TGF-β blockade improves the distribution and efficacy of therapeutics in breast carcinoma by normalizing the tumor stroma

Abstract

Although the role of TGF-β in tumor progression has been studied extensively, its impact on drug delivery in tumors remains far from understood. In this study, we examined the effect of TGF-β blockade on the delivery and efficacy of conventional therapeutics and nanotherapeutics in orthotopic mammary carcinoma mouse models. We used both genetic (overexpression of sTβRII, a soluble TGF-β type II receptor) and pharmacologic (1D11, a TGF-β neutralizing antibody) approaches to block TGF-β signaling. In two orthotopic mammary carcinoma models (human MDA-MB-231 and murine 4T1 cell lines), TGF-β blockade significantly decreased tumor growth and metastasis. TGF-β blockade also increased the recruitment and incorporation of perivascular cells into tumor blood vessels and increased the fraction of perfused vessels. Moreover, TGF-β blockade normalized the tumor interstitial matrix by decreasing collagen I content. As a result of this vessel and interstitial matrix normalization, TGF-β blockade improved the intratumoral penetration of both a low-molecular-weight conventional chemotherapeutic drug and a nanotherapeutic agent, leading to better control of tumor growth.

Breast cancer is the second leading cause of cancer death in women, with most fatalities resulting from a failure to control metastatic disease with systemically administered therapies. In addition to the induction of cellular resistance mechanisms (decreased apoptosis, increased drug efflux, etc.), impaired intratumoral drug delivery is an important physiological factor contributing toward chemoresistance (1, 2). TGF-β is an important regulator of normal mammary gland development and function, as well as of the progression of mammary carcinomas (3–8). Although the role of TGF-β in tumor progression and metastasis has been studied extensively, little is known about its impact on drug delivery.

Transport of a therapeutic agent from the circulation to cancer cells is a three-step process. Systemically administered drugs must (i) travel to different regions within a tumor via the vascular network; (ii) cross the vessel wall; and finally (iii) diffuse through the interstitial space to reach the tumor cells, with each step being hindered by the presence of an abnormal vasculature and/or matrix (1, 2, 9, 10). Tumor blood vessels are structurally and functionally abnormal, characterized by increased permeability and heterogeneous perfusion. Poor vascular perfusion decreases drug delivery and, as a result, impairs the efficacy of blood-borne antitumor agents (1, 11). In addition, the dense collagen-rich interstitial matrix further hinders drug transport to tumor cells—a feature especially relevant to larger therapeutics, such as nanoparticles (1–100 nm) (1, 10, 12, 13). The dense collagen matrix also contributes to solid stress, which compresses tumor vessels (14). Hence, depleting collagen will reduce stress and open up compressed vessels. TGF-β is a negative regulator of pericyte recruitment during blood vessel stabilization (15, 16) and also plays an important role in stimulating and regulating the synthesis of collagen—a major component of the interstitial matrix (17). As such, it is plausible that TGF-β inhibition—through enhancing pericyte recruitment and normalizing vessels and through decreasing matrix collagen content—could improve perfusion and drug delivery into tumors.

In a mouse thyroid cancer model, blocking TGF-β enhanced pericyte recruitment to vessels and decreased tumor interstitial fluid pressure (IFP)—a known cause for decreased convective transport of high-molecular-weight agents (18). However, the effects on delivery of small or large therapeutics were not measured in this study. In a pancreatic tumor model, a small molecule TGF-β receptor I inhibitor enhanced uptake of the large nanotherapeutic Doxil (90 nm) and polymeric micelles (30 and 70 nm) (19), but did not affect distribution of low-molecular-weight chemotherapeutic agent (20). Furthermore, the mechanism for the enhanced penetration of nanotherapeutics and whether long-term TGF-β blockade given as an adjunct to chemotherapy would have a sustained effect on drug delivery were not investigated. In this context, although VEGF blockade can enhance the intratumoral delivery of 12-nm nanoparticles, in contrast to TGF-β inhibition, it does not affect the penetration of larger nanoparticles (60 and 125 nm) (21). Because TGF-β sits at the crossroads of several important signaling pathways we hypothesized that TGF-β blockade might have dual effects—both inhibiting tumor progression and, particularly relevant to our investigation, enhancing drug delivery to tumor cells by normalizing the tumor vasculature and interstitial matrix. We reasoned that breast cancer represents an especially appropriate setting for investigating the effects of TGF-β blockade because the breast microenvironment is characterized by an abundant collagenous matrix, and interference with its production might have particularly beneficial effects on drug delivery.

