INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE RELATIONSHIP between dietary fat and cancer remains controversial. Although epidemiological studies previously suggested that high fat intake is positively associated with certain cancers such as colon (50, 105), prostate (37, 41, 61) and breast (39, 40, 81) cancers, recent studies have cast doubt on this relationship (27, 35, 43, 53). Confounding factors and alternative explanations have been proposed to account for the apparent association between dietary fat and cancer (103, 104). Furthermore, prospective studies to reduce dietary fat intake have also produced conflicting results, especially in regard to the incidence of breast cancer in women (13, 24, 45, 93).

On the other hand, studies in experimental animals have been more consistent in showing a positive association between dietary fat and cancer (19, 20, 42, 79). Fats derived from marine animals (fish oil, FO) have been especially investigated. These oils are enriched with polyunsaturated fatty acids (PUFAs) of the n-3 family and are believed to have protective effects at the initiation and promotion stages of carcinogenesis (74, 75). Several investigators, including our group, have also demonstrated in animal feeding studies that diets enriched with FO can reduce the growth rates of implanted tumors (54, 57, 58). These protective effects of FO are generally attributed to its n-3 PUFA content through several hypothetical mechanisms, such as modification of membrane lipids and subsequent changes in signaling pathways, membrane fluidity, prostaglandin synthesis, and ornithine decarboxylase activity (6, 7, 32, 38, 48, 72, 94).

To further delineate the molecular mechanisms that mediate the effect of FO on cell functions, studies have used cell culture techniques in which either FO or individual fatty acids are added to the growth media (6, 16, 26, 48). Recently, Collett et al. (21) demonstrated differential effects for the n-3 and n-6 PUFAs docosahexaenoic acid (DHA) and linoleic acid (LA), respectively, on oncogenic Ras activation in cultured adult mouse colon (YAMC-ras) cells. Other studies in cultured human breast cancer cells also demonstrated differential effects for n-3 and n-6 PUFAs on the activity of the nuclear receptor peroxisome proliferator-activated receptor (PPAR)- and related these effects to cell proliferation (59, 80). By showing close similarity to in vivo feeding experiments, these studies illustrate the validity of cell cultures in evaluating possible mechanisms by which dietary fats influence carcinogenesis and cell proliferation.

In previous experiments with tumor-bearing rats, we showed (54) that feeding with diets enriched with FO reduces the rate of tumor growth, consistent with findings of several other investigators (57, 58). We characterized the cell cycle kinetics of this tumor model in FO-fed and control rats by use of 5-bromo-2'-deoxyuridine (BrdU) pulse labeling and flow cytometric analysis. The most noticeable result from that study was a delay in S phase completion, which accounted for a slower rate of cell cycle progression, in the cells from FO-fed rats. In particular, although the fraction of cells estimated to enter the S phase was unaffected, the duration of S phase significantly increased after 6 wk of feeding with a diet enriched with FO. Because S phase progression is independent of extracellular regulation of the cell cycle, which is mostly achieved at the level of G₁ restriction point (2, 47), we argued that this effect of FO is distinct from its putative modulatory effects on cell surface signaling such as those of the mitogen-activated protein (MAP) kinase pathways (6, 23, 32, 70). Thus we questioned whether FO feeding could alter the DNA replication machinery, in particular the temporal/spatial organization of DNA replication origins that is the primary determinant of S phase duration (11, 22, 30, 83).

Therefore, to address this possibility, we initially tested the feasibility of reproducing a similar effect on S phase progression in a cell culture system. We chose the Chinese hamster ovary (CHO) cell line in view of the existing knowledge of DNA replication origins in this cell line. Here we characterize the cell cycle kinetics in exponentially growing CHO cells that have been treated with FO. After 5 days of treatment, FO significantly lengthens the duration of S phase of exponentially growing CHO cells, a finding consistent with our previous in vivo feeding studies (54). To determine whether this change in S phase progression is associated with altered firing at the level of DNA replication origins, we used the nascent DNA strand length assay to map the replication origin ori- of the dihydrofolate reductase (DHFR) gene locus. The region downstream from the DHFR gene in CHO cells is the most thoroughly mapped high-frequency initiation region in mammalian chromosomes. We show that in control and corn oil (CO)-treated CHO cells ori- activity localizes to the expected chromosomal site ~17 kb downstream from the DHFR gene. On the other hand, FO-treated CHO cells, which are characterized by a longer S phase duration, exhibit an altered location of this replication origin. These results suggest that FO, and possibly n-3 PUFAs, may affect cell proliferation at the level of organization of replication origins. Although the mechanism of such an effect is not yet understood, the possibility that FO can influence the chromosomal location and/or activity of replication origins in mammalian cells represents a novel finding. Understanding of the interaction between PUFAs and the DNA replication machinery will further clarify the potential role of dietary fats in cell proliferation and cancer.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cell Culture and Oil Treatment

Experimental Part I: Cell Cycle Kinetics

Growth characteristics. Estimates of the actual growth rates of the control and oil-treated CHO cells were determined from changes in cell number as function of time. After 5-day oil treatment, cells were reseeded and incubated at a concentration of 5 × 10⁵ per plate. For three consecutive days, the cells were sampled and counted each day by use of a hemocytometer and Trypan blue (GIBCO). The first-order growth rate constant (K_g) was determined by regression analysis according to the equation

(1)

where C is the total number of viable cells and C_o is the initial cell number at the time point of plating. From the growth rate constant of exponentially growing cells, the actual doubling time T_d can be calculated as

