Detection of small breast tumors using tumor penetrating-polymersomes engineered to target p32 protein

Triple negative breast cancer (TNBC) is the deadliest form of breast cancer and its successful treatment critically depends on early diagnosis and therapy. The multi-compartment protein p32 is overexpressed and present at cell surfaces in TNBC, specifically in the malignant cells and endothelial cells, and in macrophages localized in hypoxic areas of the tumor. Herein we used polyethylene glycol-polycaprolactone polymersomes that were affinity targeted with the p32-binding tumor penetrating peptide LinTT1 (AKRGARSTA) for imaging of TNBC lesions. A tyrosine residue was added to the peptide to allow for 124I labeling and PET imaging. Systemic LinTT1-targeted polymersomes accumulated in early tumor lesions more than twice as efficiently as untargeted polymersomes with up to 20% ID/cc at 24 h after administration. The PET-imaging was very sensitive, allowing detection of tumors as small as ~20mm3. Confocal imaging of tumor tissue sections revealed a high degree of vascular exit and stromal penetration of LinTT1-targeted polymersomes and co-localization with tumor-associated macrophages. Our studies show that systemic LinTT1-targeted polymersomes can be potentially used for precision-guided tumor imaging and treatment of TNBC. uFigure1 Small triple negative brast tumors can be detected with LinTT1-conjugated polymersomes. Radiolabeled LinTT1-polymersomes were intravenously injected into mice bearing small triple negative breast tumor. LinTT1 peptide binds to p32 protein expressed in the surface of tumor cells and activated macrophage/myeloid cells. LinTT1 is cleaved by urokinase type plasminogen activator (uPA) in tumor, and the processed peptide binds to NRP-1, triggering the penetration of polymersomes into the tumor tissue.


Introduction
TNBC accounts for a 15% of all breast cancer cases and shows the least favorable prognosis among the breast cancer subtypes. On average, patients with TNBC have cancer recurrence within 3 years after initial diagnosis and a life expectancy of approximately 5 years [1]. TNBC tumors are locally invasive, highly metastatic, and must be detected and treated early to prevent dissemination.
Nanoformulations offer unique advantages for drug delivery. Nanoparticles can be designed to encapsulate hydrophobic molecules that would otherwise be insoluble, and payloads that have short circulation half-life and/or need to be protected from enzymes in the bloodstream, such as esterases or nucleases [2]. Cancer diagnosis and treatment can be combined into one modality by dual-use "theranostic" nanocarriers engineered to simultaneously deliver therapeutic and imaging cargoes [3] [4]. Imaging payloads, such as fluorescent, MRI, and radio tags can be loaded in the nanosystems or coated on their surface. Nanosized polymeric vesicles (polymersomes) self-assembled from biocompatible copolymers are particularly appealing because of their versatility and unique properties. The high molecular weight of block copolymers results in the formation of highly entangled membranes displaying a high degree of resilience with elastomer-like mechanical properties. This confers the polymersomes a high flexibility [5] [6] and higher ability for tissue penetration than other vesicles self-assembled from low molecular weight entities, such as liposomes [7]. Polymersomes can be loaded with hydrophilic effector molecules, e.g. low molecular weight drugs [8] [9], proteins [10], nucleic acids [11], and imaging agents [12] [13], in their aqueous lumen and with hydrophobic cargoes within the polymer membrane [8] [14].
The surface of nanoparticles can also be modified to improve their in vivo behavior such as circulation half-life, non-specific interactions and affinity for non-target sites, and to achieve selective accumulation in target tissue(s). Affinity ligands, such as peptides [15] [16] and antibodies [17] can be coated on the nanoparticles for specific tissue and cell recognition.
Tumor-penetrating peptides [18] can be used to concentrate cytotoxic molecules and drugloaded nanoparticles in tumors and potentiate their antitumor activity [14] [19]. The AKRGARSTA peptide, referred to as "LinTT1" (linear TT1), is a 9-amino acid tumor-penetrating peptide that binds to p32 protein, which is overexpressed on the membrane of tumor cells and also on macrophage/myeloid cells in hypoxic areas of tumors [20] [21]. LinTT1 is processed by tumorderived proteases, such as uPA, to C-terminally expose the C-end rule motif of the peptide (i.e. AKRGAR), which is capable of interacting with the cell-and tissue-penetration receptor NRP-1 [20] [22]. Recently, LinTT1-functionalization was found to significantly improve the therapeutic index of iron oxide nanoworms loaded with proapoptotic effector peptide in a TNBC model [23].
In that study, the tumor accumulation of fluorescently labeled Lin-TT1 nanoparticles was evaluated by optical imaging of tissue sections. However, fluorescence imaging-based in vivo biodistribution studies remain challenging due to issues related to the low depth of light penetration, tissue autofluorescence, and the semi-quantitative nature of optical imaging [24].
PET and SPECT are clinically used for imaging of radioactive contrast agents with beta and gamma emission, respectively. In contrast to MRI, CT, and imaging using optical contrast agents, PET and SPECT are not subject to endogenous tissue background.
Encouraged by the anticancer activity of LinTT1-targeted therapeutic nanoparticles on breast tumors in mice [23], we decided to evaluate polymersomes guided with the LinTT1 peptide as a potential theranostic nanocarrier to early detect TNBC lesions. We radiolabeled LinTT1-targeted PEG-PCL polymersomes and studied, for the first time, the homing to orthotopic small breast tumors and the biodistribution of the polymersomes using PET imaging in mouse. Intravenouslyadministered p32-targeted 124 I labeled polymersomes showed good tumor selectivity and, importantly, allowed detection of tumors smaller than 20mm 3 . Our results suggest potential applications of LinTT1 engineered polymersomes for early detection of TNBC.

