Depletion of Mannose Receptor–Positive Tumor-associated Macrophages via a Peptide-targeted Star-shaped Polyglutamate Inhibits Breast Cancer Progression in Mice

Although many studies have explored the depletion of tumor-associated macrophages (TAM) as a therapeutic strategy for solid tumors, currently available compounds suffer from poor efficacy and dose-limiting side effects. Here, we developed a novel TAM-depleting agent (“OximUNO”) that specifically targets CD206+ TAMs and demonstrated efficacy in a triple-negative breast cancer (TNBC) mouse model. OximUNO comprises a star-shaped polyglutamate (St-PGA) decorated with the CD206-targeting peptide mUNO that carries the chemotherapeutic drug doxorubicin (DOX). In the TNBC model, a fluorescently labeled mUNO-decorated St-PGA homed to CD206+ TAMs within primary lesions and metastases. OximUNO exhibited no acute liver or kidney toxicity in vivo. Treatment with OximUNO reduced the progression of primary tumor lesions and pulmonary metastases, significantly diminished the number of CD206+ TAMs and increased the CD8/FOXP3 expression ratio (indicating immunomodulation). Our findings suggest the potential benefit of OximUNO as a TAM-depleting agent for TNBC treatment. Importantly, our studies also represent a novel design of a peptide-targeted St-PGA as a targeted therapeutic nanoconjugate. Significance: A peptide-targeted nanoformulation of DOX exclusively eliminates mannose receptor+ TAMs in breast cancer models, generating response without off-target effects (a drawback of many TAM-depleting agents under clinical study).


Introduction
Triple-negative breast cancer (TNBC), defined by the lack of the expression of the estrogen receptor, progesterone receptor, and HER2 (1,2), represents an aggressive breast cancer subtype with poor prognosis (3) that comprises up to 20% of all breast cancer cases (3,4). Interfering with immune checkpoints signaling [e.g., through the modulation of programmed cell death 1 (PD-1) and its ligand (PD-L1)] represents an alternative treatment strategy for several cancers and is currently being employed in combination with chemotherapy as a colitis; refs. 10,14,15). Other immune checkpoint inhibitors (ICI), including the CTL-associated antigen 4 (CTLA-4) blockers ipilimumab and tremelimumab, are currently under evaluation for TNBC treatment in combination with other drugs (clinical trial identifiers: NCT03606967, NCT02983045); however, anti-CTLA-4 treatments induce severe side effects such as endocrinopathies, myopathy, enterocolitis, and hepatitis (16)(17)(18)(19), which narrow their use. Overall, the limited success of alternative treatment options for TNBC has maintained chemotherapy as the standard of care for most patients (20).
The anthracycline drug doxorubicin (DOX), which presents high offtarget effects such as cardiotoxicity (21,22), represents a frequently employed chemotherapeutic for TNBC; however, disease relapse and metastatic development have also been associated with DOX treatment (23). M2 (antiinflammatory)-polarized tumor-associated macrophages (TAM; ref. 24) found within both primary and metastatic tumor lesions mediate both events (25); furthermore, TAMs represent the main executioners of tumor progression, immunosuppression, and invasion (24)(25)(26)(27)(28)(29), and their presence correlates with inadequate therapeutic response and poor prognosis (25). Recent efforts have focused on eliminating TAMs, and several ongoing clinical trials are currently evaluating TAM depletion in combination with treatments such as ICIs (30). The current clinical-stage gold standard for TAM depletion relies on agents that block colony stimulating factor 1 (CSF1) or its receptor CSF1R, such as the smallmolecule CSF1R inhibitor PLX3397 (31); however, microglia also expresses CSF1R (32), the inhibition of CSF1R with PLX5622 impacts M1 macrophages (33), and PLX3397 treatment causes edema (34). Clinical data suggest that anti-CSF1R antibodies induce a modest effect (35,36) and cause severe side effects that include hematologic toxicities (35) and hepatotoxicity by targeting Kupffer cells (35,36). Overall, these findings highlight the overwhelming need for new TAM depletion strategies.
For the first time, we report the effects of depleting the CD206 + subpopulation of TAMs in a metastatic TNBC mouse model through the use of a targeting agent (the mUNO peptide) for a CD206 site different from the mannosebinding site (41)(42)(43)(44). Previous studies have employed mannose to target CD206; however, mannose has other receptors besides CD206 (45,46).
We decorated a three-arm branched biodegradable multivalent polyanion with a defined negative charge and nanometer-size hydrodynamic radius (star-shaped polyglutamate or St-PGA) with mUNO peptide to function as a targeted delivery platform for a chemotherapeutic agent (DOX) conjugated through a bioresponsive linker. St-PGA-DOX-mUNO (referred to as OximUNO) efficiently depleted CD206 + TAMs, relieved immunosuppression in the tumor microenvironment (TME) and limited metastasis/tumor growth, thereby supporting OximUNO as an alternative TAM depletion strategy.
Most importantly, this study represents the first described combination of two reported technologies-the St-PGA nanocarrier and the mUNO-targeting peptide. Overall, this OximUNO proof of concept demonstrates the potential of the peptide-targeted St-PGA nanosystem. Our studies lay a foundation for future work using this nanosystem to target other receptors efficiently by changing the targeting peptide.

