|Home | About | Journals | Submit | Contact Us | Français|
Growth and differentiation of colonic epithelium are regulated in the three-dimensional (3D) physiological architecture, colonic crypt, and deregulation of 3D interactions is involved in tumorigenesis. Cell-based 3D culture systems provide a suitable approach bridging the gap between two-dimensional (2D) culture and animal models. KRAS mutations are found at high frequencies in human colorectal cancer (CRC); however, KRAS-targeted cancer therapy has not been developed. Here, we have established a 3D cell culture model resembling the colonic crypt by use of HKe3 cells, human CRC HCT116 cells disrupted at activated KRAS. In this 3D colonic crypt model, HKe3 cells showed the features of time course-dependent transit-amplifying and terminal-differentiated stages, which are characteristic of normal colonic crypt. On the basis of the features of HCT116 cells, activated KRAS inhibited normal cell polarity and apoptosis in 3D culture. The expression of DNA repair-related tumor suppressor genes including TP53, BRCA1, BRCA2, and EXO-1 was markedly suppressed by activated KRAS in 3D culture but not in 2D culture. These results together suggest that activated KRAS plays critical roles in the accumulation of genetic alterations through inhibition of DNA repair genes and apoptosis and that this 3D culture model will provide a useful tool for investigating the molecular mechanisms of CRC development.
Both cell-cell and cell-extracellular matrix interactions are critically involved in developmental programs and provide three-dimensional (3D) architectures in vivo , and deregulations of these interactions are frequently observed in cancer . Because cells grown in traditional flat two-dimensional (2D) culture often differ in morphology, cell-cell interactions, and differentiation from the cells grown in physiological 3D environments [3,4], cell-based in vitro 3D culture systems provide a suitable approach that can bridge the gap between traditional 2D cell culture and animal models [2,5]. Human cancers are derived from epithelial tissues characterized by specific cellular architectures including epithelial cell-cell junctions, which allow the separation of apical and basolateral membranes. This apical-basal cell polarity is crucial in normal cell functions, and loss of cell polarity is a critical step in tumorigenesis [6–8].
Colonic epithelium makes a 3D structure called colonic crypt, where epithelial cells migrate upward through the transit-amplifying (TA) zone in the lower-to-middle region of the crypt, before becoming terminally differentiated (TD), and are eventually shed into the lumen . Because most cell proliferation takes place in the TA region and terminal differentiation occurs distal to this region , a 3D culture model resembling colonic crypt should be needed for understanding the colorectal tumorigenesis in vivo.
Tumorigenesis is a multistep process in which genetic alterations accumulate, culminating in neoplastic phenotype [10–12]. In the adenomacarcinoma sequence model of colorectal cancer (CRC) cell, adenomatous polyposis coli gene mutations occur in the first step, followed by particular mutations of oncogenes and tumor suppressor genes, along with genetic instability [11,13–15]. KRAS mutations are frequently observed not only in CRC but also in colorectal adenomas and in pancreatic and lung cancers [16–18]. Oncogenic mutations in RAS are invariably point mutations that either interfere with Ras GAP binding or directly disrupt Ras GTPase activity, locking RAS in a constitutively active form . However, KRAS-targeted therapy has not been clinically developed, and patients with CRC bearing activated KRAS did not benefit from cetuximab, a monoclonal antibody against the epidermal growth factor receptor . Elucidation of the precise molecular mechanisms of activated KRAS in vivo should be needed for the design and development of cancer therapies.
We and colleagues reported much about activated KRAS functions through comparing human CRC HCT116 cells and HKe3 cells, HCT116 cells disrupted at activated KRAS [19–22]. Here, we have established in vitro a 3D culture model resembling colonic crypt using HKe3 cells and elucidated the relation between activated KRAS and 3D architectures and functions comparing with HCT116 cells.
2D culture for HCT116 cells, HKe3 cells, and e3-MKRas#14 cells and for HKe3-derived stable transfectants expressing activated KRAS was done as described previously [19,23]. A total of 5 x 105 cells were cultured in 10-cm culture dishes (Nunc, Rochester, NY) . For the 3D cell culture, 8 x 103 cells were cultured in LabTek 8-chamber slides (Nunc) using Matrigel, a reconstituted basement membrane (BD Biosciences, San Jose, CA), as described previously  and according to the protocol online at http://brugge.med.harvard.edu/. Briefly, Matrigel was thawed overnight at 4°C, and 25 µl per well was spread over the surface of well on ice, followed by incubation at 37°C for 1 hour. Cells were seeded in the medium containing 2% (vol/vol) Matrigel. Half of the medium was replaced with fresh medium containing 2% Matrigel every 3 days.
