Research ArticleCANCER

Nanoparticle interactions with immune cells dominate tumor retention and induce T cell–mediated tumor suppression in models of breast cancer

See allHide authors and affiliations

Science Advances  25 Mar 2020:
Vol. 6, no. 13, eaay1601
DOI: 10.1126/sciadv.aay1601


The factors that influence nanoparticle fate in vivo following systemic delivery remain an area of intense interest. Of particular interest is whether labeling with a cancer-specific antibody ligand (“active targeting”) is superior to its unlabeled counterpart (“passive targeting”). Using models of breast cancer in three immune variants of mice, we demonstrate that intratumor retention of antibody-labeled nanoparticles was determined by tumor-associated dendritic cells, neutrophils, monocytes, and macrophages and not by antibody-antigen interactions. Systemic exposure to either nanoparticle type induced an immune response leading to CD8+ T cell infiltration and tumor growth delay that was independent of antibody therapeutic activity. These results suggest that antitumor immune responses can be induced by systemic exposure to nanoparticles without requiring a therapeutic payload. We conclude that immune status of the host and microenvironment of solid tumors are critical variables for studies in cancer nanomedicine and that nanoparticle technology may harbor potential for cancer immunotherapy.


Nanoparticles provide unique opportunities and challenges for cancer therapy and diagnosis. They have the potential to interact with the immune system and solid tumor microenvironment (TME) in unexpected ways to ultimately and critically affect performance and tumor response (13). The premise that nanoscale materials can be engineered to selectively detect and destroy cancer cells in solid tumors is undergoing a critical reevaluation (411). Yet, relatively little analysis of nanoparticle fate and intratumor accumulation across biological models and immune cell or tumor compartments has been completed, particularly with histology or flow cytometry (6).

As with many cancer drug development scenarios, nanotechnology-based formulations are often tested and optimized using a specific mouse model of human cancer. These xenograft tumor studies rely on immunodeficient animal models, which provide a permissive environment for cross-species tissue grafting. Therefore, how well these models predict the potential and mechanisms for “nano-targeting” becomes a relevant question when the construct itself demonstrates strong interactions with the recipient’s immune system (13, 6).

Polysaccharide (starch or dextran)–coated iron oxide nanoparticles have been used for decades in biomedicine as agents for parenteral anemia therapy, magnetic resonance contrast, cancer hyperthermia, drug delivery, cell sorting, and most recently for inducing ferroptosis in cancer cells (4, 5, 1219). Thus, they present an interesting and important class of nanoparticles for applications in medicine.

Here, we show that host immune status and the immune components of the TME are key factors influencing retention of 100-nm hydroxyethyl starch–coated iron oxide nanoparticles in orthotopic mammary tumors. When labeled with an antibody, the nanoparticles were retained by tumors regardless of the presence of the target antigen, whereas retention of the unlabeled counterpart was not substantial. Additional experiments demonstrated that systemic exposure of tumor-bearing immune competent mice to the nanoparticles induced immune-mediated tumor growth inhibition with evidence of later infiltration by CD8+ T cells. Both plain and antibody-labeled nanoparticles initiated similar immune responses with similar tumor growth inhibition and T cell infiltration into tumors, despite different tumor retention. This suggests that complex interdependencies exist between host and tumor immune responses to nanoparticle exposure. Together, these results offer intriguing possibilities to explore nanoparticle “targeting” of the tumor immune microenvironment, and they demonstrate an exciting potential to develop nanoparticles as cancer immune therapy platforms.


Trastuzumab-labeled nanoparticles demonstrate targeted binding in vitro

We used amine-functionalized starch-coated bionized nanoferrite (BNF) nanoparticles with trastuzumab (BH), a humanized anti–human epidermal growth factor receptor 2 (HER2/neu) monoclonal antibody approved for clinical use in the management of HER2+ breast cancer (Fig. 1A). The ability of trastuzumab to target HER2+ cancer cells in tumors has been validated and well documented, as has its use for nanoparticle-targeting studies (20, 21). The precursor BNF-Plain (BP) nanoparticles comprise a magnetic iron oxide core that is coated with hydroxyethyl starch (core shell) to provide biocompatibility and colloid stability in biological media (1519).

Fig. 1 Particle design and in vitro demonstration of targeting for studies in mice.

(A) Schematic of particle chemistry showing amine functionalization of BP nanoparticles using maleimide precursors for conjugation with thiol moieties of the antibody (trastuzumab). (B) Western blot analysis showing HER2 protein expression by human breast cancer cell lines used in the study. (C) Immunofluorescence results showing HER2 protein surface expression in six human breast cancer cell lines. MDA-MB-231 is a triple-negative ER/PR/HER2- cell line. MCF7/neo and MCF7/HER2 are an isogenic pair with HER2-expressing (MCF7/HER2) variant having a single copy of HER2 gene and HER2- (MCF7/neo), which received a scrambled gene. Other cell lines are wild type and express varying amounts of HER2 protein. (D) In vitro iron content analysis (ferene-s assay) after exposure of cells to BP and BH nanoparticles shows a positive correlation with HER2 protein level and iron uptake in the breast cancer cells. For the assay, cells were incubated at 37°C for 3 hours with BP or BH nanoparticles (0.5 mg/ml) and evaluated for total iron content after washing unbound particles. Untreated cells, Herceptin alone, and BNF-IgG were used as controls. The average of three independent experiments is shown. Statistical differences among BP, BH, and BNF-IgG were obtained by two-tailed Student’s t test (*P < 0.05 and **P < 0.01). (E) Schematic of the overall study design using mouse models of human breast cancers. See text for details.

The choice of 100-nm BNF nanoparticles was motivated from our previous study, which demonstrated higher accumulation by the 100-nm nanoparticles to tumors than with 30-nm nanoparticles, despite the longer blood circulation time of the latter construct (1519). When measured by dynamic light scattering, BP nanoparticles had a mean measured (z average) hydrodynamic diameter of 99 nm (±3) with a mean polydispersity index of 0.07 (±0.02) (table S1). Zeta potential, a measure of surface charge density, was slightly negative (−2.2 ± 0.2 mV) at pH 7.4. Overall, addition of trastuzumab to the BP nanoparticles had only a modest effect on the measured physical properties of the nanoparticles. Several of antibody-labeled nanoparticles were prepared and assayed using a modified in vitro test to confirm selective binding of the BH construct. In all cases, successful binding of antibody was confirmed by a modified bicinchoninic acid assay (BCA) and immunofluorescence (tables S2 to S5 and fig. S1, A and B). A BNF–immunoglobulin G (IgG) construct was synthesized with a nonspecific human polyclonal antibody, as an additional control. The measured physical properties of the BNF-IgG nanoparticles were similar to those of BH nanoparticles (tables S2 and S4).

We limited our selection of cancer models to those for which a stable transmembrane protein/marker is well documented and for which multiple cell lines and mouse models are readily available. In general, HER2+ breast cancer biology has been extensively studied, providing numerous human and mouse cell lines to yield xenograft, syngeneic, and spontaneous models (22, 23). For in vitro and xenograft studies, we selected six human breast cancer cell lines (Fig. 1, B and C, and table S3). HER2 protein expression was verified by Western blotting (Fig. 1B). We used an isogenic pair derived from a HER2- MCF7 parental line, MCF7/HER2 (+), and MCF7/neo (−) (Fig. 1, B and C). The variable total protein and surface expression of HER2 were evident in three HER2+ lines: HCC1954, BT474, and SKBR3 from both Western blotting and immunofluorescence, whereas MDA-MB-231 cells showed no HER2 expression.

Residual iron concentration was measured in cells using a modified ferene-s assay (24) and correlated with HER2/neu expression following exposure to BH nanoparticles. Both iron concentration and HER2/neu expression followed the same progression: MCF7/HER2 < HCC1954 < BT474 < SKBR3 (Spearman correlation coefficient, ρ = 0.89, P = 0.03; Fig. 1D, inset, and fig. S1C), confirming that in vitro targeting occurred via the expected antibody-antigen binding.

Tumor retention of antibody-labeled nanoparticles depends on immune status of the host

We used two immunodeficient strains of mice [athymic nude and nonobese diabetic/severe combined immunodeficiency (NOD/scid) gamma (NSG)] engrafted with five human breast cancer cell lines: two HER2- cell lines (MDA-MB-231 and MCF7/neo) and three HER2+ lines (MCF7/HER, HCC1954, and BT474). The xenograft study design is illustrated in Fig. 1E, and details are provided in Materials and Methods and table S6. Visibly evident 24 hours after injection by discoloration of tumors, BH nanoparticles were retained by tumors to a greater extent than were BP nanoparticles (Fig. 2A).

Fig. 2 Iron content and histopathology analyses demonstrate higher accumulation of antibody-labeled nanoparticles in tumors.

