Background
Triple-negative breast cancer (TNBC) is defined by the absence of estrogen receptors, progesterone receptors and HER-2 amplification and represents an aggressive breast cancer subtype. Despite significant advancements in the treatment of other breast cancer subtypes, there is still no licensed targeted therapy available for the treatment of TNBC, and therefore little improvement in survival has been observed for this patient population over the last years [
1,
2]. This highlights the need for better understanding of TNBC and identification of mechanisms involved in disease progression and treatment response.
It is now well recognized that breast cancer progression can be influenced by a pro-tumorigenic microenvironment surrounding the malignant epithelial cells. This environment consists of a heterogeneous mixture of stromal cells, including a diversity of cancer associated fibroblasts (CAFs), a biological active network comprising the extracellular matrix (ECM), in addition to the interstitial fluid and its solutes [
3,
4]. New knowledge about the components of the microenvironment and how they interact with tumor cells can hopefully identify new biomarkers or potential targets in TNBC.
The aberrant stroma affects the physiological forces within the tumor. Indeed, the hydrostatic pressure in the tumor interstitium, known as interstitial fluid pressure (PIF), is considerably increased in the majority of solid tumors [
5], including human breast cancer [
6,
7], and this poses a major physiological barrier to transport of soluble factors within the tumor [
8].
Increased PIF has been shown to predict poor prognosis in some solid tumors [
9,
10], and can also hinder effective delivery of drugs into the tumor [
11‐
13]. Finding ways to lower tumor PIF may therefore increase efficiency of cancer therapy.
Fibroblasts can actively modify PIF and transcapillary fluid exchange (reviewed in [
8,
14,
15]) and the molecular mechanisms are outlined by collagen contraction assays [
16,
17] and heterospheroids [
18‐
20], as well as parallel in vivo experiments [
21‐
23]. Dysfunctional blood and lymph vessels will lead to fluid accumulation in the tumor interstitium, and swelling of hyaluronan and proteoglycans would in normal conditions hinder an increase in PIF [
8,
24]. Tension exerted by fibroblasts and collagen network can probably counteract this swelling, resulting in a persistent increased PIF [
14]. However, although fibroblast-mediated contraction has previous been shown to be dependent on β1-integrins [
21], fibroblast-mediated PIF influence is still not fully understood.
Integrin α11β1 is a collagen binding integrin expressed during differentiation of myofibroblasts [
25‐
27] and is involved in collagen organization [
17,
28] and tumor stiffness [
28]. As a collagen organizer and a link between contractile fibroblasts and the complex collagen network, integrin α11β1 could be a regulator of tumor PIF. Although a few studies indicate that it has a physiological role in the regulation of PIF in dermis [
29,
30], its influence on PIF in tumors remains to be demonstrated. A better understanding of the mechanisms that regulate pressure homeostasis within a tumor, can probably lead to a new insight into breast carcinogenesis, and we therefore investigated the effect of stromal integrin α11-deficiency on pressure homeostasis, ECM organization and tumor growth using two human TNBC xenograft models.
Methods
Cell lines
MDA-MB-231 (ATCC® HTB-26™) was provided by Professor James Lorens (University of Bergen, Bergen, Norway), and MDA-MB-468 (ATCC® HTB-132™) was obtained from the American Type Culture Collection (Manassas, VA., USA). The MDA-MB-231 cells were fingerprinted before use and matched with the cell line MDA-MB-231 (ATCC® HTB-26™) in the ATCC database. MDA-MB-231 was used at passage number five to nine, while the MDA-MB-468 cells were used at passage number two to five. These TNBC cell lines have high tumor take in SCID mice and slowly forming tumors, which may be more stromal dependent than more rapidly growing xenografts. Wild type (WT) and integrin α11-deficient (α11-KO) mouse embryonic fibroblasts (MEFs) were obtained from mouse embryos of embryonic day 14.5 as described previously [
31]. In order to obtain immortalized MEFs, primary MEF cultures were infected with recombinant retrovirus-transducing simian virus 40 (SV40) [
32]. All cell lines were grown in Nutrient Mixture F-12 Ham (Sigma-Aldrich, Steinheim, Germany) supplemented with 10% Foetal Bovine Serum, 100 units/ml penicillin, 100 μg/ml streptomycin, and 1–2% L-glutamine (all from Sigma-Aldrich). The cells were grown as single monolayers in a humidified incubator at 37 °C in 5% CO
2 and in all experiments used at log phase. All cell lines tested negative for mycoplasma contamination.
