The crucial role of transforming growth factor β (TGFβ) in tumor progression, metastasis and treatment has been well recognized and has become the topic of extensive research. Among the effects, TGFβ can regulate cancer cell proliferation, contribute to epithelial-to-mesenchymal transition (EMT), suppress the function of immune cells compromising immune response, contribute to the conversion of fibroblasts to myofibroblasts and cause overproduction of extracellular matrix (ECM) in the tumor. While it has been known for over two decades that anti-cancer drugs cannot penetrate deep into collagen-rich tumors (e.g., pancreatic cancers) and, more significantly, that depletion of collagen fibers can improve drug delivery, only recently TGFβ has become a target to reduce tumor fibrosis and thus, increase intratumoral drug concentration and treatment efficacy. Preclinical data of this new strategy are promising and it has already reached clinical trials. In this review, we first present a brief description of TGFβ synthesis and activation along with its signaling pathways. Following, we discuss the effects of TGFβ on tumor progression, its pathway alterations in cancer as well as its effects on EMT, immune cells function, fibroblasts behavior and ECM remodeling. Finally, based on the above, we review the barriers to the effective delivery of drugs caused by TGFβ and how regulation of TGFβ signaling can be employed to optimize delivery of therapeutic agents and overall survival (1–3).

The TGFβ superfamily encompasses around 40 secreted cytokines, including TGFβ, bone morphogenetic proteins (BMPs), activins, nodal, lefty, myostatin, anti-Müllerian hormone (AMH) and growth differentiation factors (GDFs). These cytokines regulate a plethora of biological functions such as cell proliferation and apoptosis, embryonic patterning, stem cell maintenance, cell differentiation, migration and immune surveillance. Importantly, the effects of these factors are characterized as cell-type specific as well as context dependent (1–3). The TGFβ isoforms, with most common being TGFβ1, 2 and 3, are initially synthesized as 75 kDa inactive homodimers, known as pro-TGFβ, which consist of TGFβ associated with latency-associated proteins (LAPs) at the N-terminal part of the pro-peptide. This is part of the TGFβ large latent complex (LLC), comprised of the LAPs and the latency TGFβ-binding proteins (LTBPs) (4–7), and is covalently associated to the ECM via the N-terminal region of LTBPs (8,9) (Fig. 1). While TGFβ is part of the LLC complex, it remains in an inactive form since the high affinity association of LAPs with TGFβ prevents the interaction with its receptors (10). During TGFβ activation, LAPs undergo conformational changes induced by thrombospondin-1 (TSP-1) (11,12) followed by cleavage mediated by furin convertase, plasmin or matrix metalloproteinases MMP-2/9 resulting in the release of the mature 24 kDa TGFβ dimer (13–15). The active ligand is then able to bind and activate TGFβ receptors (TGFβRs) to propagate downstream intracellular signaling events. Therefore, the processing of pro-TGFβ into the active TGFβ ligand is a critical regulatory step which determines its bioavailability.

The TGFβ and TGFβ-like cytokines mediate downstream intracellular signaling via the Smad family of proteins, which consists of eight human structurally related members (16–20) (Fig. 1). Smads can be functionally classified into three groups: the receptor activated Smads (R-Smads), which include Smad1, 2, 3, 5, 8; the common mediator Smad (Co-Smad), Smad4; and the inhibitory Smads (I-Smads), Smad6 and 7 (17,21). Three types of TGFβRs are responsible for initiating signaling; TGFβRI, II and III. There are seven TGFβRI, five TGFβRII and two TGFβRIII known so far. TGFβRIs include activin receptor-like kinases 1–7 (ALK1–7), TGFβRIIs include the TGFβRII, bone morphogenetic protein receptor II (BMPRII), activin receptor II (ACTRII), ACTRIIB, anti-Müllerian hormone receptor II (AMHRII), while beta-gycan and endoglin belong to the TGFβRIIIs (22) and mostly function as co-receptors to enhance activin signaling (23). In most tissues, TGFβ ligands function through heteromeric complex formation between two TGFβRI and two TGFβRII molecules. While both receptors possess Ser/Thr kinase activity, TGFβRIIs function as the ‘activator’ and TGFβRIs as the ‘signal propagating’ component (24). The TGFβRII-ALK5 complex transduces the signal from all three TGFβ isoforms in multiple cell types, whereas association of TGFβRII with ALK1 is involved in endothelial cells and with ALK2 in cardiovascular tissues (25). ALK5 activates Smad2 and 3 via the canonical TGFβ signaling pathway whereas ALK2, 3 and 6 can activate Smad1, 5 and 8, which are transducers of the BMP signaling pathway (26,27). The TGFβ signaling pathways can be classified in two major categories; the canonical or Smad-dependent and the non-canonical or Smad-independent pathways.