Preparation, functionalization and radiolabeling of polymersomes
Polymersomes were prepared by the film hydration method and functionalized with the Cys-Tyr-LinTT1 or Cys-FAM-LinTT1 peptide through a thioether bond between the maleimide group of the copolymer and the thiol group of the cysteine of the peptide. The number of polymersomes was determined by the ZetaView instrument and the functionalization with FAM-labeled peptide was quantified by fluorimetry. The peptide functionalization resulted in ~3.7x10 4 peptides/polymersome particle (density ~0.7 peptides/nm 2 ). For radiolabeling, the polymersomes were functionalized with the LinTT1-Tyr-Cys peptide or control Tyr-Cys dipeptide. The tyrosine residue was incorporated for radioiodination. The hydrodynamic diameter of LinTT1-Tyr-polymersomes (LinTT1-Tyr-PS), Tyr-polymersomes (Tyr-PS), polymersomes labeled with ATTO550 (LinTT1-ATTO550-PS and ATTO550-PS), and polymersomes labeled with FAM (LinTT1-FAM-PS and FAM-PS) measured by DLS, was ~130 nm for all the polymersome samples ( Figure 1B). The Z-potential was slightly negative but very close to 0 for the different polymersome preparations ( Figure 1B and S1).
For PET imaging, the LinTT1-Tyr-PS and Tyr-PS were radiolabeled with 124 I. Before purification, the efficiency of polymersome radiolabeling was determined by TLC ( Figure S2). The yield of radiolabeling after purification, measured with activimeter, was 48±9% for LinTT1-Tyr-124 I-PS and 43±2% for Tyr-124 I-PS. The low radiolabeling of PEG-PCL polymersomes without peptide indicated that 124 I present in LinTT1-Tyr-124 I-PS and Tyr-124 I-PS preparations was predominantly due to the covalent binding of 124 I to the tyrosine residue of the peptides ( Figure S2). TLC analysis after purification demonstrated that 99% of the 124 I was bound to polymersomes (

