Dendrimer size effects on the selective brain tumor targeting in orthotopic tumor models upon systemic administration

Abstract Malignant gliomas are the most common and aggressive form of primary brain tumors, with a median survival of 15–20 months for patients receiving maximal interventions. Advances in nanomedicine have provided tumor‐specific delivery of chemotherapeutics to potentially overcome their off‐target toxicities. Recent advances in dendrimer‐based nanomedicines have established that hydroxyl‐terminated poly(amidoamine) dendrimers can intrinsically target neuroinflammation and brain tumors from systemic administration without the need for targeting moieties. The size of nanocarriers is a critical parameter that determines their tumor‐targeting efficiency, intratumor distribution, and clearance mechanism. In this study, we explore the dendrimer size effects on brain tumor targeting capability in two clinically relevant orthotopic brain tumor models, the 9L rat and GL261 mouse models, which capture differing aspects of gliomas. We show that increasing dendrimers from Generation 4 to Generation 6 significantly enhances their tumor accumulation (~10‐fold greater at 24 hr), tumor specificity (~2–3 fold higher), and tumor retention. The superior tumor targeting effect of G6 dendrimers is associated with its reduced renal clearance rate, resulting in longer circulation time compared to G4 dendrimers. Additionally, the increase in dendrimer generation does not compromise its homogeneous tumor distribution and intrinsic targeting of tumor‐associated macrophages. These results validate the potential for these dendrimers as an effective, clinically translatable platform for effectively targeting tumor‐associated macrophages in malignant gliomas.

with malignant gliomas remain limited. Current standard of care for newly diagnosed patients includes maximal safe surgical resection followed by intensive radiotherapy and concomitant chemotherapy, leading to median survival times of 15-20 months for patients with glioblastoma. These survival times have improved minimally in the past few decades. [4][5][6] Innovative new strategies are necessary to address the plethora of challenges facing glioma treatment to deliver more effective therapies.
The field of nanomedicine has provided advances in tumorspecific systemic delivery of chemotherapeutics to overcome offtarget toxicities of these therapies via formulation into nanoparticles.
As opposed to local delivery, which is highly invasive and requires either surgical intervention or catheterization, systemic delivery utilizes blood circulation and the leakiness of the tumor blood vessels to traffic therapies to the target site noninvasively. 7 However, the efficacy of such nanomedicine-mediated systemic therapies have generally not translated into clinical successes due to their failure to address critical delivery challenges, including lack of transport across biological barriers, limiting of systemic distribution, and inadequate accumulation in the brain tumor. 8,9 Therefore, the size and surface attributes of systemic nanoparticles must be carefully engineered to overcome these challenges. To achieve efficient delivery and high accumulation in brain tumors, nanoparticles must achieve long systemic circulation time without significant residence in peripheral organs, while managing to cross the highly heterogeneous bloodbrain tumor barrier (BBTB). Previous studies have found that nanoparticles must be less than 20 nm in size to cross the BBTB and less than 7 nm to diffuse freely within the tumor microenvironment. 10 However, nanoparticles with smaller sizes often have shorter systemic circulation time compared to larger ones due to the size dependence of renal clearance. 11 Therefore, careful balancing of these factors is critical for achieving effective brain tumor accumulation.
In addition to effective tumor penetration and accumulation, therapies must access the cells of interest to have an effect on the tumor. 12 Tumor-associated macrophages (TAMs) have emerged as promising therapeutic targets for cancer treatment due to their abundance within tumors and their critical roles in manipulating the immune environment toward a pro-tumor state. 13 Tumors actively recruit host macrophages and monocytes and repolarize them into TAMs, which suppress immune activation and promote tumor growth, metastasis, and drug resistance. [14][15][16] Therefore, immunotherapies that can reprogram TAMs from a pro-tumor to an anti-tumor phenotype can inhibit their tumor-supporting functions while simultaneously bolstering their immune activation and antigen-presenting functions. 17,18 Based on strong preclinical results, TAMs-focused interventions are undergoing clinical trials alone or in conjunction with traditional treatment modalities (NCT02829723, NCT01349036, NCT02584647, and NCT03708224). However, translation of immunotherapies targeting macrophages has been limited by low response rates, drug resistance, and systemic toxicities associated with nonspecific immune modulation. [19][20][21] Therefore, delivery strategies that can carry immunotherapy payloads into the tumor and specifically to TAMs while remaining inactive in the rest of the body may yield positive clinical outcomes.
