Organ-restricted vascular delivery of nanoparticles for lung cancer therapy

Nanomedicines hold immense promise for a number of devastating diseases due to the ability to custom-design both the carrier and cargo. However, their clinical implementation has been hampered by physicochemical and biological barriers and off-target deposition which impair cell specific targeting, especially in internal organs. This study reports a new delivery approach using organ-restricted vascular delivery to allow for direct administration and recirculation of stimuli-responsive nanoparticles to promote cellular uptake into an organ of interest. Using this technique, nanoparticles reach the interior of dense tumors and are selectively taken up by lung cancer cells. Importantly, this surgical approach is essential as the same nanoparticles do not reach lung tumor cells upon systemic or intratracheal administration. Organ-restricted vascular delivery thus opens up new avenues for optimized nanotherapies for cancer and other diseases.


Introduction
The use of nanoparticles as therapeutic agents for cancer therapy and other diseases has attracted major attention in the past decades. The first generation of nanomedicines for cancer relied on passive targeting of cancer cells based on the concept of enhanced permeability and retention (EPR) effect observed in tumors. [1] The EPR effect has been described to promote preferential accumulation of nanoparticles in the tumor due to increased blood vessel permeability and impaired lymphatic drainage in the tumorous regions. Several FDA-approved nanomedicines have been designed with the intent of exerting their therapeutic effect by passive targeting of tumors. [2] More recently, nanoparticles incorporating active targeting strategies have emerged as an alternative approach to fine-tune nanoparticle delivery to specific cell types. [3] To achieve cellspecific targeting, nanoparticles can be functionalized with targeting agents on their surface, which can bind to receptors specifically overexpressed on tumor or tumor-associated cells. Furthermore, smart nanoparticles which can respond to specific external stimuli and release of cargo into the tumor have been explored to limit off-target effects. [4] However, the initial excitement accompanying these innovative nanomedical approaches has failed to translate into clinical success, largely due to physicochemical and biological factors which impair nanoparticle targeting in vivo. [5] In particular, increased interstitial hypertension in tumors is known to reduce convective transport and the dense extracellular matrix of the tumor environment hinders the diffusion of particles. [6] Furthermore, nanoparticles can be directly engulfed or immunologically tagged for destruction by the mononuclear phagocyte system (MPS) to be cleared in the liver or spleen, or by alveolar macrophages in the lung. [7] Accordingly, up to 99% of all systemically administered nanoparticles are deposited in the liver and spleen, representing a major limiting factor for effective application of nanomedicines. [8] In order to increase nanoparticle accumulation in solid tumors, clearance of particles by competing organs needs to be minimized and novel organ-specific methods of nanoparticle delivery should be considered. [9] While local delivery is possible for tumors of peripheral organs, such as the skin, delivery into tumors of internal organs, such as the lung, remains challenging.
Among solid tumors, nanoparticle delivery for lung cancer is one of the most challenging with the lowest targeting efficacy compared to other solid cancer types. [8] Lung cancer affects over 1.7 million patients per year and has the highest mortality rate among all cancers worldwide. [10] Although recent advances in cancer immunotherapy [11] and targeted therapy [12] have resulted in encouraging progress, the number of patients and mortalities is expected to rise for lung cancer, mainly due to increases in smoking rates and exposure to pollution. Previously, there have been attempts to locally deliver chemotherapeutics to lung tumors through either the airways [13] or the vasculature. [14] While airway delivery via inhalation has shown some degree of efficacy, particle delivery into the core of larger, solid tumors has not yet been achieved and thus inhalation therapies are limited to the treatment of smaller tumors. On the other hand, vascular delivery of chemotherapeutics exclusively to the lung, termed as isolated lung perfusion, is feasible in patients with lung tumors as demonstrated previously. [15] However, these approaches are infrequent due to high levels of pulmonary toxicity resulting from chemotherapeutics accumulating not only in tumors but also in healthy lung tissue. [14] Nanotherapy is particularly well suited for overcoming pulmonary toxicity associated with healthy tissue since with the help of the EPR effect and the recently described transcytotic pathway upregulated in endothelial cells lining the tumor vasculature [16] , nanoparticles could be successfully and selectively delivered to tumorous regions. Here, we provide proof-of-concept evidence for effective delivery of both active and passive targeting nanoparticles with pHresponsiveness into the interior of solid, dense lung tumors using a novel organ-restricted vascular delivery (ORVD) strategy. This approach overcomes unwanted accumulation of nanoparticles in off-target organs such as the liver and spleen and in alveolar macrophages of the lung by nanoparticle delivery into the surgically isolated vasculature of the organ of interest. It also eliminates extravasation of chemotherapeutics into healthy regions of the lung which is associated with pulmonary toxicity after isolated lung perfusion (Figure 1). Furthermore, ORVD results in cellular uptake and endosomal escape of nanoparticles in lung tumor cells, which is a prerequisite for nanoparticle-mediated drug delivery. ORVD is a versatile new delivery strategy which overcomes physicochemical and biological barriers for local delivery of nanomedicines to internal organs.