Xenograft models
The integrin α11-deficient heterozygous SCID mouse strain was generated as previously described [
28]. PCR-genotyping was performed on DNA extracted from ear biopsies [
32]. The animals were kept in individually ventilated cages, cared for regularly and efforts were made to age- and weight match the animals. All animal experiments were approved by the Norwegian Food Safety Authority (Permit Number 20168751) which is the competent body responsible for authorizing research projects in animals in Norway. This is in accordance with the EU directive 2010/63 article 36.
A total of 5 × 105 MDA-MB-231 or 1.5 × 105 MDA-MB-468 tumor cells in 0.15 ml PBS were injected into the fourth mammary fat pad (orthotopic), and for the MDA-MB-231 also subcutaneously on the mouse flank (ectopic). Tumor size was measured using a caliper. The tumor volume was calculated using the formula; tumor volume (mm3) = (π/6) × a2 × b, where a represents the shortest diameter and b represents the longest diameter of the tumor. All animals were anesthetized using Isofluran (Isoba®vet. 100%, Schering-Plough A/S, Farum, Denmark) and eventually sacrificed by cervical dislocation under deep anesthesia. For investigation of the primary tumor, all the MDA-MB-231 injected mice were sacrificed day 57 post injection. For the MDA-MB-468 injected mice, some of the tumors showed tendency to ulcerate the skin, and these mice were sacrificed immediately. To make the groups comparable, one mouse from the opposite group and with similar tumor load was sacrificed on the same day.
To evaluate metastatic spread to the lungs, they were processed and fixed as previously described [
33] (
n = 5 WT and 5 α11-KO and
n = 5 WT and 4 α11-KO for the MDA-MB-231 and MDA-MB-468 injected mice, respectively).
All measurements and analysis in this study were performed blinded to genotype.
Measurement of interstitial fluid pressure
The wick-in-needle technique was used to measure the tumor PIF [
34]. Briefly, a standard 23-gauge needle with a side hole filled with nylon floss and saline was connected to a PE-50 catheter, a pressure transducer and a computer for pressure registrations, using the software Powerlab chart (version 5, PowerLab/ssp. AD instruments, Dunedin, New Zealand). The needle was inserted into the central part of the tumor after calibration. After a period of stable pressure measurements, the fluid communication was tested by clamping the catheter which shall cause a transient rise and then return to pressure prior to clamping. Measurements were accepted if the pre- to post-clamping value was within ±1 mmHg.
PIF in heterospheroids was measured with the micropuncture technique described previously [
18]. Briefly, the spheroids were collected and transferred to 10-cm Lysine-coated cell culture dishes (Nunc, Thermo Fisher, Waltham, MA., USA) and left to attach for 2 h at 37 °C. PIF was measured using sharpened glass capillaries (tip diameter 3–5 μm) connected to a servo-controlled counter pressure system. The glass capillaries were filled with hypertonic saline (0.5 M) colored with Evans blue dye and inserted into the central parts of the spheroid with the help of a stereomicroscope (Wild M5, Heerbrugg, Switzerland). PIF in the cell culture medium directly outside the spheroid was defined as the zero reference pressure.