Poly(amidoamine) (PAMAM) dendrimers are highly tailorable, branched macromolecules in the sub-10 nm size scale that have been explored as targeting vectors for cancer-specific treatments and diagnostics. [22][23][24] We have previously shown that hydroxyl-terminated PAMAM dendrimers are able to cross impaired blood-brain barriers and selectively target activated microglia from systemic circulation in a variety of models for neurodegenerative diseases. [25][26][27][28][29] This hydroxyl dendrimer platform has significant potential for clinical translation due to its well-tolerated nature in vivo 30 and scalability 31 and is currently undergoing an early-stage clinical trial for a pediatric central nervous system indication (NCT03500627). We have previously reported in the 9L rat model of gliosarcoma that systemically administered hydroxylterminated Generation 4 dendrimers (G4) with~4.3 nm size and neutral surface charge cross the BBTB to penetrate uniformly throughout the solid brain tumor and specifically target TAMs. 22 However, G4 dendrimers are also quickly cleared from circulation within 24 hr through renal clearance. In this study, we explored whether Generation 6 hydroxylterminated dendrimers (G6) of similar surface charge but greater size (~6.7 nm) can achieve greater tumor accumulation while maintaining the favorable in vivo transport properties of G4 dendrimers. We investigated this in two clinically relevant brain tumor models that capture differing aspects of gliomas: the 9L rat and GL261 mouse models.
To study tumor accumulation kinetics in gliosarcoma, 9L-tumor bearing rats were systemically administered with dendrimers at 27.5 mg/kg, and tissues were collected for analysis at 15 min, 8, 24, and 48 hr postinjection. G6 exhibited significantly higher tumor accumulation compared to G4, with~30-fold greater tumor uptake at 48 hr postadministration than G4 (Figure 1a). The concentration of G6 within the tumor continued to increase in the initial 8 hr, followed by a plateau where G6 uptake was maintained at above 20 μg/g tissue Molecular weight, size, and ζ-potential of G4-OH and G6-OH   G6 exhibits significantly lower kidney levels compared to G4, indicating decreased renal clearance. G6 also exhibits slightly increased levels in the liver and spleen due to longer circulation time. ***p < .001 adaptor molecule 1 (Iba1) to label TAMs for confocal imaging ( Figure 2).
G4 dendrimers exhibited the expected co-localization with Iba1+ cells, while increasing dendrimer generation from G4 to G6 did not alter this TAMs targeting property. The altered uptake kinetics between G4 and G6 seen in Figure 1 were reflected in the timing and quantity of intracellular dendrimer signal within TAMs. G4 signal was present within TAMs in the tumor as soon as 15 minutes. G6 exhibited slower TAMs uptake, with little signal seen at 15 minutes but strong signal colocalization with TAMs after 8 hr and maintained to the later time points. Both G4 and G6 dendrimers colocalized highly and specifically with Iba1+ cells.

| G6 dendrimers show prolonged circulation time and reduced renal clearance in the 9L rat model of gliosarcoma
To understand the underlying mechanism for different tumor accumulation between G4 and G6, we compared their plasma clearance and biodistribution in major organs. We found that in plasma, G6 concentration was higher than G4 at all time points (Figure 3a). At 48 hr after injection, G6 exhibits~18-fold greater levels (3.26 ± 0.19 %ID/mL) than G4 in the plasma (0.18 ± 0.01 %ID/mL). This increased residence in circulation was associated with reduced renal clearance rate of G6 compared to G4. At 48 hr, there was~40-fold lower accumulation of G6 (27.6 ± 2.2 μg/g) than G4 (0.68 ± 0.15 μg/g) in the kidneys ( Figure 3(b)). Interestingly, dendrimer accumulation in the liver and spleen remained at low levels for both G4 and G6, indicating renal clearance as the major excretion mechanism for both G4 and G6, albeit G6 had significantly reduced renal clearance rate.

