Engineering POLYTAC Nanoparticles for Bioorthogonal Click Reaction-Enforced Protein Degradation and Breast Cancer Therapy

Haijun Yu (  hjyu@simm.ac.cn ) Shanghai Institute of Materia Medica, CAS https://orcid.org/0000-0002-3398-0880 jing gao Qiwen Zhu Shanghai Institute of Materia Medica, CAS Bo Hou Shanghai Institute of Materia Medica, CAS Lei Yang Nanjing University of Chinese Medicine Xingyu Jiang Nanjing University of Chinese Medicine Zhifeng Zou Shanghai Institute of Materia Medica, CAS Xutong Li Shanghai Institute of Materia Medica, CAS Zhiai Xu East China Normal University Tianfeng Xu Shanghai Institute of Materia Medica, CAS Mingyue Zheng Shanghai Institute of Materia Medica https://orcid.org/0000-0002-3323-3092 Huixiong Xu Tongji University


Introduction
The heterobifunctional PROteolysis TArgeting Chimeras (PROTACs) hold promising potential for cancer therapy by degrading the onco-proteins, in particular the undruggable target [1][2][3] . PROTACs are generally composed of a "warhead" binding the protein of interest (POI), a ligand hijacking endogenous E3 ubiquitin ligase, and the linker connecting the warhead and the ligand [4][5][6] . PROTAC can label the POI with ubiquitin by recognizing the E3 ligase and subsequently, degrade the POI with the ubiquitin-proteasome system (UPS) [7][8][9] . Compared to the small molecular inhibitors, PROTAC can potentially degrade any intracellular proteins, including those undruggable targets (e.g., transcriptional factors and scaffold proteins) [10][11][12] . Furthermore, PROTACs are potent for circumventing acquired drug resistance via degrading whole proteins with low drug exposure time and dosage [13][14] . Despite promising, the conventional small molecular PROTACs generally display unfavorable pharmacokinetics and lack of tumor speci city, which might cause systemic toxicity due to their non-speci c distribution in the normal tissues [15][16] . It remains a formidable challenge to achieve tumor-speci c delivery and potentiate the antitumor potency of the conventional PROTACs.
To achieve tumor-targeted delivery of the PROTACs, several ligand-modi cation strategies (e.g., antibody-PROTACs, folate-PROTAC and Aptamer-PROTAC conjugates) have been investigated in past years [17][18][19][20][21][22] . With the assistance of these ligand "gunsight", the decorated PROTACs showed increased cellular uptake in the tumor cells in vitro. In particular, the Aptamer-PROTAC conjugates displayed increased tumor accumulation and antitumor potency than the conventional PROTACs. Nevertheless, these ligand-PROTAC conjugates suffer from low serum stability, limited tumor penetration and heterogeneous expression of the receptors in different tumor cells and cancer types. Furthermore, opto-PROTACs were developed for ultraviolet light-inducible protein degradation [23][24] . These photo-activatable PROTACs were demonstrated for spatiotemporally controllable protein degradation in vitro. However, clinical translation of the opto-PROTACs was restricted by the poor tissue penetration pro le of the ultraviolet light.
Therefore, precise PROTACs delivery to the tumor and e cient POI degradation inside the tumor cells remains a formidable challenge.
To this end, we proposed a rationally-designed polymeric PROTAC (POLY-PROTAC) nanotherapeutic for tumor-targeted delivery of the conventional PROTACs and degradation of bromodomain and extraterminal (BET) protein BRD4. The POLY-PROTAC nanoparticles (NPs) were engineered by utilizing the extracellular and intracellular signals (e.g., acidity, enzyme, and reduction) in the tumor microenvironment [25][26][27] . We rst synthesized four VHL-based RPOTACs for BRD4 degradation, and then prepared a series of POLY-PROTACs by reversible addition-fragmentation chain transfer (RAFT) copolymerization approach. The amphiphilic POLY-PROTACs were self-assembled into micellar NPs for systemic delivery of the BRD4 PROTACs (Fig. 1A). To enforce tumor speci city of the POLY-PROTAC, an extracellular tumor acidity (i.e., pH = 6.5 ~ 6.8)-activatable pretargeted NP was subsequently engineered for tumor-speci c delivery of dibenzocyclooctyne (DBCO) groups. The azide-modi ed POLY-PROTAC NPs were thus trapped inside the extracellular matrix of the tumor via in-situ click reaction between the DBCO and azide derivatives. The POLY-PROTAC NPs can be relieved in the tumor mass via extracellular matrix metalloproteinase-2 (MMP-2)-mediated cleavage of the PEG corona. Upon internalization into the tumor cells, the POLY-PROTAC NPs were dissociated inside the acidic endocytic vesicles (i.e., pH = 5.5 ~ 6.5), and released the PROTAC payload via GSH-mediated reduction of the disul de bond (Fig. 1B). We further demonstrated that the clickable POLY-PROTAC NPs synergistically induced apoptosis of the tumor cells in combination with photodynamic therapy (PDT) in a mouse model of MDA-MB-231 breast cancer (Fig. 1C). This study might provide a generalizable POLY-PROTAC nanoplatform for tumor-speci c PROTAC delivery and potentiated cancer therapy.