It is well established that the multipotent actions of TGFβ are highly context dependent. The complexity of these functions is increased due to the fact that TGFβ exerts distinct effects depending on the tissue type as well as the genetic and epigenetic background of cells (85). It is clearly evident that TGFβ plays dual roles during carcinogenesis. In early stages TGFβ promotes growth inhibition and apoptosis of normal epithelial and lymphoid cells as well as pre-malignant tumors, whereas during late stages TGFβ acquires pro-oncogenic and pro-metastatic roles, which are associated with a progressive increase in the locally secreted TGFβ levels (86–88). Therefore, one of the hallmarks of cancer is that the vast majority of cases exhibits insensitivity to TGFβ-mediated growth inhibition.

Under physiological conditions, the sustained local release of basal TGFβ levels is sufficient to maintain normal tissue homeostasis. However, under conditions of tissue injury, the local TGFβ secretion from stromal cells and blood platelets is rapidly increased to facilitate wound repair as well as to prevent uncontrolled regenerative cell proliferation and inflammation (130,131). A similar situation is commonly observed in pre-malignant tumors where TGFβ is secreted in the microenvironment initially to control proliferation and cancer progression, but it is ultimately utilized by cancer cells to promote their malignant properties. Local TGFβ release produces a tumor microenvironment which is conducive to tumor growth, invasion and metastasis (132). Secretion of TGFβ can be derived from epithelial cancer cells thus regulating their own properties within the tumor mass in an autocrine or paracrine fashion (125). Moreover, infiltrating stromal cells, including fibroblasts, leukocytes, macrophages, bone-marrow derived endothelial, mesenchymal and myeloid precursor cells, is another major source of this cytokine (133). Finally, TGFβ can be stored in the ECM of the bone and can be activated during development of osteolytic metastatic lesions (134). In the following paragraphs, we summarize the effects of TGFβ on the main and better characterized components of the tumor microenvironment and particularly on fibroblasts, immune cells and the ECM.

The desmoplastic reaction of solid tumors hinders all three transport steps of the systemic delivery of drugs, namely vascular, transvascular and interstitial transport (156,163). As mentioned above, increased levels of collagen in the ECM, result in intratumoral blood vessel compression and hypo-perfusion. Hypo-perfusion, in turn, reduces the concentration of the drug that can reach the tumor site. Apart from compromised drug delivery, hypo-perfusion also decreases the supply of oxygen rendering the tumor hypoxic, which in turn reduces the efficacy of radiation therapy. Additionally, desmoplasia reduces the hydraulic conductivity of the tumor interstitial space, i.e., the ease with which the interstitial fluid percolates through the interstitial space of a tissue. High hydraulic conductivity allows fluid to rapidly flow in the interstitial space and be drained by peripheral lymphatic vessels. The accumulation of collagen and other ECM proteins in tumors decrease the available spaces for interstitial fluid flow and because the fluid cannot freely move, the interstitial fluid pressure (IFP) increases. Interstitial hypertension is a hallmark of tumor pathophysiology. IFP reaches and even exceeds micro-vascular fluid pressure, which eliminates pressure gradients across the tumor vessel wall and thus, the transvascular transport of drugs (164). Therefore, the only mechanism of transport is through diffusion (i.e., due to a concentration difference), which is inversely proportional to the size of the therapeutic agent. Chemotherapeutic agents, with a size <1 nm, are able to diffuse fast and exit the tumor vasculature. Nanoparticles, however, with sizes >60 nm cannot effectively extravasate into the tumor interstitial space (165).

Furthermore, the dense interstitial matrix of desmoplastic tumors hinders the homogeneous distribution of large nanoparticles. As with transvascular transport, nanoparticles with a size >60 nm often cannot penetrate deep into the tumor because their size is comparable to the size of the pores of the interstitial collagen network and they often get trapped (166). Therefore, even if large nanoparticles extravassate from the leaky vessels of the tumor, they will not be able to effectively diffuse into the tissue but they will concentrate in the perivascular regions, causing only local effects. Apart from these steric interactions between the interstitial matrix and nanoparticles, the increased levels of collagen and hyaluronan give rise to electrostatic interactions. Indeed, hyaluronan has a highly negative charge, while collagen fibers carry a slight positive charge. Nanoparticles of a non-neutral surface charge density can be attracted electrostatically and bind to these proteins, which further inhibits their uniform delivery inside the tumor (167).

