- Email: [email protected]
The effect of dihydrotestosterone and culture conditions on proliferation of the human prostatic cancer cell line LNCaP
The effect of dihvdrotestosterone and ctiltme conditionii on proliferation of the human prostatic cancer cell line LNCaP Hui-Zhu Chen,* Alexander Kirschenbaum,?’ John Mandeli,+ and Vincent P. Hollander*9 Departments of *Neoplastic Diseases, TUrology, SBiomathematical Sciences, and &Surgery, Mount Sinai Medical Center, New York, New York, USA
Cell density, nutritional state, and serum factors modify the growth response of LNCaP human prostatic cancer cells to dihydrotestosterone. Evaluation of growth response to dihydrotestosterone requires logarithmic transformation of cell count or thymidine incorporation data. Under conditions of dose response, growth increases with cell density but no significant interaction of dihydrotestosterone with cell density was found under optimal culture conditions. Thefrequency of media change was a sign$cant factor in modulating dose response. When cells from cultures maintained at dtzerent feeding periods were plated at diflerent cell densities of (trypan blue) viable cells, significant effects of plating density on dihydrotestosterone response were found. Dihydrotestosterone protects cells under the adverse effects of media deprivation. Under the extreme adverse effects of serum deprivation, cells respond to dihydrotestosterone even under conditions of increasing cell loss. The effects of dihydrotestosterone on final cell density were significant. In the absence of serum, the elongated cells of LNCaP assume a round shape, but many remain adherent to the culture dish and can be restored to normal morphology by serum. A number of growth factors fail to restore normal morphology that was completely restored by a combination offbronectin and dihydrotestosterone. We have not developed a practicable serumfree system for LNCaP. (Steroids sI:269-275, 1992)
Keywordw Serum-free cell growth; fibronectin; dihydrotestosterone, effect on LNCaP; LNCaP, effect of DHT; prostate, cancer cell line LNCaP; steroids
Introduction
Experimental
The human prostatic cell line LNCaP developed by Horoszewicz et al.’ was of great value in studying the response of hormone-dependent tumor. The study of androgen dependence was helped by the absence of steroid receptors other than androgen.‘**To study androgen-induced growth, other factors that regulate this response must be understood. We have studied the effect of cell density, starvation, and serum factors on morphology and the response to dihydrotestosterone (DHT).
RPMI-1640 was purchased from Whittaker Bioproducts Inc. CPSR (Sigma) was used in place of fetal bovine serum (FBS). CPSR was a commercially processed serum absorbed by dextran charcoal to remove steroid. The testosterone and e&radio1 content were at the limit of detection by radioimmunoassay. Dihydrotestosterone and trypsin-EDTA were from Sigma. All other chemicals were of reagent grade. LNCaP cells (a gift from Dr. J. S. Horoszewicz) were maintained in RPM1 with 5% FBS, 0.03% @namine, penicillin 100 U/ml, streptomycin 0.1 me/ml, and fungizone 12.5 U/ml. The cells were plated in flasks (75 cm? and, after a stated interval, the cells were washed in Hanks’, treated for 5 minutes with 0.05% trypsin/0.2% EDTA, pH 7.4, concentrated if necessary by gentle centrifugation, and then wunted in a hemocytometer chamber using trypan blue exclusion for a determination of cell viability. The cells were then plated in microwells as described for particular experiments. When the effect of DHT was to be
Address reprintrequests to Dr. Vincent P. Hollander at the Department of Neoplastic Diseases, MountSinai MedicalCenter, Box 1259, New York, NY 10029-6574,USA. Received July 25, 1991;accepted January 3, 1992.
Q 1992 Butterworth-Heinemann
Steroids,
1992, vol.
57, June
269
Papers studied it was added daily since the effect was somewhat greater than with a single addition of steroid to a final concentration of lo-’ M. The latter concentration was selected because it consistently gave maximal stimulation. The serum-free medium of Connolly and Rose for LNCaP was prepared as described) from RPMI, insulin, transferrin, selenous acid, linoleic acid, and bovine serum albumin.
Thymidine incorporation incubated cells
assay on
[3H]Thymidine (0.3 &i) was added to each well; after 3 hours, media were removed from the wells and the cells were trypsinized, harvested with a Rockefeller multichannel aspirator, and washed with distilled water and methanol. The filters were counted in 2.5 ml aqueous counting scintillant. A log transform on CPM from the thymidine incorporation test greatly improved the ability to calculate the significance of effects by ANOVA. The variances of the DHT-treated groups were significantly different from each other and the control by Bartlett’s test, so that analysis of variance would not be appropriate. A log transform caused the variances between groups to be similar by Bartlett’s test, so that planned comparisons can be made and significance of differences between means assessed by planned comparisons.
DHT CONCENTRATION
(MI
12
IO
Results Figure 1 illustrates the effect of log transformation on the validity of interpretation of the growth data. The cells were plated in 5% CPSR in RPM1 at three densities
(5 x 103, 5 x 104, and 5 x IO4 cells/well) in a 96-well plate. Each day, DHT was added as described above to bring the wells to five concentrations of DHT (0, lo-‘*, lo-lo, lo-*, and 10e6 M). Triplicate wells were harvested at 72 and 96 hours, and thymidine incorporation was measured. Figure 1A shows the data for 72 hours plotted as CPM. The expected stimulation by DHT and inhibition at 10e6 M (refs 1 and 2) was seen at the two lower densities. The error bars reveal the extreme heterogeneity of variances. Indeed, the Bartlett test was statistically significant, indicating that analysis of variance was inappropriate. The plot suggested that the greater the plating density, the greater the magnitude of the hormone response. Figure IB shows the same data after logarithmic transformation. The variances were much more homogeneous and the Bartlett test was not significant, indicating that analysis of variance on the transformed data would be meaningful. It can be seen that the response to DHT was now very similar at the three densities. Before transformation, an inappropriate analysis of variance of density with DHT interaction was highly significant (P = 0.0001); after transformation, this became 0.037. This marginal interaction was explained by the failure of low6 M DHT to inhibit the hormone effect at the higher density. The effect of time was highly significant (P = 0.003), with approximately 20% higher incorporation at 96 hours. The interaction of time with density was not significant. The examination of DHT stimulation by contrast showed that, without exception, only 10e8 M DHT gave a significantly higher thymidine incorporation than the control wells. The 1O-6 M DHT at the 5 x 104 cell density was significantly higher than the 270
Steroids,
1992, vol. 57, June
;8 z -I 6 a4t
_a***------
-‘-II ’ 0 10-Q
-+...----
*p--__._
I
I
IO_‘0
10-e
DHT CONCENTRATION
“-p
I
lo(M
1
Figure 1 Thymidine uptake as a function of DHT concentration. Cells were plated in 96-well microwells in 0.20 ml media at the five concentrations of DHT shown at 1,2, and 5 x IO4 cells/well. Dihydrotestosterone (10 ~1) was added daily to bring the cultures to the indicated concentration of DHT as if all steroid were metabolized. The cultures were incubated for 72 hours, pulsed with 3Hthymidine as described, harvested, and the filters counted. The three densities were plotted with increasing ordinate values: 1 x 104cells/well (open circle), 2 x lo4 (X), and 5 x lo4 (square). (A) The data were plotted on a semilog scale without transformation; (B) the same data after log transformation.