(2)

BrdU pulse labeling and staining procedure for kinetic analysis of cell cycle. To exponentially growing cells in oil-enriched and control media, we added BrdU (Boehringer Mannheim) at a final concentration of 25 µM. After a labeling pulse of 30 min at 37°C, cells were washed with identical fresh medium and then incubated in the corresponding medium for 1, 3, and 5 h, respectively. To obtain a good average, six plates were prepared for each time point. After the indicated BrdU-free chase time, cells were harvested, fixed in 70% cold ethanol, and stored at 20°C until further processing. The staining procedure to determine the DNA and BrdU content was described previously (96). Briefly, ~1-2 × 10⁶ fixed cells were washed and resuspended in 2 ml of PBS and then incubated in the dark with 2 ml of 4 N HCl containing 0.5% (vol/vol) Triton X-100 (Sigma) for 30 min at room temperature. This procedure produces single-stranded DNA molecules, improving the staining with anti-BrdU antibodies. Subsequently, cells were centrifuged at 500 g for 5 min and neutralized with 2 ml of 0.1 M sodium tetraborate (pH 8.5). Cells were washed twice with PBS containing 0.5% (vol/vol) Tween 20. After pelleting, cells were resuspended in 50 µl of PBS-Tween 20 and incubated in the dark with 20 µl of fluorescein isothiocyanate (FITC)-labeled monoclonal anti-BrdU antibodies (Becton-Dickinson, Mountain View, CA) for 30 min at room temperature. For staining of total DNA, cells were then washed twice with PBS-Tween 20 and resuspended in 1 ml of PBS containing 10 µg of propidium iodide (Calbiochem, San Diego, CA). Samples were ready for bivariate DNA-BrdU flow cytometric analysis after 15 min.

Flow cytometry. Double-stained cells were analyzed with a FACScan flow cytometer (Becton-Dickinson) at an excitation wavelength of 488 nm and a laser power of 15 mW. Red fluorescence from propidium iodide was collected through a 585-nm band-pass filter as total DNA content. Green fluorescence from FITC-labeled anti-BrdU antibodies was collected through a 530-nm band-pass filter. The red fluorescence was calibrated by adjustment of the G₀/G₁ peak to a fixed channel number, and the green fluorescence was calibrated by use of FITC-labeled latex beads (Polysciences, Warrington, PA). A total of 10⁵ cells were monitored. Flow cytometric data were accumulated at the highest resolution for each parameter (1,024 channels).

Kinetic analysis of flow cytometric data. Two separate procedures were used to determine the kinetic characteristics of exponentially growing CHO cells. In the first procedure, double-stained cells were subjected to an univariate DNA analysis (red fluorescence) only. The distribution of cells in each phase of the cell cycle was determined by mathematical analysis of the DNA histograms (see, e.g., Fig. 2) with the computer program ModFit (Verity Software House, Topsham, ME).

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

Experimental Part II: Mapping of DHFR Replication Origin ori- in Treated CHO Cells

Cell labeling and isolation of nascent DNA. Control and oil-treated, exponentially growing CHO cells were labeled with [³H]deoxycytidine (³H-dC, 10 µCi/ml) in the presence of 50 µM BrdU for 15 min at 37°C (91). The cells were washed three times with cold PBS and lysed in 7 ml of 0.5% sodium dodecyl sulfate, 1 M NaCl, 10 mM EDTA, and 50 mM Tris · HCl (pH 8.0). After incubation with proteinase K (200 µg/ml), DNA was isolated by phenol-chloroform (1:1) extraction and spooling on glass rods under 70% ethanol. The spooled DNA was rinsed in 95% ethanol.

Size fractionation of DNA by gradient centrifugation. The labeled and isolated DNA was dissolved in 4 ml of TE buffer [10 mM Tris · HCl (pH 8.0), 1 mM EDTA] by reversing the direction of spooling. NaOH (1 N) was added to give a final concentration of 0.2 N NaOH. The DNA was fractionated by sedimentation through a 5-20% (wt/vol) linear sucrose gradient containing 0.2 N NaOH and 2 mM EDTA for 18 h at 15°C in a Beckman SW28 rotor at 24,000 rpm. Gradients were collected in 12 fractions, and portions were taken to assay for ³H radioactivity. Fractions 2-11 were each neutralized with 2 N HCl in the presence of 0.1 M Tris · HCl (pH 7.5) and precipitated with ethanol. Fractions 1 and 12 were discarded. For convenience purposes fractions 2-11 were renumbered as fractions 1-10.

Immunoprecipitation of nascent DNA chains. BrdU-containing nascent DNA strands were purified by two rounds of immunoprecipitation with anti-BrdU monoclonal antibodies (90, 91). Briefly, DNA from each fraction was dissolved in 0.5 ml of TE buffer and denatured for 3 min at 95°C. After rapid cooling with an ice bath, each fraction was adjusted to 10 mM sodium phosphate (pH 7.0), 0.14 M NaCl, and 0.05% Triton X-100 (immunoprecipitation buffer) and incubated with 80 µl of mouse anti-BrdU monoclonal antibody (Becton-Dickinson) for 20 min at room temperature with slow agitation. Rabbit anti-mouse IgG (80 µl; 25 µg/ml) was then added to precipitate BrdU-DNA-antibody complexes. Immunoprecipitates were collected by centrifugation for 5 min and washed once with 0.5 ml of immunoprecipitation buffer. Pellets were deproteinized by overnight incubation at 37°C in 200 µl of 50 mM Tris · HCl (pH 8.0), 10 mM EDTA, and 0.5% sodium dodecyl sulfate containing 250 µg/ml proteinase K, followed by phenol-chloroform extraction (1:1) and ethanol precipitation. DNA pellets were dissolved in 100 µl of TE buffer, subjected to a second round of immunoprecipitation, and then precipitated with ethanol in the presence of 20 µg of Escherichia coli tRNA (Sigma). These nascent DNA fractions were dissolved in 50 µl of TE buffer and used for PCR amplification.

