The ancillary effects of nanoparticles and their implications for nanomedicine

Traditionally, NPs were viewed as excipients of drugs. However, the diverse assortment of payload-independent NP effects represents a potential untapped avenue of biological mechanisms which, in an appropriate context, might be leveraged for medicinal benefit. That is, rather than merely functioning as a passive excipient, an NP might itself exhibit drug-like effects. These ancillary effects could conceivably be exploited either independently as a stand-alone nanomedicine or coupled with a conventional small molecule or biological drug to improve and enhance the therapeutic efficacy. However, properties of nanomaterials can also induce potentially harmful effects at the cellular, tissue and organ system levels with detrimental consequences for patients who receive the nanomedicine (Fig. 1). The potential of certain nanomaterials to trigger inflammation and cytotoxicity is well-known¹; however, other ancillary effects can be more subtle and might have pernicious effects. This Review considers selected relevant findings of potential medical usages for clinically relevant NPs (or preclinical NPs with translational potential), as well as some notable health and safety concerns that arise from the use of NPs as a medical tool, based on the intrinsic effects arising from NP–biological interaction—the yin and yang of nanomedicine, interconnected intimately.

Effects on vasculature and biological barriers

The usefulness of drugs is highly dependent on their potential to reach the site of intended action. For example, a drug that is highly effective in vitro when directly applied to cells is therapeutically useless if it cannot reach those cells in vivo. Thus, biological barriers effectively function as tissue and organ gatekeepers, and dictate which tissues are accessible to which drugs. The foremost barrier that affects pharmaceuticals is the blood vasculature, as most agents must diffuse through it to reach the site of intended action. In particular, the blood–brain barrier (BBB) is a major challenge for drug development, as it is less permissive and more selective to drugs than are other vessels, and allows almost no paracellular diffusion¹⁷. It is known that nanomaterial characteristics, particularly particle size and shape, can influence the degree to which NPs can pass through vascular barriers¹⁸. However, certain translationally relevant NPs (notably gold, titanium dioxide and silica) exhibit ancillary effects that modulate the permeability vascular endothelial barriers. NPs could therefore be utilized as agents to facilitate or improve drug delivery at sites by modulating the permissiveness of the barrier to those tissues, which are otherwise problematic to reach for biological or small-molecule therapies. This permeabilization, however, could present unforeseen problems by dysregulating the protective functions of the BBB and other vasculature, and potentially expose the brain or other vulnerable organs to toxins, pathogens or metastatic cells that would otherwise be excluded by an intact barrier.

The integrity of microvasculature is sustained by cell–cell adhesion between vascular endothelial cells, which forms a tight ‘seal’ that prevents uncontrolled extravasation. Various types of NPs have been shown to affect the integrity of the blood vasculature in several organs, which include the BBB, in ways that can even drive metastasis. Certain NPs can induce the formation of micrometre-scale pores in the vascular endothelium, which create leaks through which dysregulated paracellular diffusion can occur. For example, in microvasculature, adherens junctions initiate and maintain cell–cell adhesion in the vascular endothelium using cell–cell adhesion proteins called cadherins and catenins. The interactions between the domains of cadherins in neighbouring cells initiates the formation of an adherens junction. Titanium dioxide NPs were shown to bind to vascular endothelial cadherin (VE-cadherin) and disrupt the interactions between the VE-cadherin molecules that form the adherens junctions, which increases the bidirectional permeability of the vascular endothelium in subcutaneous blood vessels and in healthy lung vasculature^19,20. Disruption of interendothelial vascular junctions was also observed with citrate-coated silver NPs, which triggers VE-cadherin downregulation. Gold NPs were found to increase the degradation rates of tight junction proteins in the BBB²¹. Therefore, NP ancillary effects on vasculature could both compromise the integrity of the vasculature or create opportunities to deliver drugs beyond otherwise highly selective barriers.

