Theranostic Small-Molecule Prodrug Conjugates for Targeted Delivery and Controlled Release of Toll-like Receptor 7 Agonists

We previously reported the design and synthesis of a small-molecule drug conjugate (SMDC) platform that demonstrated several advantages over antibody–drug conjugates (ADCs) in terms of in vivo pharmacokinetics, solid tumor penetration, definitive chemical structure, and adaptability for modular synthesis. Constructed on a tri-modal SMDC platform derived from 1,3,5-triazine (TZ) that consists of a targeting moiety (Lys-Urea-Glu) for prostate-specific membrane antigen (PSMA), here we report a novel class of chemically identical theranostic small-molecule prodrug conjugates (T-SMPDCs), [18/19F]F-TZ(PSMA)-LEGU-TLR7, for PSMA-targeted delivery and controlled release of toll-like receptor 7 (TLR7) agonists to elicit de novo immune response for cancer immunotherapy. In vitro competitive binding assay of [19F]F-TZ(PSMA)-LEGU-TLR7 showed that the chemical modification of Lys-Urea-Glu did not compromise its binding affinity to PSMA. Receptor-mediated cell internalization upon the PSMA binding of [18F]F-TZ(PSMA)-LEGU-TLR7 showed a time-dependent increase, indicative of targeted intracellular delivery of the theranostic prodrug conjugate. The designed controlled release of gardiquimod, a TLR7 agonist, was realized by a legumain cleavable linker. We further performed an in vivo PET/CT imaging study that showed significantly higher uptake of [18F]F-TZ(PSMA)-LEGU-TLR7 in PSMA+ PC3-PIP tumors (1.9 ± 0.4% ID/g) than in PSMA− PC3-Flu tumors (0.8 ± 0.3% ID/g) at 1 h post-injection. In addition, the conjugate showed a one-compartment kinetic profile and in vivo stability. Taken together, our proof-of-concept biological evaluation demonstrated the potential of our T-SMPDCs for cancer immunomodulatory therapies.

