A pH‐responsive nanoparticle delivery system containing dihydralazine and doxorubicin‐based prodrug for enhancing antitumor efficacy

The efficacy of nanoparticle (NP)‐based drug delivery technology is hampered by aberrant tumor stromal microenvironments (TSMs) that hinder NP transportation. Therefore, the promotion of NP permeation into deep tumor sites via the regulation of tumor microenvironments is of critical importance. Herein, we propose a potential solution using a dihydralazine (HDZ)‐loaded nanoparticle drug delivery system containing a pH‐responsive, cyclic RGD peptide‐modified prodrug based on doxorubicin (cRGD‐Dex‐DOX). With a combined experimental and theoretical approach, we find that the designed NP system can recognize the acid tumor environments and precisely release the encapsulated HDZ into tumor tissues. HDZ can notably downregulate the expression levels of hypoxia‐inducible factor 1α (HIF1α), α‐smooth muscle actin, and fibronectin through the dilation of tumor blood vessels. These changes in the TSMs enhance the enrichment and penetration of NPs and also unexpectedly promote the infiltration of activated T cells into tumors, suggesting that such a system may offer an effective “multifunctional therapy” through both improving the chemotherapeutic effect and enhancing the immune response to tumors. In vivo experiments on 4T1 breast cancer bearing mice indeed validate that this therapy has the most outstanding antitumor effects over all the other tested control regimens, with the lowest side effects as well.


S C H E M E 1
Schematic illustration of pH-responsive cRGD-Dex-DOX/HDZ nanoparticles.

INTRODUCTION
[3] The fascinating features of this technology include the encapsulation of massive chemotherapeutic agents with low water dispersibility or high biotoxicity into NP carriers, and the precise control of the release of these cargoes into tumor sites, which offer a better drug delivery system to boost therapeutic efficacy while reducing side effects.In addition, various functional NP modifications could further empower NPs with more advanced properties, such as longer blood circulation time, better tumor targeting capability, and more controllable drug release behavior, which could further enhance treatment efficacy.][12] A major reason is the underestimation of the influence of aberrant tumor stromal microenvironments (TSMs) on NP transportation.[15] Therefore, the promotion of NP permeation into deep tumor sites via the regulation of tumor microenvironments has become a critical bottleneck.
[18][19][20][21][22][23][24] For instance, Chen et al. prepared a nitric oxide (NO) nano-delivery system that can continuously release NO in tumors to relax vascular smooth muscle and normalize tumor vasculature, improving tumor vascular perfusion. [25]Similarly, Huang et al. developed a liposome system loaded with a vasodilator (dihydralazine [HDZ]).After three rounds of injections with this system, the released HDZ could dilate tumor blood vessels to improve tumor permeability. [26]Despite great progress, these systems still have some limitations, such as the need for multiple injections of vasodilators prior to the injection of chemotherapeutic agents and their inability to consistently improve tumor vascular status.Additionally, vasodilators are not released specifically in tumors, leading to the relatively mild vasodilatory efficacy of treatment, and the off-target release of vasodilators at other normal tissues may cause adverse side effects, such as hypoglycemia and edema. [27]herefore, a smart drug delivery system with high efficiency in continuously improving the status of tumor vascularity with no need for prior pretreatment is urgently needed.
Herein, we designed and fabricated an HDZ-loaded pHresponsive delivery system based on the cyclic RGD (cRGD) peptide-modified dextran-hydrazone-doxorubicin (cRGD-Dex-DOX) prodrug.(Scheme 1).Introducing the cRGD peptide could enhance the active target capability of the system to the tumor sites while incorporating the hydrazone group could lend the system a pH-responsive drug release behavior that targets the TSMs.Co-loading HDZ and DOX could endow the system with the capability of continuously normalizing tumor vascularity after each round of injection and thus constantly improving the permeation of NPs and DOX to deep tumor sites.We report that after four rounds of injections, the tumor blood vessels were considerably dilated, which further facilitated the downregulation of the expression of hypoxia-inducible factor 1α (HIF-1α), α-smooth muscle actin (α-SMA), and fibronectin expression.These changes in TSMs increased the tumor enrichment and penetration of NPs and drugs and promoted the infiltration of immune cells in tumors.Thus, treatment with this system received the most outstanding antitumor efficiency against 4T1 breast cancer while having the fewest side effects.The system described here provides an effective way to boost the tumor permeability of the NP-based drug delivery system via TSM improvement.

Synthesis and characterization of prodrugs
In this study, two pH-responsive polymer prodrugs (cRGD peptide-modified dextran-hydrazone-doxorubicin [cRGD-Dex-DOX] and dextran-hydrazone-doxorubicin [Dex-DOX]) were prepared according to the synthetic routes of Figure S1.Specifically, the two prodrugs were prepared following three steps: (1) synthesis of dextran propargyl carbonate (Dex-C≡C); (2) preparation of an azide-functionalized pHresponsive DOX derivative (DOX-hyd-N 3 ); and (3) formation of the cRGD-Dex-DOX prodrug by a CuAAC "click" reaction.The Dex-DOX prodrug was synthesized by omitting the cRGD-PEG 2000 -N 3 in the third step.cRGD peptides were introduced into the prodrug cRGD-Dex-DOX to confer the tumor-targeting capability to NPs.In the acidic microenvironments of tumor cells, the hydrazone group in the polymeric prodrug could be cleaved, leading to the release of DOX.
The key chemical structures of DOX-hyd-N 3 and cRGD-Dex-DOX were analyzed using nuclear magnetic resonance hydrogen spectrum ( 1 H NMR), ultraviolet (UV), and fourier transform infrared (FT-IR) as shown in Figure S2-S4.The characteristic peak of -NH-at δ 10.33 ppm (peak f) and the peaks of other signals indicated the successful synthesis of DOX-hyd-N 3 (Figure S2A).The cRGD peptide and DOX signals can be assigned to the protons of both chemical structures, confirming the successful synthesis of cRGD-Dex-DOX (Figure S2B).The mass percentage of cRGD in cRGD-Dex-DOX was ∼9.5% as determined by the bicinchoninic acid method.The loading efficiency of DOX to cRGD-Dex-DOX was measured to be ∼11.2% by UV spectroscopy.