(A) Gross morphology of tumors following intravenous injection with BP or BH nanoparticles shows different tissue color. Darker (black) color indicates greater particle uptake. Tumors from NOD/scid γ (NSG) mice show more BH than BP. Photo credit: Preethi Korangath, Johns Hopkins University. (B) Representative images of HER2 immunohistochemistry (IHC) from breast xenografts showing that expression correlates with in vitro expression. (C and D) Inductively coupled plasma mass spectrometry (ICP-MS) of Fe recovered from tumors excised from mice injected with BH nanoparticles demonstrates consistently higher Fe content than tumors from mice injected with BP nanoparticles regardless of HER2 status of the tumor. Recovered iron was higher in tumors excised from NSG mice (D) than that from athymic nude mice (C) (*P < 0.05, **P < 0.01, and ***P < 0.001). (E and F) Prussian blue–stained tissue slides recovered from the same tumors as in (C) and (D) and digitally analyzed for percent positive area that showed a similar trend as observed with ICP-MS. (G and H) ICP-MS analysis of Fe from the livers showed higher iron content in mice injected with BP nanoparticles than mice injected with BH nanoparticles, mirroring the results of Fe recovered from tumors (**P ≤ 0.01 and ***P < 0.0001).

Volumetric analysis of iron by inductively coupled plasma mass spectrometry (ICP-MS) recovered from tumors grown in nude mice corroborated our observations of gross tumor presentation. HER2 status/expression of tumors was confirmed by immunohistochemistry (IHC) (Fig. 2B). Higher iron concentrations were present in tumors of mice injected with BH relative to phosphate-buffered saline (PBS)– or BP-injected mice (P < 0.001) irrespective of HER2 status (Fig. 2C). In contrast, intratumor iron concentrations measured from mice receiving BP was only slightly higher than PBS-injected controls (MCF7/HER, HCC1954, and BT474; P > 0.05; see Fig. 2C). Iron recovered from nude mice bearing MCF7/neo tumors injected with BH was comparable to those recovered from MCF7/HER tumors. Comparable iron recovery in these two isogenic (HER2+/−) tumor models following injection with BH, which was higher than either PBS- or BP-injected controls, suggests that biological factors other than antibody-antigen binding were responsible for nanoparticle retention. In other words, the BH nanoparticle targeting observed in vitro was not evident in vivo.

This pattern of retention was also measured in tumors recovered from NSG mice (Fig. 2D and fig. S2); however, HER2 expression by the tumor slightly correlated more with BH retention in NSG mice than in nude mice (Fig. 3, A and B). In contrast to results obtained from nude mice, iron recovered from HCC1954 and BT474 tumors in NSG mice was slightly higher than in MCF7/HER, consistent with higher HER2/neu protein expression in these cell lines (Fig. 1, B and C).

Fig. 3 Histopathology analysis of tumors reveals antibody-labeled nanoparticles localize to regions rich with immune cells.

(A) Analysis of Prussian blue–positive (nanoparticle-rich) areas of tumors from nude mice injected with BH nanoparticles reveals only weak correlation with HER2 expression. (B) Conversely, this correlation is stronger in tumors from NSG mice. (C and D) Weak or no correlation was observed between BH nanoparticle presence and CD31+ (vascular endothelium) regions. (E) Representative histology images of sequential sections showing IBA-1+ cells associated with Prussian blue–positive areas in HCC1954 (HER2+) tumors grown in NSG mice and treated with BH (a) hematoxylin and eosin (H&E), (b) Prussian blue, (c) HER2 IHC, (d) IBA-1 IHC, (e) CD-31 IHC, (f) H&E of another area from same tumor, (g) sequential section stained for Prussian blue shows positive staining for iron nanoparticles, and (h) immunofluorescence (IF) staining for IBA-1 shows positivity in the nanoparticle accumulated region. (F and G) Iron recovery from HER2+ (HCC1954) or HER2 (MDA-MB-231) tumors is similar whether BNF nanoparticles have trastuzumab (anti-HER2) or human IgG (polyclonal), suggesting that antibody-antigen binding does not drive intratumor nanoparticle accumulation. ns, not statistically significant.

We analyzed tumor tissue sections stained with Perl’s reagent (also known as Prussian blue) to visualize the nanoparticle-rich regions across all models (Fig. 2, E and F, and fig. S3). The trends observed with gross presentation and ICP-MS were consistent with tumor histopathology (Fig. 2, C to F, and fig. S2) and also revealed notable spatial heterogeneity of iron localization. Nevertheless, all tumor models studied showed significantly more nanoparticle retention when mice were injected with BH, but localization to cancer cells was not evident.

As previously observed, a substantial amount of systemically injected nanoparticles will accumulate in the liver (611, 2527). It is widely held that resident macrophages (liver) and circulating macrophages along with other phagocytic immune cells will sequester nanoparticles of about 100 nm in diameter, clearing them from blood circulation and depositing them into the liver and other organs. Our ICP-MS analysis of iron recovered from the livers showed that all mice injected with nanoparticles exhibited higher iron concentration in the liver than PBS-injected controls. However, the livers of mice injected with BP had higher iron content than the livers of mice injected with BH (Fig. 2, G and H). We conclude that BH retention in tumors (and perhaps other tissues not assayed) contributed to the reduced liver content when compared with BP-injected mice.

Higher tumor retention of Herceptin (Her/trastuzumab)–labeled nanoparticles having varied composition and sizes (15 to 500 nm) following systemic delivery into nude female mice bearing MCF7 tumors has been noted (2831). It is worth emphasizing, however, that MCF7 cancer cells express no HER2 antigen on their membranes, begging the question of the mechanisms of targeting observed in these previous studies. Together, results reported here and elsewhere indicate that retention of nanoparticles in (xenograft human-mouse) tumors may depend on complex biological responses that are intertwined with the host immune system. We note in our results that immune status of the mouse seemed to play a role in nanoparticle retention in tumors, whereas antigen expression by cancer/tumor cells seemed to have very little influence, especially in nude mice (Fig. 2 and fig. S2). Further study was needed to determine nanoparticle association with cell type.

Analysis of xenograft tumor tissues reveals nanoparticles localized with tumor-associated immune cells

We analyzed comparable regions of stained serial tissue sections in detail by scoring to determine whether intratumor nanoparticle localization correlated with tumor-specific factors. Digitally scored Prussian blue–stained sections were compared with manual scoring of the corresponding HER2-stained tumor sections using Spearman’s rank correlation coefficient from mice injected with BH nanoparticles (Materials and Methods). A positive but weak correlation was found between BH localization and HER2/neu protein expression in nude mice (ρ = 0.3827; Fig. 3A). We measured a stronger, positive correlation between BH localization with HER2+ sections in tumors from NSG mice (ρ = 0.8462; Fig. 3B). These results were consistent with both ICP-MS and digital scoring of Prussian blue–stained slides among all tumor models (Fig. 2, C to F) further supporting our observations that immune status of the host animal was an important factor determining BH retention in tumors but not for BP (fig. S4A).

To test whether BH nanoparticle retention in tumors correlated with the tumor microvascular network, we compared Prussian blue–stained areas with corresponding sections stained with CD31 for visualizing the vascular endothelium (32). No correlation was found between BH score and CD31+ score in sections obtained from nude mice (ρ = 0.018; Fig. 3C), but a weak positive correlation was measured in sections obtained from NSG mice (ρ = 0.3241; Fig. 3D). By contrast, slight positive correlations were found with CD31+ regions in both nude and NSG mice injected with BP (fig. S4A).

Both athymic nude and NSG mice lack mature T cells, but NSG mice, in addition, also lack functional components of their innate immune system (table S6) (33). We speculated that subpopulations of innate immune cells in the TME contributed to BH retention. We compared Prussian blue–stained sections with corresponding sections stained for ionized calcium-binding adapter molecule 1 (IBA-1), a pan-(murine) macrophage marker that also labels other myeloid cells including subpopulations of dendritic cells, monocytes, activated neutrophils, and some types of endothelial cells (Fig. 3E) (34). Comparing IBA-1+ tissue sections with Prussian blue–positive regions revealed that antibody-labeled nanoparticles were found in similar locations as IBA-1+ regions within the TME in both nude and NSG mice (HCC1954 tumor grown in NSG mice, Fig. 3E; BT474 tumor grown in NSG mice, fig. S4B). However, we found no significant differences in the content (number) of IBA-1+ cells among any of the tumor models or treatment (fig. S4C).

Next, we tested the notion that antibody-antigen binding to cancer cells does not determine tumor localization of BH to tumors by using BNF nanoparticles labeled with a nonspecific human polyclonal IgG. BNF-IgG nanoparticles were intravenously injected into cohorts of both nude and NSG mice bearing HER2+ (HCC1954) and HER2 (MDA-MB-231) tumors. ICP-MS analysis of tissue iron content of tumors extracted from mice injected with BNF-IgG was similar to that measured from mice injected with BH in both tumor models and immune backgrounds of mice (Fig. 3, F and G, and fig. S5, A and B). These results support that retention of antibody-labeled nanoparticles (i.e., BH or BNF-IgG) was independent of antibody-antigen binding.

