Clin. Exp. Metastasis, 1992, 10, 25-31
Metastatic potential of human colon cancer cell lines: relationship to cellular differentiation and carcinoembryonic antigen production Hans E. Wagner, Carol Ann Toth, Glenn D. Steele Jr and Peter Thomas Laboratory of Cancer Biology, Departm6nt of Surgery, New England Deaconess Hospital and Harvard Medical School, Boston, MA 02115, USA (Received 23 September 1991; accepted 25 September 1991)
The relationship between cellular differentiation and carcinoembryonic antigen (CEA) production by human colorectal tumor cells and their ability to form hepatic metastases was studied. Eight human colon cancer cell lines were injected into athymic mice using different routes of administration to characterize their metastatic potential. The four poorly differentiated, non or low CEA producing cell lines were poorly metastatic to the liver after intrasplenic injection. After intraperitoneal implantation the same cell lines were highly tumorigenic, and subsequently metastastic to the liver. In contrast, the four moderate to well-differentiated cell lines that produced moderate to high levels of CEA were highly metastatic to the liver following intrasplenic injection. After intraperitoneal implantation they were less tumorigenic, and metastatic to the liver. We conclude that in this system poorly differentiated non or low CEA producing colorectal cell lines have a lower metastatic capacity compared to the well-differentiated high CEA producing colorectal cell lines. These data correlate directly with the pattern of metastatic spread and clinical course observed in patients with these tumors, suggesting that degree of differentiation and level of CEA production may play a role in development of site-specific metastases. Keywords: human colorectal cancer, nude mouse, carcinoembryonic antigen, differentiation, metastasis, liver and lung
Introduction Colorectal cancer is the second leading cause of cancer death in the USA . The 5-year survival rate for colorectal cancer patients is about 40-50%. Half these patients will recur regionally and up to 80% will have distant metastases primarily in the liver (70%) and lung (25%). Prognosis is influenced by the extent of the disease at diagnosis and the differentiation of the tumor . Patients with poorly-differentiated carcinomas generally have a poorer prognosis due to several factors, including advanced stage of disease at initial diagnosis and the higher rate of local and distant Address correspondence to: Peter Thomas, New England Deaconess Hospital, Laboratory of Cancer Biology, 50 Binney Street, Boston, MA 02115, USA. Tel: 617-732-9875.
~) 1992 Rapid Communications of Oxford Ltd
recurrences [3, 4]. Poorly differentiated colorectal cancer cells are more invasive both in vitro and in vivo [5, 6], resulting in larger numbers of circulating tumor cells from similar sized primaries. Large numbers of tumor cells are likely to overwhelm the defences of individual organ systems. This results in a pattern of spread typical for poorly differentiated adenocarcinoma of colon where multiple organs are involved . Because well-differentiated colorectal cancer cells tend to be less invasive, this results in fewer tumor cells entering the circulation. The factors that allow circulating tumor cells to colonize distant organs such as liver and lungs are not well understood. Carcinoembryonic antigen (CEA), a member of the immunoglobulin supergene family, is produced
Clinical & Experimental Metastasis Vol 10 No 1 25
H. E. Wagner et al.
by many colorectal cancer cells . CEA and related gene family members can function as intercellular adhesion molecules [9, 10]. Previous studies have shown that the growth potential of primary human colon carcinomas transplanted into athymic mice correlate with the patient's preoperative CEA level . High preoperative CEA levels have also been associated with poor prognosis for patients with Dukes stage B and C tumors . The highest CEA values are found in patients with advanced disease and in particular those with liver metastases . Serum CEA levels are influenced by numerous factors including tumor load, tumor differentiation and liver function . An additional factor to be considered is CEA production by the cells as we now have evidence that this may influence metastatic patterns. In this study, we examined the relationship of tumor differentiation and CEA production to the invasive and metastatic behaviour of human colon cancer cells in athymic nude mice. In addition we have studied the effect of various routes of tumor cell administration on the incidence and site of metastases.
