Loss of Kaiso expression in breast cancer cells prevents intra-vascular invasion in the lung and secondary metastasis

The metastatic activity of breast carcinomas results from complex genetic changes in epithelial tumor cells and accounts for 90% of deaths in affected patients. Although the invasion of the local lymphatic vessels and veins by malignant breast tumor cells and their subsequent metastasis to the lung, has been recognized, the mechanisms behind the metastatic activity of breast tumor cells to other distal organs and the pathogenesis of metastatic cancer are not well understood. In this study, we utilized derivatives of the well-established and highly metastatic triple negative breast cancer (TNBC) cell line MDA-MB-231 (MDA-231) to study breast tumor metastasis in a mouse model. These MDA-231 derivatives had depleted expression of Kaiso, a POZ-ZF transcription factor that is highly expressed in malignant, triple negative breast cancers. We previously reported that Kaiso depletion attenuates the metastasis of xenografted MDA-231 cells. Herein, we describe the pathological features of the metastatic activity of parental (Kaisopositive) versus Kaisodepleted MDA-231 cells. Both Kaisopositive and Kaisodepleted MDA-231 cells metastasized from the original tumor in the mammary fat pad to the lung. However, while Kaisopositive cells formed large masses in the lung parenchyma, invaded large pulmonary blood vessels and formed secondary metastases and large tumors in the distal organs, Kaisodepleted cells metastasized only to the lung where they formed small metastatic lesions. Importantly, intravascular invasion and secondary metastases in distal organs were not observed in mice xenografted with Kaisodepleted cells. It thus appears that the lung may constitute a barrier for less invasive breast tumors such as the Kaisodepleted TNBC cells; this barrier may limit tumor growth and prevents Kaisodepleted TNBC cells from invading the pulmonary blood vessels and forming secondary metastases in distal organs.

Introduction An estimated 90% of medical fatalities in cancer patients are due to metastases [1]. Carcinomas of the mammary gland [2,3], prostate gland [4], liver [5][6][7][8][9], pancreas [10], endometrium [11], thyroid gland [12,13] and Merkel cell [14] have been shown to invade lymphoid vessels and blood vessels [2,6,8,9,[11][12][13][14] resulting in metastases to distant organs, particularly the lung [7,9]. According to the systemic and pulmonary circulatory patterns, cancer cells that escape the primary tumor site and enter the bloodstream would first disseminate to the lung via the right ventricle before dispersing from the lung through the left heart to distal organs. Characterization of the mechanisms of malignant mammary carcinoma indicates that in the primary tumor, a small population of cells travel towards the blood vessels, and invade them in a complex process involving enhanced activity of genes regulating the dynamics of the actin cytoskeleton, e.g. Mena [2,[15][16][17] and LIM kinase1 [18]. In a series of in vivo experiments in mouse and rat models utilising multiphoton microscopy, migrating tumor cells were seen to be assisted by perivascular macrophages in the process of vascular invasion and metastases to distant organs, which involved chemotaxis regulated by EGF and CSF-1 [1,[15][16][17][19][20][21][22][23].
The progression of breast tumors towards an invasive behaviour and metastasis are postulated to involve several molecular factors associated with the complex epithelial-to-mesenchymal transformation (EMT) process that endows tumor cells with the ability to modulate their cell-cell adhesion and the extracellular matrix [24][25][26]47,48,[50][51][52], apparently involved in the first line of defense against metastatic tumors. EMT is frequently accompanied by loss of the epithelial marker E-cadherin, concurrent with elevated expression of vimentin [41][42][43][44][45], an intermediate filament that participates in cell motility [46], as well as increased expression of matrix metalloproteases-2 and -9 (MMP-2, MMP-9) [27-32] that are often assessed in the determination of poor prognosis in breast cancer patients [33][34][35][36]. Tissue plasminogen activator (tPA) and urokinase plasminogen activator (uPA) are known to activate pro-enzyme forms of MMP-2 and MMP-9 to active forms [37,38] and tPA and uPA have been proposed as markers for breast cancer progression [39,40].