Here, we first show that TGF-β blockade inhibits tumor growth and metastasis in two mammary carcinoma models. We then show that TGF-β blockade improves drug delivery via two distinct mechanisms. First, it normalizes the structure of tumor vasculature and increases vessel perfusion. As a result, transvascular transport of both low- and high-molecular-weight therapeutic agents is improved. Second, it decreases the density of the interstitial matrix by decreasing collagen I content and, consequently, increases tumor tissue penetration of the 100-nm liposomal doxorubicin (Doxil). Because of this enhanced drug delivery, TGF-β blockade significantly improves the efficacy of chemotherapy. Because TGF-β inhibitors are currently being tested in clinical trials, our findings may be rapidly translated to the clinic to improve the treatment of breast cancer.

Results

Blocking TGF-β Signaling Inhibits the Growth and Metastasis of Orthotopic Mammary Carcinomas.

To reflect tumor-to-tumor variability, we used two different breast carcinoma cell lines: the human MDA-MB-231 (which expresses low levels of TGF-β, 76.3 ± 26.8 pg/mg) and the murine 4T1 (which expresses high levels of TGF-β, 316.9 ± 65.0 pg/mg). We used two methods to block tumor and host TGF-β signaling in each cell line. First, we cloned a soluble TGF-β receptor II (sTβRII) construct and stably transfected both cell lines. The transfected tumor cells constitutively secreted large quantities of sTβRII protein, which competed with TGF-β receptor II for sequestering TGF-β and successfully blocked TGF-β1 and -β3 induced Smad2 phosphorylation (Fig. S1). Importantly, sTβRII did not affect tumor cell growth in vitro (Fig. S1). Second, to model treatment in the clinical setting, we used a pan–TGF-β neutralizing antibody, 1D11, to treat established orthotopic mammary carcinomas.

sTβRII transfection significantly inhibited the growth of both MDA-MB-231 and 4T1 tumors (Fig. 1 and Fig. S2). In MDA-MB-231 tumor-bearing mice, no visceral metastases were observed in either parental or sTβRII-transfected tumors. However, we found that there was a significant decrease in lymph node metastasis in sTβRII-transfected MDA-MB-231 tumors compared with controls [as determined by bioluminescence imaging (BLI)] (Fig. 1B). In 4T1 tumors, sTβRII transfection completely abolished macroscopically detectable metastatic lung nodules (Fig. 1D). In the antibody-treated group, we started 1D11 treatment 7 d after tumor implantation. At this time, mice typically bore small, but well established tumors (∼2.5 mm in diameter). 1D11 treatment significantly inhibited tumor growth and lung metastasis of the high TGF-β expressing 4T1 tumors, but did not affect the growth or lymph node metastasis of low TGF-β expressing MDA-MB-231 tumors (Fig. 1E).

Fig. 1.

Blocking TGF-β signaling inhibits the growth and metastasis of orthotopic mammary carcinoma. (A) Primary tumor growth curves in mice bearing parental (n = 9), mock (n = 8), and sTβRII-transfected (n = 6) MDA-MB-231 cells. *P < 0.0001. (B) Representative BLI of inguinal (In-LN) and axillary (A-LN) lymph nodes (n = 4). Color scale: min, 27; max, 30308. (C and D) Primary tumor growth curves (C; n = 8) and lung metastasis quantification (D; n = 12) in mice bearing parental, mock, and sTβRII-transfected 4T1 cells. (C) *P < 0.0001. (D) *P < 0.005. (E Left and Center) Primary tumor growth curves of MDA-MB-231 (Left) and 4T1 (Center) tumors treated with control IgG (13C4) or anti-TGF-β antibody, 1D11 (n = 8). *P < 0.0001. (Right) Quantification of 4T1 tumor lung metastasis (n = 8). *P < 0.005.

