Abstract
Mitemcinal is an erythromycin derivative with motilin agonistic action, developed as a gastrointestinal motor-activating agent. The characteristics of mitemcinal-induced multinuclear hepatocytes (MNHs, hepatocytes with three or more nuclei per cell) from detailed morphological observations together with the results of a study on the mechanisms of MNH formation by combining cytocentrifuge preparations with 5-bromo-2’-deoxyuridine cumulative labeling are reported. MNHs were observed only in rats in the high-dose groups of the subchronic study, with a higher incidence in females and reversibility after twenty-eight days of drug withdrawal, but not observed in dogs. In the chronic study, the incidence increased relative to the dose. Histopathologically, MNHs were preferentially observed in the centrilobular zone, without nuclear atypia or mitotic figures. In the cell kinetic study, the labeling pattern of MNHs included all-positive, all-negative, and mixed labeling patterns of nuclei. The all-negative pattern indicated that the cells were formed by fusion of nondividing cells. The current results indicate that the cell kinetic approach effectively demonstrated the mechanism of mitemcinal-induced MNHs as fusion of hepatocytes and that drug-induced disturbance of mitosis is not involved in the multinucleation of MNHs by mitemcinal.
Keywords
Introduction
Mononuclear and binuclear hepatocytes are commonly observed in the liver. The binuclear cell population is approximately 30%–40% in normal rat liver (Engelmann et al. 1981; Fujii et al. 2004; Gerlyng et al. 1993; Mossin et al. 1994; Seglen 1997; Wheatley 1972). Although the mechanisms are unclear, the ratio of mono- and binuclear hepatocytes is known to change with age (Engelmann et al. 1981; Gerlyng et al. 1993; Seglen 1997; Wheatley 1972), partial hepatectomy (Auvigne et al. 2002; Frederiks et al. 1990; Gerlyng et al. 1993; Melchiorri et al. 1993; Seglen 1997; Wheatley 1972), endocrine status (Lamers 1985; Mossin et al. 1994; Wheatley 1972), and certain hepatomitogenic and hepatocarcinogenic drugs (Auvigne et al. 2002; Gerlyng et al. 1994; Kostka et al. 1994; Kostka et al. 2000; Melchiorri et al. 1993; Palut et al. 1992; Seglen 1997).
In contrast, induction of multinuclear hepatocytes (MNHs) is restricted to a limited number of disease states and certain drugs. Some liver diseases and chemical intoxications are known to induce MNHs in infants of both humans and animals, and hepatocellular fusion has been proposed as the mechanism (Richey et al. 1977; Scampini et al. 1993; Wheeler et al. 1986). On the other hand, some chemicals that have liver oncogenic potential are known to induce MNHs. Such MNHs are considered to originate from single hepatocytes by acytokinetic mitosis (Campbell and Gilbert 1967; Harada et al. 1999; Malarkey et al. 1995; Miszurski and Doljanski 1949; Pohjanvirta et al. 1995; Rachmilewitz et al. 1950; Scampini et al. 1993). Based on this information, at least two mechanisms for induction of MNHs have been hypothesized. One is considered to be related to the development of the liver, and the other is related to chemical carcinogenicity (Richey et al. 1977; Scampini et al. 1993).
Mitemcinal is an erythromycin derivative developed as a gastrointestinal motor-activating agent because of its motilin agonistic action (Ozaki et al. 2007). In the course of the pre-clinical development of the compound, MNHs were found in rat studies. This finding was significant and prompted a toxicologic profile of the compound. In this study we have characterized mitemcinal-inducing MNHs by comparison of incidence in the preclinical studies, and from detailed morphological observations. In addition, we have applied a cell kinetic approach to attempt the clarification of the mechanisms related to the formation of MNHs induced by mitemcinal in rats using a method combining cytocentrifuge preparations with 5-bromo-2’-deoxyuridine (BrdU) cumulative labeling described previously in normal rats (Fujii et al. 2004).
Materials and Methods
Preparation and Administration of Mitemcinal
Mitemcinal was synthesized at Chugai Pharmaceutical Co., Ltd. For the rodent studies, mitemcinal was homogenized in a vehicle of 3% Arabic gum (Wako Chemical Industries Co., Ltd., Japan; Sigma Chemical Co., St. Louis, MO, USA; or Spectrum Quality Products, New Brunswick, NJ, USA) dissolved in distilled water and administered by oral gavage. The vehicle control animals were treated by the same route. For the dog studies, mitemcinal was administered in a 1/2-ounce gelatin capsule (Kasho Co., Ltd., Japan), whereas control animals were given an empty capsule.
