Monitoring Metastatic Behavior of Human Tumor Cells in Mice with Species-Specific Polymerase Chain Reaction: Elevated Expression of Angiogenesis and Bone Resorption Stimulators by Breast Cancer in Bone Metastases
Gabri Van Der Pluijm,
Department of Endocrinology and Metabolic Diseases C4-86, Leiden University Medical Center, Leiden, The Netherlands
Address reprint requests to: Gabri van der Pluijm, Leiden University Medical Center, Department of Endocrinology and, Metabolic Diseases C4-R, Albinusdreef 2, 2333 ZA Leiden, The Netherlands
Tumor-stroma interactions are of primary importance in determining the pathogenesis of metastasis. Here, we describe the application of sensitive competitive polymerase chain reaction (PCR) techniques for detection and quantitation of human breast cancer cells (MDA-MB-231) in an in vivo mouse model of experimental metastasis. Human-specific oligonucleotide primers in competitive PCR reactions were used to quantify the amount of MDA-MB-231 cells per tissue per organ. Using this species-specific (semi)quantitative PCR approach, gene expression patterns of (human) tumor cells or (mouse) stromal cells in metastatic lesions in the skeleton or soft tissues were investigated and compared. In all metastatic lesions, MDA-MB-231 cells express angiogenic factors (vascular endothelial growth factors [VEGFs]; VEGF-A, -B, and -C) and bone-acting cytokines (parathyroid hormone-related protein [PTHrP] and macrophage colony-stimulating factor [M-CSF]). In these metastases, PECAM-1-positive blood vessels and stromal cells of mouse origin are detected. The latter express angiogenic factors and markers of sprouting vessels (VEGF receptors flt-1/flk-1/flk-4 and CD31/PECAM-1). Strikingly, steady-state messenger RNA (mRNA) levels of VEGF-A and -B and the major bone resorption stimulators PTHrP and M-CSF by tumor cells were elevated significantly in bone versus soft tissues (p ≤ 0.05, p ≤ 0.0001, p ≤ 0.001, and p ≤ 0.05, respectively), indicating tissue-specific expression of these tumor progression factors. In conclusion, MDA-MB-231 breast cancer cells express a variety of factors in vivo that have been implicated in metastatic bone disease and that correlate with poor survival of patients with breast cancer. We hypothesize that the observed up-regulated expression of angiogenic and bone-resorbing factors by the breast cancer cells in the skeleton underlie the clinically observed osteotropism of breast cancer cells and pathogenesis of osteolytic bone metastases. The application of the species-specific competitive PCR-based assay in vivo can provide new information concerning the involvement of gene families in tumor progression and metastatic disease and greatly facilitates the study of tumor-stroma interactions in cancer invasion and metastasis.
BREAST CANCER is the most frequent cancer in women and the main cause of death due to malignancy. Metastatic bone disease is a frequent cause of morbidity in advanced breast cancer patients with a subsequent high incidence of skeletal complications (fractures, hypercalcemia, spinal cord compression, and severe pain).(1–7) Patients with metastatic breast cancer cannot be cured from their disease and their quality of life is adversely affected by the frequency and severity of the related morbidity. The mechanisms underlying the apparent preference of breast cancer cells for the skeleton are poorly understood. However, from the moment the cells are located in the bone microenvironment they release factors that stimulate bone resorption with a subsequent selective increase in the attraction and growth of new cancer cells to bone.(6, 8–12)
To examine processes involved in tumor progression and metastasis to bone and the role of tumor-derived factors in the destruction of its architecture, suitable experimental approaches are needed. In 1988, Arguello and coworkers introduced a nude mouse model that was designed specifically to study the pathogenesis of bone metastasis.(13) This model has been used to study the mechanisms of bone metastasis in a variety of tumors, including breast cancer. Several weeks after intracardiac inoculation of human MDA-MB-231 breast cancer cells, osteolytic bone metastases are formed.(14) These mainly are caused by enhanced formation and activity of osteoclasts. In recent years considerable advances have been made in understanding mechanisms underlying osteoclastic bone resorption by cancer cells.(6, 14–21) Through these studies it became apparent that the interactions between tumor cells and the bone marrow/bone microenvironment of the host are important determinants of these responses. New evidence further established the significant role of other processes, in particular angiogenesis, which is mediated by endothelial cells of the host, in the development and amplification of bone metastases.(21)
The Arguello nude mouse model, apart from its role in the investigation of the pathogenesis of bone metastasis, allows also the study of various agents such as antiresorptive and antiangiogenic compounds in the prevention and treatment of metastatic disease.(14, 21–23) Therefore, methods that can discriminate between tumor- and host-derived factors involved in the metastatic process are needed. Until now, the formation of metastases from human breast cancer cells was determined by radiographs (areas of osteolytic bone lesions) and histochemistry (bone and soft tissues).(13, 14, 16, 17), (20, 23) However, close monitoring and quantitation of cancer cell metastasis and particularly the study of tumor-host interactions in vivo is difficult and labor intensive. Therefore, to address these tumor-bone interactions and allow separation, it is important to develop techniques that are suitable for tumor- and host-derived factors experimentally in vivo.