We also investigated the effects of TGF-β blockade on tumor cell proliferation (Ki67 staining), apoptosis (TUNEL staining), and angiogenesis [tumor microvessel density (MVD) and VEGF expression]. In the groups where TGF-β blockade inhibited tumor progression (all except 1D11 treatment of MDA-MB-231 tumors), we found that TGF-β blockade significantly decreased tumor cell proliferation and increased apoptosis. We also observed that TGF-β–blocked tumors expressed lower concentrations of VEGF with an associated reduction in tumor MVD (Table 1). In the MDA-MB-231 group that did not respond to 1D11 treatment, consistent with the tumor growth results, tumor cell proliferation, apoptosis, VEGF levels, and MVD did not change (Table 1).

Table 1.

Effect of TGF-β blockade on tumor cell proliferation, apoptosis, MVD, and expression of VEGF

	sTβRII overexpression				1D11 treatment
Staining	231	231-sTβRII	4T1	4T1-sTβRII	231-control	231–1D11	4T1-control	4T1-1D11
Ki67*	26.1 ± 5.1	5.8 ± 3.5†	35.5 ± 4.2	9.5 ± 3.8†	25.7 ± 4.6	23.5 ± 6.3	34.7 ± 4.6	10.2 ± 4.4‡
TUNEL§	28.7 ± 7.8	42.7 ± 3.8‡	23.5 ± 5.9	38.8 ± 4.5	27.9 ± 6.8	28.8 ± 8.5	25.6 ± 5.2	37.5 ± 4.8¶
MVD||	28 ± 5.8	16 ± 6¶	65.3 ± 6.3	35.8 ± 7.1‡	26 ± 6.1	32 ± 3.2	60.4 ± 6	38.7 ± 6.9‡
VEGF**	497 ± 80	75 ± 50†	71 ± 6	10 ± 0.2¶	482 ± 50	427 ± 30	74 ± 5	15 ± 0.3¶

All analysis was performed in tumors collected at the end of the experiment.

*Data are shown as number of Ki67⁺ cells per 0.041-mm² area.

^†P < 0.001.

^‡P < 0.01.

^§Data are shown as number of TUNEL⁺ cells per 0.329-mm² area.

^¶P < 0.05.

^||Data are shown as number of CD31⁺ structures per 1.355 mm² area.

**Data shown are human VEGF (in MDA-MB-231 tumors) and mouse VEGF (in 4T1 tumors) levels (pg/mg) as determined by ELISA. Their level (pg/mL) was normalized to the amount of total protein (μg/mL).

Blocking TGF-β Signaling Normalizes the Tumor Vasculature and Improves Vessel Perfusion.

Pericytes supporting the endothelial layer of blood vessels play a key role in vessel maturation and function (22–24). In TGF-β–blocked 4T1 and MDA-MB-231 tumors, we found significantly greater colocalization of NG2 (a pericyte marker) and CD31 (an endothelial cell marker) immunostaining. Quantification of pericyte coverage confirmed that blocking TGF-β with sTβRII or 1D11 significantly increased the fractional coverage of tumor blood vessels by NG2-positive perivascular cells (Fig. 2A and Table 2). To determine whether this structural normalization translated to functional normalization, we examined the fraction of perfused vessels by injecting FITC–lectin i.v. to identify perfused tumor vessels and by staining for CD31 to detect the total number/density of blood vessels. In both sTβRII-transfected and 1D11 antibody-treated 4T1 and MDA-MB-231 tumors, TGF-β blockade significantly increased the percentage of perfused blood vessels (Fig. 2B and Table 2).

Fig. 2.