Synthesis and characterization of mesoporous silica nanoparticles
We first generated functionalized mesoporous silica nanoparticles (MSNs) [17] which can be used for active or passive targeting of specific tumor cells. Additionally, our MSNs contain a pHresponsive element allowing selective release into the tumor microenvironment or upon internalization into the endolysosomal compartment. The internal pore system of the MSNs was functionalized with thiol groups and the external particle surface with amino groups, creating a nanoparticle platform which allows for a wide range of customizable functionalizations ( Figure   2A). The pH-cleavable linker system, containing a biotin functionality, was added to the external surface of the MSN. The glycoprotein avidin was attached to the outer surface of the MSNBiotin via noncovalent association with the biotin groups (MSNAVI), acting as a stable gatekeeper of the pore system. MSNAVI nanoparticles without further functionalization were used for passive targeting of tumors in this study. Different targeting ligands can be attached to the outer surface of the avidin gatekeepers for active targeting of tumors ( Figure 2B and Figure S1A).
Synthesized core-shell functionalized MSNs (MSN-SHin-NH2,out) were amorphous ( Figure S1B) and spherical in shape with an approximate size of around 100-150 nm ( Figure 2C). The mesoporous structure was confirmed using nitrogen sorption measurements ( Figure 2D), with pore sizes of around 4 nm ( Figure S1C). Following internal and external functionalization (internal thiol external amine functionalization for MSN-SHin-NH2,out, external carboxy functionalization for MSNCOOH, external pH-responsive AK linker functionalization for MSNAK, and external biotin functionalization for MSNBiotin) ( Figure S1D), pore size and volume changed negligibly ( Figure  S1E and F). As anticipated, successful avidin capping resulted in a loss of surface porosity concomitant with a decrease in measured surface area, as the internal surfaces were no longer accessible following capping ( Figure S1C and G). We also observed a slight increase in the isoelectric point ( Figure S1H) and stabilization of surface charge across different pHs ( Figure S1I) following avidin capping ( Figure S1J). MSNAVI nanoparticles had a mean particle size of 170 nm in aqueous media ( Figure 2E), demonstrating colloidal stability. We next sought to confirm functionality of our pH-cleavable linker system using a neutral pH of 7 and a pH of 5, representing the acidic pH of the tumor microenvironment. Propidium iodide (PI) was loaded into MSNAVI (0.365 mg PI/mg MSNAVI) and the release of PI was measured over time after the pH was changed from 7 to 5. While we observed minimal PI release under neutral pH over 48 h, acidic pH induced cargo release over time with up to ~60% release after 48 h ( Figure 2F). Release at 48 h at acidic pH was significantly increased as compared to the same time point at neutral pH, indicating the specificity and stability of our pH-cleavable linkage system and avidin capping. Next, we further functionalized our MSNAVI with targeting ligands for two different receptors that are well known to be specifically elevated in distinct compartments of human lung tumors, i.e., epidermal growth factor receptor (EGFR) on lung tumor and tumor-associated cells [18] or C-C chemokine receptor type 2 (CCR2) on tumor-associated macrophages in the surrounding stroma [19] (Figure 2G and lower magnification image in Figure S1K). Active targeting of these two receptors by exploiting two different particle types can be regarded as a complementary approach to achieve an additive anti-tumor effect on cancer as well as cancer-associated cells. Here, we functionalized the artificial peptides GE11 [20] and ECL1i [21] on MSNAVI to generate particles targeting EGFR (MSNtEGFR) and CCR2 (MSNtCCR2), respectively ( Figure 2H). Importantly, colloidal stability was retained in aqueous solutions following the addition of targeting ligands for both MSNtEGFR and MSNtCCR2 ( Figure 2I and Figure S2L).

Increased uptake of actively targeted MSNs in vitro
To assess whether our active targeting strategy enhances nanoparticle uptake in vitro, we exposed A549 cells, a human lung cancer cell line with high EGFR expression ( Figure S2A), to receptortargeted MSNtEGFR and untargeted MSNAVI. We observed significantly enhanced uptake of MSNtEGFR in A549 cells as quantified by confocal microscopy ( Figure 3A) and flow cytometry ( Figure 3B). MSNtEGFR co-localized with EGFR, indicating EGFR-mediated uptake. MSNtCCR2 specificity was tested in an alveolar macrophage line, MH-S cells, which have high expression of CCR2 ( Figure S2B). We again observed significantly increased uptake of MSNtCCR2 in MH-S cells, as measured by confocal microscopy ( Figure S2C) and flow cytometry ( Figure S2D). In order to further confirm cellular uptake, we performed transmission electron microscopy (TEM) of A549 cells exposed to MSNtEGFR. Interestingly, we observed both receptor-mediated and unspecific endocytic uptake as well as endosomal escape and evidence of particle degradation ( Figure 3C). Together, these results show that the enhanced cellular uptake of the actively targeted nanoparticles in vitro is at least partially mediated via receptor-mediated uptake and confirms the functionality of our active targeting strategy.