Materials and methods Animals Six- to eight-week-old, male Swiss athymic nude mice were obtained from Taconic Farms (Germantown, NY) and housed in a pathogen-free and temperature-controlled room with a 12-h lightdark cycle. Surgical procedures were performed following sodium pentobarbital (50mg/kg i.p.) anesthesia. Cell lines The human colorectal cancer cell lines CCL 188, CCL 222, CCL 231, CCL 235 and HT-29 were obtained from the American Type Culture Collection (Rockville, MD). Clone A was provided by Dr D. Dexter (E. I. DuPont DeNemours & Co., Wilmington, DE). CX-1, isolated from the cell line HT-29, was obtained from Dr L. B. Chen (DanaFarber Cancer Institute, Boston, MA). The MIP-101 cell line was isolated in our laboratory from a poorly differentiated adenocarcinaoma of the colon . These cell lines were maintained in RPMI 1640 with 10% fetal calf serum (FCS) supplemented with L-glutamine, penicillin and streptomycin. All cell lines tested free of mycoplasma using the Hoechst stain (Sigma Chemical Co., St. Louis, MO).
Clinical & Experimental Metastasis Vol 10 No 1
For all experiments the cells were harvested mechanically by scraping from the dishes with a rubber policeman. Cells were examined microscopically to ensure they were present as single cell suspensions. Viability and degree of clumping was assessed by Trypan Blue dye exclusion. Tumor cell suspensions with greater than 90% viability and less than four cells per cluster were used for injections. Liver metastasis assay Experimental colonization of the liver was determined using the intrasplenic injection method described by Koslowski et al. . The spleen was exposed through a short incision and 2 x 10 6 tumor cells in 0.2 ml of phosphate-buffered saline (PBS) were slowly injected into the lower pole. The spleen was placed back into the abdominal cavity and the abdominal wall and skin closed by clips. The mice were killed and autopsied when sick or at 90 days after tumor cell injection. Intraperitoneal implantation assay Tumor cells (2 x 106) in 0.5 ml of PBS were injected intraperitoneally in a caudal-cranial direction as described by Morrissey et al. . Care was taken not to inject cells directly into any internal organ. The mice were killed and autopsied when sick or at 90 days following tumor cell injection. Intraperitoneal tumor burden was classified as follows: 0, no visible tumor; +, less than five nodules of any size; + + , 5-20 nodules including tumor growth in the liver hilus; + + +, peritoneal carcinomatosis including tumor growth on the diaphragm and the presence of ascites. CEA assay The level of CEA secretion by the tumor cell lines in vitro was determined by assaying supernatants from tumor cells in log phase growth for 10 days and expressed as ng/ml/106 cells. Circulating levels of CEA in tumor-bearing mice were determined from serum samples obtained by heart puncture at autopsy. CEA activity was assayed using the commercial CEA-Roche EIA test kit (a gift from Hoffman-LaRoche Inc., Nutley, N J). Histology At autopsy tissues were fixed in 10% buffered formalin, paraffin embedded, processed routinely and stained with hematoxylin and eosin. The differentiation of the tumors was based on the percentage of the tumor that formed gland-like structures. Cell lines that exhibited gland formation in
Metastatic potential of colorectal tumor cells less than one-third of the tumor were classified as poorly differentiated; moderately differentiated tumors had one-to two-thirds gland formation and well-differentiated tumors had two-thirds or greater gland formation. As an additional criterion of differentiation the tissue sections were stained for CEA using the immunoperoxidase procedure described previously . For the detection of CEA sections were deparaffinized with xylene and ethanol, rehydrated and incubated with 0.3% hydrogen peroxide for 10 min. The sections were incubated with the primary antibody D-14 (E-Z-EM, Inc., Westbury, NY) for 60min at ambient temperature, rinsed and incubated for a further 45 min with peroxidase conjugated rabbit anti-mouse IgG. The slides were washed and flooded with 3-amino9-ethylcarbazole and counterstained with hematoxylin. Ascites from an IgG producing mouse myeloma (MOPC-21) were used as a negative control antibody. Sections of known CEA producing colorectal carcinomas were used as positive controls.