Immunofluorescence experiment was done as described previously . Briefly, cells in the 3D culture were fixed using 3.7% formaldehyde in PBS and incubated at room temperature for 20 minutes. Permeabilization was done using 0.05% saponin, and the fixed cells were incubated at room temperature for 30 minutes. Blocking was done by PBS containing 10% normal goat serum (Jackson Immuno-Research, West Grove, PA) at 4°C for 1 hour, followed by incubation of primary antibodies in blocking buffer for 15 hours at 4°C. The wells were washed three times with PBS containing 0.5% NP-40 and twice with PBS, followed by incubation of Alexa-conjugated secondary antibodies (Invitrogen, Carlsbad, CA) diluted in blocking buffer to 1:200 for 1 hour at room temperature. The wells were rinsed as described here, and 50 µl of ProLong Gold antifade reagent (Invitrogen) was added. Primary antibodies used were cleaved caspase-3 (5A1; Cell Signaling Technology, Beverly, MA), laminin-5 (D4B5; Millipore, Billerica, MA), ZO-1 (1/ZO-1; BD Biosciences, San Jose, CA), and Ki-67 (Thermo Scientific, Rockford, IL). Alexa Fluor 568-conjugated to phalloidin (Invitrogen) and 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, St Louis, MO) were used for staining of F-actin and DNA, respectively. For the examination of 3D structures, TCS-SP5 Laser Scanning confocal microscopy (Leica, Wetzlar, Germany) was used. The figures show the representative data from three independent experiments.
HCT116 and HKe3 cells were cultured for 6 days in Matrigel and were stained by anti-cleaved caspase-3 antibody and DAPI at day 6. Cleaved caspase-3-positive cells were counted in a cross section of the 3D culture at maximum diameter, and the 3D structure containing more than two positive cells were defined as apoptotic. The 3D structures for HCT116 and HKe3 cells were analyzed in three independent experiments, and the average ratio of apoptotic 3D structure was calculated.
HCT116 and HKe3 cells were cultured for 6 days in Matrigel and were stained by anti-Ki-67 antibody and DAPI at the indicated day from day 1 to day 6. The ratio of Ki-67-positive cells in the total cells contacting Matrigel in the cross section of 3D structures at maximum diameter was calculated. Similar results were obtained from independent experiments.
Total RNA from cells cultured in the 2D or 3D cultures were extracted as described previously . Gene expression arrays were done using Human Genome U133 Plus 2.0 Array 6800 GeneChips (Affymetrix, Santa Clara, CA) and were analyzed by GeneSpring v7.3 software (Agilent Technologies, Santa Clara, CA) as previously described .
Real-time quantitative reverse transcription-polymerase chain reaction was done by Perfect Real-time Support System (Takara Bio, Inc, Shiga, Japan) for BRCA1 (HA113973), BRCA2 (HA119560), EXO1 (HA104791), TP53 (HA092869), CTNNB1 (HA115708), and ACTB (HA067803). Complementary DNA were synthesized from total RNA as described previously . Assays were performed on QuantiTect SYBR Green PCR Kits (Qiagen, Valencia, CA) and ABI PRISM 7900HT (Applied Biosystems, Foster City, CA), and ACTB values were used for normalization. The normalized values were quantified by the ΔΔCt method and presented as means ± SDs for triplicate samples as described previously [27,28]. Each relative expression unit (REU) was determined by the REU of HCT116 in 2D culture as 1.0.
Western blot analysis was performed as described previously . Cells grown in 3D culture were harvested by cold PBS with 5mM EDTA, and Matrigel was separated by centrifugation, followed by cell lysis in RIPA buffer. Lysates were spun at 12,000 rpm at 4°C for 10 minutes, and the supernatants were quantified using Bradford ULTRA (Expedeon, San Diego, CA). The antibodies used were p53 (DO-1; Santa Cruz Biotechnology, Santa Cruz, CA) and ERK1 (K-23; Santa Cruz Biotechnology). The figures show the representative blot from three independent experiments. Quantitative Western blot analysis of p53 was performed using a measurement module (BZ-H1M; Keyence, Osaka, Japan) for obtaining integration values of each blot (n = 3). ERK1 intensity was used as the standard. The relative intensity of each signal was normalized by the signal intensity in HKe3 cells grown in 3D culture as 1.0.