Synthesis And Characterization Of The Poly-protac Nps
The BET family proteins, in particular BRD4, have been investigated as promising antitumor targets due to their crucial role for gene transcription 28 . To design the BRD4-targeted POLY-PROTAC, we selected von Hipel-Lindau (VHL) ligand for PROTAC synthesis since the hydroxyl group in VHL can be reversibly caged via a disul de spacer. The VHL protein binding ability of the modi ed PROTACs can be restored by GSHtriggered cleavage of the disul de bond in the cytosol of tumor cells.
The POLY-PROTAC micellar NPs were then prepared via nanoprecipitation method as described previously [29][30] . At neutral pH of 7.4, dynamic light scattering (DLS) data and transmission electron microscopy (TEM) examinations revealed averaged hydrodynamic diameter of ~ 55 nm and spherical morphology of the POLY-PROTAC NPs with narrow particle size distribution (polydispersity index (PDI) < 0.2). In contrast, amorphous aggregates appeared at pH of 6.0 due to acid-induced protonation of DPA groups and dissociation of the POLY-PROTAC NPs (Figs. 2H, I).
HPLC examination displayed that in the absence of GSH, ARV771 was marginally released from the PGD7 NPs at either neutral or acidic condition (i.e., pH = 7.4 or 6.0) (Fig. 2K). In contrast, ARV771 was remarkably restored with the addition of 10 mM GSH. For instance, over 50% of ARV771 was released upon 4 h incubation with 10 mM of GSH solution at pH of 6.0, which was ~20% higher than that determined at neutral pH. This phenomenon could be explained by the increased GSH accessibility when the POLY-PROTAC NPs was dissociated at acidic condition.

BRD4 degradation with the POLY-PROTAC NPs in vitro
To demonstrate the advantage of the MMP-2-sheddable POLY-PROTAC for increased cellular uptake and deep tumor penetration ( POLY-PROTAC bearing ethylene glycol linker negligibly affected BRD4 expression (Fig. 3K), validating the crucial role of GSH-triggered reduction of the disul de bond and restoration of the VHL ligand for protein degradation inside the tumor cells. MMP-2 pretreatment of the MMP-2-sheddable POLY-PROTAC NPs remarkably reduced the DC 50 of the PGD7 and PGDM NPs, which were 1.9-and 3.1-fold lower the MMP-2insensitive PD7 and PDM counterparts respectively, and comparable to that of the free ARV771 and MZ1 (Fig. 3L). This could be attributed to increased cellular uptake of the PGD7 and PGDM NPs upon MMP-2mediated cleavage of the PEG corona ( Supplementary Fig. 33).
Co-treatment with proteasome inhibitor MG132 abolished the protein degradation ability of the small molecular PROTACs (e.g., MZ1 and ARV771) and the POLY-PROTAC NPs (e.g., PGDM and PGD7) ( Fig. 3M), verifying ubiquitin-proteasome-dependent BRD4 degradation pro le of the POLY-PROTAC NPs. CCK-8 assay further revealed increased cytotoxicity of the PGDM and PGD7 NPs than that of the MMP-2 non-responsive PGDO7 control upon 72 h incubation (Fig. 3N). Collectively, above data demonstrated that the POLY-PROTAC NPs with sheddable PEG corona and reduction-liable linker e ciently degraded the POI and suppressed tumor cell proliferation in vitro.

Biodistribution and antitumor performance of the POLY-PROTAC NP in vivo
The  4A). Noticeably, the PGDA7 NPs showed signi cantly higher intratumoral uorescence intensity and slower decline than PDA7 at all the time points. For instance, the PGDA7 group was of 2.0-fold higher tumor uorescence intensity than the PDA7 control at 36 h post-injection (Fig. 4B).
