Exploiting the cancer niche: Tumor-associated macrophages and hypoxia as promising synergistic targets for nano-based therapy

Thetumormicroenvironmenthasbeenwidelyexploitedasanactiveparticipantintumorprogression.Extensive reports have de ﬁ ned the dual role of tumor-associated macrophages (TAMs) in tumor development. The protumoral effect exerted by the M2 phenotype has been correlated with a negative outcome in most solid tu- mors.Thehighin ﬁ ltrationofimmunecellsinthehypoxiccoresofadvancedsolidtumorsleadstoachainreaction of stimuli that enhances the expression of protumoral genes, thrives tumor malignancy, and leads to the emer- gence of drug resistance. Many studies have shown therapeutic targeting systems, solely to TAMs or tumor hypoxia, however, novel therapeutics that target both features are still warranted. In the present review, we discuss the role of hypoxia in tumor development and the clinical outcome of hypoxia-targeted therapeutics, such as hypoxia-inducible factor (HIF-1) inhibitors and hypoxia-activated prodrugs. Furthermore, we review the state- of-the-art of macrophage-based cancer therapy. We thoroughly discuss the development of novel therapeutics that simultaneously target TAMs and tumor hypoxia. Nano-based systems have been highlighted as interesting strategies for dual modality treatments, with somewhat improved tissue extravasation. Such approach could be seen as a promising strategy to overcome drug resistance and enhance the ef ﬁ cacy of chemotherapy in ad- vanced solid and metastatic tumors, especially when exploiting cell-based nanotherapies. Finally, we provide an in-depth opinion on the importance of exploiting the tumor microenvironment in cancer therapy, and how this could be translated to clinical practice.


Introduction
Cancer is amongst the leading causes of mortality and morbidity worldwide. The latest statistics point to an estimated 14.1 million new cases and 8.2 million cancer-related deaths [1]. It has been characterized by an uncontrolled cell proliferation, which is often associated with vascular abnormalities and cell invasiveness. In the past few years, the hallmarks of cancer have been revised and an active participation has been assigned to the tumor microenvironment ( Fig. 1) [2]. Studies have shown that cancer cells evolved to promote angiogenesis, metastasis and survival in response to several factors within the tumor microenvironment, such as pH, growth factors, oxygen levels and the presence of immune cells [3][4][5]. Tumor cells adaptation to the environment is considered essential to maintain their survival and growth. For instance, tumor cells grow under low levels of oxygen and nutrients, develop new blood vessels, a process called 'de novo angiogenesis' to compensate for the lack of oxygen and nutrients. These newly formed blood vessels contain discontinuous endothelium, which renders them leaky in nature. This vascular hyperpermeability, in combination with impaired lymphatic drainage, are known as enhanced permeation and retention effect (EPR) [6]. Surgery, radiotherapy, and classical chemotherapy are still the first options for many types of tumors [7]. However, the associated side effects and the emergence of multidrug resistance (MDR), have limited the clinical use of most common therapeutic compounds [8]. Therefore, there is an unmet need to develop novel approaches and therapeutics to target the tumor cell and its microenvironment. The heterogeneity and complexity of tumor microenvironment promote cancer survival and progression. Lately, research has focused on the involvement of tumor-associated macrophages (TAMs) in cancer progression. Studies have explored the link between the secretion of TAM chemoattractants by the tumor cells and the consequent upregulation of tumor-promoting genes by these immune cells, in response to the microenvironment stimuli [9,10]. Additionally, hypoxia has been found as a critical factor for the survival of large tumor masses and therefore a key target for the development of targeted therapies [11]. Furthermore, high infiltration of TAMs are found in hypoxic tumor cores, promoting resistance to classical chemotherapeutics. This microenvironment may be used to develop sophisticated fine-tuned nanobased systems, capable of enhancing therapeutic extravasation into tumor hypoxic cores. Great efforts should be made to promote the rational design of delivery systems that could achieve high therapeutic efficacy by simultaneously targeting TAMs and hypoxia.

Cancer hypoxia
Structural abnormalities in tumor vessels lead to reduced oxygen diffusion to tumor cells and eventually, necrotic cores. It has been fifty some years since Thomlinson and Gray first postulated the role of hypoxia in human tumors [12]. Hypoxia has an active role in oncogenesis and contributes to the overall survival of tumors. In normal tissues, oxygen levels are heterogeneous and physiological pO2 can range between 20 mmHg in the liver and brain to 70 mmHg in the kidney (3.1-8.7% O2). In contrast, a decrease to about 10-30 mmHg of pO2 is observed in tumors. Most importantly, 82% of all oxygen readings taken from solid tumors, present a 0.33% O2 reading (as low as 2.5 mmHg) [13,14]. The level of hypoxia within tumors increases during tumor progression. Chronic hypoxic cells have been described as prone to higher proliferation and survival. Aggressive phenotype with increased resistance to therapy has been associated with patients with highly hypoxic tumors, highlighting the clinical significance of hypoxia [15].
Hypoxia-inducible factors (HIFs), essentially HIF-1 have been linked to hypoxic tumor microenvironments. It is commonly overexpressed in solid and metastatic tumors including breast, prostate, colon, lung, pancreatic, head and neck cancer [16]. This molecule functions as a heterodimeric transcription factor composed of HIF-1α and HIF-1β, whose dimerization is regulated by an oxygen-dependent prolyl hydroxylase. When oxygen levels decrease, HIF-1α accumulates and translocates to the nucleus, where it forms the active transcription factor HIF-1 by binding to HIF-1β. This molecule regulates a plethora of genes in cancer biology and metabolism, controlling the proliferation rate, metastasis, and aggressiveness of cancer cells [17]. It also potentiates tumor cells resistance to radio and chemotherapy [18][19][20]. Although some controversy exists on using HIF-1 as an actual target for hypoxia, many novel therapies have been exploited, which will be discussed below, suggesting the relevance of hypoxia in cancer therapy.