Reagents and Solutions
The peptides mUNO (sequence: CSPGAK-COOH) and FAM-mUNO (FAM-Ahx-CSPGAK-COOH) were purchased from TAG Copenhagen and DOX from Sigma-Aldrich. St Mayer's hematoxylin solution was prepared by dissolving 5 g of aluminium potassium sulphate dodecahydrate (Merck Millipore, catalog no. 1010421000) in 100 mL of water, and adding 1 g of hematoxylin (Merck, catalog no. H9627). After complete dissolution, 0.02 g of sodium iodide (Merck, catalog no. 1065230100) was added and completely dissolved. Then, 2 mL of acetic acid (Sigma-Aldrich, catalog no. 33209) was added, and then the solution was boiled and then cooled. Once ready to use, the solution was filtered using a 0.45-μm filter.

Tumor Models
Two tumor models were used for homing studies: the orthotopic TNBC model, where 1 × 10 6 4T1 cells in 50 μL of PBS (Lonza, catalog no. 17-512F) were subcutaneously injected into the fourth mammary fat pad, and the experimental metastasis of TNBC model, where 5 × 10 5 4T1 cells in 100 μL of PBS were injected intravenously into Balb/c mice.
Two tumor models were used for treatment studies: the orthotopic TNBC model where 5 × 10 4 4T1 cells in 50 μL of PBS were injected subcutaneously into fourth mammary fat pad; and the experimental metastasis of TNBC model where 2 × 10 5 4T1-GFP cells in 100 μL of PBS were intravenously injected.

Nuclear Magnetic Resonance Spectroscopy
Nuclear magnetic resonance (NMR) spectra were recorded at 27°C (300 K) on a 300 Ultrashield from Bruker. Data were processed with Mestrenova software. Sample solutions were prepared at the desired concentration in D 2 O or D 2 O supplemented with NaHCO 3 (0.5 mol/L).

UV-visible Analysis
UV-visible (UV-Vis) measurements were performed using JASCO V-630 spectrophotometer at 25°C with 1-cm quartz cells and a spectral bandwidth of 0.5 nm. Spectra analysis was recorded three times in the range of 200-700 nm.

Fluorescence Analysis
Fluorescence analysis was performed using a JASCO FP-6500 spectrofluorimeter at 25°C with 1-cm quartz cells.

Dynamic Light Scattering
Size measurements were performed using a Malvern ZetasizerNano ZS instrument, supported by a 532 nm laser at a fixed scattering angle of 173°. Nanoconjugate solutions (0.1 mg/mL) were freshly prepared in PBS (10 mmol/L phosphate, 150 mmol/L NaCl), filtered through a 0.45-μm cellulose membrane filter, and measured. Size distribution was measured (diameter, nm) for each polymer in triplicate. Automatic optimization of beam focusing and attenuation was applied for each sample.

Zeta Potential Measurements
Zeta potential measurements were performed at 25°C using a Malvern Zetasiz-erNano ZS instrument, equipped with a 532 nm laser using disposable folded capillary cells, provided by Malvern Instruments Ltd. Nanoconjugate solutions (0.1 mg/mL) were freshly prepared in 1 mmol/L KCl. Solutions were filtered through a 0.45-μm cellulose membrane filter. Zeta potential was measured for each sample per triplicate.

Molecular Dynamics Simulations
Molecular dynamics (MD) simulations of PGA chains, and mUNO peptide were carried out using the ff19SB force field (47) in the Amber20 MD engine (https://sbgrid.org/software/titles/ambertools). The nanoconjugate system was neutralized using Na + ions and hydrated to account for a total of approximately 920,000 atoms (∼300,000 TIP3P water molecules) in a truncated octahedral box. A hydrogen mass repartitioning strategy was applied on the resulting topology, allowing us a 4 fs integration time step (48). Standard minimization and equilibration protocols were used to reach 300 K and 1 atm., followed by 50 ns of production MD run. The simulations were run under the NVT ensemble [constant number of particles, volume, and temperature through Berendsen thermostat (49)], considering periodic boundary conditions. The SHAKE algorithm was used to fix hydrogen atoms (50). The nonbound cut-off value was set to Angstrom. The central moiety was parameterized using the recommended protocol for the Amber force field. It was necessary to introduce amide bond, angle, and dihedral terms using the ParmEd module to establish the bond of the central molecule to the PGA chains.

