• Open Access

Nonclonal growth of preneoplastic cells positive for glutathione S-transferase P-form in the rat liver



We investigated the process of induction of preneoplastic cells positive for glutathione S-transferase P-form (GST-P) in the rat liver. AAF (2-Acetylaminofluorene) mixed with normal rat chow at high concentration (0.04%) induced 517 000 ± 86 000 GST-P+ single hepatocytes/g liver after 2 weeks followed by induction of a few foci and nodules after 4–6 weeks. Overproduction of GST-P+ single hepatocytes was dose- and time-dependent, and the induction kinetics were typical of first-order consecutive reaction, by which induction of the positive cells was nongenetic. Quantitative analysis indicated that the estimated numbers of cells in foci and nodules at 4–6 weeks after exposure to AAF ranged from 2.7 × 104 (214.7) to 3.6 × 106 (221.7) cells, and 2.0 × 104 (214.3) to 2.7 × 106 (221.4) cells, respectively, when analyzed by using two equations. According to the initiated cell theory of Farber, foci and nodules are formed through sequential cell division of 14 to 21-times or more within a short time period. The rapid growth exceeded the rate of cell division, indicating that the growth of preneoplastic cells is based on a nonclonal penetration mechanism. (Cancer Sci, doi: 10.1111/j.1349-7006.2012.02325.x, 2012)

The molecular and cellular mechanisms of cancer initiation remain a mystery.[1-3] However, the induction process of pre-neoplastic cells could provide clues to identify the mechanism of initiation of rat chemical hepatocarcinogenesis. In 1987, Moore and coworkers reported the presence of single hepatocytes and minifoci strongly positive for glutathione S-transferase P-form (GST-P, EC in the rat liver as early as 2–3 days after i.p. administration of diethylnitrosamine (DEN).[4] Based on their phenotypic identity and the fact that the GST-P+ single hepatocytes were induced prior to the formation of foci and nodules in the animal liver, the former cell populations were considered the precursors of the latter ones.[5-7] Although the appearance of these preneoplastic cells is thought to be the result of genetic changes in hepatocytes, it remains unresolved whether genetic changes are actually involved in the expression of the marker enzyme, and whether the GST-P+ single hepatocytes could form minifoci, foci and nodules. Glutathione S-transferase P-form is an isoenzyme of GST and a member of the phase II detoxication enzymes.[8-10] Glutathione S-transferase P-form is considered to act specifically against water-soluble carcinogens.[11, 12] In addition, the induction of GST-P correlates with the formation of acrolein,[13, 14] an endogenous carcinogen, whereas GST-P acts as binding protein to the GSH-conjugate of the toxic aldehyde.[7]

On the other hand, medium-term bioassays using GST-P as a surrogate endpoint marker, have identified the formation of GST-P+ foci and nodules following exposure to various carcinogens and carcinogenic modifiers.[15-17] In addition, the expression of the GST-P gene is also modulated by various enhancers and transcriptional factors such as GPE1 enhancer and C/EBPα.[18, 19]

In the present study, we examined the process of preneoplastic cell induction by chemical kinetics. The results indicated that GST-P+ single hepatocytes can be induced nongenetically rather than genetically. Unexpectedly, animals subjected to carcinogenic stress showed non-clonal growth of preneoplastic cells at a growth rate far beyond the rate of cell division.

Materials and Methods


Affinity-purified polyclonal anti-GST-P antibody (rabbit) was obtained from Medical and Biologic Laboratories (Nagoya, Japan). Polymeric HRP-conjugated anti-rabbit IgG was purchased from Dako Cytomation (Glostrup, Denmark). AAF was purchased from Nacarai Chemical Company (Osaka, Japan). The basal diet was purchased from Oriental Food Corporation (Tokyo).

Animals and treatments

Male Sprague–Dawley rats (5-week-old) were purchased from CLEA Japan and maintained at the Institute for Animal Experiments at Hirosaki University. All animal experiments were conducted according to the Hirosaki University Guidelines for animal experimentation and the research protocol was approved in advance by the Ethics Committee. According to the protocol described in Figure 1, 6-week-old rats were fed a basal diet containing appropriate concentrations of AAF and then killed at appropriate time points.