Several therapeutic agents, such as anti-angiogenesis (VEGF inhibitors) [14,21,22], tumor MAC depleting agents [23], and androgen deprivation therapy [19], have shown to induce tumor hypoxia. Most of these drugs work by shutting the blood supply to the tumor, reducing nutrient and oxygen delivery to the cancer cells, which results in tissue hypoxia. Photosensitizers are another class of cancer therapeutics, which consumes tumor oxygen upon light activation to produce reactive oxygen species (ROS), leaving lower oxygen levels in the tumor microenvironment [24,25]. Drug-induced hypoxia has been considered indeed as a key factor that reduces the therapeutic efficacy of a wide range of anticancer treatments. However, recent studies have shown that careful selection of the combinatory treatment, as well as, their sequence, could convert hypoxia from a challenge into a potential therapeutic target. For instance, a recent study combining Topotecan (hypoxia-induced factor inhibitor) with Bevacizumab (antibody directed against VEGF) showed a synergistic antitumor activity in glioblastoma xenograft models [28]. A study by Mabjeesh et al., also identified a novel antiangiogenic compound (2ME2), which induced downregulation of HIF-1 at a post-transcriptional level [21]. The data showed a significant antitumor effect, given HIF-1 induced down-regulation of VEGF. This rational provided evidence that dually targeting HIF-1 and other indirect pathways may be a robust strategy to overcome off-target therapy, and translate HIF-1 directed therapies to clinical development. Furthermore, a new interest has focused on designing multifunctional nanoparticles to deliver a combination of photosensitizers with hypoxiaactivated prodrugs (HAPs) to tumor tissues, where induced hypoxia could activate HAPs, leading to a significant cell death and tumor growth inhibition [24,25]. The latter approach will be discussed in more details in Section 2.2 of this review.

HIF-1 inhibitors
The active role of hypoxia and HIF molecule has positioned them as novel therapeutic targets for cancer therapy [17,26]. Classical approaches to overcome low oxygenation levels at the tumor site was carried out using hyperbaric chambers. Such intervention was proven to be unsuccessful in combination with radio and chemotherapy [14]. Therefore, new approaches have been introduced to target tumor hypoxia and small molecules that specifically target HIF-1, a master transcriptional factor that regulates cancer progression and development, are highly attractive. Existing therapeutic options, consider HIF-1 as a new gold standard target to treat cancer. Several small inhibitors have been developed and can inhibit HIF-1′s activity by promoting HIF-1 protein degradation, or by blocking at least one of the following pathways: 1) HIF-1 mRNA expression; 2) HIF-1 protein translation; 3) HIF-1 DNA binding and 4) HIF-1 transcriptional activity [27,28]. The demanding role and complex network underlying the HIF molecule indirect inhibitors have been thoroughly reviewed [29]. These include molecules that target upstream or cross-talked pathways with the HIF target (e.g. inhibitors of VEGF, mTOR, epithelial growth factor receptor (EGFR), topoisomerase I and II, PI3/AKT/MAPK pathways). Despite the discovery and development of several therapies targeting HIF-1 or HIF-1 pathways, only a few have progressed into clinical development (Table 1). There have been no approved drugs that directly inhibit HIF-1, and many small molecule inhibitors have shown a high rate of late-stage clinical failure. This is attributed to the high redundancy and complexity of the tumor microenvironment. Furthermore, the desired effects of indirect inhibitors may be difficult to separate, as many different signaling pathways are linked to HIF-1 induced tumorigenesis, leading to the existence of off-target effects that are likely to be less predictable [30]. However, as mentioned above, recent studies have shown that dually targeting HIF-1 and other indirect pathways may be a strong strategy to overcome off-target therapy and translate HIF-1 directed therapies to clinical development [21,22,31].
Although there is a much better understanding of the pathway today than in the early 1990s when HIF-1 was discovered, its dual role in promoting cell survival and inducing apoptosis has caused controversy amongst researchers as to whether it should be considered as a therapeutic target [32]. Nonetheless, selective gene therapy can be achieved by designing therapeutic genes, which are controlled by response to binding of HIF-1 to HREs (hypoxia responsive elements). Shibata et al. proved this concept through the development of vectors containing a bacterial nitroreductase gene, with a responsive domain to VEGF [49]. This system was used to selectively activate an anticancer prodrug (CB1954). In a similar approach, the HRE vector was for human flavoprotein cytochrome P450 reductase, which was also used to selectively activate RSU1069 prodrug [50]. Both studies showed increased cytotoxicity in hypoxia, compared to normaxia.
Hypoxia-driven triple suicide genes have also been explored to increase the cytotoxicity of ganciclovir (GCV) and 5-fluorocytosine (5-FC) both in vitro and in vivo [51]. Additionally, antisense or small interfering RNA (siRNA) plasmids injected in hypoxic tumors has shown to eradicate EL-4 thymic lymphoma in combination with angiostatin [52]. Although these systems have shown an interesting outcome, offtarget effects are still an issue, if injected intravenously. The use of drug carriers could further improve the efficacy and safety of gene delivery to the hypoxic tumor microenvironment.

