Introduction

Combinations of surgery, radiation therapy, and chemotherapy are the conventional treatment modalities for cancer. With the growing knowledge of tumor molecular biology, however, most novel cancer therapies are focused on tumor-specific targets to increase efficacy and to minimize systemic toxicity. Immunotherapy [reviewed in 1,2,3,4] and gene therapy [reviewed in ], for example, are often directed against tumor-associated molecules, and carcinoembryonic antigen (CEA), due to its ectopic or deregulated overexpression in up to 70% of all tumors, represents a popular target [for examples, see 3,6,7,8,9,10].

Human CEA is the prototypic member of the human CEA gene family, a group of highly glycosylated homotypic/heterotypic cell surface intercellular adhesion molecules, and part of the immunoglobulin gene superfamily, which consists of seven expressed genes [reviewed in ]. These have molecular structures consisting of multiples of highly homologous units and demonstrate a wide range of expression patterns and biological activities. Their extracellular domains all consist of a V-like Ig amino-terminal domain followed by a variable number (between 0 and 6) of I-like Ig internal domains. They can be divided into two groups based on their membrane anchorage: glycophosphatidyl inositol (GPI)-anchored members (CEA (CEACAM5), CEACAM6, CEACAM7, and CEACAM8) and transmembrane members (CEACAM1, CEACAM3, and CEACAM4).

The prototype member, CEA, a well-known tumor marker¹³, is expressed mostly in the gastrointestinal tract^11,12 and is overexpressed in many human cancers, including epithelial tumors originating from the gastrointestinal tract, lung, thyroid, breast, prostate, cervix, and ovaries^11,12,14. Preclinical animal studies are commonly performed with athymic immunodeficient mice bearing human xenografted tumors. For CEA-targeted therapies, however, these animal models are inadequate for testing therapeutic response and toxicity because rodents lack the gene coding for human CEA.

Recently established immunocompetent preclinical animal models, i.e., human CEA transgenic mice bearing human CEA-transfected mouse xenografted tumors^15,16,17, represent an improvement but still have serious shortcomings. For example, anti-CEA antibodies used in various kinds of immunotherapy [for examples, see^3,7,8] can show differences in pharmacokinetics in mice versus humans depending on their specificity¹⁸. This can be due to cross-reaction with other human CEA family members, such as CEACAM6 (formerly NCA), that are absent in mice, resulting in mistargeting to CEACAM6 on circulating and tissue granulocytes and other CEACAM6-expressing tissues in humans¹⁸. This not only may limit efficacy of the therapy but also may cause unnecessary toxicity. Construction of a transgenic model with a closer human approximation, i.e., with as many human CEA family members as possible having correct spatiotemporal expression, will be necessary to ensure better preclinical assessment of CEA-targeted therapies.

Rodents have genes coding for only murine transmembrane-anchored Ceacam1 (formerly biliary glycoprotein), a homolog of human CEACAM1. CEACAM1 has a broad expression pattern in normal human and rodent tissues and is usually down-regulated in human cancers¹². During the primate radiation, the human CEA family genes evolved, presumably from a primordial CEACAM1-like gene, by duplication (F. Naghibalhossaini et al., submitted for publication), acquired new structures and functions, and developed a more restricted expression pattern¹². For example, CEACAM6 is expressed in the gastrointestinal tract, breast, and hematopoietic system¹⁹ and is also overexpressed in a variety of adenocarcinomas^11,12 and leukemias^20,21; CEACAM7 (formerly CGM2) is expressed mainly in colon and pancreas¹⁹, and CEACAM3 (formerly CGM1) is expressed only in mature neutrophils²². In humans these genes are organized into two clusters (250-kb proximal cluster and 850-kb distal cluster, relative to the centromere) on the long arm of chromosome 19 in the region 19q13.2^23,24.

Using a well-established protocol for transgenesis with bacterial artificial chromosomes (BAC) [reviewed in 25], we have successfully constructed transgenic mice with genomic insertions of an intact 187-kb human BAC containing much of the proximal human CEA gene cluster, including CEACAM5 (CEA), CEACAM3, CEACAM6, and CEACAM7 genes, along with 29.3-kb centromeric and 19.4-kb telomeric flanking sequences. While the expression of CEACAM3 is inconclusive, CEA, CEACAM6, and CEACAM7 are expressed much the same as in humans, showing the inclusion and conservation of all cis-transcriptional regulatory elements in the CEABAC transgenic mice. Their close linkage ensures cotransmission in mouse matings. Thus the human-like expression pattern of three and possibly four human CEA family genes and the ease of mouse line maintenance make this mouse the model of choice for preclinical animal testing for CEA-targeted therapies.

Results

Generation and identification of CEABAC transgenic mice

After sequencing the CEABAC ends, restriction mapping, and aligning with the known sequence of human chromosome 19, we determined the CEABAC shown in Fig. 1A to have complete coding sequences for the CEACAM5, CEACAM3, CEACAM6, and CEACAM7 genes and 29.3-kb centromeric and 19.4-kb telomeric flanking sequences. We used a NotI DNA fragment containing the 187-kb CEABAC insert to generate transgenic mice. Although over a thousand embryos were microinjected with the CEABAC insert, only 30 mice were born and only 3 of them (CEABAC-891, -892, and -1747) were positive for the CEABAC transgene by PCR using primers specific for CEA, CEACAM3, CEACAM6, and CEACAM7 (Fig. 2). However, CEABAC-891 lacks part of the CEACAM7 gene (Fig. 2) and is also a germ-line-negative mosaic that did not give any CEABAC-positive offspring (0/8). CEA, but not CEACAM6 or CEACAM7, could be detected in fecal extracts from CEABAC-891 by ELISA, whereas the expression of CEA, CEACAM6, and CEACAM7 was detected in fecal extracts from CEABAC-892 and CEABAC-1747 (Fig. 3). Thus the CEABAC-892 and -1747 transgenics with complete CEABAC transgenes express all three CEA family members tested. CEABAC-891, lacking a complete CEACAM7 gene, as expected, expressed no CEACAM7; its failure to express CEACAM6 as well suggests that regulatory elements for this family member could be present within the CEACAM7 gene. This suggestion requires confirmation with more direct assays.

