Transforming growth factor-beta 1 as a signal for induction of cell death by apoptosis.

Cell death by apoptosis is a major determinant of growth of normal tissues and tumours. The present study aimed to elucidate signal factors involved in its regulation. Epithelial cells in control liver, during regression of cyproterone acetate induced liver hyperplasia, in liver (pre)neoplasia and in uterus undergoing apoptosis in vivo show immunostaining for transforming growth factor beta 1 (TGF-beta 1) as detected by anti-pre(266-278) TGF-beta 1 antibodies. Positive immunostaining is also seen in a few intact cells of hyperplastic, regressing liver apparently preparing for apoptosis, but is virtually not found in hepatocytes of normal or growing liver nor in cells undergoing death by necrosis. Recombinant latency associated protein (rLAP, dimer of the pro-region non-covalently associated with the mature region) complex and mature TGF-beta 1 induce apoptosis in isolated hepatocytes cultured in vitro. These findings suggest an involvement of TGF-beta 1 in the induction of apoptosis in certain epithelia in vivo.

The pathogenesis of tumours in the liver and other organs has been found to include disturbance of mechanisms controlling cell death by apoptosis (Bursch et al., 1984;Columbano et al., 1984;Garcea et al., 1984;Wyllie, 1985;Schulte-Hermann et al., 1990;Henderson et al., 1991). Our previous studies on the regulation of liver growth revealed that apoptosis serves to eliminate hepatocytes during involution of hormonally induced liver hyperplasia and during carcinogenesis in preneoplastic tissues (Bursch et al., 1984;Schulte-Hermann et al., 1990). Tumour promoters inhibit apoptosis, thereby accelerating growth of preneoplastic lesions and occurrence of frank neoplasia in the liver (Bursch et al., 1984;Schulte-Hermann et al., 1990). Furthermore, in hormonedependent tumours massive apoptosis can be induced by hormone withdrawal or by hormone antagonists, resulting in rapid tumour regression (Kyprianou et al., 1990;Szende et al., 1990;Bursch et al., 1991). Therefore elucidation of signal factors that can initiate apoptosis in hyperplastic and neoplastic tissues would be of great importance. Up to now progess in understanding the regulation of apoptosis was mainly restricted to hematological cells (Wyllie et al., 1980;Duke & Cohen, 1986;Trauth et al., 1989;Savill et al., 1990;Williams et al., 1990;Koury & Bondurant, 1990;Nunez et al., 1990).
In epithelial tissues TGF-,11 was found to be a negative regulator of growth. It inhibits DN-A synthesis in liver (Carr et al., 1986;Russell et al., 1988), mammary gland (Coletta et al., 1991), uterine endometrium (Rotello et al., 1991) etc. In whole organ homogenates from prostate regressing after castration and from regressing tumours enhanced expression of TGF-,1I was found suggesting its involvement in apoptosis (Kyprianou et al., 1990;Kyprianou & Isaacs, 1989). In primary cultures of uterine endometrial cells and of hepatocytes TGF-,1I induced cell death (Rotello et al., 1991;Oberhammer et al., 1991). In the present study we asked whether TGF-P1 can be detected in individual dying cells of involuting tissues in vivo using immunohistochemical techniques with antibodies raised against two synthetic peptides of the molecule. The first corresponds to the amino terminals 30 amino acids of mature TGF-P1 (LC(1-30)), the second to amino acids 266-278 of the TGF-P1 precursor Flanders et al., 1989). Anti     stains TGF-P1 in its latent or newly synthesised form, but also cleaved parts of the precursor molecule containing amino acids 266-278.
The results obtained indicate that (1) individual cells undergoing apoptosis in vivo specifically respond with antibodies against (pre-)TGF-PI and, albeit less pronounced, against mature TGF-p1. (2) Necrotic liver cells did not show any response to these antibodies. This finding may provide a histological marker to discriminate between apoptotic and necrotic cell death. (3) With anti-pre(266-278) apparently intact hepatocytes preparing for apoptosis, but not yet in its histologically detectable stages could be visualised. So far, no other histological marker of apoptosis has been proven to detect this early stage.