Hypoxia-activated prodrugs
The concept of bio-reductive prodrugs (also known as hypoxia-activated prodrugs) has arisen, and the use of drugs that are non-toxic until they are reduced under low oxygen levels has opened the door to specific systemic treatments of solid and metastatic tumors [53]. Hypoxia-activated prodrugs are defined as cytotoxins that are metabolized by various endogenous reducing enzymes (oxireductases such as NAD(P)H, cytochrome P450) and quinone oxidoreductase (NQO1 in a one-electron or two-electron catalysis). The one step electron reduction is a selective process, as it composes a reversible step that generates a prodrug free radical that can easily back-oxidized to its original compound if oxygen is present, conferring specificity to highly hypoxic regions. Contrary to this, the two-electron reduction fails to produce an oxygen sensitive intermediate, being highly toxic, even for tissues in normaxia. The choice of which reduction occurs depends on the structure of the compound itself and the differential expression of reducing enzymes in tissues [20,54]. The different classes of hypoxia-activated prodrugs are: (nitro (hetero) cyclic compounds; aromatic N-oxides; aliphatic N-oxides; quinones and metal complexes. These drugs are thoroughly summarized in Table 2, with the indication of their overall success in clinical trials. Although the values of these prodrugs are highly dependent on their relative metabolism and penetration, many have progressed significantly and showed promising applications as selective hypoxic cytotoxins [53].
Tirapazamine (TPZ) is the most clinically advanced hypoxia-activated prodrug [45]. Early clinical studies for non-small cell lung cancer (NSCLC), head and neck-cancer, metastatic melanoma (Phase I and II) and other solid tumors, showed enhanced tumor inhibition for this agent [55,56]. This was achieved by co-administering TPZ with radiosensitizing and platinum agents. Furthermore, a dose-defining study of TPZ combined with embolization in liver cancer is now recruiting [57]. However, unexpectedly, a recent randomized clinical trial combining radiation and cisplatin, with TPZ, for cervical cancer (Phase III) [58], showed that the combined treatment was not superior to cisplatin alone, since no change in the overall survival rate was observed. Although these results seem disappointing, one must take into account the small population size for this particular trial, and that the patients may constitute a non-selective hypoxia sub-population [60]. Therefore, we believe that it is essential to overlook apparent clinical (continued on next page) failure and understand the overall limitations of these pro-drugs. This can be achieved by using in vitro systems that can accurately validate the therapeutic efficacy of these drugs. Furthermore, translation to clinical scenarios should be based on selecting hypoxia sub-populations. Evofosfamide (TH-302) is an investigational hypoxia-activated prodrug that releases the DNA alkylator bromo-isophosphoramide mustard under hypoxic conditions [66]. Phase I/II clinical trials have shown promising results for the drug as both a single agent and in combinatorial treatments with gemcitabine and nab-paclitaxel for pancreatic cancer [67,68,76]. A phase I/II Open-label Study of TH-302 and dexamethasone in subjects with relapsed/refractory multiple myeloma is now recruiting [77]. Phase II placebo-controlled study of TH-302 in combination with Pemetrexed in patients with non-squamous nonsmall cell lung cancer has started (NSCLC) [78]. PR-104 is a water soluble phosphate ester that has shown rapid hydrolysis in vivo with promising preclinical outcomes [79]. PR-104 is currently in Phase II trial for relapsed or refractory acute myeloid leukemia [71].
Although several hypoxia-activated prodrugs have progressed into clinical trials, their poor extravascular space penetration, short blood half-life, instability and poor balance between the reduction/oxidation equilibrium, have limited their accumulation in the target tissues [53]. A solution to overcome these limitations may underlie the development of novel carriers to deliver these drugs. Therefore, improving their cellular uptake, specificity, tissue penetration, pharmacokinetics and dynamics [80,81]. From a clinical point of view, the standardization of treatment, dynamic interplay between hypoxia and other molecular determinants, as well as the poor definition of patient subgroups for hypoxia-targeted treatment, has caused limited positive clinical outcomes and warrants further evaluation [16,82]. In support of these findings, recent studies have utilized nanocarriers to improve HAPs poor in vivo distribution, and to offer a combinatory treatment with photosensitizers. TPZ is well known for its moderate to severe hypoxia activation, but other HAPs seem to have a narrow window of action. The use of photodynamic therapy (PDT) could create a more selective hypoxia environment through photosensitization, with increased synergism with HAPs. Shi et al. and his co-workers presented an interesting piece of work where they validated that photosensitizers can indeed produce a strong enough hypoxia environment that can potentiate the action of HAPs [25]. In this paper, double layered silica-shell nanoparticles were developed, which were capable of co-delivering a photosensitizer (UC/PS) and TPZ. The drug-loaded nanoparticles (TPZ-UC/PS) showed excellent biocompatibility and sufficient oxygen depletion by the photosensitizer, resulting in a significant reduction in cell viability. Furthermore, they showed a remarkable reduction in tumor volume after near-infrared irradiation (NIR). These combinatorial synergistic systems could offer a potential treatment for patients with large deep-seated tumors. In the same line of work, a mesoporous silica-based theranostic platform was also developed for synergistic PDT and hypoxia cancer therapy [24]. CD44-targeted system was comprised of a layer-by-layer silica nanoparticles that were used as a drug reservoir for TPZ, with an assembled supramolecular photosensitizer (TPPS 4 ), coordinated with the paramagnetic agent gadolinium-III for tumor-targeted diagnosis and treatment. Interestingly, PDT-induced apoptosis was confirmed by enhanced ROS production and the system showed a remarkable synergistic effect under 21% oxygen conditions, where the most significant reduction in tumor volumes was observed in mice co-treated with PDT and TPZ nanoparticles. These two studies confirmed that PDT could be used as a successful approach to activate HAPs, and to induce a synergistic effect in vivo models.