| Increasing dendrimer size improves tumor accumulation in the GL261 model of glioblastoma
The 9L rat model of gliosarcoma has been largely used to study the transport of drugs across the blood-brain and blood-tumor barrier. 32 F I G U R E 4 Size dependence of tumor targeting in GL261 murine model of glioblastoma. Fluorescently labeled G4 and G6 were injected intravenously into GL261 brain tumor bearing mice. hemisphere and G6 exhibiting~25-fold difference at 24 hr. G6 also did not exhibit increased off-target brain accumulation, with accumulation in the contralateral hemisphere comparable to that of G4.
We did not explore the peritumor regions in this model due to size constraints with mice.
AUC calculations based on the quantification data corroborated these trends. G6 exhibited significantly greater tumor AUC (519.8 F I G U R E 5 Systemic biodistribution of G4 and G6 dendrimers in glioblastoma. Fluorescently labeled G4 and G6 dendrimers were injected intravenously into GL261 brain tumor bearing mice. Plasma and organs were collected at specified time points, homogenized to extract dendrimers, and measured via fluorescence spectrometry. (a) G6 dendrimers exhibit higher plasma concentration over time than G4 dendrimers. The size of dendrimers also greatly affects their rate of renal clearance and renal accumulation. Nanoparticles less than 6 nm in diameter undergo rapid renal clearance through the fenestrations in kidney glomerular capillary walls, while those greater than 8 nm are not. 40,41 Based on these considerations, G6 is able to partially escape renal clearance without altering this as its primary clearance route for prolonged circulation time, which is reflected in the measured plasma and kidney dendrimer content (Figures 3a,b and 5a,b). Many hypotheses propose the glomerular filtration barrier (GFB) as the key barrier to regulate the efficient filtration of macromolecules without clogging. 42 In one hypothesis, an electrical field is presented across the GFB, preventing charged macromolecules from clogging this barrier. 43 In another, glomerular based membrane, a dense fibrinous network that lies next to the glomerular endothelial fenestrations, acts as a permeable gel and mediates the filtration of macromolecules through a combination of convective flow and passive diffusion. 44 In this mechanism, diffusion is regarded as the major anti-clogging mechanism involved in removing the retentate from the filter. In particular, the theory suggests that smaller size macromolecules partition from the blood into these glomerular membranes to a greater extent compared to larger size macromolecules. 42 This hypothesis correlates with the higher kidney accumulation of G4 compared to G6 observed in our experiments (Figures 3b and 5b).
Given the significant decrease of renal accumulation resulting from the increase of dendrimer generation, we do not observe any alteration of dendrimer acculturation in the spleen or liver (Figures 3b   and 5b). The hepatobiliary system represents the primary route of excretion for particles with sizes beyond the renal filtration cutoff (10-20 nm). Nanoparticles above 40 nm tend to be cleared through the mononuclear phagocyte system, for example, the Kupffer cells. 11,45 These two mechanisms account for the nanoparticle accumulation in the liver and spleen. Since G6 has a size of~7 nm, which is below the size range of clearance via the hepatobiliary and mononuclear phagocyte systems, it does not exhibit significant changes in these organs compared to G4.