In addition to the ancillary effects on vascular integrity, NPs might also influence the biological processes that regulate neovascularization. Zinc oxide NPs were found to enhance the development of capillary formation both in vitro and in vivo. The metallic component of the NP catalyses the formation of intracellular reactive oxygen species (ROS) in vascular endothelial cells, which including hydrogen peroxide, a redox-active second messenger in the angiogenic signalling process. This subsequently drives sprouting and the formation of new blood vessels from the existing microvasculature²². Such an NP may function to rapidly improve circulation at an injury site, which facilitates further tissue growth and repair. Therefore, stimulation of the neovascularization by an NP could be beneficial in enhancing healing. NP-based tissue engineering approaches are particularly interesting for the frontiers of regenerative medicine, and include bone tissue regeneration, treatment of myocardial infarct injuries, scar tissue repair, cosmetic surgery and organ transplantation. It is plausible that these emerging medical technologies could benefit by incorporating NPs to harness the ancillary effects for an improved efficacy and effects otherwise potentially difficult to achieve.

Haemostatic regulation and alteration of blood parameters by NPs

NPs can modulate mechanisms that regulate the delicate balance behind haemostasis and haemostatic regulation, in both procoagulant and anticoagulant directions. Blood clotting is a complex process, typically initiated by the adhesion of platelets at a site of vascular injury, followed by an activation cascade of soluble clotting factors, which generate an insoluble network of cross-linked fibrin proteins. Some types of NPs were shown to affect platelet adhesion, the initial step. Silver NPs, either uncoated²³ or PEGylated (PEG, polyethylene glycol)²⁴, reduce platelet adhesion by altering the conformation of the adhesion protein integrin α_IIbβ₃, which has anticoagulant effects in mice by ultimately reducing the downstream generation of thrombin and fibrin. Similarly sized uncoated gold NPs have no comparable antiplatelet effects²⁴. The effects on haemostatic regulation can also be mediated by NP interaction with secreted clotting factors. In this context, cationic poly(amidoamine) dendrimers acted as a procoagulant by promoting the formation of fibrinogen aggregates via a thrombin-independent mechanism, in which the cationic dendrimer interacts directly with fibrinogen²⁵. The clot-forming potential of erythrocytes can be affected by NPs as well, as titanium dioxide NPs upregulate phosphatidylserine exposure on erythrocytes, which facilitates thrombin formation and erythrocyte aggregation in rats²⁶. Furthermore, NPs might modulate endothelial cell adhesiveness: silica NPs were observed to upregulate endothelial PECAM (platelet endothelial cell adhesion molecule) expression and increase platelet adhesion to the vascular wall, which could contribute to thrombus formation²⁷. It is unlikely that nanotherapies would function more effectively than the existing drugs for systemic haemostatic regulation; however, these effects could be exploited for localized haemostatic effects consistent with another therapeutic objective, for example, by enhancing the uptake of NPs into tumours by inhibiting intratumorally platelet activation²⁸. However, these effects are potentially also undesirable off-target effects and could contribute to risks such as brain ischaemia or cardiac infarction. NP therapies thus might be contraindicated in patients receiving anticoagulants.

NPs can potentially affect the formation and pathology of atherosclerotic plaques. In atherosclerosis, fatty lesions develop along the wall of blood vessels, which impairs blood flow and ultimately increases the risk of ischaemia and acute infarct or stoke. These lesions are typically inflamed due to infiltration by macrophages, which become laden with phagocytosed lipids, such as low-density lipoprotein, and promote disease progression. Long-term inhalation exposure to titanium dioxide²⁹ and zinc oxide³⁰ NPs, which are commonly found in many consumer goods, has been shown to induce blood dyslipidaemia in rodents by an unknown mechanism. These changes, which include the elevation of low-density lipoprotein and total cholesterol, and the elevation of serum triglycerides with a subsequent increased risk of atherosclerosis and plaque formation. Further, risks could arise from NPs meddling with inflammatory functions, as above in atherosclerotic plaques, and both possibly enhance their formation or slow their progress.

