Proc. Natl. Acad. Sci. USA Vol. 87, pp. 9042-9046, November 1990 Developmental Biology

Expression of chicken liver cell adhesion molecule fusion genes in transgenic mice (insulin promoter/neurorflament promoter/cis-regulatory sequence/altered gene expression)

MARTIN BEGEMANN, SEONG-SENG TAN*, BRUCE A. CUNNINGHAM, AND GERALD M. EDELMANt The Rockefeller University, 1230 York Avenue, New York, NY 10021

Contributed by Gerald M. Edelman, August 30, 1990

genital tracts) but remains in ectoderm and endoderm (1113). At neurulation, L-CAM expression diminishes in the neural fold, and a different Ca2"-dependent CAM, N-cadherin, appears. In the adult mouse, L-CAM continues to be expressed in a variety of epithelial tissues, such as skin and intestinal tract. We have generated transgenic mice by using different recombinant DNA constructs with the goal of directing the expression of L-CAM to regions such as the endocrine pancreas where it is expressed (14) and to the central nervous system where it is generally not detected in the adult (11, 12). The rat insulin promoter (RIP) was used to target L-CAM expression to the 8 cells of the islets of Langerhans (15) and the murine neurofilament promoter (NFP) (16, 17) was used to target expression to neural tissues. Chicken L-CAM was chosen because specific antisera that bind chicken, but not mouse, L-CAM (6) were available. Only mice carrying chicken L-CAM genomic sequences expressed the protein and the RIP targeted this expression to pancreatic 8 cells. Surprisingly, however, the murine L-CAM homolog was undetectable in the pancreatic 83 cells of these transgenic mice, but it was present in 8 cells of control animals. Animals expressing chicken L-CAM in the pancreas also expressed it in extrapancreatic tissues, including some in which mouse L-CAM is not expressed. Similarly, a construct containing NFP and genomic L-CAM sequences targeted expression of chicken L-CAM to neural tissues and to various nonneuronal tissues in which L-CAM is expressed but in which NFP has not been reported to function. These results suggest that sequences within the L-CAM gene contain tissue-specific cis-acting elements that can function even when fused to heterologous promoters/enhancers and that such combinations can lead to novel patterns of gene expression.

The tissue-specific expression of the chicken ABSTRACT liver cell adhesion molecule (L-CAM) was studied by generating transgenic mice. The rat insulin II promoter was fused to a chicken L-CAM cDNA or to chicken genomic L-CAM sequences. Mice carrying the cDNA showed no expression of L-CAM. Mice carrying L-CAM genomic sequences showed expression in the .8 cells of the pancreas, suggesting that sequences in introns or in flanking regions are required for expression. Murine L-CAM was undetectable in the ,3 cells ofthe pancreas of those transgenic mice expressing chicken L-CAM and thus appeared to be down-regulated, but expression of the mouse protein was not altered at other sites. Chicken L-CAM was also found in extrapancreatic tissues such as skin, kidney, liver, lung, intestine, blood vessels, and the choroid plexus and leptomeninges of the central nervous system. These findings raised the possibility that the chicken L-CAM gene contains cis regulatory elements that interfere with the specificity of a tissue-specific promoter such as the rat insulin promoter. To test this hypothesis, transgenic mice were produced with a construct containing the murine neurofilament promoter fused to genomic chicken L-CAM sequences. Chicken L-CAM was expressed in the brain and spinal cord, where L-CAM is not normally found, but it was also found in some nonneural tissues (kidney, liver, intestine, lung) in which L-CAM is normally expressed. The combined results suggest that tissue-specific cis-acting elements in the chicken L-CAM gene, when combined with heterologous promoters/enhancers, can generate novel patterns of gene expression.