Tumor Homing Studies
Tumors were induced as described in the tumor model section. Tumor homing studies were performed on mice bearing orthotopic TNBC or experimental metastasis of TNBC. Ten days post-induction (p.i) of the orthotopic TNBC or the experimental metastasis of TNBC model, mice were intraperitoneally injected with St-PGA-OG-mUNO (0.41 mg/0.5 mL of PBS) or St-PGA-OG (0.35 mg/0.5 mL of PBS; corresponding to 15 nanomoles of OG, absorbance measured by UV-Vis). The homing of a higher dose of St-PGA-PGA-mUNO (0.82 mg/0.5 mL of PBS) or St-PGA-OG (0.7 mg/0.5 mL of PBS; corresponding to 30 nanomoles of OG) was also analyzed and compared with the homing of FAM-mUNO (30 nanomoles/0.5 mL of PBS). In every case, nanoconjugates or free peptide were circulated for 6 hours, after which time, mice were sacrificed by anesthetic overdose followed by cervical dislocation. Organs and tumors were collected and fixed in cold 4% w/v paraformaldehyde (PFA) in PBS at +4°C for 24 hours, washed in PBS at room temperature for 1 hour and cryoprotected in 15% w/v sucrose (Sigma Life Science, catalog no. S9378) followed by 30% w/v sucrose at 4°C overnight. Cryoprotected and fixed tissues were frozen in optimal cutting temperature (OCT; Leica, catalog no. 14020108926), cryosectioned at 10-μm thickness on Superfrost+ slides (Thermo Fisher Scientific, catalog no. J1800AMNZ) and stored at −20°C. Immunofluorescence staining was performed as described earlier (42). OG was detected using rabbit anti-FITC/Oregon Green (dilution 1/100, Invitrogen by Thermo Fisher Scientific, catalog no. A889) and Alexa Fluor 647 goat anti-rabbit antibody (dilution 1/250, Invitrogen by Thermo Fisher Scientific, catalog no. A21245). CD206 was detected using rat anti-mouse CD206 (dilution 1/150, Bio-Rad, catalog no. MCA2235GA) and Alexa Fluor 546 goat anti-rat antibody (dilution 1/250, Life Technologies, catalog no. A11081). CD86 was detected using rat anti-mouse CD86 (dilution 1/100, BioLegend, catalog no. 105001) and Alexa Fluor 546 goat anti-rat secondary antibody (dilution 1/250). CD11c was detected using hamster anti-mouse CD11c antibody (dilution 1/75, BioLegend, catalog no. 117301) and Alexa Fluor 546 goat anti-hamster secondary antibody (dilution 1/200, Life Technologies, catalog no. A21111). CD31 was detected with rat anti-mouse CD31 (dilution 1/100, BD Biosciences, catalog no. 553370) as primary antibody and with Alexa Fluor 546 goat anti-rat (dilution 1/200, Invitrogen, catalog no. A11081) as secondary antibody. Slides were counterstained using 4 ,6-diamidino-2-phenylindole (DAPI, 1 μg/mL in PBS, Sigma-Aldrich, catalog no. D9542-5MG). Coverslips were mounted using mounting medium (Fluoromount-G Electron Microscopy Sciences, catalog no. 17984-25), and sections were imaged using Zeiss confocal microscope (Zeiss LSM-710) and 20× objective. The colocalization analysis between the FAM or OG channel and the CD206 channel was carried out using the "Coloc2" plugin in the Fiji program and selecting the "Pearson R value (no threshold)" coefficient. The colocalization values were obtained from at least three representative images per mouse per group and their average and SE were plotted. The OG/FAM mean signal per CD206 + cell analysis was measured using ImageJ, taking the mean OG/FAM signal, and dividing it with the number of CD206 + cells. Average values were obtained from four images per mouse. N = 3 for orthotopic TNBC and N = 2 for the homing in experimental metastasis of TNBC.

Analysis of Tumor and Liver Leakiness
Endogenous IgG immunostaining of orthotopic 4T1 tumors and livers was performed following the same Immunofluorescence (IF) protocol as described above to assess leakiness. Endogenous IgG was detected using Alexa Fluor 647 goat anti-mouse antibody (dilution 1/200, Invitrogen by Thermo Fisher Scientific, catalog no. A21235) and slides were counterstained with DAPI (1 μg/mL in PBS). The coverslips were mounted, and sections were imaged using Zeiss confocal microscope and 20× objective (N = 3 tumors).

PDL1 Expression Analysis in Orthotopic TNBC Tumors
The assessment of PDL1 expression in orthotopic 4T1 tumors followed the IF protocol described above. PDL1 was detected using rat anti-mouse PDL1 (dilution 1/100, BioLegend, catalog no. 124302) as primary antibody and Alexa Fluor 647 goat anti-rat (dilution 1/200, Invitrogen, catalog no. A21247) as the secondary antibody. Slides were counterstained with DAPI (1 μg/mL in PBS), mounted, and imaged using a Zeiss confocal microscope.

Plasma Half-life Evaluation for St-PGA-OG-mUNO
Plasma half-life studies were performed as described previously (42 DOX in medium, or free medium as a control (N = 3 wells/group). Concentrations used were calculated on the basis of DOX: 33 and 100 μmol/L.
(Of note, the dose of OximUNO used for the 33 μmol/L DOX in vitro experiments corresponds to the same dose of OximUNO used for both in vivo treatment studies). In vivo, all treated groups received injections containing 2 mg/kg of DOX, which, assuming the dilution in mouse blood, corresponds to a DOX concentration of 33 μmol/L). After incubation, wells were washed, fresh medium added, and cells incubated for 48 hours at 37°C. After 48 hours, 10 μL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, concentration 5 mg/mL, Invitrogen, catalog no. M6494) in PBS was added to each well containing culture medium and incubated for 2.5 hours at 37°C. Medium containing MTT was then removed without removing formed crystals, and 100 μL of isopropanol was added to each well to dissolve crystals. Absorbance was read at 580 nm using a plate reader (Tecan Sunrise) and the correspond- FcX for 25-45 minutes at +4°C in the dark. After that, cells were washed 2× with 200 μL of RB and read using BD Accuri 6 plus (BD Biosciences). As an isotype control, APC mouse IgG (BioLegend, catalog no. 400119) was used. For washing, plate was centrifuged at 350 × g for 7 minutes at +4°C.