Figure 1.

CEABAC DNA construct and Southern blots of CEABAC transgenic mouse genomic DNA. (A) The human CEABAC DNA was cloned into the pBeloBAC11 vector at the HindIII restriction site, which is flanked by two NotI restriction sites. Dotted and solid lines denote pBeloBAC11 vector (7.4 kb) and CEABAC (187 kb) sequences, respectively. The CEABAC contains four known genes (arrows pointing in the direction of transcription) that belong to the human CEA family, starting from the centromeric end: CEACAM7 (CC7, 44.2–29.3 kb), CEACAM5 (CEA, 64.7–86 kb), CEACAM6 (CC6, 111.5–127.9 kb), and CEACAM3 (CC3, 152.4–167.7 kb). Restriction enzyme sites: No (NotI), H (HindIII), P (PacI, 2, 7.9, 20.7, 125.8, and 157.4 kb), A (AatII, 47.5, 51, 169.2, and 181.8 kb), Pv (PvuI, 92.1 kb), and N (NruI, 137.6 and 146.7 kb) are shown. The CEABAC end sequences were (centromeric end) AAGCTTATTTATGTTCCACCTAAAGTCAGTTTTGGGAAACACTGA...(CEABAC)...GTAGCTATGGCAGTGGCAGAAGATTTTAATTAGAAAACAAAGCTT (telomeric end), where AAGCTT is the HindIII cloning site. (B) Long-range Southern blot of agarose-embedded fibroblast genomic DNA (5–30 g) from CEABAC heterozygous mice and BAC DNA digested with PacI, resolved by PFGE and probed with ³²P-labeled random primers generated from CEA cDNA. Lanes (from left to right): BAC DNA control, negative mouse control, CEABAC-892, and CEABAC-1747. As indicated in A, three bands (a 105-kb band containing CEACAM7, CEA, and part of the CEACAM6 gene; a 31.6-kb band containing part of the CEACAM6 gene; and a 29.6-kb band containing the CEACAM3 gene) were expected due to high homology between these family members. Note that the latter two bands comigrated in the unresolvable region of PFGE. Also note that the 29.6-kb band would become 31.6 kb in the CEABAC mice due to the head-to-tail configuration of multiple BAC copies. (C) Southern blot of genomic DNA (10 g) and quantified BAC-end DNA digested with EcoRI and probed with ³²P-labeled random primers generated from BAC-end DNA (3' end only). Lanes (from left to right): negative mouse control, CEABAC-892, CEABAC-1747, and BAC-end copy-number control (1, 3, 5, and 10 copies). Note the expected 4-kb band (3' end) and a 5.4-kb band (5' end + 3' end) that was due to incomplete digestion of the CEABAC concatemers in a head-to-tail configuration. By densitometric quantification, the transgene copy number of CEABAC-892 and CEABAC-1747 was estimated to be 2.3 0.1 and 9.7 0.2 copies (by combining the two bands), respectively.

Full figure and legend (119K)

Figure 2.

PCR analyses of CEABAC transgenic mouse tail DNA. Columns from left to right: CEABAC DNA control, negative mouse (1745), and transgenic founders (891, 892, and 1747). Rows from top to bottom: CEACAM7-specific primers (535 bp), CEA-specific primers (738 bp), CEACAM6-specific primers (794 bp), CEACAM3-specific primers (584 bp), and -actin (DNA level control)-specific primers (1008 bp).

Full figure and legend (73K)

Figure 3.

Expression of CEA family members from ELISA of fecal extracts. Fecal levels of CEA, CEACAM6, and CEACAM7 were detected by antibody-sandwich ELISA using RbCEA and monospecific antibodies (D14, 9A6, and BAC2, respectively). Relative levels in each experiment were calculated by normalizing to the highest absorbance at 405 nm and were averaged for three independent experiments. Black, white, and patterned series denote fecal samples collected from human, negative mouse, and CEABAC transgenic founders (891, 892, and 1747), respectively. Mean SD was plotted for each series. Fecal CEA, CEACAM6, and CEACAM7 could be detected in both 892 and 1747 transgenics, whereas only CEA could be detected in 891 transgenic fecal extracts. ^* and ^** denote significant differences between CEABAC transgenic founders and negative mice (P values <0.005 and <0.05, respectively). Error bars are not shown for CEA and CEACAM6 levels in the human samples since they represented the highest levels (used for normalizing the other levels) in all three experiments (SD = 0).

Full figure and legend (112K)

Southern blot analyses showed that the CEABAC inserts were intact in CEABAC-892 and -1747 (Fig. 1B), with approximately 2 and 10 transgene copies, respectively, in a head-to-tail configuration (Fig. 1C). CEABAC-892 and -1747 gave 50% (3/6) and 13% (6/46) positive F1 pups, respectively, indicating that CEABAC-1747 was a germ-line mosaic. Upon breeding of the F1 and F2 generations with wild-type FVB mice, CEABAC-892 and -1747 lines gave 52% (16/31) and 47% (60/127) positive offspring, respectively, following a Mendelian pattern of inheritance and thus showing removal of mosaicism from the CEABAC-1747 line and confirming close clustering of the multiple CEABAC inserts in each line. Also, in the CEABAC-1747 heterozygotes, we could detect only one labeled nuclear spot by fluorescence in situ hybridization (data not shown). We used these two lines (CEABAC-892 and -1747) to establish the CEABAC transgenic lines for further studies.