Data were presented as mean ± SDs of means of triplicate samples. Statistical analysis was performed with an unpaired Student's t-test to compare the means of multiple groups. Differences at P < .05 are considered to be statistically significant.
To elucidate the precise functions of activated KRAS in colorectal tumorigenesis, we compared the 3D structures between HKe3 and HCT116 cells grown in 3D cell culture using Matrigel containing laminin and collagen. To address the KRAS functions in 3D culture, morphological structure and apical-basal cell polarity were examined. Firstly, we examined the distribution of laminin V and F-actin, markers for basement membrane and apical membrane, respectively, in the HCT116 and HKe3 cells grown in 3D culture at day 6. Serial cross-sectional images from bottom to top showed that HKe3 cells formed a single layer of polarized cells and luminal cavity, where laminin V was placed at the basal region and F-actin was localized in the apical region (Figure 1A, from left to right). Conversely, localizations of laminin V and F-actin for HCT116 cells were disorganized, and luminal cavity was not observed (Figure 1A), indicating the loss of apical-basal cell polarity in HCT116 cells in the 3D culture. To further address cell polarity in the 3D culture, we examined the distribution of ZO-1, a tight junction marker. ZO-1 signals were localized at the border between the apical and basolateral membranes adjacent to luminal cavity of HKe3 cells but not in the HCT116 cells (Figure 1B), and distributions of ZO-1 and F-actin were colocalized in the HKe3 cells in the 3D culture (Figure W1), suggestive of the existence of normal apical-basal cell polarity in the HKe3 cells. Furthermore, e3-MKRas#14 cells, HKe3-derived stable transfectants expressing activated KRAS , showed impaired cell polarity or luminal cavity formation like HCT116 cells (Figure W2, A and B). These results together suggest that activated KRAS disturbs cell polarity and luminal cavity formation in the 3D culture.
Apoptosis is induced by detaching from the surrounding extracellular matrix, called anoikis, in the luminal epithelial cells in vivo . To address whether activated KRAS influences apoptosis in the 3D culture, we examined the apoptosis activity in the HCT116 and HKe3 cells by detecting the cleaved caspase-3 signals. The ratios of apoptotic 3D structures for HCT116 and HKe3 cells were 16.9% and 80.5%, respectively (*P < .001; Figure 1C), indicating that apoptosis activity is present for luminal cavity epithelium in the 3D culture of HKe3 cells and is inhibited in the HCT116 cells. Furthermore, apoptosis activity in e3-MKRas#14 cells was inhibited like HCT116 cells (Figure W2C). These results suggest that activated KRAS suppresses apoptosis in the 3D culture.
Continuous proliferation is one of the features of cancer cells in the traditional 2D culture. A proliferation marker Ki-67 signal for HKe3 cells grown in 2D culture was expectedly detected only in the cells located at the outer region but not at the inner region (Figure 2A). Conversely, the Ki-67 signal for the HCT116 cells grown in 2D culture was observed in both the inner and outer regions (Figure 2A). Together, these results suggest that activated KRAS suppresses the contact inhibition of cell growth in the 2D culture.
To address whether or how activated KRAS influences cell proliferation in the 3D cell culture, we examined the time course-dependent proliferation of the cells grown in the 3D culture for 6 days. Ki-67 signals were evidently detected both in the HKe3 and HCT116 3D structures from day 1 to day 3, whereas Ki-67 signals were suddenly decreased at day 4 not only in the HKe3 cells but also in the HCT116 cells (Figure 2B). Ki-67-positive cells in the 3D structures of HCT116 and HKe3 were similarly decreased time course-dependently, 97.1% and 97.6% at day 2, 80.8% and 91.9% at day 3, 64.7% and 54.6% at day 4, 34.4% and 51.2% at day 5, and 11.0% and 11.8% at day 6, respectively (Figure 2C). A significant difference in the ratio of Ki-67-positive cells between the HCT116 and HKe3 cells was not detected at any time point (P > .05; Figure 2C). These results, together, suggest that activated KRAS is rather involved in the inhibition of apoptosis than in cell proliferation. Of note is the decreased proliferation and luminal cavity formation in HKe3 cells in the 3D culture at day 4, as TA cells in the colonic crypt in vivo differentiate to TD cells at day 4 , suggesting that HKe3 cells in this 3D culture will be applied as a colonic crypt model.