Microbiotic regulation

There are relatively few identified therapeutic applications for orally administered nanomedicines. However, NPs are already abundant in consumer products, which include sunscreens, toothpaste, cosmetics, topical medications and food additives. The commonality of nanomaterial in these products means that exposure via the oral and inhalation routes is unavoidable and largely unknown. The microbial ecosystem of the gastrointestinal tract, the gut microbiome, is a highly complex and dynamic biological milieu; current research is revealing an ever-expanding effect of the gut microbiome on multiple systems. Resident microbiota can be beneficial, commensal or pathogenic to the host; thus, upsetting the delicate balance of this ecosystem might impair digestive efficiency or leave the host more vulnerable to opportunistic infections. The gut microbiome has recently been implicated in the modulation or manipulation of an unexpectedly wide array of host biological processes, which include metabolism, immunoregulation and even behaviour. Although a mechanistic understanding of the specific interactions between microbes and host remains poorly understood, NPs may also interact with bacteria or other microorganisms in the microbiome, and thus potentially impact the poorly understood interactions between host and microorganism³¹. From a product safety and public health perspective, the effects of internalized NPs on the microbiota of these organs, and their broader health consequences, remains largely an unresolved question.

Silver and silver-composition NPs possess broad antimicrobial activity and are frequently incorporated into topical medical treatments for burns as an anti-infective agent. However, different species and strains of bacteria exhibit varying degrees of sensitivity to silver toxicity³². Thus, silver NPs affect certain constituents of the bacterial species commonly found in the gastrointestinal tract more than in others: the oral intake of silver NPs alters the microbiome homeostasis in mice and rats^33,34,35,36. Gut biome perturbations are not exclusive to silver NPs; exposures to gold nanoclusters³⁷, titanium dioxide NPs³⁶ and zinc oxide NPs³⁸ have also demonstrated the alteration of gut homeostasis in mice. In addition to bacteria, bacteriophages are an integral component of the microbiome, and represent a phage virome that interacts with and modulates the gut bacterial ecology. The phage virome composition might also be impacted by NP exposure; for example, the long-term intake of silver NPs by rhesus monkeys was shown to reduce the gut population of bacteriophages known to prey on pathogenic bacteria³⁹.

The gut microbiome is also only one of many identified microbiomes in the human body. Though poorly studied, the lung microbiome is believed to have roles in the regulation of host immunity and homeostasis similarly to the gut microbiome and is also easily accessible to inhaled external nanomaterials. Microbicidal silver NPs disrupt the composition of the mouse lung microbiome in a coating-dependent manner, with citrate-coated silver NPs being the most disruptive⁴⁰. To make matters even more complex, NPs could also exert effects on tumour microbiomes, the relevance of which to cancer has only recently come into focus^41,42,43,44.

The current limited understanding of the relationship between microbiome and host biology constrains our ability to interpret the relevance of these NP-mediated perturbations of microbial populations, or to determine whether the induced changes might be beneficial, detrimental or inconsequential. However, some ancillary effects on the host were observed in a few of the aforementioned studies. For example, rats fed with silver NPs exhibited changes in behavioural tests and histopathological alterations in the brain, although it is unclear whether the NP-induced microbiome disruption was a causative factor³⁴. In another rat study, silver-NP-induced microbiome changes were accompanied by a decrease in expression of the immunomodulatory genes in intestinal tissue, which suggests a modulation of the host immune response³³.

Immunomodulation

NPs exhibit a wide range of effects on the immune system, even in translationally relevant forms. NP-mediated immunomodulation can be pro- or anti-inflammatory and downstream may affect other activities not directly related to inflammation, such as tissue remodelling or scavenging functions, as well as immune cell differentiation, maturation and senescence. To emphasize the variability of ancillary effects that arise from NP–biological interactions, we highlight notable studies that illustrate the divergent immunomodulatory effects of NPs on unwanted systemic immunomodulation. The elicited ancillary effects are highly dependent on the physicochemical properties of the NPs, the immune cell subset interacting with NPs and the nature of the surrounding environment.