Macrophages in the tumor microenvironment
The presence of dendritic cells, fibroblasts, endothelial cells, monocytes and macrophages in the tumor microenvironment has been repeatedly linked to tumor progression [83]. The high infiltration of the immune cells in the tumor microenvironment, essentially in the hypoxic cores of tumors, has been correlated with a negative clinical outcome concerning patient's survival.
Monocytes are the primary moderate phagocytic cell type that contain primary lysosomes and have an overall pro-inflammatory effect. These cells circulate for about 1 to 3 days and afterward move to tissues where they mature into more active phagocytic cells, the macrophages (MACs) [84]. These cells are native residents of organs such as the liver, lung, spleen, lymphatic nodes. They have an important immunosurveillance role, mainly responsible for tissue repair, wound healing and defense against pathogens, including tumor cells [85]. However, a differential MAC programming in cancer progression has been elucidated, and much needs to be taken into account when describing the primary role of these cells in the tumor microenvironments [86]. MACs are particularly abundant and present at all stages of tumor progression. Their role in the tumor environment has reached a concept of 'friend or foe', recognizing that a well-functioned immune system should destroy tumor cells [87]. Nonetheless, many cancer cells escape the tight immune surveillance and redirect immune cells to their advantage. This escape is due to the development of deficiencies in antigen processing and presentation in several MAC pathways, and the consequent production of immune suppressive cytokines [88]. Tumor cells are known to secrete different cytokines, which control the balance between pro-inflammatory and anti-inflammatory components that in turn are responsible for this dual shift in MAC genomics and proteomics. Chemotactic cytokines, such as CXCL1, CXCL8, CXCL12, CXCL13, CCL5, are used to attract circulating monocytes to the tumor microenvironment [89][90][91]. These residing monocytes can then differentiate into MACs and initiate an inflammatory imbalance that causes a vicious cycle in which MACs initially control tumor development, but once established, are educated to become protumoral [23,92]. Highly active against orthotopic models of human lung cancer; The drug was also observed to decrease overall total protein synthesis, with a more pronounced effect in hypoxic prostate cancer cells. [47,48] 3.2. Differential polarization and the relevance of the M2 protumoral effect Two different polarization status have been reported amongst the MACs population [93]. The M1 phenotype (classically activated form) is responsible for host defense. When activated by Interferon gamma (IFN-γ) and/or lipopolysaccharides (LPS), M1 MACs produce large amounts of pro-inflammatory cytokines, such as oxygen species (e.g. nitric oxide), high levels of MHC (Major Histocompatibility Complex) molecules, interleukin 12 (IL-12) and low levels of IL-10. These characteristics make them potent killers of pathogens and tumor cells. On the  [27,53] other hand, the M2 subtype (alternatively activated form) responds to stimulus from IL-4/IL-13, IL-10, Toll-like receptors (TLR), glucocorticoids, allowing the development of MACs involved in tissue remodeling, angiogenesis, immune-regulation (M2b or M2c) and tumor development (M2a). These cells produce high amounts of IL-10, tumor necrosis factor (TNFα) and arginase-1. They over-express scavenger receptors, mannose receptors and exhibit anti-inflammatory activity [94,95]. The definition of M1 and M2 is not a 'black and white' concept. M2 type MACs are not simply defined by their location, but also by the stimuli they receive in the specific environments, in which they reside. Although these extreme forms of polarization are seductive, it seems that the multiple tumor-associated macrophages (TAMs) phenotype is dependent on the stimuli received in the complex network of signaling in the tumor microenvironment, as well as the tumor itself [92]. TAMs have been defined to closely resemble the M2 phenotype, and to constitute up to 80% of the tumor mass [93]. They are highly accumulated in poorly vascularized, hypoxic and necrotic areas, contributing to a poor prognosis and exerting a protumoral effect in many types of solid tumors [96,97].
Over forty years ago, a study by Fidler in 1974 [98], demonstrated that intravenous injection of specifically activated MACs (M1 phenotype) inhibited lung tumor metastasis. On the other hand, Gorelik et al. [99] showed that injecting M2 MACs intravenously in mice increased the development of lung cancer nodules. In the last two decades, studies have shown that co-culturing tumor cells in vitro with M2 MACs, even in the absence of a direct contact, increased tumor angiogenesis, progression and invasiveness [100][101][102][103]. These results highlight the different roles that polarized MACs play in cancer progression.

M2 MACs as promising therapeutic targets for cancer
Several studies have shown that MACs play a fundamental role in tumor development, and, therefore, remain a respectable therapeutic target in cancer therapy [104]. M2 MACs overexpress different types of receptors, which makes them excellent targets for drug delivery. A summary of the different targeted delivery systems to these MACs is shown in Table 3.
The principal mechanisms of MAC-based anticancer therapy rely on 1) inhibiting monocyte maturation in resident tumors; 2) depleting M2 type MACs; and 3) shifting polarization to an M1 phenotype with the pro-inflammatory response.