Experiment 1: Pathological Characterization of MNHs Induced by Mitemcinal
Study Design
Table 1shows the design of the groups for the pre-clinical studies with mitemcinal. In the subchronic studies in rats and dogs, recovery groups were determined and these groups were observed for reversibility after a further twenty-eight days. The acute single dose and subchronic daily dose studies in rats and dogs and the chronic, daily dose studies in rats were conducted at Fuji Gotemba Laboratories of Chugai Pharmaceutical Co., Ltd. (Shizuoka, Japan). The chronic study in dogs was conducted at Shin Nippon Biomedical Laboratories, Ltd. (Kagoshima, Japan). Sprague-Dawley rats were purchased from Japan SLC, Inc. (Shizuoka, Japan), and were six weeks of age at study initiation. CSK beagles were purchased from CSK Research Park Co., Ltd. (Nagano, Japan), and were seven to eight months of age at study initiation. All animal experiments were approved by the Committee for Ethical Treatment of Laboratory Animals of each test facility.
Pathological Procedure
In the acute, subchronic, and chronic studies, rats were sacrificed by exsanguination from the abdominal artery under ether anesthesia, and dogs by exsanguination from the carotid artery under pentobarbital sodium anesthesia. At necropsy, the liver was removed from all animals, fixed in 10% or 20% neutral buffered formalin, and embedded in paraffin. Thin sections were prepared from the paraffin blocks, stained with hematoxylin and eosin, and submitted for histopathological examination.
Experiment 2: Cell Kinetic Analysis of MNHs Induced by Mitemcinal in Rats
All conditions were identical to those of a previous study in normal rats (Fujii et al. 2004). The labeling index, the percentage of mono-, bi-, and multinuclear hepatocytes, and the BrdU labeling of mono- and binuclear cells in the present study were compared with the data of the previous study.
Study Design
Seven female Sprague-Dawley rats were purchased from Japan SLC, Inc. (Shizuoka, Japan) and were six weeks of age at study initiation. The animal experiments were approved by the Committee for Ethical Treatment of Laboratory Animals of Chugai Pharmaceutical Co., Ltd.
BrdU labeling was carried out according to the method described by Fujii et al. (2004). Briefly, rats kept under ether anesthesia were implanted intraperitoneally with mini-osmotic pumps (Alzet Model 2002, Alza Co., Palo Alto, CA, USA) that were filled with 12% BrdU (Sigma Chemical Co., St. Louis, MO, USA) in distilled water containing 50% dimethyl sulfoxide (Sigma Chemical Co., St. Louis, MO, USA) and released at 240 μg of BrdU per hour for twenty-eight days. Two pumps per rat were implanted for the first fourteen days, and two more pumps were added for the remaining fourteen days.
The animals were given 300 mg/kg of mitemcinal. Mitemcinal was administered once daily for twenty-eight days starting on the day following the implantation of the osmotic pumps. All animals were euthanized by exsanguination from the abdominal aorta under ether anesthesia at the end of the administration period. At necropsy, livers were removed from all animals. The left lateral lobe was excised for perfusion with collagenase. The medial lobe was fixed in 10% neutral buffered formalin and embedded in paraffin for histopathological examination.