Intravenously administered MSNs are deposited in liver and spleen in flank tumor models
Next, we sought to evaluate the targeted uptake of our MSN particles upon systemic delivery in an in vivo tumor model. In order to evaluate active targeting efficacy within the same animal, we exploited a double flank tumor model in an immunologically competent mouse strain (C57BL/6) where we subcutaneously inoculated syngeneic clones derived from Lewis lung carcinoma (LLC) cells ( Figure 4A) or murine melanoma (B16F10) cells that had been genetically engineered to have high or low EGFR expression (i.e., LLC-EGFR high and LLC-EGFR low or B16F10-EGFR high and B16F10-EGFR low , respectively). [22] In both settings, cells formed tumors of similar size within two weeks with similar morphology. Fluorescently labeled (ATTO 633) MSNs (with and without targeting ligands) were then systemically applied via retro-orbital intravenous injection and analysis of the flank tumors and several internal organs that were previously shown to have enhanced uptake of systemically administered nanoparticles. [8] Both MSNAVI and MSNtEGFR were mainly localized in the liver and spleen with negligible uptake in the flank tumors, lungs, and kidneys ( Figure 4D, S4D and S5). Quantification of the immunofluorescence signal per cell nucleus across the different organs confirmed that the localization of the MSNs in the liver was much more pronounced compared to flank tumors and other organs ( Figure 4E and Figure S4E). Importantly, we did not observe any difference in the uptake of the particles with regard to whether tumors were derived from EGFR high or low expressing cells. We further confirmed enhanced uptake in the liver compared to LLC flank tumors regardless of MSN type (active versus passive) through quantification of nanoparticle-based fluorescence in tissue homogenates of flank tumors and livers ( Figure 4F). Together, these data indicate that systemic delivery of actively targeted MSNs does not result in improved uptake into tumors as compared to passively targeted MSNs. Importantly, we found that our nanoparticles did not preferentially localize to the tumors but instead were taken up mostly by the mononuclear phagocyte system of the liver and spleen.

Alveolar macrophages engulf intratracheally administered MSNs in a mouse model of lung cancer
One strategy to increase the efficiency of nanoparticle delivery to tumors is through local delivery mechanisms. [23] The lung is considered to be a particularly well-suited organ for local drug delivery as nanoparticles can be delivered via the trachea to the respiratory epithelium of the lung where they are efficiently taken up due to its large surface area, thin epithelium layer, and rich blood supply. [24] Therefore, we next evaluated intratracheal delivery of passively and actively targeted MSNs into the lungs using the previously reported Kras LA2 mutant mouse model for lung cancer. [25] This model displays clinically relevant cancer development compared to tumor cell injection models as tumors develop spontaneously and are heterogeneously distributed ( Figure 5A). We first evaluated EGFR ( Figure 5B) and CCR2 ( Figure S6A) overexpression in Kras LA2 mutant lung tumors to confirm that this model is suitable for investigating EGFR-and CCR2-specific targeting via MSNtEGFR ( Figure 5) and MSNtCCR2 ( Figure S6). MSNAVI, MSNtEGFR, or MSNtCCR2 were intratracheally instilled directly into the lungs of tumor-bearing Kras LA2 mutant mice and the biodistribution of the MSNs was evaluated three days after administration. All MSN types were found to be retained in the lungs of the Kras LA2 mutant mice with no translocation of MSNs to secondary organs ( Figure S7). However, particles localized mainly to smaller hyperplastic lesions of tumorous lungs and in the periphery of solid tumors, but no uptake was observed in solid tumors ( Figure 5C and Figure S6B). Importantly, high magnification microscopic evaluation did not reveal any obvious differences in cellular uptake of MSNAVI, MSNtEGFR, and MSNtCCR2 on the cellular level ( Figure 5D and Figure S6C). The majority of particles were engulfed by alveolar macrophages in both normal and tumorous regions of lungs irrespective of nanoparticle functionalization ( Figure 5E and Figure S6D). These data indicate impaired targeting performance of both, actively and passively targeting MSNs to lung tumors when administered via the respiratory surface of the lung due to effective alveolar macrophage clearance of these particles.

Organ-restricted vascular delivery of MSNs enables specific deposition of nanoparticles in tumors
In order to circumvent clearance by alveolar macrophages, we sought to use an alternative vascular delivery mode which also increases the cross circulatory time of the nanoparticles within the tumor vasculature while minimizing the competing effects such as MPS clearance. Isolated lung perfusion is a clinically established surgical approach which was developed for delivery of higher doses of chemotherapy to the lung through isolation of the pulmonary circulation from systemic circulation. [15] While the procedure itself is safe, it has not yet seen widespread clinical application due to damage to the neighboring healthy tissue from chemotherapeutics. Therefore, we sought to leverage our pH-responsive nanoparticles by delivering them using a modified isolated lung perfusion procedure. The combination of isolated lung perfusion with environmentally-responsive or stimuli-responsive nanoparticles could permit controllable delivery of chemotherapeutics selectively to tumorous tissue.
