Characterization and Partial Purification of Human Epithelial Transforming Growth Factor
Damien J. Dunnington, Robert G. Scott, Mario A. Anzano, and Russell Greig
Department of Cell Sciences, L-109, SmithKline Beecham Pharmaceuticals, King of Prussia, Pennsylvania 19406-2799
A polypeptide growth factor has been partially purified from medium conditioned by the human adrenocortical carcinoma cell line SW13. This factor, designated h-TGFe, stimulates anchorage-independent growth of the SW13 cells. Similar activity was observed in human milk, and in conditioned media from seven of 14 epithelial cell lines. The SW13-derived activity is stable to low pH and 8M urea but labile to dithiothreitol and 2% sodium dodecyl sulfate. Human TGFe does not bind to heparin and fails to stimulate growth of endothelial cells in monolayer culture. The apparent molecular weight of h-TGFe is 59k by size exclusion chromatography in the presence of 8M urea and the activity binds strongly to cation exchangers. The activity elutes at 15-30% acetonitrile from a C18 reverse- phase column and has been partially purified by using a four-step chromatographic procedure. TGFe appears to be a novel growth factor produced by many epithelial cells and tissues.
Key words: adrenocortical carcinoma cells, protein purification, anchorage independent growth, epithelial cell proliferation
Proliferation of epithelial cells is affected by a number of polypeptide growth factors including transforming growth factors (TGF) type alpha [1] and beta [2], fibroblast growth factor [3], and an apparently novel protein known as TGFe [4]. Production of TGFe appears to be restricted to epithelial cells and tissues [5] and is measured by quantitation of colony formation in soft agar by the adrenocortical carcinoma cell line SW13 [5]. A form of TGFe present in acid ethanol extracts of bovine kidney has been extensively characterized [4,6] but its amino acid sequence has yet to be determined. This material has a molecular weight of 22,000-25,000 and appears to be a single-chain protein requiring intrachain disulfide bonds for full activity [4]. Bovine TGFe has been discriminated from previously described growth factors by molecular weight, receptor binding studies, and physical and biological properties [4].
Received May 15, 1990; accepted July 31, 1990.
In contrast to bovine TGFe, relatively little is known about the activity produced by human cells and tissues. Acid-ethanol extracts of human carcinoma tissues con- tained acid- and heat-stable polypeptides with molecular weights of 20,000-22,000 that were active in the SW13 soft agar assay [5]. The same report described growth- stimulating activity in medium conditioned by SW13 cells, implying the presence of an autocrine loop in these cells [5]. However, the properties of the SW13-secreted factor were not examined. In the present report, we describe partial purification of the SW13-derived material and its relationship to bovine TGFe.
MATERIALS AND METHODS
Cell Cultures and Growth Factors
All cell lines were obtained from the American Type Culture Collection and were free from mycoplasma contamination as determined by the Gen Probe assay (Gen Probe, San Diego, CA 92123) and broth culture (Microbiological Associates). Isoenzyme analysis demonstrated the human origin of SW13 cells (Corning Authenti- kit). Bovine basic fibroblast growth factor (bFGF) was supplied by Collaborative Research.
Collection of Conditioned Media
Cells were grown to confluence in Dulbecco’s modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum, in roller bottles. The cells were washed with phosphate-buffered saline (PBS), incubated in serum-free DMEM for 4-5 h, and again washed with PBS. Each roller bottle received 200 ml of serum-free, phenol red-free RPMI 1640 medium (Gibco) supplemented with 10 mg/L freshly prepared ferrous sulfate and 5 ng/ml sodium selenite (Collaborative Research). Cells were incubated with serum-free medium for 2 days and medium was acidified with 5 ml/L trifluoroacetic acid (TFA) and stored at 4℃. The acidified conditioned medium was centrifuged at 1,500g for 15 min and filtered through a 0.45 um membrane (Gelman), and biological activity was determined after concentration by ultrafiltration (Amicon YM5 membrane, 5,000 mw cutoff) and dialysis against 0.05% TFA. The following cell lines were surveyed for secretion of activity: SW48, SKC01, LOVO, CACO 2, HT-29, COLO 320DM, COLO 320HSR, SW948, DLD-1, HCT-15, WIDR, SW403, and SW742 (all derived from colon carcinomas) and NRK-52E (rat kidney epithelial cells) and SW13.
Biological Assays
TGFe activity was measured by using growth of SW13 cells in soft agar as described [5] except that the base and upper layers contained 0.5% and 0.3% agarose respectively in DMEM, and 10% fetal bovine serum, and 5,000 cells were plated per 3 cm dish.