Tumor-speci c distribution of the PGDA7 NPs was con rmed by uorescence imaging of the major organs (e.g., heart, liver, spleen, lung and kidney) and the tumor tissue ex-vivo at 48 h post-injection (Fig. 4C). CLSM examination of the tumor section further illustrated that the PDA7 NPs distributed in the peripheral area of the blood vessels. In contrast, the PGDA7 NP diffused throughout the tumor tissue (Fig. 4D). Taken together, the uorescence imaging and CLSM examination data veri ed increased tumor accumulation and penetration of the PGDA7 NPs due to MMP-2-mediated cleavage of the PEG corona.
We subsequently explored the anti-tumor e cacy of the POLY-PROTAC NPs in MDA-MB-231 tumor model in vivo. The tumor-bearing BLAB/c nude mice were randomly grouped when the tumor volume reached 100 mm 3 (n = 5), and i.v. injected with PBS, ARV771 or the PGD7 NP at an identi ed ARV771 dose of 10 mg/kg. The treatments were repeated every three days for ve times (Fig. 4E). Free ARV771 marginally suppressed MDA-MB-231 tumor growth. In contrast, PGD7 NP signi cantly delayed ~ 50% of tumor growth and consequently elongated the survival of the tumor-bearing mice (Figs. 4F, G).
To elucidate the mechanism underlying the antitumor performance of the POLY-PROTAC NPs, BRD4 degradation in the tumor lysates was evaluated by western blot assay. Fig. 4K validated that the PGD7 NPs dramatically degraded BRD4 protein (Supplementary Fig. 34). Furthermore, PGD7 NP-treatment signi cantly elicited the expression of cleaved-caspase-3 both in vitro and in vivo (Fig. 4L), verifying BRD4 degradation induced apoptosis of the tumor cells since caspase-3 is a crucial executor of apoptosis 33 . BRD4 degradation-induced apoptosis of the tumor cells was further veri ed by Hematoxylineosin (H&E) staining of the tumor sections ex-vivo (Fig. 4H). There is no body weight loss was found during the whole experimental periods ( Supplementary Fig. 35), and hematoxylin-eosin (H&E) staining of the major organs revealed negligible histopathological damage of the major organs ( Supplementary  Fig. 35 The PED pretargeted NPs displayed homogeneous and spherical morphology with an averaged hydrodynamic diameter of ~ 60 nm at the neutral pH value (i.e., 7.4) (Fig. 5B), and dissociated dramatically at acidic pH (i.e., 6.5) mimicking the acidic tumor microenvironment (Fig. 5C). The PPaconjugated PED NPs displayed quenched uorescence at neutral condition (Supplementary Scheme 13 and Fig. 38). In contrast, the uorescence of PPa recovered at pH value lower than 6.6 via extracellular acidity-induced dissociation of the pretargeted NPs (Fig. 5D), validating superior tumor extracellular acidity-responsive property of the pretargeted NPs.
To demonstrate extracellular acidity-triggered bioorthogonal click reaction, the DBCO-loaded PED pretargeted NP and the azide-modi ed N 3 @PGD7 POLY-PROTAC NPs were incubated together at pH of 7.4 or 6.5, respectively. DLS examination showed uniform hydrodynamic diameter and narrow particle size distribution of the mixed pretargeted and POLY-PROTAC NPs in the FBS-containing neutral buffer solution (pH 7.4), suggesting negligible interaction between the PED and the N 3 @PGD7 NPs since the DBCO groups were encapsulated inside the hydrophobic core of the PED NPs ( Supplementary Fig. 40). In contrast, DLS and TEM examination revealed the appearance of amorphous aggregates with broad size distribution at acidic pH of 6.5 (Figs. 5E, F). This phenomenon validated the bioorthogonal click reaction between the DBCO and azide groups to form cross-linked nanostructure between the PED copolymer and azide-functionalized POLY-PROTAC NPs.
We next sought to investigate whether bioorthogonal click reaction can increase intratumoral accumulation and retention of the POLY-PROTAC NPs in vivo. Fluorescence imaging in vivo displayed obvious uorescence signal at the tumor site when the MDA-MB-231 tumor-bearing mice were i.v. injected with the PED pretargeted NPs (Fig. 5G), which was caused by tumor acidity-triggered dissociation of the PED NPs and activation of the PPa uorescence signal via protonation of the EPA tertiary amine (Fig. 5G). CLSM examination of the tumor sections showed that the PED pretargeted NPs colocalized well with the cell membrane (labeled with wheat germ agglutinin, WGA) 2~4h post-injection, verifying that the PED NPs dissociated and exposed the DBCO groups in the extracellular matrix (ECM) of the tumor tissue (Fig. 5H). which could facilitate click reaction between the DBCO and azide groups presenting on the surface of the POLY-PROTAC NPs in the ECM.