Electron microscopy of collagen fibrils in the tumor
Tumor samples were taken from the tumor periphery and were fixed and processed as previously described [
33]. A JEM-1230 Transmission Electron Microscope (TEM) (Jeol, Tokyo, Japan) was used to measure the diameter and organization of the collagen fibrils, and images from four to six different areas of the tissue were analyzed. Pictures were captured at × 100,000 magnification and analyzed using Image J 1.46 (National Institute of Health, Bethesda, MD., USA) to measure the fibril diameter. To investigate the organization of the collagen fibrils, pictures were captured at × 30,000 magnification and scored from one to four considering collagen fibril organization and alignment within the collagen fibers.
A JSM-7400F Scanning Electron microscope (Jeol) was used to study the tumor collagen fibril scaffold architecture. Five images from different areas of the tumor were captured from each tumor at × 10,000 magnification.
Immunostaining and Picrosirius-red staining
Histological analysis was performed on both paraffin embedded sections and cryosections. For paraffin embedded sections, 5 μm thick sections were deparaffinizated and rehydrated, followed by heat induced antigen retrieval at pH 6 (#S1699, Dako, Agilent, Santa Clara, CA., USA) for Ki67 (100 °C, 20 min) and αSMA (100 °C, 25 min), pH 9 (#2367, Dako) for Coll III (100 °C, 25 min) or pH 10 (#T6455, Sigma Aldrich) for PDGFRβ (110 °C, 5 min). After antigen retrieval, the sections were incubated with peroxidase block (#K006, Dako) and then primary antibody. Envision+ System-HRP (#K4006 or #K4010, Dako) was used as secondary antibody, in addition to rabbit anti-goat for collagen III (1:1000, #6164–01, Southern Biotech, Birmingham, AL., USA), and DAB was used as chromogen, except for αSMA staining, where a FITC-conjugated antibody was used. Analysis of immunohistochemistry was performed using Leica DN 2000 Led (Leica Microsystems, Wetzlar, Germany). The following primary antibodies were used on paraffin sections: rabbit anti-mouse PDGFRβ mAb (1:100, #3169, Cell Signaling Technology, Danvers, MA., USA), goat anti-mouse Type III Collagen pAb (1:100, #1330–08, Southern Biotech), anti-mouse αSMA mAb (F3777, dilution 1:200, Sigma Aldrich) and mouse anti-human Ki67 mAb (1:100, #M7240, Dako).
Cryosections with a thickness of 6 μm were fixed in ice-cold methanol (− 20 °C, 8 min) and rehydrated with PBS, followed by blocking with 10% goat serum. Afterwards, the following primary antibodies were supplied: rabbit anti-mouse integrin α11 pAb (1:200, custom-made, Innovagen AB, Lund, Sweden, [
31]), mouse anti-human cytokerain AE1/AE3 mAb (1:200, #M3515, Dako) and mouse anti αSMA mAb (1:200, #A5228, Sigma Aldrich). Goat anti-rabbit Alexa 594 (1:400, #111–585-144, Jackson ImmunoResearch, Ink., West Grove, PA., USA) and goat anti-mouse Alexa 488 (1:400, #315–545-045, Jackson ImmunoResearch) were used as secondary antibodies. Mounting was done with ProLong Gold Antifade Mountant with DAPI (#P36934, ThermoFisher). The staining results were evaluated under an Axioscope fluorescence microscope and micrographs were acquired using a digital AxioCam MRm camera (Zeiss, Oberkochen, Germany).
Picrosirius-red stain (Polysciences inc, Warrington, FL., USA) was used for a semi-quantitative measurement of collagen type I and III as previously described [
33].
Evaluation of the staining
For Picrosirius-red, collagen III, PDGFRβ and αSMA, a total of four to six pictures were captured from each tumor with × 100 magnification. Images were taken in the tumor periphery in order to avoid the necrotic central area. The software Image J 1.46 (National Institute of Health, Bethesda, MD., USA) was used to identify the amount of positive pixels.
For Ki67, the tumors were examined using light microscopy with an eye-piece grid at × 630 magnification. A total of 500 tumor cells from the tumor periphery were evaluated, and distinct nuclear staining regardless of intensity was registered as positive. Areas with necrosis, bleeding or inflammation were avoided.
