2020 Volume 45 Issue 8 Pages 475-492
By analysis of the data from the Toxicogenomics Database (TG-GATEs), histidine decarboxylase gene (Hdc) was identified as largely and commonly upregulated by three fibrates, clofibrate, fenofibrate, and WY-14,643, which are known to induce hepatocellular hypertrophy and proliferation via stimulation of peroxisome proliferator-activated receptor α (PPARα) in rodents. As histamine has been reported to be involved in the proliferation of liver cells, the present study was conducted to focus on Hdc. Among other genes related to histidine and histamine, the expression of the gene of histamine ammonia lyase (Hal) was exclusively mobilized by the three fibrates. The expression of Hdc, which was usually very low in the liver, was increased with the repeated administration of fibrates, and concomitantly, the constitutive expression of Hal was suppressed. An interpretation is that the formation of urocanic acid from histidine under the normal condition switches to the formation of histamine. The mobilization of gene expression of Hdc and Hal by PPARα agonists could not be reproduced in primary cultured hepatocytes. The Hdc mRNA appeared to be translated to a protein which is processed differently from brain but similarly to gastric mucosa. Surprisingly, the fibrates caused hepatic hypertrophy but no induction of Hdc mRNA at all in mice. These results revealed that the changes in the histidine catabolism by PPARα agonists might be partially, but not directly, involved in the hepatocyte proliferation in rats, and there is a large genetic distance even between rat and mouse.
Peroxisome proliferator-activated receptor α (PPARα) agonists are the first-choice drugs for dyslipidemic patients with high triglyceride. Their repeated administration to rodents, however, certainly causes hepatic hypertrophy within a few days and thereafter hepatic tumor in a high incidence following long-term administration. This toxicity has been considered rodent-specific and therefore considered to be no risk in humans (Desvergne and Wahli, 1999; Han et al., 2017). Hepatic hypertrophy as well as tumor has been attributed to PPARα by analysis of knockout mice (Lee et al., 1995) and thus the species difference occurs in the reaction after the step of PPARα activation, but its detailed mechanism is unknown, although the upregulation of c-Myc has been suggested to be one of the key causal genes (Gonzalez and Shah, 2008).
Our analysis of the data in the Toxicogenomics Database (TG-GATEs) in the previous study (Mizukawa et al., 2020) proposed a list of genes that are reproducibly and markedly upregulated by PPARα agonists. We suggested in the study that a) some important changes may occur before the elevation of c-Myc, b) various genes, which are scarcely related to lipid metabolism, are upregulated together with that related to lipid metabolism and to cell proliferation, c) many of the genes unrelated to lipid metabolism are not induced in the primary cultured hepatocytes, and d) the gene list might contain genes that are involved in the rodent-specific hepatotoxicity. Among them, we pay attention to histidine decarboxylase gene (Hdc), which was highly upregulated by PPARα agonists and is physiologically interesting.
Histidine decarboxylase (HDC) is the only enzyme synthesizing histamine, and is usually quite low in adult tissues except mast cells, gastric ECL cells, or basophils, where HDC is constitutively expressed and histamine is stored in the secretory vesicles (Ichikawa et al., 2010). HDC is enriched in some developing organs, and the liver is the most enriched organ in the rat fetus (Taguchi et al., 1984). HDC activity is markedly induced in the liver, lung, or spleen under pathophysiological conditions where various cytokines are involved (Endo et al., 1986). Histamine produced by induced HDC is reported to play a role in cell proliferation, such as in vascular smooth muscle cells during restenosis of coronary artery (Fang et al., 2005) and various cancer types (Blaya et al., 2010; Medina and Rivera, 2010). It is also reported that HDC is induced in the liver by partial hepatectomy (Ishikawa et al., 1970; Chang et al., 2010), which markedly stimulates hepatic cell proliferation. Consequently, we consider it to be highly possible that induction of Hdc is involved in hepatic hypertrophy caused by PPARα agonists.
