The Prostate 21:223-237 (1992)
Isolation and Characterization of Transforming Growth Factor Beta Response Variants From Human Prostatic Tumor Cell Lines Rebecca G. Watts and Joy L. Ware Department of Pathology, Medical College of Virginia, Richmond In this study we examined the relation between the response to transforming growth factor beta (TGFP,) in vitro and the growth in vivo of I-LN-PC3-1A (1-LN) human prostatic carcinoma cells. I-LN cells resistant to the growth-inhibitory effects of TGFP, were isolated after exposure to 2 ng/ml TGFP, in an anchorage-independent growth assay. Cloning of TGFP,-resistant and -sensitive populations produced 2 clones (R2-6 and 1-LN clone 4), which maintained relatively stable resistance or sensitivity, respectively, in the absence of TGFP, for up to 12 passages. Colony formation by the R2-6 cells in the presence of TGFP, was 2-10 times greater than that of 1-LN clone 4,depending upon the TGFP, concentration. Injection of 1 X lo5 R2-6 cells into athymic nude mice produced tumors with a significantly shorter latency interval as compared with I-LN clone 4 tumors ( P < 0.0001). Western immunoblotting showed that higher levels of latent TGFP, protein were secreted into the culture medium by 1-LN clone 4 cells. Acidified conditioned media from both clones inhibited mink lung epithelial cell DNA synthesis. Neutralizing monoclonal antibody to TGFP, but not TGFP, abrogated this inhibitory effect. Comparison of the different sensitive and resistant clones showed that in vitro sensitivity to TGFP, and in vivo tumor latency interval were not invariably correlated. Thus, the TGFP, response phenotype in vitro was not always predictive of growth delay in vivo. D 1992 Wiley-Liss, Inc.
Key words: nude mice, prostate cancer, TGFP,
INTRODUCTION Although most human prostatic carcinomas are initially responsive to androgen ablation therapy, they frequently recur as androgen-independent tumors [ 11. Consequently, it is imperative to identify and understand mechanisms by which nonandrogens can affect the proliferation of normal and malignant prostatic epithelial cells. Transforming growth factor beta (TGFP,) is a multifunctional growth regulatory factor which may either inhibit or stimulate cell proliferation, dependent upon the cell type [2]. Analysis of both human prostate carcinoma cell lines and limited data
Received for publication January 24, 1992; accepted June 16, 1992. Portions of this study were presented at the annual meeting of the American Association of Cancer Research, held in Washington, D.C., May 21-26, 1990. Address reprint requests to Joy L. Ware, Ph.D., Department of Pathology, Box 662, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA 23298.
0 1992 Wiley-Liss, Inc.
224
Watts and Ware
from specimens of benign prostatic hyperplasia implicates TGFP, in the regulation of prostatic epithelial cell growth and behavior. Two androgen-insensitive human prostate carcinoma cell lines, PC-3 [3] and DU145 [4], possess receptors for TGFP,, express TGFP, mRNA [ 5 ] , and secrete TGFP, into the culture media [5,6]. In contrast, the androgen-sensitive LnCAP does not contain detectable TGFP mRNA nor express high-affinity TGFP receptors [ 5 ] . Neither anchorage-independent nor anchorage-dependent growth of LnCAP is affected by TGFP, [5,7]. Thus, all prostatic tumor cell lines tested to date are either inhibited by TGFP, or unaffected. Northern analysis of human benign hyperplastic prostatic tissue reveals mRNA for both TGFP, and TGFP, [8]. If TGFP ,-mediated inhibition of prostatic tumor cell proliferation is relevant to the evolution of the prostatic tumor phenotype, cells which are resistant to this inhibition would be predicted to possess a selective growth advantage in vivo. To test this hypothesis, we have isolated and characterized TGFP ,-resistant clones of 1LN-PC-3-1A cells, a highly metastatic subline [9-111 of PC-3 [3]. TGFP, inhibited both anchorage-independent and anchorage-dependent proliferation by 1 -LN-PC-31A (designated 1-LN) cells in a dose-dependent manner. A relatively stable TGFP,resistant subpopulation, 1-LN-R2, was obtained by expansion of a colony of 1-LN cells growing in the presence of 2 ng/ml TGFP . Comparison of the TGFP I sensitivity of clones of 1-LN and subclones of 1-LN-R2 revealed that 1-LN-R2 was enriched with clones resistant to TGFP, . Both the sensitive and resistant clones were tumorigenic in athymic nude mice, yet some TGFP,-sensitive clones grew with a significantly prolonged latent period.
,
MATERIALS AND METHODS The PC-3 cell line was originally obtained from a bone marrow metastasis of human prostate cancer [3]. The subline of PC-3, designated 1-LN, was isolated from a lymph node metastasis in a PC-3 tumor-bearing athymic nude mouse and is highly metastatic in adult athymic nude mice [9-111. This human prostatic carcinoma cell line was routinely cultured in RPMI 1640 supplemented with 10% defined heatinactivated calf serum (HICS, Hyclone Laboratories, Logan, UT), glutamine, and gentamicin. Generation of TGFp,-Sensitive and -Resistant Variants
TGFf3, purified from porcine platelets was reconstituted in 4 mM HC1 containing 1 mg/ml bovine serum albumin (BSA) (R and D Systems, Minneapolis, MN). Each well of a 24-well tissue culture plate was coated with 0.6% agar-RPMI 1640. 2 X lo3 cells of the metastatic 1-LN subline were plated as an overlay in RPMI 1640 containing 0.3% agar, HICS, and the indicated final concentration of TGFP, (0.1, 0.5, 1 .O, or 2.0 ng/ml). Control wells contained no TGFP, but received an equal volume of acidified BSA. Plates were maintained at 3 7 T , 5% CO, for 12-14 days. Isolated cell colonies which grew in the presence of 2 ng/ml TGFP, were retrieved from the soft agar and 3 colonies-designated 1-LN R1, R2, and R3-were expanded in adherent culture in the absence of additional TGFP, . These cells were exposed to single-step selection to permit observation of the stability of this phenotype. Nine clones obtained by
TGFP, Response Variants in Human Prostate
225
limiting dilution cloning of colony R2 were subsequently retested for resistance to varying concentrations of TGFP in the anchorage-independent growth assay, and produced clones designated R2- 1 to R2-9. Nonselected clones of 1-LN were obtained randomly by limiting dilution cloning. These clones, designated 1-LN clone 1 to 1-LN clone 9, were expanded in culture in the absence of added TGFP, and were examined for their sensitivity to different concentrations of TGFP Effect of TGF& on Anchorage-Independent Growth
Each well of a 24-well plate was initially layered with 0.6% agar-RPMI 1640. A top layer containing 2.5 X lo3 1-LN cells or cell clones suspended in RPMI supplemented with 0.3% agar, 5% HICS, and 0-10 ng/ml of TGFP, was added to triplicate wells for each TGFP, concentration tested. Plates were cultured as previously described for 12-14 days and colonies 50 p m in diameter or greater were counted. The anchorage-independent growth assays were periodically performed on the cell clones to document continued sensitivity or resistance in the presence of TGFP
,.
Effect of TGFPl on Growth of Adherent Cells
We examined the effect of TGFP, on the growth of 1-LN clones in monolayer culture using tritiated (3H)-thymidine incorporation measured by trichloroacetic acid (TCA) precipitable counts and autoradiography . Autoradiography permits the determination of 3H-thymidine incorporation on a per cell basis; TCA precipitable 3Hthymidine labelling reflects DNA synthesis by the whole cell population. TGFP,resistant or -sensitive clones (3.5 X lo3) were plated in triplicate wells of a 96-well plate in RPMI media supplemented with 5% HICS and the indicated concentrations of TGFP,. After 48 hours, cells received 'H-thymidine (2 pCiim1; specific activity 2 CiimM). Ascorbic acid was added at 1, 6, 18, 24, 36, or 48 hours to stop DNA synthesis, and the relative radioactivity incorporated into TCA-precipitable material was determined by liquid scintillation counting. For autoradiographical analysis, clones were plated in duplicate in sterile 4-well chamber slides (4 X lo4 celldwell) in HAMSF12 culture medium supplemented with 5% calf serum. Clones were cultured in the indicated concentration of TGFP, ( 0 , 0. I , 0.5, and 2.0 ngiml). After 72 hours in culture, 'H-thymidine was added to each well (5 mCi/ml, specific activity 45-60 Ci/mmole). Forty-eight hours later, each slide was washed, fixed in methanol, and air dried. Each slide was then dipped in autoradiographic emulsion (NTB-3, Kodak, Rochester, NY) and stored at 4°C. After 7 days of exposure, each tissue culture slide was developed and counterstained with hematoxylin. A minimum of 500 cell nuclei were evaluated per well to permit the calculation of the labeling index (percent 3H-thymidine incorporation). In Vivo Behavior TGFP,-sensitive 1-LN clones 1 , 4 , 5 , and 7, and resistant clones R2-6 and R2-4 were compared for tumorigenicity and tumor growth rate in vivo. Specific pathogenfree male athymic nude mice, 6-8 weeks old, were obtained from the Massey Cancer
226
Watts and Ware
Center, Richmond, VA, Athymic Nude Mouse Colony. Tumor cells (1 X lo5, 5 X lo5, or 1 x lo6) were injected subcutaneously into the dorsal side of each mouse. Tumor growth was monitored by measurement with calipers, and tumor volume was estimated by the following formula: L x w2 2 ’ where L = length and W = width. Mice were sacrificed approximately 35 days after injection, or when the tumor burden reached 300-400 mm3. Mink Lung Assay for TGFP, Growth Inhibition
CCL-64 mink lung epithelial cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% Nu-Serum in a humidified incubator at 37°C and 5% CO, [ 121. Cells from subconfluent cultures were removed with trypsin-EDTA, washed, and adjusted to 5 X lo5 in an assay buffer consisting of DMEM with 0.2% Nu-Serum. Then, 24-well plates were seeded in duplicate with 2 x lo5 cells/well and were incubated for 1-6 hours. To prepare conditioned media, serum was eliminated from 1-LN clone 4 or R2-6 cell cultures with 4 changes of serum-free media over a period of 6-24 hours. Cells were incubated with fresh serum-free MEM containing BSA (0.5 mg/ml) and 1 pg/ml each of aprotinin, leupeptin, and pepstatin A. The media was conditioned for an additional 24 hours, after which it was removed and centrifuged at 15,000 g for 5 minutes. Unconcentrated conditioned media collected from either the TGFP -resistant or -sensitive 1-LN cell clones was added to the CCL-64 cells in each well of the 24-well plate. For comparison, each well of the conditioned media was also acidified and/or treated with 10 pg (1,000-fold excess) neutralizing antibody to TGFP, or TGFP2 (R and D Systems). The CCL-64 cells and various conditioned media were incubated for 20 hours at 37°C in a humidified incubator. 3H-thymidine (0.5 pCi/well, specific activity 40-60 Ci/mmole) was added to each well and incubated an additional 3 hours. The cells were then washed and fixed in methanol, air dried, and trypsinized. The contents of each well were transferred to scintillation vials and analyzed in a liquid scintillation counter.
