In Vitro Cell Dev. Biol. 29A:4-6, January 1993 © 1993 Tissue Culture Association 0883-8364/93 $01.50+0.00

Letter to the Editor FIBROBLAST INHIBITION OF TUMOR CELLS MAY BE MEDIATED BY TGF-~I

Dear Editor:

Several reports show that normal human fibroblasts express an anti-tumor activity in vitro. Fibroblasts, derived from embryonic or neonatal tissue, secrete a soluble anti-tumor substance that is active against a wide range of cell lines established from human and mouse solid tumors (Kirk et al., 1983; Tanaka et al., 1985; Wu et al., 1985; Shirasuna et al., 1988). We monitored the effects of a fibroblast monolayer on the clonal growth of cancer ceils seeded in an overlying agarose gel. Using neonatal human lung fibroblasts, we observe a dose-dependent inhibition of colony forming efficiency with increasing number of fibroblasts seeded (Kirk, 1983). Compa-

rable inhibitions are observed for several different human tumor lines derived from prostate (PC-3, DU-45), bladder (J82), endome- trium (HEC-1A) and brain (Hs683t).

Further studies using PC-3 cells have shown the inhibitory effect to be a transforming growth factor 81 like activity (TGF-/31). Neu- tralizing antibody to TGF-/31 caused about a 60% reversal of inhibi- tion in serum-containing co-cultures. A 100% reversal of the inhibi- tory activity was observed in serum-free conditioned medium con- centrated by uhrafiltration (5 kD cut-off). Acid treatment of the serum-free conditioned medium increased its inhibitory activity ap- proximately 20-fold. This enhanced activity was also reversed com-

CM CM Control + +

Anti-TGF B1 Anti-TGF B1 FIG. 1. Neutralization of the inhibitory activity of conditioned medium (CM) by anti-TGF/31. Serum-free CM from human lung

fibroblasts was concentrated over a YM100 ultrafihration membrane (Amicon), acid dialyzed against 1% acetic acid, lyophilized and resuspended in HEPES buffered saline. The PC-3 cells were seeded in soft agarose (2500/0.3 ml) in 24-well dishes. The CM extract was added at a 10X concentration and the neutralizing antibody against TGF ~1 (IgG fraction, R&D Systems) was added at 10 #g/ml. Cultures were photographed 20 days after treatment. Magnification X3.

FIBROBLAST SECRETED ANTI-TUMOR CELL ACTIVITY 5

pletely by anti-TGF-~l (Fig. 1). Other tests on the conditioned medium revealed that the inhibitor was heat stable and sensitive to both trypsin and reducing agents, suggesting it was indeed TGF-~ 1. One dissimilarity to TGF-~I was the fact that uhrafihration indi- cated that the active inhibitor had a molecular weight greater than 100 kD. However, subsequent purification of this activity on Mono- S and Reverse Phase HPLC chromatography showed it to have a similar molecular weight to TGF-fll (25 kD) as determined by western blot analysis. We used an acid treatment in our purification and this shift in molecular weight was comparable to that observed for acid activation of a latent form of TGF-~I (dimer of 390 amino acids) to an active form of two disulfide linked homodimers of 12 000 molecular weight (Lawrence et al., 1985).

The miniscule amount of fibroblast derived TGF-~I that was able to produce these effects was of interest. The amount of TGF- ~1 produced by the fibroblasts was estimated by comparing titra- tions of inhibitor, assayed as the crude conditioned medium concen- trate, and pure TGF-~I on PC-3 cell growth. In our serum-free system, the PC-3 cells were estimated to be 50% inhibited with only 6 pg/ml (Fig. 2). This represented about a thousand fold increase in the sensitivity of PC-3 cells to TGF-~I as reported previously (8.75 ng/ml; Fernandez-Pol et al., 1986). The increased sensitivity in our system possibly was due to the absence of complicating serum growth factors that were present in the assay of Fernandez-Pol et ai., 1986. In absolute terms, one million confluent lung fibroblasts were estimated to secrete 31.2 pg of total activated TGF-~I into serum-free medium per 24 hours. This represented about 0.00062% of the total protein secreted by lung fibroblasts. In com- parison, transformed mouse L929 cell cultures secrete about 120 times this amount, the activated TGF-fll representing 0.075% of the total protein secreted into serum-free medium (Feruandez-Pol et al., 1986). Normal human fibroblast cultures have also been reported to secrete into their supernatants less than 2% of epider- mal growth factor related transforming growth factors produced by several human tumor cell lines (Todaro et al., 1980).

ILl

2o

10.

. . . . . 0.001

2 4 '

• TGF I}1

I~ CM

\ \

\

r [ I ~ . 0~ I ' ~ ng/ml TGF I~1

. . . . Fold conc. of a 10 C M

Comparative Cormerltratlons

Fig. 2. Comparative inhibitory effects of TGF ~1 (R&D Systems) and concentrated conditioned medium from human lung fibroblasts (CM) on the growth of PC-3 ceils. The serum-free CM was processed as in Figure 1 and growth was assessed as colony forming efficiency (CFE) in serum-free aga- rose. Unrepresented standard error bars were too small to be drawn.