control (P = O.OOOl),although at the two lower densities the lO-‘j M DHT response was lower than that at 10e8 M. In a similar experiment comparing cell density at 1, 2, 4, and 10 x 10“ cells/well, at 0, 10e9, and IO-* M DHT, only 10V8 M DHT produced significant stimulation but without significant interaction between density and DHT (data not shown). Another experiment (data not shown) was also done with a single addition of DHT with harvesting at 72 hours. The response to daily dosing was greater overall (P = 0.0001) and in every individual contrast (P <
Effect of dihydrotestosterone on LNCaP: Chen et al.
0.05) than with the single addition of DHT. A similar difference was observed when total cells were counted after 72 hours. Subsequent experiments were all done with daily hormone treatment. We conclude that the stimulation of LNCaP cells by DHT in the presence of serum was a function of hormone concentration and was dependent on population density in the density range tested. Figure 2 contrasts thymidine incorporation with cell count as a function of cell density and DHT concentration. LNCaP cells were plated at 1,2, and 5 x 10’cells/ well, and triplicate wells compared at 0 and lo-* M daily DHT after 72 hours of incubation. The concordance between the measures of growth as affected by DHT and plating density was satisfactory. For the log cell count data, DHT stimulation was significant (P = O&06),but there was no interaction between DHT and density (P = 0.889). For the log CPM data, the DHT effect was highly significant (P = O.OOOl), but in this experiment, the DHT by density interaction effect approached significance (P = 0.064). The lines representing cell count and CPM in the absence of DHT crossed between the second and third densities. Since thymidine incorporation was easier and more precise, it was preferred for studies involving multiple samples. In spite of our inability to demonstrate interaction between DHT and density under standard culture conditions, we were troubled by the observation that in occasional experiments plating at low density resulted in no DHT effect, although higher plating densities did respond. Since nut~tional state might be a factor in hormonal response as a function of density, we investigated the time of media and hormone feeding on the DHT effect.
to,000 i
140
-120 E
8000
-100 8 c - 80 $
i E 6000 4 4000
- 60 : - 40-+
2000 0,” -r
- 20 I
2I
3I
4I
5I
0
CELL DENSITY ~xIO-~) Figure 2 Comparison of thymidine in~rporation with cell count. LNCaP cells were plated in triplicate for both cell count and thymidine incorporation at 1, 2, and 6 x lo’ ceils/well at either 0 (solid lines) or lo-* M DHT (dashed lines), which was added daily. Cell counts are indicated by open squares and solid circles. Thymidine incorporation is indicated by open circles and solid squares. The error bars were omitted for the higher density since they superimpose each other.
DAYS CULTURE AGE Figure 3 The effect of the frequency of media change on DHT stimuiation. Cultures in RPM1with 5% stripped serum were maintained for 4,8,12, and 18 days without media change of addition of DHT. The cells were trypainized and ptated at either 2 (Dtl or 5 (021 x $0” trypan blue viable cells in fresh 5% CPSR in RPM1 for 72 hours in the absence of hormone. The cultures were then incubated for 72 hours with daily addition of either media or DHT to a level of lo-* M. pulsed with PHlthvmidine. harvested. and the incorporation measured. The-data were plotted in the.bars as the DHT effect: the difference in CPM between the incorporation in the presence and absence of DHT at the two densities. The line graphs display the ratio of thymidine incorporation in the presence and absence of DHT at 2 x 104 (open) or 5 x l@ (filled) cells/well.
Figure 3 shows the effect of the period of time LNCaP cells were cultured without change in medium or DHT addition before plating for the experiments similar to those shown above. Cultures maintained without feeding for 4,8, 12, and 18 days were trypsinized and plated at 2 and 5 x 104 trypan blue-viable cells/well in fresh medium and stripped serum. They were then allowed to attach for 72 hours before starting treatment. Maintenance for 12 and 18 days constitutes very adverse conditions; in the 18-day-old cultures, 70% were dead by the trypan blue criterion at the time of plating viable cells. Thymidine incorporation was measured after 72 hours of treatment of the viable plated cells. Cell density, DHT, and age of culture were all si~ific~t (P < ~.~l). Interaction of density with DHT (P = 0.010) and DHT with age (P = 0.0001) were significant facto&No significant interaction of density with age was found. At each age and density studied, DHT significantly increased the thymidine incorpo ration. Thymidine incorporation decreased with the age of cultures. The response was calculated in terms of (trypan blue) viable cells. Table 1 shows the large effects of culture age. At the higher density in the presence of DHT, age did not cause lowered incorporation until after 18 days of starvation. Table 2 shows that even under extreme adverse conditions of feeding, both densities and all four ages of culture responded to DHT. Steroids, 1992, vol. 57, Juno
273
Papers Table 1
Starvation
affects of thymidine
incorporation
Density
DHT
Age (d)
2 x 104
-
4 8 12 18 4 8 12 18 4 8 12 18 4 8 12 18
_ 2 x 104
+ + + + + + + +
5 x 104
5 x 104
Mean CPM 2 SD’ 2,059 2,000 588 273 3,170 5,517 1,557 1,370 5,725 3,899 980 518 9,413 10,024 7.021 4,776
k k 2 k 2 ? 2 k 2 2 k f k f + -t
472 822 64 69 390 1,312 559 725 1,212 309 195 113 1,176 1,298 338 223
Contrast (probability)
4 vs 8 d (0.75) 4 vs 12 d (0.0001) 4 vs 18 d (0.0001) 4 vs 8 d (0.02) 4 vs 12 d (0.0001) 4 vs 18 d (0.0001) 4 vs 8 d (0.10) 4 vs 12 d (0.001) 4 vs 18 d (0.0001) 4 vs 8 d (0.77) 4 vs 12 d (0.19) 4 vs 18 d (0.004)
a Sample size, n = 3.