Microdialysis
Microdialysis was performed as previously described [
35] on the MDA-MB-231 mammary fat pad tumors. Briefly, after anesthesia with Ketalar (Pfizer Inc., NY., USA) and Dormitor (Orin Pharma AS, Espoo, Finland), one microdialysis probe was placed in the MDA-MB-231 mammary fat pad tumor (CMA12 Elite Microdialysis probe, ref.nr 8,010,434) and one in the jugular vein (CMA12 Elite Metal free, ref.nr 80,111,204). The probes were connected to a PE-50 catheter, perfused by a pump (CMA100 Microinjection pump, ref.nr 8,210,040) at a rate of 1 μl/min and left to stabilize for 30 min. After intravenous injection of 0.15 ml 0.65 MBq
3H-5FU (Nycomed Amersham, Buckinghamshire, UK), dialysate was sampled and pooled every 10 min for a total of 90 min. Scintillation counting solution (Optiphase Hisafe 3, PerkinElmer, Inc., Waltham, MA., USA) was added, and the radioactivity measured using a liquid scintillation analyzer (Tri-Carb 2900TR, PerkinElmer, Inc.). The probes and pump were delivered by CMA Microdialysis AB, Kista, Sweden.
The area under the curve (AUC) for the plasma and tumor was calculated with Graph Pad Prism 7 (GraphPad Software Inc., La Jolla, CA., USA) as the total radioactivity collected, i.e. as the product of radioactivity (counts per minute) and time. Finally, transport of 3H-5FU was expressed as AUC tumor divided by AUC plasma.
After each experiment, the probes were tested in saline with a known amount of 3H-5FU, and experiments with probes that differed more than 15% in permeability were excluded.
Heterospheroids
Heterospheroids containing a mixture of SV40-immortalized MEFs and MDA-MB-231 cells were prepared using the hanging drop method as described previously [
19]. Briefly, sub-confluent cells were trypsinized and suspended in culture medium to a concentration of 1 × 10
6/ml. The MEFs (WT or integrin α11-KO MEFs) and MDA-MB-231 cell suspensions were then mixed at a ratio of 4:1 to make WT MEFs + MDA-MB-231 and α11-KO MEFs + MDA-MB-231 spheroids. Approximately 40 drops (25 μl/ drop, 2.5 × 10
4 cells/drop) were dispensed onto a lid of a cell culture dish. The lid was then inverted and placed over a cell culture dish containing medium for humidity, and cultured in a humidified incubator at 37 °C in 5% CO
2 for 5 days.
Statistical analysis
Sigmaplot 13.0 (Systat Software Inc., Chicago, IL., USA) and Graph Pad Prism 7 (GraphPad Software) were used for statistical analysis. Either the unpaired two-tailed t-test or the Mann- Whitney U test, was used to analyze statistical differences between the two groups. Results were accepted as statistically different when p < 0.05. Data are given as mean ± SD, and number of measurements (n) refers to number of tumors or heterospheroids unless otherwise specified.
Discussion
Integrins are essential adhesion receptors necessary for intercellular communication, attachment of cells to the ECM and modulation of the tumor microenvironment [
36‐
39]. In this study, we have demonstrated that stromal integrin α11-deficiency markedly decreased PIF in vivo using two orthotopic human triple-negative breast cancer cell lines. A perturbed collagen structure was seen, with fewer aligned and thinner collagen fibrils. Furthermore, integrin α11-deficiency impeded orthotopic breast tumor growth in the MDA-MB-231 model, and the same trend was also found in the MDA-MB-468 orthotopic model. By investigating the isolated effect of integrin α11-positive fibroblasts on MDA-MB-231 tumor cells in vitro, we provide here evidence that PIF regulation is, at least partly, mediated by integrin α11-positive fibroblasts.