Growth of fetal bovine heart endothelial (FBHE) cells in monolayer culture [7] was used to measure FGF activity. Samples were added to 3 cm dishes containing 2 ml DMEM, 10% fetal bovine serum. Each dish then received 100 ul of a suspension of 5 × 104 FBHE cells in DMEM, 10% serum. Cultures were incubated for 5 days, fixed with 10% buffered formalin, and stained with 0.5% crystal violet. Cell growth was quantitated by measurement of the total colony areas for each dish by image analysis.
Analysis of TGFe Activity in Human Milk
Human milk pooled from ten donors was delipidated by three successive extractions with an equal volume of diethyl ether at 4℃ followed by centrifugation at 12,000g for 20 min. The supernatant was acidified with TFA (0.5% final concentra- tion) and centrifuged at 12,000g for 20 min before being filtered through a 0.45 pm membrane and concentrated five-fold by using an Amicon YM5 membrane. The concentrate was dialyzed against 0.05% TFA, lyophilized, reconstituted at 10 x the original concentration, and bioassayed on SW13 cells.
Characterization of SW13-Derived TGFe
The affinity of TGFe activity for heparin was determined by passage of crude conditioned medium reconstituted in 50 mM NaH2PO4, 0.25 M NaCl pH 7.0 through a 0.5 × 3 cm column packed with 1 gm of heparin-Sepharose conjugate (Pharmacia). Samples of the starting material and the column effluent and material eluted with 2M NaCl, 50 mM NaH2PO4 were bioassayed on SW13 cells. As a control, 50 ng/ml bFGF in the same buffer was passed through the same column and samples were bioassayed by using FBHE cells. Conditioned medium was bioassayed against FBHE cells and on NIH-3T3 cells by using the thymidine uptake assay [8] with platelet-derived growth factor as positive control.
The sensitivity of h-TGFe activity to denaturing agents was tested by incubating crude conditioned medium concentrate with either 2% sodium dodecyl sulfate, 0.065 M dithiothreitol, or 8M urea for 1 h. The treated samples along with an untreated control were exhaustively dialyzed against 0.05% TFA to remove the denaturing agent and bioassayed on SW13 cells.
The molecular weight of h-TGFe was estimated by gel filtration chromatography of conditioned medium concentrate on a BioGel P60 column in the presence of 8M urea. Lyophilized concentrate was dissolved in 20 ml 8M urea in 0.05% TFA and applied to a 5 x 86 cm column (Pharmacia) containing BioGel P60 resin (BioRad). The flow rate was 36 ml/h, 20 ml fractions were collected, and 100 ul aliquots were dialyzed against 0.05% TFA to remove urea and bioassayed. Molecular weight standards were ovalbumin (43k), chymotrypsinogen (25k), and ribonuclease A (13.7k).
Purification of TGFe
TGFe was purified from SW13-conditioned medium by preparative reverse- phase, size exclusion, and cation exchange HPLC. A 25 L batch of acidified SW13- conditioned medium was loaded at 75 ml/min onto a 5 × 30 cm preparative column containing Vydac 218TP C18 packing, particle size 15-20 um, by using a Waters Delta-Prep system. The column was washed with 0.05% TFA until the absorbance at 214 nm reached baseline and a gradient of 0.05% TFA in acetonitrile was run at 0-30% over 50 min, 30-50% over 20 min, and 50-100% over 10 min at a flow rate of 70 ml/min. Fractions were collected every 2.5 min, acetonitrile was removed by vacuum evaporation, and aliquots were bioassayed on SW13 cells.
Active fractions from the reverse-phase step were concentrated 40x by using Amicon YM5 membranes and concentrates were loaded onto two series connected TSK 3000 SW size exclusion columns (each 2.5 x 60 cm). Material was eluted from the columns with 0.05% TFA at a flow rate of 2 ml/min and 4 ml fractions were collected.
Sodium phosphate was added to the active material from the size exclusion step to a final concentration of 20 mM. The pH was adjusted to 6.0 with sodium hydroxide and the samples were centrifuged at 90,000g for 30 min. Supernatants were loaded onto a CM2SW cation exchange column (Beckman) and washed with 20 mM phosphate buffer pH 6.0. A linear gradient of 0-0.6M sodium chloride in phosphate buffer pH 6.0 was run at 0-95% over 90 min and 2 ml fractions were collected and bioassayed.