From the results obtained across the five human tumor xenograft models in two immunodeficient mouse strains and with two antibody nanoparticle types (trastuzumab and nonspecific IgG), we hypothesized that BNF nanoparticle retention by tumors was determined by active biological processes influenced (or directed) by cells of the innate immune system, residing within the TME and reacting to the presence of an antibody on the nanoparticle surface. Our analysis of xenograft tumors of the IBA-1–stained tissue sections provided no evidence of measurable (aggregate innate) immune cell infiltration into or depletion from the tumors following nanoparticle exposure. To the contrary, the area of IBA-1+ regions among PBS- and nanoparticle-injected cohorts was comparable (fig. S4C), indicating that tumor-associated immune cell subpopulations internalized antibody-labeled nanoparticles (trastuzumab or IgG; see Fig 3, F and G). To test whether macrophages were responsible for these observations, we depleted macrophages by treatment with clodronate liposomes in athymic nude mice growing HCC1954 tumors and injected with BH (35). Unexpectedly, macrophage depletion alone failed to decrease the amount of BH nanoparticles retained in tumors (fig. S5C), suggesting involvement by other immune cells.

Tumors growing in mice having an intact immune system retained more antibody-labeled nanoparticles

BNF nanoparticle localization in tumors across multiple xenograft mouse models suggested that immune status contributed to, and perhaps dominated, nanoparticle retention. To test this concept further, we used a syngeneic tumor model derived from the transgenic huHER2 mouse (Fig. 4A) and transplanted to NSG, nude, and immune competent FVB/N mice. HER2 protein expression in the tumors was confirmed by IHC (Fig. 4B).

Fig. 4 Comparison of nanoparticle accumulation in syngeneic transgenic (HER2+) tumors reveals that strongest correlation between tumor uptake of antibody-labeled nanoparticles (BH) occurs in immune competent mice.

(A) Schema of transgenic huHER2 tumor allograft development and IHC confirmation of HER2 protein expression on cancer cells in tumors. (B) IHC analysis demonstrates that HER2 protein expression in syngeneic huHER2 allografts is comparable among the range of immune strains of mice tested: FVB/N, athymic nude, and NSG mice. (C) Gross appearance of huHER2 allograft tumors grown to 150 to 200 mm3 in FVB/N, athymic nude, or NSG mice 24 hours after they were injected via tail vein with BP or BH nanoparticles shows that BH accumulation is greatest in tumors growing in immune competent host(s). Photo credit: Preethi Korangath, Johns Hopkins University. (D) ICP-MS results showing absolute iron recovery from tumors grown in all mice reveals highest accumulation of BH nanoparticles in FVB/N mice (*P < 0.05, **P < 0.005, and ***P ≤ 0.0001). (E) Histology analysis revealed that Prussian blue–positive area was seen in stromal area and colocalized more with IBA-1+ cells than HER2+ tumor cells.

The intensity of coloration, 24 hours after injection by BH nanoparticles into FVB/N mice, was visibly greater than that displayed by tumors in either NSG or nude mice (Fig. 4C). Iron content analysis by ICP-MS and analysis of Prussian blue–stained slides demonstrated a notable uptake of BH by huHER2 allograft tumors grown in FVB/N mice (Fig. 4D and fig. S6, A and B). Similar to results obtained from xenograft models (Fig 2), FVB/N mice showed retention of less BH in the liver than BP, and higher iron content was detected in the lymph nodes and spleens of both BP- and BH-injected mice (fig. S6, C to E). Prussian blue–positive areas appeared more prominently in stromal regions associated with IBA-1+ areas than in the HER2+ regions (Fig. 4E and fig. S7). These results provided strong evidence that immune status of mouse models is a critical biological variable for nanoparticle targeting studies; however, the nature of this interaction was unclear.

Inflammatory phagocytic innate immune cells residing in the TME display preference for antibody-labeled nanoparticles

Across all models studied, the presence of immune cells within tumors was detected. Colocalization of nanoparticles and IBA-1+ cells occurred at the tumor periphery (Fig. 3E and fig. S3) in xenograft tumors and in tumor-stromal interfaces in the immune competent huHER2 allograft model (Fig. 4E and fig. S7). It has been well documented that the cancer tissue boundary of tumors often exhibits proinflammatory features (36). We hypothesized that tumor-associated immune cells exhibiting an “inflammatory phenotype” preferentially sequestered and retained antibody-labeled nanoparticles.

To test this hypothesis and to further elucidate the mechanism of nanoparticle retention in the TME, we performed tests in vitro with murine macrophages and neutrophils. Macrophages were activated with lipopolysaccharide (LPS) and interferon-γ (IFN-γ) to mimic a T helper 1 (TH1)–type induction (M1) or with interleukin-4 (IL-4) to mimic a TH2-type induction (M2). When exposed to either BP or BH, macrophages always sequestered more BH than BP; however, M1 macrophages sequestered significantly more nanoparticles, especially BH (Fig. 5A). Uninduced neutrophils showed no preference for either construct; however, when activated with LPS (TH1-type induction), neutrophils demonstrated significantly greater preference for BH (Fig. 5B).

Fig. 5 In vitro and in vivo analysis shows preferential uptake of nanoparticles in immune cells.

(A) Undifferentiated RAW 264.7 (M0) or differentiated M1 or M2 (LPS + IFN-γ or IL-4, respectively) macrophages were incubated for 24 hours with BP or BH, and ferene-s assay was conducted to measure the total amount of iron uptake per cell. As a control, BP and Her, added together, were also used. As shown in the figure, BH nanoparticles were taken up more significantly than BP by macrophages irrespective of their phenotype. The uptake was significantly higher in M1 macrophages than either M0 or M2, which indicates that proinflammatory macrophages take up more BP and BH nanoparticles with preference toward BH. (B) Likewise, LPS-activated neutrophils (induced) preferentially sequestered BH over BP, whereas no difference in uptake was observed with naïve bone marrow neutrophils (uninduced). (C) Total cell count obtained from magnetically separated BP- or BH-injected tumors shows significant difference. Immune competent FVB/N mice (n = 3 per group, two tumors each) bearing huHER2 tumors were intravenously injected with BP or BH. After 24 hours, tumors were harvested and digested to isolate single cells and were magnetically separated to collect nanoparticle-associated cells to determine the total cell count. (D) Analysis of magnetically sorted cells obtained from in vivo tumors showed that nanoparticles were associated with immune cells, not tumor cells. Immune competent FVB/N mice (n = 5 to 8 per group) bearing huHER2 tumors were intravenously injected with PBS, BP, or BH. After 24 hours, tumors were harvested and digested to isolate single cells and were magnetically separated to collect nanoparticle-associated cells for analysis by flow cytometry. Gating strategy is provided in fig. S8. Cell numbers measured from BP- and BH-injected mice are shown as change in ratio relative to PBS-injected mice (PBS ratio = 1). (a) Populations of cancer cells were not changed in nanoparticle-associated cancer cells. Ratios of NK cells (b), monocytes (c), TAMs (d), neutrophils (e), and dendritic cells (f) are increased in nanoparticle fractions, suggesting uptake of nanoparticles by immune cells rather than tumor cells. (*P ≤ 0.05, **P ≤ 0.01, and ***P < 0.001).

Magnetic nanoparticles provide a unique tool to query biological responses because they enable magnetic sorting to isolate specific cell populations containing the nanoparticles. To further elucidate the in vivo tumor immune response to BNF nanoparticle exposure, tumor digests were placed on a permanent magnet. Cells containing nanoparticles were sedimented, whereas cells devoid of nanoparticles remained suspended. Sedimented (nanoparticle-associated) cells were collected and analyzed for total number (Fig. 5C). Consistent with in vitro results, the total number of cells containing iron was higher in tumors of mice injected with BH than in those injected with BP. To distinguish among tumor-associated cell populations that sequestered nanoparticles, both sedimented (nanoparticle associated) and suspended (supernatant, no nanoparticle) cells were collected and analyzed by polychromatic flow cytometry. Figures S8 and S9 provide graphical gating strategy and complete results of analysis, respectively. Results of magnetic sorting of equal (initial) numbers of tumor-derived cell populations are displayed in Fig. 5D (a to f) as ratios of cell number by type and fraction relative to cell numbers obtained from PBS-injected mice. PBS ratios are expressed as unity and all others as <1 or >1 depending on the number of cells detected in each fraction. Among cancer cells, it is notable that for either BP or BH, numbers were lower than from PBS-injected controls, indicating little nanoparticle association with the HER2+ cancer cells (Fig. 5D, a). This is consistent with histopathology (Fig. 3E). Following intravenous delivery, evidence indicates that nanoparticle association with cancer cells was minimal regardless of HER2+ expression, further confirming the different performance of antibody-labeled nanoparticles in vivo versus in vitro.

Systemic exposure to BNF nanoparticles alters the immune TME

On the basis of the evidence, nanoparticle retention in the studied models was likely determined by tumor-associated leukocytes, but what effect did systemic exposure to nanoparticles have on the tumor immune microenvironment? We used polychromatic flow cytometry to identify changes of individual tumor immune cell populations in huHER2 allograft tumors growing in FVB/N mice following injection with nanoparticle or free antibody (Fig. 6A and fig. S8, A and B). Twenty-four hours after intravenous injection, we measured a slight decrease of live cell populations in tumors derived from mice receiving either BP or BH relative to PBS-injected controls. No measurable differences were detected in cancer cell populations among the four cohorts, but a significant decrease in CD45+ population was noted (fig. S9B, a to c).

Fig. 6 Flow cytometry analysis demonstrates impact of nanoparticles on TME in response to intravenous nanoparticle delivery.