Table 2
Dihydrotestosterone
response
within
density-age
effects Density
Age (d)
DHT effect (CPMja
Probability
2 x 104
4 8 12 18 4 8 12 18
+1,111 +3,517 + 989 + 1,097 + 3,687 +6,126 + 6,042 + 4,259
0.048 0.0001 0.0001 0.0001 0.025 0.0001 0.0001 0.0001
5 x 104
(P = O.OOOl),and the hormone with density interaction (P = 0.033) on thymidine incorporation were significant in the absence of serum. Even with minimal growth and greatly increased cell death, DHT stimulated both thymidine incorporation and cell count. The increase in thymidine incorporation was significant (P < 0.001) only at the higher density. The apparent but not significant effect at lower density would require a very large sample size to enable demonstration of possible significance. Serum factors
a DHT effect, mean CPM for (DHT+) - (mean CPM for DHT-1, where mean CPM for DHT+ and DHT- are given in Table 1.
Figure 3 shows that the DHT effect (the difference in thymidine incorporation response between the presence and absence of DHT) was maximal at 8 to 12 days of stock growth. The ratio (incorporation with/ incorporation without DHT) rises with age and was greater at the higher density. Frequency of feeding affects the magnitude of the DHT response. The triple interaction of age, density, and DHT (P = 0.042) was significant; the DHT effect was a function of age as well as cell density. The unexpected tenacity of the hormone effect in the face of deprivation led us to study serum factors on the growth of LNCaP. Serum-free
studies
Figure 4 shows that in the absence of serum, LNCaP cells could still respond to DHT. Cells were plated at 2 and 5 x 10e4 cells in !&well plates in the absence of serum. They were incubated for 72 hours and then were treated daily with buffer or DHT to bring them to 10e8 M. At the higher density, there was increased thymidine incorporation and cell count as a result of DHT. The overall effects of DHT (P = O.OOlZ), cell density 272
Steroids,
1992, vol. 57, June
and morphology
Figure 5A shows that LNCaP cells had different morphology in the absence of serum. Figure 5A shows the stellate shape of the normal LNCaP cells in RPM1 with 5% stripped serum. Occasional round cells were present. Transfer to serum-free media produced the morphologic change seen in Figure 5B within 48 hours. The cells that were not dead were round with decreased cytoplasmic staining and resemble those seen after trypsinization except that the cells were fixed to the plastic dish and could not be dislodged by shaking. These cells were not dead since replacement of serum reversed the morphology to the appearance in Figure 5A after 48 hours. A variety of growth substances were evaluated for the ability to produce the long narrow stellate cells (type A) seen in Figure 5A. Insulin (10 pg/ml), PDGF (100 ng/ml), FGF mixed (10 ng/ml), transferrin (10 pg/ml), TGFB (1 pg/ml), selenous acid (6.25 pg/ml), linoleic acid (5.35 pg/ml), and 0.2% bovine serum albumin were without effect. “ITSLA” refers to the mixture of insulin, transferrin, selenous acid, linoleic acid, and bovine serum albumin in the quantities described above.3 Table 3 shows the result of plating LNCaP in RPM1 with 5% stripped serum for 24 hours, washing the cells with RPMI, and incubation of the mixtures described for 3
Effect of d~~ydr~testost~r~ne on ~~~~P: Chen et at.
200DHT
20tDHT
50-DHT
PLATED CELLS/WELL
50+DHT (~10’~)
Pigure 4 LNCaP cetls were plated at 2 and 5 x 104 oellslwell without addition of s8rum to RPM standard media. After 72 hours of incubation, triplicate cultures w8r8 treated drily with either RPM1or DHT in RPM sufficient to bring cultures to XI-* M. After 72 hours, the cells were counted and thymidine incorporation was measured. Bars show cpm thymidine incorporated; the line points show cell counts. Dihydrotestosterone significantly increased the incorporation only at the higher density. Because total cell count actually falls during incubation, no estimate of the significance of the counts was made.
days. The approximate fraction of LNCaP cells in the stellate form was recorded, It should be noted that control cultures of LNCaP in RPM1 and 5% serum always contain a small number of round cells indistinguishable from the round cells shown in Figure 5B in the absence of serum. Table 4 shows the effect of different media on cell growth. LNCaP (5 x lo’ cells) was plated in RPM1 with 5% serum for 24 hours and washed, and the medium was replaced as shown. At 72 hours, the cell count was recorded. ITSLA did not permit growth while serum did. We cannot explain our repeated observation of DHT response in the absence of serum and growth factors. We have not studied the DHT-induced growth of the cells after normal appearance was restored with fibronectin and DHT. The contrast could not be interpreted since the addition of fibronectin alone gives only a 20% restoration of normal stellate cells unless DHT was also added. Some factor other than androgen must be involved since deprivation of DHT does not cause the cells to assume the round shape.