Figure 1.

Schematic illustration of the experimental protocols. (A) Rats were fed basal diet containing varying concentrations of AAF (2-Acetylaminofluorene) from 0.00–0.06% for 2 weeks. (B) and (B′), Rats were fed either high-AAF (0.04%) diet or low-AAF (0.02%) diet. Open square, administration of AAF-containing basal diet; ↓, time of death. Values in parentheses denote AAF concentration in the basal diet.

The livers were excised and cut into 3 to 4-mm thick slices for fixation in 10% phosphate-buffered formalin at 4°C. The fixative was replaced the next day as reported previously.[20]

Measurement of precursor cell populations

Liver pieces were sectioned at 25-µm thickness in phosphate buffered saline (PBS) using vibratome 1500 (Vibratome Product, New York, NY, USA). Immunocytochemical staining for GST-P was performed as described previously.[20] Color images of the microslices were incorporated into Photoshop Elements 2.0 using an Epson ES-2000 scanner (Seiko-Epson, Tokyo). Photomicrographs were acquired using a Coolscope digital microscope (Nikon) equipped with a light emitting diode (LED). The cell composition and the area of GST-P+ minifoci and foci were measured using NIH image software version 1.62(Bethesda, MD, USA). GST-P-stained single hepatocytes were counted by direct visual inspection.

Statistical analysis

Values were expressed as mean ± SD. Differences between groups were examined for statistical significance using the Student's t-test. A P-value <0.05 was considered statistically significant.


AAF dose-dependent induction of GST-P+ single hepatocytes in rat liver

As illustrated in Figure 1(A), groups of three to five rats were fed basal diet containing increasing concentrations of AAF (0%, 0.01%, 0.02%, 0.03%, 0.04%, 0.05%, or 0.06%) for 2 weeks before sacrifice. As shown in Figure 2(A), the number of GST-P+ single hepatocytes induced in the liver was highly dependent on the dose of AAF.

Figure 2.

Induction of GST-P+ single hepatocytes in rat liver. (A) Dose-dependent induction of GST-P+ single hepatocytes. Rats were fed basal diet containing 0.00–0.06% AAF (2-Acetylaminofluorene) for 2 weeks as illustrated in protocol A (Fig. 1). GST-P+ single hepatocytes were analyzed as described in Materials and Methods. Data are mean ± SD. *P < 0.005, compared with AAF-fed (0.02%) rats. (B) Representative staining pattern of GST-P+ single hepatocytes induced in the liver of an animal fed high-AAF (0.04%) diet for 2 weeks (low magnification × 5). BD, bile duct. (C) Higher magnification of GST-P+ single hepatocytes (magnification: a–f: ×40). GST-P, glutathione S-transferase P-form.

Whereas GST-P+ single hepatocytes were barely detected in the livers of rats fed either 0.01 or 0.02% AAF, the number increased to a maximal count of 1200 ± 198 cells/cm2 (n = 5) when the rats were fed a diet containing 0.04% AAF. Higher dietary AAF concentrations depressed the induction of GST-P+ hepatocytes, probably due to the strong toxicity of the carcinogen. The maximum number induced by 0.04% AAF was significantly higher than that induced by 0.06% AAF; and tended to be different from those induced with either 0.03% or 0.05% AAF. According to the stereological estimation by Campbell et al.,([21]) the maximum number of 1200 ± 198 cells/cm2 and a mean diameter of 23.2 ± 4.7 µm (n = 149) equates with 517 000 ± 86 000 GST-P+ single hepatocytes/g liver (n = 5).