Therapeutic agents have been developed to target TAMs. For instance, trabectedin is an anticancer drug, licensed to treat advanced soft tissue sarcoma and recurrent ovarian cancer. It works by inhibiting macrophage colony-stimulating factor (M-CSF)-driven differentiation of monocytes into MACs, and blocking the production of CCL2, IL-6, and VEGF, causing a subsequent inhibition of tumor progression [114][115][116][117]. Prednisolone is a second class of drug (glucocorticoid) that inhibits monocyte differentiation. Systemic administration of prednisolone phosphate (PLP) encapsulated in long-circulating liposomes (LCL-PLP) significantly reduced M2 MAC levels in the tumor tissue, decreased the production of chemoattractants involved in the infiltration of monocytes into tumor microenvironment, and lowered the production of angiogenic factors by M2 MACs [118][119][120]. Metalloproteinase-9 (MMP-9) is a protein that is expressed by M2 MACs, which in turn triggers the release of VEGF, a factor important in the angiogenesis of tumors. Inhibiting tumor metalloproteinase activity, and diminishing the association of VEGF with its tyrosine kinase receptors on proliferating endothelial cells, may be an effective strategy to inhibit tumor growth. Clodronate, a bisphosphonate drug that blocks the activity of MMP-9 in TAMs. A liposome-based carriers containing clodronate (LIP-CLOD) selectively targeted MMP-9 following phagocytosis and intracellular drug release. MACs phagocytosed the LIP-CLO, which were subsequently degraded by the lysosomal phospholipases releasing the clodronate into the cell and inducing apoptosis, and reduced blood vessels formation [121].
The third mechanism in MAC-based anticancer therapy relies on shifting M2 polarization to M1 MACs [122]. TAMs express primarily M2-like phenotype, but can polarize to a classical M1-like phenotype, due to their high plasticity. For M1 polarization to occur two specific pathways have to be targeted, nuclear factor-κB (NF-κB) and the signaling transducer and activator of transcription (STAT) pathway. Toll-like receptors (TLR) are also expressed on MACs and is considered essential for the activation of NF-κB pathway, causing a reverse in MAC phenotype [88]. TLR agonists and inducers of NF-κB pathway have been exploited, causing a significant decrease in tumor growth [123]. Murphy et al. demonstrated that Azitromycin-treated J447 mouse MAC cells, produced lower levels of pro-inflammatory cytokines and higher levels of anti-inflammatory cytokines [124]. This clearly demonstrated the capacity of this antibiotic to shift the M1 polarization to M2. This study contributed to a better understanding of MAC function and polarizability in early inflammation. However, further studies to evaluate the clinical application, are still needed.
Gene therapy has also been explored to re-educate TAMs at tumor tissues. Adenoviral transduced MACs with IL-12 were used to treat orthotopic 178-2 BMA mouse prostate cancer model. The systemic administration of over-expressing IL-12 M1 MACs significantly reduced the growth of primary tumor and its metastasis in mice [97,125]. Lately, nano-carriers containing retinoic acid succeed in reverting M2 phenotype into M1. Flow cytometry and fluorescent microscopy analysis demonstrated that these nanocarriers were efficiently taken up by MACs, following systemic administration and remained engulfed in TAMs for 7 days [126]. Another study by Lobenberg group reported the capacity of doxorubicin-loaded nanoparticles to activate MACs (shifting naïve MACs to a M1 phenotype) after phagocytosis, which led to a significant cell death in cancer cells [127].

The link between hypoxia and TAMs: The promising role of targeted drug delivery systems
Evidence has connected tumor aggressiveness and poor patient survival to the hypoxic regions of tumors. TAMs infiltrate the hypoxic perinecrotic areas of the tumors, due to the increased number of cellular debris in those regions. Therefore, promote immune suppression, angiogenesis, lymphogenesis and matrix remodeling [9,10]. Such correlation has been observed in patients diagnosed with breast, prostate, ovarian and cervical cancer [128,129]. TAMs have been able to modulate the transcription factor HIF-1, several survival pathways, such as phosphatidylinositol 3-kinase (PI3K) or mitogen-activated protein kinase (MAPK) pathway, as well as regulate tumor angiogenesis and metastasis [19,20]. Recent reports have suggested that HIF-1 is overexpressed by TAMs, thus completing the vicious cycle that promotes cancer survival, progression and resistance to therapy [15] (Fig. 2). HIF-1 transcription factor not only modulates the expression of cancer-related genes in TAMs, but also provokes a metabolic shift in these immune cells, thriving the tumor development under nutrient and oxygen deprivation [130].
Individual and combinatorial systems exploiting TAMs and tumor hypoxia have been proposed. Hypoxia-responsive polymeric micelles showed a superior therapeutic efficacy in mouse cancer models, following systemic administration. This effective anti-tumor activity in vivo, was due to the high selectivity and fast intracellular release of doxorubicin from the responsive nanoparticles, under low oxygen levels. Quantitative analysis showed a 4-fold increase in the accumulation of responsive hypoxia micelles in tumor tissues, compared to normal tissues, which resulted in a slower tumor growth [131]. Another approach to target tumor hypoxia was demonstrated by Wang et al. and his team by developing micellar-based nanoparticles for the delivery of HIF-1α siRNA (EZN-2968). These nanoparticles showed specific gene knockdown both in vitro hypoxic mimicking cultures and in vivo hypoxic tumor models for prostate cancer. HIF gene silencing inhibited cell migration and angiogenesis (reduction of VEGF levels) and increased the The nano-complex reversed TAM phenotype by decreasing M2-specific genes: Arg-1, Ym1, [111] (continued on next page) sensitivity of cancer cells to doxorubicin [132]. Recently, Quan et al. reported the development of galactose-based thermosensitive nanogels as a theranostic system for hypoxic hepatocellular carcinoma [133].