Furthermore, in cultured primary hepatocytes, TGF-P1 is shown to be an inducer of apoptosis. In conclusion, the present study strongly suggests the involvement of TGF-P1 in the control of apoptosis.

Materials and methods
Animals and treatment (in vivo studies) Animals were treated according to published protocols (Bursch et al., 1984(Bursch et al., , 1989. Briefly, cyproterone acetate (CPA, Schering AG, Berlin, FRG) was dissolved in corn oil (Mazola) and administered by gavage once per day to female Wistar rats. The CPA-doses were 100 mg kg-' day-' for 3 days, followed by 130 mg kg-' day-' for 4 days (day 1-7).
Carbon tetrachloride was dissolved in corn oil (1:4, v/v) and administered as a single dose (2 ml kg-') by gavage to male Wistar rats.
Histological procedures Liver and uterus specimens were fixed in Carnoy's fixative and embedded in paraplast. Serial sections 5 gcm thick were exposed to anti-LC(1-30) and to anti-pre(266-278), kindly provided by Dr M. Sporn, NCI, Bethesda, MD, USA. The reaction was accomplished by the unlabelled antibody peroxidase-antiperoxidase technique (Sternberger et al., 1970). The sections were counterstained with Meyer's hemalum.
The specificity of the TGF-,B1 antibodies used in our studies was shown previously in control experiments per-formed by K. Flanders et al. (1989). Briefly, in various cell types a positive immunostaining with anti pre(266-278) was found to be colocalised with antibodies raised against the amino acids 46-56 of precursor TGF-P1 as well as two antibodies directed against the mature TGF-11 (antiLCI-30 and anti(50-75)). Further controls included preincubation of antibody solution with TGF-1I sepharose which was reported to result in reduced staining .
For quantitative histological counts of apoptoses, 4000-6000 hepatocytes were scored per liver; the number of apoptotic bodies was expressed as a percentage of normal hepatocytes.
Cell culture studies Hepatocytes were treated with 10 ng ml-' mature TGF-P1I and 40 ng ml-' recombinant latency associated protein (rLAP) complex (the dimer of the pro-region non-covalently associated with the mature region; provided by Bristol-Meyers-Squibb, Seattle, Wash., USA) for 48 h, other details as previously published (Oberhammer et al., 1991). Hepatocytes were fixed in 4% para-formaldehyde. Chromatin was stained with Hoechst fluorochrom H33258, cytoplasmic condensation and occurrence of apoptotic bodies was demonstrated by hematoxylin-eosin.

Results
First we studied apoptosis in rat liver. Hyperplasia was induced by repeated administration of the hepatomitogen a cyproterone acetate (CPA) as described (Bursch et al., 1984); upon cessation of CPA treatment extensive apoptoses led to regression of hyperplasia within a few days (Bursch et al., 1984).
Apoptosis is known to occur in a morphologically defined sequence of events (Kerr et al., 1972, Wyllie et al., 1980. The first of these is characterised by condensation of chromatin at the nuclear membrane ( Figure  With anti LC(1-30) faint staining for mature TGF-P11 was found in the liver, being more intense in the pericentral region as has been shown previously . Some apoptotic bodies showed a clear cut positive response with anti-LC(1 -30). Quantitative counts revealed that approx. 10% of the apoptotic bodies were positive; the majority showed at best a faint staining.