LinTT1-targeted polymersomes bind to recombinant p32 and to cultured breast tumor cells.
To evaluate the effect of LinTT1 functionalization on the tropism of polymersomes in vitro, we first tested the binding of LinTT1-Tyr-124 I-PS to recombinant p32 protein, the primary receptor of LinTT1. P32-coated magnetic beads were incubated with the polymersomes, and polymersome binding was quantified by gamma counter. Compared to non-targeted polymersomes, LinTT1-Tyr-124 I-PS showed ~10-fold increased binding to the p32 beads ( Figure 2A). This binding was specific, as the LinTT1-Tyr-124 I-PS did not bind to NRP-1 ( Figure 2A). The LinTT1 peptide does not bind to NRP-1 unless proteolytically processed by uPA [22]. These data show that the LinTT1 peptide attached to the polymersomes remains available for p32 binding to modulate polymersome tropism.
Various human and mouse tumor cell lines express p32 on the cell surface [21]. We studied the presence of cell surface p32 in 4T1 and MCF-10CA1a TNBC cells by flow cytometry and confocal microscopy, and confirmed its surface expression on both cell lines ( Figure S3).To study the uptake of polymersomes in 4T1 cells, we incubated the cultured cells for 1h with LinTT1-targeted or control polymersomes labeled with ATTO550 (LinTT1-ATTO550-PS and ATTO550-PS) ( Figure 2B). The LinTT1-functionalization increased polymersome uptake in 4T1 cells and the signal from LinTT1-ATTO550-PS partially colocalized with p32 ( Figure 2B). These experiments demonstrate that LinTT1 functionalization results in p32-enhanced uptake of polymersomes in cultured 4T1 cells. At 48 h after the injection, the tumors and organs were excised and 124 I in tissue extracts was quantified with gamma counter. The highest percentage of ID/g of both targeted and nontargeted polymersomes after 48 h was observed in spleen and tumor ( Figure 4A). Accumulation of both LinTT1 and untargeted polymersomes in spleen is consistent with the polymersome clearance by the RES. At 48 h, untargeted polymersomes showed accumulation in tumors (15±0.6% ID/g) and the functionalization with LinTT1 increased tumor accumulation of polymersomes by >70%, to 26±3% ID/g. Moreover, the percentage of ID/g of LinTT1-Tyr-124 I-PS in tumor was 2.5 times higher than in liver ( Figure 4A). Quantification of radioactivity revealed more than 2-fold higher accumulation of LinTT1-Tyr-124 I-PS than Tyr-124 I-PS in the sentinel lymph node of breast tumor mice ( Figure 4A). of Tyr-124 I-PS and 45% for LinTT1-Tyr-124 I-PS remained in the body ( Figure 4B). We have shown in a recent publication that an insignificant portion of the peptide is released from PEG-PCL polymersomes incubated with the serum of the 4T1 tumor bearing mice for 6 h [25]. We suggest that the high excretion at short time points observed is due to the renal clearance of the 124 I released from the peptide-conjugated polymersomes. It is important to note that the signal in thyroid gland ( Figure S4) -which accumulates free iodine -is similar for both targeted and untargeted polymersomes, suggesting similar leaching of iodine from polymersomes.
The effect of LinTT1 functionalization on the biodistribution and elimination rate of the polymersomes may be due to depletion of circulating LinTT1-polymersomes by the target sites: tumor tissue and macrophages. 4T1 tumor mice injected with both LinTT1 targeted and untargeted polymersomes showed similar 124 I excretion rate at short time points. However, after 6 h, the 124 I excretion rates for the targeted polymersomes became lower, likely due to preferential uptake of LinTT1-PS by p32 + tumor cells and activated macrophages.

LinTT1-polymersomes target both the tumor cells and tumor macrophages
We next studied the tissue biodistribution of i.v. administered FAM-labeled polymersomes in 4T1 orthotopic tumor mice at the cellular level. The polymersomes were injected in 4T1 tumor mice, allowed to circulate for 24 h, and the sections of tumors and control organs were analyzed by confocal immunoanalysis.
We first studied the biodistribution of p32 immunoreactivity in tissues. In a previous report, p32 was found to be upregulated in MDA-MB-435 breast tumors compared to the control organs [26] [21]. P32 immunostaining of sections of tumors and control organs from 4T1 mice demonstrated elevated expression of p32 in tumor tissue ( Figure S5). In agreement with the tissue extract-based radiography data, FAM-LinTT1-polymersomes accumulated in tumor and spleen (Fig. 3A). It was recently published that LinTT1 functionalization of nanoparticles enhances their penetration into tumor tissue [23] [15]. Here we show that at 24 h, the LinTT1-FAM-PS in tumors did not colocalize with CD31-positive blood vessels, confirming that polymersomes had extravasated and penetrated into tumor stroma ( Figure 4C, tumor inset). It has been shown that p32 is expressed by CD11b positive macrophages [26] and that LinTT1-conjugated nanoparticles colocalized with C68-positive macrophages in the breast [23], gastric, and colon tumors [15]. To study the macrophage uptake of LinTT1-PS, sections of tumors and organs were immunostained with antibodies against CD68, CD11b, and CD206 markers. CD68 and CD11b are pan-macrophage markers that label normal macrophages (including macrophages in spleen, lung, and in Kupffer cells in liver [25]), and TAMs [27]. CD206 is a marker of pro-tumor M2 macrophages [28] and thought to promote tumor progression [29]. We found that LinTT1-FAM-PS colocalized with CD68 (˃50% of colocalization), and showed partial colocalization with CD11b and CD206 (9% and 21% of colocalization, respectively) in tumors, confirming the targeting of tumor-associated macrophages ( Figure 5A and 5B). CD68positive macrophages, extensively found in sentinel lymph node and spleen, and in liver (in less extend), also showed colocalization with LinTT1-FAM-PS ( Figure S6A and S6B), which might be one of the reasons for the higher accumulation of LinTT1-polymersomes in spleen and lymph node (Fig. 4A) compared with untargeted polymersomes.
Human triple negative breast tumors and methastatic lymph nodes overexpress p32 and CD68+ macrophages.
To investigate the clinical relevance and translability of LinTT1-targeted polymersomes, we investigated the distribution of p32 and CD68 immunoreactivity on surgical cases of triple negative breast primary tumors and metastasis, in comparison to healthy breast tissue. As exemplified in figure 6A and B, p32 diplays a uniform pattern of expression on healthy breast and healthy lymphoid tissue whereas strong membranous staining is found on malignant lesions. p32 appears to be statically significantly overexpressed in primary tumors, both metastatic and non-metastatic, and in particulary on sentinel lymph nodes metastasis ( Figure   6B,C). Additionally, increased number of CD68-positive cells was found in breast tumors and in sentinel lymph nodes, both from patients with and without metastases, in comparison to healthy breast ( Figure 6D and Figure S7).