,
TGFP, Protein Preparation and Analysis
R2-6 and 1-LN clone 4 tumor clones were assessed for secretion of TGFP, protein using a modification of the procedure for Western immunoblotting [ 131. Serum-free conditioned media was collected, as described previously, from the R2-6 and 1-LN clone 4 cells and was concentrated with ultrafiltration with a YMlO membrane (Amicon Corp., Danvers, MA). Protein was electrophoresed through a 15% SDS polyacrylamide gel [ 141 and transferred to an Immobilon filter. The presence of TGFP, was detected immunologically with a polyclonal chicken anti-TGFP antibody using a modification of the procedure distributed by R & D Systems. The filter was blocked 2-1 2 hours with borate-buffered saline containing 1% bovine serum albumin (BBS/BSA). Chicken anti-TGFP, (R & D Systems, Minneapolis, MN) was diluted 1:10,000 in BBS/BSA and was incubated overnight at room temperature. Following 5 washes for 5 minutes in BBS, biotinylated rabbit anti-chicken antibody
,
TGFP, Response Variants in Human Prostate
227
TABLE I. Comparison of Anchorage-Independent Growth of PC-3, 1-LN, and 1-LN-R2 Cells in the Presence of Increasing Concentrations of TGFP, 0 PC-3 1-LN 1-LN-R2
103 C 12 121 f 22.7 87 2 13.8
0.1 101 i 4.4 176 i 22.6 158 5 25
TGFP, (ng/ml) 0.5 41.5 k 5 47 t 5.3 145 C 17.5
1
2
20 2 2.9
14.5 t 1.4 10.3 2.4 76 f 12.7
-
108 t_ 19.1
*
Results expressed as mean number of colonies/well and are the average of 2-3 experiments performed in triplicate 2 standard error of the mean.
(1:2,500, Zymed Laboratories, South San Francisco, CA) was added for 2 hours. After 5 washes in BBS, the filter was incubated with alkaline phosphatase labelled streptavidin (1:2,500, Southern Biotech, Birmingham, AL). Following 5 final washes in BBS, the filter was equilibrated in carbonate buffer and NBT/BCIP substrate was added. The reaction was stopped after 15-30 minutes by replacing the NBT/BCIP solution with BBS containing 5 mM EDTA. RESULTS Effect of TGFP, on Anchorage-Independent Growth Exposure of the parental PC-3 line and the metastatic 1-LN subline to TGFP, produced a dose-dependent inhibition of colony formation in soft agar. Low concentrations of TGFP, (0.1 ng/ml) consistently stimulated colony formation among 1-LN cells to levels 20% above control values, while a more modest stimulatory effect on parental PC-3 cells was observed (Table I). At 2 ng/ml TGFP,, colony formation by both PC-3 and 1-LN was greatly inhibited. Single colonies of 1-LN cells, which grew in the presence of 2 ng/ml TGFP, were recovered from 3 different wells and expanded in culture. When subsequently retested for sensitivity to TGFP, , only one of these, designated 1-LN-R2, produced significant numbers of colonies in the presence of 2 ng/ml TGFP1. Furthermore, 1-LN-R2 cells were not inhibited by 5 or 10 ng/ml TGFP,. As illustrated in Table I, pronounced stimulation of colony formation with 0.1-0.5 ng/ml often occurred, and colony formation only decreased slightly at 2 nglml. Comparison of the dose-response behavior of 1-LN and 1-LN-R2 exposed to varying concentrations of TGFP, is facilitated in Fig. 1, where the data from Table I are plotted. The anchorageindependent growth assays were performed periodically on passages 5 -22 to reaffirm and document the response of 1-LN and 1-LN-R2 cells to TGFP,. 1-LN cells and 1-LN-R2 cells were subcloned and 9 clones of each were isolated and retested in an anchorage-independent growth assay along with the varying concentrations of TGFP, . Colony formation of all 9 randomly chosen clones from the 1-LN line was inhibited by TGFP, at doses ranging from 0.5 to 2.0 ngiml (Fig. 2a). TGFP, did not inhibit the growth of 6 of 9 clones obtained from the TGFP,resistant R2 colony as effectively as it inhibited 1-LN clone 4 cells at concentrations of 0.5-2 ng/ml (Fig. 2b). Although the original R2 subline remained resistant to TGFP, for up to 3 months in vitro, cells of clone R2-6 spontaneously exhibited sensitivity after 12 passages. Thus, TGFP, resistance was not stable during prolonged culture. In con-
228
Watts and Ware 180;
t 1-LN-R2
I160: 160 8Oj l
J
t 1-LN
-
1404 0
-
5 1201 0
5
100;
-
t 80:n
f
2
SO: -
40 1
20 1
0'
I
I
0.1
0.5
I
1
2
TGF-61 (nglrnl)
Fig. 1. Comparison of anchorage-independent growth of 1-LN and I-LN-R2 in the presence of varying concentrations of TGFP,. Results are expressed as mean colonies 2 standard error of the mean. (Data from experiments described in Table I.)
trast, 1-LN clone 4 cells from long-term cultures remained equally sensitive to TGFP, in the anchorage-independent growth assays. Effect of TGFP, on the Growth of Adherent Cells
The monolayer growth of both R2-6 and 1-LN clone 4 in the presence of TGFP, was evaluated using 3H-thymidine incorporation as determined by TCA precipitable radioactivity and/or autoradiography . 3H-thymidine incorporation by parental 1-LN cells was inhibited by TGFP, 45-60% at 0.5-2.0 ng/ml. However, cells from the TGFP ,-resistant colony 1-LN-R2 demonstrated no decrease in 3H-thymidine uptake when exposed to any concentration of TGFP, up to 2.0 ng/ml. Two clones of 1-LN-R2 which were refractory to the inhibitory effects of TGFP, (clones 3 and 6) and 2 clones of 1-LN which were sensitive to TGFP, (clones 1 and 4) in the anchorage-independent growth assay were tested using 3H-TdR uptake to determine sensitivity to TGFP, in monolayer culture. As Fig. 3 indicates, 3H-thymidine incorporation was inhibited similarly in 3 of 4 clones. Only 1-LN-R2 clone 6 (R2-6) remained unaffected by TGFP, in adherent cell culture. In fact, at 2.0 ng/ml, 3H-thymidine incorporation by this clone was stimulated nearly 2-fold. Autoradiography of adherent cell cultures demonstrated that the number of radiolabelled tumor cell nuclei observed in the cells of TGFP ,-resistant clone R2-6 was only modestly decreased by the addition of up to 2.0 ngiml TGFP, (84% of control values). These results support the data obtained from 3H-thymidine incorporation into TCA-precipitable material of adherent R2-6 cells. Each assay determined that DNA incorporation of the radiolabel was not significantly inhibited by 2 ng/ml TGFP,. In contrast, the labeling index of the I-LN clone 4 cells was reduced to 40% and 45% in the presence of 0.5 and 2.0 ng/ml TGFP,, respectively (data not shown). Tumorigenicity
Comparisons of the tumorigenicity and growth rate in vivo of the TGFP,sensitive 1-LN clones 1, 4, 5, and 7, and resistant clones R2-4 and R2-6, were conducted. 1-LN clone 1 and clone 4 exhibited similar sensitivity to TGFP, in the
TGFP, Response Variants in Human Prostate
229
1 -LN (T9) Clones
h
400
300
I
1
TGF-8 conc. -0.1 ng/ml Lz4 0.5 ng/ml e0 1.O ng/ml 0 2 . 0 "g/ml
1
3
4
5
6
7
8
9
10
clone number B
1-LN (R2) Clones
'I
TGF-p conc. 1 0 . 1 ng/ml 0.5 ng/ml
rza 1.0 ng/ml
300
-
e
ri3 2.0 "g/rnl
c
c
8
200
%
0
x 100
-0 n
1
2
3
4 5 6 clone number
7
8
9
10
Fig. 2. A: Comparison of anchorage-independent growth of 9 different clones of 1-LN in the presence of varying concentrations of TGFP,. B: Comparison of anchorage-independent growth of 9 different clones of 1-LN-R2 in the presence of varying concentrations of TGFP,. Concentrations of TGFP, are indicated by the legend in the figure inset.