Morphogenesis and histogenesis require control of various complex processes including cell division, death, motion, adhesion, and differentiation. The interactions of these primary processes of development are place dependent and are under both genetic and epigenetic control (1). The aggregation of cell populations into functional units and various aspects of cell migration are mediated by cell-cell adhesion molecules (CAMs) and cell-substrate molecules. In the past decade, several CAMs have been isolated and characterized (2). Paradigmatic examples include the Ca2+-independent (3) N-CAM (neural cell adhesion molecule), the evolutionary precursor of which gave rise to the immunoglobulin superfamily (4), and L-CAM (liver cell adhesion molecule) (5, 6), which belongs to a family of Ca2+-dependent CAMs known as cadherins (7). Chicken L-CAM is encoded by a gene that spans -10 kilobases (kb) and contains at least 16 exons (8). Its murine homolog was discovered independently in other laboratories and was named uvomorulin (9) or E-cadherin (10). In the mouse, it is expressed at preimplantation stages from the single fertilized cell to the blastocyst (7, 9). After gastrulation, L-CAM is not present in mesoderm (except in the kidney and

MATERIALS AND METHODS Hybrid Gene Construction and Production of Transgenic Mice. The rat insulin II promoter (RIP) (15) was fused separately to 2.9 kb of L-CAM cDNA (RIP/L-CAM1) and a 12.1-kb fragment of the L-CAM gene (RIP/L-CAM2) (see Fig. 1). To construct RIP/L-CAM1, a 2.9-kb fragment of clone pEC1302 (18) was used. This cDNA encodes the complete sequence of chicken L-CAM and most of its amino-terminal precursor sequence; the signal sequence, however, was provided by a segment of N-CAM cDNA that encodes the N-CAM signal peptide and 103 amino acids of the amino terminus of N-CAM. To obtain the 2.9-kb fragment, pEC1302 was digested with BamHI, filled in with Klenow Abbreviations: CAM, cell adhesion molecule; L-CAM, liver CAM; N-CAM, neural CAM; RIP, rat insulin II promoter; NFP, neurofilament (NF-L) promoter. *Present address: Department of Anatomy, University of Melbourne, Victoria, Australia. tTo whom reprint requests should be addressed.

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. 9042

Developmental Biology: Begemann et al. fragment, and then digested with Sal I. The fragment was subcloned into the Sal I and Xho I site of pMSG (Pharmacia) and the latter was blunt-ended with the Klenow fragment to construct pMSG/L-CAM1. pMSG/L-CAM1 was digested with Nhe I and BamHI enzymes to yield a 3.6-kb fragment. RIP/L-CAM1 was constructed by replacing simian virus 40 large tumor antigen sequences in pRIPl-Tag (15) with the 3.6-kb L-CAM cDNA fragment using Xba I and BamHI restriction sites (the latter by partial digestion). For microinjection, the 4.3-kb BamHI/BamHI fragment was purified on a low melting point agarose gel. RIP/L-CAM2 was also constructed by removal of the large tumor antigen from pRIP-Tag (15) and fusion of the 3.5-kb Xba I/Sal I fragment (carrying RIP and pxf 3 vector sequences) to a 12.1-kb fragment of the chicken L-CAM genomic sequence (8). This genomic fragment, spanning 16 exons and 15 introns, had previously been subcloned into Bluescript KS (Stratagene) and was excised with Xba I and Sal I. The entire 15.6-kb construct was linearized with Sal I and used for microinjection. NFP/L-CAM2 was constructed by digestion of NF-68 (gift of Nicholas Cowan, New York University) with HindIII and Sma I to obtain the 1.7-kb fragment containing the promoter and enhancer. The fragment was treated with Klenow enzyme to produce blunt ends and subsequently Xba I linkers were added. This fragment was inserted into the Xba I site of a Bluescript KS derived plasmid containing the 12.1-kb fragment of genomic L-CAM (8) in the BamHI/EcoRI sites of Bluescript KS. DNA was microinjected into pronuclei of zygotes (19) of the following strains of mice: (C57BL/6 x DBA/2)F1 (Animal Resources Center, Western Australia) hybrid strains to generate founders 22 and 49, (C57BL/6J x CBA)F1 hybrid strains (The Jackson Laboratory) to produce founders 397, 402, 425, and 428, NIH FVB/N (Charles River) to generate founders 322, 338, 357, and 422, respectively. Transfer of embryos into pseudopregnant CD-1 mice (Charles River) and analysis of 2- to 3-week-old pups were carried out accordingly (19). Analyses of Tissues and Fluids. Immunocytochemistry was carried out as described (11) with rabbit polyclonal antichicken L-CAM IgG (6), rabbit polyclonal anti-mouse L-CAM (anti-uvomorulin, kindly provided by Rolf Kemler), fluorescein-conjugated goat anti-rabbit IgG (ICN), guinea pig anti-porcine insulin (Linco, St. Louis), and rhodamineconjugated rabbit anti-guinea pig IgG (Sigma). To quantitate L-CAM expression, measured amounts of protein (Bio-Rad) were processed for immunoblotting with 125I-labeled protein A (6). The L-CAM band was cut out of the nitrocellulose filter and assayed in a scintillation counter. Total RNA was isolated from various tissues (20), spotted on slot blots, and hybridized with a labeled probe containing the 12.1-kb chicken L-CAM genomic sequence (8, 20). Blood glucose and serum insulin were determined (21) on samples obtained by cardiac puncture from animals that were starved overnight. Serum glucose levels were determined with a hexokinase test (Ragosin, New York), urinary glucose was measured by using Diastix (Ames, Elkhart, IN), and serum insulin levels were measured by RIA (Linco). Blood glucose values were determined by AccuCheklI/Chem Strips (Boehringer Mannheim). Urinary pH and specific gravity, urine protein and glucose levels, blood urea nitrogen, and serum inorganic phosphate and creatinine levels were measured by the Animal Medical Center, New York.