Self-assembly of polymeric prodrugs
The amphiphilic prodrugs, Dex-DOX and cRGD-Dex-DOX, could self-assemble to form NPs with a hydrophobic core (DOX) and hydrophilic shell (Dex) in aqueous solution (Dex-DOX NPs and cRGD-Dex-DOX NPs).Likewise, cRGD-Dex-DOX can co-assemble with hydrophobic HDZ to form NPs (cRGD-Dex-DOX/HDZ NPs) with both DOX and HDZ as the hydrophobic core.The critical aggregation concentration (CAC) of cRGD-Dex-DOX is 167 mg L −1 , determined through fluorescence spectroscopy using pyrene as a probe (Figure S5).Using high-performance liquid chromatography (HPLC), we show that the HDZ loading content (DLC) was about 10.3%.Moreover, we characterized the morphology of the NPs using transmission electron microscopy (TEM), while the particle size distribution ( Dz ), polydispersity index (PDI), and zeta-potential of the NPs were analyzed by the Zetasizer dynamic light scattering (DLS) instrument (Table S1).TEM imaging showed that both cRGD-Dex-DOX NPs and cRGD-Dex-DOX/HDZ NPs have a spherical morphology (Figure 1A,B) and all three NPs display a similar size (Table S1).However, cRGD-Dex-DOX NPs and cRGD-Dex-DOX/HDZ NPs had weakly positive zeta-potentials (0.81 and 1.64 mV, respectively), whereas Dex-DOX NPs had a negative zeta-potential of −8.64 mV (Figure 1C).We further investigated the stability of cRGD-Dex-DOX/HDZ NPs under three conditions (Milli-Q water, phosphate-buffered saline [PBS], and fetal bovine serum [FBS]).The average diameters of cRGD-Dex-DOX/HDZ NPs remained unchanged in all tested conditions over 2 days (Figure 1D), suggesting that cRGD-Dex-DOX/HDZ NPs may have good stability during the delivery process via biological blood system to the target tissues.
To investigate the detailed encapsulation of HDZ into cRGD-Dex-DOX/HDZ NPs, we carried out coarse grained molecular dynamics (CGMD) simulation of HDZ and cRGD-Dex-DOX (see System and Methods).In the initial system, we used a total of 100 cRGD-Dex-DOX chains and 500 HDZ molecules, randomly placed in the simulation box (Figure 1E).After 3000 ns of simulation, we can observe the formation of several clusters of cRGD-Dex-DOX/HDZ NPs (Figure 1F).We have quantified the aggregation cluster in three replicas of CGMD simulations by counting the contacting DOX moieties (contact cutoff = 0.7 nm).The aggregation analysis shows that while the average cluster sizes (N ave ) increase slowly to ∼5 at the end of the simulation.Additionally, the average cluster size (N ave ) reaches a plateau after 2000 ns in all three replica systems, indicating the reasonable convergence of the coarse-grained molecular dynamics (MD) simulations of cRGD-Dex-DOX/HDZ (Figure 1G).Moreover, we calculated the radial distribution function (RDF) of DOX and HDZ respect to every DOX moiety in the last frame of the simulation (Figure 1H).The RDF represents the average distribution of atoms around any given atom within the system and is frequently used to describe the packing architectures of nanoparticles.From the results in Figure 1H, we can observe several peaks in the RDF for DOX-DOX, indicating the packing of DOX moieties.Meanwhile, we also note the peak of the RDF for HDZ-DOX, confirming the loading of HDZ molecules in the hydrophobic core of cRGD-Dex-DOX/HDZ NPs.The loading of HDZ can be clearly observed in the representative cRGD-Dex-DOX cluster structure (Figure 1I) where the packing of HDZ molecules (shown in green) and DOX moieties (shown in red) is highlighted in the enlarged snapshot.It can be noticed that HDZ molecules and DOX moieties tend to pack in a parallel stacking suggesting the potential favorable contribution of the π-π stacking to the encapsulation of HDZ into cRGD-Dex-DOX/HDZ NPs (enlarged view of Figure 1I).