Discussion
In the current study, we evaluated the LinTT1-guided biocompatible PEG-PCL polymersomes as PET contrast agent for TNBC detection. Our findings indicate that LinTT1-polymersomes can be used for sensitive and specific detection of small triple negative breast tumors. This, along with recently published reports on LinTT1-mediated targeting of therapeutic nanocarriers [15][23], suggests potential theranostic applications for the LinTT1-targeted nanocarriers for early detection and treatment of TNBC. Nanoparticles have been affinity targeted to tumors for PET imaging. For example, in a recent PET study, clinical application of RGD-targeted PET-active nanoparticles for melanoma imaging has been reported [30]. The current study documents high tumor accumulation of LinTT1polymersomes (>20% ID/cc) that translates into ability to detect very small malignant lesions, barely visible by CT. This sets our system apart from other molecular and nanoparticle PET contrast agents with reported tumor accumulation range between 5-10% ID/g [31][32] [33].
Remarkable tumor selectivity and tumor binding capacity observed for the LinTT1polymersomes is likely to be due to a combination of the tumor homing properties of LinTT1 peptide with the favorable properties of the PEG-PCL polymersome nanoplatform. LinTT1 belongs to a family of tumor homing peptides that, unlike conventional vascular homing peptides, are not limited to vascular docking sites but have access to extended tumor extravascular space [34]. P32 protein, the receptor of LinTT1 peptide, is normally expressed in the mitochondria of the cells, but it is aberrantly displayed on the surface of tumor cells and on macrophage/myeloid cells, specially in hypoxic areas of tumors[20] [21]. LinTT1 bind to the superficial p32 on the tumor cells and activated macrophages and is cleaved by uPA, enzyme involved in tumor migration and progression [35]. LinTT1 is then exposing the C-end motive (R/KXXR/K) on the C-terminal. C-end motive binds to NRP-1 protein, which is overexpress in tumor cells and tumor vasculature. The binding to NPR-1 triggers an increase of the tumor tissue permeability and the peptide together with the cargo is internalized into the tumor [36].
Another potentially contributing aspect, not addressed in the current study, is the ability of LinTT1 to increase tumor penetration of co-administered compounds and nanocarriers. LinTT1iron oxide nanoparticles were recently found to increase tumor penetration of co-administered 70kDa dextran [15]. Homing of LinTT1-nanocarriers may thus not be limited by the number of systemically accessible peptide receptors [37] and allow more nanocarriers to enter the tumor tissue for improved sensitivity of detection. Tumor accumulation of LinTT1-polymersomes may also be enhanced by physicochemical features PEG-PCL polymersomes used in the current study. On one hand, the flexibility of polymersomes [7] may contribute to tissue penetrative targeting with tumor-penetrating peptides. In addition, polymersomes are known to possess an intrinsic tumor tropism. For example, we have recently demonstrated that pH-sensitive polymersomes efficiently delivered payloads to the tumor tissue in the absence of active targeting [8]; this accumulation was further boosted by targeting with iRGD peptide [14]. Likewise, the systemic radiolabeled non-targeted polymersomes in the current study showed high accumulation in 4T1 breast tumors; this accumulation was potentiated by functionalization of polymersomes with the LinTT1 peptide by about 70%.
LinTT1 homing is likely due to a combination of both tumor cell and macrophage targeting. In the 4T1 breast tumor mice, the highest ID/g of LinTT1-polymersomes was seen in the tumor, spleen, and lymph nodes. All these tissues contain abundant macrophages, a cell population known to upregulate, upon activation, the expression of cell surface p32. TAMs are an important diagnostic and therapeutic target, which play major roles in progression of solid tumors. In the context of drug delivery, TAMs can act as slow-release reservoir of drugs encapsulated in polymeric particles [38]. LinTT1-polymersomes may be capable of targeting tumor cells and TAMs in TNBC patients, as the peptide is not species specific, and since TAMs are abundant in clinical lymph node and breast tumor samples. We show here that primary breast tumors and sentinel lymph nodes from clinical samples from patients with or without metastasis overexpress p32 protein and that the number of CD68+ macrophages is increased compared with healthy tissues. This finding supports potential translatability of the system into clinical applications.
Clinical breast tumors are heterogeneous and the cell surface p32 expression and the sensitivity to p32-targeting-based treatment is likely to differ between the patients. PET imaging with LinTT1-polymersomes can be potentially used as a companion diagnostic test for selection of patient cohort most likely to respond to p32-targeted therapies.
Lymphangiogenesis in tumor-draining lymph nodes occurs before the onset of metastasis and is associated with distant metastasis. A study using Lyp-1 (a tumor lymphatics-specific peptide and also p32-binder [21]) to image tumor-induced lymphangiogenesis [39], suggested that the pre-malignant niche is positive for p32. The accumulation of LinTT1-polymersomes in sentinel lymph nodes containing 4T1 tumor cells migrating out from the primary tumor and activated macrophages, suggests potential applications for LinTT1-polymersomes for improved detection of early metastatic dissemination of breast cancer than is possible with currently approved compounds, such as Lymphoseek [40].
The potential applications of our system extend beyond breast cancer detection and therapeutic targeting. Systemically accessible p32 is overexpressed across solid tumors, including, gastric, colon, and ovarian caricinoma [15], glioma (Säälik et al. unpublished), and in atherosclerosis lesions [41]. Systematic evaluation of the relevance of the linTT1-polymersomes for detection and/or therapy of these conditions will be a subject of follow-up studies. The solvent was then evaporated to form the polymer film. The polymersomes were assembled and the Cys-LinTT1 peptide conjugation was performed as described above.