Integrin α11β1 has arisen as a possible marker of a pro-tumorigenic subset of CAFs in the tumor microenvironment [
40,
41]. It has been found to be overexpressed in the stroma of lung cancer and head and neck cancer [
40,
42]. Further, it stimulates lung cancer cell growth in vitro [
20], and lung and prostate cancer growth in vivo [
28,
33]. However, its role in tumor growth and progression is still not clear, especially in breast tumors where we recently reported that it did not affect the growth of the murine TNBC cell line 4 T1 in vivo [
33].
In the present study, we found that stromal integrin α11-deficiency led to reduced tumor PIF in both orthotopic xenograft models. This demonstrates for the first time that integrin α11β1 has a role in maintaining an elevated PIF in solid tumors. A dense ECM, contractile fibroblasts, leaky blood vessels and dysfunctional lymphatic drainage are possible causes of increased PIF in tumors [
8]. PIF can be actively modulated through interactions between contractile fibroblasts and ECM molecules [
8,
23], where fibroblasts have been proposed to normally exert a tension on the collagen network through collagen-binding integrins [
14]. Furthermore, integrin α11β1 contracts collagen matrices experimentally [
17], and we therefore suggest that integrin α11β1-mediated PIF modifications can involve a contraction of the interstitial space mediated by direct or indirect binding of integrin α11-positive fibroblasts to collagen.
The involvement of integrin α11-positive fibroblasts in tumor PIF homeostasis is supported by our study of heterospheroids, where we observed a similar PIF reduction in spheroids composed of MDA-MB-231 cells and integrin α11-deficient fibroblasts. This simplified system allows us to investigate how fibroblasts grown together with tumor cells can influence PIF [
18‐
20]. In line with our results, a similar integrin α11β1 function in pressure regulation has previously been shown in fibroblasts/lung cancer heterospheroids [
20]. However, although these avascular spheroid studies indicate that the pressure regulatory abilities of integrin α11β1 is, at least in part, mediated by integrin α11-positive fibroblasts, the mechanisms behind integrin α11-mediated effect on PIF in heterospheroids are not investigated in detail in this study. In addition, we cannot exclude additional factors in the more complex in vivo system, such as influence of the tumor vasculature, which has been shown to have an important impact on tumor PIF [
13,
43‐
45].
Furthermore, integrin α11-deficiency led to less organized and thinner collagen fibrils in the orthotopic models, which could be a contributing factor to reduced tumor PIF. Although it has been shown that the collagen-binding proteoglycan fibromodulin promotes the formation of a dense stroma and increased tumor PIF [
46], it is nevertheless difficult to predict how different components in the extracellular matrix affect the hydraulic conductivity of tissues and thereby fluid flow and PIF [
47].
Although the present study is the first to identify integrin α11β1 as participating in regulation of pressure in solid tumors, it is already known to maintain a homeostatic PIF in dermis [
29,
30]. Furthermore, we have previously demonstrated the function of β1-integrins in the regulation of dermal PIF by inhibiting β1-integrins [
21].
Numerous studies have highlighted the role of CAFs in tumor progression, invasion and metastasis, either directly by stimulation of tumor cells via production of pro-tumorigenic growth factors or indirectly by for example remodeling the ECM (reviewed in [
48]). Here we show that integrin α11β1, known to be expressed during myofibroblast differentiation [
25,
26], seems to facilitate breast tumor growth in vivo.
In previous studies, the pro-tumorigenic abilities of integrin α11β1 have been associated with increased matrix stiffness, collagen reorganization and increased levels of IGF-2 [
28,
40]. In the present study, changes in pressure homeostasis and collagen organization could both influence tumor growth and invasion. Of interest, increased tumor PIF has been linked to tumor aggressiveness in some human cancers [
9,
49], and is an independent poor prognostic factor in cervical cancer [
10,
50].