In the present study we describe polymerase chain reaction (PCR)-based techniques that allow the in vivo detection and quantitation of human breast cancer cells and tumor- and host-derived gene transcripts at metastatic sites in the Arguello nude mouse model.(13) We applied these techniques to investigate the differential expression of bone-acting cytokines (parathyroid hormone-related protein [PTHrP] and macrophage colony-stimulating factor [M-CSF]) and angiogenic factors (vascular endothelial growth factors [VEGFs]), suggested from clinical and animal studies to be important in the formation of distant metastases including bone.
MATERIALS AND METHODS
Breast cancer cell lines
The breast cancer cell line MDA-MB-231 was purchased from the American Type Culture Collection (Rockville, MD, USA). This cell line was established from a single pleural effusion obtained from a 51-year-old white woman with poorly differentiated adenocarcinoma.(24) Cells were cultured in RPMI1640 + 10% fetal bovine serum + penicillin/streptomycin (Life Technologies-Gibco BRL, Breda, The Netherlands) in a humidified incubator at 37°C at 5% CO2 until confluency.
The murine model of experimental metastasis
MDA-MB-231 breast cancer cells (105 cells/100 μl) were injected into the left heart ventricle of 8-week-old female BALB-c nu/nu mice according to the protocol described by Arguello and coworkers.(13) After 5 weeks, the animals developed osteolytic bone metastases, which are visualized by radiographs. At the end of the experiment, the animals were killed and bone and soft tissues were removed immediately and used for RNA isolation and/or immunohistochemical staining procedures (see the following section).
Immunohistochemistry and histochemical staining procedures
Long bones of the BALB-c nu/nu mice containing visible metastatic bone lesions (radiographs) were fixed in zinc-Macrodex formalin (ZnMF) fixative (0.1 M Tris acetate [pH 4.5] containing 0.5% ZnCl2, 0.5% zinc-acetate, 5% Dextran, and 10% formalin) overnight at room temperature, washed three times with phosphate-buffered saline (PBS), and decalcified in 10% sodium EDTA for 7 days at 4°C and subsequently processed for paraffin embedding.
Immunohistochemistry was performed on 5 μm of ZnMF-fixed paraffin-embedded sections of 17-day-old fetal hindlimbs with some modifications as described.(25) Sections were rehydrated and washed with PBS followed by incubation with 40% methanol/1% H2O2 in PBS for 15 minutes to abolish endogenous peroxidase activity. After three washes with PBS, the sections were incubated with 0.5% Boehringer blocking reagent (Boehringer milk protein [BMP]) (Boehringer, Mannheim, Germany) in 0.1 M Tris-buffered saline (pH 7.4) containing 0.02% Tween 20 (BMP/TNT) for 1 h at 37°C.
This was followed by incubation with primary rat anti-mouse antibody ER-MP12 directed against murine PECAM-1, kindly provided by Dr. P. Leenen (Erasmus University Rotterdam, The Netherlands) diluted in BMP/TNT for 16 h at 4°C. After three washes with TNT, the sections were incubated with biotinylated sheep anti-rat antibody (Amersham, Den Bosch, The Netherlands) diluted in BMP/TNT for 45 minutes at 37°C and followed by incubation with horseradish-conjugated streptavidin (streptavidin-horseradish peroxidase [HRP]; Amersham) diluted in BMP/TNT for 30 minutes. The signal was amplified using biotinylated tyramids as described(26) followed by incubation with streptavidin-HRP and finally detected by the chromogen 3-amino-9-ethyl-carbazole (AEC; Sigma Chemicals, Zwijndrecht, The Netherlands).