PCR amplification conditions. Oligonucleotide primers and probes for the marker segments A, B, and C, homologous to the region of the DHFR replication origin in CHO cells (15), were chemically synthesized (Baron Biotech, Milford, CT). Nucleotide positions of each primer pair downstream of the XbaI restriction site were as follows (15, 91): segment A, 5': 81 to 100 (biotin-5'-GTG CTA GAA GTA GAT GAG AG-3'), 3': 381 to 400 (5'-AAT CCA GCA TGC GAA CAG TT-3'); segment B, 5': 2677 to 2696 (biotin-5'-TTC TCA GTG AGT CCA CTT CT-3'), 3': 2976 to 2995 (5'-CCT GGT AGG GAC TTC AGA AA-3'); segment C, 5': 4569 to 4588 (biotin-5'-AGT ATT GTA GGT ATG TGC CC-3'), 3': 4955 to 4974 (5'-GTT GTG CTT TAG TGA TAG GG-3'). PCR amplification was performed at a final concentration of 1× PCR buffer containing 50 µM dNTPs, each 5' and 3' primer at 0.1 µM, and 1 unit of AmpliTaq DNA polymerase (Perkin-Elmer, Norwalk, CT) in a total volume of 50 µl. All amplification and hybridization reactions were performed in a DNA Thermal Cycler 480 (Perkin-Elmer). The amplification temperature profile involved a denaturation step at 95°C for 0.5 min, a primer-annealing step at 58°C for 0.5 min, and an extension step at 72°C for 1 min and 30 cycles of amplification.

Quantification of PCR products by electrochemiluminescence-based detection system. An automated QPCR 5000 system (Perkin-Elmer) was used to quantify the amplified PCR products. This system is designed to detect electrochemiluminescence (ECL) generated by the reaction between tris(2,2'-bipyridine)ruthenium(II)-chelate (TBR) and tripropylamine (TPA). TBR is attached to the PCR probe, whereas the reactant TPA is present in the QPCR assay buffer. Typically, 5 µl of PCR product was mixed with 35 µl of PCR buffer [10 mM Tris · HCl (pH 8.3), 50 mM KCl] and 10 µl of 1 pmol/µl TBR-labeled probe. For hybridization, the amplified PCR products were denatured at 95°C for 3 min and hybridized with the probe at 60°C for 5 min. Twenty microliters of 2 mg/ml streptavidin-coated magnetic beads, which bind the biotin-labeled PCR sequences, was then added to separate the ECL-signal quantitatively. After 20 min at 55°C, the products of each hybridization reaction were transferred to a separate QPCR System 5000 tube containing TPA in QPCR assay buffer to measure the ECL signal. The signal measured by this system corresponds proportionally to the amount of hybridized PCR product. The TBR probes used in the hybridization reaction are as follows: segment A, 145 to 165 (TBR-5'-CTA CAA TCC TTC CTC TCT CTT-3'); segment B, 2787 to 2807 (TBR-5'-GCT GAA CTT TAT CAG TGC AGT-3'); segment C, 4643 to 4662 (TBR-5'-TG TGT AGC ACA GTC TAG TCT-3').

Statistical Analysis

For analysis of the effect of FO on ori- mapping characteristics, 95% confidence intervals for relevant ECL signal ratios were calculated from untreated (control) cells, based on a replicate number (n) of 4. Subsequently, the corresponding range of experimental values in the treated cells was evaluated in reference to the 95% confidence interval thus defined. For statistical evaluation of the overall significance of the effect of FO on ori- mapping, the null hypothesis that differences in the ECL ratios among treatment groups were due to chance was tested by ²-analysis. In this analysis, two possible outcomes were considered, i.e., whether a specific FO-treated ECL ratio was similar to or different from the expected ratio defined by the data from untreated cells. All statistical analyses were performed on a desktop computer with the Microsoft Excel 97 software package.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

FO Prolongs S Phase Duration in Exponentially Growing CHO Cells

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Growth curves of Chinese hamster ovary (CHO) cells treated for 5 days with control growth medium (circle) or medium enriched in corn oil (triangles) or fish oil (squares). After reseeding, cells were counted daily for 3 consecutive days. Data are mean values pooled from 3 experiments; error bars represent SE. Lines represent the best fit for exponential growth (Eq. 1) by use of regression analysis. The corresponding values of the actual doubling times (Td) are given in Table 3.