Immunostimulation and pro-inflammatory effects

The potential of many nanoscale materials to induce inflammation in organs and tissues is probably one of the earliest observed effects of nano–bio interactions. It is highly likely that ancillary mechanisms are integral to NP-mediated inflammatory responses. NP-triggered inflammatory-response mechanisms are highly diverse and complex, but generally can be categorized into one of three mechanistic schemes: the intracellular route, wherein NPs are taken up by sentinel immunocytes or other cells and provoke inflammation through various intracellular mechanisms; the Fc (fragment crystallizable)-mediated immune response route, wherein humoral immunogenicity of the nanomaterial results in extensive opsonization with NP-specific antibodies and subsequent Fc receptor activation; and the complement system activation route, caused by the extracellular interaction of complement immunity factors with nanomaterials. With respect to the intracellular pathway of inflammation activation, there are any number of mechanisms by which cellular binding or the uptake of nanomaterial could provoke inflammation. Such possibilities are too complex and varied to discuss in detail in this Review, and are extensively covered elsewhere¹; however, such complexity is illustrative of the diversity of NP-mediated effects in biology. Generally, an inflammatory response will be driven by the release of inflammatory mediators (tumour necrosis factor, interleukins, interferons, ATP and so on) from cells exposed to nanomaterial. Such a response could be provoked in any number of known (or unknown) ways: for example, NP uptake leads to endosomal build-up and leakage followed by NP contact with cellular machinery, which causes varied responses, including protein unfolding and interference with cell signalling cascades, disruption of cellular metabolism and increased oxidative stress, and activation of damage-associated molecular pattern receptors and autophagic pathways. These effects and others, all ultimately, lead to the release of inflammatory mediators by secretion or by cell lysis, which triggers recruitment and activation of the surrounding immune system and paracrine inflammatory responses in neighbouring cells. Importantly, the nature and magnitude of such effects are highly dependent on NP identity (both pristine and biological) and biological context. The activation of the complement system by NPs is another important ancillary effect that is also a recurring clinical problem with approved nanodrugs in current use, and therefore is described in greater detail further on in this section.

NP-induced immune activation, generally considered an unwanted side effect, could be medically useful in certain circumstances: for example, to combat infection, to provoke a host defence response against cancer or to stimulate the antigenicity of vaccines. The pro-inflammatory nature of many nanomaterials may therefore be useful as a supplemental adjuvant for vaccines. Indeed, microparticle formulations of aluminium compounds (for example, Alhydrogel) have been in clinical use in adjuvanted vaccines for decades (and gave rise to false and unwarranted but ongoing controversies). Investigations of nanoscale formulations of aluminium showed an improved adjuvant activity compared with those of microscale forms, as well as a reduced induction of local inflammation⁴⁵. This improvement was attributed to the increased potency of nanoalum in NLRP3 inflammasome activation in antigen-presenting cells (APCs)⁴⁶. Conventional microscale aluminium-based adjuvants are known to elicit a strong T helper cell 2 (T_H2)-type immune response by CD4⁺ helper T cells. Although efficient in provoking a strong and durable humoral immune response to antigens by B cells, they lack a strong induction of a T_H1-type helper T cell immune response and thus are not particularly effective at stimulating persistent cell-mediated immunity by the activation of CD8⁺ T cells. Thus, their utility is limited in vaccines against diseases for which antibody responses are not particularly effective (for example, the human immunodeficiency virus). However, nanoscale aluminium particles may be able to circumvent this deficiency. A polyacrylic acid-stabilized nanoscale formulation of Alhydrogel elicited a potent T_H1 immune response in a murine model⁴⁷, which suggests that NP adjuvant formulations could be used to appropriately tune the adaptive immune response in vaccinations for which a stronger cell-mediated immunity is required. Additionally, other nanomaterials were found to exhibit strong adjuvant activity with a minimal local inflammation, such as biocompatible lipid NP formulations^48,49,50, which could provide a promising adjuvant alternative to the irritation-producing aluminium salts or Toll-like receptor agonist adjuvants, and most recently used in COVID-19 vaccines. Immunostimulatory ancillary effects could also conceivably be useful in cancer; such effects and applications are discussed in the ‘Ancillary effects in cancer and cancer therapy’ section below.