Blocking TGF-β signaling increases tumor blood vessel pericyte coverage and perfusion. (A) Representative images and quantification of immunofluorescence double staining for endothelial cells (CD31) and pericytes (NG2) in frozen sections of 4T1 and 4T1–sTβRII tumors. Green, CD31⁺ staining; red, NG2⁺ staining; yellow, colocalization of red and green. (B) Representative images and quantification of perfused blood vessels (FITC–lectin) and immunofluorescence staining for endothelial cells (CD31) in frozen sections of 4T1 and 4T1–sTβRII tumors. Green, FITC–lectin-labeled perfused vessels; red, CD31⁺ staining in nonperfused vessels; yellow, CD31⁺ staining of perfused vessels. P < 0.001.

Table 2.

Effect of TGF-β blockade on pericyte coverage, blood vessel perfusion, collagen content, and drug penetration

	sTβRII overexpression				1D11 treatment
Measurement	231	231–sTβRII	4T1	4T1–sTβRII	231–control	231–1D11	4T1–control	4T1–1D11
NG2+/CD31,* %	36 ± 4.4	60 ± 5.2†	37 ± 6.5	60 ± 2.2†	37 ± 4.0	51 ± 2.9†	39 ± 4.7	59 ± 9.2†
Perfused vessels,‡ %	55.8 ± 6.2	70.8 ± 4.9†	49.5 ± 11.4	75.4 ± 7.1†	49.4 ± 5.1	68.8 ± 4.4†	46.7 ± 9.8	70.2 ± 7.6†
Doxorubicin penetration,§ μm	45.1 ± 10.8	64 ± 13	9.6 ± 3.7	50.7 ± 18¶	15.2 ± 4.5	64.3 ± 3.7||	11.6 ± 4.0	45.0 ± 11.5¶
Doxorubicin uptake,** %	9.6 ± 3.8	44.6 ± 12.4||	16.8 ± 3.1	46 ± 11||	15.2 ± 4.5	66.6 ± 4.9¶	18 ± 2	59.0 ± 12.8¶
Collagen I content††	9.87 ± 3.7	2.06 ± 0.8¶	24.0 ± 2.5	8.9 ± 0.8¶	7.68 ± 3.7	2.77 ± 1.0||	21.9 ± 8.4	7.2 ± 2.7¶
Doxil penetration,§ μm	N/A‡‡	N/A‡‡	17.8 ± 2.3	46.3 ± 9.4||	N/A§§	N/A§§	17.2 ± 3.3	38.8 ± 5.9||
Doxil uptake,** %	N/A‡‡	N/A‡‡	12.9 ± 8.6	54.9 ± 9.4†	18.8 ± 1.0	59.4 ± 11.1||	4.2 ± 1.1	16.8 ± 6.2†

*Data are shown as percent of NG2⁺/CD31⁺ staining quantified by an automated routine in ImageJ.

^†P < 0.01.

^‡Data are shown as percent of colocalization of FITC–lectin and CD31⁺ immunostaining.

^§An automated routine in ImageJ was used to quantify the average intensity of extravasated fluorescent doxorubicin as a function of distance from the blood vessel wall.

^¶P < 0.001.

^||P < 0.05.

**An automated routine in ImageJ was used to quantify the percentage of tumor area positive for extravasated fluorescent doxorubicin.

^††Data are shown as percent of collagen I⁺ staining per 0.329-mm² surface area quantified by an automated routine in ImageJ.

^‡‡Because transfection of sTβRII almost completely abolished tumor growth, Doxil treatment experiment was not performed in this group.

^§§Because the pattern of diffused intratumoral penetration of Doxil in 231–1D11 group, measure of penetration depth and comparison with the control group was not carried out.

Inhibition of TGF-β Signaling Improves Intratumoral Distribution of the Low-Molecular-Weight Drug Doxorubicin.