Transcriptome data of 132 chemicals, body weight and liver weight changes were obtained from the database, TG-GATEs (http://toxico.nibiohn.go.jp/) (Urushidani, 2008). The doses, the vehicles, and the abbreviations for the 132 drugs analyzed in the present study are summarized in Table 1. Most of the drugs were dissolved or suspended in 0.5% methylcellulose (MC) or corn oil and administered orally. In some cases indicated as “saline” in the column for vehicle, the drug was dosed by intravenous injection. Ethanol was diluted with distilled water. The highest dose of each drug for repeated administration was first determined by a small-scale dose finding test (repeated administration for 1 week) such that the expected survival of the animals until the end of 28 days of repeated administration was set as the first priority. In general, the highest dose was reduced by the factor of the square root of 10, to middle and to low dose, and the same dose level was employed for the single dosing. When the toxicity was expected to be highly increased in repeated administration in advance, like in anti-inflammatory, anti-cancer, or immunosuppressive drugs, higher dose levels were employed for single dose experiments such that at least one dose level was common with the repeated dosing. In some cases, because of the death or early sacrifice when moribund, the animals for high dose at 28 days were lacking and this is indicated with parentheses in Table 1. When the high dose data of these drugs were presented together with the other drugs, middle dose data were used instead. During the project, it sometimes occurred that gene expression changes in single dose were so small that even the highest dose showed almost the same results as control vehicle, and then additional experiments were performed employing higher dose(s). In two cases (CPZ and ADP) among them, the low dose was omitted and indicated as NA in the table. They are basically aligned as the order of the project number but the drugs with high agonistic potency of PPARα are dispersed in order to avoid the overlapping of the symbols on the 3D-graph.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/45_475-t1.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
The concentrations of the drugs for primary cultured hepatocytes were independently determined by a pilot test for cellular toxicity. The highest concentration was set to 10–20% of the lethal concentration as estimated by lactate dehydrogenase leakage over 24 hr. When the cells could tolerate as much as 10 mM or the level equal to the solubility limit of the compound in DMSO (allowed to add up to 0.1% in the final concentration), the highest concentration was set to either value. In general, the maximum concentration (high) was decreased by a factor of 5 to middle and low concentration (Table 1).
The precise protocols for TG-GATEs are also available from the same web page. In general, for in vivo protocol, 7-week-old male Sprague Dawley (SD) rats were treated with three doses of each test drug (low, middle, high, and vehicle: 5 rats each) and sacrificed 3, 6, 9, and 24 hr after single oral dosing as well as 24 hr after repeated dosing for 3, 7, 14, and 28 days. Toxicology data were obtained from 5 animals while gene expression analysis by Affymetrix GeneChip (Rat Expression Array 230 2.0) was performed for 3 out of 5. As for in vitro protocol, isolated hepatocytes from 7 weeks SD rats according to the standard method were cultivated with a collagen coated six-well plate for 24 hr, and 2, 8, and 24 hr after the addition of the test drug (low, middle, high, and vehicle: 2 wells each) and the cells were harvested and analyzed for gene expression (Tamura et al., 2006).
The digital image files were processed by Affymetrix Microarray Suite version 5.0 and the intensities were normalized for each chip by setting the mean intensity to 500 (per chip normalization).
Additional experiments in the present studyFor the data newly added by the present study, male 6-week-old SD rats and C57BL/6N mice were purchased from Shimizu Laboratory Supplies Co. and used after one week of acclimatization. The experimental protocol was approved by The Animal Research Committee in Doshisha Women’s College of Liberal Arts.
In order to see the reproducibility of the data in TG-GATEs, rats (N=3) received 1000 mg/kg of FFB or 30 mg/kg WY p.o. for 3 days and were sacrificed 24 hr after the last dose by exsanguination under isoflurane anesthesia. The control rats received vehicle (0.5% MC for FFB and corn oil for WY). For mice, 100 mg/kg of FFB were administered for 3 and 7 days (N=4 each), or 100 mg/kg of WY for 3 days (N=3), and the mice were sacrificed by exsanguination under isoflurane anesthesia. For real-time PCR, a part of the liver was soaked in RNA later and the remaining part was frozen by liquid nitrogen for protein analysis.
The sample for PCR was homogenized and the total RNA was extracted using NucleoSpinRNA kit (Takara Bio, Shiga, Japan). Reverse transcription reaction was performed using PrimeScript RT reagent kit with gDNAeraser (Takara Bio) to obtain cDNA and real-time PCR was done with SYBR premix Ex Taq II (Takara Bio) using Roter-Gene Q (Qiagen, Osaka, Japan). The primers used were purchased from Takara Bio; RA058783 for rat Hdc and MA131065 for mouse Hdc. For a reference gene, β-actin (RA015375 for rat and MA050368 for mouse) was employed and the results were expressed as the ratio to β-actin.
The sample for protein analysis was homogenized in 4 volumes of a homogenizing buffer (113 mM mannitol, 37 mM sucrose, 5 mM PIPES-Tris, 0.4 mM EDTA, pH=6.7, supplemented with protein inhibitor cocktail) with a sonicator (Vibra Cell, Sonics & Materials Inc., Newtown, CT, USA) for 30 sec x 4 on ice, and centrifuged at 500 x g for 10 min to remove debris. The supernatant was centrifuged at 100,000 x g for 1 hr (Optima TLX, Beckman Coulter, Tokyo, Japan), and the supernatant was taken as a cytosol fraction and the pellet was suspended in the same buffer and taken as a membrane fraction. For the standard of HDC, gastric mucosa and whole brain were taken from the rat and mouse, and subjected to the same protocol as above to get cytosol and membrane fraction.
