Isolation and Characterization of Renin-expressing Cell Lines from Transgenic Mice Containing a Renin-Promoter Viral Oncogene Fusion Construct*

We constructed transgenic mice containing a renin- promoter SV40 T antigen fusion transgene with the intention of inducing neoplasia in renin-expressing cells and isolating renin-expressing cell lines in vitro. We examined six kidney tumors from mice represent- ing three different transgenic lines and found they expressed their endogenous renin gene. Initially, five nonclonal kidney tumor-derived cell lines were estab- lished which expressed their endogenous renin gene in addition to the transgene. They retained active renin intracellularly and constitutively secreted an inactive form of renin (prorenin). One of these was cloned to This high level of

Although the cells contained an equal proportion of active and inactive renin, the species constitutively secreted into the media was predominantly (95%) prorenin. However, active renin secretion was stimulated 2.3-and 4.6-fold by treatment with 8-bromo-CAMP after 4 and 15 h, respectively. In addition, the presence of multiple secretory granules was confirmed by ultrastructural analysis. These cells, which express renin mRNA and can regulate secretion of active renin, should provide an excellent tool for studying renin gene regulation and secretion. Furthermore, these mice should provide a useful source for the establishment of renin-expressing cell lines from a variety of renin-expressing tissues.
Mouse renin genes (Ren-l', Ren-ld, and Ren-2) exhibit a complicated pattern of gene-specific expression (l-3). Although all three murine renin genes share greater than 96% homology, they exhibit overlapping yet distinct tissue-specific expression profiles (2) transgenes have shown that the cis-acting elements controlling renin expression in various organ sites are closely linked to the renin coding regions (4-7). In addition, a Ren-2 genomic segment has been identified which confers a strainspecific, estrus cycle-dependent pattern on adrenal renin expression (6). However, the detailed analysis of a large number of genomic constructs by transgenic mice may be impractical because the number of constructs that can be efficiently examined is limited.
Transfection of mouse and human renin genes into L cells, Chinese hamster ovary cells, and AtT-20 cells has permitted isolation of cell lines which contained immunoreactive renin (8,9). Furthermore, data suggesting the presence of negative and positive &-acting regulatory elements in the mouse renin genes transfected into noncognate cells has been reported, although some of these data are conflicting (10-13). However, such noncognate cells do not express their endogenous renin genes and most likely do not contain the correct complement of truns-acting factors required for gene-specific or tissuespecific expression. In addition, although the isolation of renin-expressing cell lines has been reported, many of these reports contain conflicting data regarding the quantitation of renin mRNA. This has prompted us to investigate the possibility of isolating tissue culture cells from various organ sources that express their endogenous renin gene.
We undertook an approach using transgenic mice to direct the expression of a nuclear oncogene (SV40 T antigen) specifically to renin-expressing cells. This was accomplished by fusing 4.6 kb' of the 5' flanking sequence from the Ren-2 gene to the T antigen structural gene. T antigen has the capacity to induce a neoplastic phenotype in certain cell types when expression is directed by certain tissue-specific promoters (14-16). Accordingly, we reasoned that it might be possible to isolate renin-expressing tumors and tumor-derived cell lines if T antigen expression was directed to reninexpressing cells. Such an approach has been successful in the isolation of insulin-secreting pancreatic P-cell lines when T antigen expression was driven by an insulin promoter (17). Transfection of SV40 viral DNA into JG cells isolated from a human JG cell tumor has been reported to yield an immortalized human renin-producing JG cell line, although the levels of renin secreted by these cells decreased over time in culture (18,19). In addition, renin-expressing cells have been reported from a testicular Leydig cell tumor (20,21).
We have previously reported that the 4.6-kb 5' flanking  sequence employed in this construct confers the appropriate tissue-and cell-specific expression profile characteristic of the mouse renin genes throughout fetal development and in adults (22,23). Here we report the isolation and characterization of nonclonal and clonal renin-expressing cell lines isolated from a primary kidney tumor and ascites.