For estimation of the cell cycle kinetics, we first analyzed DNA histograms in each treatment group. The one-dimensional histograms represent the total DNA stained by propidium iodide and allow determination of the distribution of cells in each phase of the cell cycle. In Fig. 2 we show representative DNA histograms for control (Fig. 2A), CO-treated (Fig. 2B) and FO-treated cells (Fig. 2C). For each group, the percentages of cells in G₀/G₁, S, and G₂/M phase are indicated as obtained by the software program ModFit. The percentages for each cell population averaged over six independent experiments are summarized in Table 1. The 5-day treatment with FO or CO did not disturb the diploid nature of the CHO cell line. In addition, flow cytometric analysis showed no evidence for cell death or apoptosis. Cells treated with FO had ~5-6% more cells in S phase (P < 0.005) and slightly fewer cells in G₁ phase compared with controls and CO-treated cells.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Representative DNA histograms for control (A) and corn oil (B)- and fish oil (C)-treated CHO cells. The total DNA content of cells (x-axis) was obtained by staining with propidium iodide. Cells were analyzed by flow cytometry. The percentage of cells in each cell cycle phase was determined with the program ModFit. Average values from 6 experiments are summarized in Table 1. The first, large peak represents population of cells (y-axis) in G0/G1 phase, the second, small peak shows population of cells in G2/M phase, and the gray area between both peaks represents cells in S phase. CV, coefficient of variation and a measure of resolution for the G0/G1 peak.

To determine kinetic parameters of the cell cycle, we analyzed cells of the three treatment groups by bivariate flow cytometry. The cells were stained for total DNA content (with propidium iodide) and for BrdU content (with FITC-BrdU antibodies). For kinetic analysis, cells were harvested and fixed at three different time points after a BrdU pulse for 30 min. Figure 3 represents a typical example of bivariate BrdU-DNA contour plots recorded at the earliest (Fig. 3A) and the last (Fig. 3B) time point. In Fig. 3A we show that populations of cells in G₀/G₁ and G₂/M phase are separated according to their DNA content and cells in S phase are distinguished according to their BrdU content at 1 h after pulse labeling. As displayed in Fig. 3B, BrdU-labeled cells have separated into labeled, divided (f_ld) and labeled, undivided (f_lu) fractions after 5 h. The percentage of cells in each fraction was determined by use of the software program IsoContour and is summarized in Table 2. At the 5-h time point, significantly fewer FO-treated cells had divided compared with either control or CO-treated cells. In the FO group only 12% of cells had completed cell division by that time and appeared in the G₁ phase population, compared with 20% and 16% in control and CO-treated groups, respectively (P < 0.05). Because each division of labeled cell results in two labeled daughter cells, the contribution of divided and undivided fractions to the total number of labeled cells is expressed by the function  (Eq. 3). This function characterizes the fraction of cells traversing a single S phase and is directly proportional to the rate of G₁/S phase transition. Numeric values of  are listed in Table 2. In each of the treatment groups, ~28% of exponentially growing CHO cells underwent DNA replication during the S phase. This finding implies that the rate of G₁-to-S phase transition in the FO-treated group was similar to that of either control or CO-treated cells.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   General example for bivariate analysis of 5-bromo-2'-deoxyuridine (BrdU)-DNA contour plots obtained by flow cytometry. After a BrdU pulse for 30 min, treated cells were incubated for 1 (A) and 5 (B) h with the corresponding BrdU-free medium. The x-axis represents the linear intensity of red fluorescence obtained from propidium iodide (total DNA content), and the y-axis represents the logarithmic intensity of green fluorescence obtained from FITC-conjugated anti-BrdU antibodies (BrdU-labeled population). Cells are separated into G0/G1 phase and G2/M phase according to their DNA content, and into labeled undivided (flu) and labeled divided (fld) subgroups according to the DNA content of BrdU-labeled cells. Estimates of the percentage of cells in both subgroups as well as the relative position of cells in G0/G1 and G2/M phase (according to their DNA content) were derived with IsoContour software. As a result of DNA synthesis, BrdU-labeled cells (S phase, A) move toward higher DNA content (flu, B) as demonstrated.

Further analysis of BrdU-DNA plots allows estimation of the S phase duration time T_S as well as the mean times spent in G₁, G₂/M, and the total cell cycle. Two analytical methods are presented, both of which are derived from RM measurements, as defined by Eq. 6. The results of relative movement at 1, 3, and 5 h after BrdU pulse labeling are summarized in Fig. 4. One hour after BrdU labeling, the RM values were significantly higher in FO-treated cells (P < 0.01 for FO vs. either control or CO). However, in cells examined 5 h after pulsing, FO-treated cells had significantly lower values of RM compared with the other treatment groups (P < 0.005 for FO vs. either control or CO). Thus the slope of the linear regression function RM(t) was significantly smaller after FO treatment as shown in Fig. 4. The actual numeric values of RM are given in Table 2 for the 5 h time point [RM(5), Eq. 6] and for the initial time point t = 0 (RM_o, Eq. 8).

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Plot of the relative movement of labeled undivided cells as function of time shown for control (circle) and corn oil (CO; triangles)- and fish oil (FO; squares)-treated cells. The relative movement of labeled cells in each treatment group was measured 1, 3, and 5 h after the BrdU pulse and calculated according to Eq. 5. Data are averaged over 6 experiments and given as mean values ± SE. The lines represent the best linear regression curve (r2 > 0.98 for each treatment group). Numeric values of the relative movement at 5 h [RM(5)] are displayed in Table 2. S phase duration times TS are calculated from Eqs. 7 and 11 and are shown in Table 3. *P < 0.05 or better for fish oil compared with either control or corn oil.

The results of determining the S phase duration times T_S, which are based on the linear fit of RM(5) and RM_o, as summarized in Table 3, show that the S phase duration time was 8.2 ± 0.2 h for FO-treated cells, which is significantly longer compared with 6.7 ± 0.2 h and 6.6 ± 0.2 h for control and CO-treated cells, respectively (P < 0.005 for FO vs. either control or CO). Table 3 also summarizes the results of T_S derived from the cubic fit of the 5-h BrdU/DNA data according to Eqs. 9-11. It is noted that estimates of T_S derived from this method were consistently smaller than those obtained from the linear model. However, the difference between treatment groups persisted, with FO-treated cells having significantly longer S phase duration times.