Owing to the difficulties in performing such a surgery in a small animal model of lung cancer, we tested the feasibility of this approach in an ex vivo model to simulate the surgical conditions of isolated lung perfusion. We explanted tumorous lungs from healthy or Kras LA2 mutant mice and introduced actively or passively targeting MSNs for 3 h at controlled flow rates using an ex vivo system containing ventilation and perfusion which we have previously described ( Figure 6A). [26] Our first data using organ-restricted vascular delivery (ORVD) indicated that nanoparticles were homogeneously distributed in healthy mouse lungs and located with close proximity to endothelial cells (CD31 positive), suggesting their retention inside healthy blood vessels ( Figure S8A). On the other hand, we observed enhanced delivery of MSNAVI, MSNtEGFR, and MSNtCCR2 into the cores of solid lung tumors in a reproducible manner, regardless of their surface functionalization for active targeting ( Figure 6B). Similar to other ex vivo murine lung setups used for pharmacologic analysis, we did not observe any significant cell death (i.e., cleaved caspase-3) over the 3 h exposure time [27] , indicating that neither the nanoparticles nor surgical ORVD technique induced cytotoxic side-effects. Additionally, we did not observe uptake of nanoparticles by lung-resident macrophages ( Figure S8B) but did observe efficient cellular uptake of the nanoparticles by cancer cells located in solid tumor cores as demonstrated by TEM analysis ( Figure 6C). In lungs exposed to MSNtEGFR via ORVD, nanoparticles localized to tumor cells with lamellar bodies, validating their distal epithelial origin ( Figure 6C). Taken together, these results demonstrate that our organrestricted approach results in enhanced cellular uptake of actively and passively targeting nanoparticles in solid lung tumors. ORVD may thus represent a useful strategy to overcome some of the existing physicochemical and biological barriers limiting nanoparticle delivery efficacy to solid tumors of the lung.

Conclusion
Despite rapid and exciting progress in the design and synthesis of increasingly sophisticated nanomedicines, overcoming the physical and biological barriers of specific organs has remained a primary challenge upon administration in vivo. Additionally, recent evidence suggests that previous reports may have overestimated the contribution of the EPR to directing nanoparticles to tumors and that additional active cellular processes, such as transcytosis, also promote extravasation of nanoparticles into tumors. [16] Many of these experiments have demonstrated proof-of-concept nanoparticle delivery in immunocompromised animals thereby underestimating deposition to immunological active off-target organs upon systemic administration, which remains a major challenge.
Therefore, in order to circumvent systemic effects associated with low efficacy of nanoparticle delivery, we developed a novel organ-restricted delivery strategy to directly administer nanoparticles into the target organ vasculature which allows recirculation. We also show that the ORVD technique promotes their extravasation from the deranged tumor vasculature into the core of solid tumors possibly via the recently demonstrated transcytosis pathway. [16] Encouragingly, we observed specific cellular uptake of nanoparticles within the core of solid lung tumors while nanoparticles were retained within the capillaries in regions with healthy tissue. This observation is fully in line with the recent work showing preferential activation of transcytotic transport of particles across the tumor vasculature. We did not observe any noticeable differences in cellular uptake between active or passively targeting nanoparticles using ORVD suggesting that the delivery method is the primary hurdle to obtain effective delivery to the tumor site.
Lung cancer remains one of the most challenging cancers to treat and its prevalence is increasing globally. [28] In addition to the fact that it is often diagnosed at advanced stages when tumors are beyond surgical treatment due to their larger size or their location, the lung is also a major site of metastasis and metastasized cancers are some of the most challenging cancers to treat. [29] Even at early detection stages, some small tumors remain inoperable and cannot be removed due to their location centrally or due to reduced pre-operative lung capacity. Additionally, patients with reduced lung capacity might not be able to undergo wedge or lobe resection. Overall survival for lung cancer is around 15% after 5 years. Currently, only 20% of lung cancer patients are eligible for resections to remove large tumors due to the location of the tumor and potential comorbidities which may present complicating factors [30] . Resection is often followed by radiation and chemotherapeutics to address hyperplastic regions and to avoid recurrence after surgery. [31] The majority of lung cancer chemotherapeutics are administered systemically and are associated with negative side effects (e.g., anemia), and are particularly toxic to the kidney, nervous system, and heart. [32] A substantial number of patients have to end or pause the treatment prematurely due to negative side effects. Recently, nanoparticles have emerged as a potential option to circumvent some of the problems associated with systemic administration of chemotherapeutics, but numerous challenges remain and therefore the efficacy of nanoparticle-mediated drug delivery to solid tumors has been limited. [33] In this study, we designed and compared cell-specific targeting efficacies of actively and passively targeted pH-responsive mesoporous silica nanoparticles for treatment of lung cancer.
The mesoporous silica nanoparticles we utilize here have several benefits which make them an attractive nanoparticle platform for treating cancer: 1) they are inert and can be loaded with virtually any cargo smaller than their pore size (i.e., 4 nm), making it compatible with a range of existing and emerging chemotherapeutics and other compounds, 2) contain a pH cleavable linker, restricting delivery of their cargo to cells within the acidic tumor microenvironment or after cellular uptake in the endo-lysosome, 3) can readily be functionalized for active targeting, and 4) are cytocompatible. [17,34] Overall, MSNs are highly adaptable and can be fine-tuned for each particular cancer microenvironment.