EXPERIMENTAL PROCEDURES
Production of Transgenic Mice-The renin T antigen fusion gene, PR~~,~TAG (Fig. l), consists of 4.6 kb of 5' flanking sequence of Ren-2d cloned upstream of the SV40 T antigen structural gene ( Fig. 1) as described (22,23). Production of transgenic mice with this construct was previously described (23).  (24). Northern blots were probed for T antigen using an [a-32P]GTP-labeled T3 polymerase antisense T antigenic transcript from the plasmid T3T7-TAG (obtained from Dr. S. Efrat, Cold Spring Harbor-Laboratory) and for renin using an la-32P1GTP-labeled SP6 antisense renin transcriut from the plasmid pSL&l (22). Dot-blots were performed as described (24. Renin Actiuity Assays-Renin activity was measured by conversion of exogenously added angiotensinogen to angiotensin I followed by radioimmunoassay for angiotensin I as previously described (25,26).
Tumor samples and tissue culture cells were homogenized in 0. were added to a lOO-~1 aliquot as above. Both the untrypsinized and trypsinized samples were then incubated with the substrate, and the angiotensin I produced was radioimmunoassayed. Experiments to examine the time course (30,60,90,120, and 180 min) and temperature (4 "C through 37 "C) for optimal trypsin activation using trypsin-Sepharose were performed on mouse tissues, following which the l-h time point and 4 "C temperature were selected (data not shown The cells were fixed in 3% glutaraldehyde, 0.1% tannic acid at 25 "C after washing twice with 3'7 "C Sorensen's buffer, pH 7.2 (0.2 M sodium phosphate). They were postfixed in 2% OsOl for 15 min at 4 "C and rinsed in buffer before dehydration and EPON embedding as described previously (28). Cells were sectioned and visualized after uranyl acetate staining by standard methods.

Characteristics of the Ren-Tag
Transgenic Mice-Transgenie mice were constructed with a fusion transgene consisting of 4.6 kb of the 5' flanking region of Ren-2d and the large T antigen gene from SV40 in an effort to initiate neoplastic transformation of renin-expressing cells (Fig. 1). In all, eight founder mice were identified by Southern blot analysis of tail DNA (data not shown), three of which either did not breed or have not developed any overt tumors. Transgenic mice from the other transgenic lines (numbers T3, T4, T6, T7, and T8) gave rise to tumors with an onset between 4 and 8 months of age. These animals exhibited tumors involving the kidney, subcutaneous tissue, adrenal gland, and testes (  sentially the same spectrum of tumor types. The difference between these lines is not necessarily surprising considering the stochastic nature of tumorigenesis. Although transgene position effects may cause a difference in the tumor frequency among lines, we feel that differences in the level of expression at the various sites among lines is a more likely explanation. Furthermore, we have previously reported qualitatively correct tissue-and cell-specific expression of this transgene (22,23). Kidney Tumor Histology-Although kidney tumors involving the renal parenchyma were the most abundant type of kidney tumor, others involving the kidney capsule and perirenal fat were also seen. The latter generally developed around 4 months of age while the intraparenchymal tumors, all but one of which were unilateral, generally did not develop until 5-8 months of age. The intraparenchymal tumors were usually large, measuring up to a few centimeters in their maximum dimension, often completely replacing the normal renal parenchyma. They presented mixed histological patterns ranging from adenocarcinomas to spindle cell sarcomas with some intermediate or transitory forms. The adenocarcinomas were composed mostly of relatively small glandular structures lined by plump low columnar to cuboidal epithelial cells with pleomorphic and hyperchromatic nuclei, and the stroma contained some small spindle cells (Fig. 2B). The intermediate form frequently assumed an angiosarcomatous pattern with nests and cords of spindle and polygonal tumor cells outlined by an irregular network of vascular channels (Fig. 2, A and C). The sarcomas were composed of loosely arranged neoplastic spindle cells with scattered multinucleated tumor giant cells (Fig.  2, D and  that they originated in nephroblastic tissue or from vascular smooth muscle. Support for the latter stems from the observation that adult mice that did not show evidence of overt renal tumors often demonstrated marked renal vascular smooth muscle cell hyperplasia (data not shown).
In addition to a renal tumor, the T7-1-6 mouse developed an ascites which probably originated from release of tumor and blood cells from the primary kidney tumor. Histological analysis of this fluid revealed atypical tumor cells, red blood cells, and lymphocytes. RNA isolated from the ascites cells revealed a low level of renin and transgene mRNAs (data not shown).