Cytocentrifuge Preparation of Isolated Hepatocytes
The isolation method used in the present study was similar to the two-step collagenase perfusion method. Briefly, the left lateral lobe was first perfused with a perfusion solution (Ca, Mg-free Hank’s solution; HBSS [Sigma] 9.5 g, HEPES [Sigma] 2.38 g, EGTA [Sigma] 0.19 g, NaHCO30.35 g per 1 L in distilled water, pH 7.4) and then a collagenase solution (HBSS 9.5 g, HEPES 2.38 g, Trypsin inhibitor [Sigma] 0.05 g, NaHCO30.35 g,CaCl2·2H20 0.74 g, collagenase [Sigma] 0.5 g per 1 L in distilled water, pH 7.4). After the collagenase perfusion, the cells were isolated and suspended in the perfusion solution. The cell suspension was then filtered through a cell strainer (100 μm, Becton Dickinson Labware, Franklin Lakes, NJ, USA), after which it was centrifuged for one minute at 60× G (twice) to isolate the hepatocytes. The hepatocyte suspension was adjusted to 4.0 × 104cells/mL and centrifuged onto glass slides by a cytocentrifuge (Cytospin® 2 Cytocentrifuge, Shandon Southern Products Ltd., Astmoor, Cheshire, UK) at 70 g for five minutes. The cytocentrifuge preparations were fixed in Carnoy’s solution and stored at –30°C. The slides stained with hematoxylin and eosin were examined under a light microscope to confirm that single cells were isolated and that the cells within the preparation were mostly hepatocytes. All analyses were performed on hepatocytes only.
Immunohistochemistry
Immunohistochemical staining was performed on the cytocentrifuge preparations using a mouse anti-BrdU monoclonal antibody/nuclease mixture (Amersham, Bucks, UK).
Cytocentrifuge preparations were air-dried and then fixed in 10% neutral buffered formalin at room temperature for five minutes. After washing in distilled water, the slides were incubated in 0.3% H2O2in methanol for thirty minutes to inactivate endogenous peroxidase activity. Following rinsing in distilled water and 0.05M Tris-buffered saline, pH7.6 (TBS), the specimens were pre-rinsed for ten minutes with 0.05M Tris-buffer, pH 7.6 (TB), for ten minutes at 37°C, and then digested in 0.1% trypsin in TB containing 0.1% CaCl2for thirty minutes at 37°C.
A labeled streptavidin–biotin method was applied after digestion of the slides. Following incubation with skim milk for thirty minutes to reduce nonspecific staining, slides were incubated successively in mouse anti-BrdU monoclonal antibody/nuclease mixture for 120 minutes at 37°C, in biotinylated anti-mouse IgG (Dako, Carpenteria, CA, USA) for forty-five minutes at room temperature, and in streptavidin-conjugated horseradish peroxidase (Dako) for forty-five minutes at room temperature. The slides were washed in TBS for fifteen minutes after each staining step. For visualization of the immunoreaction, the specimens were incubated in a medium containing 0.02% 3,3’-diaminoben-zidine tetrahydrochloride, 0.01M imidazole, and 0.005% H2O2in TB. For the final step, the slides were lightly stained with hematoxylin and eosin.
Analysis of BrdU Labeling Index of Nuclei, Nuclearity of Hepatocytes, and BrdU Labeling of Each Nuclearity Class
The number of BrdU-labeled nuclei, the number of mono-, bi-, and multinuclear (three or more nuclei) hepatocytes, and BrdU labeling of mono- and binuclear hepatocytes were counted using the Image Processor for Analytical Pathology System (IPAP-Win, Sumika Technoservice Cooperation, Osaka, Japan). A minimum of 500 hepatocytes per animal were counted, and the percentage of mono-, bi-, and multinuclear hepatocytes, the BrdU labeling index, and the percentage of BrdU labeling of mono- and binuclear hepatocytes were calculated.
Analysis of Nuclearity and BrdU Labeling of MNHs
The total number of MNHs in all slides of all animals was 955 cells. Hepatocytes with three, four, five, six, seven, eight, nine, and ten or more nuclei were counted, and the percentage was calculated for each type. The percentage of BrdU labeling was also determined for all-positive, all-negative, and mixed type nuclei.
Results
Experiment 1: Pathological Characterization of MNHs Induced by Mitemcinal
Table 2shows the incidence of MNHs in the preclinical studies with mitemcinal. MNHs were observed only in the rat and not in the dog. In the rat studies, MNHs were not observed with a single high dose of mitemcinal (acute study) but were observed in the high-dose groups of the subchronic study with reversibility after twenty-eight days of drug withdrawal. The incidence of MNHs increased in the chronic study relative to dose. With respect to the sex differences in the rat studies, a higher incidence was noted in females. Other findings in the liver, such as hypertrophy, vacuolation, and/or single cell necrosis of hepatocytes, were observed in both rats and dogs, but nuclear multiplicity was not found in any organs or tissues other than rat liver.