In Vivo Liver and Kidney Toxicology Studies with OximUNO
Three healthy 12-week-old female Balb/c mice were intraperitoneally injected once with OximUNO (0.704 mg/0.5 mL PBS or 1.408 mg/0.5 mL) and circulated for 48 hours. Then, mice were anesthetized, and blood collected through retro-orbital bleeding into Lithium Heparin tubes (BD Vacutainer, catalog no. 368494). Blood samples were centrifuged at 1,800 × g for 15 minutes at +4°C and 400 μL of plasma was collected for analysis. Samples were analyzed in Tartu University Hospital using a Cobas 6000 IT-MW (Roche Diagnostics Gmbh) machine and reagents for creatinine (CREP2, catalog no. 03263991) and alanine aminotransferase (ALTLP, catalog no. 04467388).
For histologic analysis of livers and kidneys, after sacrificing animals, tissues were frozen into block, sectioned at 10-μm thickness and kept at room temperature for approximately 30 minutes before fixing them with ice-cold methanol for 2 minutes at room temperature followed by hematoxylin and eosin (H&E) staining as described under "H&E staining on PFA-fixed cryosections." Slides were scanned using Leica DM6 B microscope and Leica Aperio Versa 8 slides scanner with 20× zoom and images were analyzed using the ImageScope (version 12.3.3). Slides were then analyzed by pathologists.

OximUNO Treatment of Orthotopic TNBC
A total of 5 × 10 4 4T1 cells in 50 μL of PBS were subcutaneously injected into the fourth mammary fat pad of 8-12 weeks old female Balb/c mice. On day 7, mice were sorted into four groups by tumor volume measured using a digital caliper (Mitutoyo). Tumor volume was calculated on the basis of the formula ( Tumor tissues were processed as described under "In vivo biodistribution studies," and the lungs and hearts were embedded in paraffin and processed for H&E staining (described below). Tumors were immunostained as described above. CD206 was detected using rat anti-mouse CD206 (dilution 1/200), CD8 using rat anti-mouse CD8 (dilution 1/75 BioLegend, catalog no. 100701), FOXP3 using rat anti-mouse FOXP3 (dilution 1/75, BioLegend, catalog no. 126401) as primary antibodies, Alexa Fluor goat anti-rat 647 (dilution 1/300 for CD206 and 1/200 for CD8, FOXP3,) was used as a secondary antibody for all markers. Slides were counterstained with DAPI (1 μg/mL in PBS) and imaged using a Zeiss confocal microscope with a 10× objective. All five tumors from each group were included in the IF analysis and at least three images per mouse per group were included. Fluorescent signal intensity was calculated using the ImageJ; to account for different amounts of tissue in the different images, only the area containing tissue was selected and the "mean signal intensity" given by the program taken (total integrated intensity divided by the selected area). For this analysis, at least three images per tumor were included.

Survival Analysis Following OximUNO Treatment of Orthotopic TNBC
For survival analysis, treatment was performed the same way as described above, with N = 5 mice in each group. Mice were sacrificed when their tumors reached 1,500 mm 3 . Survival was analyzed using GraphPad Prism (version 9.3.1) to plot Kaplan-Meier survival curves and to perform Mantel-Cox test for statistical analysis.

OximUNO Treatment of Experimental Metastasis of TNBC
A total of 2 × with flow cytometry (FC), and three full lungs and three left lungs from each group were frozen into blocks using OCT. Frozen lung tissues were cryosectioned as described earlier, fixed for 10 minutes with cold 4% PFA (CD206) or acetone (for CD8 and FOXP3), and stained as described in the following section. Immunofluorescence staining was performed using the same markers and antibodies as shown in the "OximUNO treatment in orthotopic TNBC" section.

GFP Staining and Imaging
Six lungs from each group were frozen in OCT. Ten-micron-thick sections were cut and slides were kept at −20°C until ready to use. Slides were taken out of the freezer at least 30 minutes before staining. For staining, slides were fixed with 4% PFA for 10 minutes at room temperature, washed with PBS for 10 minutes at room temperature, counterstained using DAPI (1 μg/mL in PBS) for 5 minutes at room temperature, washed 3 × 4 minutes with PBS and finally mounted using mounting medium. Permeabilization was not used in this step to improve GFP visualization. GFP was visualized using its native fluorescence. Slides were imaged using Olympus confocal microscope (FV1200MPE) with a 10× objective.

Macroscopic Analysis of GFP Signal
Lungs from each group were imaged using Illumatool Bright Light System LT-9900 (LightTool's Research) in the green channel to visualize the fluorescent signal macroscopically, and a photograph of each lung was taken. The total GFP signal of each lung was quantified by ImageJ using the "IntDen" value.