The system was used to deliver iodoazomycin arabinofuranoside (IAZA), a radiosensitizer and a hypoxic imaging agent, to hypoxic hepatocellular carcinoma cells. IAZA-loaded nanogels showed a higher sensitization enhancement ratio, compared to IAZA alone. These results indicated that using nanogel particles could be a potentially useful approach to enhance the therapeutic efficacy of hypoxia radiosensitizers in tumor tissues. Lately, trigger responsive drug delivery nanocarriers have been proposed to target the acidic tumor microenvironment that usually associated with hypoxia [11]. For example, Poon et al. developed pHresponsive sheddable nanoparticles, consisting of multiple layers, of which neutral layers shed in response to the acidic environment, exposing the charged nanoparticle surface to be easily taken up by tumor cells, while sparing healthy tissues [134]. In another study, Meng et al. successfully developed pH-responsive nanovalves [135]. The porous mesoporous silica nanospheres were loaded with the drug and the pores were plugged with β-cyclodextrin. The interaction between the β-cyclodextrin and the stalk was dependent on the pH level, eventually facilitating diffusion of the drug from nanopores in the hypoxic tissues. The latter strategy was further improved by Dong et al. who successfully developed pH-sensitive polymeric-coated porous silica nanoparticles, by using polyethylene imine (PEI) and PEI-PEG copolymers. Both polymers improved the biodistribution of silica particles in vivo, where a superior tumor accumulation was observed with sterically stabilized nanoparticles [136]. More interestingly, photodynamic therapy (PDT) has been exploited to facilitate hypoxia generation, triggering drug release from hypoxia-responsive nanocarriers [137]. In this N/A N/A CHO-K1 cell line and C567BL/6 male mice The peptide did not home to other organs and co-localization of the peptide with macrophages was observed; The peptide bound strongly to CHO-K1 cells overexpressing IL-4Rα, compared to control. CRKRLDRNC bound both human and murine cells, presenting cross-reactivity of species. [112] Anti-IL4 aptamer N/A N/A 4 T1 and MSC2 cell lines BALB/C mice FACS analysis showed that the aptamer preferentially bound its ligand in vivo, with enhanced targeting of TAMs in tumor bearing mice; the aptamer recognized and bound both murine and human MACs; It may also trigger biological activity, possibly through a pro-apoptotic activity on TAMs, as tumor growth was inhibited in mice treated with the aptamer and a dramatic decrease in secondary metastasis was also observed.
[113] work, a ROS-generating and hypoxia-sensitive 2-nitroimidazole-grafted conjugated polymer (CP-NI) was synthesized, and doxorubicin (Dox) was successfully encapsulated within these polymeric nanoparticles. These multifunctional nanoparticles generate ROS upon near-infrared irradiation (NIR), leading to a rapid oxygen consumption within the tumor tissues, resulting in an efficient drug release from the nanocarriers. Such system showed promising in vitro and in vivo results, where mice treated with CP-NI-Dox in combination with NIR irradiation showed a complete tumor growth inhibition, with high levels of apoptosis as confirmed by TUNEL assay. Combining PDT with HAPs has been also considered as a promising approach to enhance the therapeutic efficacy of anti-cancer therapy, as discussed in Section 2.2.
The concept of using MACs as 'trojan horses' has proven to be an interesting approach in cancer therapy [138,139]. Choi et al. developed MACs loaded with gold nano-shell particles, which efficiently targeted the hypoxic regions of T47D breast cancer spheroids in vitro, and successfully slowed down the spheroid growth in combination with NIR [140]. A second study by Hirschberg et al. showed the infiltration of gold-silica nanoshells-loaded MACs into glioma tumor spheroids [141]. Similarly, small gold nanorod-laden MACs (sAuNRs-laden MACs) with a smaller diameter of~7 nm were able to infiltrate inaccessible tumor hypoxic regions, enhancing photothermal tumor ablation. The data showed that sAuNRs had higher cellular uptake (52.5%) and lower cytotoxicity in MACs, compared to other gold-based particles. RAW264.7 macrophages incubated with AuNRs and irradiated with 808 nm NIR laser, showed a significant cell death (higher ratio of cell apoptosis (32.6% ± 0.87) compared to the control group. Interestingly, these results were translated in vivo, where 95% tumor growth inhibition was observed in HepG2 tumor-bearing nude mice, with no visible recurrence [142]. Another study by Choi et al. reported the development of drug-loaded MACs to treat cancer [143]. This novel therapeutic system was composed of mouse peritoneal MACs loaded with liposomal doxorubicin (macrophage-LP-Dox). Liposomal nanoparticles protected the MACs from the encapsulated doxorubicin in vitro, while delaying the tumor growth in animal models. These approaches highlight the potential use of MAC-based drug delivery systems to treat cancer. An Interesting work by Jiang et al. using bone marrow derived MACs (BMDM) loaded with doxorubicin polymeric nanoparticles, demonstrated the deep penetration of doxorubicin into the tumor tissues, which was significantly higher in infrared irradiated tumors (IR) [144]. Such approach could be particularly suited to treating IR-induced recurrent tumors.
Model simulations have suggested that, compared with conventional chemotherapy, MAC-based therapy preferentially targets tumor cells and diminishes the hypoxic cells in the tumor core [145]. In agreement with these simulations, Griffiths et al. developed a successful approach using a gene-dependent enzyme prodrug therapy (GDEPT), where MACs were transfected with adenoviral vectors that express cytochrome P450 [146]. The transfected MACs highly infiltrated tumor spheroids and enhanced the therapeutic activity of cyclophosphamide, a prodrug that induces tumor cell killing by the bystander effect. Treatment of spheroids with cyclophosphamide, after co-culture with transfected MACs, showed a significant reduction in spheroids volume. This approach became more promising after engineering adenoviral construct with a hypoxia-responsive promoter, where cytochrome P450 was only expressed in the hypoxic region of the tumor, highlighting the specificity of the treatment. Subsequent exposure of the tumor spheroids to cyclophosphamide and MACs transfected with a hypoxiaresponsive construct showed a specific inhibition of cell proliferation under hypoxic conditions. Furthermore, Lewis' lab previously used MACs to efficiently deliver oncolytic viruses to hypoxic prostate cancer tissues in vivo. The engineered MACs were co-transduced with a hypoxia-regulated E1A/B construct and an E1A-dependent oncolytic adenovirus, whose proliferation was restricted to prostate tumor cells using prostate-specific promoter elements from the TARP, PSA, and PMSA genes [138]. A single systemic injection of the oncolytic virus-loaded MACs resulted in a marked inhibition of tumor growth and reduction of pulmonary metastases. Furthermore, such approach enhanced the therapeutic efficacy of chemo-and radiotherapy in prostate cancer models [147]. This data demonstrated the high potential of multifunctional nanotherapeutics which target tumor hypoxia and MACs.