Protein samples were solubilized in a buffer (2% SDS, 65 mM dithiothreitol, 0.25 mM EDTA, 10% glycerol, 15 mM Tris, pH=6.8) and subjected to 10% SDS-polyacrylamide electrophoresis according to standard protocol. After the electrophoresis, the proteins were transferred to PVDF membrane, blocked with 5% skim milk in 0.5% Tween 20-PBS (TPBS), and incubated with 1/100 diluted primary antibody (rabbit polyclonal anti-HDC) in blocking solution at 4°C, overnight. The membranes were washed 3 times with TPBS and incubated with 1/10,000 diluted secondary antibody (Donkey polyclonal HRP-anti-rabbit IgG, Sigma Aldrich Japan, Tokyo, Japan) in TPBS at room temperature for 2 hr. After 3 times washing with TPBS secondary antibody was visualized by Western Lightning ECL Pro (Perkin-Elmer Japan, Yokohama, Japan) and observed by an illuminator (C-DiGit, LI-COR). We used two commercially available antibodies recognizing both rat and mouse HDC. One (Ab-A) was from Abcam plc (Cambridge, UK) and it was against near the C terminal 20 amino acids (620 - 639) of HDC (662 amino acids in total). The other (Ab-B) was from Atlas Antibodies (Bromma, Sweden) and it was raised against near the N-terminal 132 amino acids (89 - 220) of HDC.
Another pair of rats receiving 30 mg/kg WY or its vehicle control for 3 days were subjected to histological analysis. The rats were perfused via left ventricle with PBS followed by 1% paraformaldehyde under isoflurane anesthesia. The fixed liver was subjected to paraffin sections (Kyoto Institute of Nutrition & Pathology, Inc., Kyoto, Japan). The sections were deparaffinized and rehydrated. In the preliminary experiments, it was found that the antigenicity to both Ab-A and Ab-B was lost by the fixation and embedding, so antigen retrieval was performed by heating the sections soaked in 10 mM citrate buffer (pH=6.0) for 12 min by a microwave using a pressure cooker. The sections were blocked with 5% skim milk in TPBS for 1 hr at room temperature, they were incubated with 1/50 dilution in the blocking solution of Ab-A at 4°C overnight, washed 3 times with TPBS and then incubated with 1/100 dilution in TPBS of Alexa Fluor 488 anti-rabbit IgG (Sigma Aldrich Japan) at room temperature for 2 hr. Immunofluorescence was observed by using EOS FL (Thermo Fischer Scientific, Tokyo, Japan) with constant settings throughout the observation of the sections.
Data presentation and statistical analysisData are presented as means + SEM. When one pair was compared, Student’s t-test was employed. When multiple comparison against one control was made, Dunnett’s test was used. A difference of p < 0.05 was considered statistically significant.
From the database, TG-GATEs, we extracted genes that were commonly and largely upregulated by three fibrates, CFB, FFB, and WY, at 24 hr after the single dosing, and sorted them by their induction rates (Mizukawa et al., 2020). Of these, the 4th ranking was Hdc mRNA detected by the probe set, 1370491_a_at. Because of the scarce expression in the vehicle control, the induction rates were so high that 67.6, 74.1, and 54.7 fold by CFB, FFB, and WY, respectively. Figs. 1A-C show upregulation of Hdc mRNA by these three agonists at all doses and time points. It is obvious from the figures that very low expression of Hdc was observed at any points in the liver of control rats, suggesting no circadian or feeding-related changes occurred. In fact, all the data from control rats showed “absent call” by GeneChip. In the groups receiving PPARα agonists, upregulation of Hdc expression increased dose- and time-dependently after single administrations. In the case of CFB, further increase did not occur by repeated administrations, whereas the expression kept increasing by repeated dosing both in FFB and WY.
In order to see the relationship between Hdc induction and hypertrophy of the liver, data of body and liver weight were harvested from the database and the relative liver weight per body weight (%) was calculated as shown in Figs. 1D-F. The enlargement of the liver started as early as 24 hr after a single dose and became obvious after 3 successive doses or more in all three fibrates. These changes are roughly in parallel with that of Hdc expression.
Data of primary cultured hepatocytes are also available to compare with in vivo data. However, all the expression values in both control and treated groups were very low (< 70) with the judgment of “absent call” (data not shown). It is obvious that fibrates lose their ability to induce Hdc mRNA in the cultured hepatocytes.