For active targeting studies, we explored a combinatorial approach to address complementary compartments of solid tumors: [35] the parenchyma via EGFR targeting and the stroma via CCR2 targeting, since these two surface receptors have been previously shown to be overexpressed in lung cancer and associated with poor prognosis of the disease. [18][19]36] As has been shown in previous studies for MSNs and other nanoparticles [2,37] , MSNs which contained active targeting ligands were taken up by cells at a significantly higher rate than passively targeted nanoparticles in vitro. However, we did not observe any preferential uptake of actively targeted nanoparticles when they were administered intravenously in vivo.
In order to minimize the likelihood that our active targeting strategy failed due to differential uptake between tumor types, we used two different in vivo flank tumor models in this study: 1) a murine lung carcinoma line (LLCs) to mimic primary lung tumors and 2) melanoma line (B16F10) to mimic melanoma metastasis in the lung. Furthermore, we included an internal control for each animal by generating paired flank tumors on the left and right side of each animal using two different clones which were genetically engineered to differentially express high or low amounts of EGFR, respectively. Of note, we did not observe any differences in tumor uptake for active or passively targeting MSNs, nor between tumors derived from tumorous cells engineered for high or low levels of EGFR. Finally, our flank models were syngeneic and thus generated in immunocompetent mouse models, making them relevant to the clinical scenario where a number of immune associated biological barriers play a role in limiting nanoparticle efficacy. [8,33] In all of these flank models, we found that the majority of nanoparticles were taken up in the liver and spleen, independent of the presence or absence of a targeting ligand, which is in line with some previous work. [7a] Local administration of chemotherapeutics and nanoparticles via the airways has been previously attempted for lung cancer. [2,13] Therefore, we next tested our nanoparticles via intratracheal administration in the Kras LA2 mutant mouse model, which spontaneously forms lung tumors over a period of several months. [25] This model is highly representative of the clinical scenario as the tumors are heterogeneous in size and distributed throughout the lung. Furthermore, Kras is the most frequent oncogene driver mutation in lung cancer [38] for which no approved drugs exist. After lung tumors were allowed to develop in these animals, we administered the nanoparticles intratracheally. Although, we did not observe any nanoparticle localization in other internal organs 3 days post-instillation, the overwhelming majority of MSNs were taken up nonspecifically in vivo by alveolar macrophages in both normal and tumorous lungs. There was no obvious difference in uptake between active and passively targeting nanoparticles. Intriguingly, our in vitro experiments showed higher uptake of the receptor-targeted nanoparticles, which suggests that the loss of targeting in vivo is a result of several additional factors such as the inherent capacity of the MPS to efficiently clear systemically or locally delivered particles. Our data are partly in contrast to other studies that observed enhanced tumor delivery using active versus passively targeted nanoparticles in vivo. [39] Differences between our findings and those of other groups may be due to the targeting ligands chosen, nanoparticle shape and composition, the animal model used, target organ/tumor architecture, and how tumor uptake and drug delivery efficacy was assessed. [8] Our work demonstrates the potential of administering nanoparticles in an organ-restricted manner to increase delivery and cellular uptake into solid lung tumors. The ORVD approach represents a novel strategy for locally delivering nanoparticles into the lung or other highly vascularized organs. By using controlled perfusion and recirculation of nanoparticles we promote extravasation from leaky, tumor-associated blood vessels and subsequently promote retention and cellular uptake in targeted tumor cells. As we did not observe any significant cellular damage in our ORVD setup, this indicates that the cellular uptake we observed was likely due to either the EPR effect or increased transcytosis in tumorous regions and not due to the induction of vascular damage in our ex vivo setup. We chose to perform our initial ORVD experiments in an ex vivo model due to the difficulties in performing isolated lung perfusion in murine animals in vivo. Ex vivo models and evaluation of potential therapies are an emerging research area and provide a unique opportunity to perform preclinical testing for situations which are difficult to first mimic in vivo. [27,40] While these data provide proof-of-principle evidence that ORVD is applicable for nanoparticle-based targeting of solid lung tumors, it will be important in future studies to validate this approach in larger animals using chemotherapeutic-loaded nanoparticles in vivo ultimately aiming at the clinical application of this concept in cancer patients.
Isolated lung perfusion uses extracorporeal circulation techniques to isolate the pulmonary vasculature from the systemic circulation. It has already been used for delivering lung cancer chemotherapeutics with the intent of increasing local delivery concentrations and reducing systemic toxicity. [14] However, despite the fact that the surgical approach is safe, one of the major limiting factors for more widespread use of this surgical technique has been the toxicity associated with the surrounding healthy tissue which is also exposed to chemotherapeutics [14] . Smart, stimuliresponsive nanoparticles, such as those used in this study, offer the unique advantage of controlled and selective release of drugs into the tumor microenvironment. Furthermore, application of nanoparticles under controlled flow parameters during ORVD may allow for further fine-tuning of delivery to optimize nanoparticle uptake under EPR conditions. In addition, it will be of interest to further explore other nanoparticle formulations, including those which showed safety but no efficacy in clinical trials when administered systemically.