Knowing the S phase duration time T_S, which is equivalent to the DNA replication time, we calculated the potential doubling time T_pot for the three treatment groups according to Eqs. 4 and 5. Corresponding numeric values are listed in Table 3, which also contains the estimates of the total cell cycle time (T_c), the mean times spent in G₁ and G₂/M phase, and the actual doubling time (T_d) derived from the growth curves (Fig. 1). These results show that FO treatment, in addition to the effect on S phase, also significantly increased the mean time spent in G₂/M phase. On the other hand, estimates of the total cell cycle time and potential doubling time, derived from the cubic fit model, were more variable within each treatment group and therefore did not achieve statistical significance between the treatment groups. We also note that the estimates of total cell cycle time were in close agreement with and almost equivalent to the corresponding estimates of potential doubling time. This finding indicates that all the cells were cycling (growth fraction equal to 1), with no significant increases in the nonproliferating cell pool resulting from oil treatments. Furthermore, estimates of the actual doubling times were longer than those of the potential doubling times, indicating significant cell loss from this CHO cell line. However, comparison of the ratio of T_pot to T_d was similar in all the treatment groups. Thus the degree of cell loss, which amounted to 43-50% under the conditions of this experiment, was not affected by the oil treatment, as noted by the parameter  in Table 3 (P = not significant). On the basis of these results, we deduce that most of the difference in the observed proliferation potential in FO-treated cells compared with the control and CO-treated cells is accounted for by the longer S phase duration. Factors that reduce the proliferation potential by halting cells at the G₁ restriction point, thus reducing the growth fraction, or increasing the rate of cell death do not seem to account for a significant part of the effect of FO in this kinetic model.

We conclude from the kinetic cell cycle studies that FO significantly slows the progression of proliferating cells through the S phase, and possibly through mitosis. However, the net effect on actual cellular growth is smaller and more difficult to measure. This apparent discrepancy between the effect on S phase and total growth may be explained by compensatory changes in other cell cycle regulatory mechanisms. These possibilities were not addressed by the current study.

Effect of FO on Replication Origin ori-

View larger version (10K): [in this window] [in a new window]   Fig. 5.   Schematic representation of the organization of the replication initiation zone between the dihydrofolate reductase (DHFR) and 2BE2121 genes (55 kb) in CHO cells. The high-frequency initiation sites (ori- and ori-) as well as the low-frequency site (ori-') are indicated. Furthermore, the position of the 3 marker segments relative to ori- and ori-' are shown in greater detail.

Preliminary experiments were initially performed and repeated with untreated (control) CHO cells to determine the reproducibility of the fractionation and PCR conditions for delineating the putative ori- replication origin, as previously documented in exponentially growing CHO cells (91). Subsequently, two independent experiments, consisting of three treatment groups each, were performed. The relative intensities of the ECL signal generated by each of the three hybridized probes, expressed as the average value from the two experiments where all three treatments were conducted simultaneously, are presented in Fig. 6 as a function of fraction number. In the control and CO-treated cells, segment B continues to be represented in the smallest DNA fractions (highest fraction numbers in Fig. 6B), whereas segments A and C are diminished significantly in lighter DNA fractions, starting noticeably in fraction 7 (Fig. 6, A and C). These results are very similar to those reported previously by Vassilev et al. (91). It is noted that segment C, which also diminishes starting in fraction 7, is undetectable in fractions 9 and 10. In contrast to control and CO-treated cells, cells treated with FO continue to exhibit significant amounts of segment A even in the highest fraction numbers containing the shortest DNA fragments (Fig. 6A). Segment B decreases significantly in fractions 9 and 10 in the FO-treated cells (Fig. 6B), whereas segment C behaves similarly in FO-treated cells as well as in the two control groups (Fig. 6C). These results indicate that in the chromosome region of control and CO-treated cells, DNA replication is initiated closest to segment B, i.e., closest to the replication origin ori- as expected from previous studies (60, 91). In FO-treated cells, however, this activity is suppressed. Instead, the initiation of DNA replication in the FO-treated cells appears to favor a location closer to segment A.

View larger version (32K): [in this window] [in a new window]   Fig. 6.   Summary of quantitative PCR analysis for the 3 marker segments A , B, and C (A-C, respectively) flanking the DHFR replication origin ori- in treated CHO cells. Nascent BrdU-labeled DNA was purified by 2 rounds of immunoprecipitation and separated by use of a sucrose gradient. Fractions are numbered starting with the largest DNA fragments (fraction 1). Segments A, B, and C were amplified by PCR in each fraction and subsequently quantified by an electrochemiluminescence (ECL) signal. In each experiment, ECL was highest for the largest DNA fragments (fractions 1 and 2). Here, the mean ECL value of 2 experiments is shown for each fraction and treatment group. Data are normalized to the average ECL signal in fractions 1 and 2 (measured counts  blank) and expressed as a percentage. Each fraction shows control, fish oil-treated, and corn oil-treated cells.