Despite remarkable advancements in our understanding of cellular and molecular mechanisms involved in tumor growth and intravascular invasion at primary tumor sites, relatively little is known about how malignant tumors travel to distant organs. Since all lymph and venous blood from the body flows via the right heart ventricle to the lung, it is expected that most if not all primary metastases of carcinomatous tumors are trapped in small pulmonary blood vessels. However, the relevant questions regarding the pathogenesis of metastatic cancer remain; how do secondary metastases travel to other distant organs, and what are the mechanisms involved in the generation of these secondary metastases?
To gain insight into the mechanisms involved in breast tumor metastases to distant organs, we studied the pathogenesis of secondary metastases of parental, Kaiso positive and experimental Kaiso depleted MDA-231 TNBC cells in immunocompromised mice [53]. Kaiso is a dualspecificity transcription factor that is highly expressed in, and linked with the aggressive features of breast, prostate, colon and pancreatic carcinomas [54][55][56][57]. We previously reported that Kaiso depletion strongly inhibited the metastasis of TNBC cells to distal organs [53]. Herein, we describe histological analyses of the metastases generated by parental Kaiso positive and Kaiso depleted MDA-231 cells.
We found that Kaiso positive MDA-231 breast cancer cells implanted into the mammary fad pad of immunocompromised mice [53] formed large tumor masses in the lung parenchyma, invaded large blood vessels and metastasised to other distant organs where it also formed large masses. In contrast, Kaiso depleted tumor cells formed small clusters only in the lung parenchyma and did not invade blood vessels and did not metastasize to distant organs. While the role of high Kaiso expression in the metastatic activity of human breast cancer cells was reported in our previous paper [53], we describe here the critical importance of high Kaiso expression in the propagation of breast tumors to distal organs beyond the lung, which we call "secondary metastases". We propose that the lung serves as the second line of defense against carcinomas with metastatic potential; tumors with less malignant characteristics such as Kaiso depleted MDA-231 cells are trapped, do not progress and perhaps are eliminated. In contrast, malignant tumors such as Kaiso positive MDA-231 cells thrive in the lung to form large masses that then, invade the large pulmonary blood vessels, travel via the left heart ventricle and lodge in small blood vessels of a variety of peripheral organs where they form secondary metastases.

Generation of stable Kaiso-depleted MDA-231 cells
The MDA-231 TNBC cell line was purchased from the American Type Culture Collection (ATCC) (Manassas, VA, USA), and cultured in Dulbecco's Modified Eagle's Medium (DMEM; Lonza BioWhittaker 1 , Walkersville, MD, USA) supplemented with 10% Fetal Bovine Serum (FBS; Hyclone, Logan, Utah, USA), 0.1 mg/mL penicillin/streptomycin and 0.25 g/mL Fungizone (Invitrogen, Grand Island, NY, USA). Cells were passaged every 2 or 3 days and grown in a 5% CO 2 humidified incubator at 37˚C. Stable Kaiso-depletion was achieved by using multiple Kaiso-specific short hairpin RNAs (shRNA) that target the Kaiso mRNA specifically as previously described [49]. A scrambled shRNA that does not target the Kaiso mRNA was used as a control. Stable control (Kaiso positive ) and Kaiso depleted (sh-K) MDA-231 cells were maintained in DMEM-supplemented media treated with Puromycin (Invitrogen) at 0.8 μg/mL.