RESULTS The constructs shown in Fig. 1 were used to generate transgenic mice and a number of pedigrees were established. The cDNA construct RIP/L-CAM1 was used to establish two

9043

Proc. Natl. Acad. Sci. USA 87 (1990) RIP/L-CAM 1 5 UT N-CAM

ATG

F

i

Bam Hi

COOH SV 40

NH2

. .W-.

RIP P

Xba 1 Nhe

-

II 3 UT

I

RIP/L-CAM 2

ATG i Xba 1

Barn Hi

KD

5 UT

Sal I

AAAA

_

COOH AAAA

NH2

_ _T tk~~~~~~~~~~~~~~~[ 2

3

5

Kb

7

9

4 6 8

11

13

10 12

15

16

14

Sal

Barn HI

5 UT BAT

NFP/L-CAM 2 3

2 Sail 1

Xba I

Xba 1

COOH AAAA

NH2 Kb

5 4

9 6 8

11 10

13 15

12

14

16 Sal

FIG. 1. Hybrid chicken L-CAM constructs used to generate transgenic mice. The construct RIP/L-CAM1 contains the rat insulin II promoter fused to chicken L-CAM cDNA (22), whereas the construct RIP/L-CAM2 contains the RIP fused to chicken L-CAM genomic sequence (8). The RIP and the murine NFP are indicated, as are 3' and 5' untranslated sequences (UT) of L-CAM, coding sequences of chicken N-CAM (solid boxes) and simian virus 40 (SV40) splice and polyadenylylation site, intron sequences or sequences 3' of the polyadenylylation site of the chicken L-CAM gene (heavy lines), and prokaryotic sequences (narrow line). Figure is not to scale.

independent lines (322 and 338), and the genomic construct RIP/L-CAM2 was used to generate three lines (22, 49, and 402) and two founders (357 and 397). Founder 357 (female) gave no transgenic offspring among 45 pups examined from four litters, and was, therefore, assumed to be germ-line mosaic. Founder 397 (male) was sterile. Founders 357 and 397 were sacrificed at 7 and 5.5 months of age, respectively. The construct NFP/L-CAM2 was used to establish three lines (422, 425, and 428). Immunohistochemistry of Islets of Langerhans of Animals Carrying Different RIP/L-CAM Constructs. Offspring of the established lines and founders with RIP/L-CAM1 and RIP/ L-CAM2 were sacrificed and the pancreases were examined for L-CAM expression by indirect immunohistochemistry (Fig. 2 a-f). Chicken L-CAM was detected on the cell surface of all pancreatic p cells of animals from lines 22, 49, and 402 carrying RIP/L-CAM2 (Fig. 2 a-c). In the mosaic founders 357 and 397, p cells expressing chicken L-CAM were seen in clusters in the islets of Langerhans. No expression ofchicken L-CAM was observed in the islets of Langerhans of C57BL/6J mice or in those of three offspring of line 322 and two offspring of line 338 bearing the cDNA construct RIP/ L-CAM1 (Fig. 2 d-f). No obvious abnormality of islet morphology and no biochemical evidence ofdiabetes mellitus were observed in mice expressing chicken L-CAM. Glucagon-producing a cells, normally localized in the periphery of the islets of Langerhans, did not costain for chicken L-CAM (data not shown). Down-Regulation of Mouse L-CAM in .8 Cells. Mouse L-CAM was detectable on the surface of cells of the exocrine as well as the endocrine pancreas of nontransgenic animals (Fig. 2g) and animals bearing RIP/L-CAM1. Transgenic mice bearing RIP/L-CAM2, however, showed a loss of this expression on the surface of the , cells (see Fig. 2h and i). This down-regulation below our level of detection was observed uniformly in two offspring of founder 402 and in patches in an independent founder, 357. Expression of Chicken L-CAM in Extrapancreatic Tissues. Antibodies that recognize chicken L-CAM but not murine L-CAM were used to localize the protein in extrapancreatic tissues (Fig. 3). In contrast to nontransgenic controls (Fig. 3