anchorage-independent growth assays and a similar delayed latent interval after subcutaneous (sc) injection into nude mice (Fig. 4). 1-LN clones 5 and 7, although exhibiting reduced colony formation in the presence of TGFP, , demonstrated rapid growth in nude mice. 1-LN clone 4 was selected for intensive analysis, because TGFP consistently inhibited both anchorage-independent and anchorage-dependent growth. Clone R2-4 displayed resistance to TGFP, in the anchorage-independent growth assay, yet exhibited a variable rate of growth in vivo. R2-6 was chosen for additional study, because TGFP, was unable to inhibit both monolayer and colony formation. SC injection of 5 X lo5 cells of each clone produced tumors in all mice, but there was a slight increase in the time to tumor formation in mice injected with 1-LN clone 4 cells. Comparison of athymic nude mice injected sc with 1 X lo5 1-LN clone 4 cells at 30 days showed a significantly longer latency period compared to mice injected with 1 X lo5 clone R2-6 cells (Figs. 5 and 6). There was a significant tumor size difference between the 2 groups of mice at day 30 ( P < 0.0001, t-test). By day 25 post-injection, only one mouse injected with 1-LN clone 4 cells had reached a
,
230
Watts and Ware
I P)
5
Fig. 3. Comparison of 3H-thymidine incorporation by clones in monolayer culture exhibiting resistance or sensitivity to growth inhibition by TGFP, in the anchorage-independent growth assay. The plot is representative of 2 experiments.
tumor volume of 50 mm3, while the remainder of mice had tumors of only 15 mm3 or less (Fig, 6b). Conversely, all mice injected with 1 X lo5 R2-6 cells quickly reached a tumor volume of 50 mm3 by an average of 19 days postinjection (Fig. 6a). At 38 days, the average tumor volume of the R2-6 tumors was 5-8 times greater than tumors resulting from subcutaneous injection of 1-LN clone 4 cells. When tumors from mice injected with 1-LN clone 4 cells were allowed to continue to grow past 38 days, they entered log phase growth and grew to a size comparable to that observed in animals injected with R2-6 cells. Thus, in vivo growth was retarded but not ablated. Gross metastases were not observed in mice injected with any number of tumor cells. Of 12 mice injected with 1 X lo5 1-LN clone 4 cells included in the study, tumors in 2 mice exhibited anomalous proliferation, similar to that seen in animals injected with R2-6. However, even these 2 faster growing tumors had an increased time to tumor formation. TGFP, and Secretion Both 1-LN clone 4 and R2-6 cells secreted TGFP, into the culture as assessed by Western immunoblotting, following SDS-PAGE under reducing conditions (Fig. 7). When an estimated 700, 350, and 100 pg of total proteidlane was loaded from both 1-LN clone 4 and R2-6 supernatants, immunologically reactive bands were detected at approximately 25 kDa, with weaker bands observed at 35-40 kDa and 82-90 kDa. In lanes 1 and 2, containing 700 pg total protein, it is probable that the anomalous migration of TGFP, resulted from protein overloading. The bands at approximately 40 and 90 kDa may represent binding of the polyclonal TGFP, antibody to the pro region of the TGFP ! precursor molecule and the TGFP precursor, respectively [ 151. Loading 350 pg total protein produced a strongly immunoreactive band in lane 3, containing 1-LN clone 4 supernatant, with slight immunoreactivity observed in the comparably loaded lane containing R2-6 supernatant. We conclude that R2-6 conditioned media contained quantitatively less TGFP, protein than 1-LN
,
TGFP, Response Variants in Human Prostate
231
1000
1 I
100 a -
-€E W
I
3
9 U
0
5+
10
1
TIME (days)
Fig. 4. Growth of 3 1-LN clones sensitive to the inhibitory effects of TGFP, in vitro after subcutaneous injection of 1 X lo5 cells in athymic nude mice. Clone 1, A; clone 5 , 0 ;and clone 7 , . Symbols indicate individual mice.
clone 4. Relative to purified TGFP, (R & D Systems), the estimated migration of the dominant 1-LN clone 4 TGFP, immunoreactive band was approximately 25 kDa. Mink Lung Cell Inhibition by TGFP,
We assessed the presence of secreted TGFP, in conditioned media from the cultures of 1-LN clone 4 cells and R2-6 cells using the CCL-64 mink lung epithelial cell inhibition bioassay. Untreated conditioned media from R2-6 or 1-LN clone 4 cells did not produce a significant inhibition compared to control media alone (Fig. 8). In contrast, acidified serum-free conditioned media from R2-6 inhibited CCL-64 growth by 83% and 1-LN clone 4 cells inhibited DNA synthesis by 87% compared with unconditioned control media. Untreated media in the presence of a 1,000-fold excess of neutralizing antibody to TGFP, exhibited no differences from control. In contrast, acidified conditioned media of R2-6 and 1-LN clone 4 in the presence of neutralizing antibody to TGFP, produced only 35% and 25% inhibition, respectively, compared to control media. Pretreatment of R2-6 and 1-LN clone 4 conditioned media with anti-TGFP, antibody had no effect on the inhibitory activity of either conditioned media. Despite inclusion of a 1,000-fold excess of neutralizing antibody to TGFP,, acid-activated conditioned media of R2-6 and 1-LN clone 4 cells inhibited CCL-64 cell proliferation by 89.5% and 94.4%, respectively.
Watts and Ware
232
B
A 1000
1000
900 800
900 800
700
700
600 500
I
600 500
400
400
300
300
200
200
-EE
E E
-
UJ
w
I
3
p
I 100
3
p
100
a:
90 80
5
70
LT
60
f
I-
50
50
40
40
30
30
0
20
0
f
90 80 70 60
20
10
10
- 1
1
10
20
30
TIME (days)
40
50
10
20
30
40
50
TIME (days)
Fig. 5 . Growth of TGFP,-resistant clone R2-6 (A) and TGF@,-sensitive 1-LN clone 4 (B) after subcutaneous injection of 5 X l o 5 cells in athymic nude mice. Each symbol indicates an individual mouse.
DISCUSSION
In this study, we have examined the impact of TGFP, on the proliferative response of a metastatic derivative of the androgen-independent human prostatic carcinoma cell line PC-3. Although both anchorage-independent and adherent monolayer growth of the 1-LN cells was inhibited by TGFP, concentrations greater than 0.1 ngiml (Table I), this subline was composed of cells which exhibited heterogeneous sensitivity to this factor. Both TGFP,-sensitive and -resistant clones (I-LN clone 4 and R2-6, respectively) secreted immunodetectable TGFP, (Fig. 7) in latent form (Fig. 8). Our finding that the growth of 1-LN cells in monolayer cultures, as well as colony formation in soft agar, was inhibited by exogenous TGFP, is consistent with the reports of others who demonstrated that TGFP, inhibited the growth of the parental PC-3 line in monolayer [ 161 and suppressed anchorage-independent proliferation of both PC-3 and DU-145 [ 5 ] . Our report not only extends their observations to a highly metastatic variant of PC-3 [9-111, but also demonstrates the isolation of spontaneously occurring TGFP,-resistant subpopulations of 1-LN cells, such as R2 and clones R2-6. In addition, in vivo analysis demonstrated no consistent
TGFP, Response Variants in Human Prostate A
B 1000
900 800 700
--
1000
-
700 -
600 -
400 -
400 -
300 -
300
-
200
-
100
--
500
500
200
-
-
" 5
E E w
I
3
9
a: 0
f +
-
900 800 -
600 -
--
233
E E
w
2
90 80 70 -
3
100
9 8 fI-
60 50 40
-
30
-
90 80
1
20 10 1
-
1 1
I
I
I
I
10
20
30
40
50
1 10
20
30
40
50
TIME (days)
TIME (days)
Fig. 6. Growth of TGFP,-resistant clone R2-6 (A) and TGFP,-sensitive 1-LN clone 4 (B) after subcutaneous injection of 1 X lo5 cells in athymic nude mice. Statistical significance, P < 0.0001, t-test. Each symbol indicates an individual mouse. Two mice out of 12 injected with 1 X 105 1 -LN clone 4 cells exhibiting anomalous growth were not included in the graphed results but are discussed in the text.