Results This study examined the metastatic potential of eight human colorectal cell lines exhibiting varying degrees of differentiation and levels of CEA production (Table 1). All eight cell lines were 100% tumorigenic in the nude mouse after subcutaneous injection. As expected poorly differentiated tumors stained negatively for CEA while positive staining was seen in sections from the moderate or welldifferentiated tumors. CEA staining was particularly prominent in gland-forming areas of the tumors. The four more poorly differentiated cell lines (MIP-101, Clone A, CCL 222 and CCL 231) produced zero or very low amounts of CEA in
vitro. In contrast the moderate to well-differentiated cell lines (CX-1, HT-29, CCL 188 and CCL 235) produced moderate to high levels of CEA under our culture conditions. The amount of CEA staining in the xenografts correlated with the amount of CEA the tumor cell lines produced in culture (Table 1). The incidence of hepatic metastases and the serum CEA levels for nude mice injected intrasplenically with the eight colorectal cell lines is summarized in Table 2. All the cell lines produced splenic tumors. The poorly differentiated cell lines (MIP-101, Clone A and CCL 222) rarely gave liver metastasis (0-6%) after intrasplenic injection. CCL 231, a poor-moderately differentiated cell line produced hepatic metastases in 13% of the animals. In contrast, the moderate-well-differentiated cell lines colonized the liver in 46-93% of the mice. The difference between the incidence of hepatic metastases in the two pooled groups; moderate-well differentiated (51/77, 66%) and poormoderately differentiated (5/77, 7%) was statistically significant (P < 0.001, stratified MantelHaenszel test). The number of distinct nodes of tumor tissue in the liver of positive mice was measured. A correlation between the number of animals with tumor and the number of tumor nodes per liver was observed (Table 2). The extent of local tumor growth, incidence of liver metastases and circulating CEA levels in the intraperitoneally injected mice as summarized in Table 3. The four more poorly differentiated cell lines showed a high intraperitoneal tumor take rate with extensive peritoneal carcinomatosis. As a pooled group 89% of the mice had peritoneal tumor (39/44). The three poorly differentiated cell lines (MIP-101, Clone A and CCL 222) all showed a statistically significant increase (Fisher exact test,
Table 1. Characterization of human colon cancer cell lines Cell l i n e s
CEA production (ng/106 cells/10 days)
Tumorigenicity by subcutaneous injection (%)
MIP-101 Clone A CCL 222 CCL 231 CX-1 HT-29 CCL 188 CCL 235
Poor Poor Poor Poor-moderate Moderate Moderate-well Moderate-well Moderate-well
0 0 1.1 (+ 0.4) 0 34.9 (_+ 2.9) 6.8 (-+ 1.8) 51!9.0 (+ 865.0) 6.5 (_+ 4.5)
IO0 100 100 100 100 100 100 100
+1 +1 +2 +2
Clinical & Experimental Metastasis Vol 10 No 1 27
H. E. Wagner et al. Table 2. Incidence of liver metastases after intrasplenic injection Cell lines
Liver metastases [median (range)] ~
Serum CEA levels (ng/ml) b
MIP-101 Clone A CCL 222 CCL 231 CX-1 HT-29 CCL 188 CCL 235
1/18 [1,(1)] 1/24 [1,(1)1 0/12  3/23 [1,(1)] 14/15 [6,0-15)] 9/12 [2,(1-4)] 12/26 [3,(1-6)] 16/24 [3,0-7)]
0.5 (0.1-0.9) 0.4 (0-5.3) 1.4 (0.0-13.4) 0.0 (0.0.1) 324.0 (107-2300) 12.0 (1-830) 3.8 (0.0-133) 40.0 (0.0-1280)
a The median number of tumor nodules with the range (in parentheses) for animals with tumor. b CEA levels on serum obtained by heart puncture at sacrifice.
Table 3. Tumor growth and incidence of liver metastases after intraperitoneal injection Cell lines
No. of mice with tumor/group
No. of mice with liver metastases/group [median (range)] a
MIP-101 Clone A CCL 222 CCL 231 CX-1 HT-29 CCL 188 CCL 235
10/10 6/6 17/19 6/9 10/10 4/6 6/6 9/11
+++ +++ +++ +++ ++ ++ +++ +
8/10 [1,(1-2)1 5/6 [1,(1-2)] 8/19 [1,(1-2)] 3/9 [1,(1-2)] 5/10 [3,(1-3)] 1/6 [1,(1)] 1/6 [1,(1)1 0/11 
Serum CEA levels (ng/ml) b 3.9 (1.1-7.3) 6.8 (0.6-9.1) 9.8 (0.0-217) 800 ng/ml), biliary tract obstruction was present. Surprisingly, mice injected intraperitoneally with the non-CEA-producing cell lines, MIP-101 and Clone A, had detectable amounts of serum CEA, although these cell lines do not produce detectable serum C E A levels when grown subcutaneously.