Among the TLRs, TLR7 has served as an attractive and productive druggable target for synthetic TLR agonist development. To date, a large number of TLR7 agonists have been reported [17,18]. Primarily expressed in the cells of the innate immune system that are virtually present in all solid tumors, TLR7 is an intracellular target. As such, one of the essential criteria for TLR7-targeted drug design is that the drug payload must be able to cross the cellular membrane to activate TLR7, which in tandem will interact with myeloid differentiation protein 88 (MyD88) and translocate the NF-κB transcription factor into the nucleus to release pro-inflammatory chemokines and cytokines [19]. Most clinically tested TLR7 targeting small-molecule agonists are imidazoquinoline derivatives (Scheme 1), among which imiquimod is the first Food and Drug Administration (FDA)-approved drug (in 1997) to treat basal cell carcinoma and genital warts [20,21]. Followed by imiquimod's approval, many other potent imidazoquinoline derivatives, such as resiquimod and gardiquimod, have been reported with improved potency and solubility [21,22]. However, while the highly potent imidazoquinoline agents can be potentially used as immunomodulatory agents and vaccine adjuvants for infectious diseases, the dose limiting toxicities and general inflammatory responses of the host observed in their clinical trials are so severe that the systemic administration (i.e., untargeted infusion into the vein) has been abandoned [23][24][25][26][27]. In addition, the agents exhibited poor pharmacokinetic characteristics mainly due to their poor solubility [28]. Consequently, to reach an effective dose for anti-tumor treatment, current strategies require the agents to be directly injected into tumors multiple times on a weekly basis. Given the logistical barriers, risks of bleeding/organ damage/infection, this practice significantly hampers the clinical application of these agents. In addition, the intra-tumoral injection is not applicable to deep-seated metastases, and the fact that tumors are inherently heterogeneous makes the approach of little clinical value. However, while the highly potent imidazoquinoline agents can be potentially used as immunomodulatory agents and vaccine adjuvants for infectious diseases, the dose limiting toxicities and general inflammatory responses of the host observed in their clinical trials are so severe that the systemic administration (i.e., untargeted infusion into the vein) has been abandoned [23][24][25][26][27]. In addition, the agents exhibited poor pharmacokinetic characteristics mainly due to their poor solubility [28]. Consequently, to reach an effective dose for antitumor treatment, current strategies require the agents to be directly injected into tumors multiple times on a weekly basis. Given the logistical barriers, risks of bleeding/organ damage/infection, this practice significantly hampers the clinical application of these agents. In addition, the intra-tumoral injection is not applicable to deep-seated metastases, and the fact that tumors are inherently heterogeneous makes the approach of little clinical value.
We reason that targeted systemic delivery of the TLR7 agonists via our previously reported small-molecule drug conjugate (SMDC) platform [29] could surmount the roadblock that impedes their clinical application. In this work, we present a uniquely designed chemically identical pair of theranostic small-molecule prodrug conjugates (T-SMPDCs), [ 18/19 F]FB-AMP-TZ(PEG 3 -Lys-Urea-Glu)-PEG 6 -LEGU-GARD ([ 18/19 F]F-TZ(PSMA)-LEGU-TLR7), constructed on a tri-modal molecule, 2,4,6-trichloro-1,3,5-triazine (TZ), where the halogen functionality was leveraged for modular synthesis of the conjugation (Scheme 2). The water-soluble short polyethylene glycol (PEG) linkers were employed to optimize the in vivo kinetics of the conjugate as necessary [30]. Urea-based Lys-Urea-Glu [31], which is a commonly used targeting moiety for prostate-specific membrane antigen (PSMA), serves as a model vector for cancer-targeted delivery of TLR7 agonists. Of note, the well-observed PSMA-mediated internalization mechanism upon ligand-binding [32] is leveraged for the intracellular delivery of the TLR7 payload carried by the T-SMPDCs. To enable the designed controlled release [33][34][35], we incorporate a legumain-cleavable linker, Azido-PEG 4 -Ala-Ala-Asn(Trt)-PAB-PNP (Azido-PEG 4 -LEGU), which is stable in the blood, as reported in many antibody drug conjugates (ADCs), to minimize the off-target toxicities [36,37], but undergoes a traceless release (self-immolation) of drugs upon interacting with legumain, a lysosomal endopeptidase. For the proof-of-concept, in this work we chose gardiquimod (GARD) as a model TLR7 agonist to construct our prodrug conjugates, which was coupled to the legumain-cleavable linker. At the last step, an 18

Structural Design and Synthesis
To enable the modular synthesis of T-SMPDCs and avoid potential steric hindrance that might result from the assembly of three designed functionalities, we introduced three different spacers between the tri-modal TZ core and each functional moiety (Scheme 2, Schemes S1-S3, and Figures S1-S21). In brief, propargyl-PEG2-amine was conjugated with TZ in (1:1) stoichiometric fashion to afford monosubstituted TZ-PEG2-PROP, in which the

Radiochemistry
The multistep radiosynthesis of [ 18 F]T-SMPDC was initiated with~55.5 GBq of [ 18 F]fluoride which was helpful for reproducible radiochemical yield and molar activity. The synthetic strategy developed for [ 19

Legumain-Enzyme-Induced GARD Release from [ 19 F]F-TZ(PSMA)-LEGU-TLR7
To test if the T-SMPDC platform is capable of controlled release of GARD upon legumain-catalyzed cleavage, we performed an in vitro GARD release assay using a previously published method [41]. [ 19 F]F-TZ(PSMA)-LEGU-TLR7 (150 μg) was dissolved in 15 μL of DMF and diluted to 400 μL with the cleavage buffer (0.1 M citrate pH 5.5) followed by the addition of 300 μg of cysteine. Murine legumain (7 μg in 25 mM tris buffer) was added to the mixture. After 1 h incubation at 37 °C, the sample showed a quasi-molecular ion peak of 365.07 [M + 2H2O] + , the molecular mass of hydrated GARD ( Figure S23) released from [ 19 F]F-TZ(PSMA)-LEGU-TLR7 [42]. The legumain-controlled release of GARD was further confirmed by the absence of the molecular ion peak at 365.07 when [ 19 F]F-TZ(PSMA)-LEGU-TLR7 was incubated with the buffer only under the same condition ( Figure S24).