Immune competent FVB/N mice (n = 5 to 8 per group) bearing huHER2 tumors were intravenously injected with PBS, BP, BH, or Herceptin (Her). After 24 hours, tumors were harvested and digested to isolate single cells and evaluated by polychromatic fluorescence-activated cell sorter (FACS). Gating strategy is provided in fig. S8. (A) Relative decreases in T cell (a) and B cell (b) populations were observed following injection of nanoparticles. By contrast, relative increases were measured in many innate immune cell populations within the TME: NK cells (c), neutrophils (d), TAMs (e), and monocytes (f) 24 hours after nanoparticle exposure. Except for TAMs, no significant increase was seen in any other immune cell population after Her injection. (*P ≤ 0.05 and **P ≤ 0.01). (B) Graphic representation of distributions of nanoparticle-associated CD45+ immune cells among the cohorts.

Nanoparticle exposure induced many changes across a number of tumor immune cell lineages, with a notable decrease in T cells and an increase in the relative fraction (i.e., ratio) of innate immune cells initiating a restructuring of the immune compartment of the TME (Fig. 6A, a). B cell populations also decreased in BH- and Her-treated groups (Fig. 6A, b). Relative to PBS controls, natural killer (NK) cell and monocyte fractions increased following BH injection but not in mice receiving BP or Her (Fig. 6A, c and f). Populations of other phagocytic innate immune cells, specifically neutrophils, and tumor-associated macrophages (TAMs) increased with either BP or with BH injection relative to controls (Fig. 6A, d and e, and fig. S9B), but dendritic cell populations remained relatively unchanged 24 hours after injection (fig. S9B, d) as did the fraction of γδT cells (GD T cells) (fig. S9B, e). However, we found no evidence in histology data indicating that depletion or infiltration of innate immune cells carrying nanoparticles to or from tumors occurred after nanoparticle injection, suggesting “capture” of nanoparticles by the resident population(s) of innate immune cells in the TME (Fig. 3 and fig. S4C) (37). Nevertheless, for conclusive quantification of this process, further study is needed. Exposure to free trastuzumab (Her) elevated TAMs, reflecting a specific interaction (Fig. 6A, e).

Trastuzumab is a humanized monoclonal antibody with a human IgG1 (hIgG1) that can elicit a response in murine macrophages (38). Furthermore, it is recognized that Fcγ receptors on murine macrophages can recognize hIgG1 (38), and the response observed in our flow cytometry with free trastuzumab (Her) is consistent with this observation (Fig. 6A, e). Note that, however, macrophages were the only tumor immune population that elevated within 24 hours following injection with free trastuzumab, whereas multiple immune cell subpopulations responded to BP and BH exposure (Fig. 6, A and B, and fig. S9B). The tumor immune response to BH was more complex than that to free trastuzumab (Her)—including T cells, NK cells, monocytes, neutrophils, dendritic cells, and macrophages—and it was similar to that of BP. Thus, while the potential exists for specific interactions between murine macrophages and hIgG1-containing nanoparticles, our evidence demonstrates that labeling the surface of a nanoparticle with a hIgG1 monoclonal antibody alters the immune response to recognize the nanoparticle-antibody construct as an entity distinguishable from free antibody.

The data indicate that, in addition to macrophages (TAMs), many other lineages of phagocytic innate immune cells—NK cells, monocytes, neutrophils, and dendritic cells—reside in the TME sequestered nanoparticles (Fig. 5D, b to f, and fig. S9A, b to i). It seemed that an intact immune system is a critical component in determining the retention of nanoparticles in solid tumors. To challenge this notion, we pretreated tumor-bearing mice with a pan-leukocyte inhibitor, azathioprine (39, 40), before injecting with BH. Iron recovered from tumors in azathioprine-treated mice was significantly reduced and similar to BP-injected mice (fig. S10, A and B), confirming the role of a wider immune involvement in nanoparticle retention.

These results support a model that tumor-associated phagocytic immune cells significantly influence the degree of retention of systemically delivered nanoparticles within the TME. Furthermore, our results demonstrate that an intact host immune system can manifest decidedly different tumor retention when compared with comparable immunodeficient models, raising an important question about clinical relevance of studies performed in the latter. Depending on environmental chemical cues, tumor-associated leukocytes may display a greater sensitivity to the chemical signatures of nanoparticles than their counterparts residing in other tissues. This offers potential for tumor targeting with nanomedicines.

In a complex manner, while the restructuring of the immune compartment of the TME, likely mirroring a systemic immune response to nanoparticle exposure, was similar for both BH and BP nanoparticles, it is only the BH nanoparticles that were significantly retained within the TME. These complex and seemingly contradictory immune responses may indicate potential for anticancer effects.

Systemic exposure to nanoparticles inhibits tumor growth and increases CD8+ T cell infiltration

To explore the potential clinical relevance of our findings, we used the huHER2 allograft tumor model to ascertain effects of nanoparticle exposure on tumor growth in FVB/N and athymic nude mice. Five days after implantation of huHER2 tumors, FVB/N or athymic nude mice received a single intravenous injection of PBS, BP, BH, or Her as previously described. Exposure to either BP or BH significantly delayed tumor growth in FVB/N mice but not in athymic nude mice (Fig. 7, A to C, and fig. S11, A to C). As expected, trastuzumab alone was effective to significantly inhibit tumor growth in both FVB/N and athymic nude mice, however, its mode of action involves direct binding via HER2 antigen to cancer cell membranes. Our evidence shows that neither BP nor BH nanoparticles associated appreciably with cancer cells in vivo; thus, the therapeutic effect seen only in FVB/N mice due to nanoparticle exposure must involve an alternate mechanism that we hypothesized was mediated by the adaptive immune system. To gain further insight, we repeated the experiment in FVB/N mice and conducted flow cytometry analysis of immune populations in tumors 3, 7, and 14 days after injection. Beginning at 7 days after injection, significant increases in activated T cells (CD3+/CD4+/CD8+) were measured in tumors, reversing the depletion observed at 24 hours and 3 days and supporting a model of immune-mediated tumor suppression induced by systemic exposure to nanoparticles (Fig. 7, D and E, and figs. S11D to S14). Immune cells known to be involved in adaptive immune signaling, i.e., dendritic and γδ T cells, displayed a complex time-dependent pattern—increasing to day 3 and decreasing thereafter—consistent with adaptive immune signaling response (Fig. 6 and figs. S9 and S13) (41, 42). On the other hand, phagocytic effector immune cells, i.e., macrophages and monocytes, initially displayed relatively elevated numbers at day 1 but displayed no such increases afterward relative to PBS controls (Fig. 6 and fig. S9 and S14). These complex and time-dependent immune cell responses observed in the TME resemble systemic responses observed in mice following acute and nonlethal infection by some pathogens, i.e., Listeria monocytogenes, which can also lead to anticancer immune stimulation (41, 42). Note that both BH and BP nanoparticles induced similar effects on tumor immune cell populations and on tumor growth, despite the fact that BP nanoparticles were not significantly retained within the tumor. This suggests that exposure to nanoparticles has the potential to induce both systemic and local (tumor) effects, which bear further study and offer potential for developing another paradigm in cancer nanomedicine (fig. S15).

Fig. 7 Nanoparticles stimulate T cell infiltration into the TME, inhibiting tumor growth.

(A) Female FVB/N mice bearing huHER2 allograft tumors (n = 7 to 18 per group) were intravenously injected with either PBS, BP, BH (5 mg per mouse), or Herceptin (175 μg per mouse) 5 days after tumor implantation (day 0). Growth of the tumors was monitored by caliper measurements twice per week for 4 weeks (means ± SEM). Final tumor weight is given in inset (**P < 0.005 and &P ≤ 0.0001). (B) On day 28, all mice were euthanized, and representative images of tumors are shown. Photo credit: Preethi Korangath, Johns Hopkins University. [C (a and b)] Female athymic nude mice bearing huHER2 allograft tumors (n = 6 to 7 per group) were similarly treated as above, and 3 weeks of tumor growth and tumor weight is reported (means ± SEM, *P < 0.05). [D (a and b) and E (a and b)] Flow analysis of tumors: As in (A), mice (n = 5 per group) were intravenously injected with either PBS, BP, BH (5 mg per mouse), or Herceptin (175 μg per mouse) on the 10th day after tumor implantation. Seven days after injection, mice were euthanized; tumors were harvested, and single cells were isolated and evaluated by FACS. Infiltration of CD3+ T cells with increases in CD8+ T cells was measured following nanoparticle exposure, likely leading to growth inhibition observed in (A) (*P < 0.05). FITC, fluorescein isothiocyanate.