Cell density may influence response to growth factors in culture.“-‘j Kaighn and associates noted that two androgen-independent prostatic cancers, PC-3 and DU-145, responded to some growth factors only at high cell densities’ and growth under serum-free conditions was population dependent. Santen and associates studied the growth of the androgen-responsive tumor R3327G line by a mo~homet~c procedure. Increases of cell density in cultures using 1% castrated rat serum from 400 cell/cm2, which showed no DHT effect, to 4,000 cells/cm2 increased hormone effect to a limited but significant (P < 0.01) degree in 6 days growth.8 It was difficult to interpret such growth curves without transformation of the data. In the studies reported here, densities 50-fold higher were shown to increase the DHT effect on cell number more than two-fold in 3 days. If cells were plated at a density permitting exponential growth, cell number at any time before plateau should be a function of the initial plating density. If Steroids, 1992, vol. 57, June
273
Papers Table 3
Effect of serum on morphology Percentage of type A cells after 3 days
Media RPM1 RPM1 RPM1 RPM1 RPM1
+ + f + +
Table 4
ITSLA ITSLA + DHT ITSLA + fibronectin (10 &g/ml) ITSLA + fibronectin + DHT 0.4% stripped serum
Serum-free
media and dihydrotestosterone
Media RPMI + ITSLA
RPM1 + ITSLA + DHT RPM1 + 5% serum KPSR)
Figure 5 (A) LNCaP cells grown in standard media containing 5% stripped serum. In addition to the normal stellate or elongated forms there were always a few round cells resembling those seen in panel 8. Coverslips were placed in vessels before plating and adherent cells were stained in Wright-Giemsa. (Approximate magnification, x 150.1 (B) l.NCaP cells grown as above, but in the absence of serum for48 hours. The cells were adherent to plastic and cannot be dislodged by gentle shaking. ~Approximate magnification, x 150.1
DHT stimulates some fraction of the plated cells or some fraction of their growth rate, the cell count in the presence of DHT will be a function of the plating density, However, a highly significant effect of density on cell count does not imply that increased density sensitizes the cells to DHT. In the experiments in the presence of serum reported here, almost no significant DHT with density interaction was noted under ordinary culture conditions, implying that an increase in density did not sensitize cells to DHT stimulation. However, experiments using larger sample sizes might reveal some interaction. 274
Steroids, 1992, wt. 57, June
Cells/Well
effect
after 72 hours
4.5 x 104 4.7 x 104 1.2 x IO”
With serum deprivation and retention of the capacity to respond to DHT, or by inadequate feeding, density with DHT interaction was significant; only cells at higher density respond. Dihydrotestosterone appears to protect the cells under adverse conditions. A serum factor required for response may also be produced by the cells under adverse conditions, but no studies of conditioned medium have yet been carried out. Significant deviation from correspondence between cell count and thymidine incorporation has been noted in the growth of FAM breast cancer cells9; indeed, there was no basic reason why the response of these parameters should be identical. In the present work, we confirmed a satisfactory relationship and justified using thymidine inco~oration alone in some experiments, In any study in which hormonal pe~urbation affects the cell growth, cell count should be regarded as the standard. If thymidine incorporation correlates with cell count, the incorporation study will be simpler and more precise. The tenacity of LNCaP cells to retain androgen sensitivity during starvation was su~~sing; our data suggest that the absence of nut~ent may either select or induce cells whose response to DHT was greater than the initial population. The morphologic change in LNCaP on serum withdrawal was not nonspecific; many growth factors other than fibronectin fail to convert the cells to the normal form. The unknown factor required for the stellate cell form was neither DHT nor ~b~onectin; DHT and testosterone were unmeasurable in our stripped serum. Fibronectin was made by breast cancer cells,‘0 was involved in substrate interaction by breast cancer cells, f1and was used in defined media for such cells. I* Fibronectin frequently reverses the rounded shape of transformed cells to that of the wild type.13 Connolly and Rose reported serum-free growth of LNCaP cells using ITSLA.3 In this medium, our
Effect of dihydrotestosterone on LNCaP: Chen et al. LNCaP cells were in the round form in the presence and absence of DHT. LNCaP cells have many variants; we assume that such differences in our line allowed us to demonstrate a role of fibronectin. We have not yet developed a practicable serum-free system for longterm maintenance of LNCaP. Further studies in this direction are in progress. Cell density, DHT, and nutritional factors modify tumor growth in culture. In the patient, the role of androgen has been exploited in therapy. It is likely that tumor cell density as well as autocrine and par-amine growth factors will be exploited as a result of further study. It is difficult to relate heterogenous tumor masses with the control of the growth of a cell line.
References 1. Horoszewicz
2. 3.
5.
6.
7.
8.
Acknowledgment Grateful appreciation is expressed for her invaluable assistance.
4.
to Eileen J. Trager
JS, Leong SS, Kawinski E, Karr JP, Rosenthal H, Ming Chu T, Mirand EA, Murphy GP (1983). The LNCaP model of human prostatic carcinoma. Cancer Res 43:1809-1818. Sonnenschein C, Olea N, Pasanen ME, Soto AM (1989). Negative controls of cell proliferation: human prostate cancer cells and androgens. Cancer Res 49:3474-3481. Connolly JM, Rose DP( 1990). Production of epidermal growth factor and transforming growth factor-alpha by the androgenresponsive LNCaP human prostate cancer cell line. Prostate 16~209-218.
9.
10.
11.
12.
13.
Burke JM (1989). Stimulation of DNA synthesis in human and bovine RPE by peptide growth factors: the response to TNFa and EGF was dependent upon culture density. Curr Eye Res 8~1279-1286. Chen YY, Rabinovitch PS (1998). Mitogen response and cell cycle kinetics of Swiss 3T3 ceils in defined medium: differences from human fibroblasts and effects of cell density. Exp Cell Res 190:145-150. Paulsson Y, Beckmann MP, Westermark B, Heldin CH (1988). Density-dependent inhibition of cell growth by transforming growth factor-beta 1 in normal human fibroblasts. Growth Factors 1:19-27. Kaighn ME, Kirk D, Szalay M, Lechner JF (1981). Growth control of prostatic carcinoma cells in serum-free media: interrelationship of hormone response, cell density, and nutrient media. Proc Nat1 Acad Sci USA 78:56735676. Santen RJ, Mowszowicz I, Portois MC, Mauvais-Jarvis P (1987). Androgen dependence of the Dunning R3327G cell line in monolayer culture. Prostate 11:377-387. Jozan S, Gay G, Marques B, Mirouze A, David JF (1985). Oestradiol was effective in stimulating ‘H-thymidine incorporation but not on proliferation of breast cancer cultured cells. Cell Tissue Kinet l&457-464. Stampfer MR, Vlodavsky I, Smith HS, Ford R, Becker FF, Riggs J (1981). Fibronectin production by human mammary cells. J Nat1 Cancer Inst 67~253-261. Noel A, Calle A, Emonard H, Nusgens B, et al. (1988). Antagonistic effects of laminin and fibronectin in cell-to-cell and cellto-matrix interactions in MCF-7 cultures. In Vitro Cell Dev Biol24:373-380. Calvo F, Brower M, Camey DN (1984). Continuous culture and soft agarose cloning of multiple human breast carcinoma cell lines in serum-free medium. Cancer Res 44~4553-4559. Hynes RO, Yamada KM (1982). Fibronectins: multifunctional modular glycoproteins. J Cell Biol%:369-377.