The presence of osteoclasts in bone metastases in mouse long bones was shown using tartrate-resistant acid phosphatase (TRAP) activity as described previously.(27)
Isolation of cellular RNA
Five weeks after inoculation of the breast cancer cells, the animals were killed and radiographs were taken. Bone/bone marrow (long bones from extremities) and soft tissues (brain, liver, lungs, and kidneys) were explanted and used for RNA isolation procedures.
Visible skeletal metastases were excised surgically and/or bone marrow was collected from the animals as described earlier.(28) In brief, bone marrow was obtained by removing both ends of all limbs and subsequent flushing with ice-cold 0.9% NaCl.(28) In this experimental setup, cells are obtained from the central bone marrow compartment while cells at the bone surface do not detach and are not included.(28) Bone marrow cells were resuspended in 0.9 ml 4 M guanidinium isothiocyanate lysis buffer and stored at −20°C until RNA isolation.
RNA was isolated from bone and soft tissues according to the method described by Chomczynski and Sacchi.(29) Briefly, cells and tissues were homogenized in the lysis buffer, extracted with phenol and chloroform, precipitated at −20°C with 100% isopropanol, resuspended in autoclaved denatured water, and stored at −80°C. RNA concentration was determined spectrophotometrically assuming 40 μg/ml per optical density at a wavelength of 260 nm (1 cm path length) and quality was checked on a 1% agarose gel containing 0.5 μg ethidium bromide (EtBr)/ml.
Denatured RNA (3 μg; 5 minutes at 70°C and quick-chilled on ice) was reverse transcribed into complementary DNA (cDNA) in a 50-μl reaction volume containing first-strand buffer (75 mM KCl, 3 mM MgCl2, and 50 mM Tris-HCl [pH 8.3]), 10 mM dithiothreitol (DTT), 0.5 mM deoxynucleoside triphosphate (dNTP), 600 ng random hexanucleotide primers, 1 U RNasin/μl (Promega, Leiden, The Netherlands), and 2.5 U murine myeloid leukemia virus (M-MLV)/μl (Life Technologies-Gibco BRL). Reverse transcription was performed at 37°C for 90 minutes and 70°C for 10 minutes, followed by quick chilling on ice.
Species-specific competitive PCR
To correct for any variation in RNA content, cDNA synthesis, and the presence of tumor or host cells between the different preparations, samples were equalized on the basis of their β2-microglobulin content.
Human-specific primer sets were used to identify tumor-derived (human) steady-state messenger RNA (mRNA) levels for housekeeping genes and tumor progression factors (Table 1). Similarly, mouse-specific primers were used to identify and quantitate murine steady-state mRNA levels (Table 2).
Table Table 1.. Human-Specific S and AS Primers Used to Amplify cDNA and the IS pQA1 Sequences
Table Table 2.. Mouse-Specific S and AS Primers Used to Amplify cDNA and IS pMUS Sequences
The 5 ng cDNA was coamplified with human- or mouse-specific β2-microglobulin primers over 32 cycles in the presence of 4-fold serial dilutions of the internal standards pMUS (mouse specific) or pQA1 (human specific) as described previously.(28, 30, 31) To analyze human or mouse-derived mRNA expression, 2 ng cDNA was amplified over 32 cycles (semiquantitative PCR) unless stated otherwise. An outline of the species-specific competitive PCR technique is depicted in Fig. 1.