Expression of CEA and CEACAM6

The expression levels of CEA and CEACAM6 in various tissues of CEABAC-1747 (10 transgene copies) were shown by immunoblots to be higher than those of CEABAC-892 (2 transgene copies) by approximately fivefold using serial dilutions of colon protein extracts (data not shown). Although both transgenics exhibited similar spatiotemporal expression patterns, we show immunoblot results for CEABAC-1747 only (Fig. 4). While CEA was highly expressed in part of the gastrointestinal tract and vagina (Fig. 4, low exposure), CEACAM6 was highly expressed in the bone marrow and vagina (Fig. 4, low exposure). We could also detect CEA at a much lower level in the trachea, salivary gland, penis, cervix, and breast (Fig. 4, high exposure). Similarly, we could detect CEACAM6 at lower levels in the trachea, lung, part of the alimentary tract, cervix, breast, and spleen (Fig. 4, high exposure). The much higher level of CEA relative to CEACAM6 expression in colonic tissue in the CEABAC mice is mirrored in CEA vs CEACAM6 expression in normal human colonic tissue²⁶.

Figure 4.

Tissue-specific expression of CEA and CEACAM6 from immunoblots. Lanes 1 to 13 and 15 to 27: immunoblots using mAb B18 (specific for both CEA and CEACAM6) for detection of CEA (the lower M_r CEACAM6 region of the blot is not shown; identical but much more weakly detectable bands in the higher M_r region were obtained using a CEA-specific mAb, data not shown) and mAb 9A6 for detection of CEACAM6, after electrophoretic resolution of 50 g protein from extracts of the indicated tissues (200 g protein from peripheral blood) of CEABAC mice. Lanes 14 and 28: protein extracts of colon and bone marrow, respectively, from wild-type (WT) control littermates. The lower image of each of the CEA and CEACAM6 immunoblots represents a longer film exposure of the same blot by a factor of 300 for CEA and 300 (left) or 30 (right) for CEACAM6. The presence of a prominent background band (70 kDa) in most of the samples (except small intestine), including the WT controls (lanes 14 and 28), is due to ubiquitous endogenous mouse IgG molecules (not detected when rabbit anti-CEA Ab and anti-rabbit secondary antibody was used—not shown).

Full figure and legend (170K)

Molecular weights of the CEA family members expressed in human cells and in transfectants of various animal cell lines have often been found to be variable due to variable glycosylation²⁷, giving multiple broad bands on immunoblots. In the CEABAC mice, the molecular weights of CEA in most tissues (180 kDa) were comparable to known human values¹². The molecular weight of CEACAM6 in bone marrow (50 and 90 kDa) was also similar to that of human granulocytes¹². The molecular weight of the major band of CEACAM6 in many tissues, however, was 75–90 kDa, which is broader than that found in human tissues¹². In rectal tissue, both CEA and CEACAM6 had surprisingly higher molecular weights than expected¹². Since splice variants giving variable molecular weights have never been observed for CEA and CEACAM6, these results are most likely indicative of higher levels of glycosylation. Thus, although the spatiotemporal expression patterns of CEA and CEACAM6 were highly conserved between humans and mice, machinery responsible for posttranslational modifications may not be completely conserved, leading to differences in glycosylation of some CEA family members in some tissues.

Expression of CEACAM3 and CEACAM7

We expected CEACAM3 and CEACAM7 to be expressed at a very low level in mature neutrophils and colonocytes, respectively, and in fact they could not be easily detected by immunoblots of total protein extracts (data not shown). Hence, we assessed the expression patterns of CEACAM3 and CEACAM7 by RT-PCR in selected tissues. We could not detect CEACAM3 in any tested tissues, including spleen, in which the proportion of mature neutrophils is high relative to other hematopoietic organs (data not shown). Since CEACAM3 mRNA could not be detected in a human spleen RNA sample using the same RT-PCR conditions, the expression of CEACAM3 in mature neutrophils was inconclusive. In the gastrointestinal tract, we could detect CEACAM7 only in the colon (Fig. 5). CEA and CEACAM6 RT-PCR results are included in Fig. 5 as positive controls. Thus, the expression of CEACAM7 was highly restricted, as observed in humans¹².

Figure 5.

Tissue-specific expression of CEA family members from RT-PCR. From left to right: 100 bp ladder marker (MW); RT-PCR buffer control; RT-PCR products of RNA extracted from esophagus, stomach, ileum, colon, and liver of CEABAC mice; RT-PCR control of RNA extracted from the colon of a wild-type control littermate; RT-PCR control of human colon RNA; MW markers. From top to bottom: RT-PCR for CEA mRNA (expected size 338 bp)—positive for esophagus, stomach, small intestine, and colon; RT-PCR for CEACAM6 mRNA (expected size 353 bp)—positive for stomach and colon; RT-PCR for CEACAM7 mRNA (expected size 1423 bp)—positive for colon only; and RT-PCR for mouse -actin mRNA (expected size 405 bp)—positive for mouse samples only.