In protocol A, GST-P+ minifoci composed of 2–50 cells were hardly detectable. Microscopic examination showed GST-P+ single hepatocytes dispersed throughout the rat liver (Fig. 2B) as noted previously.[4, 5, 20] The cytosolic portion was strongly positive for the marker enzyme, but the nuclei were negative (Fig. 2C, a–h). Most of the single hepatocytes showed features of significant injury and contained enlarged cytosolic compartments. Neither shrunk nuclei nor apoptotic bodies were detected in the GST-P+ hepatocytes and debris. Several binuclear hepatocytes (Fig. 2C, e–h), along with the majority of mononuclear hepatocytes (Fig. 2C, a,b), responded to strong carcinogenic stress. The proportion of binuclear hepatocytes [(number of binuclear hepatocytes)/(total number of hepatocytes) × 100], was 12.3% (391/3170 cells) in the carcinogen-treated rat livers, compared to 4.6% (315/6817) in the control rat livers (P < 0.001).

Cell induction kinetics of GST-P+ single hepatocytes

As illustrated in protocols B and B′ (Fig. 1), groups of three to five rats were fed diet containing either high AAF (0.04%) or low AAF (0.02%), for 12 weeks. Figure 3 shows the time course for the induction of GST-P+ single hepatocytes in the rat liver on high- and low-AAF diet. High-AAF diet was associated with the appearance of 517 000 ± 86 000 cells/g of liver after 2 weeks of feeding. However, the number of positive cells then decreased by 6–7 weeks and thereafter became undetectable.

Figure 3.

Time course of induction of GST-P+ single hepatocytes in rat liver. Rats were fed high-AAF (2-Acetylaminofluorene) (0.04%) diet over 12 weeks (closed circles). For comparison, the low-AAF (0.02%) diet data (open circles) are represented in part from those reported previously.[23] The number of GST-P+ single hepatocyte induced in the liver and the percent transformation were examined as a function of time. n.d., not detectable. Data are mean ± SD. GST-P, glutathione S-transferase P-form.

In contrast, the proportion of GST-P+ single hepatocytes increased gradually in the low-AAF diet group to reach 171 000 ± 69 000/g liver (n = 4) by 6 weeks, but it decreased subsequently. These results point to a transient induction of the positive cells in the livers of rats exposed to the carcinogen. The percentage induction of normal hepatocytes into GST-P+ single hepatocytes in the high-AAF diet rats was [(number of GST-P+ single hepatocytes/g)/(number of normal hepatocytes/g)]×100 = (517 000 ± 86 000)/1.69 × 108]×100 = 0.31 ± 0.05%, and the frequency of induction was (3.1 ± 0.5) × 10–3, where 1.69 × 108 represents the number of normal hepatocytes/g liver according to the study of Weibel et al.[22] Importantly, the transient induction of GST-P+ single hepatocytes analyzed in this study was found to be typical of the first-order consecutive reaction, as discussed in a later section.

Induction process of foci and nodules

We tentatively defined GST-P+ single hepatocytes, minifoci, foci and nodules to represent 1, 2–50, 50–1000, and > 1000 GST-P+ cells, respectively, when stained with GST-P antibody as reported.[5, 20] Whereas numerous GST-P+ single hepatocytes were noted after 2 weeks of high-AAF diet (0.04%), the numbers of GST-P+ minifoci and foci induced after 3–4 weeks were both <100/g liver and the percent GST-P+ area was <0.1%. On the other hand, a few but relatively large foci and nodules were noted after 4–6 weeks. These cell populations were tightly bound to the bile ducts and ductules, as described previously.[20, 23] Several GST-P+ and GST-P foci and nodules were induced after 6–10 weeks; however, compared to rats fed low-AAF diet, the numbers of GST-P+ foci and nodules were reduced after 6–10 weeks in rats fed high-AAF diet due to the increase in foci area (%) (Fig. 4B).

Figure 4.

Induction of GST-P+ minifoci and foci in the rat liver. Rats were fed high-AAF (0.04%) diet over 12 weeks as illustrated in protocol B (Fig. 1). (A) Macroscopic observation of minifoci and foci; (a–f) represent typical clonal expansion induced in the liver within 12 weeks. Time points of death are indicated for individual liver samples. (B) Quantification of GST-P+ minifoci and foci induced in the rat liver. The area of GST-P+ foci was quantified as described in Materials and Methods. Closed bars, high-AAF (0.04%) diet; open bars, low-AAF (2-Acetylaminofluorene) (0.02%) diet. Data are mean ± SD. *P < 0.005, compared with animals fed low-AAF diet for 12 weeks. GST-P, glutathione S-transferase P-form.