correlation between in vitro sensitivity to TGFP, and initiation of tumor growth in athymic nude mice among the different sensitive and resistant clones tested. Thus, the TGFP, response phenotype in vitro was not always predictive of growth delay in vivo. In vivo behavior of clones with different TGFP, in vitro response phenotypes was variable. Two of 4 randomly isolated clones of I-LN which were inhibited by TGFP, in soft agar (clone 4 and clone 1) exhibited a longer latency period for tumor formation compared to 2 other TGFP -sensitive clones when injected subcutaneously into athymic nude mice. These other 1-LN clones (clones 5 and 7) sensitive to the growth inhibitory effects of TGFP in the anchorage-independent growth assay exhibited the same latency and grew in vivo at the same rate as the TGFP,-resistant R2-6 clones (Fig. 4). Detection of differences between the growth of sensitive and resistant clones in vivo was dependent upon the number of tumor cells initially injected. Subcutaneous injection of 1 X lo5 1-LN clone 4 cells resulted in significantly longer latent periods
,
,
234
Watts and Ware
Fig, 7. Detection of TGFP, by Western immunoblot. Anti-TGFP, antibody at 1:10,000 dilution was used for the detection of secreted TGFP, in conditioned media from 1-LN clone 4,lanes 1, 3, 5 ; and R2-6, lanes 2, 4,6. Lane 7 contains purified TGFP, (6 ng, R & D Systems). This figure represents 1 of 3 similar experiments.
to tumor formation relative to R2-6 (Fig. 6b). However, sc injection of 1 X lo6 or 5 x los resistant or sensitive 1-LN tumor cells generated similar growth curves that were not statistically different. We speculate that regulatory factors present in vivo may be responsible for observed differences in tumor latency. Thus, the microenvironment at the site of injection may have influenced the establishment of the tumor. It is possible that the increased TGFP, levels secreted by both the 1-LN clone 4 tumor cells and tissues in the microenvironment may have produced a combined threshold concentration that initially inhibited the growth of small numbers of TGFP,-sensitive cells. In effect, the transiently high local levels of TGFP, may act in an autocrine fashion to downregulate the growth of TGFP,-sensitive tumor cells. The ability to elude this regulatory mechanism, as observed in most of the TGFP,-resistant R2-6 tumors, may have conferred a growth advantage in vivo. As all tumors enlarged, albeit the 1-LN clone 4 tumors more slowly, tumor cell numbers would finally increase above a critical number and favor exponential tumor growth. Clearly, additional investigations are necessary to quantify the local concentrations of TGFP, at the site of prostatic tumor injection. We propose that augmented secretion of TGFP, by 1-LN clone 4 cells in vivo may have altered the levels of TGFP, in the vicinity of tumor cell injection. This hypothesis is supported by the Western blot data with 3 different concentrations of conditioned media, which revealed that immunodetectable 25 kD TGFP was greater in supernatant obtained from 1-LN clone 4 cells. In addition, conditioned media from TGFP,-sensitive 1-LN clone 4 inhibited CCL-64 mink lung epithelial cell DNA synthesis by 88%. This acid-activated inhibition of 1-LN clone 4 media media was abrogated 60% with the use of a neutralizing antibody to TGFP, but not by a neutralizing antibody to TGFP,. The increased TGFP, protein secretion of 1-LN
,
TGFP, Response Variants in Human Prostate
235
Fig. 8. TGFP, activity in conditioned media from TGFP,-resistant R2-6 and TGFP,-sensitive I-LN clone 4 cells. Data are expressed as a percent of control k SEM. The mean percent of control was determined from duplicate samples.
clone 4 cells suggests that these cells may produce TGFP, that then acts in an autocrine fashion to regulate tumor cell growth in vitro and in vivo. Based upon this autocrine hypothesis of growth regulation by growth factors, we expected that TGFP,-sensitive 1-LN clone 4 tumors would exhibit TGFP, resistance in vitro after successful in vivo passage [ 171. However, the 1-LN clone 4 cells recovered from in vivo tumors remained as sensitive to TGFP, in the anchorage-independent growth assays as they were prior to in vivo growth. Finally, we concluded that resistance to TGFP, was not stable because R2-6 clones cultured 12 or more passages in vitro spontaneously displayed sensitivity to the presence of TGFP, . Our observations are consistent with multiple investigators who have shown that clones exhibit greater phenotypic variability and instability than the parental population from which the clone was derived [18, reviewed in 19-21]. Obviously, the combination of positive and negative growth factors produced in vitro plays a role in the regulation of R2-6 tumor cells. In summary, this paper described the first isolation of TGFP,-resistant subclones from a TGFP,-sensitive parental human prostatic carcinoma cell line. This distinct subpopulation of 1-LN cells refractory to the inhibitory effects of TGFP, may represent an important stage in the progression of neoplastic transformation in prostatic malignancy. These results, combined with the observed variation in phenotypic stability among these cells, suggest that the impact of local TGFP, levels will be
236
Watts and Ware
greatest at low tumor cell population size. Consequently, TGFP, may be more significant in modulating the growth of prostatic tumor micrometastases than of the primary tumor. ACKNOWLEDGMENTS
This work was supported by NCI grant CA.50609. We thank Winston C. Chandler for his technical expertise, and Sharon R. Collins for her assistance with the CCL-64 bioassays. REFERENCES I . Cunha GR, Donjacour AA, Cooke PS, Mee S, Bigsby RM, Higgins SJ, Sugimura Y: The endocrinology and developmental biology of the prostate. Endocrin Rev 8:338-362, 1987. 2. Sporn MB, Roberts AB, Wakefield LM, DeCrombrugghe B: Some recent advances in the chemistry and biology of transforming growth factor-p. J Cell Biol 105:1039-1045, 1987. 3. Kaighn ME, Naragan K S , Ohmuki Y, Lechner JF, Jones LW: Establishment and characterization of a human prostatic carcinoma cell line (PC-3). Invest Urol 17:16-23, 1979. 4. Stone KR, Mickey DD, Wunderli H, Mickey GH, Paulson DF: Isolation of a human prostate carcinoma cell line (DU145). Int J Cancer 21:274-281, 1978. 5. Wilding G, Zugmeier G, Knabbe C, Flanders K, Gelmann E: Differential effects of transforming growth factor p on human prostate cancer cells in vitro. Mol Cell Endocrinol 62:79-87, 1989. 6. Ikeda T, Lioubin MN, Marquardt H: Human transforming growth factor type P2: Production by a prostatic adenocarcinoma cell line, purification, and initial characterization. Biochemistry 26:24062410, 1987. 7. Schuurmans ALG, Bolt J , Mulder E: Androgens and transforming growth factor-p modulate the growth response to epidermal growth factor in human prostatic tumor cells (LNCaP). Mol Cell Endocrinol 60:101-104, 1988. 8. Mori H, Maki M, Jaye M, Igarashi K, Yoshida 0 , Hatanaka M: Increased expression of genes for basic fibroblast growth factor and transforming growth factor type p2 in human benign prostatic hyperplasia. Prostate 16:71-80, 1990. 9. Ware JL, Paulson DF, Mickey GM, Webb KS: Spontaneous metastasis of cells of the human prostate carcinoma cell line PC-3 in athymic nude mice. J Urol 137:1304-1306, 1982. 10. Ware JL, Delong ER: Influence of tumour size on human prostate tumor metastasis in athymic nude mice. Br J Cancer 51:419-423, 1985. 1 1. Ware JL, Lieberman AP, Webb KS, Vollman RT: Factors influencing phenotypic diversity of human prostate carcinoma cells metastasizing in athymic nude mice. Exp Cell Biol 53:163-169, 1985. 12. Danielpour D, Dart LL, Flanders KC, Roberts AB, Sporn MB: Immunodetection and quantitation of the two forms of transforming growth factor-beta (TGF-P1 and TGF-P2) secreted by cells in culture. J Cell Physiol 138:79-86, 1989. 13. Florini JR, Roberts AB, Ewton DZ, Falen SL, Flanders KC, Sporn MB: Transforming growth factor-beta. A very potent inhibitor of myoblast differentiation, identical to the differentiation inhibitor secreted by Buffalo rat liver cells. J Biol Chem 261:16509-16513, 1986. 14. Laemmli UK: Cleavage of structural proteins during the assembly of bacteriophage T4. Nature (Lond.) 227:660-685, 1970. 15. Gentry LE, Lioubin MN, Purchio AF, Marquardt H: Molecular events in the processing of recombinant type I pre-pro-transforming growth factor beta to the mature polypeptide. Mol Cell Biol 8:4162-4168, 1988. 16. Okutani T, Nishi N, Kagawa Y, Takasuga H, Takenaka 1, Usui T, Weda F: Role of cyclic AMP and polypeptide growth regulators in growth inhibition by interferon in PC-3 cells. Prostate 18:73-80, 1991. 17. Sporn MB, Todaro GT: Autocrine secretion and malignant transformation of cells. N Engl J Med 303:878-880, 1980. 18. Poste G , Doll J, Fidler IJ: Interactions among clonal subpopulations affect the stability of the
TGFP, Response Variants in Human Prostate
237
metastatic phenotype in polyclonal populations of B16 melanoma cells. Proc Natl Acad Sci USA 78:6226-6230, 1981. 19. Heppner GH: Tumor heterogeneity. Cancer Res 44:2259-2265, 1984. 20. Miner KM, Kawaguchi T, Uba GW, Nicholson GL: Clonal dnft of cell surface, melanogenic, and experimental metastatic properties of in vivo-selected, brain meninges-colonizing murine B- 16 melanoma. Cancer Res 424631-4638, 1982. 21. Poste G , Tzeng J , Doll J, Grieg R, Rieman D, Zeidman I: Evolution of tumor cell heterogeneity during progressive growth of individual lung metastases. Proc Natl Acad Sci USA 79:6574-6578, 1982.