Full figure and legend (98K)

Immunohistochemical analysis of CEA family expression

Although individual CEA family members could be distinguished using specific mouse monoclonal antibodies, ubiquitous endogenous mouse immunoglobulins rendered their localization technically difficult. Therefore, we carried out immunohistochemistry using polyclonal rabbit anti-CEA antibody, which detects all human CEA family members but fails to bind to any of the mouse Ceacam1 proteins (Fig. 4, WT colon and bone marrow lanes and data not shown), with results shown in Fig. 6; the actual family member(s) detected can be deduced by referral to the immunoblots of Fig. 4. We confirmed these results using mouse monoclonal antibodies specific for CEA and CEACAM6 with special provision for the presence of mouse immunoglobulins, but they are not shown here due to the persistence of some background. The highest expression of CEA family members was in colon followed by stomach and vagina (in agreement with levels detected by immunoblots). The expression of CEA (CEACAM6 and CEACAM7 are expressed at much lower levels in the colon—Figs. 4 and 5) was localized to the apical surface of differentiated colonocytes at the top of colonic crypts; expression decreased toward the proliferative zone at the base of the crypts (Fig. 6H). Along the alimentary tract (Figs. 6A–6H), CEA and/or CEACAM6 were also detected on the fusiform epithelium of the tongue (6A), the luminal surface of the salivary glands (6B) and esophagus (6C), the mucosal surface of the gastric stomach (6D), the pits of the pyloric stomach (6E), and the crypts and villi of the small intestine [with a higher level in the duodenum (6F) than the ileum (6G), contrary to the immunoblot result]. We also detected CEA and/or CEACAM6 in other epithelial tissues (Figs. 6J–6R): the ductal epithelium of the female breast (6J), in the bronchial epithelium of the trachea (6K) and lung (6L), the mucosal surface of the urinary bladder (6M), the glandular epithelium of the cervix (6N), the squamous epithelium of the vagina (6O), the luminal surface of the lateral and dorsal prostatic glands (6P), and the epithelia of the male penile urethra (6Q) and skin (6R). Importantly, CEACAM6 can be detected in the bone marrow where positive immature myelopoietic cells and a small number of mature monocytes and neutrophils were found (Figs. 6S, 6T, and 6U). Based on the nuclear morphology, cell size, and presence of the azurophilic granules (appearing purplish with Wright's stain, data not shown), CEACAM6, similar to human expression²⁸, was first seen at low levels in the early promyelocyte stage (Fig. 6S), then at gradually increasing levels during neutrophilic differentiation (Fig. 6T), and finally back to low levels in mature neutrophils (Fig. 6U). We could also detect CEACAM6 in cells of the monocytic lineage, but not in other lineages (data not shown). We could detect no staining (except for neutrophils that were present) in other tissues, including kidney, heart, thymus, and liver (data not shown).

Figure 6.

Patterns of tissue expression from immunohistochemistry. All sections were incubated at room temperature with RbCEA as primary antibody, HRP-conjugated anti-rabbit antibody as secondary antibody, and DAB substrate and then counterstained with hematoxylin. RbCEA antibody dilution: 1:2500 (A–C, G, J–N, P–U), 1:5000 (D–F and O), and 1:10,000 (H and I). Original magnification: 400 (A–R) and 1000 (S–U). (A) Tongue epithelium—positive cluster of cells in the fusiform epithelium; (B) serous salivary glands (left) and mucous salivary gland (right)—positive epithelial cells; (C) esophagus—weak positive luminal surface; (D) gastric stomach adjacent to esophagus—positive mucosal surface only; (E) pyloric stomach—positive staining only in the pits, not in the glands; (F) duodenum adjacent to the pyloric stomach—positive staining throughout the epithelium of the crypts and villi; (G) ileum—positive staining on the apical surface of the epithelial cells throughout the crypts and much less in the villi; (H) colonic epithelium—intense mucosal surface staining and only on the apical surface of the cells with decreasing gradient from top to bottom; (I) colonic epithelium from a wild-type control littermate—no background staining; (J) breast—positive epithelial cells and luminal content; (K) trachea—positive epithelial cells; (L) lung—positive bronchiole; (M) urinary bladder—positive staining of the cuboidal cells of the transitional epithelium; (N) cervix—positive glandular epithelium; (O) vagina—positive stratified squamous epithelium; (P) dorsal prostate (left) and lateral prostate (right)—positive staining on the luminal surface of the prostatic glands; (Q) male urethra—weak positive staining on the luminal surface; (R) penile skin—clusters of positive cells in the stratified squamous epithelium; (S) bone marrow immature granulocyte (early promyelocyte stage)—weak surface staining (left) and wild-type control (right); (T) bone marrow immature neutrophil (band cell)—strong surface staining (left) and wild-type control (right); and (U) bone marrow mature neutrophil—intermediate surface staining (left) and wild-type control (right). Note that only wild-type controls for colon and immature or mature granulocytes are presented; however, parallel staining of all tissue sections from the wild-type control mice were completely negative (data not shown).

Full figure and legend (486K)

The tissue-specific expression patterns of the 3 CEA family members are summarized in Table 1. They show remarkable similarity to the human pattern, with a few notable exceptions that are discussed below.

Table 1 - Tissue-specific expression of the human CEA family members in the CEABAC transgenic mice.

Full table

Discussion

The CEA gene family is a large group of homotypic/heterotypic intercellular adhesion molecules¹², each of which has a specific tissue distribution with presumably specific functions in various tissues. CEA, overexpressed in as many as 70% of all human cancers^11,12,14, is a popular target for novel cancer therapies, including cancer vaccines, cellular immunotherapy, radioimmunotherapy, antibody therapy, and gene therapy^3,6,7,8,9,10. Although preclinical animal data can show promising effects, clinical outcomes have commonly been unfavorable, such as low therapeutic response or the presence of associated toxicity. This discrepancy suggests improper or insufficient assessment by present animal models. Although previously constructed transgenic mice bearing the CEA gene alone can improve the validity of preclinical tests^15,16,17, a transgenic model with a closer human approximation is necessary to ensure valid pharmacokinetics of test agents and to avoid possible treatment toxicity and/or reduced efficacy from cross-reaction with other highly homologous human CEA family members.