Cell counts in foci and nodules

The numbers of cells in foci and nodules induced in rats fed the high-AAF diet were determined quantitatively. Assuming that the foci and nodules were all spherical and cut into two exact halves in the liver section, the cell count of a focus or nodule can be derived from the following two equations, respectively.

display math(1)
display math(2)

where V is the volume of a focus or nodule (µm3), v is the mean cell volume of the component cells of foci or nodules, D is the diameter of a focus or nodule, d is the mean diameter of the component cells of foci or nodules, and N0 is the number of normal hepatocytes per g liver (1.69 × 108/µm3) according to Weibel et al.[22] The mean cell diameter of the component cells of foci measured was, d = 20.3 ± 5.9 µm (n = 43), and that of normal hepatocytes 19.6 ± 3.0 µm (n = 30) as reported previously,[7] where n is the number of cells examined. In equation (1), V is the sum of v. In equation (2), the cellular constitution of foci and nodules was assumed to be identical to that of normal hepatocytes based on the finding of identical mean diameter of the cell forming the foci and normal hepatocytes. The calculated values for foci and nodules (a–f) marked on Fig. 4A are shown in Table 1.

Table 1. Cell counts in foci and nodules marked in Figure 4A
Focus or noduleDiametera (D, mm)Volume (V, μµm3)nA (Cell number)nB (Cell number)
  1. a

    Mean diameter of a focus or nodule measured in eight different directions. Values denote mean ± SD.

a0.61 ± 0.02(1.2 ± 0.1)×10−4

(2.7 ± 0.3)×104

(1.0 ± 0.1)×214.7

(2.0 ± 0.2)×104

(1.0 ± 0.1)×214.3

b0.78 ± 0.05(2.5 ± 0.5)×10−4

(5.7 ± 1.1)×104

(1.0 ± 0.2)×215.8

(4.2 ± 0.8)×104

(1.0 ± 0.2)×215.4

c1.62 ± 0.08(5.3 ± 0.8)×10−4

(5.1 ± 0.8)×105

(1.0 ± 0.2)×219.0

(3.8 ± 0.6)×105

(1.0 ± 0.2)×218.5

d1.45 ± 0.08(3.8 ± 0.7)×10−4

(3.0 ± 0.5)×105

(1.0 ± 0.2)×218.5

(2.7 ± 0.5)×105

(1.0 ± 0.2)×218.0

e2.65 ± 0.11(2.3 ± 0.3)×10−3

(2.2 ± 0.3)×106

(1.0 ± 0.1)×221.1

(1.7 ± 0.2)×106

(1.0 ± 0.1)×220.7

f3.12 ± 0.15(3.8 ± 0.6)×10−3

(3.6 ± 0.5)×106

(1.0 ± 0.2)×221.8

(2.7 ± 0.4)×106

(1.0 ± 0.2)×221.4


Induction of GST-P+ preneoplastic cells has so far been explained solely by the initiated cell theory proposed by Farber[1, 24, 25] in rat chemical hepatocarcinogenesis, where rare initiated cell mutants grow clonally giving rise to resistant hepatocytes forming foci and nodules. However, this hypothesis does not explain the rapid growth of these cells following carcinogenic stress.

Overproduction of GST-P+ single hepatocytes in rat liver by AAF

The induction of GST-P+ single hepatocytes varied according to the dose of AAF. As many as 517 000 ± 86 000 of GST-P+ single hepatocytes/g liver were induced in rats fed high AAF (0.04%) diet for 2 weeks. In contrast, low AAF (0.02%) diet induced approximately one-third that number of positive cells after 6 weeks. The maximum number of cells was one to two orders of magnitude greater than our previously observed values of 12 400 cells/g liver induced by DEN[5] and 17 600 cells/g liver induced by NNM.[6] The estimated frequency of induction of normal hepatocytes into GST-P+ single hepatocytes was 3.1 × 10–3 (Fig. 3). Based on previous studies,[1, 24] the carcinogenic process is initiated by rare genetic changes that occur at a rate of 10–6 to 10–7 induced by carcinogens. Thus, the frequency of induction measured in this study was three to four orders of magnitude greater than that of genetic changes, indicating that the induction of GST-P+ cells was nongenetic.