Complement activation

In addition to leukocytes, the complement system is affected by NPs as well. Complement is a biochemical defence that comprises secreted factors, which act in parallel to and in conjunction with the other arms of the immune system. The complement system is driven by the interaction of complement factors with biochemically foreign surfaces and has three main functions: direct lysis of foreign cells (the ‘classical pathway’), opsonization of foreign cells and debris to facilitate clearance (the ‘alternative pathway’) and induction of inflammation to attract phagocytes (the ‘lectin pathway’). Complement immunity factors are often found within NP coronas; interaction with the NP surface or with other corona biomolecules can trigger activation of complement factors, which potentially activate any of the three complement pathways⁵¹. However, activation of the alternative pathway appears to be the most prevalent and best-characterized effect of NP interaction with this system^52,53. Alternative pathway activation by dextran-coated iron oxide NPs has been closely studied. The initial step in this process appears to be the adsorption of complement factor C3 onto the NP corona. The resultant molecular interaction between C3 and other molecules on the NP surface can induce spontaneous hydrolysis of C3 into two subunits, C3a and C3b. C3b within the NP corona can further drive complement activation by forming an enzyme complex on the NP surface with other complement factors; this complex (AP convertase) catalyses the cleavage of intact C3 into C3a and C3b, as well as the cleavage of the complement factor C5 into C5a and C5b. The cleavage products of C3 and C5 can activate complement factor receptors in many tissues, and potentially induce a cascade of biological responses (Fig. 2). C3a and C5a are also known anaphylatoxins that elevate the risk of severe immune reactions. The potential of iron oxide NPs to activate the alternative pathway appears to be independent of the dextran structure, functionalization and dextran/iron molar ratio, which suggests that the corona is the primary driver of C3 cleavage⁵⁴.

Complement activation is especially problematic for nanomedical purposes, as it can contribute to suboptimal NP pharmacokinetics and biodistribution, inflammation, allergy and potentially a serious anaphylaxis-like condition, CARPA (complement activation-related pseudoallergy), which has been observed with several clinical nanomedicines. The complement system can be activated even by highly biocompatible NPs, such as poly(lactic-co-glycolic acid) (PLGA)⁵⁵, carbohydrate-coated iron oxide NP⁵⁶ or liposomes;⁵⁷ with respect to the last of these, increased activation was associated with a high negative surface charge and morphological deviation from sphericity. CARPA-like patient reactions have occurred with several clinical nanomedicines, including Doxil (PEG-liposomal doxorubicin)⁵⁸ and Feraheme (ferumoxytol)⁵⁹, which received an FDA black box warning as a result.

Although generally considered undesirable, NP complement activation could conceivably be leveraged for therapeutic benefit as well. When co-administered with a model antigen, rod-shaped PEG NPs promoted the alternative pathway of complement activation, which provided an adjuvant-like boost to the immune response and improved humoral immunogenicity⁶⁰. Chitosan NPs have also exhibited adjuvanticity via the activation of complement and were investigated extensively^61,62,63. Such particles could form the basis of vaccine nanoadjuvants free of metals and endotoxin derivatives.

Immunomodulation, immunosuppression and anti-inflammatory effects

We now understand that NPs can be immunoregulatory or immunosuppressive in addition to being inducers of inflammation⁶⁴. Immunosuppressive characteristics of NPs have been observed in interactions with multiple immune cell classes. Such effects can include immune cell cytotoxicity or senescence, decreased antigen presentation, diminished or altered cytokine release profiles and attenuated reactivity to pathogen-associated molecular signals.