Characterization of the DOX release from cRGD-Dex-DOX/HDZ NPs
The in vitro cumulative release of DOX and HDZ from cRGD-Dex-DOX/HDZ NPs were studied under three pH conditions (pH 7.4, 6.0, and 5.0) to assess the pH-responsive release behavior of cRGD-Dex-DOX/HDZ NPs (Figure 2A,  B).At pH 7.4, only ∼34.7% and ∼34.4% of the total DOX and HDZ were released from cRGD-Dex-DOX/HDZ NPs after 72 h of incubation, demonstrating that cRGD-Dex-DOX/HDZ NPs could largely avoid the release of DOX and HDZ from NPs in the blood circulation.The DOX and HDZ release percentage increased to ∼59.5% and ∼64.2%, respectively, as pH decreased to 6.0.Astonishingly, the DOX and HDZ release percentage can hit ∼91.6% and ∼88.7% after 72 h of incubation when the pH value was further decreased to 5.0.The hydrazone group linker is rapidly broken down in acid media, which is responsible for the controlled release of DOX. [28,29]The controlled release of HDZ could be owed to its protonation and reduced hydrophobic interactions among HDZ molecules and between HDZ and DOX moieties.Furthermore, these drug release behaviors were corroborated by monitoring the particle size changes of cRGD-Dex-DOX/HDZ NPs using DLS (Figure S6).The intelligent controlled release of DOX and HDZ in acidic environments resembling the tumor environment suggests that this system may have excellent pH responsiveness in tumors.
To further study the pH effects on the aggregations of DOX and HDZ, we have performed 100 ns all-atom molecular dynamics simulation of a first system composed of neutral DOX and HDZ molecules (pH = 7.4) and a second system composed of protonated DOX and HDZ molecules (pH = 5.0) (see System and Methods).As is shown in Figures 2C and D, we noticed the aggregations between neutral DOX and HDZ (pH = 7.4) during the simulation of the first system, while the majority of protonated HDZ molecules remain in the solutions during the simulation of the second system.The contact number between HDZ and DOX moieties was calculated over the last 50 ns simulations where a contact was counted if any two heavy atoms are within 0.4 nm.The average number of HDZ in contact with DOX molecules is about 8.9 at pH = 7.4, while the average number of HDZ in contact with DOX molecules at pH = 5.0 is about 4.8, which is significantly lower (Figure S8).Our MD simulation results thus suggest that the release of HDZ from the hydrophobic core of nanoparticles under low pH conditions should be attributed to the electrostatic repulsions among the protonated HDZ as well as electrostatic repulsions between the protonated HDZ and protonated DOX.These simulations give a first explanation for the controlled release of HDZ from cRGD-Dex-DOX/HDZ NPs at lower pH conditions (i.e., the pH responsiveness).

Cellular uptake and cytotoxicity assay
We studied the cellular uptake and cellular drug release behaviors of cRGD-Dex-DOX/HDZ NPs and Dex-DOX NPs using inverted fluorescence microscopy with the free DOX treatment group as the control.The nuclei of 4T1 cells were stained with Hoechst (blue), and DOX showed red fluorescence.Figure 3A shows that in the first 2 h, the DOX fluorescence in the Dex-DOX NPs and cRGD-Dex-DOX/HDZ NPs treatment groups was distributed in the cytoplasm, indicating that 4T1 cells can quickly uptake both Dex-DOX NPs and cRGD-Dex-DOX/HDZ NPs.However, the weak DOX fluorescence in the nuclei suggests that most of the DOX had not been released from the NPs.Indeed, after 6 h the DOX fluorescence was remarkably intensified in the nuclei of 4T1 cells, suggesting that a large number of DOX was released from NPs and sequentially entered the nucleus.The observed intracellular drug release dynamics were in a way consistent with that in the in vitro drug release tests (Figure 2A).The rapid intracellular release of DOX from carriers is crucial for promoting the therapeutic effects of the system.Notably, the intensities of DOX fluorescence measured at 2 and 6 h in the cRGD-Dex-DOX/HDZ NPs treatment group were higher than those in the Dex-DOX NPs treatment group, which could be attributed to the coupling effects of the cRGD peptide.This peptide could specifically target integrin αvβ3, which is overexpressed on the surface of 4T1 tumor cells, and enhance the internalization of cRGD-Dex-DOX/HDZ NPs.As expected, free DOX can be rapidly taken up by 4T1 cells and then enter and accumulate in the nucleus rather than in the cytoplasm after 6 h of co-incubation, obeying the typical cellular uptake characteristic of free DOX.These results were further confirmed by the results from flow cytometry analysis (Figure S9).Next, we examined the cytotoxicity of Dex-DOX NPs, cRGD-Dex-DOX NPs, and cRGD-Dex-DOX/HDZ NPs to 4T1 cells and human umbilical vein endothelial cells (HUVECs) with free DOX as the control.As expected, free DOX elicited a much higher cell viability loss to HUVEC and 4T1 cells than the NPs (Figures 3B,C).In general, cRGD-Dex-DOX NPs and cRGD-Dex-DOX/HDZ NPs almost had the same cytotoxicity level against 4T1 cells and higher cytotoxicity than Dex-DOX NPs due to the fact that the cRGD peptide on the surface of cRGD-Dex-DOX NPs and cRGD-Dex-DOX/HDZ NPs enhanced the enrichment of cRGD-Dex-DOX NPs in 4T1 cells.Not surprisingly, Dex-DOX NPs, cRGD-Dex-DOX NPs, and cRGD-Dex-DOX/HDZ NPs showed comparable cytotoxicity to HUVECs, because HUVECs cannot overexpress integrin αvβ3, leading to the disappearance of the active targeting function of the cRGD peptide.In addition, the cytotoxicity of free HDZ toward HUVEC and 4T1 cells was also measured.The data demonstrated that HDZ had excellent cytocompatibility with both cell lines.The cell viability of the two cell lines was still higher than 70% even as the treatment concentration was increased to 10 mg L −1 (Figure S10).This factor could be the main reason why cRGD-Dex-DOX NPs and cRGD-Dex-DOX/HDZ NPs had similar cytotoxicity.

Blood retention time and biodistribution
Free DOX and cRGD-Dex-DOX NPs were injected into BALB/C mice through the tail vein at the same DOX dose of 7.5 mg kg −1 to evaluate whether cRGD-Dex-DOX NPs could have a longer blood circulation time than free DOX.The results clearly show that cRGD-Dex-DOX NPs had a longer blood retention time than free DOX, increasing the accumulation of the drug in tumors (Figure 4A).Furthermore, we examined the time-dependent bio-distributions of free DOX, Dex-DOX NPs, and cRGD-Dex-DOX NPs.4T1 tumor-bearing mice were sacrificed at 24 and 48 h after the intravenous injection of the different drugs.Then, some important organs were removed for tissue grinding and extraction for fluorescence analysis.Overall, the enrichment rates of Dex-DOX-NPs and cRGD-Dex-DOX NPs in various organs and tumors were higher than that of free DOX at both time points (Figure 4B).At 24 h, the enrichment rates of Dex-DOX NPs and cRGD-Dex-DOX NPs in tumors were ∼1.5and ∼1.8-fold higher than that of free DOX, respectively.The enrichment rates of Dex-DOX NPs and cRGD-Dex-DOX NPs further increased to ∼1.9 and ∼2.2 folds compared with that of free DOX, as the circulation time was extended to 48 h.Apparently, cRGD-Dex-DOX NPs had a higher tumor enrichment rate than Dex-DOX NPs at both monitoring points, which agrees very well with the cellular uptake results (Figures 3A,B).Therefore, the in vitro and in vivo results demonstrated that the cRGD peptide enhances the tumor-targeting capability of cRGD-Dex-DOX NPs, leading to their great potential as an intelligent nanocarrier for the effective treatment of tumors.