9044

Developmental Biology: Begemann et al.

FIG. 2. Expression of chicken L-CAM and mouse L-CAM in islets of Langerhans of transgenic mice. Pancreas sections from first generation offspring of founder 402 bearing RIP/L-CAM2 as shown in matched bright field and double stained (a), for chicken L-CAM (b), and for insulin (c). Comparable sections of the pancreas from a first generation offspring of founder 338 bearing RIP/L-CAM1 as shown in d-f. (g-i) Fluorescein staining of pancreatic sections. (g) Uniform expression of mouse L-CAM in the pancreas C57BL/6J mice. (h and i) Down-regulation of mouse L-CAM in islet cells of two different transgenic offspring of founder 402, both of which carried RIP/L-CAM2. (Bars = 2 gm.)

a and c), the offspring of founder 402 showed membrane staining in liver parenchymal cells (Fig. 3b), especially on membranes of apposing hepatocytes, as well as in glomeruli and the tubular epithelium of the kidney (Fig. 3d). Cellsurface staining was pronounced in glomeruli, presumably on endothelial cells of capillary loops. Staining of the tubular epithelium was localized to the basolateral surface. Sections of brains of founder 402 and of founder 357 offspring showed staining in the choroid plexus as well as in blood vessels and leptomeninges, but not in neuronal tissue (data not shown). Sections of the intestines obtained from RIP/L-CAM2-

bearing animals showed cell-surface staining of the epithelial cell layer lining the villi and crypts along the intestine as illustrated for founder 357 in Fig. 3e. The staining pattern shows patches of chicken L-CAM-expressing tissue distributed within nonexpressing tissue. Tissue extracts of the offspring of founder 402 were also tested for L-CAM expression by immunoblots (Fig. 4). Chicken L-CAM was detected in pancreas extracts at low levels (Fig. 4 A and B) relative to extracts of brain, liver, and kidney (Fig. 4A), and intestine and spinal cord (data not shown); it was not detected in blood cells or in serum. Relative to the amount of L-CAM in chicken liver, the amounts of chicken L-CAM in transgenic mice were '1% in total pancreas, 10% in kidney, 5% in liver, and 2% in brain. Founder 357 showed expression of chicken L-CAM in brain, kidney, and intestine, whereas founder 397 showed extrapancreatic expression only in liver. Consistent with their failure to express chicken L-CAM in the pancreas, two animals from two different lines carrying RIP/L-CAM1 showed no expression in any tissue examined by immunohistochemistry or by immunoblots (Fig. 4). Furthermore, liver, kidney, and brain extracts of five adult

Proc. Natl. Acad. Sci. USA 87 (1990)

FIG. 3. Expression of chicken L-CAM in extrapancreatic tissues of RIP/L-CAM2 carrying animals. Sections of liver from a C57BL/6J mouse (a) and a second generation of founder 402 offspring (b) and sections of kidney from the C57BL/6J control (c) and transgenic animal (d) were stained to detect chicken L-CAM (11). (e) Expression of chicken L-CAM in the intestinal epithelium of mosaic founder 357 carrying RIP/L-CAM2 (longitudinal section). (Bars = 5 ,um.)