To evaluate whether the in-situ click reaction enhance tumor accumulation of the PROTAC molecules in vivo, the N 3 @PGD7 POLY-PROTAC NPs was i.v. administrated 2 h post-injection of the pretargeted NPs.
Fluorescence imaging in vivo displayed much brighter uorescence signal at the tumor site of the PED + N 3 @PGDA7 group compared to the N 3 @PGDA7-injected mouse group (Fig. 5I). For instance, PED + N 3 @PGDA7 group showed ~ 2.0-fold higher uorescence signal than the N 3 @PGDA7 NP alone at 24-36h post-injection (Fig. 5K).
Increased tumor distribution of the POLY-PROTAC NPs was validated by uorescence imaging of the major organs and tumor tissues ex-vivo 48 h post-injection (Fig. 5L). CLSM examination of the tumor sections demonstrated increased tumor accumulation and deep penetration of the N 3 @PGDA7 POLY-PROTAC NPs when administrated post i.v. injection of the pretargeted PED NPs (Fig. 5J). HPLC examination of ARV771 distribution in-vivo further demonstrated increased intratumoral distribution and retention of the POLY-PROTAC NPs via in-situ click reaction. In comparison with free ARV771 and N 3 @PGDA7 NPs, the combination of PED + N 3 @PGDA7 NP remarkably enhanced tumor distribution of ARV771 by 3.9-and 1.9-fold respectively, when examined at 36 h post-injection (Fig. 5M). Both uorescence imaging and HPLC data provided consistent evidence that the in-situ click reaction signi cantly enhanced tumor accumulation and retention of ARV771 PROTAC at the tumor site.

Bioorthogonal POLY-PROTAC NPs regressed breast tumor growth in vivo
Given signi cantly improved tumor distribution of the clickable POLY-PROTAC NPs, we next investigated their antitumor performance in MDA-MB-231 tumor model in vivo (Fig. 6A). Compared to free ARV771 or N 3 @PGD7 NPs alone, the combination of the PED pretargeted and N 3 @PGD7 NPs much more e ciently suppressed ~70% of MDA-MB-231 tumor growth (Fig. 6B), and elongated the survival of the tumorbearing mice (Fig. 6C), which can be attributed to increased ARV771 distribution in the tumor tissue.
In previous studies, we had explored intracellular acidity-activatable PDT for circumventing multidrug resistance of the breast tumor 34 . To explore the potential of the POLY-PROTAC NPs as a generalizable nanoplatform for combinatory therapy, we next sought to combine the bioorthogonal NPs for PDTenforced antitumor therapy (Fig. 6D). The N 3 @PGDA7 NPs integrating the ARV771 POLY-PROTAC and PPa was prepared by co-assembling N 3 @PGDH, PGD7 and PGDA diblock copolymers. Western blot assay displayed that combination of BRD4 degradation and PDT in vitro signi cantly activated the caspase-3 protein in vitro, implying cumulative antitumor performance of the N 3 @PGDA7 NPs (Fig. 6E and Supplementary Figs. 41, 42).
The anti-tumor performance of combinatory PDT and BRD4 degradation was next investigated in vivo. The MDA-MB-231 tumor-bearing BALB/c nude mice were randomly grouped when the tumor volume reached ~ 100 mm 3 and treated with PBS, ARV771, PED + N 3 @PGDA + Laser, PED + N 3 @PGD7, PED + PGDA7 + Laser, or PED + N 3 @PGDA7 + Laser, respectively (Fig. 6F). The N 3 @PGDA7 NP was i.v. administrated 2h post-injection of the PED NP, and 671 nm laser irradiation was applied 36h postinjection of the N 3 @PGDA7 NP. Fig. 6G demonstrated that PDT or ARV771 alone marginally suppressed proliferation of the MDA-MB-231 tumor. In contrast, the combination of the bioorthogonal NPs (PED + N 3 @PGDA7) and PDT dramatically regressed 90% of tumor growth, 1.5-fold more e cient than BRD4 degradation alone by PED + N 3 @PGDA7.
Furthermore, combination of the PED pretargeted NPs with N 3 @PGDA7 and PDT elongated the survival of the tumor-bearing mice by 40% compared to that of the PED + N 3 @PGDA7 group, with 40% animal survived in 100-days post treatment (Fig. 6H). TUNEL staining of the tumor sections revealed obvious apoptosis of the tumor cells in the PED + N 3 @PGDA7 + PDT group, suggesting combination of PDT and BRD4 degradation with the bioorthogonal NPs cumulatively induced apoptosis of the tumor cells in vivo (Fig. 6J). Semi-quantitative analysis of the TUNEL staining data further revealed that PED + N 3 @PGDA7 + Laser 15.2-and 4.2-fold more e ciently induced apoptosis of the tumor cells than ARV771 and PED + N 3 @PGDA7, respectively ( Supplementary Fig. 27). Treatment-induced apoptosis of the tumor cells was further con rmed by H&E staining of the tumor sections ex-vivo (Fig. 6K). Moreover, negligible body weight loss and histopathological damage of the major organs were observed during the experimental period, verifying satisfying biosafety of the POLY-PROTAC NPs (Fig. 6I and Supplementary Fig. 43).