Legumain-Enzyme-Induced GARD Release from [ 19 F]F-TZ(PSMA)-LEGU-TLR7
To test if the T-SMPDC platform is capable of controlled release of GARD upon legumain-catalyzed cleavage, we performed an in vitro GARD release assay using a previously published method [41]. [ 19 F]F-TZ(PSMA)-LEGU-TLR7 (150 µg) was dissolved in 15 µL of DMF and diluted to 400 µL with the cleavage buffer (0.1 M citrate pH 5.5) followed by the addition of 300 µg of cysteine. Murine legumain (7 µg in 25 mM tris buffer) was added to the mixture. After 1 h incubation at 37 • C, the sample showed a quasi-molecular ion peak of 365.07 [M + 2H 2 O] + , the molecular mass of hydrated GARD ( Figure S23) released from [ 19 F]F-TZ(PSMA)-LEGU-TLR7 [42]. The legumain-controlled release of GARD was further confirmed by the absence of the molecular ion peak at 365.07 when [ 19 F]F-TZ(PSMA)-LEGU-TLR7 was incubated with the buffer only under the same condition ( Figure S24).

Discussion
The currently broadly-utilized immunotherapy for cancer with immune checkpoint inhibitors relies on a pre-existing immune response. Non-inflamed, immunologically cold tumors are typically resistant to this treatment approach and the primary reason for treatment failure. The capability of inducing de novo immune responses and enhancing otherwise modest benefits is therefore an unmet clinical need in the field of cancer immunotherapy and subject to significant interest and drug development efforts. Although numerous agonists of the TLR pathways have been designed, developed, and tested clinically in past decades [43], their clinical development has been hampered by systemic side effects when delivered systemically or by the need for serial intratumoral injections. TLR7, however, remains an attractive therapeutic target, since it is virtually always present in the microenvironment of solid tumors. TLR7 agonists delivered into the tumor microenvironment can then further diffuse into bystander cells, thus causing a proinflammatory field effect. The antibody-drug conjugate (ADC) strategy has been well-explored for targeted delivery of drug payloads to reduce systemic toxicities [44][45][46]. However, due to the inherent large molecular weight and low tumor-cell penetration ability of antibodies, ADCs are not an optimal choice to deliver TLR agonists to their cytoplasmic targets.
To overcome those limitations, in this work we present a T-SMPDC platform as a feasible solution [29,47]. The prodrug conjugate was designed to reduce the systemic exposure by covalent linkages (to prevent premature drug release) and targeted delivery of the drug payload by a small-molecule platform that can effectively penetrate into tumor microenvironments [29,48]. Additionally, the T-SMPDC platform builds on the mechanism of tumor-specific receptor/antigen-mediated cell internalization to enable intracellular delivery and an enzyme-cleavable prodrug linker to realize the desired controlled release of the drug onto its action target. For the proof-of-concept study, we constructed a theranostic T-SMPDC system, which consists of a model vector, Lys-Urea-Glu, to target cancer-specific PSMA [49], which in tandem initiates cell internalization upon the vector binding [50,51]. The internalized T-SMPDC would then release its drug payload, GARD, via legumain-catalyzed self-immolation of the prodrug linkage in a traceless fashion [33].
For the practicality of T-SMPDC synthesis, we started from 2,4,6-trichloro-1,3,5-triazine and obtained the key intermediate AMP-TZ(Lys-Urea-Glu)-PEG6-LEGU-GARD after five steps of synthesis with the overall yield of ~2.3%. Of note, the yield from each of the