Competitive PCR was performed in 25 μl of reaction volume containing reaction buffer [75 mM Tris-HCl, pH 9.0, 20 mM (NH4)2SO4, 0.01% (wt/vol) Tween 20], 2.0 mM MgCl2, 200 μM dNTPs, 0.25 μM sense (S) and antisense (AS) primer (Eurogentec, Seraing, Belgium), 0.125 U Goldstar DNA polymerase (Eurogentec,), and cDNA and internal standard pMUS or pQA1 to be coamplified. Negative controls in which cDNA or both cDNA and internal standard were omitted were run in parallel in each experiment. After one cycle of 30 s at 94°C, cDNA and internal standard were coamplified by repeated cycles of which one cycle consisted of a 30-s denaturing step at 94°C, a 30-s annealing step at 56°C, and a 1-minute primer extension step at 72°C. This was followed by one cycle of 2 minutes at 72°C. Aliquots of 20 μl of each amplified sample and the 100-base pair (bp) DNA ladder (Life Technologies-Gibco BRL) were subjected to electrophoresis on a 1% agarose gel containing 0.5 μg EtBr/ml and photographed. The intensity of each band was measured using computerized densitometry. The intensity of the cDNA and internal standard pMUS or pQA1 amplicons could be measured separately on gel because of their difference in size. The internal standard amplicons had sizes of 300 bp (pMUS) or 370 bp (pQA1) while those of the cDNA amplicons were 222 bp (mouse β2-microglobulin) or 268 bp (human β2-microglobulin), respectively. All the internal standards contained the same specific nucleotide sequence for β2-microglobulin and each of the mRNA transcripts to be examined and therefore could be amplified with the corresponding primer set.
The amplification efficiency of the internal standards pQA1 and pMUS was compared with that of cDNA for β2-microglobulin and M-CSF. cDNA and a competitive dilution of pQA1 were coamplified in the presence of 3 μCi of [α-32P]deoxycytidine triphosphate (dCTP)/100 μl of reaction volume over a range of PCR cycles (22-38 cycles depending on the specific mRNA amplified). Ten-microliter aliquots were subjected to electrophoresis on a 5% polyacrylamide gel. Gels were dried on a gel dryer model 583 (Bio-Rad Laboratories, Veenendaal, The Netherlands). Autoradiographs were prepared using Kodak XAR-5 film and the intensifying screen at −70°C. Intensity of the bands was quantified using computerized densitometry. The use of one primer set allowed equal amplification efficiency of the cDNA and internal standard pQA1 (see Results section) and pMUS (results not shown) as described earlier.(28, 30, 31)
In quantitative measurements, the cDNA/internal standard ratio remained constant throughout the amplification process even into the plateau phase (see Results section). This implies that in the case of a cDNA/internal standard ratio of one, the initial amount of cDNA was equal to the initial amount of the internal standard. Coamplification of cDNA with serial dilutions of the internal standard and plotting the cDNA/internal standard ratio against the number of copies of internal standard allowed us to determine the number of copies of internal standard with which the cDNA sample could compete. The point at which the cDNA/internal standard ratio is equal to one is independent of the linearity of EtBr staining because loss of linearity will flatten the slope of the ratio between the intensities of the amplicons but will not affect the point at which the intensities of the amplicons are equal to each other. Therefore, relative differences in mRNA expression were determined by comparing the number of copies of internal standard with which cDNA could compete.
The internal standards for mice (pMUS) and humans (pQA1) were generous gifts of Dr. D. Shire (Sanofi Recherche, Labège, France).(30, 31) Specific S and AS primers to amplify both mouse cellular cDNA and internal standard pMUS were purchased from Eurogentec and Isogen Bioscience BV (Maarssen, The Netherlands). The primer sets used crossed intron/exon boundaries with such large introns that putative contaminations with genomic DNA will not be amplified in the amplification process.
Steady-state mRNA expression of tumor- and mouse-derived factors
After normalization for the housekeeping gene β2-microglobulin, steady-state mRNA levels were determined for a variety of tumor- and host-derived factors (Tables 1 and 2). Both semiquantitative and quantitative analyses were performed for mouse and tumor-derived transcripts. PCR reactions were performed in an end volume of 25 μl per tube (32 or 38 cycles, 56°C annealing temperature, and 2.0 mM MgCl2) using a Perkin Elmer 9700 (Perkin Elmer, Nieuwenkerk aan den Ijssel, The Netherlands).
S and AS primers were designed to hybridize specifically with either tumor (human) or host transcripts (mouse) during PCR reactions under the conditions specified previously. Human and mouse internal standards were used to standardize cDNA with species-specific primers for the housekeeping gene β2-microglobulin.(28, 30, 31)
ELISA for human VEGF
At confluency, MDA-MB-231 cells were cultured in RPMI1640 + 10% fetal bovine serum + penicillin/streptomycin in a humidified incubator at 37°C at 5% CO2 in 25-cm2 culture flasks for 48 h (Costar, Cambridge MA, USA). The VEGF concentration in the cell culture supernatant was determined by a quantitative sandwich enzyme immunoassay technique according to manufacturer's specification (Quantikine, human VEGF; R & D Systems, Inc., Minneapolis, MN, USA).