Conclusion
The heterobifunctional PROTACs with protein degradation ability has been recently investigated for cancer therapy 4 . However, clinical translation of the small molecular PROTACs severely suffers from their insu cient bioavailability and low tumor speci city. It remains one priority to develop novel PROTACs for tumor-speci c protein degradation and minimizing the on-target but off-tumor adverse effects. A tumormicroenvironment-activatable POLY-PROTAC nanoplatform was thus engineered herein for tumor-speci c delivery and potentiating antitumor performance of PROTACs.
In comparison with small molecular PROTAC counterparts, the POLY-PROTAC NPs possessed several distinct advantages for tumor-speci c protein degradation and enhanced antitumor potency. First, the POLY-PROTAC NPs with MMP-2-liable PEG corona elongated blood circulation of the small molecular PROTACs, while the PEG corona was cleaved via intratumoral MMP-2 for facilitating tumor-speci c ambulation and retention of the PROTACs. Second, the POLY-PROTAC NPs can be disintegrated in the intracellular acidic microenvironment. The PROTAC prodrug can thus be restored in the cytosol via GSHmediated cleavage of the disul de bond. Thirdly, the PROTAC encapsulation capacity of the POLY-PROTAC NPs can be readily tuned by adjusting the polymerization degree of PROTACs. Furthermore, we demonstrated that the POLY-PROTAC NPs can be adapted to bioorthogonal click reaction-enforced tumorspeci city of the POLY-PROTACs. Remarkably, the azide-modi ed POLY-PROTAC NPs showed 3.9-fold higher tumor accumulation than the small molecular counterpart via in-situ click reaction with the pretargeted NPs, and therefore further boost PROTAC-based cancer therapy. Last but not least, other kind of stimuli-activatable chemical bonds (e.g., thioketone, selenic or boric acid bond) 35,36 can be utilized to achieve spatial-temporally PROTAC activation and reinforce the therapeutic performance.
In summary, a POLY-PROTAC prodrug strategy was developed for tumor-speci c delivery of PROTAC in this study. We demonstrated for the rst time that the extracellular MMP-2, intracellular acidity and reduction multiple stimuli-activatable POLY-PROTAC nanoplatform can achieve increased tumor accumulation, deep tumor penetration and enhanced protein degradation performance over the small molecular counterpart. The POLY-PROTAC nanoplatform with extracellular tumor acidity-triggered bioorthogonal click reaction remarkably enhanced tumor-speci c PROTAC delivery. The clickable POLY-PROTAC nanoplatform can be further enforced with PDT for protein degradation and combinatory therapy, which completely eradicated the MDA-MB-231 TNBC tumor. Extracellular acidic and MMP-2 microenvironment has been well-documented in various solid tumors. Therefore, the acidity-activatable bioorthogonal POLY-PROTAC nanoplatform can be further extended for combinatory therapy of a broad spectrum of cancer by integrating multiple therapeutic regimens (e.g., chemotherapeutics and immunotherapeutics). Taken together, the ingenious POLY-PROTAC nanoplatform might provide a novel avenue for potentiating PROTAC-based cancer therapy. Figure 1 Schematic illustration of tumor microenvironment-activatable POLYTAC nanoparticles for bioorthogonal click reaction-enhanced cancer therapy. a, Synthesis of POLYTAC via RAFT polymerization approach, the PROTAC molecule was modi ed with disul de linker and thereafter copolymerized with DPA using a MMP-2-liable macromolecular chain transfer agent; b, Self-assemble of the BDCO-conjugated pretargeted nanoparticles, and the azide-modi ed POLYTAC nanoparticles; c, Schematic demonstration of bioorthogonal strategy-promoted tumor distribution and combinatory tumor therapy with the POLYTAC nanoparticles, which displayed enhanced tumor distribution and deep penetration through in-situ click reaction with the pretargeted nanoparticles and deshielding of the PEG corona in the extracellular matrix of tumor. The POLYTAC nanoparticles were then internalized with the tumor cells for BRD4 degradation and combinatory cancer therapy with PDT.