To approximate the location of a replication origin in reference to a marker, one determines the smallest length of nascent DNA that includes the marker in question. In the control cells, where the replication origin appears to lie closer to segment B, this is best illustrated by examining the ratio of B to A. Because longer segments of nascent DNA contain both markers, the ratio B/A is expected to be ~1.0 in fractions containing larger fragments (low fraction numbers). For smaller nascent DNA fragments, the marker farther away from the replication origin has not been replicated yet, so that no signal can be detected and thus, theoretically, the ratio tends to go to infinity. In actual data, however, DNA fractions contain a range of lengths and therefore, the change in fragment ratios is more gradual. Vassilev et al. (91) previously showed that the transition in hybridization ratios (equivalent to ECL signal in Fig. 6) is likely to occur at values between 1.0 and 3.0. In Fig. 7A, we show the experimental values of the ratio B/A derived for the three treatment groups. In the control and CO groups, B/A achieves a value >3.0 in fraction 7 (control) and fraction 8 (CO), respectively. In the FO group, B/A varies between 1.0 and 0.1, with A/B achieving a value of 3 in fraction 9 (see Fig. 7B). Plots of the ratios C/A (Fig. 7C) and C/B (Fig. 7D) show variations between 1.0 and 10⁴ in each of three treatment groups because the ECL signal for segment C is very similar for all groups and is almost undetectable for the smallest DNA fragments. ECL signal strength ratios A/C and B/C are also presented in Fig. 7. As noted above and in Fig. 6C, segment C did not amplify in fractions 9 and 10, a finding that could result either from the distance of this marker relative to the actual initiation point or from reduced PCR efficiency for this segment. Therefore, values of the ratios A/C and B/C in fractions 9 and 10 are inappropriate because segment C was not quantitatively different from 0. However, despite this limitation, we note that B/C shows an increase from 1.0 to 3-4 in fractions 7 and 8 in the CO and control cells, but not in the FO-treated cells. On the other hand, A/C tends to increase in fraction 8 in FO cells but remains equivalent to 1 in the other two groups.

View larger version (20K): [in this window] [in a new window]   Fig. 7.   Plot of individual ratios of ECL signals for marker segments A, B, and C from Fig. 6. Ratios B/A (A), A/B (B), C/A (C), C/B (D), A/C (E), and B/C (F) for control, corn oil-treated, and fish oil-treated cells are shown. A/C and B/C in fractions 9 and 10 were deleted from the figure (missing data denoted by asterisk) because estimates of the PCR product of this segment were not different from zero in 2 uppermost DNA fractions (see text).

It is noteworthy that differences in ECL signal strength ratios in the FO group compared with the other two groups were significantly larger in fractions 9 and 10, with a coefficient of variation in these measurements ranging between 10 and 25%. In fractions 7 and 8, however, differences related to the FO treatment were more subtle. As noted in DISCUSSION, the actual size of nascent DNA may have a significant impact on the relative importance of these fractions in origin mapping. In the absence of experimental information about DNA fraction sizes in the current study, it is important to further evaluate the smaller differences in ECL ratios noted in fractions 7 and 8. In the control cells, four separate sets of ECL measurements were available (2 from preliminary experiments in untreated exponentially growing CHO cells and 2 in the control groups of the main study). Thus it is possible to define 95% confidence intervals for specific ECL ratios in untreated CHO cells (n = 4). The results of this analysis are presented in Fig. 8 (95% confidence intervals depicted in boxes) for the key ratios B/A, B/C, and A/C in fractions 7 and 8. The range of experimental values obtained in the corresponding fractions in CO- and FO-treated cells, as well as reference values from Vassilev et al. (91), are also depicted in Fig. 8. We note that the 95% confidence ranges for B/A and B/C in fractions 7 and 8 from untreated CHO cells were similar in the current study to the previously reported values (91). It is also noted in Fig. 8 that, in the FO-treated cells, measured B/A and B/C ratios were consistently excluded from the 95% confidence intervals defined by data from the control cells. In contrast, the corresponding values from the CO-treated cells overlapped with these hybridization ratio ranges. Figure 8 also shows that A/C from FO-treated cells, but not those from CO-treated cells, were significantly outside the 95% confidence range of untreated cells in fraction 8. These results, together with the larger differences already noted in fractions 9 and 10 (Fig. 7), imply a possible difference in the initiation of DNA replication in FO-treated CHO cells in reference to the expected location of ori-.

View larger version (19K): [in this window] [in a new window]   Fig. 8.   Plot of the ratios B/A, B/C, and A/C in fractions 7 and 8 in FO- and CO-treated CHO cells in reference to the expected ranges of these ratios in untreated cells. Boxes indicate the 95% confidence intervals (n = 4) derived from untreated control cells in the current study. Data shown as open circles with range lines represent the corresponding values reported in Ref. 91 from 2 independent experiments in exponentially growing CHO cells similar to the control cells of the current study. Mean values for FO- and CO-treated cells, respectively, from 2 independent experiments in the current study are also shown, with lines representing the range of values.