Validation of species-specific competitive PCR and detection of tumor cells in vitro and in vivo
The species-specific competitive PCR method developed to detect and quantitate tumor load and tumor- and host derived transcripts is depicted in Fig. 1. Human-specific oligonucleotide primers for the housekeeping gene β2-microglobulin were used to validate the species-specific PCR and investigate the possibilities for quantitative detection of tumor cells at different organs or tissues in vivo.
The species-specific nature of all the used oligonucleotide primer sets (Tables 1 and 2) was checked using appropriate negative controls (cDNA from other species). No amplification took place when human primer sets were used in PCR reactions with mouse cDNA and vice versa. An example of this species specificity is shown in Fig. 2 in which increasing amounts of human MDA-MB-231 breast cancer cells were added to a fixed number of mouse cells (MC3T3-E1 osteoblastic cells) and seeded in 6-well plates. The mixed populations were harvested and cellular RNA was isolated and reverse-transcribed into cDNA. Subsequently, standardized PCR reactions were performed for each sample with human-specific β2-microglobulin primers loaded onto agarose gels and separated by electrophoresis (Fig. 2A). The densities of each band correlated in a linear fashion with tumor cell numbers (r2 = 0.986; Fig. 2B). Furthermore, no amplification of mouse cDNA took place, indicating that the PCR reaction was indeed species (human)-specific (Fig. 2A, second lane). Conversely, mouse cells could be detected in an identical manner (r2 = 0.990) using mouse-specific β2-microglobulin primers (results not shown). These results show that species-specific PCR can be applied to detect sensitively either tumor (human) or host (mouse) cells in a mixed population of cells. Moreover, they indicate that tumor and host transcripts also may be detected separately in metastases in vivo.
Because amplification is exponential, small sample-to-sample concentrations and loading differences are amplified as well. Therefore, PCR requires careful optimization when used for quantitative cDNA analysis. This can be achieved by competitive PCR procedures involving adding a known amount of PCR-amplifiable standard into a cDNA sample and then amplifying the standard and target DNA in the same reaction. We have used human pQA1 and mouse pMUS internal standards for competitive PCR procedures. The internal standards contain the same binding sites for the used target sequences (β2-microglobulin in pQA1 and pMUS) and the amplicon size differs between the target sequences and the internal standards (Tables 1 and 2).
Next we investigated if similar PCR reactions could be applied to in vivo cDNA samples derived from various metastatic sites. Using human-specific primers for the housekeeping gene β2-microglobulin, tumor-derived steady-state mRNAs could be detected in soft tissues and bone. The following distribution of the breast cancer (micro)metastases in the Arguello nude mouse model was found; 74% of the mice have one or more bone metastases and 67% of mice contain soft tissue metastases (lung and brain, 56% and 22% of the mice, respectively).
When cDNA samples derived from in vivo bone metastases were used, amplification proceeded with the same efficiency for both the cDNA samples (mixture of human and mouse cDNA) and their respective internal standards, pQA1 (human) and pMUS (mouse) for β2-microglobulin as previously described(28, 30, 31); that is, the ratio between the amplification efficiency of cDNA and internal standards in the exponential phase is approximately one (Fig. 3). Fig. 3 depicts an example of coamplification of cDNA (20 ng) and pQA1 (197 × 103 copies) for human β2-microglobulin over a wide PCR cycle range (22-38 cycles) in the presence of [α-32P]dCTP. The difference between the intensity of the amplicons of cDNA and the internal standard pQA1 was constant throughout the exponential phase of the competitive PCR and remained constant in the nonexponential and plateau phases as expected.(30) This allowed us to perform the competitive PCR up to the nonexponential phase of the amplification process and to visualize the amplicons with the nonradioactive method of EtBr staining.(28, 30, 31) Similarly, no differences in amplification efficiencies between the mouse internal standard pMUS and cDNA samples were found for β2-microglobulin (slopes for cDNA and pMUS are 11.00 and 11.63, respectively; amplification efficiency = 0.95), as expected.(28)
Histology and species-specific competitive PCR in vivo
The sensitivity of the species-specific competitive PCR-based method to detect and quantify human metastatic breast cancer cells in different tissues was tested in mRNA samples obtained from the Arguello nude mouse model of experimental metastasis. Five weeks after inoculation of human MDA-MB-231 breast cancer cells into the left cardiac ventricle, osteolytic bone metastases could be detected radiographically (Fig. 4a). Many TRAP-positive (TRAP+) resorbing multinucleated osteoclasts were observed at the bone surface surrounding the metastatic bone lesions (Fig. 4b). Many blood vessels were observed after staining of bone metastases with the rat-anti-mouse PECAM-1 antibody for vascular endothelial cells (sprouting blood vessels; Fig. 4c).