Materials
DLS and Z-potential measurements (Zetasizer Nano ZS, Malvern Instruments, USA) were used to assess the average size, polydispersity, and surface charge of polymersome preparations.
The size was measure at a concentration of 1mg polymer/mL in PBS (10mM of phosphate and 137mM of NaCl). The z-potential was measured at 0.2mg of polymer/mL in NaCl 10mM). TEM was used to assess the size and morphology of assembled vesicles. Briefly, polymersomes were deposited from a water solution onto copper grids at 1mg/mL, stained with 0.75% phosphotungstic acid (pH 7), air-dried, and imaged by TEM (Tecnai 10, Philips, Netherlands).
The number of polymersomes in the suspension was measured using the ZetaWiew instrument (Particle Metrix GmbH, Germany).

In vitro binding of polymersomes to recombinant p32 protein
Recombinant hexahistidine-tagged p32 was bacterially expressed and purified as previously described [20]. For protein binding assays, Ni-NTA magnetic agarose beads (Qiagen, Germany) in binding buffer (50mM Tris pH 7.4, 150mM NaCl, 5mM imidazole) were coated with p32 protein at 15µg of protein/10µL beads. Radiolabeled polymersomes were incubated with the p32-coated beads in binding buffer containing 1% BSA at room temperature for 1 h. The magnetic beads were washed with binding buffer and resuspended in a final volume of 1mL of binding buffer. The radioactivity of each sample was quantified by automatic gamma counter (2470 Wizard 2, Perkin Elmer).

Biodistribution studies
For biodistribution studies, the tumor, blood, and organs were weighed and the radioactivity was measured using the automatic gamma counter. A standard curved was generated using 124 I to determine the relationship between cpm and Bq. The biodistribution was expressed as percentage of injected dose per gram of tissue (ID/g). To determine the elimination rate of polymersomes, the radioactive signal in the whole mouse body was measured from the PET images at 10 min, 2, 6, 12, 24, and 48 h post-injection and normalized by the signal at 10 min post-injection. The elimination rate was expressed as signal in mouse x 100 divided by the signal at time zero.

Tissue immunofluorescence and confocal microscopy
Balb/c mice were orthotopically injected with 1 million of 4T1 cells in the mammary gland and

Statistical analysis
All the statistical analysis was performed with the Statistica 8 software, using the one-way ANOVA, Fisher LSD test.