Isolation and Characterization of Transforming Growth Factor Beta Response Variants From Human Prostatic Tumor Cell Lines Rebecca G. Watts and Joy L. Ware Department of Pathology, Medical College of Virginia, Richmond In this study we examined the relation between the response to transforming growth factor beta (TGFP,) in vitro and the growth in vivo of I-LN-PC3-1A (1-LN) human prostatic carcinoma cells. I-LN cells resistant to the growth-inhibitory effects of TGFP, were isolated after exposure to 2 ng/ml TGFP, in an anchorage-independent growth assay. Cloning of TGFP,-resistant and -sensitive populations produced 2 clones (R2-6 and 1-LN clone 4), which maintained relatively stable resistance or sensitivity, respectively, in the absence of TGFP, for up to 12 passages. Colony formation by the R2-6 cells in the presence of TGFP, was 2-10 times greater than that of 1-LN clone 4,depending upon the TGFP, concentration. Injection of 1 X lo5 R2-6 cells into athymic nude mice produced tumors with a significantly shorter latency interval as compared with I-LN clone 4 tumors ( P < 0.0001). Western immunoblotting showed that higher levels of latent TGFP, protein were secreted into the culture medium by 1-LN clone 4 cells. Acidified conditioned media from both clones inhibited mink lung epithelial cell DNA synthesis. Neutralizing monoclonal antibody to TGFP, but not TGFP, abrogated this inhibitory effect. Comparison of the different sensitive and resistant clones showed that in vitro sensitivity to TGFP, and in vivo tumor latency interval were not invariably correlated. Thus, the TGFP, response phenotype in vitro was not always predictive of growth delay in vivo. D 1992 Wiley-Liss, Inc.
Key words: nude mice, prostate cancer, TGFP,
INTRODUCTION Although most human prostatic carcinomas are initially responsive to androgen ablation therapy, they frequently recur as androgen-independent tumors [ 11. Consequently, it is imperative to identify and understand mechanisms by which nonandrogens can affect the proliferation of normal and malignant prostatic epithelial cells. Transforming growth factor beta (TGFP,) is a multifunctional growth regulatory factor which may either inhibit or stimulate cell proliferation, dependent upon the cell type [2]. Analysis of both human prostate carcinoma cell lines and limited data
Received for publication January 24, 1992; accepted June 16, 1992. Portions of this study were presented at the annual meeting of the American Association of Cancer Research, held in Washington, D.C., May 21-26, 1990. Address reprint requests to Joy L. Ware, Ph.D., Department of Pathology, Box 662, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA 23298.
0 1992 Wiley-Liss, Inc.
224
Watts and Ware
from specimens of benign prostatic hyperplasia implicates TGFP, in the regulation of prostatic epithelial cell growth and behavior. Two androgen-insensitive human prostate carcinoma cell lines, PC-3 [3] and DU145 [4], possess receptors for TGFP,, express TGFP, mRNA [ 5 ] , and secrete TGFP, into the culture media [5,6]. In contrast, the androgen-sensitive LnCAP does not contain detectable TGFP mRNA nor express high-affinity TGFP receptors [ 5 ] . Neither anchorage-independent nor anchorage-dependent growth of LnCAP is affected by TGFP, [5,7]. Thus, all prostatic tumor cell lines tested to date are either inhibited by TGFP, or unaffected. Northern analysis of human benign hyperplastic prostatic tissue reveals mRNA for both TGFP, and TGFP, [8]. If TGFP ,-mediated inhibition of prostatic tumor cell proliferation is relevant to the evolution of the prostatic tumor phenotype, cells which are resistant to this inhibition would be predicted to possess a selective growth advantage in vivo. To test this hypothesis, we have isolated and characterized TGFP ,-resistant clones of 1LN-PC-3-1A cells, a highly metastatic subline [9-111 of PC-3 [3]. TGFP, inhibited both anchorage-independent and anchorage-dependent proliferation by 1 -LN-PC-31A (designated 1-LN) cells in a dose-dependent manner. A relatively stable TGFP,resistant subpopulation, 1-LN-R2, was obtained by expansion of a colony of 1-LN cells growing in the presence of 2 ng/ml TGFP . Comparison of the TGFP I sensitivity of clones of 1-LN and subclones of 1-LN-R2 revealed that 1-LN-R2 was enriched with clones resistant to TGFP, . Both the sensitive and resistant clones were tumorigenic in athymic nude mice, yet some TGFP,-sensitive clones grew with a significantly prolonged latent period.
,
MATERIALS AND METHODS The PC-3 cell line was originally obtained from a bone marrow metastasis of human prostate cancer [3]. The subline of PC-3, designated 1-LN, was isolated from a lymph node metastasis in a PC-3 tumor-bearing athymic nude mouse and is highly metastatic in adult athymic nude mice [9-111. This human prostatic carcinoma cell line was routinely cultured in RPMI 1640 supplemented with 10% defined heatinactivated calf serum (HICS, Hyclone Laboratories, Logan, UT), glutamine, and gentamicin. Generation of TGFp,-Sensitive and -Resistant Variants
TGFf3, purified from porcine platelets was reconstituted in 4 mM HC1 containing 1 mg/ml bovine serum albumin (BSA) (R and D Systems, Minneapolis, MN). Each well of a 24-well tissue culture plate was coated with 0.6% agar-RPMI 1640. 2 X lo3 cells of the metastatic 1-LN subline were plated as an overlay in RPMI 1640 containing 0.3% agar, HICS, and the indicated final concentration of TGFP, (0.1, 0.5, 1 .O, or 2.0 ng/ml). Control wells contained no TGFP, but received an equal volume of acidified BSA. Plates were maintained at 3 7 T , 5% CO, for 12-14 days. Isolated cell colonies which grew in the presence of 2 ng/ml TGFP, were retrieved from the soft agar and 3 colonies-designated 1-LN R1, R2, and R3-were expanded in adherent culture in the absence of additional TGFP, . These cells were exposed to single-step selection to permit observation of the stability of this phenotype. Nine clones obtained by
TGFP, Response Variants in Human Prostate
225
limiting dilution cloning of colony R2 were subsequently retested for resistance to varying concentrations of TGFP in the anchorage-independent growth assay, and produced clones designated R2- 1 to R2-9. Nonselected clones of 1-LN were obtained randomly by limiting dilution cloning. These clones, designated 1-LN clone 1 to 1-LN clone 9, were expanded in culture in the absence of added TGFP, and were examined for their sensitivity to different concentrations of TGFP Effect of TGF& on Anchorage-Independent Growth
Each well of a 24-well plate was initially layered with 0.6% agar-RPMI 1640. A top layer containing 2.5 X lo3 1-LN cells or cell clones suspended in RPMI supplemented with 0.3% agar, 5% HICS, and 0-10 ng/ml of TGFP, was added to triplicate wells for each TGFP, concentration tested. Plates were cultured as previously described for 12-14 days and colonies 50 p m in diameter or greater were counted. The anchorage-independent growth assays were periodically performed on the cell clones to document continued sensitivity or resistance in the presence of TGFP
,.
Effect of TGFPl on Growth of Adherent Cells
We examined the effect of TGFP, on the growth of 1-LN clones in monolayer culture using tritiated (3H)-thymidine incorporation measured by trichloroacetic acid (TCA) precipitable counts and autoradiography . Autoradiography permits the determination of 3H-thymidine incorporation on a per cell basis; TCA precipitable 3Hthymidine labelling reflects DNA synthesis by the whole cell population. TGFP,resistant or -sensitive clones (3.5 X lo3) were plated in triplicate wells of a 96-well plate in RPMI media supplemented with 5% HICS and the indicated concentrations of TGFP,. After 48 hours, cells received 'H-thymidine (2 pCiim1; specific activity 2 CiimM). Ascorbic acid was added at 1, 6, 18, 24, 36, or 48 hours to stop DNA synthesis, and the relative radioactivity incorporated into TCA-precipitable material was determined by liquid scintillation counting. For autoradiographical analysis, clones were plated in duplicate in sterile 4-well chamber slides (4 X lo4 celldwell) in HAMSF12 culture medium supplemented with 5% calf serum. Clones were cultured in the indicated concentration of TGFP, ( 0 , 0. I , 0.5, and 2.0 ngiml). After 72 hours in culture, 'H-thymidine was added to each well (5 mCi/ml, specific activity 45-60 Ci/mmole). Forty-eight hours later, each slide was washed, fixed in methanol, and air dried. Each slide was then dipped in autoradiographic emulsion (NTB-3, Kodak, Rochester, NY) and stored at 4°C. After 7 days of exposure, each tissue culture slide was developed and counterstained with hematoxylin. A minimum of 500 cell nuclei were evaluated per well to permit the calculation of the labeling index (percent 3H-thymidine incorporation). In Vivo Behavior TGFP,-sensitive 1-LN clones 1 , 4 , 5 , and 7, and resistant clones R2-6 and R2-4 were compared for tumorigenicity and tumor growth rate in vivo. Specific pathogenfree male athymic nude mice, 6-8 weeks old, were obtained from the Massey Cancer
226
Watts and Ware
Center, Richmond, VA, Athymic Nude Mouse Colony. Tumor cells (1 X lo5, 5 X lo5, or 1 x lo6) were injected subcutaneously into the dorsal side of each mouse. Tumor growth was monitored by measurement with calipers, and tumor volume was estimated by the following formula: L x w2 2 ’ where L = length and W = width. Mice were sacrificed approximately 35 days after injection, or when the tumor burden reached 300-400 mm3. Mink Lung Assay for TGFP, Growth Inhibition
CCL-64 mink lung epithelial cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% Nu-Serum in a humidified incubator at 37°C and 5% CO, [ 121. Cells from subconfluent cultures were removed with trypsin-EDTA, washed, and adjusted to 5 X lo5 in an assay buffer consisting of DMEM with 0.2% Nu-Serum. Then, 24-well plates were seeded in duplicate with 2 x lo5 cells/well and were incubated for 1-6 hours. To prepare conditioned media, serum was eliminated from 1-LN clone 4 or R2-6 cell cultures with 4 changes of serum-free media over a period of 6-24 hours. Cells were incubated with fresh serum-free MEM containing BSA (0.5 mg/ml) and 1 pg/ml each of aprotinin, leupeptin, and pepstatin A. The media was conditioned for an additional 24 hours, after which it was removed and centrifuged at 15,000 g for 5 minutes. Unconcentrated conditioned media collected from either the TGFP -resistant or -sensitive 1-LN cell clones was added to the CCL-64 cells in each well of the 24-well plate. For comparison, each well of the conditioned media was also acidified and/or treated with 10 pg (1,000-fold excess) neutralizing antibody to TGFP, or TGFP2 (R and D Systems). The CCL-64 cells and various conditioned media were incubated for 20 hours at 37°C in a humidified incubator. 3H-thymidine (0.5 pCi/well, specific activity 40-60 Ci/mmole) was added to each well and incubated an additional 3 hours. The cells were then washed and fixed in methanol, air dried, and trypsinized. The contents of each well were transferred to scintillation vials and analyzed in a liquid scintillation counter.