Figure 1shows the morphologic characteristics of the MNHs induced by mitemcinal. The cytoplasm of the MNHs was similar to that of surrounding normal and hypertrophied hepatocytes with light eosinophilic to granular, pale staining in the HE sections. The morphology of the nuclei in MNHs was similar to that of normal hepatocytes, oval-shaped with scattered chromatin, without any sign of atypia or mitotic figures. Concerning the structure of the MNHs, some cells had nuclei aligned linearly forming a palisading pattern, whereas others were aligned in a circle and showed a syncytium-like conformation. The morphological characteristics of the MNHs were similar in the sub-chronic, chronic, and carcinogenicity studies.
In female rats from the subchronic and chronic study, MNHs were preferentially observed in the centrilobular zone, with less in the mid-zone, and were rare in the periportal zone. In male rats, MNHs tended to be distributed in the mid-zone, but the lower incidence in male rats precluded precise evaluation.
Experiment 2: Cell Kinetic Analysis of MNHs Induced by Mitemcinal in Rats
Morphology of Hepatocytes and Percentage of Multinuclear Hepatocytes in Cytocentrifuge Preparations
Although the incidence of MNHs showed an increase in six of seven animals from histopathological examination in HE-stained sections, MNHs were observed in all animals from cytocentrifuge preparations. The hepatocytes in the cytocentrifuge preparations were well isolated, enabling the observation of the number of nuclei per hepatocyte (Figure 2). MNHs were readily distinguished, with a tendency to contain more than three nuclei per cell (Figure 2). The histopathological findings of Experiment 2 such as hypertrophy and single cell necrosis of hepatocytes were similar to those of the subchronic study.
The percentage of each cell type was also changed (Table 3). The percentage of MNHs was less than 1% of all hepatocytes in normal rats but was 4% in treated rats. The percentage of mononuclear hepatocytes was high (74.2%), whereas the percentage of binuclear hepatocytes was markedly low (21.8%) compared to normal rats.
Labeling Index and BrdU Labeling of Mono- and Binuclear Hepatocytes
The BrdU labeling index (LI) of nuclei in the rats treated with mitemcinal was markedly increased compared to normal animals (Table 4). The labeling percentages of mono-and binuclear hepatocytes are shown in Figure 3. The percentage of positive mononuclear hepatocytes was high compared to normal rats. With the binuclear hepatocytes, three patterns of labeling were observed; cells with two positive nuclei (WP), cells with two negative nuclei (WN), and cells with one positive and one negative nuclei (PN) (Figure 2A). The percentage of WP cells was high (14.5%), whereas that of WN cells was low (6.5%). The PN cells, which are difficult to detect in normal rats because of their low prevalence, were readily detected in the present study (0.5% of all hepatocytes).
Analysis of Nuclearity and BrdU Labeling of MNHs
The nuclearity of MNHs varied, with both odd and even numbers of nuclei (Figures 2and 4), and a variance of 3–40 nuclei per cell. The most frequent number of nuclei was three or four, with fewer cells of the higher nuclearity classes (Figure 4). The percentage of odd- and even-numbered cells was similar. Three patterns of BrdU labeling were observed (Figures 2and 5). The most frequently observed pattern was the mixed pattern (Figure 2C), followed by the all-positive pattern (Figure 2B). The all-positive pattern included cells with an odd number of nuclei (Figure 2B). The mixed-labeling pattern also included cells with an odd number of positive nuclei (data not shown). The all-negative pattern (Figure 2D) was observed at a lower frequency compared to the two other types (Figure 5).
Discussion
In the preclinical studies of mitemcinal, MNHs were observed only in the studies using rats (subchronic and chronic) and not in the studies with dogs. In the rat liver, MNHs were not induced by single high-dose (acute) administration, but the incidence increased with prolongation of the dosing duration (subchronic to chronic). Sex-dependent incidental difference was also noted, with a higher prevalence in females. Additionally, the carcinogenicity studies revealed that mitemcinal does not have hepatocarcinogenic potential in rodents (Fujii et al. 2007). The morphological characteristics revealed that mitemcinal-induced MNHs had mere multiplicity of nuclei without atypical findings in either the nuclear or cytoplasmic components. From the morphometric analysis, obvious zonal preference was noted in female rats, although the low incidence in males precluded precise evaluation in males.