Normalization of vessels and improvements in vascular perfusion can enhance the intratumoral distribution of low- and high-molecular-weight drugs in tumors (24–26). We first studied the effect of TGF-β blockade on the intratumoral distribution of a low-molecular-weight drug, doxorubicin. In the 4T1 group, we examined doxorubicin distribution when tumors reached 200 mm³; in the MDA-MB-231 sTβRII-transfected group, we examined doxorubicin distribution on day 28 and in the 1D11 treatment group after the last treatment. In parental- and control-treated tumors, red fluorescence (doxorubicin) was observed closely surrounding blood vessels, whereas in TGF-β–blocked tumors, fluorescence was more broadly distributed (Fig. 3). Quantitative analysis confirmed that TGF-β blockade significantly increased the distribution of intratumoral doxorubicin (shown as the percentage of tumor area positive for doxorubicin) (Fig. 3C and Table 2).

Fig. 3.

Blocking TGF-β signaling improves intratumoral distribution of doxorubicin in orthotopic mammary carcinoma models. (A and B) Representative images of doxorubicin intratumoral distribution in 4T1 (A) and 4T1–sTβRII (B) tumors. Green, FITC–lectin-labeled perfused vessels; red, fluorescent doxorubicin; blue, DAPI. (C) Quantification of the fraction of tumor area positive for doxorubicin (n = 12 sections, with 3 sections per tumor). *P < 0.001.

Blocking TGF-β Signaling Decreases Collagen I Content and Improves Transvascular and Interstitial Transport of the Nanotherapeutic Doxil.

Poor perfusion and diminished transvascular drug transport are not the only barriers to efficient drug delivery in tumors. The dense interstitial matrix further inhibits the ability of large therapeutics to reach the cancer cells. We and others have shown repeatedly that a decrease in collagen content improves interstitial transport of large-molecular-weight drugs (>10 nm; e.g., nanospheres, liposomal agents—Doxil) (27–31). TGF-β plays an important role in the stimulation and regulation of collagen synthesis. We found that blocking TGF-β signaling by transfection of sTβRII or the 1D11 antibody treatment significantly decreased collagen I content in orthotopic 4T1 and MDA-MB-231 tumor models (Fig. 4A and Table 2). We next examined the effect of TGF-β blockade on the interstitial penetration of Doxil (liposomal doxorubicin). In control tumors, doxorubicin (red fluorescence, released from the liposome) was observed inside and closely surrounding blood vessels; in TGF-β–blocked tumors, the fluorescent signal was seen penetrating further away from the vessel. Quantitative analysis confirmed that, as a result of enhanced vessel perfusion, the amount of intratumoral Doxil (shown as the percentage of tumor area positive for doxorubicin) was significantly higher compared with the control group (Table 2). Moreover, because of its effect on decreasing the density of the interstitial matrix, TGF-β blockade significantly increased the penetration distance of Doxil away from blood vessels and into the tumor tissues (shown as micrometers from the nearest blood vessel) (Fig. 4B and Table 2).

Fig. 4.

Blocking TGF-β signaling decreases collagen I content and improves Doxil tissue penetration. (A) Representative images and quantification of collagen I immunofluorescent staining in 4T1 and 4T1–sTβRII tumors. Red, collagen I staining; blue, DAPI (×20). (B) Representative images and quantification of Doxil intratumoral distribution in 4T1 and 4T1–sTβRII tumors. Green, FITC-lectin labeled perfused vessels; red, fluorescent doxorubicin; blue, DAPI (n = 12 sections, with 3 sections per tumor). *P < 0.001.

Blocking TGF-β Signaling Enhances Efficacy of Doxil in an Orthotopic Mammary Carcinoma Model.

Finally, we examined the growth inhibitory effects of Doxil, with and without TGF-β blockade. Among the genetically modified tumors, the transfection of sTβRII almost completely abolished MDA-MB-231 tumor growth; therefore, Doxil treatment was only administered to mice bearing 4T1 and 4T1–sTβRII tumors. Both cell lines were studied for effects of combined 1D11 and Doxil therapy. As shown in Fig. 5, TGF-β blockade (sTβRII or 1D11), combined with Doxil treatment was significantly more effective in inhibiting tumor growth. Immunohistochemical staining confirmed that in the Doxil and TGF-β blockade combined group, there were significantly more apoptotic tumor cells compared with the Doxil monotherapy group (Table 3).

Fig. 5.