Discussion
The currently broadly-utilized immunotherapy for cancer with immune checkpoint inhibitors relies on a pre-existing immune response. Non-inflamed, immunologically cold tumors are typically resistant to this treatment approach and the primary reason for treatment failure. The capability of inducing de novo immune responses and enhancing otherwise modest benefits is therefore an unmet clinical need in the field of cancer immunotherapy and subject to significant interest and drug development efforts. Although numerous agonists of the TLR pathways have been designed, developed, and tested clinically in past decades [43], their clinical development has been hampered by systemic side effects when delivered systemically or by the need for serial intratumoral injections. TLR7, however, remains an attractive therapeutic target, since it is virtually always present in the microenvironment of solid tumors. TLR7 agonists delivered into the tumor microenvironment can then further diffuse into bystander cells, thus causing a proinflammatory field effect. The antibody-drug conjugate (ADC) strategy has been well-explored for targeted delivery of drug payloads to reduce systemic toxicities [44][45][46]. However, due to the inherent large molecular weight and low tumor-cell penetration ability of antibodies, ADCs are not an optimal choice to deliver TLR agonists to their cytoplasmic targets.
To overcome those limitations, in this work we present a T-SMPDC platform as a feasible solution [29,47]. The prodrug conjugate was designed to reduce the systemic exposure by covalent linkages (to prevent premature drug release) and targeted delivery of the drug payload by a small-molecule platform that can effectively penetrate into tumor microenvironments [29,48]. Additionally, the T-SMPDC platform builds on the mechanism of tumor-specific receptor/antigen-mediated cell internalization to enable intracellular delivery and an enzyme-cleavable prodrug linker to realize the desired controlled release of the drug onto its action target. For the proof-of-concept study, we constructed a theranostic T-SMPDC system, which consists of a model vector, Lys-Urea-Glu, to target cancer-specific PSMA [49], which in tandem initiates cell internalization upon the vector binding [50,51]. The internalized T-SMPDC would then release its drug payload, GARD, via legumaincatalyzed self-immolation of the prodrug linkage in a traceless fashion [33].
For the practicality of T-SMPDC synthesis, we started from 2,4,6-trichloro-1,3,5-triazine and obtained the key intermediate AMP-TZ(Lys-Urea-Glu)-PEG 6 -LEGU-GARD after five steps of synthesis with the overall yield of~2.3%. Of note, the yield from each of the steps can be further improved by modifying the linker reactivity. The key intermediate, AMP-TZ(Lys-Urea-Glu)-PEG 6 -LEGU-GARD, is stable and can be long-stored for the construction of various T-SMPDCs as necessary. In addition, from the chemistry perspective, it is feasible to scale up the synthesis of AMP-TZ(Lys-Urea-Glu)-PEG 6 -LEGU-GARD. Further step-wise addition of the prodrug moiety, azido-PEG 4 -LEGU-GARD, and the installation of the theranostic [ 18/19 F]SFB functionality, were proven straightforward to afford the chemically identical product pair, [ 18/19 F]F-TZ(PSMA)-LEGU-TLR7, with reasonable yields. [ 18/19 F]F-TZ(PSMA)-LEGU-TLR7 showed the anticipated properties (e.g., serum stability, PSMA binding affinity, PSMA-mediated internalization, and legumain-mediated drug release) and biological behavior (e.g., PSMA-specific uptake and retention in PSMA + tumors), which validate the design concept of our T-SMPDCs for targeted delivery and controlled release of GARD for immunomodulatory therapies. However, we acknowledge that a more clinically relevant animal model must be developed and used in further in vivo theranostic evaluation of the conjugate for immunomodulatory therapies, because TLR7 is expressed in immune cells but our proof-of-concept in vivo studies presented herein were conducted in NOD.CB17-Prkdc scid /NCrHsd mice, which lack mature T and B lymphocytes. Therefore, a syngeneic mouse model that retains intact immune systems and carries PSMA positive tumor grafts will be needed to evaluate the immunotherapy potential of our T-SMPDCs. In addition, the in vivo stability of our T-SMPDCs still needs to be improved. A detailed metabolite assay will have to be performed with [ 19 F]F-TZ(PSMA)-LEGU-TLR7 in the blood and liver in order to identify the metabolite fragments. Once the chemical bonds are identified as being labile in vivo, we will take corresponding chemical strategies to optimize the structure of the T-SMPDC platform.
Without incorporating an albumin-binding moiety [52], the relatively slow in vivo clearance and tissue-distribution profiles of the T-SMPDCs could be advantageous for the therapeutic applications of the prodrug conjugates because the prolonged high plasma concentration facilitates the targeted accumulation and unidirectional internalization to reach the intended intracellular target protein.
While the in vivo imaging evaluation revealed a relatively high tumor-to-muscle ratio (~4.7) for [ 18 F]F-TZ(PSMA)-LEGU-TLR7, it also showed high uptake levels in the heart, lung, and liver. The off-target accumulation likely reflects the lipophilic nature of [ 18 F]F-TZ(PSMA)-LEGU-TLR7, which is due to the presence of aromatic functionalities in the structure of T-SMPDCs. Linkers/spacers in the T-SMPDCs can be readily leveraged to overcome the issue [53]. It is noteworthy that the high off-target accumulation of T-SMPDCs is expected to be non-toxic or at least less toxic than the free molecule of GARD, because GARD is covalently loaded to the conjugate. However, whether premature release of GARD would occur or not in the organs need to be investigated. Notably, the bone uptake of [ 18 F]F-TZ(PSMA)-LEGU-TLR7 was low (0.79 ± 0.07% ID/g), indicative of the desired in vivo stability of the theranostic moiety.
The chemical platform of [ 18/19 F]F-TZ(PSMA)-LEGU-TLR7 is versatile. The modular synthesis we present in this work can be readily adapted for other targeted therapy systems for cancer precision medicine.  Figure S14. Purity of the compound (>99%) was assessed by reverse-phase analytical HPLC.