Steroids,
1992, vol. 57, June
275
Cell density, nutritional state, and serum factors modify the growth response of LNCaP human prostatic cancer cells to dihydrotestosterone. Evaluation of growth response to dihydrotestosterone requires logarithmic transformation of cell count or thymidine incorporation data. Under conditions of dose response, growth increases with cell density but no significant interaction of dihydrotestosterone with cell density was found under optimal culture conditions. Thefrequency of media change was a sign$cant factor in modulating dose response. When cells from cultures maintained at dtzerent feeding periods were plated at diflerent cell densities of (trypan blue) viable cells, significant effects of plating density on dihydrotestosterone response were found. Dihydrotestosterone protects cells under the adverse effects of media deprivation. Under the extreme adverse effects of serum deprivation, cells respond to dihydrotestosterone even under conditions of increasing cell loss. The effects of dihydrotestosterone on final cell density were significant. In the absence of serum, the elongated cells of LNCaP assume a round shape, but many remain adherent to the culture dish and can be restored to normal morphology by serum. A number of growth factors fail to restore normal morphology that was completely restored by a combination offbronectin and dihydrotestosterone. We have not developed a practicable serumfree system for LNCaP. (Steroids sI:269-275, 1992)
Keywordw Serum-free cell growth; fibronectin; dihydrotestosterone, effect on LNCaP; LNCaP, effect of DHT; prostate, cancer cell line LNCaP; steroids
Introduction
Experimental
The human prostatic cell line LNCaP developed by Horoszewicz et al.’ was of great value in studying the response of hormone-dependent tumor. The study of androgen dependence was helped by the absence of steroid receptors other than androgen.‘**To study androgen-induced growth, other factors that regulate this response must be understood. We have studied the effect of cell density, starvation, and serum factors on morphology and the response to dihydrotestosterone (DHT).
RPMI-1640 was purchased from Whittaker Bioproducts Inc. CPSR (Sigma) was used in place of fetal bovine serum (FBS). CPSR was a commercially processed serum absorbed by dextran charcoal to remove steroid. The testosterone and e&radio1 content were at the limit of detection by radioimmunoassay. Dihydrotestosterone and trypsin-EDTA were from Sigma. All other chemicals were of reagent grade. LNCaP cells (a gift from Dr. J. S. Horoszewicz) were maintained in RPM1 with 5% FBS, 0.03% @namine, penicillin 100 U/ml, streptomycin 0.1 me/ml, and fungizone 12.5 U/ml. The cells were plated in flasks (75 cm? and, after a stated interval, the cells were washed in Hanks’, treated for 5 minutes with 0.05% trypsin/0.2% EDTA, pH 7.4, concentrated if necessary by gentle centrifugation, and then wunted in a hemocytometer chamber using trypan blue exclusion for a determination of cell viability. The cells were then plated in microwells as described for particular experiments. When the effect of DHT was to be
Address reprintrequests to Dr. Vincent P. Hollander at the Department of Neoplastic Diseases, MountSinai MedicalCenter, Box 1259, New York, NY 10029-6574,USA. Received July 25, 1991;accepted January 3, 1992.
Q 1992 Butterworth-Heinemann
Steroids,
1992, vol.
57, June
269
Papers studied it was added daily since the effect was somewhat greater than with a single addition of steroid to a final concentration of lo-’ M. The latter concentration was selected because it consistently gave maximal stimulation. The serum-free medium of Connolly and Rose for LNCaP was prepared as described) from RPMI, insulin, transferrin, selenous acid, linoleic acid, and bovine serum albumin.
Thymidine incorporation incubated cells
assay on
[3H]Thymidine (0.3 &i) was added to each well; after 3 hours, media were removed from the wells and the cells were trypsinized, harvested with a Rockefeller multichannel aspirator, and washed with distilled water and methanol. The filters were counted in 2.5 ml aqueous counting scintillant. A log transform on CPM from the thymidine incorporation test greatly improved the ability to calculate the significance of effects by ANOVA. The variances of the DHT-treated groups were significantly different from each other and the control by Bartlett’s test, so that analysis of variance would not be appropriate. A log transform caused the variances between groups to be similar by Bartlett’s test, so that planned comparisons can be made and significance of differences between means assessed by planned comparisons.
DHT CONCENTRATION
(MI
12
IO
Results Figure 1 illustrates the effect of log transformation on the validity of interpretation of the growth data. The cells were plated in 5% CPSR in RPM1 at three densities
(5 x 103, 5 x 104, and 5 x IO4 cells/well) in a 96-well plate. Each day, DHT was added as described above to bring the wells to five concentrations of DHT (0, lo-‘*, lo-lo, lo-*, and 10e6 M). Triplicate wells were harvested at 72 and 96 hours, and thymidine incorporation was measured. Figure 1A shows the data for 72 hours plotted as CPM. The expected stimulation by DHT and inhibition at 10e6 M (refs 1 and 2) was seen at the two lower densities. The error bars reveal the extreme heterogeneity of variances. Indeed, the Bartlett test was statistically significant, indicating that analysis of variance was inappropriate. The plot suggested that the greater the plating density, the greater the magnitude of the hormone response. Figure IB shows the same data after logarithmic transformation. The variances were much more homogeneous and the Bartlett test was not significant, indicating that analysis of variance on the transformed data would be meaningful. It can be seen that the response to DHT was now very similar at the three densities. Before transformation, an inappropriate analysis of variance of density with DHT interaction was highly significant (P = 0.0001); after transformation, this became 0.037. This marginal interaction was explained by the failure of low6 M DHT to inhibit the hormone effect at the higher density. The effect of time was highly significant (P = 0.003), with approximately 20% higher incorporation at 96 hours. The interaction of time with density was not significant. The examination of DHT stimulation by contrast showed that, without exception, only 10e8 M DHT gave a significantly higher thymidine incorporation than the control wells. The 1O-6 M DHT at the 5 x 104 cell density was significantly higher than the 270
Steroids,
1992, vol. 57, June
;8 z -I 6 a4t
_a***------
-‘-II ’ 0 10-Q
-+...----
*p--__._
I
I
IO_‘0
10-e
DHT CONCENTRATION
“-p
I
lo(M
1
Figure 1 Thymidine uptake as a function of DHT concentration. Cells were plated in 96-well microwells in 0.20 ml media at the five concentrations of DHT shown at 1,2, and 5 x IO4 cells/well. Dihydrotestosterone (10 ~1) was added daily to bring the cultures to the indicated concentration of DHT as if all steroid were metabolized. The cultures were incubated for 72 hours, pulsed with 3Hthymidine as described, harvested, and the filters counted. The three densities were plotted with increasing ordinate values: 1 x 104cells/well (open circle), 2 x lo4 (X), and 5 x lo4 (square). (A) The data were plotted on a semilog scale without transformation; (B) the same data after log transformation.