Metastatic potential of colorectal tumor cells
Discussion In this study we have shown that the poorly differentiated human colorectal cancer cell lines were more tumorigenic than their well or moderately differentiated counterparts. This highly tumorigenic capacity and the greater capacity to invade [5,6], correlates with in vitro studies. Poorly differentiated cell lines showed a higher rate of movement through synthetic basement membranes, reduced laminin production and increased laminin receptor expression [5, 18]. Patients with poorly differentiated colorectal cancers generally have a worse prognosis due to a higher incidence of local and distant recurrence [3, 4]. The results obtained in this study using the experimental liver metastases assay, are consistent with these findings . The metastatic pattern of colorectal cancer cells is related to their degree of morphologic differentiation. Clinically poorly differentiated colorectal cancer cells are more invasive, grow rapidly and form metastases at multiple organ sites. Moderate to well-differentiated colorectal cells have a more specific pattern of spread, liver being the primary site for metastases followed by lung. This study also suggests that these cells have a selective advantage to form tumors in the liver and lungs. It is not the ability to grow in these organs that makes the poorly differentiated tumor clinically more aggressive, but its accelerated growth and increased capacity to invade. In the intrasplenic injection model the increased incidence of cellular differentiation correlated with an increased incidence of liver metastases. The well-differentiated, high CEA producing cell lines were more metastatic to the liver in this model than the poorly differentiated low CEA producing cell lines. This is counterbalanced in the intraperitoneal injection model by their reduced capacity to invade. Similar observations have been made with murine melanoma cell lines where the capacity to form lung metastases correlates positively with cellular differentiation . In patients with advanced colorectal cancer the two predominant sites of metastases are the liver and lung. Patients with metastases to these sites tend to have the highest CEA values . It is possible that well-differentiated colon cancer cells, secreting or expressing high CEA levels on their surface, may be more easily captured by the liver and lungs. Receptors for CEA similar to those on the Kupffer cell are also present on lung alveolar macrophages . It has been shown that liver
macrophages can bind colon cancer cells without causing cell death . Other reports have shown that unactivated Kupffer cells can be both cytostatic or cytotoxic to some tumor cells . The data presented here show an association between CEA production and metastases in the intrasplenic injection model. Other experimental studies have suggested that CEA may be directly involved in the metastasis of colorectal cancer cells to the liver and lungs . Preinjection of CEA into athymic mice prior to intrasplenic injection of human colon cancer cells enhanced the metastatic capability of several weakly metastatic cell lines [6, 23]. CEA has been shown to act as an intercellular adhesion molecule allowing tumor cells to cluster via homotypic interactions [9,24]. This could potentially enhance the metastatic potential. These observations suggest a possible role for CEA in colon cancer metastases formation . A potential mechanism for CEA involvement requires that CEA acts as an intercellular adhesion molecule, between tumor cells and hepatic macrophages. CEA is rapidly bound by surface proteins on Kupffer cells . It is possible that CEA can act as a bridging molecule between Kupffer cells and tumor cells. This would result in arrest of these cells in the hepatic sinusoid. In the present study serum CEA values in the nude mouse correlated with CEA production by the cell lines in vitro and with the tumor load. Very high serum CEA levels are observed in mice with obstructive jaundice due to the presence of liver metastases. This follows the clinical pattern in patients where the highest CEA levels are also associated with liver metastases and cholestasis . Interestingly, the poorly differentiated, nonCEA-producing cell lines (MIP-101 and Clone A) gave elevated serum CEA levels in the mice, when they grew intraperitoneally. These cells do not produce detectable serum CEA values when grown subcutaneously . It is possible that the implantation site selects for subsets of cells able to secret small amounts of CEA. This may be a factor in the production of hepatic and lung metastases following intraperitoneal implantation. Thus, the subpopulation of tumor cells with intrinsic properties to resist host defences and implant in the organs is most likely to metastasize . The athymic mouse provides a good experimental model to study hepatic implantation by human colorectal cancer cells [29, 30]. Orthotopic implantation of tumor cells in the cecum may best mimic the sequential steps of the metastatic cascade, but this technique is demanding in mice .