In summary, we have shown that the extended recirculation of actively and passively targeted mesoporous silica nanoparticles in an organ-restricted fashion via isolated pulmonary perfusion results in enhanced localization of the nanoparticles in solid lung tumors. These results bring optimism to be able to offer a new treatment option for patients with inoperable lung tumors and those who are unable to cope with chemotherapy due to negative side effects. Our findings represent a clinically relevant and promising strategy to direct the nanoparticles to solid tumors in highly vascularized tissues for an enhanced therapeutic efficacy. Written informed consent was obtained from all subjects.

Experimental Section
Cell culture: The human non-small-cell lung cancer cell lines, A549 and H520, and the mouse alveolar macrophage cell line, MH-S, were obtained from the ATCC (American Type Culture Collection). A549 and H520 cells were maintained in DMEM medium supplemented with 10% FBS and 1% Pen/Strep. MH-S cells were maintained in RPMI 1640 medium supplemented with 10% FBS and 1% Pen/Strep. MH-S cells were further supplemented with 1 mM sodium pyruvate, 10 mM HEPES, and 50 µM β-mercaptoethanol (all AppliChem). All cells were grown at 37°C in a sterile humidified atmosphere containing 5% CO2.
Immunocytofluorescence: A549 and MH-S cells which were grown on coverslips were exposed to Western blotting: A549, H520, and MH-S cells were lysed in RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with cOmplete protease inhibitor cocktail. Protein content was determined using the Pierce BCA protein assay kit (Thermo Scientific). For Western blot analysis, equal amounts of protein were subjected to electrophoresis on 10% SDS-PAGE gels and blotted onto PVDF membranes (Bio-Rad).
Membranes were treated with antibodies using standard Western blot techniques. The ECL Plus detection reagent (GE Healthcare) was used for chemiluminescent detection and the membranes were analyzed with the ChemiDoc XRS+ (Bio-Rad).
Flow cytometry: 5x10 5 A549 or MH-S cells were plated on 6 well plates and incubated overnight.
The next day, the cells were exposed to ATTO 488-or ATTO 633-labeled MSNs for 1 h.

In vivo biodistribution studies
Intravenous application: Two weeks after subcutaneous inoculation of EGFR-high and EGFRlow LLC and B16F10 tumor clones, 1 mg ATTO 633-labeled MSNAVI or MSNtEGFR was applied to each mouse retro-orbitally. The mice were sacrificed with an overdose of isoflurane three days after the administration. In vivo live animal imaging experiments were carried out to analyze the pharmacokinetics and ex vivo organ distribution of ATTO 633-labeled MSNs. Fluorescence imaging of living mice was done using an IVIS Lumina II imager (Perkin Elmer, Santa Clara, CA).
Mice were anesthetized using isoflurane and serially imaged at various time-points: before, immediately at, 3, 6, 24, and 48 h post-injection of MSNs. Retro-orbital venous sinus injection, comparable to tail-vein injection, was used, in order to avoid potential animal distress and/or retention of significant amounts of dose in the tail. The images were acquired and analyzed using Living Image v4.2 software (Perkin Elmer, Santa Clara, CA). Flank tumors were selected as specific regions of interest and photon flux within these regions were measured. Intratracheal application: 12 week-old Kras LA2 mutant mice were intratracheally instilled with ATTO 633labeled MSNAVI, MSNtEGFR, and MSNtCCR2, as previously described [34] . Three days postinstillation, the mice were sacrificed with an overdose of ketamine (188.3 mg/kg) and xylazin hydrochloride (4.1 mg/kg) (bela-pharm). Lung lobes from each group (n=5 mice per group) were excised and prepared for cryoslicing.
Organ-restricted vascular delivery: Heart-lung blocks from WT and Kras LA2 mutant mice were extracted and placed into the ex vivo perfusion/ventilation system as described by Bölükbas et al. [26] and schematically depicted in Fig 5A. The system was protected from light for all of the experiments containing the fluorescent MSNs. The ex vivo heart-lung blocks were submerged in cell culture media supplemented with 10% FCS and 1% Pen/Strep in the internal incubation chamber kept at 37°C and were mechanically ventilated with a respiratory frequency of 100 strokes/minute and 100 μL stroke volume throughout the 3 h nanoparticle exposure. The blocks were perfused with PBS for 10-15 min before intra-arterial administration of the particles via the pulmonary trunk. 400 µL of 1 mg/mL concentrated ATTO 633-labeled MSN suspension was dispersed in 20 mL cell culture media in the external chamber protected from light. The MSN solution from the external chamber was fed into the heart-lung block for 10 min at 0.5 mL/min flow rate using a peristaltic pump for the initial loading of the system with 100 µg MSNs. Then, the feeding was stopped and the particle suspension was re-circulated through the inner loop of the system for an additional 160 min. After the exposure, the lungs were filled with OCT for cryosectioning and histological observation. Histological preparations and immunofluorescence imaging: For the intravenous systemic delivery experiment, internal organs as well as flank tumors were dissected and placed in 4% PFA overnight after which the suspension medium was exchanged to PBS. Representative parts of the organs were frozen in OCT and kept at -80°C. For the lungs obtained from the intratracheal delivery as well as the ORVD experiments, the airways were immediately filled with OCT by intratracheal administration. Later, the lobes were separated, transferred into cryomolds, and covered with additional OCT. Samples were left to freeze on dry ice and then stored at −80°C. For both experiments, 5 μm thick cryo-sections were sliced with a cryostat (Zeiss Hyrax C 50) and placed on superfrost plus adhesion slides. Immediately before staining, all cryo-sections were fixed with 4% PFA for 10 min, then washed with PBS, and permeabilized with 0.5% Triton-X. The sections were incubated with Roti-Block for 1 h at room temperature, and then with the primary antibody at 4°C overnight; i.e., EGFR (Abcam, ab52894) and CCR2 (Novus Biologicals, NBP1-48338). Afterwards, the sections were washed with PBS, incubated with Alexa Fluor 488 secondary antibody for 1 h at room temperature. After another PBS wash, the sections were finally stained with DAPI. In case phalloidin staining was used, the sections were first incubated with phalloidin for 45 min and then with DAPI for 10 min at room temperature directly after the fixation and washing step. The sections were mounted using fluorescence mounting medium (DAKO) and analyzed using confocal microscopy (LSM710, Carl Zeiss). Quantification of the cellular uptake of the MSNs in the tissues was conducted using the IMARISx64 software (version 7.6.4, Bitplane).