Competitive PCR analyses were performed on the cDNA samples (bone and soft tissues) obtained from these nude mice. For this, human or mouse internal standards (pQA1 and pMUS, respectively) were used with the appropriate primers. Figure 5 depicts an example of competitive PCR analyses (pQA1 as an internal standard) of tumor-derived β2-microglobulin steady-state mRNA levels in metastatic lesions (bone and soft tissues) in the nude mice. The data show coamplification of the human-specific pQA1 internal standard and the cellular DNA during competitive PCR. There was no amplification of cDNA from control mice that were not inoculated with tumor cells.
Conversely, we investigated the application of competitive PCR analyses of stromal cell-derived β2-microglobulin steady-state mRNA levels in a metastatic lesion in nude mice that were intracardiacally (ic) injected with human tumor cells (MDA-MB-231 breast cancer cells). Competitive PCR analysis of β2-microglobulin mRNA levels in 4-fold serial dilutions of a bone metastasis sample is depicted in Fig. 6 (right panel). The cDNA derived from the bone metastasis or cDNA of in vitro cultured MDA-MB-231 cells were coamplified (32 cycles) with 5 serial 1:4 dilutions of the mouse-specific internal standard pMUS. The data show that steady-state mRNA levels for mouse β2-microglobulin could be quantified in metastatic bone lesions, which consist of a mixture of human and mouse cells, whereas no PCR amplification of cDNA took place with human cDNA (from cultured MDA-MB-231 cells). Therefore, tumor- and host-derived gene transcripts can be measured separately in a species-specific and quantitative manner, thus offering the possibilities to investigate gene expression at different metastatic sites.
Gene expression at different metastatic lesions/tumor-derived mRNA transcripts
To correct for any variation in RNA content and cDNA synthesis between the different preparations, samples were equalized based on β2-microglobulin.(28, 30, 31) Five nanograms of cDNA was coamplified with human-specific β2-microglobulin primers in the presence of the human-specific internal standard pQA1.
After quantitative analyses (see examples in Figs. 1 and 5), all tissue samples were normalized for the amount of tumor cells for further analyses of tumor-derived progression factors (Fig. 7). Subsequently, semiquantitative analyses were performed on the expression of tumor cDNAs at a fixed number of PCR. Figure 7A depicts the expression of PTHrP by the metastatic breast cancer cells in bone and soft tissues (lung and brain). The cancer cells express detectable levels of PTHrP, which is believed to enhance locally osteoclastic bone resorption. Interestingly, PTHrP expression by the breast cancer cells was strongly up-regulated in bone when compared with soft tissue (brain and lung; p ≤ 0.001).
In bone and soft tissue metastatic lesions from MDA-MB-231 breast cancer cells steady-state mRNA transcripts of M-CSF were detected (Fig. 7B). However, the level of expression of M-CSF by the cancer cells differed among the metastatic sites. In bone metastases, steady-state mRNA levels were significantly higher than in the soft tissues (p ≤ 0.05).
Figure 8 depicts an example of coamplification of cDNA (80 ng) and pQA1 (197 × 103 copies) for human M-CSF over a wide PCR cycle range (30-38 cycles) in the presence of [α-32P]dCTP. No differences in amplification efficiencies between the human internal standard pQA1 and cDNA samples for were found for M-CSF, thus validating the use of competitive PCR procedures for tumor-derived M-CSF gene transcript quantitation. Figure 9 shows an example of quantitative detection of tumor-derived M-CSF by coamplifying the internal standard pQA1 with a cDNA sample obtained from a skeletal metastasis using human-specific M-CSF.