Nanomaterial immunosuppression and anti-inflammation have potential applications, especially in the context of injurious immune overactivation. Clinical immunomodulatory agents, such as corticosteroids and mTOR inhibitors, are often used to mitigate localized pathological inflammatory responses; however, these drugs have broad off-site and off-target systemic effects, which result in a complicated side-effect profile. Even without targeting, NPs are frequently preferentially incorporated into monocytes, macrophages and other myeloid-derived cells with immunoregulatory functions, and thus represent a potential passive targeting tactic for nanomaterials with immunomodulatory properties. For example, negatively charged particles composed of the biodegradable PLGA copolymer, a nanomaterial, have shown the capability to exhibit immunomodulatory effects in inflammatory monocytes. In particular, PLGA particles exhibited therapeutically beneficial anti-inflammatory and immunosuppressive effects in multiple disease models. PLGA NPs passively target inflammatory monocytes that express the scavenger receptor MARCO (macrophage receptor with collagenous structure), which promotes splenic sequestration and apoptosis and results in a depletion of these cells from distal sites of inflammation. PLGA particles were shown to act therapeutically as an anti-inflammatory agent in murine models of West Nile virus-induced encephalitis, experimental autoimmune encephalomyelitis, cardiac infarction, kidney reperfusion injury and inflammatory bowel disease⁶⁵. These effects were achieved without apparent systemic immunodeficiency, a common and potentially severe side effect of anti-inflammatory drugs.

Additionally, internalized PLGA nanomaterial can suppress co-stimulatory signals from APCs, such as macrophages and dendritic cells. The mechanism behind these PLGA-mediated ancillary effects in the immune system is not well understood, but appears to be dependent on acidification caused by degradation of the particle into lactic acid and glycolic acid; therefore, it is probably a product of the pristine, rather than the biological, identity of the NP⁶⁶. Highly promising applications for NPs with immunosuppressive properties (and perhaps one of the most promising for nanomedicines to date) are in the development of tolerogenic NPs for the treatment of allergy and autoimmune disease, and in minimizing rejection in human organ transplantation without systemic immune suppression. By exposing an APC to an antigen under immunosuppressive conditions, it is possible to teach the adaptive immune system to ‘forget’ a pathological adaptive immune response, as the antigen is presented to T cells in the absence of a co-stimulatory signal, which results in ablation of antigen-reactive T cells⁶⁷ (Fig. 3). This requires coupling an antigen to an immunosuppressive delivery NP; the delivery of the antigen to an APC (with the activation suppressed by the internalized NP) can cause the reactivity to a known antigen to be diminished or eliminated. Separate administration of the inhibitory agent has not been found to be effective; co-delivery of both the antigen and immunosuppressive agent as a unified entity appears to be required. Although the co-packaging of an mTOR inhibitor or other immunosuppressive drug payload is probably the most clinically practical configuration for a tolerogenic NP, the induction of tolerance has been demonstrated using only NPs with immunosuppressive qualities. Antigen-conjugated PLGA NPs have been shown to tolerize allergic airway inflammation in murine models^68,69,70. Furthermore, PLGA NPs decorated with donor peptides reduced allograft rejection in several murine transplant models, which includes in skin⁷¹ and pancreatic islet cell⁷² transplants, a strategy that could be an alternative to broadly immunosuppressive antirejection drugs, such as sirolimus or tacrolimus.

Ancillary effects in cancer and cancer therapy

NP effects are especially interesting in the context of cancer, because prototypic nanomedicines are frequently designed as potential cancer therapeutics (usually as a drug delivery system and often coupled with targeting and/or imaging modalities). However, the ancillary effects of the NP carrier itself on cancer biology are often not considered. The biology of cancer is atypical, dysregulated and extremely heterogeneous, and therefore presents a unique environment for NP-mediated ancillary effects. Furthermore, the uptake of NPs into tumours is often amplified due to a poorly defined vascular phenomenon described as the enhanced permeation and retention (EPR) effect, wherein the biodistribution of administered NPs is increased in tumours⁷³. Thus, biomedical NPs infused into to a patient are likely to be deposited into the tumour microenvironment at an elevated concentration (although the magnitude of this effect is highly variable and often less pronounced in humans)⁷⁴. Therefore, we must consider the interaction of the NP with biology in a cancer setting. Here we highlight instances in which ancillary effects were observed that may have translational relevance in cancer therapies, or instances in which they may facilitate cancer growth and progression (Fig. 4).