GST-P+ single hepatocyte cell induction kinetics

Interestingly, induction of the GST-P+ single hepatocyte was typical of the unstable intermediate type based on chemistry and physical chemistry kinetics.[26, 27] Unstable intermediates are formed neither in simple reactions of A → B, nor in Michaelis-Menten typical consecutive reaction. Rather, they are formed through first-order consecutive reaction, A → B → C, where A, B, and C represent starting material (GST-P normal hepatocytes), intermediate (GST-P+ single hepatocytes), and product (dead cells), respectively (Fig. 5).

Figure 5.

Induction of GST-P+ single hepatocytes in the rat liver. (a) Schematic illustration of GST-P+ single hepatocytes. (b) kinetic pattern, (c) formulation of the first-order consecutive reaction.[26] (A, B, and C) are the amounts of the starting material, intermediate, and product, corresponding to GST-P normal hepatocytes, GST-P+ single hepatocytes, and dead cells, respectively, as illustrated in (a). (1), (2), and (3) are equations of A, B and C as a function of time, where A0 = 100% at t = 0. GST-P, glutathione S-transferase P-form.

Although the largest proportion for B was very low (0.31% of the total hepatocyte), the cell induction kinetics observed in Figure 3 are essentially identical to consecutive reaction. This is probably due to the fact that GST-P is induced in certain hepatocytes yielding GST-P+ single hepatocytes but not in all hepatocytes. Since consecutive reactions are irreversible, almost all the GST-P+ single hepatocytes could be eliminated by necrosis. Thus, the process of induction of the GST-P+ single hepatocytes can be illustrated in the schematic diagram shown in Figure 5. The first step (k1) is the rate limiting step in the induction of GST-P in hepatocytes (up to 0.29 ± 0.1 mM of GST-P subunits).[7] The second step (k2) is the loss of positive cells through cell death by necrosis. Therefore, both steps include critical changes. Unfortunately, the necessary conditions are still lacking for the kinetic equations to be solved. For example, it is difficult to estimate the particular fraction of hepatocytes (A0 value) that undergo GST-P induction. Nevertheless, our results indicate that the induction was an irreversible and extreme. That chemical kinetics follows the reaction of chemical compounds implies that the GST-P induction process is essentially a biochemical reaction that is unrelated to genetic changes. The present kinetic data thus indicate that GST-P expression is controlled nongenetically by certain transcription factors[18, 19] or by epigenetic mechanisms.[7]