In general, MNHs are known to be induced by several hepatocarcinogenic chemicals, such as colchicine (Miszurski and Doljanski 1949), thiourea (Rachmilewitz et al. 1950), aflatoxin B1(Svoboda et al. 1971), adriamycin (Cuccurullo et al. 1979), 2,3,7,8-tetrachlorodibenzo-p-dioxin (Kociba et al. 1976; Pohjanvirta et al. 1995), and chlordane (Malarkey et al. 1995). For these chemically induced MNHs, it is considered that the MNHs originate from single hepatocytes with disturbed cell division because the chemicals affect the nuclear components (Richey et al. 1977; Scampini et al. 1993), and mitosis, or polyploidization, considered to be caused by karyokinesis in the absence of cytokinesis (van Zwieten and Hollander 1997), is frequently found (Malarkey et al. 1995; Pohjanvirta et al. 1995; Rachmilewitz et al. 1950). On the other hand, several hepatocytic abnormalities found in infants such as giant cell hepatitis (Bird et al. 1963), fetal ascites (Fletcher et al. 1964), liver cell adenoma/carcinoma (McDougal and Gatzimos 1957; Roth and Duncan 1955; Wheeler et al. 1986), idiopathic iron storage (Goldfisher et al. 1981), and rubella (Plotkin et al. 1965) are known to induce MNHs. Since a wide variety of causative factors can induce MNHs in infant liver, it has been considered a nonspecific response to injury (Richey et al. 1977; Scampini et al. 1993; Wheeler et al. 1986). Taken together with the fact that the cell membrane and/or resembling structures were absent and the microorganelles in the cytoplasm of the MNHs observed under electron microscopy appeared normal (Elsner 1973; Ruebner and Miyai 1963; Schaffner and Popper 1963), it was concluded that the MNHs in infants was caused by fusion of immature regenerative hepatocytes (Elsner 1973; Richey et al. 1977; Ruebner and Miyai 1963; Scampini et al. 1993; Wheeler et al. 1986).
The characteristics of mitemcinal-induced MNHs were markedly different from those induced by hepatocellular oncogenic chemicals. Mitemcinal does not show oncogenic potential, and the MNHs induced did not show nuclear atypia.
Interestingly, there have been a few reports describing MNH-inducing chemicals without oncogenic potential. Particularly, rifabutin, an antibiotic of the rifamycin class, is well described as an inducer of MNHs without oncogenic potential, and the morphologic characteristics of its induced MNHs have been described in detail (Scampini et al. 1993). According to the report, (a) rifabutin induced MNHs only in rats, (b) the incidence of MNHs increased with subchronic and chronic exposure but decreased with lifelong exposure, (c) rifabutin did not show oncogenicity, (d) morphological atypia did not accompany multiplicity of nuclei of MNHs, (e) sex-dependent prevalence was observed (more prevalent in males), and (f) MNHs were evenly distributed in the lobules without specific zonal distribution. From these observations, with the exceptions of zonal distribution and the prevailing sex, the characteristics of MNHs induced by mitemcinal and rifabutin are quite similar. Furthermore, electron microscopy of MNHs induced by rifabutin confirmed the absence of cell membrane structures and normal microorganelles in the cytoplasm. Based on these observations, the authors of the report proposed the mechanism to be fusion of hepatocytes (Scampini et al. 1993).
Taking the accumulated knowledge into consideration, it is intriguing to hypothesize that mitemcinal and rifabutin belong to a chemical class that induces MNHs without oncogenic potential and that the mechanism for the induction of MNHs is the fusion of hepatocytes. For further investigation, since the electron microscopy results are rather circumstantial evidence, a cell kinetic approach was thought to provide more conclusive evidence. Therefore, we carried out a cell kinetic analysis to attempt clarification of the mechanisms related to the formation of MNHs induced by mitemcinal in rats using the method combining cytocentrifuge preparations with BrdU cumulative labeling.