An active sample of TGFe from the cation exchange step was analyzed by reverse-phase HPLC. The sample was acidified to pH 2.5 with 1% TFA and loaded onto a 0.46 x 30 cm column containing Vydac C18 218TP 10 um packing. A gradient of 0-50% 0.05% TFA in acetonitrile was run at 0.25%/min and 2 ml fractions were collected and bioassayed after removal of the acetonitrile by vacuum evaporation. Aliquots of fractions 30-40 were run on a 15% polyacrylamide gel in the presence of 2-mercaptoethanol and examined by silver staining (BioRad). Recovery of activity during purification was monitored by constructing dose-response curves using the purified material. One unit of activity was defined as the dose of test sample needed to give a response equal to an ED50 of a standard preparation of concentrated SW13- conditioned medium run in the same assay. Protein concentration was measured by the BCA method (Pierce), or by electronic integration of absorbance peaks at 214 nm by using ovalbumin as a standard.
RESULTS
Presence of TGFe Activity in Human Tissues, Human Milk, and Cell Supernatants
TGFe-like activity was detected in human milk and SW13, SW403, LOVO, NRK-52E, WIDR, and HCT-15 cells released activity into their respective media (Table I) Smaller but detectable amounts were produced by SKC01 and SW48 cells, while DLD-1, HT-29, SW948, COLO 320 (DM or HSR), CACO 2, and SW742 failed to release observable amounts of TGFe. Activity was also detected in unacidified conditioned media from SW13 and LOVO cultures, suggesting that acid activation is not essential for the release of TGFe activity in contrast to TGFB [9].
| Cell line | Colonies > 20 µm | Cell line | Colonies > 20 um |
|---|---|---|---|
| SW48 | 95 | NRK-52E | 598 |
| SKC01 | 70 | HCT15 | 280 |
| LOVO | 840 | WIDR | 360 |
| CACO2 | 20 | SW403 | 650 |
| HT-29 | 18 | COLO320 HSR | 20 |
| COLO320 DM | 12 | SW742 | 35 |
| SW948 | 15 | Controlª | 15 |
| DLD-1 | 18 | SW13 | 370 |
*Conditioned media were collected from each cell line as described (Materials and Methods), dialyzed against 0.05% TFA, reconstituted at 10x original concentration, and assayed at 100 ul/dish on SW13 cells in soft agar. Human milk extract gave 392 colonies in the same assay.
“Control was medium incubated in the absence of cells. Background colony formation in this assay ranged from 0 to 20 colonies (mean of 20 observations = 6 colonies).
Discrimination of SW13-Derived h-TGFe From Fibroblast Growth Factor
When SW13-conditioned medium was concentrated 50-fold by ultrafiltration and dialyzed against 0.05% TFA, the resulting material produced a dose-related increase in colony formation by SW13 cells in the soft agar assay (Fig. 1). The ED50 of this crude concentrate was 29 µg protein/ml, giving a specific activity of 34.5 half- maximal units per mg protein. Since basic fibroblast growth factor (bFGF) is also known to stimulate colony formation in soft agar by SW13 cells [4,5], we examined the heparin affinity of h-TGFe and its ability to stimulate growth of endothelial cells and 3T3 fibroblasts, both of which respond to bFGF.
A sample of SW13-conditioned medium concentrate was passed through a heparin-Sepharose column. As shown in Figure 2A, most of the activity was recovered in the column breakthrough fraction, indicating that it had little affinity for heparin. In contrast, bFGF applied to the same column was completely retained, as expected from the known affinity of FGF-like growth factors for heparin [9]. The biological activity of h-TGFe could also be separated from that of bFGF. When an aliquot of h-TGFe from the reverse-phase purification step (see below) was assayed on FBHE cells, no stimulation of proliferation was observed (Fig. 2B), in contrast to an eight- to nine-fold increase in cell growth produced by bFGF. When bFGF was assayed on FBHE cells in the presence of TGFe this proliferative response was not diminished, indicating that the absence of effect of h-TGFe alone was not due to contamination of the samples with inhibitors or toxins. Furthermore, partially purified h-TGFe did not stimulate thymidine uptake in quiescent 3T3 cells in contrast to bFGF (not shown).