Together, the results presented in Figs. 6-8 show a consistent pattern of hybridization ratios for the FO group in each of the two independent experiments where all three treatments were carried out simultaneously. On the other hand, CO and control cells were consistent with the expected location of ori- in both experiments. These findings led us to propose that the DHFR replication origin ori-, which is normally in close proximity to marker segment B, is suppressed in FO-treated cells (P = 0.083, by ²-analysis based on differences in a single ECL ratio). Instead, the initiation site of DNA replication near segment B in FO-treated cells appears to be closer to marker segment A, which is located upstream from segment B (see Fig. 5). However, it is important to note that in view of the limitations in the mapping method used here, especially when ori- is no longer in proximity to segment B (see below), further experiments with more detailed mapping markers are needed to confirm the results of this study.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This study primarily addresses the mechanism by which FO modulates cell proliferation. Sufficient evidence in the literature indicates that FO, possibly because of its high content of n-3 PUFAs, has a protective effect against carcinogenesis. The potential mechanisms that account for this effect are numerous and have been addressed by many studies in the past 10-15 years. Here, we have focused on the S phase in view of previous in vivo animal experiments in our laboratory, which have shown that FO feeding lengthens the DNA replication time in a breast cancer tumor model. This is an unexpected result because the cell cycle is primarily regulated at specific checkpoints that control exit and entry of cells from one phase of the cell division cycle into the next phase (47, 67). Progression through S phase is generally regarded to be independent of extrinsic regulation (52). We previously hypothesized that changes in the S phase duration can be explained by alteration in the spatial and/or temporal organization of replication origins. However, these replication origins, which are known to be DNA sequence specific in yeast and lower eukaryotic cells, are less well understood in mammalian cells. Furthermore, because of experimental difficulties in higher eukaryotic cells, it was important to address the possibility of an effect on the DNA replication machinery in a system in which replication origins have already been characterized. Thus we have chosen the CHO cell line, in which the replication origin ori- has been described in detail by several investigators using various techniques.

Our initial approach to this problem was to examine the cell cycle kinetics of the CHO cell line to determine the feasibility of reproducing an effect on S phase progression similar to that previously observed in animal feeding studies. Measurement of BrdU uptake in newly synthesized DNA and the rate of progression of this label in S phase allow estimation of the DNA replication time and the generation rate of new cells (defined by T_pot) according to the analytical methods of White and his group (88, 97, 102). In the current study, we report estimates of the DNA synthesis and potential doubling times derived from two separate analytical methods. As noted by White et al. (98), the linear RM(t) model tends to overestimate these parameters for sampling times close to T_S, whereas the cubic fit method is more accurate under these conditions. Data summarized in Table 3 are consistent with these expectations, showing consistently smaller T_S values in the cubic fit model. However, both techniques point to the lengthening of S phase duration as the main cell cycle kinetic effect of FO treatment. We note that despite theoretical simplifications, these methods provide significant information about cell cycle regulatory sites, as demonstrated by the current study. Such information would not be available from cell cycle distribution percentages as usually achieved from simple DNA histograms. For example, the slight increase in percentage of cells in S phase in FO-treated cells could possibly be explained by either an increase in the actual number of proliferating cells or an increase in S phase duration relative to the total cell cycle time. Use of bivariate BrdU-DNA analysis in the current study points toward the latter explanation, thus suggesting a reduction in the proliferation potential with FO treatment. The alternative possibility that an increased S phase population implies a larger growth fraction is inconsistent with the finding of reduced growth rate in the FO-treated cells.

Proliferating cells respond to extracellular factors through an intricate network of signal transduction pathways that start at receptor sites within the plasma membrane. In mammalian cells, growth factor stimulation triggers a cascade of events that lead to the expression of early-response genes, many of which are transcription factors for delayed-response genes, such as cyclins D and E, cdks 2, 4, and 6, and E2Fs. Through regulated protein phosphorylation/dephosphorylation events, cdk/cyclin D and cdk/cyclin E complexes control the passage of cells across the G₁ restriction point and activate the transcription of S phase cyclins and subsequent degradation of the S phase inhibitor. The latter event allows the S phase cdk/cyclin complexes to phosphorylate the regulatory sites of DNA prereplication origins, which are already assembled on replication origins during G₁. Therefore, the net regulatory event arising from the extracellular environment is a decision about the transition of G₁ cells into DNA synthesizing S phase cells (22, 95). Within the S phase, checkpoint regulation of DNA replication is primarily a protective mechanism that averts the replication of damaged DNA (1).

We examined the effects of FO on the proliferation kinetics of CHO cells to determine whether fats with a high n-3 fatty acid content can actually inhibit growth, as has been previously claimed (38, 74, 76, 82). Dietary fats have long been suspected to be involved in cancer causation (18, 34, 49, 84) and cellular growth (26, 38), although controversy still prevails (34, 84, 103). Because lipid molecules are important components of cell membranes and signaling pathways, proposals have been made about possible mechanisms by which dietary fats alter cellular function (6, 9, 56, 68, 86, 87). The majority of the proposed mechanisms focus on the actions of phospholipase C and second messengers diacylglycerol and inositol trisphosphate. Activation of this system initiates a series of events that lead to the activation of protein kinases, release of intracellular calcium, and subsequently activation of MAP kinase and its translocation into the nucleus (6, 32, 64, 73). Inside the nucleus, an active MAP kinase induces the transcription of early growth-response genes (c-fos, c-jun, c-myc), which are transcription factors necessary for the expression of G₁ cyclins (17, 25, 26, 73), as noted above. Thus, at least theoretically, dietary fats have the ability to modulate cell proliferation inasmuch as they influence signal transduction pathways. A large volume of literature has addressed the salutary properties of n-3 fatty acids in marine oils in regard to cell proliferation and cancer (23, 38, 76). By altering the lipid composition of cell membranes in favor of n-3 fatty acids (e.g., linoleic over arachidonic acid), FO reduces the activity of protein kinase C and the subsequent events, outlined above, that activate the early- and delayed-response genes (25, 26, 33, 46, 73, 76, 86). In addition, FO feeding alters the composition of eicosanoid metabolites, for example, reducing the concentration of prostaglandin E₂, which then modulates the activity of several protein kinase systems. The experimental evidence supporting this hypothesis is well documented in various cellular and tissue systems (6, 9, 36, 87).