Effects of PPARα agonists on the expression of the genes related to histidine and histamineThe following enzymes are involved in the metabolism of histamine or its precursor, histidine (Chang et al., 2010), and their data were extracted from the database.
a) D-ribose 5-phosphate is phosphorylated by phosphoribosyl pyrophosphate synthetase 2 to form 5-phosphoribosyl-1-pyrophosphate. The mRNA of this gene (Prps2) can be detected by the probe sets 1368551_at, 1375932_at, and 1398262_at in GeneChip.
b) Phosphoribosyl pyrophosphate aminotransferase transfers α-amino group of glutamine to 5-phosphoribosyl-1-pyrophosphate to form histidine. The mRNA of this gene (Ppat) can be detected by 1369785_at.
c) Histidine is broken down into urocanic acid by histidine ammonia lyase (HAL). The mRNA of this gene (Hal) can be detected by 1387307_at.
d) Histidine is alternatively turned into histamine by HDC. The mRNA of Hdc is detected by 1370491_a_at.
e) Histamine is inactivated by histamine N-methyltransferase (HNMT) and diamine oxidase (DAO). The expression of HNMT gene (Hnmt) can be detected by 1385762_at and 1387382_at., and that of DAO gene (Dao) can be detected by 1369491_at.
The first probe set (1368551_at) for Prps2 appeared to be low efficient and all the data showed “absent call”. The latter two sets (1375932_at and 1398262_at) detected the expression but no drug-related changes were observed (data not shown). The expression of Ppat detected by 1369785_at showed circadian changes in the control group receiving vehicle, but the fibrates failed to affect them (data not shown).
Figures 2A-C show the expression changes of Hal by the three fibrates. Although not obvious in the single dosing, its basal expression was markedly suppressed by the repeated administration of the fibrates for 3 days and longer. The changes in Hal expression were not observed for in vitro either, same as the case of Hdc (data now shown).
In contrast to histidine metabolism, histamine metabolism did not appear to be affected by fibrates. The expression of Hnmt showed a very low value with “absent call” at every dose and time point by both probe sets. The expression of Dao was found to be quite high reflecting the fact that the liver is the main organ for deaminating histamine (Blaya et al., 2010), but it was not affected by the administration of fibrates at all. These were observed in vitro as well as in vitro (data not shown).
As for other histamine-related genes, probe sets are designed for four types of receptor, i.e., 1369855_at and 1370338_at for H1, 1369914_at for H2, 1368451_at and 1388080_a_at for H3, and 1387590_at for H4. However, they were all very small with “absent call” at every dose and time point.
Specificity of Hdc expression changesIt is known that the expression change of a certain gene affects the expression of its neighbor gene on the genome with various mechanisms. According to comprehensive analysis of the interrelationship among neighboring gene expression (Sémon and Duret, 2006), significant interaction is mostly limited to two genes next to each other. We then examined the genes right next to Hdc (Fig 3). Rat Hdc locates in chromosome 3, and GA binding protein transcription factor, beta subunit 1 (Gabpb1) and ATPase phospholipid transporting 8B4 (Atp8b4) exist immediately upstream and downstream, respectively, on the same strand. Slc27a2 is the closest to Hdc and ubiquitin specific peptidase 8 (Usp8) locates in the position of so-called “promoter upstream transcription” (Lloret-Llinares et al., 2016). Of these, expression of neither Gabpb1 nor Usp8 was changed by the fibrates at any dose or time point (data not shown). Atp8b4 was judged as “absent call” by GeneChip at most of the points and no consistent results were obtained. Fig. 3 also shows the genomic location in mouse genes and this is used in the discussion.
As shown in Figs. 4A-C, Slc27a2 was the only gene among them upregulated by the fibrates. SLC27A2 is not a simple transporter but converts free long-chain fatty acids into fatty acyl-CoA esters, which thereby plays a key role in lipid metabolism and is actually known to be induced by PPARα agonists (Rakhshandehroo et al., 2010). This gene was highly expressed in the control and the three agonists further increased throughout the repeated dosing period.
Effects of the fibrates on the expression of Slc27a2 were also confirmed in vitro. As shown in Fig. 5, no effect of the fibrates was observed until 6 hr after their exposure, whereas the induction effect of the drug emerged at 24 hr as the basal expression in the control cells decreased with time, resulting in a 2.5 to 3 fold increase.