Notably, some NPs may exhibit carcinogenicity by the induction of cellular changes that promote cancer development in normal tissues. This is a well-known property of nanomaterials with limited clinical relevance, such as carbon nanotubes, which have been shown to persist in the lungs of rodents after inhalation, and eventually cause mesothelioma or mesothelioma-like lung injury^75,76,77,78. However, studies indicate a potential mutagenicity or genotoxicity from commonly used NPs, such as titanium dioxide NPs, which are nearly ubiquitous in consumer products. Although there is no conclusive evidence of a direct causal effect in carcinogenesis via typical exposure routes and at reasonable exposure levels and durations, titanium dioxide (inclusive of nanoscale forms) has recently been classified as a Group 2B substance by the International Agency for Research on Cancer, which recognizes its possible carcinogenicity⁷⁹.

The immune cell compartment of tumours plays an important role in cancer development, as many cancer types can effectively ‘reprogramme’ or suppress infiltrating leukocytes to facilitate growth and metastasis. However, the modulation of immune cell responses is also an ancillary NP effect that could have interplay with tumour-mediated immunomodulation, and opsonized NPs are readily scavenged by the macrophages, neutrophils, dendritic cells and monocytes typically present in the tumour microenvironment. Consequential NP ancillary effects on the immune cell compartment within a tumour, and subsequent effects on the tumour itself, have so far been mostly overlooked, but there is increasing evidence that NP-mediated ancillary effects can, indeed, impact the surrounding cancer biology. Ferumoxytol, a clinical carboxymethyldextran-coated iron oxide NP, has shown promise as an agent to activate macrophages and counteract tumour-mediated manipulation of tumour-associated macrophages (TAM) phenotype. High-dose ferumoxytol repolarized TAM and liver Kupffer cells to an M1-like pro-inflammatory, cancer-inhibiting phenotype in a murine breast cancer–liver metastasis model; this slowed tumour growth and shielded the liver from metastatic seeding⁸⁰. TAM repolarization is believed to occur through cellular iron enrichment. Abundant intracellular iron acts as a Fenton chemistry catalyst, which generates superoxide radicals and drives the adoption of the M1 phenotype (Fig. 5a). Thus, this effect is primarily elicited by the composition of the particle (pristine identity), whereas the passive targeting of the particle to TAM is driven largely by the biomolecule corona of ferumoxytol (biological identity).

Myeloid-derived suppressor cells (MDSCs), an immunosuppressive type of immature monocyte that often resides within tumours, are activated by gold NPs functionalized with a biomimetic high-density-lipoprotein-like surface via the binding and activation of SCARB1 (scavenger receptor class B type 1), which pushes myeloid cells out of the pathological MDSC phenotypic niche. This shift reduced the tumour growth in a lung cancer model, and both metastasis and growth in a melanoma model. The reduction in the tumour MDSC burden also resulted in an increase in active tumour-infiltrating cytotoxic T lymphocytes, which presumably increased the immune surveillance of the tumours⁸¹. These effects are interesting notably because lipoproteins are a common constituent of NP coronas. High-density apolipoproteins do not directly bind SCARB1; however, cholesteryl esters contained within high-density lipoprotein will specifically bind and activate the receptor, which raises the possibility that other NPs could elicit this effect as well. As this gold NP was not synthesized with cholesteryl esters, the biological activity of the particle that activates SCARB1 is probably derived from sterol esters incorporated into the corona from serum.