Although the minimal transcriptional promoter (included in the 3.3-kb 5'-upstream sequences used in previous CEA transgenic mice) of the CEACAM5 gene was shown to confer spatiotemporal expression in previous studies^29,30, the promoters/enhancers of the CEACAM6 gene³¹ and other family members are incompletely or not at all characterized, not to mention the possibility of cis interactions of transcription factors between these closely linked and highly homologous genes. Thus, the use of the entire human CEA family gene locus in transgenics should provide more accurate models than those obtainable by crossing multiple transgenic lines and, in any event, no CEACAM6 gene transgenic has been reported. Although multiple linked-gene transgenesis can be achieved using YACs (yeast artificial chromosomes)³², a 187-kb BAC that holds part of the proximal CEA gene cluster containing four of the seven expressed CEA family genes was used for the following reasons. BAC clones, unlike YAC clones, are rarely hybrids of unrelated genomic regions and do not usually undergo deletions or rearrangements when propagated in bacteria^33,34. YAC DNA, being usually longer, is also more prone to fragmentation during purification, microinjection, and genomic integration³². Like YAC clones, BAC expression is unlikely to be influenced by the genomic integration site in transgenics, simply because of size³². Moreover, the CEABAC contains most of the human family genes except CEACAM1 (although the mouse homologue of this gene is present in the transgenics), CEACAM8 (another GPI-anchored family member, formerly denoted CGM6 and expressed only in granulocytes), and CEACAM4 (formerly denoted CGM7 and similar in structure and limited expression pattern to CEACAM3)¹². Thus, although some human CEA family genes were omitted, those that are more widely expressed and expressed at higher levels are all included.

Two CEABAC transgenic lines (892 and 1747) expressing CEA family members were successfully generated and individual identification in mouse line maintenance could be easily achieved by PCR using CEA-specific primers, since CEA-positive offspring would inherit the entire CEABAC locus. The expression patterns of CEA, CEACAM6, and CEACAM7 in both CEABAC transgenic lines were shown to be very similar to those observed in humans, both spatially and in relative level. Expression levels were proportional to their CEABAC copy numbers, indicating, as expected, little or no influence of their different integration sites. In brief, CEA was found to have a more restricted expression pattern, with the highest levels in colon, stomach, and vagina, than CEACAM6, which was shown to have a broader expression pattern with the highest levels in bone marrow (primarily immature myelocytes) and vagina. Expression of CEACAM7 was shown to be highly restricted to colorectum. The presence of CEACAM3 in mature neutrophils remained inconclusive due to the fact that it is expressed at low levels in a small population of cells in a given tissue in humans and, presumably, in the CEABAC transgenics. However, the fact that CEACAM3 was weakly detected in occasional tissue neutrophils using CEACAM3-specific monoclonal antibody suggests that it is, in fact, expressed in CEABAC mouse neutrophils (data not shown). The expression of human CEA and CEACAM6 in the penile urethra and vagina of the CEABAC mice was unexpected and suggests the possibility of similar expression in humans, where apparently no such studies have been reported. However, unlike the human situation, CEA and CEACAM7 could not be detected in the pancreas. This may be due to the sensitivity of the detection methods employed or intrinsic differences between humans and mice. Actually, the anatomical structures of the pancreas of humans (a well-defined organ) and mice (ill-defined fat-like tissue) are radically different so that different expression patterns, either qualitative or quantitative, are perhaps not surprising. Based on the conservation of the complex spatiotemporal expression pattern, presumably due to the conservation in evolution of most, if not all, of the relevant trans- and cis-transcriptional regulating elements, our CEABAC mice are suitable for preclinical trials of CEA-targeted agents using existing protocols [for examples, see^15,16,17], as explained earlier.

Since CEACAM6 is also overexpressed in a variety of adenocarcinomas^11,12 and leukemias^20,21, CEACAM6-targeted therapies, such as radioimmunotherapy for acute leukemia using radiolabeled anti-CEACAM6 antibodies³⁵, are possible. Hence, this mouse model can be used for preclinical trials of novel CEACAM6-targeted cancer therapy.

The expression of the CEA family genes in gastrointestinal, breast, respiratory, urogenital, and hematopoietic systems in the CEABAC mice could facilitate the search for therapeutic agents against various kinds of cancer and other diseases associated with the human CEA family and also provides an adequate model for basic studies. Examples include: (1) CEA and CEACAM6 have been shown to contribute directly to tumorigenesis^12,36,37,38 and metastasis^12,39,40,41 in various model systems. Overexpression of CEA and CEACAM6 blocks cellular differentiation¹², disrupts cellular polarization and normal colonic tissue architecture³⁶, and leads to an inhibition of anoikis^37,38, i.e., death by apoptosis of cells detached from their extracellular matrix. Thus, reversal of these effects represents a novel therapeutic approach and can be exploited in this present mouse model. (2) Neisseria gonorrhoeae infection, a sexually transmitted disease, uses CEA family members as major binding receptors⁴². Thus agents that inhibit this interaction in vivo may be useful for prevention, and their effectiveness can be assessed in this mouse model. (3) Human CEA family members can activate neutrophils and increase their adhesion to endothelial cells and extracellular matrices^43,44,45. They can also induce cytokine release from immune cells^46,47. These properties suggest their role in inflammatory reactions and their potential use as targets for anti-inflammatory drugs.

In conclusion, the human-like expression pattern of three and possibly four human CEA family genes along with a simple strategy for mouse line maintenance should make this novel transgenic mouse the model of choice for therapeutic trials of CEA-targeted therapies. This may also lead to better understanding of various diseases associated with these human-specific CEA family members and, eventually, development of more successful therapies.

Materials and Methods

Primers and antibodies

Primers specific for the genes and cDNAs of CEACAM5, CEACAM3, CEACAM6, and CEACAM7 are listed in Table 2. Primers specific for the T7 (5'-TACCCGGGGATCCTCTAGAGTC-3') and SP6 (5'-TTCCGGCTCGTATGTTGTGTGG-3') sequences on the pBeloBAC11 vector were used to sequence both ends of the 187-kb CEABAC insert.

Table 2 - Primer sequences and expected sizes of PCR products.