In order to examine if the induction of Hdc is specific for PPARα stimulation, expression of Hdc was examined for 132 chemicals available in the database and shown in Fig. 6A, where the data were narrowed down to control and high dose (the data of middle dose was employed when lethal cases occurred at high dose) at 24 hr after single dosing and repeated dosing for 3, 7, 14 and 28 days, and the name of drugs that induced more than 300 at any time point are indicated on the graph. In addition, expression of acyl-CoA thioesterase 1(Acot1), which is one of the most inducible genes by PPARα agonist in rats, is shown in Fig. 6B in the same manner as for Hdc except the drugs that induced more than 1,000 at any time points were indicated. The bar graphs for Hdc and Acot1 are quite similar for the specified drugs. It is obvious that four fibrates in the database, i.e., CFB, FFB, WY and GFZ, all showed marked induction of both Hdc and Acot1. Furthermore, BBr, AM, SST, and BZD, which were reported to possess PPARα agonistic activity (Kunishima et al., 2007; Seo et al., 2008) induced Hdc as well as Acot1. Non-steroidal anti-inflammatory drugs, which are known to contain PPARα agonists (Lehmann et al., 1997), ASA, IBU, MLX, BDZ, and LNX upregulated Hdc, whereas ASA, BDZ, and NPAA induced Acot1 gene, but others such as DFNa, IM, NPX, MEF, SUL, and NIM did not affect either gene expression. We could not find supporting reports for CMP, EBU, and TBF as PPARα agonists, but they appeared to stimulate PPARα since they clearly stimulated both Acot1 and Hdc. Repeated dosing of MP, CCl4, or ETH, and single dosing of AAF, which all appear to be PPARα agonists in respect to the induction of Acot1, failed to induce Hdc, whereas repeated dosing of HCB, MDP, and DOX induced Hdc but not Acot1. In summary, the drugs stimulating Acot1 but not Hdc expression were NPAA, MP, CCL4, ETH, and AAF, and that stimulating Hdc but not Acot1 expression were IBU, MLX, LNX, HCB, MDP, and DOX. A simple explanation would be that the former drugs are PPARα agonists with a direct inhibitory activity on Hdc expression, and the latter drugs have an activity of inducing Hdc expression independent on PPAR. Although there are some exceptions as above, Figs. 6A and 6B are quite similar as a whole, indicating that both Hdc and Acot1 are induced by the same mechanism, i.e., stimulation of PPARα.
Next, expression of Hal, which was downregulated by the three fibrates in vivo, was examined for 132 chemicals. As the inhibitory effect is difficult to see in a graphic presentation like Fig. 6, the results are summarized as Table 2, where the values of ratio to control of high dose (middle dose was employed when lethal cases occurred) at 24 hr after single dosing and repeated dosing for 3, 7, 14 and 28 days are sorted by the value of 28 days. In order to see possible relationship between Hal expression and hepatic hypertrophy, the data of relative liver weight per body weight were extracted from the toxicological reports in the database, and the point where the relative liver weight significantly increased is shaded, and where it significant decreased is black-and-white inverted, in the corresponding column of Table 2. For the 20 drugs with Hdc-inducing activity named in Fig. 6A, the corresponding column for the drug name is shaded. Of these 20 drugs, 16 are ranked in the upper half of the list, suggesting a certain relationship between Hdc induction and Hal inhibition. Although most of the drugs that induced Hdc inhibited Hal, there are many drugs inhibiting Hal without inducing Hdc. Hepatic hypertrophy tends to occur where Hal expression is low (in the higher rank of Table 2), while atrophy tends to occur where it is unchanged or high, but there are too many exceptions.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/45_475-t2.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
The expression changes of Hal in vitro were reviewed as well. As stated above, the fibrates showed no effect on Hal expression up to 24 hr treatment. Among 132 drugs with any effect on Hal expression in vitro, the largest decrease was observed for COL (Fig. 7A) and the largest increase was by DOX (Fig 7B). In Table 2, COL also showed a strong inhibition at 24 hr in vivo, whereas DOX did not show any effect on Hal expression, showing an apparent discrepancy between in vivo and vitro. In order to see the correlation between in vivo and vitro, values of ratio to control at 24 hr after high dose are shown as a scatter plot (Fig 7C). It is obvious from the figure that there is no correlation between in vivo and in vitro. Effects of the expression changes in the neighboring genes could be excluded since there are no PPARα–responsive genes around Hal on the rat genome.
Figures 8A, B show the confirmation of data from the database by real-time PCR. As the sensitivity of PCR is higher than GeneChip, Hdc expression was low but detectable in the control rats. Both FFB 1000 mg/kg (A) and WY 30 mg/kg (B) were administered for 3 days and they markedly induced Hdc expression. As the control values were close to the detection limit, the value of ratio to control was relatively imprecise, but the value ca. 30 fold is comparable to the one in the database, 50 to 70 fold.