Western blot analysis
Stable control Kaiso positive and Kaiso depleted (sh-K1 and sh-K2) MDA-231 cells were cultured until~80-90% confluent, washed twice with cold PBS, and then harvested by cell scraping into cold microfuge tubes. Control, sh-K1 and sh-K2 MDA-231 cells were then pelleted by centrifugation, lysed, and protein isolated as previously described [59]. Rabbit anti-Kaiso polyclonal (1:5,000 dilution; a generous gift from Dr. A. Reynolds), and mouse anti-β-actin monoclonal (1:50,000 dilution, Sigma Aldrich, Oakville, ON, Canada) primary antibody incubations were performed overnight at 4˚C. Secondary antibody incubations were performed with goat antirabbit-or donkey anti-mouse-horseradish peroxidase-conjugated secondary antibodies. (1;10,000 dilution, Jackson ImmunoResearch Laboratories, West Grove, PA, USA). Signals were then amplified with Clarity Western Enhanced Chemiluminescence substrate as previously described [53]. The sh-K2 MDA-231 cells, hereafter referred as sh-K or Kaiso depleted MDA-231 cells, where chosen for all subsequent studies as these cells displayed the most efficient Kaiso knockdown.

Animal studies
All animal studies were approved and performed at McMaster University, Ontario, Canada according to the guidelines by the Canadian Council for Animal Care. Extensive description of animal studies has been outlined previously [53]. Briefly, 4.5 x 10 6 Kaiso positive or Kaiso depleted MDA-231 cells were injected subcutaneously into the mammary fat pad of 6-8 week old female NOD SCID Gamma (NSG, Jackson Laboratories) mice (n = 5 each per condition), and allowed to form prominent subcutaneous masses up to 3,300 mm 3 in volume. Non-invasive monitoring of mice was performed once a week, and increased to 2-3 times per week upon tumor appearance. Tumor growth was monitored externally with vernier calipers and tumor volume (in mm 3 ) measured using the following formula; length/2 × width 2 , 2-3 times per week [53]. This tumor mass volume was achieved before the onset of serious clinical signs such as body weight loss, dehydration and lethargy that would require the application of the endpoint and euthanasia. The endpoint tumor volume of 3,300 mm 3 was determined in pilot experiments prior to the study. At endpoint, the mice were euthanized by overdosing with intraperitoneal injection of sodium pentobarbital (100 mg/kg body weight), perfused when deeply anaesthetized, and fixed in 10% formalin and all routine tissues collected for histological examination.

Protein expression
Expression of Kaiso in MDA-231 cells was abundant but remarkably reduced in sh-K1 (to 28%) and sh-K2 (to 6%) (see Fig 1) as the result of the stable transfection of the Kaiso-specific shRNA in these cells. Sh-K2 cells were utilized as Kaiso depleted in the xenograft studies.

Clinical observations and histological analyses
The subcutaneous masses in the mammary gland fat pad reached the endpoint volume of 3,300 mm 3 within 8 weeks in Kaiso positive tumors and 12 weeks in Kaiso depleted tumors [60]. Further characterization of the Kaiso positive and Kaiso depleted MDA-231 phenotypes revealed that the difference in tumor growth was due to Kaiso-depletion effects on cell proliferation, anchorage-independence and apoptosis [60].
The injection of the Kaiso positive and Kaiso depleted MDA-231 cells into the mammary fat pad of immunocompromised mice resulted in the formation of large subcutaneous masses (Fig 2Ai and 2Aiii) formed by large, pleomorphic cells with high mitotic index (Fig 2Aii and  2Aiv). Primary tumor masses formed by both types of mammary carcinoma cells were morphologically indistinguishable from each other (Fig 2A). Veins and lymphatic vessels in vicinity to some subcutaneous masses were distended with clusters or single tumor cells scattered throughout the lumen (Fig 2B).
In the lung, Kaiso positive cells formed large, often coalescing, non-circumscribed tumor masses with obliteration of the alveolar architecture ( Fig 3A) formed by large, pleomorphic cells with a high mitotic index. Scattered neutrophils infiltrated the periphery of the tumor masses and the surrounding alveolar tissue. A proportion of large blood vessels encompassed by or  adjacent to tumors (Fig 3A) had masses of tumor cells protruding into the lumen, with the segmental concurrent obliteration of the vascular wall by tumor cells forming a continuity of perivascular and intravascular tumors (Fig 3B-3D). The apparent vascular invasion of Kaiso positive tumors was often associated by formation of thrombi infiltrated by tumor cells (Fig 3D). In contrast, Kaiso depleted tumor cells formed small aggregations scattered in the alveolar parenchyma (Fig 3E and 3G), sometimes adjacent to large blood vessels but with no invasion of the vascular wall or the lumen (Fig 3F). Kaiso depleted tumor aggregations were often infiltrated by scattered neutrophils (Fig 3G).