In most of the tumors, expression of VEGFs is a prerequisite for the induction of angiogenesis, which is needed for tumor growth and metastasis. In the similar set of bone and soft tissue metastases, MDA-MB-231 cells expressed different VEGF types, VEGF-A, -B, and -C, in both bone and soft tissue metastatic lesions (Fig. 10). Strikingly, the expression of VEGF-A(121,165,189), -B, and, -C was up-regulated in the bone/bone marrow microenvironment in comparison to soft tissues (Figs. 10A-10C). Under similar experimental conditions, no detectable levels of VEGF-D were observed (Fig. 10D). In line with the observed VEGF-A expression at the transcriptional level, accumulation of VEGF165 protein into a conditioned media of MDA-MB-231 cells (6.514 pg/106 cancer cells/48 h) was found.
Gene expression at different metastatic lesions: mouse-derived mRNA transcripts
Bone and soft tissue metastasis samples were used to study the steady-state mRNA expression patterns of tumor-associated murine stromal cells. For this, gene expression of the endothelial marker PECAM-1 and four different VEGF-receptors by tumor-associated murine stromal cells was investigated after normalization with mouse-specific competitive PCR procedures for the mouse β2-microglobulin as described previously.
PECAM-1, a marker of endothelial cells in sprouting blood vessels,(32) was expressed strongly in all tissues harboring MDA-MB-231 breast cancer cells (Fig. 11A). In line with the immunohistochemical data presented in Fig. 4C, surgically excised bone metastases strongly express PECAM-1, indicating that this endothelial marker is expressed by the tumor-associated endothelial cells. Moreover, receptors for VEGF (flt-1, flk-1, flt-4, and NP-1) are expressed in all samples containing the metastatic lesions (Fig. 11B). Therefore, our data suggest that markers of sprouting vessels can be detected in metastatic lesions and that the murine stromal cells, in particular endothelial cells, express receptors for different VEGFs.
In the present study competitive species-specific PCR techniques were used for the detection and quantitation of human tumor cells in a mouse model of experimental metastasis with the human housekeeping gene (β2-microglobulin) as a target sequence. Accurate quantitation of different steady-state mRNA transcripts in metastatic lesions in vivo also could be accomplished with competitive PCR and involved addition of a known amount of PCR-amplifiable internal standard into a cDNA sample and then amplifying the standard and target DNAs in the same reaction. We have used internal standards (pQA1 and pMUS) that share the same PCR primer binding sites with the endogenous target sequence (e.g., β2-microglobulin), but whose PCR products are distinguishable from that of the targets because of differences in size.(28, 30, 31) Moreover, successful competitive PCR requires that the internal standards be amplified at the same rate as the endogenous cDNA target. We found that amplification proceeded with the same efficiency for both the cDNA and the internal standards (pMUS and pQA1) for β2-microglobulin and other transcripts, for example, human M-CSF; that is, the ratio between the amplification efficiency of cDNA and internal standard in the exponential phase was approximately one. Provided that amplification efficiencies of the internal standard and cDNA do not differ among the gene transcripts of interest, competitive PCR with a species-specific internal standard and corresponding primers circumvents potential errors that may be introduced by sample-to-sample and loading differences. In addition, the species-specific oligonucleotide primers in PCR procedures provide the means to study separately the involvement of tumor and stromal cell gene transcripts in experimental tumor invasion and metastasis.
We show for the first time in experimental conditions differences in the level of expression of bone-acting cytokines (PTHrP and M-CSF) and angiogenic factors (VEGF types) by the tumor cells in metastatic lesions in bone and soft tissues. For instance, the expression of PTHrP by breast cancer cells is up-regulated strongly and significantly in skeletal metastases when compared with soft tissues. Our data provide the first direct experimental evidence that PTHrP is up-regulated in the bone microenvironment and support the clinical observations that PTHrP also is expressed strongly in the majority of bone metastases (73-92%) from breast cancer, whereas only 17-20% of the metastases in nonskeletal sites express PTHrP.(33, 34) It is now clear that PTHrP is a major stimulator of osteoclastic bone resorption in bone metastatic lesions from breast cancer, which explains the presence of increased numbers of TRAP+ osteoclasts surrounding the skeletal lesions.(15, 18) Recently, Yin and coworkers(20) showed that transforming growth factor β (TGF-β), which is released and activated from the bone matrix during bone resorption, triggers breast cancer cells via the TGF-β receptors to stimulate PTHrP production leading, in turn, to osteoclastic bone resorption that is increased locally. Our findings are consistent with current hypotheses concerning locally enhanced PTHrP expression and subsequent enhanced tumor-induced osteoclastic bone resorption in metastatic breast cancer. In addition to PTHrP, there was increased expression of M-CSF, an essential factor for osteoclast development,(35–37) in metastatic bone lesions.