There have been reports suggesting that increased tumor PIF can both facilitate and inhibit tumor progression. First, major pressure gradients due to increased tumor PIF can enhance interstitial fluid flow at and lymph drainage from the tumor margins, which probably increase the risk of cancer cells leaving the tumor. Increased flow can also facilitate tumor progression indirectly by either mechano-modulation of the tumor stroma or by changing the host immune response and thereby promote immunological tolerance (reviewed in [
51,
52]). Further, in vitro elevation of tumor PIF increased proliferation of human osteosarcoma [
53] and oral squamous cell carcinoma cells [
54]. Similarly, in vivo lowering of tumor PIF, and thereby reduction of mechanical stretch for 24 h, reduced tumor cell proliferation in vulva and lung xenograft tumors [
55]. However, contrary to these findings, increased tumor PIF may also limit uptake of nutrition and growth factors into the tumor and thereby inhibit tumor cell progression [
8]. In the context of breast cancer, MDA-MB-231 cells have actually been shown to invade towards regions of higher pressure in vitro [
56,
57], indicating that the elevated tumor PIF may in fact restrain breast tumor invasion. In summary, these findings demonstrate that maintenance of a high tumor PIF may be a contributing factor to integrin α11β1’s pro-tumorigenic effects, but at the same time, it can have opposite effects during tumorigenesis, pinpointing the need for further preclinical investigation.
Although increased tumor PIF can be a major barrier in cancer treatment, lowering of tumor PIF by integrin α11-deficiency did not increase the uptake of the low molecular weight drug
3H-5FU into MDA-MB-231 tumor interstitium. Low molecular weight compounds are transported by both diffusion and bulk flow/convection, and we have previously shown that reducing PIF can increase the uptake of the small molecular weight drugs
3H-5FU [
11,
58] and
51Cr-EDTA [
12,
59] into the tumor interstitium. However, in parallel with the results in the present study, it is evident that lowering of PIF will not always increase the uptake of low molecular weight drugs [
35,
60]. Similarily, Flessner et al. showed that decapsulation of ovarian xenografts markedly decreased PIF to zero, but did not increase penetration of the high molecular weight drug trastuzumab into the tumor [
61]. In summary, probably other features of the tumor microenvironment can also contribute to the failure of transport within solid tumors [
5,
61].
Our data show that integrin α11-deficiency leads to thinner and less organized collagen fibrils in the orthotopic tumor stroma. Changes in collagen composition and organization are already known to influence tumorigenesis and can predict breast cancer behavior [
3]. For example, progressive deposition of collagen [
62] and increased collagen fiber linearization [
63,
64] are associated with breast cancer aggressiveness.
While integrin α11-deficiency influenced tumor growth and reduced PIF with concomitantly more disorganized collagen fibrils in the orthotopic tumors, no effect was seen in the MDA-MB-231 ectopic tumors. Interestingly, there was similar amount of integrin α11β1 expression in both the MDA-MB-231 models. In a previous study, we also observed that while integrin α11-deficiency reduced RM11 tumor growth, but did not affect 4 T1 tumor growth, the integrin α11β1 expression was not higher in RM11 than in 4 T1 tumors [
33]. Thus, differences in integrin α11β1-expression cannot explain the contrasting effect seen in these in vivo models.
The different effects seen between the MDA-MB-231 orthotopic and ectopic tumors show that tumor location significantly influences the effect of integrin α11β1 in vivo. The tumor microenvironment displays a significant heterogeneity [
65], and the subcutaneous location probably does not always give rise to a representative tissue-specific stromal infiltration [
66‐
68]. Supporting the fact that the organ-specific fibroblasts influence breast tumor growth differently, co-injection of breast fibroblast with breast tumor cells increased tumor growth, whereas no enhancement was seen with the co-injection of skin fibroblasts [
69]. The significance of the local microenvironment illustrates the complexity of in vivo studies, and may explain some of the discrepancies seen with different mouse models. This underlines the importance of choosing the appropriate preclinical model, particularly when investigating the tumor microenvironment. If possible, orthotopic models should be preferred rather than ectopic ones.