Blocking TGF-β signaling enhances Doxil efficacy in orthotopic mammary carcinoma models. (A and B) Primary tumor growth of 4T1 and 4T1–sTβRII tumors, with or without Doxil treatment (A) and 4T1 (B) and MDA-MB-231 (C) tumors treated with saline (control), Doxil alone (9 mg/kg, weekly), 1D11 alone (5 mg/kg, three times a week), or combined Doxil and 1D11 (n = 8 in all groups).

Table 3.

Effects of combined TGF-β blockade and Doxil on tumor cell apoptosis and proliferation

Parameter	4T1	4T1–sTβRII	4T1	4T1–1D11	231–control	231–1D11
Apoptosis*
Saline	14.5 ± 2.1	38.8 ± 4.5†	11.6 ± 2.6	37.5 ± 4.8†	27.9 ± 6.8	28.8 ± 8.5
Doxil	46.3 ± 7.4‡	52.6 ± 8.3§	36.4 ± 2.6‡	72.4 ± 3.2§	46.3 ± 7.4†	72.4 ± 3.2§
Proliferation¶
Saline	65.5 ± 4.2	28.5 ± 3.8†	72.1 ± 7.4	23.2 ± 4.4‡	75.7 ± 4.6	63.5 ± 6.3
Doxil	17.0 ± 3.8‡	12.0 ± 4.5	22.3 ± 1.6‡	14.2 ± 3.6	34.7 ± 4.6	27.3 ± 5.1

*Data are shown as number of TUNEL⁺ cells per 0.329-mm² area.

^†P < 0.01 compared with parental tumor-bearing mice–saline group.

^‡P < 0.001 compared with parental tumor-bearing mice–saline group.

^§P < 0.05 compared with parental tumor-bearing mice–saline group.

^¶Data are shown as number of Ki67⁺ cells per 0.041-mm² area.

Discussion

Several strategies to block TGF-β are being tested in cancer and pulmonary fibrosis and are at various stages of investigation, from preclinical research to phase III clinical trials (32). These strategies fall into two classes: direct inhibition of TGF-β (using TGF-β–neutralizing antibodies or TGF-β receptor inhibitors) and interference with downstream signaling (e.g., peroxisome proliferator-activated receptor-γ agonists) (33–42). In our study, we used two strategies to block TGF-β signaling: (i) in a genetic model, TGF-β signaling was inhibited by stable transfection of tumor cells with a soluble TGF-β receptor (sTβRII), which functions as a “TGF-β trap,” competing with TGF-β1 and -β3 for binding to TGF-β receptor ΙΙ; and (ii) in a pharmacologic model, TGF-β signaling was inhibited by a neutralizing antibody that blocks all three isoforms of TGF-β. The 4T1 tumors are known to express elevated levels of TGF-β and are sensitive to TGF-β blockade (4, 8). We observed that both sTβRII transfection and 1D11 antibody administration significantly inhibited 4T1 tumor growth and metastasis. To represent breast cancer variability, we also used MDA-MB-231 tumors, which express low levels of TGF-β and depend on other growth factors for their progression. Administration of 1D11 did not affect its tumor growth; however, overexpression of sTβRII almost completely blocked the growth of MDA-MB-231 tumors. The differential effect between the two models of TGF-β blockade in the MDA-MB-231 group may be due to (i) limited intratumoral delivery of 1D11 antibody—such that the local concentration of sTβRII produced by cancer cells within the tumor is greater than that of antibody administered via i.p. injection; and (ii) a difference between an early vs. late treatment effect—overexpression of sTβRII blocks TGF-β signaling from implantation on, whereas 1D11 treatment starts when the tumors are established and larger. Additional studies in which the timing of onset of ID11 treatment is varied would elucidate this possibility further.