Protein expression of HDC was examined by western blotting using antibodies with different epitopes. It is known that HDC is translated as a 74 kDa precursor protein and post-translationally cleaved to a 54 kDa species, which forms a 110 kDa dimer (Ichikawa et al., 2010). In our preliminary experiments, the optimal condition was determined for the antibodies obtained from Abcam (Ab-A) and Atras (Ab-B), using samples from rat brain and gastric mucosa, known to be enriched in HDC. HDC protein appeared to be labile as reported (Ichikawa et al., 2010) especially in the solubilized condition, and it was quite difficult to obtain reproducible results regarding its fragmentation. The most reliable results were as follows. Ab-A recognized a 74 kDa band in the brain membrane fraction and 32 kDa band in the gastric cytosol fraction. Ab-B recognized a faint 74 kDa and a 45 kDa band in the brain membrane fraction and a 110 kDa band in the brain cytosol. Ab-B also recognized a faint 74 kDa and 65 kDa bands in the gastric membrane fraction but nothing in the gastric cytosol fraction. These are in accordance with the proposed mechanism that the full length (74 kDa) of HDC is cleaved by caspase 9 into 54kDa active enzyme and 20 kDa C-terminal fragment (Tanaka et al., 1998; Furuta et al., 2007), and this is consistent with the epitopes of the antibodies, i.e., C-terminal for Ab-A and N-terminal for Ab-B, considering the inaccuracy of the molecular weight on SDS-PAGE. This indicated that the post-translational processing of HDC is different in tissues. Figs. 9A, B show the western blotting analysis of the membrane and cytosol fractions of rat liver using Ab-A as primary antibody. In the cytosol fractions from FFB-treated rats (Fig. 9A), clear 32 kDa bands, which are usually present in the gastric cytosol, are visible, and these bands are absent from the control. When the sensitivity of the detector increased, this band became visible in the membrane fraction of FFB-treated, but not in the control liver. This band could not be detected in the membrane fraction of gastric mucosa, in contrast to cytosolic fraction (Fig. 9B). Ab-B also visualized bands in the position similar to that of gastric mucosa exclusively in FFB-treated rats (data now shown).
Figures 10A, B show the immunostaining of rat liver sections. The cytosol of the hepatocytes from WY-treated rats showed positive staining by Ab-A compared with control rats. No characteristic staining of vascular endothelium, bile ducts, or Kupffer cells were seen.
From the results described above, it was revealed that fibrates strongly induced Hdc mRNA that was translated to HDC protein in the rat liver. One important question is whether HDC or histamine is involved in the pharmacology and/or toxicology of PPARα agonists. In order to investigate this point, the most powerful strategy would be the use of HDC-, or histamine receptor-knockout mice that are now available (Ichikawa et al., 2010; Ohtsu et al., 2001; Masaki et al., 2005). Therefore, we examined if the same phenomena in rats occur also in mice, which have been considered to be the same in respect to the pharmacology and toxicology of fibrates, such as hypolipidemic effects and hepatic hypertrophy leading to tumor (Desvergne and Wahli, 1999).
Looking at the rat data in Fig. 1E, increase in the relative liver weight per body weight by FFB 100 mg/kg for 3 days was almost the same extent as that by 1000 mg/kg. Because of the high viscosity of the suspension for 1000 mg/kg, the physical burden of this dose was considered to be too heavy for mice. We then chose 100 mg/kg of FFB for the mouse study and administered it for 3 and 7 days. As was in the rats, the expression of Hdc was low but detectable in the liver of control mice by PCR. Surprisingly, Hdc was never induced by FFB at all in spite of the fact that the hypertrophy started after 3 days and became marked after 7 days (Figs. 11A, B). Considering the possibility of insufficiency of FFB dosage, we employed 100 mg/kg of WY, which was higher than the dose showing the maximal effect on both Hdc induction and hepatic hypertrophy in rats. As shown in Fig. 11C, D, however, WY did not elicit any inducing effects on Hdc, whereas the relative liver weight per body weight was weakly, but significantly increased. As expected, HDC proteins could not be detected by western blotting in the liver of the mice treated with FFB or WY (data not shown).
In the normal condition, the expression of Hdc was very low with “absent call” by GeneChip and barely detectable by real-time PCR in matured rat liver, and thus the induction rate by the fibrates was observed to be quite large like several tens to hundred fold as in other tissues where Hdc is pathophysiologically induced (Ichikawa et al., 2010; Kikuchi et al., 1997).
Overviewing other genes involving histidine and/or histamine, Hal was the only gene of which expression was affected by fibrates. The induction of Hdc was obvious 24 hr after a single dosing of fibrates, whereas the downregulation of Hal emerged after 3 days of administration. Histidine is converted to histamine by HDC or to urocanic acid by HAL. The present results could be interpreted that HDC is induced to produce histamine resulting in the reduction of histidine and then the pathway to produce urocanic acid is suppressed to preserve the former pathway, namely, the switching of histidine metabolism occurs.
All the subtypes of histamine receptor (H1 to H4) could not be quantified by GeneChip irrespective of stimulation by fibrates. This does not necessarily mean that all the receptors for histamine are absent in the liver, since usually the number of the receptor is small and the turnover rate is low compared with enzymes, resulting in the small amount of mRNA. In fact, α1 and β2 receptors are known to exist and play important roles in the liver (Aggerbeck et al., 1983; Ghosh et al., 2012), but mostly their mRNAs were judged as “absent call” by GeneChip (data not shown). Another possibility is a putative intracellular histamine binding site that was reported to be involved in the DNA synthesis stimulated by partial hepatectomy (Brandes et al., 1992). However, this protein has not been identified (Blaya et al., 2010) and could not be verified at present.