Taken together, our results demonstrate that G6 exhibits a size optimal for balancing renal clearance and tumor penetration, resulting in greater tumor accumulation in both gliosarcoma and glioblastoma than G4 dendrimers. These findings are consistent with qualitative trends observed in recent studies, which indicate that larger dendrimers and other nanoparticles fail to penetrate the solid tumor while smaller ones rapidly clear out, with the ideal size being in the generation 5-6 range. 7,10 While dendrimers have been explored for targeting brain tumors, relatively few studies have performed quantitative analysis of dendrimer accumulation in orthotopic brain cancer models. [46][47][48] Compared to other types of dendrimers injected via tail vein in flank tumor models, our hydroxyl-terminated PAMAM dendrimer results exhibit similar levels of tumor accumulation despite needing to overcome physical transport barriers in orthotopic brain tumors that flank tumors lack. [49][50][51] Compared to other types of nanoparticles explored in orthotopic brain tumors, G6 shows a~100-fold greater in tumor accumulation than liposomal 52 nanoparticles and~10-fold greater than G4 and reported tumor levels of gold 53,54 and PEGylated iron oxide 55 nanoparticles. Studies based on traditional polymeric nanoparticles with~135 nm size showed only a 50% improvement in the delivery of temozolomide in a rat brain tumor model. 56 In another study in orthotopic rat brain tumors, liposomes of 80 and 200 nm sizes were unable to penetrate the BBB and distributed heterogeneously in the solid brain tumor, with dense peripheral tumor deposition and incomplete tumor core penetration. 57 Due to the limitations in passive targeting with these nanoparticles, active targeting strategies are often adopted in larger nanoparticle systems to improve their tumor penetration. 58 Surface modifications with targeting ligands such as angiopep-2 and nestin have been employed to achieve even greater tumor accumulation, 59,60 a strategy we are exploring to further improve the tumor targeting of these PAMAM dendrimers. Additionally, these dendrimers achieve greater specificity for the tumor compared to healthy brain tissue. At 24 hr after injection, G6 dendrimers exhibited~10-fold and~15-fold greater accumulation in tumor compared to the contralateral hemisphere in the 9L and GL261 models, respectively, while G4 dendrimers exhibited a~5-fold greater tumor accumulation in tumor compared to contralateral hemisphere in both models. In comparison, gelatin-conjugated polylysine dendrimers and PEGylated iron oxide nanoparticles exhibited <3-fold specificity. 51,55 In addition to tumor specificity, these PAMAM dendrimers also specifically target TAMs, with >80% of the dendrimer signal in the tumor localized within TAMs ( Figure S2a). This means the measured tumor accumulation represents dendrimer quantity not just in the tumor but specifically within the target cells of interest. To our knowledge, no quantitative analyses of nanoparticle accumulation in orthotopic brain tumors have explored that quantification on a celltype level. Therefore, the hydroxyl-terminated PAMAM dendrimers are a nontoxic, translatable targeting platform that not only exhibits high tumor accumulation, but also enables selective delivery to TAMs.
Gliosarcoma and glioblastoma have long been considered clinically indistinguishable and undergo similar treatment regimens. [61][62][63] However, recent literature suggests that there are a number of critical distinctive characteristics that warrant separate study and, potentially, intervention. 64,65 The sarcomatous elements in gliosarcoma result in firm, well-defined tumors while a hallmark of glioblastoma is its diffuse, highly infiltrative tumor border. 65,66 Despite this invasiveness, glioblastoma is highly localized to the brain while gliosarcoma is metastatic. 65 Clinically, patients with gliosarcoma face poorer prognoses. 64 Sarcomatous elements in tumors have been shown to correlate with increased PD-L1 expression by tumor cells and suppression of the cytotoxic T-cells. 67 This suppression creates an anti-inflammatory immune environment that polarizes macrophages into TAMs, which exhibit efficient phagocytosis and endocytosis. [68][69][70] This difference in immune environment arising from sarcomatous elements within the tumor may account for the differences between the two models observed with the G6 dendrimer at 48 hr after injection. The increased TAMs activation in gliosarcoma results in greater tumor accumulation and retention due to greater and more efficient internalization activity. In contrast, the G6 levels within TAMs in glioblastoma peak at 24 hr and then drop off as they are exocytosed into the extracellular space. Further exploration of the mechanism behind this difference in TAMs retention is warranted but may arise due to G6 being carried out with the secretion of intercellular signaling extracellular vesicles from TAMs, 71,72 which may then exhibit strong interactions with the tumor extracellular matrix to result in the signal pattern observed. 73 The difference in tumor accumulation kinetics in gliosarcoma compared to glioblastoma arising from sarcomatous content may have implications for the design of dendrimer-mediated interventions and suggests that they should be considered separately for the development of effective treatments.