Tumor microenvironment remodeling and cRGD-Dex-DOX/HDZ NPs penetration
We examined whether cRGD-Dex-DOX/HDZ NPs could further regulate tumor microenvironments (e.g., reduce the hypoxia level or lower the expression of intratumoral ECM markers (α-SMA and tumor-associated fibroblasts [TAFs] generate fibronectin) by relieving the pathological status of tumor vessels.32] We examined the changes in tumor HIF-1α expression levels in mice after the treatment with the three different regimens (Figure 6A).As shown in Figure 6A, cRGD-Dex-DOX/HDZ NPs treatment can more strongly decrease the tumor hypoxic level compared with the other two treatments.For instance, the hypoxic positive region in the cRGD-Dex-DOX/HDZ NPs treatment group was only ∼41.7%, which was much smaller than those in the cRGD-Dex-DOX NPs (∼71.1%) and cRGD-Dex-DOX NPs + HDZ (∼62.0%)groups.In addition, the fibronectin-positive area was only ∼4.2% in the cRGD-Dex-DOX/HDZ NPs treatment group but ∼6.6% and ∼11.4% in the cRGD-Dex-DOX NPs and cRGD-Dex-DOX NPs + HDZ treatment groups, respectively (Figure 6B).A similar decrease trend in α-SMA expression was observed.The α-SMA-positive area (green) in the cRGD-Dex-DOX/HDZ NPs group was ∼3.7%, which was also remarkably lower than those in the cRGD-Dex-DOX NPs + HDZ (∼7.9%) and cRGD-Dex-DOX NPs (∼9.6%) groups (Figure 6C).These data undoubtedly demonstrate that cRGD-Dex-DOX/HDZ NPs can more effectively reduce the pathosis of tumor microenvironments.
We further probed the penetration and distribution of DOX (red) in tumors (Figure 6D).In the cRGD-Dex-DOX NPs and cRGD-Dex-DOX NPs + HDZ groups, DOX was mainly distributed near blood vessels (green), suggesting the relatively poor penetration of DOX under these treatments.In sharp contrast, the DOX fluorescence in the cRGD-Dex-DOX/HDZ NPs group diffused to much larger areas far from the blood vessels.Quantitatively, the DOX-positive area in the cRGD-Dex-DOX/HDZ NPs group was ∼1.9%, which was about ∼2.4and ∼2.7-fold higher than those in the cRGD-Dex-DOX NPs + HDZ (∼0.8%) and cRGD-Dex-DOX NPs (∼0.7%) groups.Hence, cRGD-Dex-DOX/HDZ NPs can remarkably improve NP penetration in tumor tissues.Likewise, cRGD-Dex-DOX/HDZ NPs can also notably enhance the infiltration of CD3 + T cells, CD4 + T cells, and CD8 + T cells (Figure 6E and Figure S12).For instance, the CD3 + T cell-positive area was only about ∼12.1% in the cRGD-Dex-DOX NPs group.Meanwhile, cRGD-Dex-DOX NPs + HDZ treatment could mildly enhance the infiltration of CD3 + T cells (∼20.5%), but its effect was still much weaker than that of cRGD-Dex-DOX/HDZ NPs (∼48.5%).These results clearly show that cRGD-Dex-DOX/HDZ NPs can more effectively improve tumor vasculatures and remark-ably downregulated the expression levels of HIF-1α, α-SMA, and fibronectin, which ultimately enhanced the permeability of DOX and the infiltration of T cells in tumors.
Free DOX treatment only showed a very mild tumor inhibition effect (∼17.4%) compared with the PBS control group (group 1).However, they can cause severe toxic effects in mice (i.e., body weight loss).Compared with free DOX, Dex-DOX NPs demonstrated a slightly higher tumorinhibiting effect (∼28.7%) and improved biocompatibility (i.e., inducing body weight increase) due to the enrichment of Dex-DOX NPs at tumor sites and the sustained release of DOX from Dex-DOX NPs.As expected, cRGD-Dex-DOX NPs had a better tumor suppression effect (∼35.3%) than Dex-DOX NPs, because the modified targeting peptide cRGD on the surface of cRGD-Dex-DOX NPs could enhance the active enrichment of NPs in tumors.cRGD-Dex-DOX NPs + HDZ treatment further improved the antitumor efficacy (∼46.3%)compared with cRGD-Dex-DOX NPs, because free HDZ can promote the tumor accumulation of cRGD-Dex-DOX NPs.Most striking was the remarkable antitumor efficiency (∼65.8%) and very moderate side effect (i.e., body weight increase steadily) of cRGD-Dex-DOX/HDZ NPs.These results were further confirmed by observations of tumor sections (Figure 7E).The remarkable antitumor advantages of the cRGD-Dex-DOX/HDZ NPs treatment may be attributed to the ability of cRGD-Dex-DOX/HDZ NPs to accurately deliver a large number of HDZ and DOX to the tumor site and the intelligently controlled release of HDZ in the mildly acidic environment of the tumor, which expanded tumor microvessels and enhanced the subsequent enrichment of NPs at the tumor site.The expansion of tumor microvessels reduced the expression levels of HIF-1α, α-SMA, and fibronectin and enhanced the permeability of DOX and the infiltration of CD3 + T cells in tumors.
Histological examination of tumor tissue sections demonstrated that the tumor cells in the control group had a large, spindle-shaped nucleus, whereas those in the cRGD-Dex-DOX/HDZ NPs treatment group shrunk most distinctly and underwent massive necrosis (Figure 7E).Moreover, spleen and liver tissue sections had no obvious pathological symptoms (Figure S13).These observations strongly indicated that cRGD-Dex-DOX/HDZ NPs had the most superior tumorsuppressive effect over all other tested regimens in vivo.
The systemic toxicity of cRGD-Dex-DOX/HDZ was further assessed through hemolysis, blood biochemistry, and routine blood tests.As shown in Figure S14, when the concentration of cRGD-Dex-DOX/HDZ NPs reached 200 μg mL −1 , its hemolysis ratio remained below 5%, indicating excellent erythrocyte compatibility.Further examinations revealed that free DOX led to significant decreases in white blood cell (WBC), lymphocyte (LYMPH), and platelet (PLT) counts while increasing neutrophil (NEUT) counts in mice (Figure S15 and Table S2), underscoring the severe immunosuppressive effects of free DOX.Additionally, free DOX elevated the levels of aspartate aminotransferase (AST), urea nitrogen (BUN), and alanine aminotransferase (ALT) in the blood of mice, indicative of acute inflammation and injury in the liver and kidneys.In contrast, cRGD-Dex-DOX/HDZ had minimal impact on these functions, suggesting a significant reduction in the systemic toxicity associated with free DOX.