Tumors require a blood supply for growth, expansion, and metastasis.(38–42) The density of microvessels in primary breast tumors and nodal metastases are correlated strongly with survival.(43–50) The process of new capillary formation from preexisting vessels, angiogenesis, or neovascularization is regulated tightly by inducers and inhibitors of endothelial proliferation, migration, and differentiation.(40, 41) In malignant tumors this occurs when the balance is shifted in favor of angiogenesis either by increased levels of angiogenic inducers or decreased levels of angiogenic inhibitors. VEGFs are specific endothelial cell mitogens, inducers of angiogenesis,(51, 52) and act as survival factors for endothelium.(53) Our immunohistochemical and PCR-based studies show the presence of many PECAM-1-positive blood vessels in metastatic lesions, using PECAM-1 antibodies and mouse-specific oligonucleotide primer combinations for PECAM-1(CD31) in PCR reactions with cDNA obtained from bone and soft tissue metastases. Moreover, here, we show that MDA-MB-231 cells strongly express VEGF-A, -B, and -C in various metastatic lesions, including bone. The steady-state mRNA levels for the VEGF-A splice variants VEGF121, VEGF165, and VEGF189 and VEGF-B in cancer cells were all significantly elevated in bone metastases compared with soft tissue metastases. In the same in vivo bone and soft tissue samples, various host-derived VEGF receptors were identified: VEGFR-1 (flt-1), VEGFR-2 (flk-1/KDR), VEGFR-3 (flt-4), and VEGF165R (neuropilins). VEGF-A isoforms, which are expressed by the breast cancer cells in vivo, may exert their actions through binding to receptors VEGFR-1 (flt-1), VEGFR-2 (flk-1/KDR), and VEGF165R (neuropilin), which also are expressed in the same metastatic lesions presumably by the PECAM-1-positive endothelial cells.(51, 52, 54–56) Moreover, tumor-derived VEGF-B, which is expressed highly in the metastases, can bind to VEGFR-1(57, 58) whereas VEGF-C binds to VEGFR-3 (flt-4).(59, 60) Collectively, our findings provide strong evidence for the existence of angiogenic factors that are capable of eliciting neovascularization in these tumors.
It may be that enhanced VEGF expression by breast cancer cells in bone microenvironment contributes to their successful growth. Thus, the bone/bone marrow compartment may provide factors capable of stimulating PTHrP(20) and certain VEGF types by the breast cancer cells that eventually will favor tumor growth in bone. At present, the mechanisms underlying the differential expression of the VEGF-A and -B types by the breast cancer cells are not known. Interestingly, TGF-β was reported previously to induce the expression of VEGF-A in breast cancer cells.(61) These data suggest that—besides enhanced expression of PTHrP by breast cancer cells in the bone microenvironment because of activation of latent bone matrix-bound TGF-β by osteoclasts(20)—comparable mechanisms of TGF-β activation also may account for the observed enhanced production of VEGF by the tumor cells in bone. Clearly, more research is warranted to address these issues.
In conclusion, species-specific primer combinations in (competitive) PCR reactions can be used to measure tumor load in vivo, facilitating the study of tumor-stroma interactions in cancer-induced bone resorption, tumor invasion, metastasis, and tumor-induced angiogenesis. Our PCR method is based on selective (species specific) hybridization of oligonucleotide primers to target sequences and therefore may have a much broader application. The data obtained with these PCR-based procedures and their application in xenografted tumor may not only generate information regarding tumor metastatic behavior, but also may be of value in the design of intervention studies with antitumor and antiangiogenic agents.
This work is supported by a grant from the Dutch Royal Academy of Sciences (KNAW) and the Dutch Cancer Society (NKB).