In summary, targeting nanoparticles has been a topic of considerable debate, even controversy, in the cancer nanomedicine community (17, 16, 2531). In most previous studies, the biology of tumor and/or host was not studied in detail with analysis of tissue histology and flow cytometry, thus motivating our efforts to understand the role of host biology in nanoparticle-tumor interactions (610). Across all models studied, we found strong evidence pointing to immune status of the host as a key factor determining the retention of antibody-labeled nanoparticles in tumors. Using an immune intact model, we discovered that the retention of nanoparticles in tumors was dominated by multiple lineages of tumor-associated immune cells when the nanoparticles included an antibody and found no in vivo evidence supporting a mechanism of antibody-antigen binding (i.e., the mechanism operating in vitro) to cancer cells in the tumor. Yet, the amount of nanoparticle retained by the tumor within 24 hours was most pronounced in an immune intact model, further emphasizing the significance of an intact immune system in studies of nanoparticle delivery to solid tumors. Our results demonstrate that the host immune system can be a substantial factor in studies of cancer nanomedicine and that macrophages are only one among many immune cell lineages that determine nanoparticle fate. It was only when we pharmacologically inhibited the entire host immune system that we measured a reduced retention of the BH nanoparticles. While these findings reveal new insights, they also raise many questions regarding complexities of nanoparticle–immune cell interactions in vivo across the many biological models used in cancer research and how immune cell receptors distinguish among nanoparticle coatings.

Related to this, but in a different manner, we observed that the immune response to nanoparticle exposure measured in tumors was equally profound and seemed insensitive to nanoparticle composition (BP or BH). As measured by population changes of immune cells in the TME, the immune response included an initial T cell depletion and later T cell infiltration into the tumor with significant tumor growth inhibition.

The presence of immune cells within an established solid tumor implies that immune cells are performing surveillance and homeostasis functions to support the growth and maintenance of the tumor. Our results show that exposure to nanoparticles can disrupt this delicate balance, potentially enabling a transient immune recognition of the tumor. In an immune-intact model of cancer, the systemic delivery of a nanoparticle construct can initiate a complex immune response, which can affect tumor growth regardless of retention. These results highlight the notion that the biology of the host and cancer tumor forms an interconnected and inextricably linked biological network that interacts in complex ways to determine the biological fate and retention of nanoparticles. Host immune status and, consequently, composition of the immune compartment(s) within the TME are critical variables in developing and testing the performance of cancer nanomedicines. Results presented here motivate more questions of mechanism of host and tumor immune cell interactions with nanoparticles. They also point to new possibilities for nanoparticle anticancer immunotherapy technologies.


Cell lines and reagents

MDA-MB-231 [ER/PR/HER2 (−) negative], MCF7 [ER/PR (+) positive/HER2 (−) negative], and BT474 [ER/PR/HER2 (+) positive] were purchased from the American Type Culture Collection (ATCC; Manassas, VA) and maintained according to the supplier’s recommendations. They were grown in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS). HCC1954 [ER/PR (−) negative/HER2 (+) positive] was grown in RPMI containing 10% FBS. MCF/neo and MCF7/HER were provided by K. Osborne (University of Texas Health Science Center). All cell lines were authenticated using short tandem repeat analysis (data provided upon request) and matched against ATCC and Deutsche Sammlung von Mikroorganismen und Zellkulturen databases to ensure the genetic origins.

Iron oxide nanoparticles

The nanoparticles used for this study are commercially available aqueous suspensions of hydroxyethyl starch–coated magnetite (Fe3O4) core-shell particles (BNF; Micromod Partikeltechnologie GmbH, Rostock, Germany). The synthesis and physical characterization of the BNF particles have been extensively documented (1519). Briefly, BNF particles were produced by precipitating ferric and ferrous sulfate salts from solution at high pH in a high-pressure homogenization reaction vessel, which controls both crystal formation and aggregation. According to the manufacturer, they have a mean hydrodynamic diameter of ~100 nm and an iron content of >50% (w/w) [or iron oxide of >70% (w/w)].

Physical characterization of BNF nanoparticles

The mean hydrodynamic diameter of the magnetic iron oxide nanoparticles (BNF) and their zeta potential were measured in 1 mM PBS buffer (pH 7.4) with a Zetasizer Nano ZS90 (Malvern Instruments Limited, UK) at an iron concentration of 0.1 mg/ml. The mean particle diameter Z(Ave) is given as a result of the cumulative analysis of the autocorrelation function. The polydispersity index is a measure of the quality of the size distribution. Monodisperse suspensions have a polydispersity index of <0.25.

Synthesis of BNF-trastuzumab nanoparticles

The monoclonal anti-HER2/neu antibody (Her), or trastuzumab (trade name) (Genentech, South San Francisco, CA), was purchased from Johns Hopkins Pharmacy and was shipped to micromod for conjugation with BNF particles to form BH. The Her was formulated according to the prescribing information. The lyophilized powder that contained 440 mg of Her was dissolved in 20 ml of bacteriostatic water for injection (provided). The Her solution was purified by washing with PBS buffer (pH 4) using a desalting column (PD-10, GE Healthcare, UK) to remove the stabilizing agents. The obtained Her solution was thiolated with Traut’s reagent (2-iminothiolane) as follows: The antibody solution (390 μl, 1.7 mg/ml in PBS buffer) was mixed with 160 μl of 1.4 mM 2-iminothiolane in 450 μl of PBS-EDTA buffer. After shaking for 1 hour at room temperature, the excess of 2-iminothiolane was removed by washing with PBS-EDTA buffer (PBS buffer, 1 mM EDTA) in a desalting column (G-25, GE Healthcare, UK). In parallel, an aqueous suspension of 80-nm BNF-starch nanoparticles with amino groups on the surface (2.25 ml, [Fe] = 8.0 mg/ml; product code: 10-01-801, micromod Partikeltechnologie GmbH) was mixed with 250 μl of 10× PBS-EDTA buffer. Sulfosuccinimidyl 4-(N-maleimidomethyl) cyclohexane-1-carboxylate (sulfo-SMCC) (3.6 mg) was dissolved in 100 μl of dimethyl sulfoxide and added to the BNF-starch suspension. After 1 hour of shaking at room temperature, the excess of sulfo-SMCC was removed by washing with PBS-EDTA buffer in a PD-10 desalting column. The maleimide-functionalized nanoparticles were mixed with the thiolated antibody solution and shaken for 3 hours at room temperature. Then, 200 μl of 20 mM cysteine solution was added to quench the remaining maleimide groups on the nanoparticle surface. Last, the nanoparticles were washed by magnetic separation in a high-gradient magnetic field column (QuadroMACS with LD columns, Miltenyi Biotec GmbH, Bergisch-Gladbach, Germany) with 5 ml of PBS-Tween buffer (pH 7.4, 0.05% Tween 20) and 5 ml of PBS buffer (pH 7.4) per column filling. The magnetic column was removed from the magnet, and the nanoparticles were eluted with 2 ml of water per column filling. The high gradient magnetic field (HGMF) wash was repeated until the suspension was completely washed. The suspension was filtered using 0.22-μm polyethersulfone filter (Carl Roth GmbH, Karlsruhe, Germany).

After conjugation, BH nanoparticles were rigorously characterized for their physical and biological properties in vitro to ensure nanoparticle stability, and BNF-Her binding was successful and retained sufficient protein. Antibody immunoreactivity of the BH construct was separately tested using a cell culture–based assay (see below).

The iron content of the antibody-conjugated nanoparticles (BH) was determined after the digestion of a 20 μl of sample with 80 μl of concentrated HCl. After addition of 4.9 ml of a citrate phosphate buffer (pH 3.6), the iron concentration was calorimetrically determined with the Spectroquant Kit (Merck, Germany) against a Titrisol Iron Standard (Merck, Germany).

The amount of the conjugated antibody in the sample was measured by a modified BCA method. The BCA reagents were obtained from Thermo Fisher Scientific (Germany). The calibration curve was obtained by adding increasing amounts of an albumin standard solution to aminated BNF-starch particles (without antibody on the surface) at a constant iron concentration of 0.25 mg/ml. The antibody-conjugated nanoparticles were adjusted to the same iron concentration of 0.25 mg/ml and developed with the BCA reagent together with the calibration curve for 2 hours at 37°C.

Synthesis of BNF-IgG nanoparticles

Polyclonal normal hIgG was purchased from R&D Systems (Minneapolis, MN) for conjugation with BNF nanoparticles for BNF-IgG nanoparticles. Methods to conjugate the IgG antibody to BNF nanoparticles were same as for trastuzumab, except that proportions of reagents were altered to accommodate differences between the antibodies. The lyophilized hIgG (2 mg) was dissolved in 1 ml of PBS buffer (pH 7.4) and purified by washing with PBS buffer (pH 4) using a desalting column (G-25, GE Healthcare, UK). The antibody solution used was 510 μl (1.3 mg/ml) in PBS buffer and was mixed with 160 μl of 1.4 mM 2-iminothiolane in 330 μl of PBS-EDTA buffer. After shaking for 1 hour at room temperature, the excess 2-iminothiolane was removed by washing with PBS-EDTA buffer in a desalting column. In parallel, BP nanoparticles with amino groups on the surface were prepared as described above. The maleimide-functionalized nanoparticles were mixed with the thiolated antibody solution, reacted, washed, and purified as above.