Pharmacological inhibition of TGFβ has been used in preclinical and clinical studies as a therapeutic strategy to either hinder tumor progression directly or modify the tumor micro-environment in order to improve perfusion and drug delivery and thus, increase indirectly the efficacy of the treatment. There is a large number of TGFβ inhibitory drugs employed in these studies (137). Particularly, targeting with TGFβ agents (e.g., 1D11, AP12009, SD-208) as well as non-specific targeting with other TGFβ inhibitory drugs (e.g., tranilast) have shown to reduce tumor progression and metastasis in vivo, mainly owing to augmentation of the immune response and inhibition of EMT (132,168–171). However, there are also studies that relate inhibition of TGFβ with promotion of tumor progression owing to an increase in inflammatory cell infiltration (172). Particularly, it has been shown that inflammatory infiltrates mediate the pro-tumorigenic functions of fibroblasts that lack TGFβ signalling. Clinical trials for the use of TGFβ inhibitory drugs have been in progress (ClinicalTrials.gov identifiers: NCT00368082, NCT01582269 and NCT00844064), but their results are not conclusive yet, presumably owing to differences in the degree of desmoplasia among tumor types or even among tumors of the same type, but also owing to the various effects of TGFβ on tumor biology.

Targeting of TGFβ to reduce desmoplasia has the ability to alleviate physical forces in tumors, decompress tumor blood vessels and improve perfusion (Fig. 2B) (160). Restoration of tumor perfusion, however, can increase nutrients supply to the tumor, and thus, increase its growth rate. Also, the decompressed vessels could allow more metastatic cells to leave the primary tumor. Indeed, in some cases, inhibition of TGFβ has been shown to facilitate tumor progression and metastases in mouse tumor models (173,174), whereas other studies, not related to TGFβ, have shown a correlation between improved perfusion and increased metastases (175,176). Therefore, based on this rationale, judicious doses of TGFβ inhibitory drugs should be used to alleviate physical forces, decompress blood vessels and improve perfusion when these agents are combined with cytotoxic treatments, such as chemo-, nano-, immuno- and radiotherapy. In these combined treatments the role of the anti-TGFβ drug is to enhance the delivery of the cytotoxic agent and thus, optimize its efficacy. This therapeutic strategy is known as stress alleviation treatment (156,163,165).

Detailed in vivo studies have shown that re-purposing the anti-hypertensive, angiotensin receptor blocker (ARB) drug losartan reduced expression of TGFβ1 and decreased stromal collagen and hyaluronan production, in doses that did not affect blood pressure. Reduction of collagen and hyaluronan, in turn, reduced stress levels in the tumor decompressing intratumoral blood vessels and improving perfusion. Furthermore, reduction of the ECM components improved the interstitial fluid flow and thus, reduced levels of IFP. Improved perfusion and reduced IFP enhanced the delivery and efficacy of chemotherapy in orthotopic breast and pancreatic murine tumor models (160). Also, in another study combined treatment of mice bearing tumors with losartan and nanomedicine (Doxil) increased the distribution of the drug and the overall survival of the mice (177). Furthermore, retrospective analyses of clinical data have shown increased survival in patients with lung or renal cancers treated with ARBs (178,179). Similar retrospective analysis has shown that patients with pancreatic ductal adenocarcinomas (PDACs) receiving ARBs survived ~6 months longer than those who did not (180). These preclinical and clinical data have led to a phase II clinical trial with losartan and FOLFIRINOX in PDAC patients (ClinicalTrials.gov identifier NCT01821729). Apart from the use of ARBs, the TGFβ neutralizing antibody 1D11 improved the distribution and efficacy of therapeutics in breast carcinomas by reducing the tumor stroma (181). Additionally, re-purposing the drug pirfenidone, a TGFβ inhibitor clinically approved for the treatment of idiopathic pulmonary fibrosis, was shown to suppress desmoplasia in mice bearing pancreatic tumors and improve the efficacy of chemotherapy (182). Apart from chemotherapy, radiation therapy has been also improved after treatment with TGFβ inhibitors. Efficacy of radiotherapy depends on the oxygenation of the tissue, which is regulated by tumor perfusion (183,184).

Owing to the pleiotropic effects of TGFβ on tumor microenvironment and progression, targeting TGFβ signaling to directly treat tumor growth remains controversial. Recent studies have suggested an alternative therapeutic strategy, which involves the use of anti-TGFβ agents in a stress alleviation treatment. The scope of this strategy is to hinder but not completely inhibit the activation of TGFβ ultimately aiming to reduce tumor desmoplasia and particularly the levels of collagen. As described in this review, reduced collagen levels can lead to improved delivery of both chemo- and nano-therapeutics by alleviating mechanical forces and decompressing intratumoral blood vessels. Thus, blocking of TGFβ can improve indirectly the efficacy of conventional treatments. It is promising that many anti-TGFβ agents exist that are already clinically approved for other diseases (e.g., ARBs for hypertension). Re-purposing of these drugs can lead to more effective anti-cancer therapies. Therefore, we need to identify safe and well-tolerated pharmaceutical agents that may complement the treatment regimen of cancer patients. Anti-TGFβ agents are not the only drugs that have the ability to modify the tumor microenvironment. In principle, any clinically approved agent that has the ability to reduce collagen levels could be employed as an alternative strategy. Also, collagen is not the only target for the stress alleviation treatment. Reduction of stromal cells or hyaluronan has also the potential to enhance drug delivery through the same mechanism (157).