Although Kaiso positive and Kaiso depleted cancer cells both formed subcutaneous masses and invaded nearby veins and lymphatics resulting in pulmonary metastases, Kaiso-depletion markedly supressed the ability of cancer masses to grow expansively in the lung, and eliminated their ability to invade blood vessels thereby restricting their spread to other organs. Indeed, tumors were not observed in any other organ except in the lung of mice injected with Kaiso depleted cells. In contrast, in mice injected with Kaiso positive cells, large tumors were observed in the liver (Fig 4A and 4B), kidney (Fig 4C and 4D), myocardium (Fig 4E and 4F), and infrequently in the adrenal gland and leptomeninges of the brain (not shown). Thus, Kaiso appear to play a regulatory role to in the; (i) expansive growth of metastatic tumors in the lung and (ii) invasion of the pulmonary blood vessels to spread to other organs supports the notion of Kaiso as a crucial factor in highly aggressive subtypes of breast cancer [53,54,60]. Our findings suggest that Kaiso could be a target for therapeutic strategies in the treatment of aggressive breast cancers. Some large blood vessels within a tumor mass or adjacent to it, such as in the liver (Fig 4A), or in kidney (Fig 4C) had intraluminal invasion of tumor cells with formation of thrombus and obliteration of the adjacent segment of the wall of the blood vessel (Fig 4B and 4D). In the myocardium, protrusions of the tumor cells into the lumen of ventricles ( Fig 4E) resulted in the formation of fibrinous thrombus (Fig 4F) infiltrated by tumor cells. Large, sometimes coalescing tumor masses obliterated the organ architecture and were composed of large pleomorphic cells similar morphologically and immunohistochemically to those described in the subcutaneous mass and in the lung of mice injected with Kaiso positive cells. Intravascular (Fig 5A) or cardiac intraventricular (Fig 5B) invasion by the Kaiso positive tumor cells often resulted in formation of a thrombus whose surface in some cases was apparently covered by endothelium (Fig 5Aii and 5Bii), delineating the remaining lumen of the blood vessel or the ventricle. Other thrombi however, where not delineated by endothelium but often by a layer of neutrophils (Fig 5Aiii and 5Biv). In some blood vessels adjacent areas of thrombi were or were not endothelialized while in others apparently endothelium-free small clusters or individual cancer cells were present in the lumen (Fig 5Aiii). The above data indicate that the subcutaneously implanted Kaiso positive MDA-231 cells efficiently penetrated the vascular wall and invaded the lumen of large blood vessels in the lung and other organs (Figs 4 and 5). This is in line with other studies that have also demonstrated vascular taxis and intravascular invasion of breast tumors implanted subcutaneously into mice [1,[15][16][17][19][20][21][22][23]. The active penetration of the vascular wall by Kaiso positive cells lead to their accessing of the lumen with frequent formation of thrombus. This pathogenesis implies two potential mechanisms; (1) tumor cells breached the endothelium of the tumor-invaded blood vessel, which may have led to thrombosis; (2) the tumor cells then invaded the thrombus as the convenient substrate, which lead to the increase of the intravascular load of cancer cells destined to metastasize to other organs. We also observed endothelium lining of the intravascular tumor masses with or without thrombosis. We consider that neo-endothelialization of the intravascular tumor masses and tumor thrombi may serve as a defense mechanism preserving the patency of the blood flow. We thus postulate that this putative defense mechanism may actually be subverted by the invading tumor cells allowing for the increase of their intravascular load and presumably leading to a greater chance of success of secondary metastases.