If the expression of Hdc gene has a physiological significance, it should be translated to protein and this was confirmed by western blotting. HDC protein has a size of 74 kDa in full length and is known to be processed after translation (Ichikawa et al., 2010; Tanaka et al., 1998; Furuta et al., 2007). The 74 kDa precursor is labile and easily destroyed by ubiquitin-proteasome, while its C-terminal fragment is cleaved off by caspase 9 to make the 54 kDa active form on the cytoplasmic reticulum membrane. This was known from an analysis by the transfection of fragments into cell lines, but not in the physiological condition. In the present study, the interpretation of the results was difficult because HDC protein was quite unstable especially under the solubilized condition such that unreproducible bands appeared in the western blotting using two types of commercially available antibodies. According to the manufacturers’ data sheets, antibody from Abcam (Ab-A) recognizes C-terminal and that from Atlas (Ab-B) recognizes N-terminal portion of HDC. Repeated experiments using freshly prepared samples to get relatively reproducible results indicated that a 30 kDa fragment was detected by Ab-A in the cytosol and 74 (faint) and 65 kDa bands were detected by Ab-B in the membrane fraction of gastric mucosa. As the apparent difference on the SDS-PAGE was within a reasonable range, it could be suggested that the processing of HDC in the rat gastric mucosa occurs according to the previous reports (Ichikawa et al., 2010; Tanaka et al., 1998; Furuta et al., 2007), i.e., the full length form and its activated form are on the membrane leaving the C-terminal fragment in the cytosol. On the other hand, the pattern was different in the brain samples. Ab-A only recognized a 74 kDa band in the membrane fraction, whereas Ab-B recognized 75 (faint) and 45 kDa bands in the membrane fraction and a 110 kDa band in the cytosol of rat brain. It appears that the processing of HDC is different in the brain and it is difficult to interpret the results by Ab-B, and we did not elucidate this issue further in the present study. In any event, HDC induced in the liver by fibrates appears to be processed as in the gastric mucosa except for the fact that a small amount of C-terminal fragment is remained in the membrane fraction.
It is known that hepatomegaly caused by PPARα agonists is due to the increase in the cell size associated with enzyme induction as well as that in cell proliferation (Maronpot et al., 2010). If HDC or its product, histamine, is involved in hepatomegaly, it could play a role in cell proliferation, since it is unlikely that HDC protein is the main part of induced enzymes, and/or histamine induces many enzymes in hepatocytes. In fact, many genes related to cell proliferation were found to be upregulated by fibrates at the time when Hdc was induced (Tamura et al., 2006; Mizukawa et al., 2020).
HDC is practically absent in the adult rat liver, but highly expressed in the liver of fetuses (Taguchi et al., 1984) and considered to be involved in cell proliferation during development or regeneration like after partial hepatectomy (Ishikawa et al., 1970). Chang et al. (2010) investigated expression changes in the gene related to histidine catabolism in the proliferating liver using a partial hepatectomy model of rat. They showed that expression of Hdc was induced in hepatocytes, vascular endothelial cells and bile duct epithelia, and the expression of Hal was increased in Kupffer cells under the increased hepatic proliferation by hepatectomy. They suggested that histamine produced by the former pathway stimulates sinusoidal endothelial cells proliferation,and urocanic acid produced by the latter acts on Kupffer cells itself and dendritic cells to generate immune suppression by autocrine and paracrine modes. In the present study, the upregulation of Hdc was quite obvious whereas the expression of Hal was rather decreased. This could be due to the difference that the data from the database are the measurements of whole liver, while the measurements by Chang et al. (2010) were done on the separated cell types. In fact, expression of Hal in the hepatocyte fraction was markedly decreased 3 days after the hepatectomy in their data.
Considering the reports of hepatectomy described above, it could be postulated that compensatory cell proliferation due to primary tissue damage by fibrates leads to the activation of Hdc expression. However, upregulation of Hdc expression was almost limited to the drugs having PPARα stimulation, whereas there were drugs causing cell damage followed by hepatic hypertrophy without inducing Hdc expression. For example, TAA or CMA, which caused severe necrosis (Uehara et al., 2008a, 2008b) and subsequent increase in relative liver weight per body weight (Table 2) showed no upregulation of Hdc. It is thus suggested that the expression of Hdc associated with cell proliferation is not the result of tissue damage, but is directly related to PPARα stimulation. Mitogens are reported to induce HDC in the liver (Endo, 1983) although this is only in the study of mice. Taken together, it is concluded that Hdc induction appeared to be restricted to the direct stimulation of cell proliferation by mitogens, hepatectomy, or PPARα stimulation. The expression of Hal did not appear to be closely related to hepatic hypertrophy in general, but it seemed to be self-regulation to compensate for the reduction of histidine in the hepatocyte under PPARα stimulation.