Full table

Rabbit polyclonal anti-CEA antibody (RbCEA) recognizes all human CEA family members. The mouse monoclonal antibodies used were D14, specific for CEA⁴⁸; B18, specific for CEA, CEACAM6, and human CEACAM1⁴⁸; 9A6, specific for CEACAM6¹⁹; and BAC2, specific for CEACAM7¹⁹. None of these antibodies binds to endogenous mouse CEA family members (murine Ceacam1).

Confirmation of CEABAC sequence

The CEABAC (GenBank Accession No. BC627193), cloned into the HindIII site of the 7.4-kb pBeloBAC11 vector³³, was obtained from a human BAC library (Research Genetics, Inc., Huntsville, AL, USA). The CEABAC clone was verified by PCR using primers specific for CEACAM5, CEACAM3, CEACAM6, and CEACAM7 and by restriction mapping using pulsed-field gel electrophoresis (PFGE; Bio-Rad Laboratories, Inc., Hercules, CA, USA). The CEABAC ends were subcloned by digestion with restriction enzyme AvrII (New England Biolabs, Inc., Beverly, MA, USA), which cuts the CEABAC but not the vector, followed by ligation. The CEABAC ends were then sequenced with an automated DNA nucleotide sequencer using T7 and SP6 primers and the sequences were compared with the DNA sequence of human chromosome 19q13.2 (GenBank Accession No. NT_011139) to identify the exact location of the CEABAC.

Generation of CEABAC transgenic mice

The CEABAC insert was removed from the pBeloBAC11 construct by digestion with NotI (New England Biolabs Inc.), separated from the vector by PFGE, and purified by -agarase digestion (Invitrogen Corp., Carlsbad, CA, USA). The DNA solution was dialyzed and adjusted to 1 ng/l with microinjection buffer (10 mM Tris, pH 7.4, 0.1 mM EDTA). This DNA solution was microinjected, minimizing shearing forces, into pronuclei of FVB embryos, which were then implanted into pseudo-pregnant CD-1 females, as described previously³². CEABAC transgenic founders were identified by PCR on genomic DNA using CEACAM5-, CEACAM3-, CEACAM6-, and CEACAM7-specific primers and tested for expression by ELISA with specific monoclonal antibodies applied to fecal protein extracts, as described below. Gene copy numbers were determined by Southern blots of genomic DNA digested with EcoRI, calibrated with known levels of CEABAC-end DNA digested with EcoRI, and detected with ³²P-labeled DNA probes generated by random priming on the CEABAC ends using the High Prime DNA Labeling Kit (Roche Diagnostics, Inc., Laval, QC, Canada). Intactness of the integrated CEABAC sequences was determined by long-range Southern blots of genomic DNA digested with PacI (New England Biolabs Inc.) in agarose blocks and resolved by PFGE, probing with ³²P-labeled DNA probes generated by random priming on the CEA cDNA that lacks the Alu sequence at the 3' end using the High Prime DNA Labeling Kit (Roche Diagnostics Inc.).

Tissue protein extraction, immunoblot analysis, and ELISA

Tissues were homogenized on ice and lysed with SDS lysis buffer (100 mM Tris, pH 8.0, 10% glycerol, 2% SDS). Fifty micrograms of tissue extracts was resolved by SDS–PAGE, and Western blot analyses were performed using B18 and 9A6 antibodies as described previously⁴⁹. Mouse fecal pellets were homogenized on ice with PBS containing 1% TX-100 and sonicated for 30 s. Two hundred microliters of fecal extracts were analyzed by antibody-sandwich ELISA⁵⁰ using D14, 9A6, and BAC2 as coating antibodies and RbCEA as detecting antibody.

Tissue RNA extraction and RT-PCR

Tissues were removed from animals and preserved in RNALater (Ambion, Inc., Austin, TX, USA) at -20°C. Bone marrow, spleen, and pancreas were prepared fresh for RNA extraction. Twenty to sixty milligrams of tissue was homogenized and RNA was isolated using the RNAqueous extraction kit (Ambion, Inc.) following the manufacturer's instructions and stored at -80°C. Human RNA samples were purchased and stored at -80°C (Ambion, Inc.). Total RNA (0.5–1 g) was used for each RT-PCR by MMTV-RT (Invitrogen Corp.) according to the manufacturer's procedures.

Immunohistochemical analyses

Tissues were cut into pieces not thicker than 5 mm, fixed with 4% paraformaldehyde for 16 h, infused with PBS–0.5 M sucrose for 12 h at 4°C, and quickly frozen with isopentane at -70°C. Frozen sections of 7–10 m were obtained using a cryostat at -20 to -30°C. Immunohistochemical staining was performed using Envision Reagents (DAKO Diagnostics Canada, Inc., Mississauga, ON, Canada) with RbCEA antibody at a dilution of 1:2500 to 1:10,000. Sections were developed with 3',3'-diaminobenzidine for 1 to 10 min and were counterstained with Mayer's hematoxylin (Sigma–Aldrich Canada Ltd., Oakville, ON, Canada).