Synthesis of AMP-TZ(PEG 3 -Lys-Urea-Glu)-PEG 2 -PROP
Protected AMP-TZ(PEG 3 -Lys-Urea-Glu)-PEG 2 -PROP (20 mg, 0.018 mmol) was dissolved in DCM (1 mL), followed by the addition of TFA (1.5 mL) under N 2 . The reaction mixture was stirred for 12 h at room temperature and the product formation was moni-tored via ESI-MS. Then, the reaction solvent was evaporated under reduced pressure in a rotary evaporator to afford a crude product, which was then purified by a reverse-phase HPLC (10% acetonitrile/90% H 2 O to 50% acetonitrile/50% H 2 O over 12 min; all solvents contained 0.1% TFA). Fractions from HPLC were combined from multiple single injections and lyophilized to yield pure compound as a sticky liquid (10 mg, 61%). MS (ESI) m/z calcd for C 39   F]F-TZ(PSMA)-LEGU-TLR7 was collected in a flask, first mixed with 80 mL of water and then passed through a C18 Plus Sep-Pak cartridge to trap the radiolabeled product. The cartridge was first dried with N2 flow and then eluted with EtOH/H 2 O (90/10; v/v; 1 mL) to obtain the final product in two-necked flask, which was finally transferred into product vial.

Preparation of 125 I-Labeled Lys-Urea-Glu Analog
The precursor phloretic acid-PEG 3 -Lys-Urea-Glu (1 mg) was dissolved in 100 µL water to produce a stock solution of 10 µg/µL. A pierce pre-coated iodination tube was rinsed with 1 mL of tris-iodination buffer (5×) and drawn off. Then, 100 µL of tris buffer (5×) was added to the pre-coated tube, followed by 43 MBq of Na 125 I. The iodide was activated for 6 min at room temperature by swirling the tube every 30 s. The activated iodide mixture was then transferred to a reaction vial containing 30 µg of precursor (6) and incubated for 9 min at room temperature. Then, the reaction mixture was purified through a semi-preparative HPLC (5% CH 3