In order to investigate the direct relation between the Hdc expression and cell proliferation induced by PPARα agonists, primary cultured hepatocytes should be a useful tool. However, the effects of PPARα agonists could not be reproduced in vitro at all. As previously reported (Tamura et al., 2006; Mizukawa et al., 2020), the genes extensively upregulated by fibrates contained a large number of the genes that did not show any expression changes in vitro. This cannot be explained by fast metabolic elimination of fibrates in vitro, since metabolic enzymes in vitro are rapidly decreased (Berry et al., 1997) and found to be less stimulated by fibrates compared with that in vivo (Tamura et al., 2006). Moreover, if the difference is due to pharmacokinetics, the difference should be even for all of the genes.
In general, in vivo specific responses are attributed to the change in the non-parenchymal cells (Qu et al., 2010), most of which are removed from the culture. However, in the case of Hdc, it was found to be expressed in the parenchymal cells by immunostaining. Therefore, it is necessary to consider the involvement of other factors, such as autonomic nerves, endocrine/paracrine system, or something from food, to explain this large difference.
Another possibility is that the gene expression is stimulated by separation and/or cultivation of the hepatocytes such that the stimulatory effects by drugs are masked. This type of phenomenon was observed in the case of cell proliferation-related genes in the previous study (Mizukawa et al., 2020). In the present study, Slc27a2 expression was stimulated by fibrates as early as 3 hr after administration in vivo (Fig. 4), whereas their effects were only noticeable after 24 hr in vitro (Fig. 5). Looking at the control values, they decreased with time of culture such that the stimulatory effects emerged at 24 hr. However, this was not the case for Hdc, since its expression was kept very low throughout the culture time.
After the cultured cell system was found to be inappropriate, we were obliged to employ another strategy. There are various gene-targeting mice available including Hdc-knockout (Ichikawa et al., 2010; Ohtsu et al., 2001), histamine receptor-knockout (Ichikawa et al., 2010; Masaki et al., 2005), or PPARα-knockout mice (Lee et al., 1995), and the combination of these should be a powerful tool. Before using these mice, we administered FFB or WY to wild-type mice, but unexpectedly, no induction of Hdc occurred. As HDC activity is reported to be induced by mitogens in mice (Endo, 1983), proliferation-related induction of Hdc should also exist in mice. Furthermore, both the rat and the mouse, as rodents, have been commonly used in pharmacological as well as toxicological research (Desvergne and Wahli, 1999). This inconceivable result cannot be explained by the difference in the metabolic enzymes for fibrates because the change of the drug concentration is expected to affect all of the genes evenly.
It is well known that the expression of a certain gene affects that of its neighboring gene, and in most cases, the nearest pair genes interact with each other (Sémon and Duret, 2006). Among them, “PROMPT”, where the promoter regions of two neighboring genes exist in a head-to-head position in the complementary strands and both genes are transcribed at the same time, has been extensively studied (Lloret-Llinares et al., 2016; Mizukawa et al., 2020). In the case of rat Hdc, the nearest one is PPARα-inducible Slc27a2 (Rakhshandehroo et al., 2010), which locates not the position of PROMPT but on the tail-to-tail position on the complementary strand. It could be postulated that during the induction of Slc27a2, which is the primary target gene of PPARα agonists, the expression of Hdc is concomitantly stimulated by e.g., histone acetylation (Asano et al., 2010) or another mechanism. However, this cannot explain the species difference. Fig. 3 shows the genomic structure around Hdc of rat and mouse. Although some additional unidentified genes are reported in the mouse genome, their relative position is highly conserved in both species. It was impossible in the present study to elucidate the mechanism of Hdc expression upregulated by PPARα agonists in rat liver, but the present findings might inspire future study of the modulation of gene expression by PPARα.
We focus on Hdc as a candidate for the causal gene of hepatic cell proliferation and/or hypertrophy induced by PPARα agonists, but it may not be the proximal cause insofar as the data of mice are concerned. Looking at the extent of hepatic hypertrophy in rats estimated by relative liver weight per body weight, where Hdc was upregulated, it was higher than that in mice, where Hdc was not induced. Although a simple comparison is difficult, there remains some possibility that HDC plays a partial role in the hepatic hypertrophy, considering the known role of histamine in the proliferation of various cell types in rodents (Ichikawa et al., 2010; Medina and Rivera, 2010). At present, the most feasible interpretation would be that the hypertrophy caused by PPARα agonists is mainly due to their well-known enzyme-inducing effect (Maronpot et al., 2010), and the physiological role of Hdc induction on cell proliferation is possible but still unknown. Further study is necessary for comprehensive understanding of the pharmacology and toxicology of PPARα agonists, and we believe that the present results contribute to the research providing a new clue.
We thank Prof. Atsushi Ono (Okayama University) for fruitful discussion. We thank Mr. John H. Jennings for editing manuscript.
Conflict of interestThe authors declare that there is no conflict of interest.