Anti-cancer compounds have been shown to induce systemic toxicities, liver toxicity in particular; therefore, the biodistribution and accumulation of nanoparticles must be considered in addition to their effectiveness in targeting the disease site. 74,75 Renal filtration is the primary route of clearance for G4 and G6 dendrimers, so, as expected, both dendrimers are present at their highest levels in the kidneys. For all other organs, both dendrimers do not exhibit accumulation and by 48 hr show less than 0.5 %ID/g, indicating limited systemic exposure to the therapeutic payload. This is in contrast to free drug administration of chemotherapies, which quickly clear from tumor tissue while accumulating in the liver at much greater concentrations. 60,76 These PAMAM dendrimers also compare favorably to gold nanoparticles, PEGylated iron oxide nanoparticles, and other classes of dendrimers, which exhibit slightly higher kidney levels but a significantly greater liver accumulation of 2-70% of the initial dose retained. 49,50,54,55,77 Taken together with their tumor accumulation results, this biodistribution profile indicates that these dendrimers may yield a large therapeutic window for effective and safe treatments for brain cancers.

| Tumor inoculations
All animals were housed at the Johns Hopkins University animal facilities and were given free access to food and water. All animal experi-   Dendrimer brain distribution images were obtained using a Zeiss LSM710 confocal microscope (Hertfordshire, UK). Settings were optimized to avoid background fluorescence using untreated animal tissues. Calibration curves for G4 and G6 ( Figure S3) showed that for the same concentration, G4 exhibited~2-fold greater fluorescence intensity than G6. This difference was offset by adjusting the parameters such as magnification, laser intensity, gain, and offsets in imaging Cy3 and Cy5 channels within each model. Zenlite 2011 software was used to process the obtained images, and any adjustments to brightness and contrast were kept constant across all compared images.

| Fluorescence quantification of dendrimers in tissues
For semi-quantitative fluorescence analyses of intracellular versus extracellular dendrimer levels in confocal images, ImageJ software was utilized. Intracellular dendrimer signal was determined by selecting TAMs as regions of interest (ROIs) to measure the integrated density parameter. Extracellular signal was determined by setting the entire image as an ROI to quantify the total fluorescence signal and subtracting the intracellular signal. The amount of extracellular signal as a percentage of total dendrimer signal was calculated via the following formula: (total signalintracellular signal)/total signal. Total tumor signal was determined by selecting the whole tumor in lowmagnification tile scans as an ROI to measure signal intensity.

| CONCLUSIONS
In this study, we explored the effect of increasing dendrimer generation from G4 to G6 on their intrinsic tumor and TAM targeting properties in the 9L rat model of gliosarcoma and the GL261 mouse model of glioblastoma. In these two orthotopic models, both dendrimers achieved high tumor specificity, intrinsic TAMs targeting, and thorough tumor penetration. The larger G6 dendrimers showed significantly enhanced accumulation within the brain tumor (~10-fold greater at 24 hr) and greater tumor specificity (~2-3-fold greater) compared to G4 dendrimers. These results demonstrated G6 dendrimers as potentially improved delivery vehicles compared to G4 dendrimers for targeted delivery of immunotherapies into brain tumors, and specifically to TAMs from systemic administration, increasing therapeutic efficacy while reducing side effects. Future work will focus on evaluating how this improved tumor accumulation will translate into superior effects when delivering a therapeutic payload.