control (P = O.OOOl),although at the two lower densities the lO-‘j M DHT response was lower than that at 10e8 M. In a similar experiment comparing cell density at 1, 2, 4, and 10 x 10“ cells/well, at 0, 10e9, and IO-* M DHT, only 10V8 M DHT produced significant stimulation but without significant interaction between density and DHT (data not shown). Another experiment (data not shown) was also done with a single addition of DHT with harvesting at 72 hours. The response to daily dosing was greater overall (P = 0.0001) and in every individual contrast (P <
Effect of dihydrotestosterone on LNCaP: Chen et al.
0.05) than with the single addition of DHT. A similar difference was observed when total cells were counted after 72 hours. Subsequent experiments were all done with daily hormone treatment. We conclude that the stimulation of LNCaP cells by DHT in the presence of serum was a function of hormone concentration and was dependent on population density in the density range tested. Figure 2 contrasts thymidine incorporation with cell count as a function of cell density and DHT concentration. LNCaP cells were plated at 1,2, and 5 x 10’cells/ well, and triplicate wells compared at 0 and lo-* M daily DHT after 72 hours of incubation. The concordance between the measures of growth as affected by DHT and plating density was satisfactory. For the log cell count data, DHT stimulation was significant (P = O&06),but there was no interaction between DHT and density (P = 0.889). For the log CPM data, the DHT effect was highly significant (P = O.OOOl), but in this experiment, the DHT by density interaction effect approached significance (P = 0.064). The lines representing cell count and CPM in the absence of DHT crossed between the second and third densities. Since thymidine incorporation was easier and more precise, it was preferred for studies involving multiple samples. In spite of our inability to demonstrate interaction between DHT and density under standard culture conditions, we were troubled by the observation that in occasional experiments plating at low density resulted in no DHT effect, although higher plating densities did respond. Since nut~tional state might be a factor in hormonal response as a function of density, we investigated the time of media and hormone feeding on the DHT effect.
to,000 i
140
-120 E
8000
-100 8 c - 80 $
i E 6000 4 4000
- 60 : - 40-+
2000 0,” -r
- 20 I
2I
3I
4I
5I
0
CELL DENSITY ~xIO-~) Figure 2 Comparison of thymidine in~rporation with cell count. LNCaP cells were plated in triplicate for both cell count and thymidine incorporation at 1, 2, and 6 x lo’ ceils/well at either 0 (solid lines) or lo-* M DHT (dashed lines), which was added daily. Cell counts are indicated by open squares and solid circles. Thymidine incorporation is indicated by open circles and solid squares. The error bars were omitted for the higher density since they superimpose each other.
DAYS CULTURE AGE Figure 3 The effect of the frequency of media change on DHT stimuiation. Cultures in RPM1with 5% stripped serum were maintained for 4,8,12, and 18 days without media change of addition of DHT. The cells were trypainized and ptated at either 2 (Dtl or 5 (021 x $0” trypan blue viable cells in fresh 5% CPSR in RPM1 for 72 hours in the absence of hormone. The cultures were then incubated for 72 hours with daily addition of either media or DHT to a level of lo-* M. pulsed with PHlthvmidine. harvested. and the incorporation measured. The-data were plotted in the.bars as the DHT effect: the difference in CPM between the incorporation in the presence and absence of DHT at the two densities. The line graphs display the ratio of thymidine incorporation in the presence and absence of DHT at 2 x 104 (open) or 5 x l@ (filled) cells/well.
Figure 3 shows the effect of the period of time LNCaP cells were cultured without change in medium or DHT addition before plating for the experiments similar to those shown above. Cultures maintained without feeding for 4,8, 12, and 18 days were trypsinized and plated at 2 and 5 x 104 trypan blue-viable cells/well in fresh medium and stripped serum. They were then allowed to attach for 72 hours before starting treatment. Maintenance for 12 and 18 days constitutes very adverse conditions; in the 18-day-old cultures, 70% were dead by the trypan blue criterion at the time of plating viable cells. Thymidine incorporation was measured after 72 hours of treatment of the viable plated cells. Cell density, DHT, and age of culture were all si~ific~t (P < ~.~l). Interaction of density with DHT (P = 0.010) and DHT with age (P = 0.0001) were significant facto&No significant interaction of density with age was found. At each age and density studied, DHT significantly increased the thymidine incorpo ration. Thymidine incorporation decreased with the age of cultures. The response was calculated in terms of (trypan blue) viable cells. Table 1 shows the large effects of culture age. At the higher density in the presence of DHT, age did not cause lowered incorporation until after 18 days of starvation. Table 2 shows that even under extreme adverse conditions of feeding, both densities and all four ages of culture responded to DHT. Steroids, 1992, vol. 57, Juno
273
Papers Table 1
Starvation
affects of thymidine
incorporation
Density
DHT
Age (d)
2 x 104
-
4 8 12 18 4 8 12 18 4 8 12 18 4 8 12 18
_ 2 x 104
+ + + + + + + +
5 x 104
5 x 104
Mean CPM 2 SD’ 2,059 2,000 588 273 3,170 5,517 1,557 1,370 5,725 3,899 980 518 9,413 10,024 7.021 4,776
k k 2 k 2 ? 2 k 2 2 k f k f + -t
472 822 64 69 390 1,312 559 725 1,212 309 195 113 1,176 1,298 338 223
Contrast (probability)
4 vs 8 d (0.75) 4 vs 12 d (0.0001) 4 vs 18 d (0.0001) 4 vs 8 d (0.02) 4 vs 12 d (0.0001) 4 vs 18 d (0.0001) 4 vs 8 d (0.10) 4 vs 12 d (0.001) 4 vs 18 d (0.0001) 4 vs 8 d (0.77) 4 vs 12 d (0.19) 4 vs 18 d (0.004)
a Sample size, n = 3.