,
TGFP, Protein Preparation and Analysis
R2-6 and 1-LN clone 4 tumor clones were assessed for secretion of TGFP, protein using a modification of the procedure for Western immunoblotting [ 131. Serum-free conditioned media was collected, as described previously, from the R2-6 and 1-LN clone 4 cells and was concentrated with ultrafiltration with a YMlO membrane (Amicon Corp., Danvers, MA). Protein was electrophoresed through a 15% SDS polyacrylamide gel [ 141 and transferred to an Immobilon filter. The presence of TGFP, was detected immunologically with a polyclonal chicken anti-TGFP antibody using a modification of the procedure distributed by R & D Systems. The filter was blocked 2-1 2 hours with borate-buffered saline containing 1% bovine serum albumin (BBS/BSA). Chicken anti-TGFP, (R & D Systems, Minneapolis, MN) was diluted 1:10,000 in BBS/BSA and was incubated overnight at room temperature. Following 5 washes for 5 minutes in BBS, biotinylated rabbit anti-chicken antibody
,
TGFP, Response Variants in Human Prostate
227
TABLE I. Comparison of Anchorage-Independent Growth of PC-3, 1-LN, and 1-LN-R2 Cells in the Presence of Increasing Concentrations of TGFP, 0 PC-3 1-LN 1-LN-R2
103 C 12 121 f 22.7 87 2 13.8
0.1 101 i 4.4 176 i 22.6 158 5 25
TGFP, (ng/ml) 0.5 41.5 k 5 47 t 5.3 145 C 17.5
1
2
20 2 2.9
14.5 t 1.4 10.3 2.4 76 f 12.7
-
108 t_ 19.1
*
Results expressed as mean number of colonies/well and are the average of 2-3 experiments performed in triplicate 2 standard error of the mean.
(1:2,500, Zymed Laboratories, South San Francisco, CA) was added for 2 hours. After 5 washes in BBS, the filter was incubated with alkaline phosphatase labelled streptavidin (1:2,500, Southern Biotech, Birmingham, AL). Following 5 final washes in BBS, the filter was equilibrated in carbonate buffer and NBT/BCIP substrate was added. The reaction was stopped after 15-30 minutes by replacing the NBT/BCIP solution with BBS containing 5 mM EDTA. RESULTS Effect of TGFP, on Anchorage-Independent Growth Exposure of the parental PC-3 line and the metastatic 1-LN subline to TGFP, produced a dose-dependent inhibition of colony formation in soft agar. Low concentrations of TGFP, (0.1 ng/ml) consistently stimulated colony formation among 1-LN cells to levels 20% above control values, while a more modest stimulatory effect on parental PC-3 cells was observed (Table I). At 2 ng/ml TGFP,, colony formation by both PC-3 and 1-LN was greatly inhibited. Single colonies of 1-LN cells, which grew in the presence of 2 ng/ml TGFP, were recovered from 3 different wells and expanded in culture. When subsequently retested for sensitivity to TGFP, , only one of these, designated 1-LN-R2, produced significant numbers of colonies in the presence of 2 ng/ml TGFP1. Furthermore, 1-LN-R2 cells were not inhibited by 5 or 10 ng/ml TGFP,. As illustrated in Table I, pronounced stimulation of colony formation with 0.1-0.5 ng/ml often occurred, and colony formation only decreased slightly at 2 nglml. Comparison of the dose-response behavior of 1-LN and 1-LN-R2 exposed to varying concentrations of TGFP, is facilitated in Fig. 1, where the data from Table I are plotted. The anchorageindependent growth assays were performed periodically on passages 5 -22 to reaffirm and document the response of 1-LN and 1-LN-R2 cells to TGFP,. 1-LN cells and 1-LN-R2 cells were subcloned and 9 clones of each were isolated and retested in an anchorage-independent growth assay along with the varying concentrations of TGFP, . Colony formation of all 9 randomly chosen clones from the 1-LN line was inhibited by TGFP, at doses ranging from 0.5 to 2.0 ngiml (Fig. 2a). TGFP, did not inhibit the growth of 6 of 9 clones obtained from the TGFP,resistant R2 colony as effectively as it inhibited 1-LN clone 4 cells at concentrations of 0.5-2 ng/ml (Fig. 2b). Although the original R2 subline remained resistant to TGFP, for up to 3 months in vitro, cells of clone R2-6 spontaneously exhibited sensitivity after 12 passages. Thus, TGFP, resistance was not stable during prolonged culture. In con-
228
Watts and Ware 180;
t 1-LN-R2
I160: 160 8Oj l
J
t 1-LN
-
1404 0
-
5 1201 0
5
100;
-
t 80:n
f
2
SO: -
40 1
20 1
0'
I
I
0.1
0.5
I
1
2
TGF-61 (nglrnl)
Fig. 1. Comparison of anchorage-independent growth of 1-LN and I-LN-R2 in the presence of varying concentrations of TGFP,. Results are expressed as mean colonies 2 standard error of the mean. (Data from experiments described in Table I.)
trast, 1-LN clone 4 cells from long-term cultures remained equally sensitive to TGFP, in the anchorage-independent growth assays. Effect of TGFP, on the Growth of Adherent Cells
The monolayer growth of both R2-6 and 1-LN clone 4 in the presence of TGFP, was evaluated using 3H-thymidine incorporation as determined by TCA precipitable radioactivity and/or autoradiography . 3H-thymidine incorporation by parental 1-LN cells was inhibited by TGFP, 45-60% at 0.5-2.0 ng/ml. However, cells from the TGFP ,-resistant colony 1-LN-R2 demonstrated no decrease in 3H-thymidine uptake when exposed to any concentration of TGFP, up to 2.0 ng/ml. Two clones of 1-LN-R2 which were refractory to the inhibitory effects of TGFP, (clones 3 and 6) and 2 clones of 1-LN which were sensitive to TGFP, (clones 1 and 4) in the anchorage-independent growth assay were tested using 3H-TdR uptake to determine sensitivity to TGFP, in monolayer culture. As Fig. 3 indicates, 3H-thymidine incorporation was inhibited similarly in 3 of 4 clones. Only 1-LN-R2 clone 6 (R2-6) remained unaffected by TGFP, in adherent cell culture. In fact, at 2.0 ng/ml, 3H-thymidine incorporation by this clone was stimulated nearly 2-fold. Autoradiography of adherent cell cultures demonstrated that the number of radiolabelled tumor cell nuclei observed in the cells of TGFP ,-resistant clone R2-6 was only modestly decreased by the addition of up to 2.0 ngiml TGFP, (84% of control values). These results support the data obtained from 3H-thymidine incorporation into TCA-precipitable material of adherent R2-6 cells. Each assay determined that DNA incorporation of the radiolabel was not significantly inhibited by 2 ng/ml TGFP,. In contrast, the labeling index of the I-LN clone 4 cells was reduced to 40% and 45% in the presence of 0.5 and 2.0 ng/ml TGFP,, respectively (data not shown). Tumorigenicity
Comparisons of the tumorigenicity and growth rate in vivo of the TGFP,sensitive 1-LN clones 1, 4, 5, and 7, and resistant clones R2-4 and R2-6, were conducted. 1-LN clone 1 and clone 4 exhibited similar sensitivity to TGFP, in the
TGFP, Response Variants in Human Prostate
229
1 -LN (T9) Clones
h
400
300
I
1
TGF-8 conc. -0.1 ng/ml Lz4 0.5 ng/ml e0 1.O ng/ml 0 2 . 0 "g/ml
1
3
4
5
6
7
8
9
10
clone number B
1-LN (R2) Clones
'I
TGF-p conc. 1 0 . 1 ng/ml 0.5 ng/ml
rza 1.0 ng/ml
300
-
e
ri3 2.0 "g/rnl
c
c
8
200
%
0
x 100
-0 n
1
2
3
4 5 6 clone number
7
8
9
10
Fig. 2. A: Comparison of anchorage-independent growth of 9 different clones of 1-LN in the presence of varying concentrations of TGFP,. B: Comparison of anchorage-independent growth of 9 different clones of 1-LN-R2 in the presence of varying concentrations of TGFP,. Concentrations of TGFP, are indicated by the legend in the figure inset.