To explore the differentially expressed genes regulated by activated KRAS in the 3D culture, we performed expression microarrays using HCT116 and HKe3 cells grown in the 2D or 3D culture for 6 days. Differentially expressed genes were cut off by three-fold up or three-fold down, and these are shown as a Venn diagram (Figure 3A), and the genes detected are listed (Tables W1–W3). Of note is the existence of 3D-specific differentially expressed genes. The numbers of 3D-specific three-fold up- and three-fold down-expressed genes were 328 and 1397, respectively (Figure 3A; Table W4).
To address the functions of 3D-specific downregulated genes by activated KRAS, we performed gene ontology (GO) analysis for the genes (Table W4). The significant GO for the 3D-specific downregulated genes by activated KRAS was classified into the GOs such as “DNA metabolism” (with a P value of 7.6e-24), “DNA repair” (P = 1.4e-19), “response to DNA damage stimulus” (P = 3.8e-18), and “cell cycle checkpoint” (P = 1.9e-16; Table W4B). However, the most significant GO for the 3D-specific upregulated genes by activated KRAS was “entrainment of circadian clock” with a P value of 1.1e-4 (data not shown). From the viewpoint of tumorigenesis and statistical significance, 3D-specific downregulated genes by activated KRAS are of great interest. The 3D-specific downregulated genes, including BRCA1 [12,31–35], BRCA2 [12,31,33,34,36,37], EXO1 (exonuclease-1) [38–40], and TP53 (p53) [12,34,41,42], were commonly listed among the GO categories related to “DNA metabolism,” “cell cycle,” “DNA repair,” and “response to DNA damage stimulus” (Table W4C), and these four genes are known as tumor suppressor genes [1,33,38]. Interestingly, in the 2D culture, expressions of these four genes in HCT116 cells, compared with those in HKe3 cells, were 4.15-, 4.27-, 4.60- and 1.19-fold upregulated, respectively (Table W4, A and D). Furthermore, 62 of the 3D-specific downregulated 72 genes classified as “DNA metabolism” were rather upregulated in the 2D culture, indicating the opposite expression profiles for these genes between 2D and 3D cultures (Table W4D).
We performed real-time quantitative reverse transcription-polymerase chain reaction to validate the microarray results of BRCA1, BRCA2, EXO-1, and TP53. In HKe3 cells for the 3D culture, the expression levels of BRCA1, BRCA2, EXO-1, and TP53 were increased to 11.22, 6.38, 6.20, and 9.31 times, respectively (*P < .001) compared with HCT116 cells in the 3D culture (Figure 3B), suggesting that activated KRAS in the 3D culture exactly inhibits the messenger RNA (mRNA) expressions of these four genes. In the 2D culture, the expression levels of BRCA1, BRCA2, and EXO-1 in HKe3 cells, compared with those in HCT116 cells, were decreased to 0.21, 0.28, and 0.28 times, respectively (P < .001), whereas TP53 was not significantly changed in the 2D culture (1.09 times, P = .31; Figure 3B).
These four gene expressions in HCT116 cells in the 3D culture, compared with those in the 2D culture, were decreased to 0.05 (BRCA1), 0.05 (BRCA2), 0.09 (EXO1), and 0.04 (TP53) times, respectively (P < .001), whereas these gene expressions in HKe3 cells in the 3D culture, compared with those in the 2D culture, were increased to 8.11 (BRCA1), 3.84 (BRCA2), 2.03 (EXO-1), and 1.28 (TP53) times, respectively (P < .05). Together, these results suggest the critical roles for activated KRAS in the regulation of expression of these tumor suppressor genes.
Considering the importance of p53 in tumorigenesis, we examined p53 protein level by Western blot. Although the p53 protein level in HCT116 cells in the 2D culture, compared with HKe3 cells in the 2D culture, was slightly increased to 1.23 times (**P < .005), p53 protein level in HCT116 cells in the 3D culture, compared with HKe3 cells in the 3D culture, was decreased to 0.75 times (*P < .05; Figure 3C). Furthermore, p53 expression in HKe3 cells in the 3D culture, compared with that in the 2D culture, was increased to 1.43 times (***P < .0001; Figure 3C). Together, these results suggest that the p53 protein level is increased in the 3D culture through loss of activated KRAS.