FC Analysis
Three mice were sacrificed using anesthetic overdose, perfused with PBS and right lung tissues were placed in cold RPMI1640 medium supplemented with 2% v/v FBS. Lungs were cut into small pieces on ice in a solution containing collagenase IV (160 U/mL, Gibco, catalog no.17104019)/dispase (0.6 U/mL, Gibco, catalog no. 17105-041)/DNase I (15 U/mL; AppliChem, catalog no. A3778) mixture. To obtain a single-cell suspension, lung pieces were incubated in 10 mL of the same mixture at 37°C on a rotating platform for 45-60 minutes, pipetting every 10 minutes to improve digestion. The cells were washed with 5 mL of RB, centrifuged (350 × g, 7 minutes, 4°C), and red blood cells were lysed with 3 mL of ammonium-chloride-potassium lysing buffer at room temperature. A total of 10 mL of RB was added; cells were centrifuged and filtered using a 100-μm cell strainer (Falcon, catalog no. 352360). Cells were counted using the bright-field mode of LUNA Automated Cell counter (Logos Biosystems

H&E Staining on PFA-fixed Cryosections
Ten-micron-thick sections were cut from unfixed tissues in a frozen block; sections were stored at −20°C until ready to use. When ready, slides were taken out of the freezer 30 minutes before staining and stained within an hour for optimal results. Room temperature slides were fixed with cold 4% PFA for 10 minutes at room temperature followed by washing in PBS for 10 minutes at room

Statistical Analysis
All statistical analysis was carried out using one-way ANOVA and Fisher LSD (Least Significant Difference) tests, using the Statistica program (release 7), except for survival analysis, where GraphPad Prism (version 9.3.1) was used to perform Kaplan-Meier survival curves and Mantel-Cox for statistical analysis.

Data Availability
All data needed to evaluate the conclusions on the article are presented in the article and/or the Supplementary Data. Additional data related to the findings of this study are available from the corresponding author.

Design and Structural Modeling of St-PGA-OG-mUNO
To characterize and explore the function of OximUNO, we first developed an mUNO-targeted St-PGA labeled with the OG fluorescent dye (referred to as St-PGA-OG-mUNO; Fig. 1A; Supplementary Scheme S1). We conjugated OG   Table S1; Fig. 1B), while both nanoconjugates exhibited highly negative charges (−42 and −48 mV, respectively) as shown by Zeta potential analysis (Supplementary Table S1); an expected result given the glutamic acid nature of the polymer carrier. Analysis of mUNO loading (Supplementary Table S1) indicated the presence of approximately seven mUNO peptides in St-PGA-mUNO nanoconjugate, which would allow multivalent receptor binding.
We next assessed the structure of unlabeled and untargeted St-PGA in water using MD simulations to access information at the atomic scale. We assumed an initial helical conformation for the three PGA chains. The studied system consisted of a fully hydrated St-PGA and the Na + counterions (∼920,000 atoms) and was built after initial minimization under vacuum conditions. We simu- To investigate whether mUNO can engage with the CD206 receptor when grafted onto St-PGA, we modeled the structure and mobility of St-PGA-mUNO using computational analysis. To attain a computationally feasible system, we simulated only single branches of St-PGA-mUNO. We placed three equidistant mUNO peptides on a PGA single branch and fully solvated the system. We observed that three mUNO peptides remained exposed to the solution available for receptor binding (Fig. 1D). The rotation of mUNO around PGA, tracked by the angle formed by a proline aromatic carbon within mUNO ( Supplementary   Fig. S2, green sphere), a pyridyldithiol linker nitrogen ( Supplementary Fig. S2, blue sphere), and a glutamic acid aromatic carbon ( Supplementary Fig. S2, light blue sphere) revealed angles between 50°and 180° (Fig. 1E). This value supports the ability of mUNO peptides to interact with their receptor (43). Comparisons with an undecorated PGA branch demonstrated the minimal alterations of secondary structure dynamics in the presence of mUNO peptides-turning alpha helices (Fig. 1F, green) into random coils (Fig. 1F, brown) at regions where they are placed; however, the PGA chain structure remained mainly helical except in the middle, where a slight kink formed (Fig. 1F).