anchorage-independent growth assays and a similar delayed latent interval after subcutaneous (sc) injection into nude mice (Fig. 4). 1-LN clones 5 and 7, although exhibiting reduced colony formation in the presence of TGFP, , demonstrated rapid growth in nude mice. 1-LN clone 4 was selected for intensive analysis, because TGFP consistently inhibited both anchorage-independent and anchorage-dependent growth. Clone R2-4 displayed resistance to TGFP, in the anchorage-independent growth assay, yet exhibited a variable rate of growth in vivo. R2-6 was chosen for additional study, because TGFP, was unable to inhibit both monolayer and colony formation. SC injection of 5 X lo5 cells of each clone produced tumors in all mice, but there was a slight increase in the time to tumor formation in mice injected with 1-LN clone 4 cells. Comparison of athymic nude mice injected sc with 1 X lo5 1-LN clone 4 cells at 30 days showed a significantly longer latency period compared to mice injected with 1 X lo5 clone R2-6 cells (Figs. 5 and 6). There was a significant tumor size difference between the 2 groups of mice at day 30 ( P < 0.0001, t-test). By day 25 post-injection, only one mouse injected with 1-LN clone 4 cells had reached a
,
230
Watts and Ware
I P)
5
Fig. 3. Comparison of 3H-thymidine incorporation by clones in monolayer culture exhibiting resistance or sensitivity to growth inhibition by TGFP, in the anchorage-independent growth assay. The plot is representative of 2 experiments.
tumor volume of 50 mm3, while the remainder of mice had tumors of only 15 mm3 or less (Fig, 6b). Conversely, all mice injected with 1 X lo5 R2-6 cells quickly reached a tumor volume of 50 mm3 by an average of 19 days postinjection (Fig. 6a). At 38 days, the average tumor volume of the R2-6 tumors was 5-8 times greater than tumors resulting from subcutaneous injection of 1-LN clone 4 cells. When tumors from mice injected with 1-LN clone 4 cells were allowed to continue to grow past 38 days, they entered log phase growth and grew to a size comparable to that observed in animals injected with R2-6 cells. Thus, in vivo growth was retarded but not ablated. Gross metastases were not observed in mice injected with any number of tumor cells. Of 12 mice injected with 1 X lo5 1-LN clone 4 cells included in the study, tumors in 2 mice exhibited anomalous proliferation, similar to that seen in animals injected with R2-6. However, even these 2 faster growing tumors had an increased time to tumor formation. TGFP, and Secretion Both 1-LN clone 4 and R2-6 cells secreted TGFP, into the culture as assessed by Western immunoblotting, following SDS-PAGE under reducing conditions (Fig. 7). When an estimated 700, 350, and 100 pg of total proteidlane was loaded from both 1-LN clone 4 and R2-6 supernatants, immunologically reactive bands were detected at approximately 25 kDa, with weaker bands observed at 35-40 kDa and 82-90 kDa. In lanes 1 and 2, containing 700 pg total protein, it is probable that the anomalous migration of TGFP, resulted from protein overloading. The bands at approximately 40 and 90 kDa may represent binding of the polyclonal TGFP, antibody to the pro region of the TGFP ! precursor molecule and the TGFP precursor, respectively [ 151. Loading 350 pg total protein produced a strongly immunoreactive band in lane 3, containing 1-LN clone 4 supernatant, with slight immunoreactivity observed in the comparably loaded lane containing R2-6 supernatant. We conclude that R2-6 conditioned media contained quantitatively less TGFP, protein than 1-LN
,
TGFP, Response Variants in Human Prostate
231
1000
1 I
100 a -
-€E W
I
3
9 U
0
5+
10
1
TIME (days)
Fig. 4. Growth of 3 1-LN clones sensitive to the inhibitory effects of TGFP, in vitro after subcutaneous injection of 1 X lo5 cells in athymic nude mice. Clone 1, A; clone 5 , 0 ;and clone 7 , . Symbols indicate individual mice.
clone 4. Relative to purified TGFP, (R & D Systems), the estimated migration of the dominant 1-LN clone 4 TGFP, immunoreactive band was approximately 25 kDa. Mink Lung Cell Inhibition by TGFP,
We assessed the presence of secreted TGFP, in conditioned media from the cultures of 1-LN clone 4 cells and R2-6 cells using the CCL-64 mink lung epithelial cell inhibition bioassay. Untreated conditioned media from R2-6 or 1-LN clone 4 cells did not produce a significant inhibition compared to control media alone (Fig. 8). In contrast, acidified serum-free conditioned media from R2-6 inhibited CCL-64 growth by 83% and 1-LN clone 4 cells inhibited DNA synthesis by 87% compared with unconditioned control media. Untreated media in the presence of a 1,000-fold excess of neutralizing antibody to TGFP, exhibited no differences from control. In contrast, acidified conditioned media of R2-6 and 1-LN clone 4 in the presence of neutralizing antibody to TGFP, produced only 35% and 25% inhibition, respectively, compared to control media. Pretreatment of R2-6 and 1-LN clone 4 conditioned media with anti-TGFP, antibody had no effect on the inhibitory activity of either conditioned media. Despite inclusion of a 1,000-fold excess of neutralizing antibody to TGFP,, acid-activated conditioned media of R2-6 and 1-LN clone 4 cells inhibited CCL-64 cell proliferation by 89.5% and 94.4%, respectively.
Watts and Ware
232
B
A 1000
1000
900 800
900 800
700
700
600 500
I
600 500
400
400
300
300
200
200
-EE
E E
-
UJ
w
I
3
p
I 100
3
p
100
a:
90 80
5
70
LT
60
f
I-
50
50
40
40
30
30
0
20
0
f
90 80 70 60
20
10
10
- 1
1
10
20
30
TIME (days)
40
50
10
20
30
40
50
TIME (days)
Fig. 5 . Growth of TGFP,-resistant clone R2-6 (A) and TGF@,-sensitive 1-LN clone 4 (B) after subcutaneous injection of 5 X l o 5 cells in athymic nude mice. Each symbol indicates an individual mouse.
DISCUSSION
In this study, we have examined the impact of TGFP, on the proliferative response of a metastatic derivative of the androgen-independent human prostatic carcinoma cell line PC-3. Although both anchorage-independent and adherent monolayer growth of the 1-LN cells was inhibited by TGFP, concentrations greater than 0.1 ngiml (Table I), this subline was composed of cells which exhibited heterogeneous sensitivity to this factor. Both TGFP,-sensitive and -resistant clones (I-LN clone 4 and R2-6, respectively) secreted immunodetectable TGFP, (Fig. 7) in latent form (Fig. 8). Our finding that the growth of 1-LN cells in monolayer cultures, as well as colony formation in soft agar, was inhibited by exogenous TGFP, is consistent with the reports of others who demonstrated that TGFP, inhibited the growth of the parental PC-3 line in monolayer [ 161 and suppressed anchorage-independent proliferation of both PC-3 and DU-145 [ 5 ] . Our report not only extends their observations to a highly metastatic variant of PC-3 [9-111, but also demonstrates the isolation of spontaneously occurring TGFP,-resistant subpopulations of 1-LN cells, such as R2 and clones R2-6. In addition, in vivo analysis demonstrated no consistent
TGFP, Response Variants in Human Prostate A
B 1000
900 800 700
--
1000
-
700 -
600 -
400 -
400 -
300 -
300
-
200
-
100
--
500
500
200
-
-
" 5
E E w
I
3
9
a: 0
f +
-
900 800 -
600 -
--
233
E E
w
2
90 80 70 -
3
100
9 8 fI-
60 50 40
-
30
-
90 80
1
20 10 1
-
1 1
I
I
I
I
10
20
30
40
50
1 10
20
30
40
50
TIME (days)
TIME (days)
Fig. 6. Growth of TGFP,-resistant clone R2-6 (A) and TGFP,-sensitive 1-LN clone 4 (B) after subcutaneous injection of 1 X lo5 cells in athymic nude mice. Statistical significance, P < 0.0001, t-test. Each symbol indicates an individual mouse. Two mice out of 12 injected with 1 X 105 1 -LN clone 4 cells exhibiting anomalous growth were not included in the graphed results but are discussed in the text.