CONCLUSIONS
Development of an effective and safe protocol to enhance the penetration of nanomedicines in tumors is a prerequisite to ensure that chemotherapeutic agents are effective in treating tumors.In this study, we successfully design and synthesize a pH-responsive cRGD peptide-modified dextran-g-doxorubicin (cRGD-Dex-DOX) prodrug.The prodrug could self-assemble with HDZ in aqueous solution to form a uniform and greatly stable NP system with a particle size of 130 nm.In vitro cellular assays revealed that the NPs modified with cRGD could cooperatively deliver antitumor drugs to tumor tissue and achieve "spatiotemporally pinpointed" drug release intracellularly with minimal offtarget effects on normal cells.Moreover, the system has a pH-responsive drug release behavior leading to the controlled release of DOX in a weakly acidic environment similar to tumor tissue.The biodistribution studies further confirmed that cRGD-Dex-DOX NPs substantially enhanced the drug accumulation rate up to 2.2-fold higher than that of free drugs at the tumor sites.After two rounds of treatments, in vivo accumulation assays demonstrated that HDZ encapsulated with cRGD-Dex-DOX/HDZ NPs could effectively enhance drug accumulation (∼1.4 folds) at the tumor sites, because HDZ could be effectively released to tumor sites, leading to tumor vasodilation.The improved vascularity reduced the level of tumor hypoxia and decreased the intratumor ECM markers α-SMA and fibronectin.These changes effectively improved the permeability of NPs (∼1.9%) and the infiltration of CD3 + T cell infiltration (∼48.5%) in tumors, thus enhancing the antitumor efficacy of the treatment.The tumor inhibitory rate of this treatment was found to be much higher (∼65.8%)than those of other groups.Therefore, our drug delivery system may provide an effective and easy way to modulate the abnormal tumor microenvironment, enhancing drug penetration and T-cell infiltration in the tumor that leads to a significant improvement in antitumor efficacy.

Synthesis of dextran propargyl carbonate (Dex-C≡C)
Dex-C≡C can be synthesized in a very simple and effective way.Briefly, 4-hydroxyl-1-butyne (0.25 g, 3.60 mmol) in 2 mL of dry CH 2 Cl 2 was added dropwise to carbonyldiimidazole (CDI) in 4 mL of dry CH 2 Cl 2 and further reacted for another 1 h at room temperature.The reaction solution was leached three times with saturated sodium chloride solution to remove the excess CDI.The collected organic layer was dried overnight by anhydrous Na 2 SO 4 and evaporated to gain a powder.The powder was dried under vacuum at room temperature for 24 h to obtain the pure product.
In the second step, 0.317 g alkynyl butyl carbonyl imidazole (1.93 mmol) and 2 g dextran (Dex, 0.4 mmol) were dissolved in 8 and 20 mL of dry DMSO, respectively.Alkynyl butyl carbonyl imidazole was added dropwise to Dex and further reacted at room temperature under N 2 atmosphere for 24 h.The product solution was dialyzed in deionized water for 3 days to remove the unreacted alkynyl butyl carbonyl imidazole and DMSO by cellulose tubular membrane (MWCO 2000).Then, the solution was lyophilized to obtain Dex-C≡C (1.464 g, yield: 63.2%).

4.3
Syntheses of cRGD-Dex-DOX and Dex-DOX cRGD-Dex-DOX was prepared via the CuAAC "click" reaction.First, 9.79 mg CuBr (0.0682 mmol) and 23.66 mg PMDETA (0.1365 mmol) were added into 6 mL of DMSO for 10 min in nitrogen atmosphere.Second, 200 mg Dex-C≡C (0.03746 mmol) and 20 mg cRGD-PEG 2000 -N 3 (0.0149 mmol) were added, separately.The mixture solution was stirred at room temperature for 8 h.Third, 50 mg DOXhyd-N 3 (0.0683 mmol) was added and stirred for another 8 h.After the reaction, the mixture solution was dialyzed against DMSO for 12 h to remove the unreacted DOX-hyd-N 3 and cRGD-PEG 2000 -N 3 and then dialyzed against Milli-Q water for 48 h to remove DMSO.Finally, cRGD-Dex-DOX was obtained by freeze-drying the last mixture solution (0.232 g, yield: 85.4%).Dex-DOX was synthesized in the same way as cRGD-Dex-DOX, except that cRGD-PEG 2000 -N 3 was not added in the second step.