St-PGA-OG-mUNO targets CD206 + TAMs and Displays Low Hepatic Accumulation
We next evaluated the potential of St-PGA-OG-mUNO to target CD206 + TAMs in a TNBC model-induced by orthotopic inoculation (referred to as "orthotopic TNBC") or by intravenous inoculation (referred to as "experimental metastasis of TNBC") of 4T1 cells. We administered St-PGA-OG-mUNO or St-PGA-OG intraperitoneally, allowed circulation for 6 hours, and then analyzed tumor homing using confocal fluorescence microscopy. Our previous study provided the rationale for the intraperitoneal administration route, where we demonstrated that the intraperitoneally administered mUNO peptide exhibited a substantially longer half-life than intravenously administered mUNO in the same mice (same strain, sex, and age) used in this study (42).
In the orthotopic TNBC, we observed a high colocalization of OG/CD206 ( Fig Fig. S3A and S3B). We employed confocal image acquisition parameters throughout this study to visualize CD206 in the tumor without signal saturation. Given the higher levels of CD206 in the tumor, imaging with associated settings provides low CD206 visualization in the liver. Using a higher image intensity, we observed the expected CD206 signal in the liver (as expected from Kupffer cells and sinusoid vessels; Supplementary  Fig. S4A) and a saturated CD206 signal in the tumor ( Supplementary Fig. S4B).
One of the rationales behind the design of OximUNO was to increase mUNO targeting through increased avidity and plasma half-life. To evaluate these aspects, we compared the homing of St-PGA-OG-mUNO with a monomeric, carboxyfluorescein-labeled mUNO peptide (FAM-mUNO). We note that even given the different nature of the fluorescent labels (OG on St-PGA-OG-mUNO and fluorescein on FAM-mUNO), we did not use their native fluorescence as a readout; instead, we used an antibody that recognizes both FAM and OG; therefore, we do not expect biases from potential differences in FAM and OG emissions.
We discovered that St-PGA-OG-mUNO ( Supplementary Fig. S8A 2O)]. In addition, we found that the OG/FAM mean signal per CD206 + cell was five times higher for St-PGA-OG-mUNO than FAM-mUNO (Fig. 2P). These findings suggest that conjugating mUNO to the St-PGA backbone greatly improved receptor binding.
Plasma half-life analysis for intraperitoneally administered St-PGA-OG-mUNO revealed a 4.5-hour half-life (Fig. 2Q), a value over two times longer than that observed after the intraperitoneal administration of FAM-mUNO in our previous study (42). We previously showed that the plasma half-life of systemically administered St-PGA is approximately 12 hours (53), that negligible degradation of FAM coupled to mUNO through an amide bond (FAM-mUNO) occurs in serum (42), and that the fluorescence of FAM-UNO was not affected by serum from mice bearing 4T1 tumors (41). On the basis of these antecedents, we here attributed the plasma fluorescence of Fig 2Q, to St-PGA-OG-mUNO.
Overall, this finding suggests that conjugating mUNO to St-PGA increased the plasma half-life of mUNO peptide, a desirable feature that will improve in vivo ligand targeting.
We next compared tumor homing of St-PGA-OG-mUNO with that of a therapeutic mAb by intravenously injecting anti-PDL1 in orthotopic 4T1 tumor-bearing mice and allowing circulation for 24 hours. We observed that administered anti-PDL1 accumulated in the tumor rim (Fig. 2R, TR) but not in the tumor core (Fig. 2R, TC) even given expression of the receptor (PDL1) in the tumor core (Fig. 2S, TC). The observed accumulation of St-PGA-OG-mUNO in the tumor core (Fig. 2T, TC) and receptor colocalization (Fig. 2U), supported the implementation of our platform as an efficient alternative to antibody-based therapies such as anti-PDL1 or antibody-drug conjugates.

OximUNO Enhances the In Vitro Cytotoxicity of DOX on M2-resembling Macrophages
St-PGA displays a large surface with multiple sites available for the conjugation of proapoptotic or cytotoxic cargoes via bioresponsive polymer-drug linkers (54,55). To selectively deplete CD206 + TAMs, we conjugated an apoptotic chemotherapeutic agent (DOX) to St-PGA-mUNO to form St-PGA-DOX-mUNO (designated "OximUNO"; Fig. 3A, Scheme S2). We conjugated DOX to St-PGA-mUNO using a hydrazone bond (54) to allow for site-specific drug release in the acidic milieu of the endosomes or lysosomes (54,56).
To evaluate the effect of mUNO targeting, we included St-PGA-DOX as an untargeted control. We employed 1 H NMR and UV-Vis analyses to evaluate the chemical identity of nanoconjugates ( Supplementary Fig. S10A and S10B).
OximUNO displayed DOX and mUNO loadings of approximately 10% and approximately 4% in weight, respectively, corresponding to around four DOX and seven mUNO molecules for every OximUNO. OximUNO exhibited a size of approximately 40 nm and a highly negative surface charge of −40 mV (Supplementary Table S2; Fig. 3B). We obtained similar DOX loading, size by DLS, and surface charge values for St-PGA-DOX (Supplementary Table S2; Fig. 3B).
The pH-sensitive hydrazone linker and the intrinsic biodegradability of St-PGA by lysosomal protease cathepsin B are expected to secure DOX release from OximUNO after cell internalization (57). Hence, we studied DOX release kinetics from OximUNO in the presence of acidic pH (pH 5) and cathepsin B using LC/MS ( Supplementary Fig. S11A-S11G). As we aimed for the intraperitoneal administration of OximUNO, we assessed DOX release in intraperitoneal fluid (Fig. 3C). At pH 5, we observed a sustained DOX release in the first 8 hours (reaching a plateau at 15%), thereby demonstrating the suitability for endosomal-lysosomal drug delivery. DOX release in the presence of cathepsin B displayed comparable values in the first 8 hours (∼13%), followed by a plateau and a reduced rate in the following hours (∼13% cumulative release at 72 hours). Importantly, OximUNO exhibited negligible drug release in both physiologic conditions evaluated (PBS and intraperitoneal fluid; Fig. 3C).
Bertani and colleagues (58) observed the same pattern of CD206 expression in PBMC-derived macrophages polarized under similar conditions.
Because the in vivo concentration that provided optimal CD206 + TAM targeting with minimal hepatic accumulation was 30 μmol/L in OG, here we focused our interest on conjugates at 33 μmol/L of DOX. Our previous studies comparing other mUNO-targeted versus untargeted polymeric nanosystems (44) demonstrated that the highest targeted uptake in primary MCSF + IL4 polarized macrophages occurred after an interval of 10 to 30 minutes. For this reason, we used an incubation time of 15 minutes for these experiments.
These results provide evidence that OximUNO displayed increased toxicity toward M2-resembling macrophages when compared with St-PGA-DOX or DOX alone.
We also evaluated the hepatic and renal safety profile of a single administration of OximUNO (at doses corresponding to 2 and 4 mg/kg of DOX) by analyzing creatinine (Crea) and alanine aminotransferase (ALAT) levels 48 hours after intraperitoneal administration in healthy mice (Supplementary Table S3). These doses did not induce toxic levels of Crea or ALAT compared with the values reported in the literature (61) Supplementary Fig. S12A), and IFNγ IF did not detect a clear increase that would indicate inflammatory changes ( Supplementary Fig. S12B).
In summary, the conjugation of mUNO and DOX to the St-PGA backbone to yield OximUNO, enhanced the in vitro efficacy of DOX toward M2-resembling macrophages with no in vivo renal or hepatic toxicity observed.