Ferene-s assay

The detailed protocol for conducting the modified ferene-s measurement of iron associated with cells after exposure to BNF nanoparticles has been previously described (24). Briefly, cells were trypsinized and washed with PBS thoroughly and were incubated at 37°C with BP (0.5 mg/ml), BH, or trastuzumab (Her 2 μg/ml) alone for 3 hours in growth media (DMEM + 10% FBS) with occasional shaking/tapping of tubes to maximize distribution and prevent settling of cells. After incubation, cells were pelleted by centrifugation and washed with PBS to remove unbounded particles and again pelleted by centrifugation. This washing with PBS was repeated three more times. The final cell pellet was resuspended in PBS and counted using a Cellometer (Nexcelom Bioscience, Lawrence, MA) to obtain the total number of cells. The cells in the tubes were then centrifuged, and the supernatant was removed to add working solution (acetate buffer with ascorbic acid). Cell pellets were digested in working solution by incubating at room temperature for at least 20 hours before reading in a colorimeter. A known quantity of ferene-s was used along with other external standard reference materials to quantify the iron concentration of the test samples according to previously described procedures (24). For the entire study, we used only those batches of BH showing more than fourfold retention by SKBR3 cells, as measured by iron concentration with the ferene-s assay when compared to BP (table S5). In addition, we used MDA-MB-231 (HER2-, control) to confirm that nonspecific binding of BH particles by those cells was minimal (<1 pg of Fe per cell).

Fluorescence microscopy

Cells were trypsinized and washed in PBS and incubated in DMEM + 10% FBS at 37°C with trastuzumab (2 μg/ml) for 3 hours with occasional shaking/tapping of tubes to maximize distribution and prevent settling of cells. After incubation, cells were washed four times with PBS and plated on poly-lysine–coated coverslips in six-well plates. After overnight incubation, they were washed with PBS, fixed with methanol for 10 min, and blocked with 1% bovine serum albumin for 30 min at 37°C. Dye-labeled secondary antibody (anti-human Alexa Fluor 488, Life Technologies, Eugene, OR) was added and incubated for 1 hour in the dark at room temperature, followed by washing three times in PBS and mounting with mounting media containing DAPI (4′,6-diamidino-2-phenylindole). They were then visualized and photographed using a fluorescent microscope (Zeiss Axioimager Z1, Carl Zeiss Microscopy GmbH, Jena, Germany). To visualize BNF-HER nanoparticles alone, 30 μl of BNF-HER or BP nanoparticles was separated on a magnet for 2 hours at 4°C. The particles suspended in 1 ml of PBS volume and the concentration of BH nanoparticle suspensions were incubated with anti-human Alexa Fluor 488 secondary antibody (1:1000) for 1 hour at room temperature. The particles were then separated on a magnet for 1 hour, washed with PBS, and dropped on a clean slide to mount and visualize with a fluorescent microscope.

Western blotting

Cells were lysed with radioimmunoprecipitation assay buffer (Sigma-Aldrich, St. Louis, MO) containing protease and phosphatase inhibitors on ice for 30 min. The lysates were centrifuged at 13,000 rpm for 15 min. The supernatant was collected and quantified by BCA (Thermo Fisher Scientific, Waltham, MA) assay. Thirty to 50 μg of total protein were used for SDS–polyacrylamide gel electrophoresis gel after being heated with sample buffer. The proteins were then transferred to nitrocellulose membranes. After blocking with 5% milk solution in PBS-T (1% Tween 20) for 30 min, the membranes were blotted with primary antibody (anti-human HER2 antibody, 1:1000; Cell Signaling Technology, 29D8) overnight and with secondary horseradish peroxidase (HRP)–conjugated antibody (GE Healthcare, UK) for 1 hour. The membranes were developed using chemiluminescence reagent (Amersham Biosciences, Marlborough, MA).

Macrophage activation

RAW264.7 cells were purchased from the ATCC (Manassas, VA) and maintained in DMEM with 10% heat-inactivated FBS. Low-passage cells were used for the study (P3 to P5). For M1 macrophage activation, cells were treated with LPS (100 ng/ml; Sigma-Aldrich, St. Louis, MO) and IFN-γ (50 ng/ml; Miltenyi Biotech, Germany) for 24 hours. To differentiate cells into M2, phenotype cells were treated with IL-4 (10 ng/ml; Miltenyi Biotech, Germany) for 24 hours (43). Induced and uninduced cells (1 million) were collected and treated with either BP or BH nanoparticles (0.5 mg/ml) or cotreated with BP and Her (16.3 μg/ml; equivalent to protein content of BH) for 24 hours. After incubation, cells were washed thoroughly with PBS four times and processed for iron content analysis with the ferene-s assay as described above. Experiments were repeated three times.

Neutrophil activation

Neutrophils were activated in vivo with LPS by the method described by Rönnefarth et al. (44). Briefly, 50 μg of LPS was intraperitoneally injected into FVB/N mice (n = 3). After 18 hours, activated peritoneal neutrophils were collected by injecting 5 ml of PBS to peritoneum, cells were harvested, and red blood cells (RBCs) were lysed with ammonium-chloride-potassium (ACK) lysis buffer and thoroughly washed. Naïve neutrophils were prepared using methods described by Mócsai et al. (45). For this, bone marrow cells were collected to Hanks’ balanced salt solution (HBSS) from femur and tibia of FVB/N mice (n = 3). RBCs were lysed from bone marrow cells with ACK lysis buffer, and cells were passed through a 70-μm strainer. These cells were then centrifuged after layering on 62.5% freshly prepared Percoll in HBSS for 30 min at 1000g without brake. The cloudy pellet of neutrophils was collected. Uninduced bone marrow–derived neutrophils and activated peritoneal-derived neutrophils were incubated with BP or BH nanoparticles (0.5 mg/ml) for 24 hours, and ferene-s assay was conducted to measure the amount of iron uptake per cell as described above.

In vivo orthotopic xenograft experiments

All animal studies were approved by the Institutional Animal Care and Use Committee at Johns Hopkins University and were conducted using female mice. All mice were fed normal diet and water ad libitum. They were maintained in the normal 12-hour light/12-hour dark cycle. All animals were closely monitored for any distress or pain throughout the study period. The weight range of animals during the study was 20 to 30 g. Strains of mice used in all studies were athymic nude (Charles River Laboratories, Frederick, MD), NSG (Sydney Kimmel Comprehensive Cancer Center colony, Johns Hopkins University School of Medicine, Baltimore, MD), and FVB/N (Jackson laboratory, Bar Harbor, ME); all mice were aged 6 to 8 weeks. The characteristics of cell lines and mice used are provided above and in tables S3 and S6. A schematic of the xenograft tumor study design is provided in Fig. 1E. An overview of the numbers of mice divided by strain and group used for the studies is provided in table S10. Depending on cohort, 3 × 106 MDA-MB-231 or HCC1954 or 5 × 106 MCF-7(HER/neo) or BT474 cells were suspended in 50 μl of PBS and Matrigel (1:1) and injected into the fourth mammary gland on either side of female mice under anesthesia. For MCF-7(HER/neo) and for BT474 xenograft studies, mice received estrogen by implanting a 60-day release estrogen pellet (0.72 mg per pellet; Innovative Research of America, Sarasota, FL) 5 days before cell line injection on the dorsal neck region through a small subcutaneous insertion made using sterile scissors while mice were under ketamine/xylazine anesthesia[ketamine (10 mg/ml) Vedco Inc., St. Joseph, MO] and xylazine (2 mg/ml; Lloyd Inc., Shenandoah, IA) mixed in sterile PBS and intraperitoneally injected at 0.01 ml/g body weight. Tumor volume was monitored by caliper measurements twice weekly when tumors became palpable. When the measured tumor volume was 125 to 200 mm3, mice were randomly assigned into cohorts containing five animals in each group. Group I received intravenous (tail vein) injections of PBS and served as (negative) control. Group II received intravenous injections to tail vein of BP (5 mg of Fe per animal), and group III received intravenous tail vein injections of BH (5 mg of Fe per animal). Group IV received injections of BNF-IgG (intravenous tail vein injections; 5 mg of Fe per animal) only for mice bearing either MDA-MB-231 or HCC1954 xenografts. The total volume of injection was 150 μl in all cases. Twenty-four hours after injection, all mice were euthanized to collect tumors and liver for analysis.

Macrophage depletion

Athymic nude mice growing HCC1954 tumors (n = 3 with two tumors each) were treated with two consecutive doses of clodronate liposome (CL) (Liposoma, Netherlands) via intraperitoneal (300 μl per animal) injection. After the second dose of CL, BH nanoparticles were injected (5 mg of Fe per mouse intravenously) and euthanized 24 hours later to harvest tumors for ICP-MS.

Inductively coupled plasma mass spectrometry

The second half of each tumor and whole livers were weighed, lyophilized, and stored at −20°C until analysis by ICP-MS using methods previously described (46). Briefly, each tissue sample was transferred to a 7-ml Teflon microwave digestion vessel (Savillex Corporation, Eden Prairie, MN) to which 1 ml of optima-grade HNO3 (Fisher Scientific, Columbia, MD) was added. The vessel was sealed and placed into a 55-ml Teflon microwave digestion vessel (CEM Corporation, Matthews, NC) to which 10 ml of ultrapure H2O (Millipore Corporation, Billerica, MA), and samples were digested in a MARS5 Xpress microwave (CEM Corporation, Matthews, NC) using a single-stage ramp-to-temperature of 15-min ramp to 130°C with a hold of 30 min. After cooling, each sample was diluted: 35 μl of sample digest and 300 μl of HNO3 were added to 14.665 ml of ultrapure H2O to achieve a final HNO3 concentration of 2% (w/v). External reference standards scandium (CPI Incorporated, Santa Rosa, CA) and Seronorm Trace Elements Whole Blood (SERO AS, Billingstad, Norway) were added to normalize instrument counts and sample iron content, respectively. In addition, four reagent blanks were digested and analyzed in each run to correct for background iron content.