References

1. Ribas, A., Butterfield, L. H., Glaspy, J. A. and Economou, J. S. (2003). Current developments in cancer vaccines and cellular immunotherapy. J. Clin. Oncol. 21: 2415–2432. | Article | PubMed | ISI | ChemPort |
2. von Mehren, M., Adams, G. P. and Weiner, L. M. (2003). Monoclonal antibody therapy for cancer. Annu. Rev. Med. 54: 343–369. | Article | PubMed | ISI | ChemPort |
3. Goldenberg, D. M. (2003). Advancing role of radiolabeled antibodies in the therapy of cancer. Cancer Immunol. Immunother. 52: 281–296. | PubMed | ISI | ChemPort |
4. Trail, P. A., King, H. D. and Dubowchik, G. M. (2003). Monoclonal antibody drug immunoconjugates for targeted treatment of cancer. Cancer Immunol. Immunother. 52: 328–337. | PubMed | ChemPort |
5. Kellen, J. A. (2002). Gene therapy in cancer. J. Exp. Ther. Oncol. 2: 312–316. | Article | PubMed | ChemPort |
6. Marshall, J. (2003). Carcinoembryonic antigen-based vaccines. Semin. Oncol. 30: 30–36. | Article | PubMed | ChemPort |
7. Kousparou, C. A., Epenetos, A. A. and Deonarain, M. P. (2002). Antibody-guided enzyme therapy of cancer producing cyanide results in necrosis of targeted cells. Int. J. Cancer. 99: 138–148. | Article | PubMed | ISI | ChemPort |
8. Chester, K. A., et al. (2000). Recombinant anti-carcinoembryonic antigen antibodies for targeting cancer. Cancer Chemother. Pharmacol. 46: S8–S12. | Article | PubMed | ChemPort |
9. Fong, L., et al. (2001). Altered peptide ligand vaccination with Flt3 ligand expanded dendritic cells for tumor immunotherapy. Proc. Natl. Acad. Sci. USA 98: 8809–8814. | Article | PubMed | ChemPort |
10. Koch, P. E. (2001). Augmenting transgene expression from carcinoembryonic antigen (CEA) promoter via a GAL4 gene regulatory system. Mol. Ther. 3: 278–283. | Article | PubMed | ChemPort |
11. Hammarström, S. (1999). The carcinoembryonic antigen (CEA) family: structures, suggested functions and expression in normal and malignant tissues. Semin. Cancer Biol. 9: 67–81.
12. Stanners, C. P. (1998). Cell Adhesion and Communication Mediated by the CEA Family: Basic and Clinical Perspectives1. Harwood Academic: Amsterdam.
13. Ballesta, A. M., Molina, R., Filella, X., Jo, J. and Gimenez, N. (1995). Carcinoembryonic antigen in staging and follow-up of patients with solid tumors. Tumour Biol. 16: 32–41. | PubMed | ChemPort |
14. Chevinsky, A. H. (1991). CEA in tumors of other than colorectal origin. Semin. Surg. Oncol. 7: 162–166. | PubMed | ChemPort |
15. Mizobata, S., Tompkins, K., Simpson, J. F., Shyr, Y. and Primus, F. J. (2000). Induction of cytotoxic T cells and their antitumor activity in mice transgenic for carcinoembryonic antigen. Cancer Immunol. Immunother. 49: 285–295. | Article | PubMed | ChemPort |
16. Wilkinson, R. W., et al. (2002). Evaluation of a transgenic mouse model for anti-human CEA radioimmunotherapeutics. J. Nucl. Med. 43: 1368–1376. | PubMed |
17. Xu, X., et al. (2000). Targeting and therapy of carcinoembryonic antigen-expressing tumors in transgenic mice with an antibody-interleukin 2 fusion protein. Cancer Res. 60: 4475–4484. | PubMed | ISI | ChemPort |
18. Oriuchi, N., et al. (1998). Antibody-dependent difference in biodistribution of monoclonal antibodies in animal models and humans. Cancer Immunol. Immunother. 46: 311–317. | Article | PubMed | ChemPort |
19. Schölzel, S., et al. (2000). Carcinoembryonic antigen family members CEACAM6 and CEACAM7 are differentially expressed in normal tissues and oppositely deregulated in hyperplastic colorectal polyps and early adenomas. Am. J. Pathol. 156: 595–605.
20. Hansen, L., Meyer, K. and Hokland, P. (1998). Flow cytometric identification of myeloid disorders by asynchronous expression of the CD14 and CD66 antigens. Eur. J. Haemotol. 61: 339–346.
21. Boccum, P., et al. (1998). CD66c antigen expression is myeloid restricted in normal bone marrow but is a common feature of CD10⁺ early-B-cell malignancies. Tissue Antigens 52: 1–8. | PubMed |
22. Nagel, G., et al. (1993). Genomic organization, splice variants and expression of CGM1, a CD66-related member of the carcinoembryonic antigen gene family. Eur. J. Biochem. 214: 27–35. | Article | PubMed | ISI | ChemPort |
23. Thompson, J., et al. (1992). Long-range chromosomal mapping of the carcinoembryonic antigen (CEA) gene family cluster. Genomics 12: 761–772. | Article | PubMed | ChemPort |
24. Brandriff, B. F., et al. (1992). Order and genomic distances among members of the carcinoembryonic antigen (CEA) gene family determined by fluorescence in situ hybridization. Genomics 12: 773–779. | Article | PubMed | ISI | ChemPort |
25. Dewar, K., Birren, B. W. and Abderrahim, H. (1997). Bacterial artificial chromosomes and animal transgenesis. In: Houdebine L.M. (Ed). Transgenic Animals: Generation and Use. Harwood Academic: Amsterdam. 283–287.
26. Cournoyer, D., Beauchemin, N., Boucher, D., Benchimol, S., Fuks, A. and Stanners, C. P. (1988). Transcription of genes of the carcinoembryonic antigen family in malignant and nonmalignant human tissues. Cancer Res. 48: 3153–3157. | PubMed | ISI | ChemPort |
27. Yamahita, K. and Kobata, A. (1996). Cancer cells and metastasis: carcinoembryonic antigen and related normal antigens. In: Montreuil J, Vliegenthart J. F. G., and Schachter H. (Eds). Glycoproteins and Disease. Elsevier: Amsterdam. 229–239.