Clinical & Experimental Metastasis Vol 10 No 1 29
H. E. Wagner et al. Intravenous, as well as intrasplenic injections, allow the investigation of terminal steps (implantation, invasion) of the metastatic cascade . This eliminates the complications of invasion and release of t u m o r cells into the circulation, the effectiveness of which m a y rely on a completely different set of parameters. The alternative model of implantation in the peritoneal cavity allows the t u m o r cells to grow, invade and metastasize to distant sites. This enables an assessment of the properties required for both invasion and extravasion.
Acknowledegements We thank D r B. C. Wolf for pathology services, D r N o r m a n Z a m c h e c k for C E A assays, Vincent King for animal care and Michael Ierardi, Diane Harrington and Shayla Sharp for technical assistance. This study was supported by Grants CA44583 and CA44704 from the National Cancer Institute and B R S G funds from the New England Deaconess Hospital.
References 1. Silverberg E and Lubera J, 1987, Cancer Statistics. CA. 37, 2-19. 2. Saenz NC, Cady B, McDermott WV and Steele GD Jr, 1989, Experience with colorectal carcinoma metastatic to the liver. Surg. Clin. North America, 69, 361-370. 3. Olson RM, Perencevich NP, Malcolm AW, Chaffey JT and Wilson RE, 1980, Patterns of recurrence following curative resection of adenocarcinoma of the colon and rectum. Cancer, 45, 2969-2974. 4. Steinberg SM, Barwick KW and Stablein DM, 1986, Importance of tumor pathology and morphology in patients with surgically resected colon cancer. Cancer, 58, 1340-1345. 5. Daneker G, Piazza A, Steele G and Mercurio A, 1989, Relationship between extracellular matrix interactions and degree of differentiation in human colon carcinoma cell lines. Cancer Research, 49, 681-686. 6. Wagner H, Thomas P, Wolf B, Zamcheck N, Jessup J and Steele G, 1990, Characterization of the tumorigenic and metastatic potential of a poorly differentiated human colorectal cancer cell line. Invasion and Metastasis, 10, 253-266. 7. Willet C, Tepper J and Cohen A, 1984, Failure patterns following curative resection of colonic carcinoma. Annals of Surgery, 200, 685-692. 8. Boucher D, Cournoyer D, Stanners C and Fuks A,
Clinical & Experimental Metastasis Vol 10 No 1
1989, Studies on the control of gene expression of carcinoembryonic antigen family in human tissue. Cancer Research, 49, 847-852. Benchimol S, Fuks A, Jothy S, Beauchemin N, Shirota K and Stanners C, 1989, Carcinoembryonic antigen, a human tumor marker, functions as an intercellular adhesion molecule. Cell, 57, 327-334. Turbide C, Rojas M, Stanners C and Beauchemin N, 1991, A mouse carcinoembryonic antigen gene family member is a calcium dependent cell adhesion molecule. Journal of Biological Chemistry, 266, 309-315. Jessup JM, Giavazzi R, Campbell D, Cleary K, Morikawa K and Fidler I, 1988, Growth potential of human colorectal carcinomas in nude mice: association with the preoperative serum concentration of carcinoembryonic antigen in patients. Cancer Research, 48, 1689-1692. Goslin R, Steele G Jr, MacIntyre J, Mayer R, Sugarbaker P, Cleghorn K, Wilson R, and Zamcheck N, 1980, The use of preoperative plasma CEA levels for the stratification of patients after curative resection of colorectal cancers. Annals of Surgery, 192,747-751. Steele G Jr and Zamcheck N, 1985, The use of carcinoembryonic antigen in the clinical management of patients with colorectal cancer. Cancer Detection and Prevention, g, 421-427. Thomas P and Zamcheck N, 1983, Role of liver in clearance and excretion of circulating carcinoembryonic antigen (CEA). Digestive Diseases and