Table 2
Dihydrotestosterone
response
within
density-age
effects Density
Age (d)
DHT effect (CPMja
Probability
2 x 104
4 8 12 18 4 8 12 18
+1,111 +3,517 + 989 + 1,097 + 3,687 +6,126 + 6,042 + 4,259
0.048 0.0001 0.0001 0.0001 0.025 0.0001 0.0001 0.0001
5 x 104
(P = O.OOOl),and the hormone with density interaction (P = 0.033) on thymidine incorporation were significant in the absence of serum. Even with minimal growth and greatly increased cell death, DHT stimulated both thymidine incorporation and cell count. The increase in thymidine incorporation was significant (P < 0.001) only at the higher density. The apparent but not significant effect at lower density would require a very large sample size to enable demonstration of possible significance. Serum factors
a DHT effect, mean CPM for (DHT+) - (mean CPM for DHT-1, where mean CPM for DHT+ and DHT- are given in Table 1.
Figure 3 shows that the DHT effect (the difference in thymidine incorporation response between the presence and absence of DHT) was maximal at 8 to 12 days of stock growth. The ratio (incorporation with/ incorporation without DHT) rises with age and was greater at the higher density. Frequency of feeding affects the magnitude of the DHT response. The triple interaction of age, density, and DHT (P = 0.042) was significant; the DHT effect was a function of age as well as cell density. The unexpected tenacity of the hormone effect in the face of deprivation led us to study serum factors on the growth of LNCaP. Serum-free
studies
Figure 4 shows that in the absence of serum, LNCaP cells could still respond to DHT. Cells were plated at 2 and 5 x 10e4 cells in !&well plates in the absence of serum. They were incubated for 72 hours and then were treated daily with buffer or DHT to bring them to 10e8 M. At the higher density, there was increased thymidine incorporation and cell count as a result of DHT. The overall effects of DHT (P = O.OOlZ), cell density 272
Steroids,
1992, vol. 57, June
and morphology
Figure 5A shows that LNCaP cells had different morphology in the absence of serum. Figure 5A shows the stellate shape of the normal LNCaP cells in RPM1 with 5% stripped serum. Occasional round cells were present. Transfer to serum-free media produced the morphologic change seen in Figure 5B within 48 hours. The cells that were not dead were round with decreased cytoplasmic staining and resemble those seen after trypsinization except that the cells were fixed to the plastic dish and could not be dislodged by shaking. These cells were not dead since replacement of serum reversed the morphology to the appearance in Figure 5A after 48 hours. A variety of growth substances were evaluated for the ability to produce the long narrow stellate cells (type A) seen in Figure 5A. Insulin (10 pg/ml), PDGF (100 ng/ml), FGF mixed (10 ng/ml), transferrin (10 pg/ml), TGFB (1 pg/ml), selenous acid (6.25 pg/ml), linoleic acid (5.35 pg/ml), and 0.2% bovine serum albumin were without effect. “ITSLA” refers to the mixture of insulin, transferrin, selenous acid, linoleic acid, and bovine serum albumin in the quantities described above.3 Table 3 shows the result of plating LNCaP in RPM1 with 5% stripped serum for 24 hours, washing the cells with RPMI, and incubation of the mixtures described for 3
Effect of d~~ydr~testost~r~ne on ~~~~P: Chen et at.
200DHT
20tDHT
50-DHT
PLATED CELLS/WELL
50+DHT (~10’~)
Pigure 4 LNCaP cetls were plated at 2 and 5 x 104 oellslwell without addition of s8rum to RPM standard media. After 72 hours of incubation, triplicate cultures w8r8 treated drily with either RPM1or DHT in RPM sufficient to bring cultures to XI-* M. After 72 hours, the cells were counted and thymidine incorporation was measured. Bars show cpm thymidine incorporated; the line points show cell counts. Dihydrotestosterone significantly increased the incorporation only at the higher density. Because total cell count actually falls during incubation, no estimate of the significance of the counts was made.
days. The approximate fraction of LNCaP cells in the stellate form was recorded, It should be noted that control cultures of LNCaP in RPM1 and 5% serum always contain a small number of round cells indistinguishable from the round cells shown in Figure 5B in the absence of serum. Table 4 shows the effect of different media on cell growth. LNCaP (5 x lo’ cells) was plated in RPM1 with 5% serum for 24 hours and washed, and the medium was replaced as shown. At 72 hours, the cell count was recorded. ITSLA did not permit growth while serum did. We cannot explain our repeated observation of DHT response in the absence of serum and growth factors. We have not studied the DHT-induced growth of the cells after normal appearance was restored with fibronectin and DHT. The contrast could not be interpreted since the addition of fibronectin alone gives only a 20% restoration of normal stellate cells unless DHT was also added. Some factor other than androgen must be involved since deprivation of DHT does not cause the cells to assume the round shape.