correlation between in vitro sensitivity to TGFP, and initiation of tumor growth in athymic nude mice among the different sensitive and resistant clones tested. Thus, the TGFP, response phenotype in vitro was not always predictive of growth delay in vivo. In vivo behavior of clones with different TGFP, in vitro response phenotypes was variable. Two of 4 randomly isolated clones of I-LN which were inhibited by TGFP, in soft agar (clone 4 and clone 1) exhibited a longer latency period for tumor formation compared to 2 other TGFP -sensitive clones when injected subcutaneously into athymic nude mice. These other 1-LN clones (clones 5 and 7) sensitive to the growth inhibitory effects of TGFP in the anchorage-independent growth assay exhibited the same latency and grew in vivo at the same rate as the TGFP,-resistant R2-6 clones (Fig. 4). Detection of differences between the growth of sensitive and resistant clones in vivo was dependent upon the number of tumor cells initially injected. Subcutaneous injection of 1 X lo5 1-LN clone 4 cells resulted in significantly longer latent periods
,
,
234
Watts and Ware
Fig, 7. Detection of TGFP, by Western immunoblot. Anti-TGFP, antibody at 1:10,000 dilution was used for the detection of secreted TGFP, in conditioned media from 1-LN clone 4,lanes 1, 3, 5 ; and R2-6, lanes 2, 4,6. Lane 7 contains purified TGFP, (6 ng, R & D Systems). This figure represents 1 of 3 similar experiments.
to tumor formation relative to R2-6 (Fig. 6b). However, sc injection of 1 X lo6 or 5 x los resistant or sensitive 1-LN tumor cells generated similar growth curves that were not statistically different. We speculate that regulatory factors present in vivo may be responsible for observed differences in tumor latency. Thus, the microenvironment at the site of injection may have influenced the establishment of the tumor. It is possible that the increased TGFP, levels secreted by both the 1-LN clone 4 tumor cells and tissues in the microenvironment may have produced a combined threshold concentration that initially inhibited the growth of small numbers of TGFP,-sensitive cells. In effect, the transiently high local levels of TGFP, may act in an autocrine fashion to downregulate the growth of TGFP,-sensitive tumor cells. The ability to elude this regulatory mechanism, as observed in most of the TGFP,-resistant R2-6 tumors, may have conferred a growth advantage in vivo. As all tumors enlarged, albeit the 1-LN clone 4 tumors more slowly, tumor cell numbers would finally increase above a critical number and favor exponential tumor growth. Clearly, additional investigations are necessary to quantify the local concentrations of TGFP, at the site of prostatic tumor injection. We propose that augmented secretion of TGFP, by 1-LN clone 4 cells in vivo may have altered the levels of TGFP, in the vicinity of tumor cell injection. This hypothesis is supported by the Western blot data with 3 different concentrations of conditioned media, which revealed that immunodetectable 25 kD TGFP was greater in supernatant obtained from 1-LN clone 4 cells. In addition, conditioned media from TGFP,-sensitive 1-LN clone 4 inhibited CCL-64 mink lung epithelial cell DNA synthesis by 88%. This acid-activated inhibition of 1-LN clone 4 media media was abrogated 60% with the use of a neutralizing antibody to TGFP, but not by a neutralizing antibody to TGFP,. The increased TGFP, protein secretion of 1-LN
,
TGFP, Response Variants in Human Prostate
235
Fig. 8. TGFP, activity in conditioned media from TGFP,-resistant R2-6 and TGFP,-sensitive I-LN clone 4 cells. Data are expressed as a percent of control k SEM. The mean percent of control was determined from duplicate samples.
clone 4 cells suggests that these cells may produce TGFP, that then acts in an autocrine fashion to regulate tumor cell growth in vitro and in vivo. Based upon this autocrine hypothesis of growth regulation by growth factors, we expected that TGFP,-sensitive 1-LN clone 4 tumors would exhibit TGFP, resistance in vitro after successful in vivo passage [ 171. However, the 1-LN clone 4 cells recovered from in vivo tumors remained as sensitive to TGFP, in the anchorage-independent growth assays as they were prior to in vivo growth. Finally, we concluded that resistance to TGFP, was not stable because R2-6 clones cultured 12 or more passages in vitro spontaneously displayed sensitivity to the presence of TGFP, . Our observations are consistent with multiple investigators who have shown that clones exhibit greater phenotypic variability and instability than the parental population from which the clone was derived [18, reviewed in 19-21]. Obviously, the combination of positive and negative growth factors produced in vitro plays a role in the regulation of R2-6 tumor cells. In summary, this paper described the first isolation of TGFP,-resistant subclones from a TGFP,-sensitive parental human prostatic carcinoma cell line. This distinct subpopulation of 1-LN cells refractory to the inhibitory effects of TGFP, may represent an important stage in the progression of neoplastic transformation in prostatic malignancy. These results, combined with the observed variation in phenotypic stability among these cells, suggest that the impact of local TGFP, levels will be
236
Watts and Ware
greatest at low tumor cell population size. Consequently, TGFP, may be more significant in modulating the growth of prostatic tumor micrometastases than of the primary tumor. ACKNOWLEDGMENTS
This work was supported by NCI grant CA.50609. We thank Winston C. Chandler for his technical expertise, and Sharon R. Collins for her assistance with the CCL-64 bioassays. REFERENCES I . Cunha GR, Donjacour AA, Cooke PS, Mee S, Bigsby RM, Higgins SJ, Sugimura Y: The endocrinology and developmental biology of the prostate. Endocrin Rev 8:338-362, 1987. 2. Sporn MB, Roberts AB, Wakefield LM, DeCrombrugghe B: Some recent advances in the chemistry and biology of transforming growth factor-p. J Cell Biol 105:1039-1045, 1987. 3. Kaighn ME, Naragan K S , Ohmuki Y, Lechner JF, Jones LW: Establishment and characterization of a human prostatic carcinoma cell line (PC-3). Invest Urol 17:16-23, 1979. 4. Stone KR, Mickey DD, Wunderli H, Mickey GH, Paulson DF: Isolation of a human prostate carcinoma cell line (DU145). Int J Cancer 21:274-281, 1978. 5. Wilding G, Zugmeier G, Knabbe C, Flanders K, Gelmann E: Differential effects of transforming growth factor p on human prostate cancer cells in vitro. Mol Cell Endocrinol 62:79-87, 1989. 6. Ikeda T, Lioubin MN, Marquardt H: Human transforming growth factor type P2: Production by a prostatic adenocarcinoma cell line, purification, and initial characterization. Biochemistry 26:24062410, 1987. 7. Schuurmans ALG, Bolt J , Mulder E: Androgens and transforming growth factor-p modulate the growth response to epidermal growth factor in human prostatic tumor cells (LNCaP). Mol Cell Endocrinol 60:101-104, 1988. 8. Mori H, Maki M, Jaye M, Igarashi K, Yoshida 0 , Hatanaka M: Increased expression of genes for basic fibroblast growth factor and transforming growth factor type p2 in human benign prostatic hyperplasia. Prostate 16:71-80, 1990. 9. Ware JL, Paulson DF, Mickey GM, Webb KS: Spontaneous metastasis of cells of the human prostate carcinoma cell line PC-3 in athymic nude mice. J Urol 137:1304-1306, 1982. 10. Ware JL, Delong ER: Influence of tumour size on human prostate tumor metastasis in athymic nude mice. Br J Cancer 51:419-423, 1985. 1 1. Ware JL, Lieberman AP, Webb KS, Vollman RT: Factors influencing phenotypic diversity of human prostate carcinoma cells metastasizing in athymic nude mice. Exp Cell Biol 53:163-169, 1985. 12. Danielpour D, Dart LL, Flanders KC, Roberts AB, Sporn MB: Immunodetection and quantitation of the two forms of transforming growth factor-beta (TGF-P1 and TGF-P2) secreted by cells in culture. J Cell Physiol 138:79-86, 1989. 13. Florini JR, Roberts AB, Ewton DZ, Falen SL, Flanders KC, Sporn MB: Transforming growth factor-beta. A very potent inhibitor of myoblast differentiation, identical to the differentiation inhibitor secreted by Buffalo rat liver cells. J Biol Chem 261:16509-16513, 1986. 14. Laemmli UK: Cleavage of structural proteins during the assembly of bacteriophage T4. Nature (Lond.) 227:660-685, 1970. 15. Gentry LE, Lioubin MN, Purchio AF, Marquardt H: Molecular events in the processing of recombinant type I pre-pro-transforming growth factor beta to the mature polypeptide. Mol Cell Biol 8:4162-4168, 1988. 16. Okutani T, Nishi N, Kagawa Y, Takasuga H, Takenaka 1, Usui T, Weda F: Role of cyclic AMP and polypeptide growth regulators in growth inhibition by interferon in PC-3 cells. Prostate 18:73-80, 1991. 17. Sporn MB, Todaro GT: Autocrine secretion and malignant transformation of cells. N Engl J Med 303:878-880, 1980. 18. Poste G , Doll J, Fidler IJ: Interactions among clonal subpopulations affect the stability of the
TGFP, Response Variants in Human Prostate
237
metastatic phenotype in polyclonal populations of B16 melanoma cells. Proc Natl Acad Sci USA 78:6226-6230, 1981. 19. Heppner GH: Tumor heterogeneity. Cancer Res 44:2259-2265, 1984. 20. Miner KM, Kawaguchi T, Uba GW, Nicholson GL: Clonal dnft of cell surface, melanogenic, and experimental metastatic properties of in vivo-selected, brain meninges-colonizing murine B- 16 melanoma. Cancer Res 424631-4638, 1982. 21. Poste G , Tzeng J , Doll J, Grieg R, Rieman D, Zeidman I: Evolution of tumor cell heterogeneity during progressive growth of individual lung metastases. Proc Natl Acad Sci USA 79:6574-6578, 1982.