Sciences, 28,216-224. Niles R, Wilhelm S, Steele G, Burke B, Christensen T, Dexter D, O'Brien M, Thomas P and Zamcheck N, 1987, Isolation and characterization of an undifferentiated human colon carcinomas cell line (MIP101). Cancer Investigations, 5,545-552. Koslowski J, Fidler I, Campbell D, Xu Z, Kaighn M and Hart I, 1984, Metastatic behaviour of human tumor cell lines grown in the nude mouse. Cancer Research, 44, 3522-3529. Morrisey L, Sidky Y and Auerbach R, 1980, Regional differences in the growth of tumor cells injected intraperitoneally into syngeneic adult mice. Cancer Research, 40, 2197-2201. Daneker G, Mercurio A, Guerra L, Wolf B, Salem R, Bagli D and Steele G, 1987, Laminin expression in colorectal carcinomas varying in degree of differentiation. Archives of Surgery, 122, 1470-1474. Bennet D, Dexter T, Ormerod E and Hart I, 1986, Increased metastatic capacity of a murine melanoma following induction of differentiation. Cancer Research, 46, 3239-3244. Toth CA, Rapoza A, Zamcheck N, Steele G and Thomas P, 1989, Receptor mediated endocytosis of carcinoembryonic antigen by rat alveolar macrophages in vitro. Journal of Leukocyte Biology, 45, 370-376. Gjoren T, Seljelid R and Kolset S, 1989, Binding of metastatic colon carcinoma cells to liver macrophages. Journal of Leukocyte Biology, 45,370-376.
Metastatic potential of colorectal tumor cells 22. Keller F, Wild M and Krin A, 1989, In vitro properties of unactivated rat Kupffer cells. Journal of Leukocyte Biology, 35,467-474. 23. Hostetter RB, Augustus IB, Mankarious R, Chi KF, Fan D, Toth C, Thomas P and Jessup JM, 1990, Carcinoembryonic antigen as a selective enhancer of colorectal cancer metastasis. Journal of the National Cancer Institute, 82,380-385. 24. Liotta L, Kleinerman J and Saidel G, 1976, The significance of hematogenous tumor cell clumps in the metastatic process. Cancer Research, 36, 889-894. 25. Jessup J and Thomas P, 1989, Carcinoembryonic antigen: function in metastases by human colorectal carcinoma. Cancer Metastases Review, 8,263-280. 26. Thomas P and Toth CA, 1989, Site of carcinoembryonic antigen binding to Kupffer cells. Biochemical Society Transactions, 17, 1121-1122. 27. O'Brien M, Bronstein B, Zamcheck N, Saravis C, Burke B and Gottlieb LS, 1980, Cholestasis and hepatic metastases: a factor contributing to extreme elevations of carcinoembryonic antigen. Journal of the National Cancer Institute, 64, 1291-1294. 28. Weiss L, Grundmann E, Torhorst J, Hartveit F,
Moberg I, Eder M, Fenoglio-Preiser C, Napier J, Home C, Lopez M, Shaw-Dunn R, Sugar J, Davies J, Day D and Harlos J, 1986, Hematogenous metastatic patterns in colonic carcinoma: an analysis of 1541 necropsies. Journal of Pathology, 150, 195203. Kyriazis A, DiPersio L, Michael G, Pesce A and Stinnett J, 1978, Growth patterns and metastatic behaviour of human tumors growing in athymic mice. Cancer Research, 38, 3186-3190. Giavazzi R, Jessup J, Campbell D, Walker S and Fidler I, 1986, Experimental nude mouse model of human colorectal cancer liver metastases. Journal of the National Cancer Institute, 77, 1303-1308. Bresalier R, Hujanen E, Raper S, Roll J, Itzkowwitz S, Martin G and Kim Y, 1987, An animal model for colon cancer metastasis: establishment and charcterization of murine cell lines with enhanced liver metastasizing ability. Cancer Research, 47, 1398-1406. Giavazzi R, Campbell D, Jessup J, Cleary K and Fidler I, 1986, Metastatic behaviour of tumor cells isolated from primary and metastatic human colorectal carcinomas implanted into different sites in nude mice. Cancer Research, 46, 1928-1933.
Clinical & Experimental Metastasis Vol 10 No 1 31