4.4
The bicinchoninic acid method First, we procured the BCA Protein Quantification Kit, which contains BCA Reagent A, BCA Reagent B, and a protein standard (BSA).The BCA working solution was prepared by adding 1 volume of BCA Reagent B to 50 volumes of BCA Reagent A (A:B = 50:1).The components were thoroughly mixed.We added 100 μL of different concentrations of BSA and the samples to be tested into separate reaction tubes.Next, we added 2.0 mL of the BCA working solution to each tube and mixed the contents thoroughly.The tubes were then placed in an incubator set at 37 • C for 30 min.This incubation allowed the formation of a colored complex in proportion to the amount of protein present.After incubation, we removed 200 μL of solution from the tubes and detected the absorbance using a spectrophotometer set at 562 nm.The absorbance of the BSA standard (ODtest) was obtained.To calculate the final reading, we subtracted the OD of the blank well from ODtest.A standard curve was plotted based on the final reading and the standard protein concentration.Using the standard curve and considering the sample dilution, we calculated the protein concentration of the samples under investigation.This method allowed us to accurately quantify the protein content in our samples and was an essential step in our experimental procedures.

Coarse-grained and full-atom system models
The coarse-grained model cRGD-Dex-DOX was built using the Martini 3 force field. [33]Initially, the cRGD model was created using Pymol, and then the coarse-grained parameters were generated with the assistance of the vermouth python library. [34]the dextran and polyethylene glycol (PEG) chains were built using the Polyply software, [35] a Python package designed to facilitate the model construction of polymer chains.Meanwhile, the DOX, triazole moieties and dihydralazine are based on the similar structures of tetracene, imidazole and 1-methoxynaphthalene, respectively, in the Martini small molecules dataset. [36]The details of the coarse-grained parametrization of Dex-DOX, dihydralazine and cRGD-Dex are shown in Figure S7.Consistently with the average weight of dextran used in the experiment, we adopted a dextran chain with 31 monomers.On average, the cRGD-Dex-DOX chain consists of two parts: one Dex-DOX moiety and one Dex-cRGD moiety.The cRGD-PEG and DOX moieties were linked to the 10th and 23rd Dex monomer, respectively.The force field parameters for cRGD-Dex-DOX and HDZ are available at https://github.com/huangjianxiangzju/HDZ-DOX-parameters/tree/main/ Coarsed-grained_Molecular_Dynamics_Simulations.
To build the initial system for the coarse-grained MD simulations, 100 cRGD-Dex-DOX chains and 500 dihydralazine molecules were inserted randomly in a box with the size of 30 × 30 × 30 nm 3 .The simulation box was solvated by water and additional chloride ions were further added to neutralize the systems.
For the all-atom model of HDZ, DOX, HDZ 2+ and DOX + molecules, we used OPLS-AA/1.14*CM1Aforce field and the force field parameters were obtained through the LigParGen web server. [37]In particular, the CM1A charges of neutral molecules (HDZ and DOX) are scaled by a factor of 1.14 as suggested in. [37]Snapshots of HDZ, DOX, HDZ 2+ and DOX + are displayed in Figure S4.These all-atom models were used to build two systems (HDZ/DOX and HDZ 2+ /DOX + ).The first system includes 10 HDZ and 10 DOX molecules inserted into a simulation box with the size of 64.5 × 64.5 × 64.5 nm 3 , which was further solvated by water.The second system consists of 10 HDZ 2+ and 10 DOX + molecules inserted into a simulation box with the size of 46.5 × 46.5 × 46.5 nm 3 , which was further solvated by water and neutralized by chloride anions.The force field parameters for HDZ, DOX, HDZ 2+ , and DOX + molecules are available at https://github.com/huangjianxiangzju/HDZ-DOX-parameters/tree/main/All Atom_Molecular_Dynamics_Simulations.

4.7
Simulation protocol and structural analysis CGMD simulations were carried out using GROMACS 2020.6 [38] in three replicas, and simulation snapshots were rendered with VMD. [39,40]The temperature (T = 310 K) and pressure (P = 1 atm) were maintained using a stochastic velocity rescaling thermostat [41] and Parrinello-Rahman barostat [42] with coupling parameters of 1.0 and 12 ps −1 , respectively.Periodic boundary conditions were applied to all the systems in three directions.All three replicas of CGMD simulations were simulated for 3000 ns with a time step of 30 fs.Short-range electrostatic and van der Waals interactions were calculated with a cut-off distance of 1.1 nm.The long-range electrostatic interactions were treated using reaction-field potential. [43]imilarly, all the atomistic MD simulations were performed by using the GROMACS 2020.6 package. [38]The atomistic MD simulations of HDZ/DOX and HDZ 2+ /DOX + with the OPLS force field were conducted with a cut-off of 1.2 nm for the van der Waals interactions and electrostatic interactions.While the long-range electrostatics were treated with the particle mesh ewald method. [44]Energy minimizations were conducted with the steepest descent algorithm and MD simulations were run with the default leap-frog integrator using a time-step of 2 fs for the integration in the isothermalisobaric ensemble ensemble.The temperature (T = 300 K) and pressure (P = 1 atm) were maintained using a stochastic velocity rescaling thermostat [41] and Parrinello-Rahman barostat [42] , respectively.Both the atomistic MD simulations of HDZ/DOX and HDZ 2+ /DOX + last for 100 ns.