A second interesting approach was explored by Huang et al. and coworkers [148,149], who exploited the use of monocyte-mediated delivery of polymeric bubbles to potentiate re-oxygenation of tumor hypoxic areas, whilst enhancing the therapeutic efficacy of chemotherapy. In the first study, Huang et al. presented an innovative strategy for overcoming the limited activity of photodynamic therapy (PDT) in hypoxic tumor tissues using bone marrow-derived monocytes as cellular vehicles for co-transport of oxygen and a light activated photosensitizer, chlorin e6 (Ce6). Superparamagnetic iron oxide nanoparticles (SPION)/Ce6/oxygen-loaded polymer bubbles were internalized into tumortropic monocytes. In this study, results showed that intratumoral administration of therapeutic monocytes exhibited a superior activity in inhibiting tumor growth in Tramp-C1 tumor-bearing mice upon the treatment with magnetic field and light laser. Histological examinations of the tumor sections confirmed the successful cellular transport of the therapeutic payloads to tumor hypoxia. This study demonstrated that oxygen/therapeutic co-delivery via tumortropic monocytes enhances tumor penetration, relieves tumor hypoxia after external hyperthermia trigger, and sensitizes cells to PDT [148]. In the second study [149], the same principle was applied for the co-delivery of polymeric bubbles and doxorubicin-loaded polymeric vesicles. Here, focused ultrasound was applied to trigger the release of drug-loaded vesicles from the monocytes within the hypoxic cores of tumors following cell infiltration. Once again, in vivo and ex vivo fluorescence imaging showed an appreciable accumulation of the doxorubicin-loaded monocytes at the tumor site. With this, a high payload of drug was delivered deeper within the tumor mass, resulting in a pronounced cytotoxic effect. These studies clearly highlight the capability of MACs to deliver a wide range of therapeutics to deep tumor tissues, where the therapeutics efficacy was enhanced in combination with external triggers, such as laser irradiation, hyperthermia or ultrasound, which facilitated drug release from the nanocarriers.
Recently, manganese dioxide nanoparticles (MnO 2 NPs) have been used to target TAMs and enhance tumor oxygenation [150]. In this study, Song et al. developed novel nanoparticles that target TAMs in the hypoxic regions of the tumor. The high reactivity of MnO 2 NPs toward hydrogen peroxide (H 2 O 2 ) allowed the simultaneous production of O 2 and regulation of pH in the hypoxic core. These nanoparticles were further modified with mannose and hyaluronic acid (HA) to promote M2 targeting and TAMs repolarisation into M1 phenotype, respectively. The results showed higher association of the targeted particles to M2 mannose receptor in vitro, compared to untargeted particles (5-fold increase). Interestingly, targeted manganese dioxide particles showed a steady tumor uptake in vivo after intravenous injection and relatively low liver uptake. Flow cytometry assays, suggested the positively induced polarization of M2 to M1, with consequent production of high levels of H 2 O 2 by these cells. The system improved the overall oxygenation and regulation of local hypoxic pH, in which the tumor showed 50.3% less tissue hypoxia, 49.3% decrease in the expression of HIF-1α, and 31.8% decrease in the expression of VEGF, 4 days after treatment. This study provided the opportunity to enhance chemotherapy response in 4 T1 xenograft model by using combining Man-HA-MnO 2 NPs and doxorubicin, where a significant reduction in xenograft tumor growth was observed.
In summary, we have demonstrated here that targeting hypoxia and TAMs is, indeed, a promising therapeutic targets for cancer therapy. Their mechanisms of action are summarized in Fig. 3. Nanomedicine is a smart strategy to reassess old drugs and improve their pharmacokinetics and tumor targeting. Therefore, we believe that special considerations should be taken into account to design a combinatory treatment, capable of exerting synergistic effects in the tumor microenvironment.
Several studies reviewed here took the advantages of oxygen generation/consumption to manipulate the hypoxia environment and to sensitize cancer cells to different treatment strategies.