RIP/L-CAM1-bearing animals (three from line 322, including its founder, and two from line 338) contained no chicken L-CAM as assessed by immunoblotting. To determine whether the lack of expression of RIP/LCAMi-carrying animals was due to pretranslational or posttranslational control, the presence of chicken L-CAM mRNA was assessed by slot blot analyses. Hybridization with the chicken L-CAM genomic sequence yielded no signal in samples from pancreas, liver, or brain of C57BL/6J mice, or from samples of the same tissues from RIP/L-CAM1carrying offspring of founders 322 and 338. However, chicken L-CAM mRNA was detected in all samples obtained from transgenic offspring of founder 402, which carried RIP/LCAM2. These results demonstrated that chicken L-CAM mRNA is absent in tissues of RIP/L-CAM1-carrying animals but is present in tissues of RIP/L-CAM2-carrying animals. Expression of Chicken L-CAM in NFP/L-CAM2-Bearing Mice. To assess whether ectopic expression would also be found when the L-CAM gene was fused to another promoter, extracts of various organs obtained from animals carrying the transgene NFP/L-CAM2 were examined for L-CAM expression. As expected, ectopic expression was observed in extracts of brain and spinal cord (Fig. 4C), but in all three lines, chicken L-CAM was also observed in nonneural tissues including liver, kidney, intestine, lung, and diaphragm; no expression was seen in spleen and heart. Immunohistochemistry (Fig. 5) indicated that chicken L-CAM in the offspring of line 425 was localized to the basolateral surface of the tubular epithelium of the kidney and to the epithelium in the intestine and the bronchial epithelium of the lung, but it was not detected in the lung parenchyma or in the spleen, striated, smooth, or cardiac muscle (data not shown). Urine and serum analyses of RIP/L-CAM2- and NFP/L-CAM2-bearing mice gave no indication of kidney dysfunction.

Developmental Biology: Begemann et A A

1

2 3

5 6

7

8

9

4 5 6

7

8

9

4

10

Proc. Natl. Acad. Sci. USA 87 (1990)

9045

l

f

11697.4- *

66 - -:4

B

1

2

3

116-

1

11

FIG. 5. Expression of chicken L-CAM in NFP/L-CAM2-bearing animals. Sections stained for chicken L-CAM were as follows: (a) olfactory bulb, (b) intestine, (c) lung of a transgenic offspring of founder 425. (Bars = 2 Am.)

n

97.466-K C

10

4

2

3

4

5

6

7

8

9

10

11

205-

116-

97.466

_W

_

mE

FIG. 4. Expression of chicken L-CAM in tissues of transgenic mice. (A and B) Immunoblots of tissue extracts from mice carrying RIP/L-CAM1 and RIP/L-CAM2 with exposure times of 2 days (A) and 14 days (B). Chicken L-CAM in extracts of tissues from an offspring of founder 402 carrying RIP/L-CAM2 (lanes 4-7) and an offspring of founder 338, carrying RIP/L-CAM1 (lanes 8-11), is compared to chicken liver extracts. Tissues include pancreas (lanes 4 and 8), liver (lanes 5 and 9), brain (lanes 6 and 10), and kidney (lanes 7 and 11). Each lane contained the same amount of total protein except for lane 2, which was a 1:10 dilution of the sample in lane 1, and lane 3, which was a 1:100 dilution of that in lane 1. The immunoreactive components smaller than the 124-kDa L-CAM correspond in size to known proteolytic fragments of L-CAM (6). (C) Expression of chicken L-CAM in tissues of a NFP/L-CAM2-bearing animal. Immunoblots of extracts of second generation offspring of founder 425. Equal amounts of protein from brain (lane 1), liver (lane 2), kidney (lane 3), heart (lane 4), lung (lane 5), spleen (lane 6), intestine (lane 7), spinal cord (lane 8), and diaphragm (lane 9). Chicken liver (lane 10) and brain (lane 11) served as positive and negative controls, respectively. Exposure time was 2 days; the intense signals in lanes 7 and 10 were due to the much greater expression of chicken L-CAM in the intestine of the transgenic mouse and in normal chicken liver.