16

12

4 CM TGF 81

. . . . . . . . . . . . . . . . . . . . . . . . . . . nQ/mlTGF pl 0 0.001 0.01 0.1

r 1 ~ i i i I i i i I FOld cOI~" Of 0 2 4 S 8 10 CM

Compara t ive Concentrat ions

Fig. 3. Comparative stimulatory effects of TGF ~1 and CM (see Figure 2) on the growth of human lung fibroblasts. Fibroblasts were seeded in 24-well dishes (10 O00/well) in serum (2%) supplemented PFMR-4 me- dium. Test material was added 24 h later and cultures were pulsed with tritiated thymidine (4 gei/ml) for 5 h and then processed for the incorpora- tion of tritium into acid insoluble material using standard methods.

The inhibitory factor was purified as follows. Conditioned media was concentrated 450-fold using a SLY100 spiral cartridge fol- lowed by a YM100 membrane (Amicon). The retentate was diluted with water, then concentrated to remove low molecular weight me- dia components. The retentate was brought to 0.25 M in acetic acid and stored at 4 ° C overnight. After centrifugation to remove insolu- ble material, the supernatant was lyophilized. The factor was then subjected to chromatography on a Mono-S column (0.5 >( 5 em, Pharmacia) in 25 mM Tris, pH 8.6. The column was developed with a linear gradient of sodium chloride. Those fractions showing inhibitory activity were concentrated and applied to a reversed phase column (Vydac C4) which was then developed with a linear gradient of 88% acetonitrile, 2% 2-propanol in 0.1% trifluoroace- tic acid. Inhibitory fractions were again detected with the in vitro bioassay. When examined in a Western format these fractions re- sponded positively toward antibodies to TGF-fll (prepared at Am- gen, Inc.) and had a molecular weight corresponding to authentic TGF-~I (supplied by Dr. Keith Westcott). Without regard to losses which might occur during purification, the yield was approximately 5 ng/liter of conditioned medium. Although the inhibitory factor was impure at this stage, the coincidence of bioactivity with TGF-/31 strongly implicated that they were the same. Considerable effort would have been needed to prepare sufficient material for further purification and characterization.

Interestingly, TGF-~I and the crude preparation of fibroblast inhibitor both showed a dose-dependent stimulation of the clonal growth of the producer fibroblasts (Fig. 3). However, the amounts necessary for half-maximal stimulation were 2 -3 times greater than those required for half-maximal inhibition of the PC-3 cells.

In summary, the previously reported inhibition of PC-3 cells by the fibroblasts seems to be mediated by a TGF-/31 like molecule. The dynamics of how the latent fibroblast TGF-~I becomes acti- vated in unknown. The inhibition possibly is effected by the small

KIRK ET AL.

amount of active TGF-/~I already present in the untreated serum- free fibroblast conditioned medium. However, interaction between the secreted fibroblast products and the tumor cells resulting in activation of additional TGF-~I cannot be ruled out. The autocrine stimulation of fibroblasts by the activated TGF-/~I in the fibroblast/ tumor cell cocultures would obviously result in the production of more TGF-~I and would explain the correlation between fibroblast growth rate and PC-3 inhibition observed previously (Kirk and Kaighn, 1980). This study highlights the importance and the com- plexity of stromal-tumor cell interactions and predicts that augmen- tation of tissue levels of active TGF-~I would protect against tumor development.

ACKNOWLEDGEMENTS

Partially funded by the Weingart Foundation and the Flintridge Founda- tion. We thank Mr. Robert Olson for his help with the photography.

REFERENCES

Fernandez-Pol, J. A.; Klos, D. J.; Gram, G. A. Purification and biological properties of type B transforming growth factor from mouse trans- formed cells. Cancer Res. 46:5153-5161; 1986.

Kirk, D.; Kagawa, S.; Vener, G. Comparable growth regulation of five human tumor cell lines by neonatal human lung fibroblasts in semis- olid culture media. Cancer Res. 43:3754-3758; 1983.

Kirk, D.; Kaighn, M. E. Non-reciprocal interactions in normal-neoplastic human ceils. A quantitative, kinetic approach to cell interactions. Cell Biology International Reports 4:599-608; 1980.

Lawrence, D. A.; Pircher, R.; Juilien, P. Conversion of a high molecular weight latent beta-TGF from chicken embryo fibroblasts into a low molecular weight active beta-TGF under acidic conditions. Bio- chem. Biophys. Res. Commun. 133:1026-1034; 1985.

Shirasuna, K.; Morioka, S.; Watatani, K., et al. Growth inhibition and dif- ferentiation of human salivary adenocarcinoma cells by medium conditioned with normal human fibroblasts. Cancer Res. 48:2819- 2824; 1988.

Tanaka, A.; Matsuoka, H.; Uemura, H., et al. Production and characteriza- tion of tumor-degenerating factor. J. Natl. Cancer Inst. 74:575- 581; 1985.

Todaro, G. J.; Fryling, C.; De Larco, J. E. Transforming growth factors produced by certain human tumor ceils: polypeptides that interact with epidermal growth factor receptors. Proc. Natl. Acad. Sci. USA 77:5258-5262; 1980.

Wu, K-F.; Pope, J. H.; Ellem, K. A. O. Inhibition of growth of certain human tumor cell lines by a factor derived from human fibroblast- like cell lines. 1. Demonstration by mixed culture and by use of cell washings. Intl. J. Cancer 35:477-482; 1985.

David Kirk Todd Broberg Juan C. Irwin

William C. Kenney Lawrence W. Jones

Huntington Medical Research Institutes Pasadena, California 91101 (D. K., T. B., J. C. I., L. W. J.) and

Amgen Corp. Thousand Oaks, California 91320 (W. C. K.)

(Received 10 August 1992)