Fluorescence dosimetry of MSNs in organ homogenates:
The dose of ATTO 633-labeled MSNs in the flank tumors and livers was determined with quantitative fluorescence analysis similar to the validated method described by Rijt et al. [34] Briefly, aliquots of the tissue were dried at low power setting in a microwave oven (SEVERIN, MW7803; 30% power; 270 Watt) until no change of tissue mass was observed anymore. Aliquots of dried tumor and liver tissue (ca. 10 mg) were diluted by 1:90 (w/v) and 1:60 with PBS, respectively (i.e., 1 mg of dried tissue was diluted by 89 and 59 µL PBS, respectively). The diluted samples were mechanically homogenized with a highperformance disperser (T10 basic ULTRA-TURRAX ® ) on ice until no tissue pieces were visible anymore (ca. 3-5 min with short breaks to avoid undue heating of the samples). Residual tissue was rinsed off the disperser using 200 μL of PBS. Samples were vortexed immediately prior to pipetting four 75 μL aliquots (quadruple determination) from each of the samples into a black 96well plate for quantitative fluorescence analysis with a standard multiwell plate reader (Tecan, Safire 2; excitation and emission wavelengths: 630 nm and 660 nm). The fluorescence signals were related to the corresponding MSN mass using standard curves, which were prepared from blank liver and flank tumor tissues of non-exposed mice spiked with known amount of MSN and processed according to the same protocol described above (cage control). The prerequisite for reliable dosimetry is that the homogenization and drying process does not destroy the fluorescence    Mann-Whitney test. c) Different modes of nanoparticle uptake (i.e., receptor-mediated, blue arrow at (1) and unspecific endocytosis, red arrow at (2)) and endosomal escape, orange arrow at (3) observed in TEM micrographs of A549 cells exposed to MSNtEGFR for 3 h. Cell membrane is shown with the black arrow. Scale bars = 2 μm (upper left), 500 nm for insets 1-3. tumor-bearing mouse model. LLC clones with different EGFR expression (LLC-EGFR high with high basal EGFR expression and LLC-EGFR low following shEGFR modification; confirmed by Western blot) were injected subcutaneously and developed over two weeks. b) Representative fluorescence images of mice receiving 1 mg of MSNAVI or MSNtEGFR before, immediately after, and 48 h after retro-orbital administration. c) Quantification of the fluorescence intensity obtained from the individual flank tumors of the mice treated with the MSNs over time. Values given are an average of signal obtained from five independent mice at each time point ± standard error of the mean. d) Histological analysis of the biodistribution of intravenously administered MSNAVI and MSNtEGFR in LLC-EGFR high and LLC-EGFR low tumors, livers, spleens, lungs, and kidneys of the mice by confocal microscopy. Nuclear staining (DAPI) is shown in blue, cellular morphology via actin staining (phalloidin) in green and ATTO 633-labeled MSNs in red in the merged image, and in gray in the single channel. Images shown are representative for three different regions from each mice (n = 5 mice treated). Scale bar = 100 μm. e) Quantification of the MSNAVI and MSNtEGFR uptake per nuclei observed in histological analyses in LLC-EGFR high and LLC-EGFR low tumors, kidneys, lungs, spleens, and livers, respectively. **** p < 0.0001, two-way ANOVA, Tukey's multiple comparisons test, n = 5. f) Quantitative analysis of MSNAVI and MSNtEGFR biodistribution in tissue homogenates of treated animals shows increased uptake in the liver versus either LLC-EGFR high or LLC-EGFR low tumors. Values given are an average of five different samples per MSN type ± SEM. ** p = 0.0022, *** p = 0.0006, **** p < 0.0001, Two-way ANOVA, Sidak's multiple comparisons test.