On the opposing end of the immune response spectrum, NPs may also modulate tumour immune cells in ways that aid cancer progression. Studies of unloaded liposomes that lack a drug payload are uncommon; in one rare study, it was shown that the liposome structure itself might have an adverse effect on cancer through interactions with immune cells⁸². PEGylated liposomes without a drug cargo elicited a tumour-promoting M2-like phenotype in TAM in a papillomavirus-induced tumour model, which resulted in a higher secretion of T_H2 cytokines and local immunosuppression in the tumour microenvironment. Treatment with unloaded liposomes consequently increased the tumour growth rate⁸². Thus, consideration must be given to the NP composition for drug-delivery nanopores intended for cancer therapy, as the NPs themselves might provoke responses that are counterproductive to the therapeutic mechanism.

Outside the immune cell compartment, other types of tumoral stroma cells are susceptible to antitumorigenic effects mediated by nanobiological interaction, albeit indirectly. In the absence of any other drug payload or concomitant therapy, 20 nm gold NPs reduce growth and metastasis of pancreatic ductal adenocarcinoma in mice. Although the effect mechanism is not immediately clear, analysis of the secretome of pancreatic cancer cells demonstrated a marked downregulation of many progrowth factors and cytokines. Correspondingly, there was a downstream reduction in extracellular matrix secretion by stromal pancreatic stellate cells, both of which reduce cancer aggressiveness⁸³. Gold NP monotherapy has also shown efficacy in rodent models of ovarian cancer, as it reduces both tumour proliferation and metastasis by interruption of the epithelial-to-mesenchymal transition (EMT) of cancer cells, possibly by inhibition of mitogen-activated protein kinase signalling through an unknown mechanism⁸⁴. In both models, gold NPs were not cytotoxic to non-cancerous cells, which suggests that the effect was due to a specific interaction of the NP or NP-associated biomolecules with the dysregulated biology of cancer cells.

Furthermore, NPs may directly contribute to programmed cell death pathway activation in cancer cells. Recently, our group showed that iron-containing NPs can modulate an iron-dependent cell death pathway called ferroptosis (Fig. 5b). The overload of ferroportin-deficient acute myeloid leukaemia cells with iron via ferumoxytol uptake generates large amounts of ROS by Fenton chemistry, which causes a build up of oxidized lipids and so triggers ferroptotic cell death in the context of low ferroportin expression¹⁵. Similarly, fluorescent silica carbon-quantum dots (C-dots) functionalized with a melanoma targeting peptide also induce ferroptosis in melanoma under cysteine-deficient conditions (Fig. 5d). Remarkably, pristine C-dots contain no iron; however, C-dots apparently accumulate iron from the biological environment, perhaps incorporating metallic ions into the outer silica layer or the corona¹⁶. Therefore, proferroptotic NPs might be useful as a novel therapeutic for cancers with cellular iron export defects or be used in combination with another inducer of oxidative stress.

Perhaps more consequentially, vascular integrity perturbation by NPs might actually facilitate metastasis. The effects described by Peng et al. wherein titanium dioxide NPs disrupted VE-cadherin binding and induced leakage in the lung vasculature by (described in ‘Effects on vasculature and biological barriers’) were also applicable to the tumour vasculature; the increase in vascular permeability was shown to facilitate both extravasation of the tumour cells from the primary tumour and to promote the ‘seeding’ of metastatic sites by an increased permissiveness of the circulating tumour cells in the normal vasculature (Fig. 5c)¹⁹. Further evidence suggests that this effect is not unique to titanium dioxide, and could occur with gold, silver and silica NPs as well¹⁹. An additional study determined that PEGylated gold NPs of various sizes are predominantly taken up into tumours via the transcellular route through active transport, and not via the passive paracellular diffusion process commonly attributed to EPR; these internalized NP conceivably could exert ancillary effects on endothelial cell biology from within⁸⁵. Both of these findings are potentially concerning for cancer nanotherapy development, as the EPR effect may not be a completely passive process; diffusing NPs might actually influence or alter the vascular endothelium through which they pass. However, this effect might be therapeutically exploitable to facilitate the uptake of other NPs or drugs out of the blood compartment, but must be carefully balanced against a possible increased metastasis. It remains to be seen which NP characteristics mediate this effect, and whether this effect is recapitulated by NPs of other classes.