Cell Culture and Animal Model
All animal studies were performed in accordance with relevant guidelines and regulations through an animal protocol (APN: 2020-102858; effective from 26 May 2020 to 26 May 2023) approved by the institutional animal care and use committee (IACUC) at the University of Texas Southwestern Medical Center (Dallas, TX, USA). The cell lines, PC3-PIP (PSMA positive) and PC3-Flu (PSMA negative), used in this work have been extensively used in PSMA-targeting agent development [54,55]. They were obtained from the laboratory of Prof. Dr. Martin G. Pomper at John Hopkins University (Baltimore, MD, USA). The sublines of the androgen-independent PC3 human-prostate-cancer cell line, derived from an advanced androgen-independent bone metastasis, were engineered to express a high level of PSMA (PC3-PIP) and maintain no expression of PSMA (PC3-flu) [56,57]. The cells were cultured in RPMI media with the addition of 10% fetal bovine serum, penicillin/streptomycin, and 1 µg/mL of puromycin. For tumor development, cells were cultured at 37 • C under 5% CO 2 and passaged at 75-90% confluency. Cell suspensions were injected subcutaneously (1.0 × 10 6 cells in 100 µL Hank's Buffered Salt Solution) into the thighs of male severe combined immunodeficient (SCID) mice (NOD.CB17-Prkdc scid /NCrHsd, 6-8 weeks). The mice were housed in laminar flow cages kept at~22 • C with~55% relative humidity in a 12-h light/dark cycle. Throughout the experiment, mice had unrestricted access to autoclaved water and commercial food, and they were checked every other day for general observations and tumor burdens.

Serum Stability Assay
In vitro stability of [ 18 F]F-TZ(PSMA)-LEGU-TLR7 was analyzed with human serum. For this purpose, [ 18 F]F-TZ(PSMA)-LEGU-TLR7 (1.85 MBq, 20 µL PBS) was added with 400 µL of human serum in a 5 mL quartz glass vial and incubated at 37 • C for 1, 2, and 4 h. Then, 100 µL solution from each was diluted with 1 mL ethanol and centrifuged for 5 min; after the supernatant was filtered out with a 0.2 µm filter, stability was analyzed by radio-HPLC.

In Vitro GARD Release Assay by Legumain
Mouse legumain was purchased from Sino Biological Inc. (Chesterbrook, PA, USA) and reconstituted in 25 mM tris-buffered saline (0.15 M NaCl, pH 7.4). GARD release assay was performed in 0.1 M citrate buffer, pH 5.5 (cleavage buffer). 150 µg [ 19 F]F-TZ(PSMA)-LEGU-TLR7 was dissolved in 15 µL DMF and diluted to 400 µL with cleavage buffer followed by the addition of 300 µg cysteine. Legumain (7 µg from stock solution) was added to the mixture and incubated at 37 • C. The first aliquot for HPLC and LC/MS was taken after 1 h. The remainder of the sample was set at 37 • C and analyzed by HPLC and LC/MS at different time intervals. A control GARD-release assay was performed under the same condition without legumain protein. MBq, in 100 µL PBS) was intravenously injected to tumor-bearing SCID mice (n = 3) for small animal PET/CT imaging. Imaging was executed with a Siemens Inveon PET/CT Multimodality System (Knoxville, TN, USA) and the mouse was sedated with 2% isoflurane anesthesia throughout the scan. Static 15 min PET scans were conducted at 1 hr p.i. followed by 7 min CT image acquisition at 80 kV and 500 µA with a focal spot of 58 µm. All the PET and CT data were reconstructed, and regions of interest (ROIs) were marked as displayed by CT to quantify the tracer uptake as percent injected dose per gram of tissue (%ID/g).

Conclusions
We have successfully designed and developed a unique class of T-SMPDCs for PSMAtargeted delivery and controlled release of TLR7 agonists for immunomodulatory therapies. Further structural optimizations are required to fully unleash the translational potential of this prodrug conjugate system.