In vitro drug release
We studied the controlled release behavior of cRGD-Dex-DOX/HDZ NPs (0.6 mg mL −1 ) in PBS with different pH (7.4,6.0, and 5.0).cRGD-Dex-DOX/HDZ NPs solution (5 mL) was transferred into a dialysis membrane (MWCO: 12,000-14,000) and placed in a centrifuge tube containing 20 mL of the different medium buffers as the dialysate.
The centrifuge tube was placed in a shaker at 37 • C and shaken continuously.At predetermined time points, 5 mL of dialysate was removed, and an equal volume of corresponding fresh buffer solution was added.The released DOX concentration was determined by a fluorescence spectrophotometer (Cary Eclipse, Agilent Technologies) at the excitation wavelength of 488 nm, emission wavelength of 560 nm, and slit width of 10 nm.All release experiments were conducted in three repeat groups, and the results were expressed as mean with standard deviation.

Pharmacokinetic study in vivo
BALB/c mice were intravenously injected with free DOX or cRGD-Dex-DOX/HDZ NPs solution.Mouse blood (10 μL) was collected at intervals through the eyeball and placed in a 1.5 mL centrifuge tube.Methanol extraction solvent (1 mL) and trifluoroacetic acid solution (50 μL) were added to the tube.After 5 min of ultrasound, the samples were incubated at −20 • C for 12 h, and the solution were centrifuged at 12,000 RPM for 5 min.The fluorescence of the supernatant solution was measured using a microplate reader (SYNERGY NEO) with an excitation wavelength of 488 nm and an emission wavelength of 560 nm.

Biodistribution and organ fluorescence measuring
When the 4T1 tumor volume reached approximately 300 mm 3 , BALB/c mice were intravenously injected with free DOX, Dex-DOX NPs, and cRGD-Dex-DOX NPs solution (7.5 mg kg −1 on DOX basis).Various tissues (heart, liver, spleen, lung, kidney, and tumor) from mice were extracted and observed using a 3D imaging instrument (IVIS Spectrum) at 24 and 48 h.The excitation wavelength was set to 488 nm within a wavelength range of 460-500 nm, while the emission wavelength was set to 560 nm within a wavelength range of 540-600 nm.Afterward, various tissues were weighed and ground.Methanol (1 mL) was added to the grinding solution as the extraction solution of DOX.The solution was incubated at −20 • C for 12 h.Finally, the solution was centrifuged at 12000 rpm for 5 min, and the fluorescence of the supernatant solution was measured by a microplate reader (SpectraMax) with an excitation wavelength of 480 nm and an emission wavelength of 560 nm.

In vivo antitumor efficacy
Mice bearing 4T1 tumor (∼300 mm 3 ) were randomly divided into six groups for the different treatment strategies.Tumor volume (V) and body weight of mice were measured by vernier calipers and electronic balance every other day.V was calculated using Equation 2 where a and b are the major and minor axes of the tumor, respectively.Mice were sacrificed on the 12th day after mice were injected with drugs.The tumor inhibitory rate was calculated on the last day using Equation 3.
Tumor inhibitory rate = Vcon − Vtest Vcon * 100%, where Vcon and Vtest are the tumor volumes of the control group and other text groups, respectively.The antitumor effects in each group were assessed using tissue section analysis via hematoxylin and eosin staining.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no competing financial interest.

F I G U R E 1
Transmission electron microscopy (TEM) and dynamic light scattering (DLS) images of cRGD-Dex-DOX nanoparticle (NP) (A) and cRGD-Dex-DOX/HDZ NPs (B).Scale bar = 200 nm.(C) Zeta potential of three kinds of NPs.(D) Size change of cRGD-Dex-DOX/HDZ NPs in three different media: Milli-Q water, PBS, and FBS (10%).(E and F) Initial and final snapshots of the simulated self-assembly of cRGD-Dex-DOX and HDZ.The DOX, cRGD, and HDZ are shown in red, yellow and green spheres, respectively, while the dextran and polyethylene glycol (PEG) chains are shown in purple and cyan sticks, respectively.(G) Time evolution of the average cluster sizes of the three replicas coarse grained molecular dynamics (CGMD) simulations.The clusters were calculated by counting the contacted DOX moieties using a contact cutoff with a center-of-mass distance of 0.7 nm.(H) Radial distribution functions (RDF) of DOX moieties and HDZ molecules with respect to every DOX moiety in the last frame of the simulation.(I) The representative cluster structure of cRGD-Dex-DOX/HDZ and the parallel packing of HDZ (shown in green) to DOX moieties (shown in red) in a closer view.

F I G U R E 2
In vitro DOX (A) and HDZ (B) release curves of cRGD-Dex-DOX/HDZ nanoparticle (NP) at pH 5.0, 6.0, and 7.4.Initial (left) and final (right) system snapshots of the MD simulation of the neutral HDZ and DOX molecules (C), and the protonated HDZ 2+ and DOX + molecules (D).