Immunohistochemical analyses
Kaiso positive primary tumor tissues, as expected, stained positive for Kaiso (Fig 6A), which localized to both the nucleus and cytoplasm. Further analysis of the Kaiso positive tumor tissues for other molecular markers implicated in tumor metastasis revealed that similar to our previous observations in vitro [53], Kaiso positive MDA-231 primary tumor tissues stained moderately for Vimentin ( Fig 6B) and negative for E-cadherin (Fig 6C). We also examined the Kaiso depleted primary tumor tissues for Kaiso, Vimentin and E-cadherin expression. While we expected little to no Kaiso staining in the Kaiso depleted MDA-231 tumors as per our western blot results (Fig 1), we were surprised to observe weak Kaiso staining in the primary tumors which was predominantly cytoplasmic (Fig 6D). The weak Kaiso staining could be due to deselection of the Kaiso depleted MDA-231 cells as the mice were not subjected to constant treatment with Puromycin that would ensure selection and maintenance of shRNA plasmid in the stable cells in vitro. Nonetheless, the Kaiso staining observed in Kaiso depleted tumor tissues was still remarkably reduced compared to the staining observed in the Kaiso positive tumor tissues (Fig 6A and 6D). Kaiso depleted MDA-231 tumor tissues also stained weakly for Vimentin ( Fig 6E) as expected from in vitro findings in our previous report [53]. In contrast, while we had observed increased E-cadherin expression in the Kaiso depleted MDA-231 cells in vitro [53], the Kaiso depleted MDA-231 tumor tissues stained negative for E-cadherin (Fig 6F). This lack of E-cadherin staining could be due to the weak levels of Kaiso expression observed in the Kaiso depleted MDA-231 primary tumors (Fig 6D) or due to other in vivo factors in the tumor microenvironment that are independent of Kaiso's expression or Kaiso's effect on E-cadherin expression.
Analysis of the tumor masses observed in the lung (Fig 7A) of mice injected with the Kaiso positive and Kaiso depleted MDA-231 cells also revealed that the Kaiso positive tumor cells that metastasized to the lung exhibited strong Kaiso (Fig 7Ai) and Vimentin (Fig 7Aii) staining but no E-cadherin staining (Fig 7Aiii), while Kaiso depleted tumor metastases in the lung displayed weak Kaiso (Fig 7Aiv) and Vimentin (Fig 7Av) staining, but no E-cadherin staining (Fig 7vi). These findings suggest a persistence of the molecular phenotype of both Kaiso positive and Kaiso depleted MDA-231 cells as they metastasized to the lung. We further analyzed the Kaiso positive and Kaiso negative MDA-231 lung metastases for the expression of MMP-2 and MMP-9.  Although, MMP-2 and MMP-9 had not previously been examined in the Kaiso positive and Kaiso negative MDA-231 cells in vitro, they have been implicated in EMT and breast cancer progression [61][62][63]. Consistent with the EMT phenotype in the Kaiso positive tumor cells, Kaiso positive lung metastases displayed strong MMP-2 and MMP-9 expression (Fig 7Bi and  7Bii). In contrast, Kaiso depleted lung metastases displayed reduced MMP-2 (Fig 7Biii) but not MMP-9 (Fig 7Biv) expression.
Examination of the Kaiso positive tumor cells in secondary metastases (Fig 7C) also revealed strong Kaiso (Fig 7Ci) and Vimentin (Fig 7Cii) staining but no E-cadherin staining (Fig 7Ciii) in liver metastases. Similar to the lung metastases, the Kaiso positive liver metastases also displayed strong MMP-2 (Fig 7Civ) and MMP-9 (Fig 7Cv) staining. Likewise, the Kaiso positive tumor masses in the myocardium displayed strong Kaiso (Fig 7Cvi) and Vimentin (Fig 7Cvii) staining but no E-cadherin staining (Fig 7Cviii), as well as strong MMP-2 (Fig 7Cix) and MMP-9 (Fig 7Cx) staining. These findings also imply a persistence of the molecular phenotype of the Kaiso positive MDA-231 cells as they metastasized to other distal organs.