Anti-pre(266-278) detected TGF-P1 in endothelial cells, resulting in pronounced staining of the lining of the sinusoids and correlating with the site of TGF-P1 gene expression (Nakatsukasa et al., 1990, Jirtle & Meyer, 1991 (Figure lb,c). In addition, most apoptotic cells and residues stood out clearly positive outside and inside the negative parenchymal cells (Figure lb,c; cf. with d, non-immune IgG control). Due to its short duration (1-2 min; Bursch et al., 1990) relatively few apoptotic cells were in the stage of chromatin condensation. Of 60 cells found among 142,000 hepatocytes scored 85% were positive for pre (266-278)  (t), not positive endothelial cells (tt); d, liver exposed to non-immune rabbit serum, note negative endothelial cells and apoptotic body (4); e, intact hepatocytes, positive for anti-pre(266-278) (4); f, balloonised (4) hepatocyte negative for anti-pre(266-278) at 12 h after a single oral dose of 2 ml kg-' carbon tetrachloride; positive hepatocyte (t). Remarkably, a number of apparently intact, vital hepatocytes were found which unequivocally stained positive for pre(266-278) TGF-P1 (Figure le). Their incidence significantly correlated with that of apoptotic bodies in different growth states, being virtually absent in growing liver of 3 week old rats, low in older control animals, and highest in the regression phase following CPA withdrawal (Figure 2). These hepatocytes were 15% smaller than the negative ones and frequently showed concave walls (Figure le), indicating cell shrinkage which is known to begin during early apoptosis (Kerr et al., 1972, Wyllie et al., 1980. We therefore postulate that these hepatocytes are preparing for apoptosis, but have not yet entered its histologically visible stages. Preneoplastic foci in the liver are known to exhibit high apoptotic activity (Bursch et al., 1984;. Many of the apoptotic bodies as well as intact hepatocytes in these foci stained positive for pre (266)(267)(268)(269)(270)(271)(272)(273)(274)(275)(276)(277)(278) TGF-P1, similar to phenotypically normal liver. Likewise, apoptoses in liver tumours were also found to contain pre)266-278) TGF-P1 (not shown). Apparently, the involvement of TGF-P1 in the control of apoptosis is preserved during hepatocarcinogenesis.
The positive immunostaining for (pre-)TGF-P1 in apoptotic liver cells does not prove its role for induction of cell death, and if so, which part of the molecule is the active factor. To address these questions we treated isolated primary hepatocytes with recombinant mature TGF-,B or latency associated protein (rLAP)-complex; the latter consists of the pro region of the precursor (30 to 278) and mature TGF-P1 (279-390) (Miyazano et al., 1988;Gentry et al., 1988). We have already shown that mature TGF-P1 can induce cell death in cultured hepatocytes (Oberhammer et al., 1991). We now found that both factors are active and produce the characteristic signs of apoptosis, namely condensation of chromatin with aggregation at the nuclear membrane ( Figure  3a), and fragmentation of nucleus and cell, giving rise to AB (Figure 3b,c). Electron microscopy revealed intact hepatocellular organelles within these cell fragments (Oberhammer et al., 1992), a major characteristic of apoptosis (Kerr et al., 1972;Wyllie et al., 1980). In control cultures apoptotic cells and AB were found very rarely. Mature TGF-P1 at the concentration investigated increased the incidence of nuclei with condensed chromatin to 2,2% (Figure 3d). The equimolar dose of LAP-complex also induced an increase, yet about five times lower (0,37%; Figure 3d).
Physiological organ involution through apoptosis occurs in the uterus post partum (Afting & Elce, 1978). In a single experiment, in rat uteri 2 days post partum about 70 apoptotic bodies among 14,000 cells scored were found, 67% of them stained positive for pre(266-278) TGF-pl; no or only a weakly positive reaction was observed with anti LC(I-30). Thus, positive immunostaining or apoptotic bodies with antibodies directed against TGF-P1 appears not to be restricted to apoptoses during regression of chemically induced liver hyperplasia.

Discussion
The present study aimed to test whether TGF-P1 is involved in the control of apoptosis. In a first series of experiments, antibodies raised against the mature and pre-form of TGF-Pl were used to ask whether TGF-P1 can be detected in individ- ual dying cells in vivo. After exposing liver sections to the antibody against the mature form of TGF-PI in 10% of apoptotic bodies a positive response could be unequivocally detected. With antipre(266-278) a pronounced staining of apoptotic cells was found. Quantitative histological analysis revealed that almost all apoptotic hepatocytes exhibiting chromatin condensation (early histological stage of apoptosis) as well as 70% of apoptotic bodies (ABs) contain the precursor form of TGF-p1. The negative response in 30% of ABs may be due to advanced intracellular degradation in later stages of the apoptotic process (Wyllie et al., 1980;Bursch et al., 1990). Thus most if not all apoptotic hepatocytes seem to contain the epitopes of anti-pre(266-278).