Effects of Denaturing Agents on h-TGFe Activity
Crude conditioned medium was treated with 0.065M dithiothreitol, 2% sodium dodecyl sulfate, or 8M urea. After removal of the denaturing agent by dialysis, bioassay
1400
1
1200
Colonies > 60um
ED50=29 µg/ml
Log(R/Rmax-R)
1000
800
0
600
400
200
-1
0.60
1.10
1.60
2.10
Log( ug/ml protein)
400
Colonies > 60 um
300
200
100
A
0
CONTROL
SM
BT
WASH
200
Total colony area (mm2)
150
100
50
B
0
CONTROL
bFGF
TGFe
TGFe+ bFGF
indicated that the activity was stable to 8M urea but labile to dithiothreitol or sodium dodecyl sulfate (Table II). In addition, the activity was labile to quaternary ammonium salts such as tetrabutylammonium phosphate (not shown).
Purification of h-TGFe
A 25 L batch of SW13-conditioned medium was loaded onto a preparative C18 column and eluted with an acetonitrile gradient. Two peaks of activity were observed
| Dose (pl/dish) | Controla | 2% NaDodSO,b | 0.065M DTT | 8M urea |
|---|---|---|---|---|
| 50 | 18 | 5 | 4 | 16 |
| 100 | 71 | 5 | 4 | 39 |
| 200 | 150 | 3 | 21 | 166 |
*SW13 conditioned medium was concentrated 10-fold and treated with the indicated concentrations of denaturing agents for 1 h at room temperature. Samples were exhaustively dialyzed against 0.05% TFA and assayed at the indicated dose. Figures are the number of colonies > 40 um in diameter. ªUntreated conditioned medium concentrate.
“Detergent was removed by dialysis against 8M urea followed by 0.05% TFA.
in the resulting fractions: Peak I eluted at 25-30% acetonitrile and Peak II at 35-40% (Fig. 3). Fractions 17-20 (Peak I) were pooled, concentrated, and applied to a size exclusion column. Activity eluted at an apparent molecular weight of 59k (Fig. 4). Further purification was accomplished by HPLC on a cation exchange column. Activity eluted at 0.3M NaCl (Fig. 5) and was subsequently run on an analytical reverse-phase column. TGFe eluted as a single activity peak at approximately 15% acetonitrile (Fig. 6). This peak was well separated from the majority of contaminating protein as judged by the UV absorbance profile, and analysis by gel electrophoresis and silver staining (Fig. 7).
Fractionation of the Peak II material (fractions 23-26 in Fig. 3) on a Bio Gel P60 column in the presence of 8M urea yielded a peak of activity at an apparent molecular weight of 59k as seen with the Peak I material.
The purification statistics (Table III) show excellent recovery of activity from SW13-conditioned medium by preparative reverse phase HPLC. Over 70% of the total activity was distributed in roughly equal amounts between Peaks I and II. Recovery was 45% of total Peak I activity after the size exclusion step and 29% after cation exchange HPLC. This latter step resulted in greater than 100-fold purification of the material from the size exclusion column. Insufficient material was present after the analytical reverse phase step to determine recovery or degree of purification at this stage.
DISCUSSION
Previous studies by Halper and Moses [5] showed that SW13 cells secrete an autostimulatory growth factor and that similar activity was present in extracts of human colon carcinoma tissue. Our findings confirm the production of active material by SW13 cells and demonstrate similar activity in media conditioned by six of 13 colon carcinoma cell lines, a rat kidney epithelial cell line, and in human milk. We have also detected activity in extracts of human colon carcinoma tissue (not shown). Thus production of TGFe-like activity is not restricted to the SW13 cell line. These findings suggest an association of TGFe activity with epithelial cells and tissues, but we do not yet know whether production of, or response to, TGFe is epithelial-specific.
Bovine TGFe has been distinguished from bFGF by its biochemical and biologi- cal properties and by its inability to bind to bFGF receptors [4]. In the present study, human TGFe did not bind to heparin and was ineffective in promoting growth of endothelial cells or 3T3 fibroblasts in contrast to bFGF which has a high affinity for heparin [10] and readily stimulates proliferation of these cell types. Biochemical
800
100
600
80
Colonies > 40 um
acetonitrile
60
400
40
200
0º
20
0
0
0
10
20
30
Fraction No.
800
VOID
43k
25k
13.7k
600
Colonies > 60 um
400
200
0
10
20
30
Fraction No.
800
0.500
600
0.375
Colonies > 60 µ m
0.35M NaCl gradient
Absorbance (214nm)
400
0.250
200
0.125
0.2
8
0.000
10
20
30
40
Fraction No.
800
0.120
Colonies > 60 µ m
600
acetonitrile
Absorbance (214nm)
0.090
400
0.060
30%
200
0.030
00
5
0.000
10
15
20
25
30
35
40
45
50
Fraction No.