DISCUSSION In this study, constructs made from chicken L-CAM genomic DNA, but not those from cDNA, gave rise to both the synthesis of chicken L-CAM and its mRNA in transgenic mice, suggesting that sequences in introns or 3' flanking segments are required for expression of the L-CAM gene. In RIP/L-CAM2-bearing animals, the expression of chicken L-CAM in the cells of the pancreas was accompanied by down-regulation of mouse L-CAM. Chicken L-CAM was also expressed in extrapancreatic tissues where L-CAM is normally expressed in mice but where RIP has not been reported to be active; it was also found in tissues in which RIP is not active nor is L-CAM normally expressed (11, 23). Similarly, NFP/L-CAM2-bearing animals showed expression of chicken L-CAM in the nervous system as expected, but also in tissues of predominantly epithelial origin in which NFP is not active. A summary of the sites of expression seen in transgenic lines and founders is shown in Table 1. These results suggest that sequences in the chicken L-CAM gene contain tissue-specific regulatory elements that can influence the specificity of other control elements such as RIP and NFP. Previous reports have shown that introns and 3' flanking sequences are required for gene expression in

transgenic mice (24, 25), although some cDNAs encoding other proteins fused to RIP have been used successfully to target the expression to pancreatic p cells in transgenic mice (26-29). Our RIP/L-CAM1 construct lacked the 5' and 3' untranslated sequences present in RIP/L-CAM2 and it also contained a short N-CAM sequence (Fig. 1) that might influence expression. Studies with this N-CAM/L-CAM chimeric construct in transfected cells, however, have shown that the protein is synthesized and that the N-CAM segment is properly removed with the L-CAM precursor sequence so that normal L-CAM is expressed on the cell surface (18). In all tissues in which chicken L-CAM was detected, it was localized on the cell surface. In transgenic offspring, it was on all cells of the same type but, in two mosaic founders, cells expressing chicken L-CAM formed discrete clusters as seen in the islets of Langerhans and the intestine. Such clusters might have arisen by aggregation and sorting out of the cells expressing chicken L-CAM from other cells as has been shown in vitro (30); alternatively, they may be derivatives of a common progenitor (31). The mechanism responsible for the down-regulation of mouse L-CAM in the ,8 cells of transgenic animals containing RIP/L-CAM2 is unknown. Similar down-regulation has been observed for other genes in transgenic mice (32-36) and competition for trans-acting factors, chromosomal position effects, or negative feedback have been suggested as possible causes. Whatever mechanism is involved in the downregulation of L-CAM in islet cells described here, it must Table 1. Expression of chicken L-CAM in tissues of transgenic mice

NFP/L-CAM2t

RIP/L-CAM2* 402 Pancreas Exocrine Endocrine Liver

Kidney Intestine Lung Brain

357

-

-

-

+ + + + +

+ + + +

+ +

-

-

Leptomeninges Choroid plexus Skin

+ + +

+ +

Heart

-

-

Neuronal

397

-

422

425

428

+ + +

+ + + +

+ +

+

+

+

+ -

-

-

-

Skeletal muscle

Spleen Chicken L-CAM was detected by immunoblotting or immunohistochemistry in transgenic mice carrying the rat insulin II promoter (*) or the murine neurofilament (NF-L) promoter (t) and chicken genomic DNA. 357 and 397 are mosaic founders.

9046

Developmental Biology: Begemann et al.