OximUNO Treatment of Orthotopic TNBC Depletes CD206 + TAMs, Inhibits Tumor Progression, and Attenuates Immunosuppression
The  (Fig. 4A, black line). Furthermore, OximUNO treatment did not affect mouse bodyweight, whereas treatment with DOX induced a significant decrease in mouse bodyweight starting from day 21 p.i. until the end of the treatment (Fig. 4C).
Histologic analysis of lungs from treated mice ( Supplementary Fig. S13 shows an H&E stain from a healthy lung for comparison) revealed that OximUNO showed a decreasing trend in the metastatic lung area and nodule number (Fig.  4D-F). Meanwhile, IF microscopy revealed no significant changes in CD31 expression in tumors ( Fig. 4G and H), but significantly fewer CD31 + structures in the OximUNO-treated mice compared with DOX-treated mice ( Fig. 4G and I), suggesting that the reduction in nodule number in the OximUNO group (of Fig. 4F) may be mediated by the lower vascularization in the primary tumor. We suggest that partial vascular homing of St-PGA-DOX in the tumor (as suggested by the tumor homing of its OG equivalent; Supplementary Fig.  S14) contributes to the blood vessel reduction observed in this group. Importantly, histologic analysis revealed no cardiotoxicity in any treatment groups ( Supplementary Fig. S15). IF analysis revealed that only OximUNO significantly reduced the CD206 expression (assigned to CD206 + TAMs), compared with PBS ( Fig. 4J and K). Interestingly, treatment with DOX upregulated CD206 expression ( Fig. 4J and K), which agrees with previous reports that demonstrated an increase in the number of CD206 + TAMs following chemotherapy (24).
Notably, only OximUNO treatment significantly increased CD8 expression [a marker of cytotoxic T lymphocyte (CTL)] compared with PBS and DOX treatment ( Fig. 4L and M). Unexpectedly, St-PGA-DOX treatment increased the expression of FOXP3, a marker for regulatory T cells (Treg; Fig. 4N and O).
Analysis of the CD8/FOXP3 expression ratio revealed that OximUNO treatment resulted in a 5-fold increase compared with St-PGA-DOX or DOX treatment (Fig. 4P), suggesting that OximUNO stimulated a shift in the immune landscape toward immunostimulation. Of note, in all cases, we normalized the quantification of marker expression using immunofluorescent images to the tissue area to account for different amounts of tissue in different images.
A repetition of this treatment study, monitoring primary tumor growth and survival, showed the slowest tumor growth kinetics in the OximUNO group ( Supplementary Fig. S16A) and Kaplan-Meier curves showed a significantly prolonged survival for OximUNO-treated mice compared with untreated mice ( Supplementary Fig. S16B-S16D).
By targeting CD206 + TAMs with DOX via OximUNO treatment, we increased the efficacy and reduced the toxicity of DOX in the orthotopic TNBC. Our results also suggest that the depletion of CD206 + TAMs by OximUNO elicited an immunostimulatory shift.