F
I G U R E 3 (A) Inverted fluorescence microscopy images of 4T1 cells incubated with Dex-DOX nanoparticle (NP) and cRGD-Dex-DOX/HDZ NPs for 2 and 6 h (6-h free DOX data were also included for comparison).For each panel, the images from left to right show the cell nuclei stained by Hoechst 33342 (blue), DOX fluorescence in cells (red), and overlays of the blue and red images.Each scale bar is 100 μm.Cell viability of free DOX, Dex-DOX NPs, cRGD-Dex-DOX NPs, and cRGD-Dex-DOX/HDZ NPs incubated with 4T1 cells (B) and human umbilical vein endothelial cell (HUVEC) cells (C) at 48 h.
In the next analysis, we evaluate the effect of HDZ encapsulated in cRGD-Dex-DOX/HDZ NPs on the tumor enrichment of NPs via vasodilation modulation.Here, mice were treated with three different therapeutic regimens: (a) cRGD-Dex-DOX NPs, (b) cRGD-Dex-DOX NPs + HDZ, and (c) cRGD-Dex-DOX/HDZ NPs.In all three groups, cRGD-Dex-DOX NPs and cRGD-Dex-DOX/HDZ NPs were intravenously injected into 4T1 tumor-bearing mice at the DOX dose of 7.5 mg kg −1 .In the cRGD-Dex-DOX NPs + HDZ group, HDZ was intraperitoneally injected into mice at the dose of 6.9 mg kg −1 to identify the influence of the HDZ administration route on the tumor enrichment of NPs.Since the in vitro drug release curve of cRGD-Dex-DOX/HDZ NPs suggests that the drug release dynamics converged fairly after 48 h, the mice received two rounds of treatments on days 0 and 2 (Figure5A).Two rounds of treatments can better exploit the vasodilation function of HDZ loaded in cRGD-Dex-DOX/HDZ NPs.On days 3 and 4, the tissues of mice were removed and monitored by a 3D imaging instrument (IVIS Spectrum) to explore the biodistributions of NPs at 24 and 48 h after the second round of injection.At 24 h, cRGD-Dex-DOX NPs + HDZ treatment only resulted in a very mild enhancement in DOX fluorescence intensity at tumor sites compared with cRGD-Dex-DOX NPs (Figure5B).Impressively, cRGD-Dex-DOX/HDZ NPs could remarkably intensify the DOX fluorescence intensity by ∼1.3 and ∼1.2 folds compared with those of cRGD-Dex-DOX NPs and cRGD-Dex-DOX NPs + HDZ, respectively.At 48 h, the tumor DOX enrichment rate in the cRGD-Dex-DOX NPs + HDZ group increased by ∼1.2 folds than that in the cRGD-Dex-DOX NPs group, pronouncing that HDZ administrated via intraperitoneal injection could help increase the tumor enrichment of NPs.Moreover, the fluorescence intensity in the cRGD-Dex-DOX/HDZ NPs group had a more remarkable increase, which was ∼1.4and ∼1.3-fold higher than those of the cRGD-Dex-DOX NPs and cRGD-Dex-DOX NPs + HDZ group, respectively.These results were further verified by the fluorescence assays of ground mouse organs and tissues (FigureS11).Therefore, these results show that HDZ encapsulated in cRGD-Dex-DOX/HDZ NPs have much higher efficacy in promoting the tumor enrichment of cRGD-Dex-DOX/HDZ NPs than free HDZ.In order to understand the underlying mechanisms for the higher tumor enrichment of cRGD-Dex-DOX/HDZ NPs, we investigate the status of the blood vessels by CD31 immunofluorescence staining.As shown in Figure5C(green), the tumor blood vessels in the cRGD-Dex-DOX NPs + HDZ group are slightly dilated compared with that in the cRGD-Dex-DOX NPs group.In sharp contrast, cRGD-Dex-DOX/HDZ NPs treatment considerably improved the status of tumor blood vessels, making them considerably more dilated than those in the cRGD-Dex-DOX NPs and cRGD-Dex-DOX NPs + HDZ groups.Quantitatively, the cRGD-Dex-DOX/HDZ NPs treatment group had a higher percentage of vessel area (∼5.1%) than the other two groups (∼4.1% and ∼3.4% for cRGD-Dex-DOX NPs + HDZ and cRGD-Dex-DOX NPs, respectively), suggesting that cRGD-Dex-DOX/HDZ NPs can expand more effectively the microvessels of tumors, facilitating the further accumulation of NPs.
Dex-DOX (cRGD-Dex-DOX) can self-assemble in Milli-Q water to form Dex-DOX NPs (cRGD-Dex-DOX NPs).We dissolved 16 mg Dex-DOX or cRGD-Dex-DOX in 2 mL of Milli-Q water and then sonicated it for 10 min with moderate stirring.The mixture solution was stirred for another 4 h.The amphiphilic prodrug cRGD-Dex-DOX and HDZ can self-assemble in Milli-Q water.Briefly, 16 mg cRGD-Dex-DOX and 1.6 mg HDZ were dissolved in 2 mL of Milli-Q water, followed by 10 min of ultrasonication under moderate stirring.Finally, the HDZ-loaded nanoparticle (cRGD-Dex-DOX/HDZ NPs) mixture solution was stirred for another 4 h.The morphologies of cRGD-Dex-DOX/HDZ NPs were measured by TEM (HT7700, Hitachi).The TEM sample was prepared by the freeze-drying method.The average particle size and PDIs of cRGD-Dex-DOX/HDZ NPs in Milli-Q water were measured by a DLS instrument (Zetasizer Nano ZS, Malvern) at 25 • C. The DOX loading content (DLC) was calculated using equation (

A
U T H O R C O N T R I B U T I O N S R.Z. conceived and designed the research.L.Z. synthesized the title material and performed characterization.L.Z., P.H., Z.Y. performed the experiments.J.X. carried out molecular dynamics simulations.L.Z., D.B., and R.Z. co-wrote the paper.All authors discussed and commented on the manuscript.A C K N O W L E D G M E N T S This work was partially supported by the National Key R&D Program of China (grant numbers: 2021YFA1201200 and 2021YFF1200404), the National Natural Science Foundation of China (grant numbers: U1967217 and 22176137), the National Independent Innovation Demonstration Zone Shanghai Zhangjiang Major Projects (grant number: ZJZX2020014), the National Center of Technology Innovation for Biopharmaceuticals (grant number: NCTIB2022HS02010), Shanghai Artificial Intelligence Lab (grant number: P22KN00272), the Starry Night Science Fund of Zhejiang University Shanghai Institute for Advanced Study (grant number: SN-ZJU-SIAS-003), The Natural Science Foundation of the Jiangsu Higher Education Institutions of China (grant number: 20KJA150010), and the Natural Science Foundation of Zhejiang Province (grant number: 2022LQ22H220001).