glioma conditioned media for 72 h in the presence of vehicle (water), LLL12, or G6-LLL12H at concentrations of 0.004 μg/mL, 0.04 μg/mL, or 0.4 μg/mL (Figure 6A). Flow cytometry analysis revealed a significant reduction in Ly6C⁺ M-MDSC populations following treatment with either compound (Figure 6B, left column, Figure 6C, see supplementary Figure S28 for detailed gating strategy). At concentrations of 0.4 μg/mL and 0.04 μg/mL, both LLL12 and G6-LLL12H led to a statistically significant reduction in Ly6C⁺ population compared to vehicle control (***p < 0.0008; **p < 0.0016). To assess the impact on APC maturation, we measured the frequency of CD11c⁺ cells (Figure 6B, middle column, Figure 6D, Figure S28, see SI for detailed gating strategy). Both treatments led to a dose-dependent increase in CD11c expression, with the highest concentration (0.4 μM) resulting in a statistically significant increase compared to vehicle (****p < 0.0001). We next evaluated APC activation by the expression of CD86, a key costimulatory molecule required for T-cell activation (Figure 6B, right column, Figure 6E, Figure S28, see SI for detailed gating strategy). Treatment with LLL12 and G6-LLL12H produced a similar dose-dependent increase in CD86⁺ cells, consistent with enhanced APC activation. These findings highlight the comparable immunomodulatory effects of G6-LLL12H to LLL12, as both treatments induced a reduction in M-MDSC frequency and promoted APC maturation.

CONCLUSIONS

In summary, we developed a series of G6-LLL12 nano-formulations using pH-sensitive carbamate, amide, and hydrazone linkers in excellent yields (91−95%). These nanoconjugates were thoroughly characterized for their physicochemical properties using NMR, DLS, and HPLC. Conjugation of STAT3 inhibitor LLL12 to G6 hydroxyl-terminated PAMAM dendrimer through either of the linkers improved water solubility without largely affecting the size or charge of the dendrimer, facilitating systemic administration. In vitro drug release studies demonstrated that all conjugates remained stable under extracellular physiological conditions but released the drug effectively in acidic intracellular conditions. All three nanoconjugates effectively inhibited STAT3 signaling in a dose-dependant manner. Among them, the hydrazone-linked conjugate demonstrated the most efficient drug release and STAT3 inhibition, comparable to free LLL12. The amide-linked conjugate displayed moderate activity, while the carbamate conjugate showed lower potency but also the lowest cytotoxicity. Importantly, all formulations significantly reduced LLL12-associated cytotoxicity and broadened the therapeutic window by 4.3-fold, 6-fold, and 2.8-fold increase for G6-LLL12C, G6-LLL12S, and G6-LLL12H, respectively. Finally, the lead candidate G6-LLL12H promoted APC maturation and reduced M-MDSC expansion ex vivo. The results of this study set the stage for future in vivo evaluation of dendrimer-based STAT3 inhibitors.

Supplementary Material

Supporting Information

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.biomac.5c01648.

Copies of NMR spectra and HPLC profiles for all the new compounds, number-based particle size distribution for dendrimer-LLL12 conjugates, THP-1_STAT3‑Luc reporter cell line establishment and THP-1_STAT3‑Luc rhIL-6 responsive curve, flow cytometry gating strategy (PDF)

Data Availability Statement

All data supporting the findings of this study are available within the article and its Supporting Information files and directly from F.Z. upon reasonable request.