The finding of prolonged S phase duration is unexpected given the effects of FO known so far and the established links between the extracellular environment, signal transduction pathways, and nuclear events. The current understanding of the DNA replication machinery, its assembly, and function in eukaryotic cells is primarily derived from studies on the various replication proteins and is reviewed in detail by Waga and Stillman (95). Key events include the identification of specific DNA origins, the unwinding of DNA and primosome assembly, the formation of RNA primers, and the activation of polymerases. Binding of the replication protein A (RPA) to prereplication centers, which subsequently serve as replication foci, is the initial step believed to occur in G₁. Subsequent instructions to the assembled DNA replication machinery are mainly derived from S phase cdk/cyclin complexes. Once DNA replication begins at a fully assembled replication fork (origin of bidirectional replication), it proceeds very efficiently until all genomic DNA is replicated. During the progression of S phase, further regulation is achieved by a checkpoint mechanism that protects the cell against excessive DNA damage (1, 69). The components of this signal transduction pathway have been best characterized in yeast and are thought to involve several proteins (known as Rad proteins) and the protein kinase Chk1. Damage to DNA by ionizing radiation causes phosphorylation of Chk1 by a mechanism that requires Rad proteins and leads to inhibition of Cdc25. This results in a decrease in Cdc2 tyrosine-15 dephosphorylation. Furthermore, Chk1 induces the export of Cdc25 from the nucleus, thus reducing the Cdc2/cyclin B complex. This signaling cascade leads to inhibition of mitosis. Other S phase regulatory proteins (e.g., Cds1) are known in yeast. These are activated by hydroxyurea in a mechanism that also involves Cdc25, as well as on Cdc2, through an effect on the two protein kinases Wee1 and Mik1 (1, 12, 14, 47, 78). Although FO treatment of cultured cells could possibly lead to DNA damage through induction of free radicals by PUFAs, the results presented here are unlikely to be explained on the basis of this S phase checkpoint regulation. Cells treated with FO do not show evidence of DNA damage by flow cytometry (Figs. 2 and 3); they appear to start DNA synthesis normally and can still complete DNA replication, albeit at a slower rate; and they go through mitosis without evidence of an interruption. An additional support of these arguments is given by the CO treatment group, which was also exposed to PUFAs and thus to possible free radical formation. However, these cells proceeded through the cell cycle unaffected and in a similar manner as the control cells.

Approximately 3 × 10⁹ base pairs comprise genomic DNA in mammalian cells, which is normally replicated in ~8 h, at a fork rate of only 100 bp/s. If each replication fork were active for the total duration of S phase, ~1,000 initiation origins would be needed to replicate the whole genome. In fact, autoradiographic studies suggest that between 10,000 and 100,000 replicons participate in the replication process. Thus replication at each origin is active only during a fraction of the S phase. This implies that the pattern of spatial and temporal distribution of replicons is a major factor in determining the time needed for completion of DNA replication. Therefore, to explain the alteration in S phase kinetics in FO-treated cells, we hypothesize that these cells have an altered pattern of firing of replication origins. This alteration could be spatial (i.e., the location and/or number of active replication origins could be affected), temporal (i.e., the timing of firing might be affected), or a combination of both. More than 30 years ago, Huberman and Riggs (51) demonstrated that the total duration time of S phase is a characteristic parameter of each cell type. They showed that DNA replication is initiated simultaneously at multiple locations. Furthermore, it was clear from this early work that the duration of S phase is determined by the spatial and temporal frequency of firing of the replication origins rather than by the rate of elongation of newly synthesized DNA (51). Since this pioneering work, major advances in understanding of DNA replication as well as in the structural organization of the replication origins have been made. In yeast and lower eukaryotic cells, DNA replication starts at chromosomal sites with specific DNA sequences. However, in higher eukaryotic cells the identification of sequence-defined DNA replication origins has been more elusive (28-30, 65). Studies using two-dimensional electrophoresis in higher eukaryotic cells suggest that DNA replication begins in a broad zone that is unlikely to be defined by specific DNA sequences. However, subsequent studies with PCR-based techniques using nascent DNA strand analyses were able to localize DNA initiation origins to discrete sites. In CHO cells, a replication initiation zone, which comprises ~55 kb and is located between the DHFR gene and the 2BE2121 gene, has been thoroughly investigated by several research groups using a variety of techniques (4, 15, 31, 55, 60, 63, 91, 92). Three major initiation sites have been characterized, two of which are high-frequency start sites (called ori- and ori-) and one a low-frequency start site that was discovered recently (ori-') (60). Ori- is located ~17 kb downstream from the 3' end of the DHFR gene, and the active initiation zone has been consistently determined as 2 kb; ori-' lies ~5 kb further downstream from ori-; and ori- is found at ~40 kb downstream from the DHFR gene (Fig. 5). Despite this characterization of the DHFR replication origin locations, the exact significance of specific DNA sequences remains incompletely understood. Kalejta et al. (55) demonstrated that, in CHO cells, restoration of the 3' end missing sequence of a DHFR gene knock-out without the ori- locus did support replication normally but was suppressed in variants lacking the 3' end. These investigators concluded that it is the 3' end of the DHFR gene, rather than a specific DNA sequence, that is required for initiating DNA replication at ori- in the DHFR gene locus in CHO cells. On the other hand, Altman and Fanning (4) were able to initiate replication in random ectopic chromosomal sites by transfecting a 5.8-kb fragment containing ori