An Agilent 7500ce ICP-MS (Agilent Technologies, Santa Clara, CA) was used to measure iron content of each sample. Measurements were blank-corrected using the average iron value of the reagent blanks and corrected using external standard reference materials. An eight-point calibration curve (0, 1, 5, 10, 50, 100, 500, and 1000 μg/liter) was obtained from Standard Reference Material (SRM) measurements. The analytical limit of detection (LOD) was calculated by multiplying the SD of the lowest detectable calibration standard (1 μg/liter) by three. For samples with values below the analytical LOD, one-half of the LOD was substituted (46).

Tissue IHC

Fresh tumors were fixed in 10% formalin and sectioned on positively charged slides. For HER2 staining, a VECTASTAIN ABC kit (Vector Laboratories, Burlingame, CA) was used to perform IHC. After hydration with serial dilutions of ethanol, antigen retrieval was performed using 10 mM citrate buffer. The sections were then treated with 3% hydrogen peroxide for 10 min and incubated with normal serum to block nonspecific binding. The sections were later incubated overnight with anti-human HER2 antibody (1:400; Cell Signaling Technology, 29D8). Secondary antibody (provided in the kit) was added the next day after washing, followed by incubation with ABC reagent and developed with 3,3’-Diaminobenzidine (DAB) (DAB peroxidase substrate kit, Vector Laboratories, Burlingame, CA) reagent and counterstained with hematoxylin (Dako North America Inc., Carpinteria, CA.) as specified by the manufacturer. For CD31 (Dianova, DIA 310), and IBA-1 (Wako, 019-19741), after deparaffinization and hydration, the slides were steamed in HTTR or EDTA buffer for 45 min in a steamer followed by washing in PBS containing Tween. They were then blocked in peroxidase solution and incubated with CD31 (1:40) or IBA-1 (1:2500) antibody for 45 min at room temperature. After washing, sections were incubated with secondary antibody (PowerVision Poly-HRP anti-Rabbit IHC Detection Systems Novocastra, Leica Biosystems, Buffalo Grove, IL) for 30 min at room temperature followed by washing. The slides were then washed and developed with DAB fast (Sigma-Aldrich, St. Louis, MO) for 20 min at room temperature and counter stained with hematoxylin.

Tissue histology imaging acquisition and analysis

One-half of each tumor was fixed with 10% formalin and submitted for paraffin embedding and sectioning for hematoxylin and eosin (H&E) staining, Prussian blue (also known as Perls reagent) staining to visualize nanoparticle (iron oxide) distributions, and IHC (HER2, CD31, and IBA-1). All stained slides were evaluated by a pathologist (B.W.S.) and quantitated in a blinded study. For manual analysis, HER2 immunostains were semiquantiatively scored to determine the percentage of tumor cells with positive, membranous staining. For automated image analysis, whole slides were digitized using the Aperio ScanScope At or CS system (Aperio, Vista CA) at 40× magnification. Analysis was performed using Aperio ImageScope software (version with the included Positive Pixel Count algorithm. Images were manually annotated to select a region of interest representing a full cross section of each graft and a 50-μm border of surrounding subcutaneous tissue. Artifacts and necrotic regions of the tumor were excluded from analysis. Default hue values (brown, positive; blue, negative) were used for immunostains (DAB Chromogen) and were adjusted for Prussian blue (blue, positive; pink, negative). Digital analysis settings that were used are provided in tables S7 and S8. One slide per condition per tumor was analyzed, and results represent as percent positive pixels over negative pixels in region of interest.

huHER2 allograft transplantation model

Transgenic (huHER2) mice (FVB/N background) that develop mouse mammary tumor virus–driven mammary-specific human HER2–overexpressing tumors were provided under a material transfer agreement (Genentech, South San Francisco, CA). These mice are well characterized for their tumor development and response to trastuzumab as described elsewhere (22, 23). The primary tumor from a donor mouse was collected in normal media and finely minced. Approximately 3 to 4 mm3 of the mash were implanted into the fourth mammary gland on either side of FVB/N females (Jackson laboratory, Bar Harbor, ME) at 6 to 8 weeks of age under anesthesia. Tumor growth was monitored twice weekly by caliper measurements. When the measured tumor volume was ~1000 mm3, tumors were collected and minced to repeat the transplantation into other FVB/N recipient mice for expansion by serial transplantation for up to six generations. At each generation, a section of tumor was fixed in formalin and was analyzed for tumor morphology by H&E and (human) HER2/neu expression by IHC. Nanoparticle uptake studies commenced when a sufficient number of tumors was established to ensure completion of the huHER2 study design. To establish tumors for the nanoparticle studies, huHER2 tumors were collected from five to eight FVB/N donor mice and minced. Portions of the mashes (3 to 4 mm3) were implanted into the fourth mammary gland on either side of female recipient mice comprising immune strains FVB/N, athymic nude, or NSG (18 to 24 animals in each group) under anesthesia. When the measured tumor volume reached 150 to 200 mm3, animals were randomly assigned into cohorts comprising five to nine animals in each group and treated according to their cohort as described for the xenograft studies (see above). For tumor growth delay, huHER2 allografts were implanted in either FVB/N or athymic nude mice and intravenously treated with PBS, BP, BH (5 mg per animal, or Her 175 μg/ml, equivalent dose of Her on BH particles) 5 days after implantation (day 0). Tumors were measured and recorded twice weekly up to 28 days. On day 28, all animals were euthanized, tumors were collected, and weight was recorded.

Flow cytometry

FVB/N female mice (five to eight animals per group) growing single huHER2 allograft tumors of 150 to 200 mm3 received intravenous (tail vein) injections of PBS, BP, or BH (5 mg of Fe per mouse) or trastuzumab (Her; 175 μg per mouse). Mice were euthanized 24 hours after injection. For the later time point (14 day after injection) flow analysis, mice bearing huHER2 allograft tumors were injected with PBS, BP, BH, or Her (same concentrations as above) 10 days after tumor implantation. Tumors were minced with a sterile blade in a petri dish and transferred to a 50-ml conical tube containing digestion media [DMEM + 10% FBS (heat inactivated) and 0.1% collagenase + 0.005% hyaluronidase]. The tubes were rotated at 37°C for 30 min, and the dissociated tissue was filtered through a 100-μm filter. After centrifugation at 1400 rpm for 10 min, the pellets were washed with 30 ml of DMEM with 10% serum and centrifuged again for 10 min. Supernatant was discarded, and RBCs were lysed with ACK lysis buffer at room temperature for 3 to 5 min. An additional 30 ml of media was added, and the mixture was centrifuged at 1400 rpm for 10 min. The supernatant was discarded, and 500 μl of media was added to resuspend the pellet. An aliquot of this whole tumor was removed and labeled with the panel of flow cytometry antibodies (see table S9 for list). The remaining single cells were incubated on a rare-earth permanent magnet at 37°C for 30 min. After incubation, the supernatant was carefully separated to a fresh tube for flow cytometry. Three milliliters of media was added to the adhered (remaining) fraction cells, which contained magnetic (nanoparticle) material. Media was added to all sample tubes to make up a final volume of 3 ml, and they were then centrifuged at 1400 rpm for 10 min. The supernatant was discarded, and pellets were suspended in PBS and counted to measure the total number of cells. Cells (1 × 106 to 2 × 106) were collected and treated with Fc blocker (2 μl of 100 μl of PBS; anti-CD16/32, BioLegend, San Diego, CA) and incubated on ice for 10 min. Samples were then centrifuged at 1400 rpm for 3 min, after which the cells were incubated with LIVE/DEAD solution (1 μl of 100 μl of PBS; Zombie Aqua, BioLegend, San Diego, CA) at room temperature in the dark for 30 min and then centrifuged. After centrifugation, cells were washed with 100 μl of PBS with 5% heat-inactivated FBS and again centrifuged. Cells were then stained with 50 μl of the solution of panel of antibodies (table S9) in appropriate dilution and incubated at room temperature in the dark for 30 min, centrifuged at 1400 rpm for 3 min, washed with 100 μl of PBS and 5% heat-inactivated FBS, and again centrifuged. Cells were then fixed with 50 μl of fix/perm solution and incubated at room temperature in the dark for 30 min, centrifuged, and washed. Two hundred microliters of PBS with 5% heat-inactivated FBS was added, and the suspensions were stored at 4°C until they were measured by polychromatic flow cytometry (LSR-II, BD Biosciences, San Jose, CA). The gating and selection of quadrants were based on fluorescence minus one controls. Analysis of data was done using FlowJo (version 10) software. Gating strategy is provided in figs. S8 and S11.