Cell density may influence response to growth factors in culture.“-‘j Kaighn and associates noted that two androgen-independent prostatic cancers, PC-3 and DU-145, responded to some growth factors only at high cell densities’ and growth under serum-free conditions was population dependent. Santen and associates studied the growth of the androgen-responsive tumor R3327G line by a mo~homet~c procedure. Increases of cell density in cultures using 1% castrated rat serum from 400 cell/cm2, which showed no DHT effect, to 4,000 cells/cm2 increased hormone effect to a limited but significant (P < 0.01) degree in 6 days growth.8 It was difficult to interpret such growth curves without transformation of the data. In the studies reported here, densities 50-fold higher were shown to increase the DHT effect on cell number more than two-fold in 3 days. If cells were plated at a density permitting exponential growth, cell number at any time before plateau should be a function of the initial plating density. If Steroids, 1992, vol. 57, June
273
Papers Table 3
Effect of serum on morphology Percentage of type A cells after 3 days
Media RPM1 RPM1 RPM1 RPM1 RPM1
+ + f + +
Table 4
ITSLA ITSLA + DHT ITSLA + fibronectin (10 &g/ml) ITSLA + fibronectin + DHT 0.4% stripped serum
Serum-free
media and dihydrotestosterone
Media RPMI + ITSLA
RPM1 + ITSLA + DHT RPM1 + 5% serum KPSR)
Figure 5 (A) LNCaP cells grown in standard media containing 5% stripped serum. In addition to the normal stellate or elongated forms there were always a few round cells resembling those seen in panel 8. Coverslips were placed in vessels before plating and adherent cells were stained in Wright-Giemsa. (Approximate magnification, x 150.1 (B) l.NCaP cells grown as above, but in the absence of serum for48 hours. The cells were adherent to plastic and cannot be dislodged by gentle shaking. ~Approximate magnification, x 150.1
DHT stimulates some fraction of the plated cells or some fraction of their growth rate, the cell count in the presence of DHT will be a function of the plating density, However, a highly significant effect of density on cell count does not imply that increased density sensitizes the cells to DHT. In the experiments in the presence of serum reported here, almost no significant DHT with density interaction was noted under ordinary culture conditions, implying that an increase in density did not sensitize cells to DHT stimulation. However, experiments using larger sample sizes might reveal some interaction. 274
Steroids, 1992, wt. 57, June
Cells/Well
effect
after 72 hours
4.5 x 104 4.7 x 104 1.2 x IO”
With serum deprivation and retention of the capacity to respond to DHT, or by inadequate feeding, density with DHT interaction was significant; only cells at higher density respond. Dihydrotestosterone appears to protect the cells under adverse conditions. A serum factor required for response may also be produced by the cells under adverse conditions, but no studies of conditioned medium have yet been carried out. Significant deviation from correspondence between cell count and thymidine incorporation has been noted in the growth of FAM breast cancer cells9; indeed, there was no basic reason why the response of these parameters should be identical. In the present work, we confirmed a satisfactory relationship and justified using thymidine inco~oration alone in some experiments, In any study in which hormonal pe~urbation affects the cell growth, cell count should be regarded as the standard. If thymidine incorporation correlates with cell count, the incorporation study will be simpler and more precise. The tenacity of LNCaP cells to retain androgen sensitivity during starvation was su~~sing; our data suggest that the absence of nut~ent may either select or induce cells whose response to DHT was greater than the initial population. The morphologic change in LNCaP on serum withdrawal was not nonspecific; many growth factors other than fibronectin fail to convert the cells to the normal form. The unknown factor required for the stellate cell form was neither DHT nor ~b~onectin; DHT and testosterone were unmeasurable in our stripped serum. Fibronectin was made by breast cancer cells,‘0 was involved in substrate interaction by breast cancer cells, f1and was used in defined media for such cells. I* Fibronectin frequently reverses the rounded shape of transformed cells to that of the wild type.13 Connolly and Rose reported serum-free growth of LNCaP cells using ITSLA.3 In this medium, our
Effect of dihydrotestosterone on LNCaP: Chen et al. LNCaP cells were in the round form in the presence and absence of DHT. LNCaP cells have many variants; we assume that such differences in our line allowed us to demonstrate a role of fibronectin. We have not yet developed a practicable serum-free system for longterm maintenance of LNCaP. Further studies in this direction are in progress. Cell density, DHT, and nutritional factors modify tumor growth in culture. In the patient, the role of androgen has been exploited in therapy. It is likely that tumor cell density as well as autocrine and par-amine growth factors will be exploited as a result of further study. It is difficult to relate heterogenous tumor masses with the control of the growth of a cell line.
References 1. Horoszewicz
2. 3.
5.
6.
7.
8.
Acknowledgment Grateful appreciation is expressed for her invaluable assistance.
4.
to Eileen J. Trager
JS, Leong SS, Kawinski E, Karr JP, Rosenthal H, Ming Chu T, Mirand EA, Murphy GP (1983). The LNCaP model of human prostatic carcinoma. Cancer Res 43:1809-1818. Sonnenschein C, Olea N, Pasanen ME, Soto AM (1989). Negative controls of cell proliferation: human prostate cancer cells and androgens. Cancer Res 49:3474-3481. Connolly JM, Rose DP( 1990). Production of epidermal growth factor and transforming growth factor-alpha by the androgenresponsive LNCaP human prostate cancer cell line. Prostate 16~209-218.
9.
10.
11.
12.
13.
Burke JM (1989). Stimulation of DNA synthesis in human and bovine RPE by peptide growth factors: the response to TNFa and EGF was dependent upon culture density. Curr Eye Res 8~1279-1286. Chen YY, Rabinovitch PS (1998). Mitogen response and cell cycle kinetics of Swiss 3T3 ceils in defined medium: differences from human fibroblasts and effects of cell density. Exp Cell Res 190:145-150. Paulsson Y, Beckmann MP, Westermark B, Heldin CH (1988). Density-dependent inhibition of cell growth by transforming growth factor-beta 1 in normal human fibroblasts. Growth Factors 1:19-27. Kaighn ME, Kirk D, Szalay M, Lechner JF (1981). Growth control of prostatic carcinoma cells in serum-free media: interrelationship of hormone response, cell density, and nutrient media. Proc Nat1 Acad Sci USA 78:56735676. Santen RJ, Mowszowicz I, Portois MC, Mauvais-Jarvis P (1987). Androgen dependence of the Dunning R3327G cell line in monolayer culture. Prostate 11:377-387. Jozan S, Gay G, Marques B, Mirouze A, David JF (1985). Oestradiol was effective in stimulating ‘H-thymidine incorporation but not on proliferation of breast cancer cultured cells. Cell Tissue Kinet l&457-464. Stampfer MR, Vlodavsky I, Smith HS, Ford R, Becker FF, Riggs J (1981). Fibronectin production by human mammary cells. J Nat1 Cancer Inst 67~253-261. Noel A, Calle A, Emonard H, Nusgens B, et al. (1988). Antagonistic effects of laminin and fibronectin in cell-to-cell and cellto-matrix interactions in MCF-7 cultures. In Vitro Cell Dev Biol24:373-380. Calvo F, Brower M, Camey DN (1984). Continuous culture and soft agarose cloning of multiple human breast carcinoma cell lines in serum-free medium. Cancer Res 44~4553-4559. Hynes RO, Yamada KM (1982). Fibronectins: multifunctional modular glycoproteins. J Cell Biol%:369-377.
Steroids,
1992, vol. 57, June
275