30 3132 33 34 35 36 37 38 3940
4 3
25.7
18.4
14.3
6.2
| Sample | Volume (mL) | Activity (U/mL) | Total activity | mg/ml protein | SP act. (U/mg) | Fold purification | % recovery |
|---|---|---|---|---|---|---|---|
| Orig CM | 25,000 | 1.4 | 34,500 | 0.04 | 34.5 | 100 | |
| Peak I | 500 | 22.2 | 11,100 | 0.19 | 116.8 | 3.4 | 32.2 |
| Peak II | 400 | 33.2 | 13,280 | 0.71 | 46.8 | 1.4 | 38.5 |
| TSK 3000ª | 205 | 24.2 | 43,966 | 0.14 | 172.8 | 5.0 | 14.4 |
| CM2SWª | 60 | 54.0 | 3,244 | 0.003 | 18,018 | 522 | 9.4 |
2Peak I material was used for the subsequent purification steps.
characterization of h-TGFe indicates that it is stable to low pH and 8M urea but labile to sodium dodecylsulfate and reducing agents. The molecular mass estimated by gel permeation chromatography under denaturing conditions is 59 kD. This figure differs from the values reported by Halper and Moses both for bovine TGFe [4] and for the activity in acid ethanol extracts of human tissues [5]. In our hands, the bovine form of TGFe, when run under similar conditions, elutes from a BioGel P60 column as a broad peak close to the 25k marker [11]. These findings suggest that h-TGFe differs in size from its bovine counterpart. As expected from the lability of h-TGFe to sodium dodecyl sulfate, activity could not be recovered from slices of polyacrylamide gels, under conditions where bovine TGFe remained active, as seen previously [4]. How- ever, this lability of h-TGFe further suggests that differences exist between bovine and human TGFe’s.
We have undertaken preliminary purification of h-TGFe from 25 L of SW13- conditioned medium by using reverse phase, size exclusion, and cation exchange HPLC. Although two peaks of activity were recovered from the preparative C18 column, it is likely that the second (more hydrophobic) peak is an artifact of the HPLC process, since upon treatment with 8M urea and fractionation on a BioGel P60 column, it co-elutes with the Peak I activity. This result suggests that the Peak II material is an aggregate of h-TGFe with unidentified proteins. A similar phenomenon was reported for bovine TGFe [4]. The behavior of h-TGFe on a carboxymethyl cation exchange column indicated that h-TGFe is a highly basic protein. These results suggest that cation exchange and reverse phase methods will be useful in the purification of h-TGFe. Although analysis of reverse phase HPLC fractions by gel electrophoresis and silver staining showed good separation of activity from major contaminating proteins, it did not permit identification of a specific band that could be correlated with SW13 colony formation. In our hands, bovine TGFe does not readily stain with silver reagents and it is possible that h-TGFe may share this property.
In summary, we have characterized and partially purified h-TGFe from SW13 cells. This material differs from bovine kidney-derived TGFe both in behavior on size exclusion columns under denaturing conditions and stability towards ionic detergents. Further studies will be necessary to elucidate the structural basis of these differences.
ACKNOWLEDGMENT
We thank Jo Anne Mackey for excellent secretarial assistance.
REFERENCES
1. Derynck R: J Cellular Biochem 32:293-304, 1986.
2. Sporn MB, Roberts AB, Wakefield LM, deCrombrugghe B: J Cell Biol 105:1039-1045, 1987.
3. Gospodarowicz D, Ferrara N, Schweigerer L, Neufeld G: Endocrinol Rev 8:95-114, 1987.
4. Halper J, Moses HL: Cancer Res 47:4552-4559, 1987.
5. Halper J, Moses HL: Cancer Res 43:1972-1979, 1983.
6. Parnell PG, Wunderlich J, Carter B, Halper J: J Cell Biochem 42:111-116, 1990.
7. Gospodarowicz D: J Biol Chem 250:2515-2520, 1975.
8. Bowen-Pope DF, Ross R: J Biol Chem 257:5161-5171, 1982.
9. Pircher R, Jullien P, Lawrence DA: Biochem Biophys Res Commun 136:30-37, 1986.
10. Lobb R, Sasse J, Sullivan R, Shing Y, D’Amore P, Jacobs J, Klagsbrun M: J Biol Chem 261:1924-1928, 1986.
11. Dunnington DJ, Pinsky S, Mattes D, Prichett W, Earl C, Greig R, Anzano M: Anal Biochem 174:257-264, 1988.