The failure to detect mature TGF-P1 in the majority of apoptotic cells can probably be explained by its rapid degradation. The biological half-life of the pre-form is about 2 h, that of mature TGF-P1 is only 2 min. (Coffey et al., 1987;Wakefield et al., 1990). Thus, once mature TGF-P1I is formed, its rapid degradation should result in low levels in apoptotic cells that may be insufficient for its unequivocal immunocytochemical detection with the antibodies used. Furthermore, we cannot rule out that anti-pre(266-278) cross-reacts with other molecules involved in apoptosis. However, we observed a colocalisation, albeit in only some apoptotic bodies, of responses with anti-pre(266-278) and anti-LC(1-30). A further line of evidence for the involvement of TGF-P1 in apoptosis is provided by the cell culture studies which show that both forms of TGF-P1 can induce apoptosis of hepatocytes. These observations support our hypothesis that our immunocytochemical findings indicate the presence of forms of TGF-P1 in apoptotic hepatocytes.
In the present study also histologically intact hepatocytes a b c 0 n m m m apparently preparing for apoptosis could be visualised with the anti-pre(266-278) antibody. The duration of this preparative, pre(266-278) TGF-,B1 positive stage can be estimated by assuming that all positive hepatocytes are on the pathway to apoptosis and hence their incidence is proportional to the duration of this stage. Since this incidence was similar to that of AB which persist in a visible form for about 3 h ) the presence of TGF-11 in hepatocytes prone to apoptosis likewise may last approximately 3 h before chromatin condensation starts. Up to now this preparative stage has not been detectable. The use of (pre-)TGF-P1 antibodies as potential marker for preapoptotic cells may render studies on early events in apoptosis possible. Necrotic cells occurring in the liver after carbon tetrachloride or after N-nitrosomorpholine were found to be negative for anti-pre(266-278) TGF-,13. However, in some apoptotic bodies and apparently intact hepatocytes pre-TGF-P1 was detected. Thus the corresponding epitopes may specifically occur in apoptotic cells. This observation may be of considerable interest in view of existing difficulties to discriminate between apoptosis and necrosis. Furthermore, our observations support previous reports that both necrosis and apoptosis may take place after toxic liver injury (Wyllie, 1987).
In an experiment with regressing rat uterus post partum we also observed a positive immunostaining of apoptotic bodies with anti-pre(266-278). These findings in vivo are consistent with recent observations in vitro on induction of apoptosis by TGF-,13 in primary endometrial cells (Rotello et al., 1991).
They show that the appearance of pre(266-278) TGF-P1 in apoptotic cells is not specific for hepatocytes, but may occur in other epithelia as well.
In addition to the immunocytochemical analysis, studies with primary hepatocyte cultures provided support for an involvement of TGF-P1 in apoptosis. It was found that (1) TGF-P1 can actively induce apoptosis; (2) TGF-B1 can induce apoptosis as an extrinsic factor; (3) the greater potency of mature TGF-P1 which was confirmed in four independent experiments suggests that this may be the active form. At present, we do not know the site of TGF-P1 synthesis and whether the increase of TGF-PI protein is regulated at the transcriptional or post-transcriptional level. In previous studies non-parenchymal cells of the liver but not hepatocytes were found to express TGF-P1 (Carr et al., 1989;Nagy et al., 1989). The apparent occurrence of (pre-)TGF-PI selectively in hepatocytes involved in apoptosis could be explained by uptake of TGF-PI through the mannose 6phosphate receptor  or by specific synthesis. Studies on this question are in progress and will be published elsewhere. In any event, the present in vivo and in vitro data strongly suggest an involvement of TGF-,1I in the initiation of apoptosis in hyperplastic and (pre)neoplastic liver and possibly other epithelial tissues.