HKe3 cells, human CRC HCT116 cells disrupted at activated KRAS, formed a colonic cryptlike 3D structure and showed time course-dependent proliferation and differentiation under the 3D microenvironment (Figure 4A). In the 3D cell culture for HKe3 cells, HKe3 cells divided like TA cells in the colonic crypt from day 1 to day 3 and displayed the polarized structures like TD cells after day 4, followed by apoptosis in luminal epithelium (Figures 1 and and4A).4A). In contrast, HCT116 cells having activated KRAS and HKe3-derived e3-MKRas#14 cells expressing activated KRAS lost apical-basal cell polarity and apoptosis/anoikis (Figures 1 and and4B;4B; Figure W2). After day 4, proliferative cells detected by Ki-67 signals were evidently decreased both in HCT116 and in HKe3 cells (Figure 2, B and C). All these results suggest that HKe3 cells in the 3D culture show the features of colonic crypt, and this 3D cell culture system will be useful for elucidating the molecular mechanisms of CRC and novel functions for activated KRAS in vivo.
From the viewpoint of the regulated genes by activated KRAS, DNA repair-related tumor suppressor genes, including TP53 [1,12,34,42], BRCA1 [12,31–35], BRCA2 [12,31,33,34,36,37], and EXO-1 [38–40], were downregulated in a 3D-specific manner, suggesting that activated KRAS plays critical roles in the accumulation of genetic alterations through the suppression of DNA repair genes (Figure 4B), and this suppression leads to the disruption of a barrier to tumor progression in precancerous lesions .
Because colonic crypts in vivo are localized in an acidic microenvironment and are continuously exposed to oxidative stress, colonic crypt cells possess higher DNA repair activity [44,45]. Cells grown in the 3D culture here might reflect microenvironment in vivo to react against acidic condition, culminating in an induced expression of DNA repair genes and activated KRAS might inhibit DNA repair activity, leading to the accumulation of genetic alterations. The reason why DNA repair activity is remarkable in the 3D culture, but not in the 2D culture, may reflect the difference between cellular conditions induced by microenvironments of the 2D and 3D cultures. Thus, microenvironmental conditions should be much taken into account for elucidating molecular mechanisms of cancer. In this study, we focused on interactions of epithelial cells and basement membranes in the 3D microenvironment; therefore, evaluation of other microenvironmental conditions including stroma cells should be needed for a better understanding of molecular mechanisms underlying the development of CRC in vivo.
The down-regulation of DNA repair-related tumor suppressor genes was evidently detected at the mRNA level (Figure 3B), whereas the p53 protein level was not completely concomitant with the mRNA level (Figure 3C), suggesting that particular mechanisms do exist in the regulation of p53, such as protein degradation and translation regulated by microRNA . Further examination of the differentially expressed genes including microRNA and signaling on this 3D culture system will lead to a better understanding of CRC development.
Another issue in this study is that the molecular mechanisms of tumorigenesis should be reevaluated in the 3D microenvironment . Most of the downregulated genes classified as “DNA metabolism” by activated KRAS in the 3D culture were rather upregulated in the 2D culture, i.e., opposite expression profiles between the 2D and 3D cultures exist (Table W4D). Researchers should adequately consider cell culture conditions; otherwise, molecular mechanisms of the tumori-genesis in vivo may be misunderstood.
We have established a 3D colonic crypt model using HKe3 cells, HCT116 cells disrupted at activated KRAS, and demonstrated that activated KRAS is involved in deregulation of apical-basal cell polarity, luminal cavity formation with apoptosis, and 3D-specific suppression of DNA repair-related tumor suppressor genes including TP53, BRCA1, BRCA2, and EXO-1. Activated KRAS plays critical roles in the accumulation of genetic alterations through the inhibition of DNA repair genes and apoptosis, resulting in the disruption of the barrier-to-tumor progression. Our 3D colonic crypt model will provide a useful tool for investigating the molecular mechanisms of CRC development in vivo and design of CRC therapies.
The authors thank T. Danno and Y. Hirose for technical assistance.