OximUNO Treatment of Experimental Metastasis of TNBC Reduces CD206 + TAMs Number, Tumor Burden and Attenuates Immunosuppression
We next evaluated the effect of OximUNO on experimental metastasis of TNBC using GFP-labeled 4T1 cells. We treated mice every other day with intraperitoneal injections of OximUNO, St-PGA-DOX, or DOX, starting from day 4 p.i. and sacrificed mice on day 18 p.i. Analysis of whole lung fluorescence in the green channel revealed that OximUNO treatment induced the lowest GFP fluorescence, indicating a lower level of pulmonary metastases (Fig. 5A). Representative macroscopic images also provided evidence for a reduction in metastases (Fig. 5B). Confocal fluorescence microscopy of lungs confirmed the trend observed with whole lung fluorescence, showing fewer GFP fluorescent nodules in the OximUNO-treated group (Fig. 5C). Furthermore, histologic analysis of lungs displayed the lowest number of pulmonary nodules for OximUNO-treated mice ( Fig. 5D and E). Mice treated with the untargeted St-PGA-DOX and free DOX showed a significant decrease in bodyweight, resulting in a 19% (Fig. 5F, blue line) and 17% loss (Fig. 5F, purple line), respectively; meanwhile, OximUNO-treated mice displayed lower bodyweight loss (Fig. 5F, red line).
OximUNO treatment significantly lowered the percentage of M2 TAMs (CD206 + ; Fig. 5G) but did not significantly impact the percentage of M1 TAMs, CTLs, or Tregs ( Fig. 5H-J). We observed the same trend when we expressed these populations as total cell counts ( Supplementary Fig. S17-S20).
To evaluate whether OximUNO affected CD206 + macrophages other than M2 TAMs, we analyzed the state of splenic macrophages from this treatment study using FC. This analysis revealed no significant differences in the CD206/CD86 populations between the OximUNO-treated mice and PBS-treated mice ( Supplementary Fig. S21A-S21C).
While FC analysis informs on the immune status of the whole lung, it does not provide specific information regarding the TME. To characterize the immune landscape of the TME, we next analyzed the expression of markers for TAMs, CTLs, and Tregs in pulmonary nodules using IF. This analysis revealed significantly lower CD206 expression in OximUNO-treated mice than PBS-treated mice ( Fig. 5K and L), providing evidence for a robust reduction in the number of CD206 + TAMs in the TME. Importantly, and similarly to OximUNO treatment in the orthotopic TNBC, OximUNO elicited the highest expression of CD8 ( Fig. 5M and N). OximUNO-treated and St-PGA-DOX-treated mice demonstrated significantly lower lung FOXP3 expression than PBS-and DOXtreated mice ( Fig. 5O and P). OximUNO-treated mice displayed between a two and three times higher CD8/FOXP3 expression ratio than St-PGA-DOX and DOX, and nearly seven times higher than PBS (Fig. 5Q). Therefore, our IF analysis in the pulmonary tumor nodules suggested that OximUNO triggered a shift in the immune profile of the TME toward immunostimulation.
By targeting DOX to CD206 + TAMs in experimental metastasis of TNBC, we increased the efficacy and reduced the toxicity of DOX, as OximUNO treatment associated with the presence of fewer pulmonary tumor lesions and less bodyweight loss when compared with treatment with untargeted St-PGA-DOX and DOX. Our results suggest that the observed therapeutic effect derived from CD206 + TAM depletion, which elicited an immunologic shift in the TME.
We recently identified and described a short peptide called mUNO (sequence: CSPGAK) that targets mouse (41) and human CD206 (43) at a different binding site than for mannose on CD206 (43). We identified mUNO from an in vivo screen using a peptide library in mice bearing metastatic breast cancer; we subsequently described how mUNO homed to CD206 + TAMs in other solid tumor models (41,78) and in early-stage models of TNBC (42) displaying low hepatic accumulation.
We envisioned that conjugating mUNO to St-PGA would significantly enhance targeting through the avidity effect and increased plasma half-life (79).
Compared with synthetic polymers such as N-(2-hydroxypropyl) methacrylamide, polypeptide-based nanocarriers show several benefits, including biodegradability, lower immunogenicity, and a lack of long-term accumulation, and the number of polypeptide-based constructs reaching clinical evaluation has significantly increased in recent years (80)(81)(82)  through a higher hydrodynamic volume that reduces rapid renal clearance (53,84). Of note, an extended plasma half-life will be advantageous when targeting the continuously replenished TAM cell type (85,86).  (25), and CD11c + DCs participate in antigen presentation (87). In line with these observations, the computational analysis indicated that mUNO peptides are available to a receptor and sweep a vast space (130°) around PGA. Altogether these data demonstrate the benefit of conjugating mUNO to St-PGA. While previous studies have reported the St-PGA nanocarrier (53,83) and the mUNOtargeting peptide (42), this work represents a novel design of a peptide-targeted St-PGA nanosystem. Regarding the administration route of peptide-guided St-PGA nanosystems, in the future we also wish to evaluate the intravenous route, which, barring the case of intraperitoneal chemotherapy, represents a more clinically translatable route to deliver cancer therapies.
In the OximUNO system, drug release studies revealed only 15% DOX release, which agrees with our previous studies (54, 56) but suggests room for improvement, which may come from using longer polymer-drug linkers such as EMCH (N-ε-maleimidocaproic acid hydrazide) moiety (54,56) or from the use of external triggers (88)(89)(90). Unexpectedly, we failed to observe a significant increase in DOX release in the presence of cathepsin B with respect to the hydrolytic conditions; we hypothesize that the nanoconjugate conformation slows down proteolytic degradation, hampering in vitro quantification within the studied timeframe (53).
Our in vivo efficacy studies showed that, strikingly, the sole depletion of CD206 + TAMs with OximUNO alleviated tumoral immunosuppression and reduced dissemination and growth, confirming the protumoral and immunosuppressive roles assigned to CD206 + TAMs in the literature and reaffirming the importance of targeting this particular TAM subset. In addition, the observed reduction in the number of CD206 + TAMs and CD31 + structures for OximUNO agrees with the established angiogenic role of CD206 + TAMs (24). Beyond TAM depletion, we show that St-PGA-mUNO represents an attractive platform to carry additional therapeutic payloads other than DOX, which could include M2→M1 polarizing agents such as TLR7 agonists (44,96), betaemitting radiotherapeutic agents such as dodecanetetraacetic acid-chelated 177 Lu (97), or photosensitizers used in photodynamic therapy (88)(89)(90). We also envisage the combination of TAM depletion via OximUNO administration together with current chemotherapy regimens to prevent dissemination and relapse, or the use of OximUNO prior to surgery, that is, as neoadjuvant chemotherapy.
Taking OximUNO as a proof of concept, our data support the peptide-targeted St-PGA design reported here as a new targeted nanosystem that could target other receptors by exchanging the targeting peptide.