Tumoral Renin Expression-Renin and T antigen expression were next assayed to determine whether tumorigenesis was the direct result of T antigen expression in a reninexpressing cell. As shown in Fig. 3, expression of the endogenous Ren-1' gene and the transgene was evident in the primary tumor (loA) and in tumors propagated in athymic nude mice (Nu-1). In these mice, the Ren-I' gene is the naturally occurring renin allele and encodes the circulating renin. Renin mRNA-containing cells were easily detected in tumor sections by in situ hybridization although the levels of renin message per cell were highly variable (data not shown). In addition, we examined two other kidney tumors from T7 mice, one tumor from a T3 mouse, and bilateral tumors from the T6 founder, and all of them were found to contain renin mRNA.
Renin activity was detected in extracts of the primary tumors and tumors grown in nude mice (Fig. 4A). Since little additional activity was detected after limited proteolysis by trypsin, a procedure which has been shown to activate prorenin efficiently (25,26), we suggest that the primary species of renin stored by these tumor samples was active renin. Similarly, active renin was the predominant form of renin found in an independent primary kidney tumor from a different T7 mouse.
Isolation and Characterization of Cell Lines-In order to isolate renin-expressing cells in uitro, both lobes of the primary kidney tumor from T7-1-6 were first propagated in nude mice and then were screened for renin expression. The resultant tumors and the primary ascites cells were then seeded into culture dishes. In most cases, the dissociated tumor cells quickly attached to the culture flasks, although it took several months to establish them as continuously growing cell lines. The origin of each cell line is described in Table II. When sufficient numbers of cells were harvested, Northern blots (Fig. 3) and T antigen immunocytochemistry were performed. The results showed that all the culture lines expressed the endogenous renin gene mRNA (Fig. 3A, TCl-5) at levels as great as 25% of the primary tumors (Fig. 3B). The size of the transgene mRNA (Fig. 3C) and the endogenous renin gene mRNA (Fig. 3A)   1-A variability in renin mRNA levels in subcultures from some cell lines recovered from frozen stocks. In addition, there was a low percentage of non-T antigen-containing cells within some of the cultures when examined immunocytochemically, and in situ hybridization revealed fluctuations in renin mRNA across the hybridization slide. Nevertheless, some sections of the slide, most likely containing clonal populations, had equal hybridization intensity among cells. The above observations, and the fact that the level of renin expression in these nonclonal cultures decreased over long periods of continuous culture, prompted us to isolate clonal cell lines. Single-well isolated colonies from TC-2 were picked and allowed to grow in culture. One of these colonies (clone 4.1) expressed renin mRNA at a level approximately IO-fold greater than the original nonclonal line and maintained this level, with only minor variability among culture flasks, for over 3 months of continuous culture (Fig. 311). Fig. 5 illustrates the cellular morphology and nuclear localization of immunoreactive T antigen. Since they uniformly contained T antigen (Fig. 5R) and maintained high level expression of renin mRNA after long term continuous culture. we have considered this line homogeneous.
Renin Synthesis in Tumor-derived Cell Lzncs-The renin activity measured in the cell extracts and media of the nonclonal cells is presented in Fig. 4, H and C. Although the amount of renin was highly variable between cell lines, the culture cells, like the tumors, appeared to store active renin with very little prorenin (Fig. 4H). In contrast, the level of prorenin in the media exceeded that of active renin by more than lo-fold (Fig. 4C). These results suggest that under the conditions described herein, without stimulating the regulated pathway, the cells constitutively secrete prorenin. These experiments were repeated with clone 4.1. In this set of experiments, the cells contained an approximate equal proportion of renin and prorenin (SE. is *lo%). However, similar to the nonclonal cultures, 95% of the renin in the media was prorenin consistent with the notion that the cells constitutively secrete prorenin. The reason for the difference in intracellular renin storage between the clonal and nonclonal cell lines is unclear; however, variations in the level of TC-I TC-5 TC-3 TC-I AtT-20 cells, cells which contain both constitutive and regulated secretory pathways (8). Renin mRNA transcription was stimulated approximately 3-fold after 15 h of treatment (Fig. 6, inset) possibly due to depletion of intracellular stores of renin. Ultrastructural analysis (Fig. 7) of clone 4.1 cells revealed the presence of numerous secretory granules morphologically similar to secretory granules of renin-secreting cells of the kidney (29,30). These results, when taken together, strongly suggest that these cells can regulate active renin secretion.