Cell count of magnetically separated cells

FVB/N female mice (three animals per group with two tumors each) growing huHER2 allograft tumors at ~100 mm3 received intravenous (tail vein) injections of BP or BH (5 mg of Fe per mouse). Mice were euthanized 24 hours after injection. Tumors were processed as above for flow cytometry and placed on permanent magnet for 30 min. After discarding supernatant, total numbers of magnetically attached cells were counted in a cell counter (Nexcelom, MA).

Pan leukocyte depletion

huHER2 allograft tumors were grown in FVB/N female mice (n = 4). When the tumor volumes reached ~150 mm3, mice were treated with subcutatneous injections of azathioprine (Sigma-Aldrich, St. Louis, MO), a pan leukocyte inhibitor at a dose of 200 mg/kg body weight for three consecutive days (39, 40). On the third day and 4 hours after azathioprine injection, mice were intravenously injected with BH (5 mg of Fe per animal). Mice were euthanized 24 hours after BH injection to collect tumor for ICP-MS analysis.

Statistical analysis

Results of all statistical analyses are provided in tables S11 to S33. ICP-MS and Prussian blue Aperio scored data were modeled as following log-normal distributions on the basis of proper exploratory analyses. Generalized mixed-effect models were used, with treatment, strain, and cell line as the fixed effects and mice as the random effect (intercept), such that the ratios between any two levels of fixed effects may be reported directly from the mixed-effect models. Models with fixed effect only and with two-way and three-way interactions were considered. Multiple comparison adjustments were made using the Bonferroni method to strongly control the overall family-wise type I error at 0.05.

Flow cytometry and tumor growth analysis. For flow cytometry and tumor growth data, it seemed unreasonable to assume commonly used parametric distributions. Therefore, pure nonparametric comparisons were made across all comparisons with Dwass-Steel-Critchlow-Fligner procedure for the pairwise comparisons to properly adjust for the potential inflation of family-wise type I errors.

In vitro cell count and ferene-s assay. All in vitro cell count and ferene-s assay data were analyzed by Student’s t test.

Correlation analysis. Rank-based, nonparametric Spearman correlation was performed using SAS 9.4, R, and Graphpad software.


Supplementary material for this article is available at

Supplementary Materials and Methods

Table S1. Summary of analytical data of (80 nm) BP nanoparticles.

Table S2. Summary of analytical data of all BNF-HER nanoparticles prepared.

Table S3. Characteristics of breast cancer cell lines used in the study.

Table S4. Summary of analytical data of BNF-IgG nanoparticles.

Table S5. Summary of analytical data of BNF-HER nanoparticles that passed in vitro qualification testing.

Table S6. Summary of immune modifications in mouse strains used for study.

Table S7. Summary of Aperio imaging settings used for digital analysis of tissue sections.

Table S8. Definitions of parameters used for Aperio imaging settings.

Table S9. Antibodies used for flow cytometry and their dilutions.

Table S10. Summary of numbers and strains of mice used in the study.

Table S11. Summary of one-factor model statistical analysis of iron measurements in xenograft models.

Table S12. Summary of two-factor model statistical analysis of iron measurements in xenograft models.

Table S13. Summary of three-factor model statistical analysis of iron measurements in xenograft models.

Table S14. Summary of one-factor model statistical analysis of Prussian blue histopathology analyses in xenograft models.

Table S15. Summary of two-factor model statistical analysis of Prussian blue histopathology analyses in xenograft models.

Table S16. Summary of three-factor model statistical analysis of Prussian blue histopathology analyses in xenograft models.

Table S17. Summary of statistical analysis of whole tumor digests flow cytometry in huHER2 allograft model.

Table S18. Summary of statistical analysis of nanoparticle-associated fractions (magnetic-sorted sediment) from flow cytometry in huHER2 allograft model.

Table S19. Summary of statistical analysis of nanoparticle-depleted fractions (magnetic-sorted supernatant) from flow cytometry in huHER2 allograft model.

Table S20. Summary of statistical analysis of iron measurements (ICP-MS) obtained from the livers of xenograft models.

Table S21. Ratio of Fe level between groups (treatment).

Table S22. Ratio of Fe level between groups (strains).

Table S23. Statistical analysis of ICP-MS huHER2-FVB/N lymph node data.

Table S24. Statistical analysis of ICP-MS huHER2-FVB/N spleen data.

Table S25. Statistical analysis of ICP-MS huHER2-FVB/N liver data.

Table S26. Ratio of percent positive between groups.

Table S27. Statistical analysis of tumor weight in huHER2-FVB/N.

Table S28. Statistical analysis of tumor growth in huHER2-FVB/N.

Table S29. Statistical analysis of whole tumor flow data third day.

Table S30. Statistical analysis of whole tumor flow data seventh day.

Table S31. Statistical analysis of whole tumor flow data 14th day.

Table S32. Statistical analysis of tumor weight–huHER2 allograft in nude mice.

Table S33. Statistical analysis of tumor growth–huHER2 allograft in nude mice (from initial day to 21st day).

Fig. S1. Representative images showing immunofluorescence staining of BH particles.

Fig. S2. Subtracting endogenous iron using PBS controls reveals little tumor retention of plain nanoparticles, and retention of BH nanoparticles is independent of tumor expression of the target antigen HER2.

Fig. S3. Retention of Herceptin-labeled BNF nanoparticles by xenograft tumors depends on immune strain of host.

Fig. S4. Weak correlations were found between deposits of plain nanoparticles and HER2, CD31+, or IBA-1+ regions in tumors of mice injected with BP nanoparticles.

Fig. S5. BNF nanoparticles labeled with a nonspecific IgG polyclonal human antibody were retained by tumors.

Fig. S6. Histopathology data support ICP-MS results for tumor retention of nanoparticles, and ICP-MS data show nanoparticles accumulated in lymph nodes, spleens, and livers of injected mice.

Fig. S7. Within tumors, nanoparticles localized in stromal regions rather than in cancer cell–rich regions.

Fig. S8. Gating for flow cytometry was conducted to ascertain immune cell populations residing in tumors.

Fig. S9. Flow cytometry analysis of huHER2 tumors harvested from immune competent mice reveals tumor immune microenvironment changes, and magnetically sorted tumor immune cell populations demonstrates impact of nanoparticles on tumor immune cells in response to intravenous nanoparticle delivery.

Fig. S10. Pan-leukocyte inhibition abrogates BH nanoparticle retention in tumors.

Fig. S11. Systemic exposure to BNF nanoparticles resulted in tumor growth inhibition but only if the host has an intact (adaptive) immune system (i.e., T cells).

Fig. S12. Following systemic exposure to nanoparticles, intratumor T cell populations decline through the third day and then increase by day 7 relative to PBS controls.

Fig. S13. Exposure to nanoparticles induces changes in adaptive immune signaling in tumors of nanoparticle-treated mice.

Fig. S14. Changes in innate cell population in tumors of nanoparticle-treated mice.

Fig. S15. Data suggest that systemically delivered BNF nanoparticles are preferentially sequestered by inflammatory immune cells within the TME, resulting in immune recognition of the tumor.

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: Funding: The authors wish to express their sincere gratitude to the Jayne Koskinas Ted Giovanis Foundation for Health and Policy for the generous award to support this project. J.D.B. and E.T.H. received partial support from NIH 5R25GM109441. Animal, biostatistics, and flow cytometry core resource facilities were supported in part by the NIH/NCI grant 5P30 CA006973. The authors wish to acknowledge T. L. Nilles for assistance in flow cytometry experiments. Author contributions: P.K. designed and conducted experiments, analyzed data, and prepared the manuscript. J.D.B. and A.S. conducted ICP-MS experiments and ICP-MS data analysis. E.T.H. and B.W.S. conducted histology analysis. J.S., E.T.H., C.-T.Y., and S.-H.Y. conducted tissue staining, ferene-s assays, microscopy, and sample preparation for ICP-MS. S.K.K. and M.H. conducted ferene-s assays. C.-T.Y. and J.S.C. performed in vitro macrophage nanoparticle uptake assay. T.D.A and E.J. provided guidance for flow cytometry experiments. C.G. conducted nanoparticle synthesis and characterization and prepared the manuscript. X.C.Z., W.F., and C.H. performed the statistical analysis of data. S.S. designed the studies and prepared the manuscript. B.W.S. designed the histology studies, conducted manual histology analysis, and prepared the manuscript. R.I. conceived and designed the experiments, conducted analysis, and prepared the manuscript. Competing financial interests: C.G. is an employee of micromod Partikeltechnologie, GmbH, manufacturer of BNF nanoparticles used in the studies. R.I. is an inventor on several issued and pending patents. All patents are assigned to Johns Hopkins University or Triton BioSystems/Aduro Biotech Inc. R.I. consults for Imagion Biosystems, a company developing imaging with magnetic iron oxide nanoparticles. R.I. serves on editorial boards of journals published by Taylor & Francis and IOP Publishing and has received stipends for his activities. C.H. consults/has consulted for Merck & Co. on technology/projects unrelated to research described in this manuscript. All other authors report no conflicts of interest and declare that they have no competing interests. Data and materials availability: huHER2 breeding pair was provided by Genentech, Inc. by a Materials Transfer Agreement with R.I. and JHU. All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article