Table of Contents (ToC)
Despite their immense potential, clinical translation of nanomedicines has been hampered by physicochemical and biological barriers impairing cell-specific targeting especially in solid tumors. This study reports a novel approach using organ-restricted vascular delivery (ORVD) with direct administration and recirculation of nanoparticles to enhance nanoparticle uptake in lung tumors. ORVD opens up new avenues for optimized nanotherapies.    bearing mouse model that was generated by subcutaneous injection of genetically modified melanoma clones (B16F10) for basal EGFR expression (B16F10-EGFR low ) versus overexpression (B16F10-EGFR high ). b) Representative fluorescence images of mice receiving 1 mg of MSNAVI or MSNtEGFR before, immediately after, or at 3, 6, 24, and 48 h after retro-orbital administration. c) Quantification of the fluorescence intensity obtained from the individual flank tumors of the mice treated with the MSNs in time course. Values given are an average of signal obtained from five independent mice at each time point ± standard error of the mean. * p = 0.0324, Two-way ANOVA, Tukey's multiple comparisons test. d) Histological analysis for the biodistribution of the intravenously administered MSNAVI and MSNtEGFR in B16F10-EGFR low and B16F10-EGFR high tumors, livers, spleens, lungs, and kidneys of the mice visualized by confocal microscopy. Nuclear staining (DAPI) is shown in blue, cell morphology via actin staining (phalloidin) in green and ATTO 633-labeled MSNs in red in the merged image, and in gray in the single channel. Images shown are representative for three different regions from each mice (n = 5 mice treated). Scale bar = 100 μm. e) Quantification of the MSNAVI and MSNtEGFR uptake per nuclei observed in histological analyses in B16F10-EGFR low and B16F10-EGFR high tumors, kidneys, lungs, spleens, and livers, respectively. In the HBSS control, animals only received HBSS and no particles. Values given are average of three different images per each treated mice ± standard error of the mean (n = 5 per MSN type). **** p < 0.0001, Two-way ANOVA, Tukey's multiple comparisons test. Figure S5. Intravenously administered MSNs are deposited to liver in a syngeneic LLC tumor model in vivo. Histological analysis for the biodistribution of the retro-orbitally administered a) MSNAVI and b) MSNtEGFR in the LLC-EGFR high tumors, LLC-EGFR low tumors, and livers of each treated mice by confocal microscopy. Nuclear staining (DAPI) is shown in blue, cell morphology via actin staining (phalloidin) in green, and ATTO 633-labeled MSNs in red. Images shown are representative for three different regions from each mice (n = 5 mice per MSN type). Scale bar = 100 μm. Figure S6. Intratracheally administered MSNtCCR2 are engulfed by alveolar macrophages in a mouse model of lung cancer in vivo. a) Immunohistochemistry staining of CCR2 (pink) in Kras LA2 mutant tumorous mouse lungs. b) Representative histological analysis of intratracheally instilled ATTO 633-labeled MSNtCCR2 uptaken in solid tumor cores versus their edges, and in hyperplastic or in tumor-free regions of the tumorous mouse lungs, after 3 days. Nuclear staining (DAPI) is shown in blue, cell morphology via actin staining (phalloidin) in green, and ATTO 633-labeled MSNs in red in the merged images, and in gray in the single channels. Five mice were analyzed per group with five random sections and three images per section in a blinded manner. Images shown are representative for three different regions from each group of mice (n = 5 per MSN type). Scale bar = 100 μm. Immunofluorescence co-staining for CCR2 in c) tumorous versus d) tumorfree regions the mutant lungs treated with ATTO 633-labeled MSNAVI versus MSNtCCR2. Nuclear staining (DAPI) is shown in blue, cell morphology via actin staining (phalloidin) in red, CCR2 staining in green, and ATTO 633-labeled MSNs in gray. Images shown are representative for three different regions from each group of mice (n = 5 per MSN type). Scale bar = 25 μm. Figure S7. MSNAVI and MSNtEGFR do not localize to the liver, spleen or kidney when administered intratracheally into Kras LA2 mutant mice. Representative histological analysis of intratracheally administered ATTO 633-labeled MSNAVI and MSNtEGFR are not present in livers, spleens, and kidneys of the Kras LA2 mutant mice 3 days. Nuclear staining (DAPI) is shown in blue, cell morphology via actin staining (phalloidin) in green, and ATTO 633-labeled MSNs in red. Images shown are representative for three different regions from each mice (n = 5 per MSN type). Scale bar = 100 μm. Figure S8. Organ-restricted vascular delivery of MSNs. a) Representative histological analysis of the lungs from WT mice exposed to ATTO 633-labeled MSNAVI, MSNtEGFR, and MSNtCCR2 (all gray). Nuclei were stained with DAPI (blue), epithelial cells labeled with E-Cadherin (green) and endothelial cells labeled with CD31 (red) shown in merged and corresponding single channel images for the nanoparticles. Scale bar = 100 μm. b) Histological analysis of the solid tumors from the Kras LA2 mutant lungs exposed to ATTO 633-labeled MSNtEGFR (gray). Nuclei were stained with DAPI (blue), monocytes/macrophages labeled with CD68 (green) and apoptotic cells labeled with cleaved caspase-3 (red) shown in merged and corresponding single channel images. Scale bar = 100 μm.