In this study we analyzed the metastatic progression of Kaiso positive and Kaiso negative malignant mammary carcinomas using in vivo transplantation experiments in a mouse model. We found that Kaiso positive and Kaiso negative MDA-231 cancer cells both formed subcutaneous masses and invaded nearby veins and lymphatics apparently leading to pulmonary metastases. However, Kaiso-depletion was associated with remarkable suppression of the growth of cancer cells in the lung. In contrast, high Kaiso-expressing tumor cells thrived in the lung, invaded large pulmonary blood vessels and metastasized to other organs. These findings suggest that Kaiso plays a key role in metastatic activity of MDA-231 cancer cells. The penetration of the vascular wall and invasion of the lumen of large blood vessels abundant in the lung and other organs appears to be the fundamental factor of malignancy of the Kaiso positive tumor cells. This is in line with other studies that have also demonstrated vascular taxis and intravascular invasion of breast tumors implanted subcutaneously into mice [1,[15][16][17][19][20][21][22][23]. Immunohistochemical characterization of Kaiso positive and Kaiso depleted tumors in this study revealed that Kaiso positive tumor cells exhibit more features associated with malignancy (increased Kaiso, Vimentin, MMP-2 and MMP-9 expression) than the Kaiso depleted tumor cells, which only displayed increased MMP-9 expression. Moreover, Kaiso expression seem to correlate positively with Vimentin and MMP-2 but not MMP-9 expression. Indeed Kaiso-depletion resulted in decreased Vimentin and MMP-2 but not MMP-9 expression. The similar staining of MMP-9 and lack of E-cadherin staining in both Kaiso positive and Kaiso depleted tumor cells, which is a marker of EMT, may explain the ability of both cell types to metastasize to the lungs. However, considering that only the Kaiso positive tumor cells were capable of surviving in the lungs, invading blood vessels and forming macrometastases in other distal organs, the higher Kaiso expression in concert with the increased Vimentin and MMP-2 expression could be considered as the critical determinants that allowed the Kaiso positive tumors to thrive after metastasis to the lung, and other distant organs.
We thus propose that the lung serves as the second line of defense against carcinomas with metastatic potential where tumors with less malignant characteristics, e.g. with reduced expression of Kaiso, are trapped, do not progress and perhaps are eliminated. Consequently, secondary metastases to the distant organs are prevented. We further postulate, that malignant tumors such as human breast tumors with high Kaiso expression can overcome this defensive mechanism, thrive in the lung and form large masses whose cells invade the blood vessels, travel via the left heart ventricle to lodge in small blood vessels of a variety of peripheral organs, and initiate multiple secondary metastatic tumors leading to accelerated demise (Fig 8). If this hypothesis is proven to be correct, potential cancer-suppressive tissue mechanisms in the lung should be considered in the pathogenesis of cancer metastasis. Also, models used to evaluate

Conclusions
In this report, we analyzed the metastatic progression of Kaiso positive and Kaiso negative malignant mammary carcinomas using in vivo transplantation experiments in a mouse model. Although this study utilizes the end point metastasis analysis of disseminated breast tumor cells, it highlights potential novel mechanisms involved in secondary metastases and provides detailed histological evidence of different behaviour of MDA-231 malignant breast cancer cells depending on the expression level of Kaiso. Both Kaiso positive and Kaiso depleted tumor types; (1) formed subcutaneous masses of cells with morphological features of malignancy; (2) invaded adjacent veins and lymphatic vessels; and (3) metastasized to the lung. However, while Kaiso positive cells; (i) formed large pulmonary tumors; (ii) actively invaded pulmonary blood vessels apparently leading to (iii) secondary metastases and tumors in a variety of distal organs, Kaiso depleted tumors formed only small aggregates in the lungs, did not invade pulmonary blood vessels and did not form secondary metastases. Thus, Kaiso may be a potent factor enabling breast cancer cells to overcome apparent inhibitory mechanisms in the lung and to send secondary metastases throughout distant organs.