Nonclonal growth of GST-P+ cell populations

Quantitative analysis in our study confirmed that the cell growth of foci and nodules was nonclonal and nongenetic in nature. For instance, foci and nodules (a–f) shown in Figure 4A contained 2.7 × 104 (214.7) to 3.6 × 106 (221.8) and 2.0 × 104 (214.3) to 2.7 × 106 (221.4) cells, by using two different equations (Table 1). These values are underestimates as the actual size of foci and nodules could be much larger than the observed ones on the slide. Thus, it may be appropriate to conclude that the cell number was 104 to 106 cells (214 to 221 cells), or more, per focus or nodule. Assuming a single cell origin, foci and nodules are formed by repeated cell divisions at least for 14 to 21-times or more within the 4- to 6-week time period. In other words, one ‘pre-malignant’ cell must grow into more than one million cells expansively within a short period of time. The growth rate is almost identical to that described in malignant transplantable hepatomas such as Novikoff hepatoma and Yoshida Ascites hepatoma AH130 that grow freely and rapidly in the nutritious ascitic fluid of the animals.[28, 29] In the liver, the initial GST-P+ induced cells are surrounded by a large number of normal cells within the confinement of the extracellular matrix. The latter make the rapid growth of induced cells difficult with respect to the supply of nutrients and excretion of waste products, with the exponential increase in the metabolic requirements accompanying the rapid cell division for more than 20 times. In particular, the mechanical counter pressure would also increase exponentially by more than 106-fold. The rapid growth is much more prominent in the Solt-Farber protocol than this case, since larger foci and nodules are inducible in rat livers within 2 weeks after 2/3PH.[20, 30] According to the initiated cell theory, foci and nodules have to divide more than 20-times exponentially. However, any increase in liver mass mainly represents hypertrophy of hepatocytes, and hepatocytes are regarded to divide only once or twice at the most.[31] In fact, ‘resistant’ foci and nodules are known to have relatively low growth potentials based on examination by BrdU incorporation, TUNEL-labeling, and proliferation cell nuclear antigen-staining.[25, 32, 33] Thus, the initiated cell theory does not seem applicable to the induction of preneoplastic cells in the rat liver, and GST-P+ single hepatocytes appear nonproliferative irrespective of genetic changes. Since the growth of foci and nodules in the carcinogen-fed rats was far beyond the rate of cell division, the growth rate of these cells was considered to be mainly non-clonal in nature.

Is ‘penetration mechanism’ involved in preneoplastic cell induction?

One theory for the induction of preneoplastic cells is the penetration mechanism. In this mechanism, acrolein plays a key role as both a penetrant and denaturant. Exposure to carcinogens results in injury of membrane lipids and organelles, which results in the release of endogenous carcinogens in the hepatocytes.[1, 34, 35] Previous studies described GST-P as a binding protein, which binds to acrolein, a GSH conjugate, to inactivate its toxicity.[7, 11, 12] Furthermore, Devi and Devaraj[13] reported that the acrolein-generating cyclophosphamide[36, 37] induced both weak and strong GST-P expression in all rat hepatocytes, and indicated that the strongly GST-P+ single hepatocytes and minifoci were distributed along the hepatocyte strands. Since acrolein is excreted through the GS-X pump as a GSH conjugate,[38, 39] this finding indicated that the induction of GST-P expression is mediated through injury of the transporter by the toxic action of acrolein. Thus, the induction of GST-P expression serves to inactivate acrolein toxicity, resulting in the formation of GST-P+ single hepatocytes, as reported previously.[7] In other words, GST-P+ cells contain both exogenous and endogenous carcinogens. Assuming that the hydrophilic acrolein is released out of foci and nodules as penetrant to the surrounding hepatocytes, injury of GS-X transporter, deformation of the excretory routes of exogenous and endogenous carcinogens, accumulation of carcinogens and toxic bile acids, and GST-P induction, are then induced serially in the denatured cell populations. It can be seen that the process of penetration is the reproduction and propagation of GST-P+ hepatocytes and the penetrant in the regions surrounding foci and nodules. Thus, pre-neoplastic cells grow nonclonally without cell division. Based on functional analysis, it is conceivable that foci and nodules represent denatured but activated hepatocytes to detoxicate exogenous and endogenous carcinogens. For example, why preneoplasmic foci and nodules specifically store GSH, glycogen, and/or fat has long been considered an enigma.[40, 41] Such accumulation must be the natural outcome for the preneoplasms, assuming that the cells are but denatured hepatocytes and accumulation of GSH, glycogen, and/or fat is characteristic of the liver parenchymal cells under appropriate conditions. Although preneoplastic cells have so-far been regarded to grow clonally according to the cell initiated theory of Farber,[1, 24, 42] the present nonclonal mechanism warrants further examination.


The authors are thankful to Dr Ichiro Hatayama (Aomori Prefectural Institute of Health and Environment) for the helpful suggestions and discussions. This study was supported by Nippon Boehringer-Ingelheim Company (Osaka, Japan), and Hirosaki University Fund for Promotion of International Scientific Research (Hirosaki, Japan).

Disclosure Statement

The authors have no conflict of interest.