Tumorigenesis
in the Renin-T Antigen Transgenic Mace-We set out to isolate renin-expressing cell lines derived from renal and extrarenal sites of renin synthesis with the aim of developing a library of cell lines that could be used to study differential regulation of the mouse renin genes at the various organ sites. So far, we have succeeded in isolating reninexpressing cell lines derived from the kidney. Renin expression in several human JG cell tumors, in congenital mesoblastic nephroma and in certain Wilms' tumors, has been decribed previously (31)(32)(33)(34). To our knowledge, only a few cell lines have been previously reported to express their endogenous renin mRNA including human chorionic cells, human tumoral JG cells, and mouse Leydig tumor cells (18)(19)(20)(21)35). However, a detailed analysis of renin secretion from some of these lines has been hindered because the level of intracellular renin and renin released from them decreased after long term continuous culture.
Although the histopathology of the kidney tumors was consistent with a nephroblastic origin, the positive identification of the cell types which gave rise to them is uncertain. A microscopic examination of nontumorous kidneys from at least two transgenic lines revealed marked hyperplasia in the renal vasculature affecting smooth muscle cells and mesangial cells (data not shown). The site of this pathology is consistent with the pattern of renin expression. For instance, renin expression has previously been shown to occur throughout the renal arterial tree during fetal development (22), and cells further up in the arterial tree from JG cells can be recruited to express renin under certain conditions in adult rodents (36,37). In addition, renin synthesis has been reported in cultured glomerular mesangial cells (38). These cells may have undergone the secondary events necessary to transform them to a tumorigenic phenotype.  (39), and synthesis and release of various proteins produced by cells through these pathways has been shown to occur widely (40). It is thought that only specialized cells are capable of secreting proteins via the regulatory pathway while constitutive secretion may be applicable to a wide range of cell types. Nevertheless, positive identification of the cells as JG cells will require further analysis.
Eight of forty-six (17%) T7 transgenic mice exhibited overt intraparenchymal kidney tumors, suggesting that a stochastic event was required in order for the initiation of a tumor. Stochastic tumor formation in many other transgenic mouse systems has been previously observed and probably reflects the fact that tumorigenesis is a multistep process (41). It is possible that expression of T antigen in the renin-expressing cells serves to predispose the cells to tumorigenesis.
It is also likely that parameters which affect the frequency of tumorigenesis in these mice include the concentration of T antigen and the rate at which other secondary mutations occur.
pressing cell lines from the kidney in uitro. That the cells are capable of regulating secretion of active renin suggests they should offer a useful tool for a detailed examination of its secretory pathways. In addition, they may present an important source of trans-acting factors that interact with renin regulatory sequences, and as such could prove useful for a detailed analysis of the sequences which regulate murine renin gene expression. Furthermore, these mice provide a reproducible source of renin-expressing kidney tumors as well as adrenal, testicular, subcutaneous, and submandibular gland neoplasias. The isolation of renin-expressing cell lines from renal and extrarenal tissues and with different renin gene specificities are feasible by this approach and studies toward this end are in progress. Isolation of Renin-expressing and Renin-secreting Cell Lines-We have thus far established and characterized 5 nonclonal cell lines and 1 clonal cell line originating from a kidney tumor and ascites. We are currently establishing other cell lines derived from independent kidney and subcutaneous tumors. We have estimated the renin mRNA content of the 4.1 cell line to be on the order of IOOO-2000 copies per cell as estimated by comparing the relative intensity of the Northern Blot hybridization signal to tissues with known renin mRNA content (1). This value is approximately 5-fold lower than the estimated renin mRNA content of JG cells previously reported (1). However, since the level of JG renin expression in a normal kidney can vary substantially depending on the animal's physiological status, it is difficult to accurately estimate the level of renin expression in these cells prior to the onset of tumorigenesis or during overt tumor formation.
Intracellularly, the nonclonal cell lines and primary tumors contained exclusively active renin. The clonal line 4.1, on the other hand, contained an equal mixture of active and inactive renin. It has previously been reported that cultured human JG cells and chorionic cells generally store less than 3% of the total renin secreted over a 24-h period (19, 35). Although our assay procedures make a direct comparison difficult, our results clearly indicate that there was significant intracellular active renin and stimulation of active renin secretion in response to secretagogue. Moreover, the presence of dense core secretory granules in these cells strongly suggests they are capable of intracellular storage and secretion, although the identification of renin inside these granules will require immunoelectron microscopy.