occur under special thresholding conditions because it was not seen in other tissues, such as the kidney, in which chicken L-CAM was strongly expressed. No gross morphologic perturbations were detected in any tissue in which chicken L-CAM was detected. The presence of endogenous CAMs might render histogenesis refractory to perturbation by additional CAMs or overexpression of CAMs, an interpretation consistent with reported results in other systems (37, 38). Moreover, an excess of a CAM is not necessarily a test of its function (37). Alternatively, the L-CAM transgene used in the present experiments may be expressed too late or at too low a level to influence development in some tissues. In all lines carrying chicken genomic L-CAM sequences fused to RIP or NFP, L-CAM was detected in tissues of predominantly epithelial origin (summarized in Table 1) in which the promoters are not known to be active, but where L-CAM is normally expressed (7, 11, 23). It seems unlikely that position effects (39) account for this expression because the same pattern was seen in three separate lines and two founders carrying RIP/L-CAM2 and in three lines carrying NFP/L-CAM2. These observations suggest that the 12.1-kb fragment of the chicken L-CAM gene used in these studies contains cis-acting elements responsible for tissue-specific expression of L-CAM, even when fused to heterologous promoters. In the RIP/L-CAM2 transgenic animals, however, chicken L-CAM was not detected in the exocrine pancreas, where endogenous L-CAM is normally expressed. Moreover, in RIP/L-CAM2-bearing animals, chicken L-CAM was also expressed in other extrapancreatic tissues, such as kidney glomeruli, endothelial cells of blood vessels, the choroid plexus, and the leptomeninges of the brain of adult animals (Fig. 4). RIP has not been reported to be active in these tissues and L-CAM is not normally expressed in these tissues in either chickens (11) or mice (23). Thus, the combination of RIP and L-CAM genomic sequences leads to expression at new sites. While the expression of some RIP fusion genes is restricted to the pancreas (15, 26, 29), others are expressed in extrapancreatic tissues (26, 28, 29, 40). It has also been shown that reporter genes or other cis-acting elements can interfere with the tissue specificity of some enhancers and change the expression of certain transgenes or the tropism of viruses (41, 42). The present findings taken together with those of others (41, 42) suggest that tissue-specific enhancers might arise by recombination of preexisting elements that by themselves have different expression patterns. Indeed, during evolution, nonhomologous recombination of cis-acting elements controlling the expression pattern of morphoregulatory molecules (43) such as CAMs or cell-substrate molecules might change their expression in time and space, contributing to alterations in the form of the developing embryo. Such rearrangements in the genome might provide an additional molecular mechanism for heterochrony (1). Examination of this hypothesis should be facilitated by producing additional transgenic animals with ectopic CAM expression that show perturbation in gross morphogenesis or histogenesis. We thank Dr. Douglas Hanahan (University of California, San Francisco) and Dr. Nicholas Cowan (New York University) for providing the plasmids pRIP1-Tag and pNF-68, respectively, and Dr. Rolf Kemler (Max-Planck-Institut, Freiburg, F.R.G.) for antibodies to uvomorulin. Dr. Joseph A. Gally provided invaluable guidance and criticism. Mses. Amy Kreisberg, Talia Spanier, Stephanie