Clinical implications and future perspectives
Hypoxia has become one of the most attractive targets in cancer [11]. Many small molecule inhibitors, HAPs and hypoxia-responsive nanosystems have shown a great promise in cancer therapy. However, hypoxia-targeted therapy has shown controversial results in mouse models and humans. These high attrition rates of failure are attributed not only to the high complexity and redundancy of the tumor microenvironment, but also due to lack of veracity and fidelity in existing preclinical models and patient subsets. Therefore, substantial efforts have been made to develop more reliable in vitro methods that can then be successfully translated to in vivo set-up and in patients [30]. For instance, the potential of these novel therapies must be validated at early stages, by fully establishing cell culture models that include comprehensive immunohistochemistry (IHC) staining, western blot analysis, and mRNA expression of direct and indirect pathways linked to hypoxia [27]. Limited molecular biological testing has been carried out due to the lack of well-established robust biomarkers associated to hypoxia, and the nonexistence of in vitro models that can actively mimic the complex acidic, immune infiltrated tumor microenvironment, with heterogeneous expression of hypoxia-related genes. Nevertheless, biomimetic 'organ/ tumor-on-a-chip' tools have presented themselves as promising new in vitro tools, that could bring together the advantages of tumor spheroids [151], with the aid of microfluidic systems that can offer information on tumor heterogeneity, interstitial flow, cell binding and nanoparticle/drug accumulation [152]. Despite these challenges, efforts have been made to create more realistic pre-clinical and clinical models that are essential to fully understand, validate and successfully predict the outcome of these targeted therapies. More importantly, many open discussions on the matter have highlighted the importance of patient pre-selection for hypoxia and tumor microenvironment targeted therapies [30]. Previous clinical set-ups have failed to show relevant positive outcomes, due to inadequate patients standardization. Many patient subsets did not show a relevant expression of HIF-1. Also, a good percentage of the patients showed mixtures of well-oxygenated and hypoxic tumors, or poorly established immune microenvironment that conferred reduced benefit of the applied treatment. Clearly, there is still much to be done to perform adequate evaluation of these treatments, but recent efforts to establish xenograft models that constitutively represent HIF-1 expression encourage future studies and reliable clinical translation [143,153]. Furthermore, the new interest to combine HAPs with drugs that are known to lowering tumor oxygen levels and potentiating tumor hypoxia, such as some VEGF inhibitors and photosensitizers, could enhance the therapeutic efficacy of the combinatory treatments, and overcome the oxygen tumor heterogeneity.
Similar to hypoxia, tumors have been known for their heterogeneity in blood supply, which results occasionally in inevitable disappointment once translated to clinical trials. This tumor heterogeneity makes it difficult to predict drug efficacy in tumor models, or to extrapolate the clinical efficacy of drugs in humans based on in vivo mice models. In support of this, Danhier et al. has recently highlighted the need to reassess the concept of nanomedicine in a clinical aspect, mainly regarding its benefits when exploiting the enhanced permeation and retention (EPR) effect [154]. Danhier argues that although the EPR is probably one of the most cited and important concepts in nanomedicine, it is a heterogeneous process, highly variable between tumors and within the same tumor mass, and not the main attribute for the high success of novel nano-therapies. Nevertheless, as aforementioned, we highlight the importance of using more relevant models that can translate the results obtained in murine models to cancer patients. For example, patient-derived tumor explant (PDX) models, can alternatively provide an accurate model of morphology, complexity, and heterogeneity of human tumors [155,156]. More promisingly, extensive work has been published in the last two decades to overcome the EPR hurdles, using mild hyperthermia and sonoporation to improve drug accumulation and penetration within the tumor mass [157][158][159][160]. We believe that the future prospects and success of tumor microenvironment targeted therapies rely on establishing strong validated models that truly represents tumors in cancer patients. Additionally, the use of combinatorial synergistic chemotherapeutics, that nanomedicine could offer, may provide a means to overcome tumor complexity, providing a more favorable realistic outcome.
Our review suggests that drug combination and time of therapy are the key factors that should be taken into consideration when designing combinatorial therapies for hypoxia, especially when translated to clinical trials. Previous pre-clinical studies have provided evidence that angiogenesis inhibition through TAM targeting can indeed inhibit tumor growth [161,162]. Most importantly, TAM depletion which potentiates tumor hypoxia could be overcome if correctly combined with angiogenesis and/or hypoxia-targeted therapies [161,162]. Although targeting TAMs and hypoxia individually has shown a great promise, we believe that developing combinatorial therapies exploiting both targets may ensure strong synergistic effects, and render solid tumors more susceptible to conventional cancer therapies. Such strategies can provide a sophisticated approach to target multiple signaling pathways in tumorigenesis, and overcome off-target effects from positive loop feedbacks generated between TAMs, hypoxia and the tumor microenvironment. These studies provide evidence that simultaneously targeting TAMs and hypoxia is an interesting novel approach to exploit in cancer therapy.
Despite the apparent controversy and disappointing results discussed above, researchers must not forget the high number of newly approved drug delivery systems and ongoing clinical trials [152]. Therefore, we support aforementioned opinions that targeting the tumor microenvironment must be revised, and translation to clinic must be supported by adequate pre-clinical evaluation alongside standardized patients selection. Researchers need to focus not only on showing sophisticated nanomedicine with enhanced efficacy, but they should discover new techniques that can identify new tumor microenvironment biomarkers that could offer a personalized medicine for cancer patients. This can lead to the development of new targeted systems, and also aid in identifying tumors that will respond better to certain drug regimen. Efforts are in place to take nanomedicine closer to clinical translation by producing more reliable and positive treatment outcomes.

Conclusions
Significant progress has been made on deciphering the role of hypoxia and MACs infiltration in tumor malignancy. In the present work, we give an in-depth review on the active role of the tumor microenvironment in cancer survival and the development of drug resistance. TAMs have been considered as potential therapeutic targets for cancer therapy, since they are strongly linked to hypoxia and cancer progression. MACs have been envisioned as 'trojan horses', given their enhanced extravasation in hypoxic tumor cores. These features offer a great opportunity to design highly selective and sophisticated drug delivery systems that simultaneously target TAMs and tumor hypoxia. We believe that adopting this new approach is anticipated to overcome drug resistance and enhance the efficacy of chemotherapy in advanced solid and metastatic tumors. Finally, with validated tumor models and standardized patient pre-selection, and targeting hypoxia and TAMs, combinatory nanomedicine can become a more reliable therapeutic tool with more favorable clinical outcomes in the future.