For the cell kinetic analysis, female rats were selected because the incidence of MNHs tends to be higher in female animals. Cytocentrifuge samples were prepared from the livers of the female rats treated with mitemcinal for twenty-eight days. MNHs were readily distinguished in the cytocentrifuge slides, in which detailed analysis of individual cells was possible. Cytocentrifuge preparations consist of whole cells which, unlike paraffin sections, enable an accurate analysis of nuclearity. Although MNHs were detected in only six of seven animals in the paraffin sections, they were readily detected in seven of seven animals in the cytocentrifuge slides. BrdU labeling of the MNHs was also clearly detected in the cytocentrifuge slides. We have reported previously that cytocentrifuge preparations combined with BrdU cumulative labeling makes it possible to analyze normal hepatocytes in rats of various ages (Fujii et al. 2004). The present study indicates that the same method is applicable for analysis of the labeling index, nuclearity, and labeling patterns of drug-induced MNHs. Labeling of hepatocytes is a relatively well-used method for studying the changes in the ploidy of cells (Styles et al. 1993). However, there are no reports on the method being used to study the cell kinetics of drug-induced MNHs.
Nuclearity as well as the percentage of MNHs showed an increase compared to normal rats, a finding that was confirmed in the cytocentrifuge preparations. To date, detailed reports concerning the nuclearity of hepatocytes are not available. In the present study, we report that the nuclearity varied from three to forty nuclei per cell, indicating that the number of nuclei in an MNH is variable, with as many as forty nuclei per cell.
The BrdU labeling index was higher with animals administered mitemcinal compared to normal rats and was thought to reflect the regenerative response of hepatocytes to the single-cell necrosis observed in the paraffin-embedded slides. This response may also be the cause of the increase in the percentage of mononuclear hepatocytes. The ratio of mononuclear to binuclear hepatocytes, along with the ploidy classes, is known to shift in rat liver when regeneration or proliferation occurs (Seglen 1997). Several studies indicate that the type of shift is dependent on the cause (Seglen 1997). Although the ploidy was not considered in the present study, the sharp shift toward mononuclear cells was thought to be similar to shifts seen in regenerating rat liver after partial hepatectomy (Auvigne et al. 2002; Frederiks et al. 1990; Gerlyng et al. 1993; Melchiorri et al. 1993; Seglen 1997; Wheatley 1972), which is similar to shifts with certain drugs (Gerlyng et al. 1994; Seglen 1997).
Binuclear hepatocytes in normal rat liver are considered a part of a process toward polyploidization (Seglen 1997; Guidotti et al. 2003), in which a mononuclear cell will go through a cycle of acytokinetic mitosis, giving rise to a binuclear cell, and then move into a mitotic phase and give rise to two daughter cells with nuclei of a higher ploidy class (Guidotti et al. 2003). We have reported previously that almost all binuclear hepatocytes in normal rat liver are WP or WN cells after cumulative labeling with BrdU for one, two, and four weeks.
When administered continuously, BrdU is incorporated into the nuclei of cells that have gone through the S phase. This means that after one, two, and four weeks of cumulative labeling, positive cells have gone through the S phase during these periods. WP cells have two nuclei that have gone through the S phase, confirming that binuclear hepatocytes are mainly formed by acytokinetic mitosis (Fujii et al. 2004). Similarly, in the present study, WP binuclear cells were dominant, indicating that at least a part of the binuclear cells were formed by acyto-kinetic mitosis.
PN cells also exist in the normal rat liver, although they are very rare (Fujii et al. 2004). So-called “normal” binuclear hepatocytes are formed by acytokinetic mitosis. The presence of PN cells suggests that instead of one nucleus dividing into two nuclei, such cells have one nucleus that has undergone mitosis during the cumulative labeling period and one that has not. This finding was thought to strongly indicate that a fusion process is also present in the normal rat liver. A notable feature of the binuclear hepatocytes found in the mitemcinal-administered rats was the increase in PN cells. The increase indicates a shift toward the fusion process in rats administered mitemcinal.
With the MNHs in the present study, all-negative cells were observed. As MNHs with three or more nuclei are very rarely observed in normal rat liver, the cells noted in this study were thought to be formed during the mitemcinal treatment and BrdU labeling. The nuclei were negative, indicating they had not undergone mitosis. Therefore, a fusion process was thought to be involved in the formation. The most frequent type was the mixed type, with cells showing odd numbers of BrdU-positive nuclei, further indicating that a fusion process is present.
From the current results, including the data from the toxicity studies and the cell kinetic data, we have concluded that drug-induced disturbance of mitosis is not involved in the multinucleation of MNHs by mitemcinal.