Consistent with our hypotheses, our data reveal that TGF-β blockade is a powerful tool to enhance chemotherapy and nanoparticle delivery into tumors, operating within the stroma at two distinct levels: the tumor vasculature and interstitial matrix. These two stromal effects stand separately from the direct antitumor effect observed after TGF-β inhibition. First, TGF-β blockade normalizes tumor vasculature—manifested by a reduction in vessel density, improved vessel maturity, and increased perfusion of vessels—and improves chemotherapy delivery into tumors (26). TGF-β blocks pericyte recruitment during blood vessel stabilization (15, 16). Our findings with regard to vessel density and structure are consistent with previous data from TGF-β blockade studies of thyroid carcinoma (18). In addition, our study, as well as the work of others, shows that TGF-β blockade decreases angiogenic gene expression, including VEGF and IL-8 (43, 44). Anti-VEGF therapies are known to “normalize” tumor blood vessels (22, 25), suggesting that TGF-β blockade may directly and indirectly (via inhibition of VEGF) exert its effect on pericyte recruitment. Indeed, in MDA-MB-231 tumors, 1D11 treatment did not change VEGF expression; however, it still led to increased pericyte coverage, indicating that, independently from inhibiting VEGF signaling, TGF-β blockade can directly induce vessel normalization. More importantly, we observed that TGF-β blockade leads to functional normalization—an increase in the fraction of perfused vessels. It is now well established that the IFP of most solid tumors is elevated and hinders transcapillary transport of high-molecular-weight agents in tumors (1). Vessel normalization lowers IFP and restores pressure gradients across vessel walls. The restored pressure gradient increases transvascular bulk fluid flow and contributes to increased penetration of drugs (22). It has been shown that inhibition of TGF-β signaling reduces tumor IFP (18). This finding together with our findings suggest that TGF-β blockade enhances the transvascular transport of both low- and high-molecular-weight drugs (doxorubicin and Doxil) by means of its effects on (i) inducing vessel normalization, (ii) increasing vessel perfusion, and (iii) decreasing tumor IFP.

Second, TGF-β blockade normalizes the interstitial matrix by reducing collagen density and, hence, improves transport of nanoparticles from the perivascular space into the deeper parts of a tumor. Tumor interstitial matrix—composed of collagens, proteoglycans, and other molecules—is a potent barrier to the intratumoral delivery and penetration of large biopharmaceuticals, such as monoclonal antibodies, therapeutic proteins, and nanoparticles (12, 28–30, 45, 46). Once extravasated, drugs must diffuse through the interstitial matrix to reach the cancer cells. Elevated levels of collagen have been shown to contribute to solid stress known to compresses tumor vessels and to reduce the delivery of large therapeutics (14, 27–30, 47, 48). To this end, our findings demonstrate that TGF-β blockade decreases collagen I content. The reduction in collagen levels combined with the increased vascular perfusion and extravasation are likely mechanisms for the improved intratumoral penetration of Doxil, a 100-nm agent.

In addition to normalization of the stromal compartment, TGF-β blockade inhibits cancer cell proliferation and increases apoptosis in vivo. In vitro studies have shown that doxorubicin penetrates poorly through high-density multicellular layers (31), and agents that induce apoptosis and reduce cell density improve drug penetration (49, 50). In 4T1 tumors, TGF-β blockade alone reduces cancer cell proliferation and enhances apoptosis; it further enhances doxorubicin-induced cell apoptosis, suggesting that TGF-β blockade reduces cell density and functions through dual tumor and stromal mechanisms to improve drug delivery. However, in the MDA-MB-231 group, treatment with the 1D11 antibody alone did not modify cancer cell apoptosis or proliferation but still significantly increased doxorubicin penetration, suggesting that in the MDA-MB-231 model, 1D11 likely functions through its stromal effects to improve drug delivery. These studies show that even in tumors that do not exhibit a growth delay after TGF-β blockade, it can still improve the delivery and outcome of chemotherapy. As such, we eagerly await the results of ongoing trials examining the effects of anti–TGF-β therapies in breast cancer and other human tumors.

Materials and Methods

The effects of TGF-β inhibition on tumor vascular morphology and perfusion, drug delivery, and tumor progression were studied in two orthotopic breast tumor models. For additional information regarding penetration of chemotherapeutic agents, treatment protocols, and statistical analysis, see SI Materials and Methods.