28. Watt, S. M., et al. (1991). CD66 identifies a neutrophil-specific epitope within the hematopoietic system that is expressed by members of the carcinoembryonic antigen family of adhesion molecules. Blood. 78: 63–74. | PubMed | ChemPort |
29. Eades-Perner, A.-M., et al. (1994). Mice transgenic for human carcinoembryonic antigen gene maintain its spatiotemporal expression pattern. Cancer Res. 54: 4169–4176. | PubMed | ChemPort |
30. Clarke, P., Mann, J., Simpson, J. F., Richard-Dickson, K. and Primus, F. J. (1998). Mice transgenic for human carcinoembryonic antigen as a model for immunotherapy. Cancer Res. 58: 1469–1477. | PubMed | ChemPort |
31. Koops, D., Thompson, J., Zimmermann, W. and Stanners, C. P. (1998). Transcriptional regulation of the non-specific cross-reacting antigen gene, a member of the carcinoembryonic antigen gene family up-regulated in colorectal carcinomas. Eur. J. Biochem. 253: 778–786. | Article | PubMed |
32. Huxley, C. (1998). Exploring gene function: use of yeast artificial chromosome transgenesis. Methods. 14: 199–210. | Article | PubMed | ISI | ChemPort |
33. Kim, U.-J., et al. (1996). Construction and characterization of a human bacterial artificial chromosome library. Genomics. 34: 213–218. | Article | PubMed | ISI | ChemPort |
34. Song, J., Dong, F., Lilly, J. W., Stupar, R. M. and Jiang, J. (2001). Instability of bacterial artificial chromosome (BAC) clones containing tandemly repeated DNA sequences. Genome. 44: 463–469. | Article | PubMed | ChemPort |
35. Burke, J. M., Jurcic, J. G. and Scheinberg, D. A. (2002). Radioimmunotherapy for acute leukemia. Cancer Control. 9: 106–113. | PubMed |
36. Ilantzis, C., Demarte, L., Screaton, R. A. and Stanners, C. P. (2002). Deregulated expression of the human tumor marker CEA and CEA family member CEACAM6 disrupts tissue architecture and blocks colonocyte differentiation. Neoplasia. 4: 151–163. | Article | PubMed | ISI | ChemPort |
37. Soeth, E., et al. (2001). Controlled ribozyme targeting demonstrates an antiapoptotic effect of carcinoembryonic antigen in HT29 colon cancer cells. Clin. Cancer Res. 7: 2022–2030. | PubMed | ChemPort |
38. Ordoñez, C., Screaton, R. A., Ilantzis, C. and Stanners, C. P. (2000). Human carcinoembryonic antigen functions as general inhibitor of anoikis. Cancer Res. 60: 3419–3424.
39. Wirth, T., Soeth, E., Czubayko, F. and Juhl, H. (2002). Inhibition of endogenous carcinoembryonic antigen (CEA) increases the apoptotic rate of colon cancer cells and inhibits metastatic tumor growth. Clin. Exp. Metastasis. 19: 155–160. | Article | PubMed |
40. Kim, J. C., Gong, G., Roh, S. A. and Park, K. C. (1999). Carcinoembryonic antigen gene and carcinoembryonic antigen expression in the liver metastasis of colorectal carcinoma. Mol. Cell. 9: 133–137. | ChemPort |
41. Leconte, A., et al. (1999). Involvement of circulating CEA in liver metastases from colorectal cancers re-examined in a new experimental model. Br. J. Cancer. 9: 1373–1379.
42. Dehio, C., Gray-Owen, S. D. and Meyer, T. F. (1998). The role of neisserial Opa proteins in interactions with host cells. Trends Microbiol. 6: 489–495. | Article | PubMed | ISI | ChemPort |
43. Skubitz, K. M., Campbell, K. D. and Skubitz, A. P. N. (2001). Synthetic peptides from the N-domains of CEACAMs activate neutrophils. J. Pept. Res. 58: 515–526. | Article | PubMed | ChemPort |
44. Nair, K. S. and Zingole, S. M. (2001). Adhesion of neutrophils to fibronectin: role of the CD66 antigens. Cell. Immunol. 208: 96–106. | Article | PubMed | ISI | ChemPort |
45. Kuijpers, T. W., et al. (1992). CD66 nonspecific cross-reacting antigens are involved in neutrophil adherence to cytokine-activated endothelial cells. J. Cell Biol. 118: 457–466. | Article | PubMed | ChemPort |
46. Filella, X., et al. (2001). Influence of AFP, CEA and PSA on the in vitro production of cytokines. Tumor Biol. 22: 67–71. | Article | ChemPort |
47. Gangopadhyay, A., Bajenova, O., Kelly, T. M. and Thomas, P. (1996). Carcinoembryonic antigen induced cytokine expression in Kupffer cells: implications for hepatic metastasis from colorectal cancer. Cancer Res. 56: 4805–4810. | PubMed | ChemPort |
48. Zhou, H., Stanners, C. P. and Fuks, A. (1993). Specificity of anti-carcinoembryonic antigen monoclonal antibodies and their effects on CEA-mediated adhesion. Cancer Res. 53: 3817–3822. | PubMed | ChemPort |
49. Screaton, R. A., Penn, L. Z. and Stanners, C. P. (1997). Carcinoembryonic antigen, a human tumor marker, cooperates with Myc and Bcl-2 in cellular transformation. J. Cell Biol. 137: 939–952. | Article | PubMed | ChemPort |
50. Hornbeck, P. (1991). Enzyme-linked immunosorbent assays. In: Coligan J. E., Kruisbeek A. M., Margulies D. H., Shevach E. M., and Strober W. (Eds). Current Protocols in Immunology. Greene Publishing and Wiley-Interscience: New York. 2.1.2–2.1.22.

Acknowledgements

We thank Janice E. Penney for technical assistance on microinjection of the CEABAC DNA construct. This work was supported by a grant from the National Cancer Institute of Canada with funds from the Canadian Cancer Society. C.C. was supported by an M.D./Ph.D. Studentship from the Canadian Institutes of Health Research.