Proc. NatL Acad Sci. USA 87 (1990) Blank, and Deborah Glassman and Mr. Will Greer provided excellent technical assistance. This work was supported by U.S. Public Health Service Grants HD 09635, HD 16550, and DK 04256. M.B. is a Fellow of the Charles H. Revson Foundation. 1. Edelman, G. M. (1988) Topobiology: An Introduction to Molecular Embryology (Basic Books, New York). 2. Edelman, G. M. & Cunningham, B. A. (1990) Cold Spring Harbor Symp. Quant. Biol. 55, in press. 3. Brackenbury, R., Rutishauser, U. & Edelman, G. M. (1981) Proc. Natl. Acad. Sci. USA 78, 387-391. 4. Edelman, G. M. (1987) Immunol. Rev. 100, 11-45. 5. Bertolotti, R., Rutishauser, U. & Edelman, G. M. (1980) Proc. Natl. Acad. Sci. USA 77, 4831-4835. 6. Gallin, W. J., Edelman, G. M. & Cunningham, B. A. (1983) Proc. Natl. Acad. Sci. USA 80, 1038-1042. 7. Takeichi, M. (1988) Development 102, 639-655. 8. Sorkin, B. C., Hemperly, J. J., Edelman, G. M. & Cunningham, B. A. (1988) Proc. Natl. Acad. Sci. USA 85, 7617-7621. 9. Hyafil, F., Morello, D., Babinet, C. & Jacob, F. (1980) Cell 21, 927-934. 10. Ogou, S.-I., Yoshida-Noro, C. & Takeichi, M. (1983) J. Cell Biol. 97, 944-948. 11. Thiery, J.-P., Delouv6e, A., Gallin, W. J., Cunningham, B. A. & Edelman, G. M. (1984) Dev. Biol. 102, 61-78. 12. Crossin, K. L., Chuong, C.-M. & Edelman, G. M. (1985) Proc. NatI. Acad. Sci. USA 82, 6942-6946. 13. Takeichi, M. (1987) Trends Genet. 3, 213-217. 14. Eidelman, S., Damsky, C. H., Wheelock, M. J. & Damjanov, 1. (1989) Am. J. Pathol. 135, 101-110. 15. Hanahan, D. (1985) Nature (London) 315, 115-122. 16. Lewis, S. A. & Cowan, N. J. (1985) J. Cell Biol. 100, 843-850. 17. Byrne, G. W. & Ruddle, F. H. (1989) Proc. Natl. Acad. Sci. USA 86, 5473-5477. 18. Edelman, G. M., Murray, B. A., Mege, R.-M., Cunningham, B. A. & Gallin, W. J. (1987) Proc. NatI. Acad. Sci. USA 84, 8502-8506. 19. Hogan, B., Costantini, F. & Lacy, E. (1986) Manipulating the Mouse Embryo (Cold Spring Harbor Lab., Cold Spring Harbor, NY). 20. Ausubel, F. A., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, K., eds. (1990) Current Protocols in Molecular Biology (Greene and Wiley-Interscience, New York). 21. Epstein, P. N., Overbeek, P. A. & Means, A. R. (1989) Cell 58, 10671073. 22. Gallin, W. J., Sorkin, B. C., Edelman, G. M. & Cunningham, B. A. (1987) Proc. Natl. Acad. Sci. USA 84, 2808-2812. 23. Vestweber, D., Kemler, R. & Ekblom, P. (1985) Dev. Biol, 112, 213-221. 24. Brinster, R. 1L., Allen, J. M., Behringer, R. R., Gelinas, R. E. & Palmiter, R. D. (1988) Proc. Natl. Acad. Sci. USA 85, 836-840. 25. Trudel, M. & Costantini, F. (1987) Genes Dev. 1, 954-961. 26. Bohme, J., Haskins, K., Stecha, P., vanEwijk, W., LeMeur, M., Gerlinger, P., Benoist, C. & Mathis, D. (1989) Science 244, 1179-1183. 27. Edwards, R. H., Rutter, W. J. & Hanahan, D. (1989) Cell 58, 161-170. 28. Roman, L. M., Simons, L. F., Hammer, R. E., Sambrook, J. F. & Gething, M.-J. H. (1990) Cell 61, 383-396. 29. Miller, J., Daitch, L., Rath, S. & Selsing, E. (1990) J. Immunol. 144, 334-341. 30. Friedlander, D. R., Mege, R.-M., Cunningham, B. A. & Edelman, G. M. (1989) Proc. Natl. Acad. Sci. USA 86, 7043-7047. 31. Ponder, B. A. J., Schmidt, G. H., Wilkinson, M. M., Wood, M. J., Monk, M. & Reid, A. (1985) Nature (London) 313, 689-691. 32. Alexander, W. S., Schrader, J. W. & Adams, J. M. (1987) Mol. Cell. Biol. 7, 1436-1444. 33. Shani, M., Dekel, I. & Yoffe, 0. (1988) Mol. Cell. Biol. 8, 1006-1009. 34. Uematsu, Y., Ryser, S., Dembic, Z., Borgulya, P., Krimpenfort, P., Berns, A., von Boehmer, H. & Steinmetz, M. (1988) Cell 52, 831-841. 35. Rosenbaum, H., Webb, E., Adams, J. M., Cory, S. & Harris, A. W. (1989) EMBO J. 8, 749-755. 36. Ferrick, D. A., Sambhara, S. R., Ballhausen, W., Iwamoto, A., Pircher, H., Walker, C. L., Yokoyama, W. M., Miller, R. G. & Mak, T. W. (1989) Cell 57, 483-492. 37. Kintner, C. (1988) Neuron 1, 545-555. 38. Elkins, T., Zinn, K., McAllister, L., Hoffman, F. M. & Goodman, C. S.

(1990) Cell 60, 565-575.

39. Allen, N. D., Cran, D. G., Barton, S. C., Hettle, S., Reik, W. & Surani, M. A. (1988) Nature (London) 333, 852-855. 40. Alpert, S., Hanahan, D. & Teitelman, G. (1988) Cell 53, 295-308. 41. Rochford, R., Campbell, B. A. & Villarreal, L. P. (1987) Proc. Natl. Acad. Sci. USA 84,449-453. 42. Russo, A. F., Crenshaw, E. B., III, Lira, S. A., Simmons, D. M., Swanson, L. W. & Rosenfeld, M. G. (1988) Neuron 1, 311-320. 43. Edelman, G. M. (1988) Biochemistry 27, 3533-3543.