In vivo protein expression changes in mouse livers treated with dialyzed coffee extract as determined by IP-HPLC

Background Coffee extract has been investigated by many authors, and many minor components of coffee are known, such as polyphenols, diterpenes (kahweol and cafestol), melanoidins, and trigonelline, to have anti-inflammatory, anti-oxidant, anti-angiogenic, anticancer, chemoprotective, and hepatoprotective effects. Therefore, it is necessary to know its pharmacological effect on hepatocytes which show the most active cellular regeneration in body. Methods In order to determine whether coffee extract has a beneficial effect on the liver, 20 C57BL/6J mice were intraperitoneally injected once with dialyzed coffee extract (DCE)-2.5 (equivalent to 2.5 cups of coffee a day in man), DCE-5, or DCE-10, or normal saline (control), and then followed by histological observation and IP-HPLC (immunoprecipitation high performance liquid chromatography) over 24 h. Results Mice treated with DCE-2.5 or DCE-5 showed markedly hypertrophic hepatocytes with eosinophilic cytoplasms, while those treated with DCE-10 showed slightly hypertrophic hepatocytes, which were well aligned in hepatic cords with increased sinusoidal spaces. DCE induced the upregulations of cellular proliferation, growth factor/RAS signaling, cellular protection, p53-mediated apoptosis, angiogenesis, and antioxidant and protection-related proteins, and the downregulations of NFkB signaling proteins, inflammatory proteins, and oncogenic proteins in mouse livers. These protein expression changes induced by DCE were usually limited to the range ± 10%, suggesting murine hepatocytes were safely reactive to DCE within the threshold of physiological homeostasis. DCE-2.5 and DCE-5 induced relatively mild dose-dependent changes in protein expressions for cellular regeneration and de novo angiogenesis as compared with non-treated controls, whereas DCE-10 induced fluctuations in protein expressions. Conclusion These observations suggested that DCE-2.5 and DCE-5 were safer and more beneficial to murine hepatocytes than DCE-10. It was also found that murine hepatocytes treated with DCE showed mild p53-mediated apoptosis, followed by cellular proliferation and growth devoid of fibrosis signaling (as determined by IP-HPLC), and subsequently progressed to rapid cellular regeneration and wound healing in the absence of any inflammatory reaction based on histologic observations. Electronic supplementary material The online version of this article (10.1186/s40902-018-0183-z) contains supplementary material, which is available to authorized users.


Background
Coffee is one of the most popular daily drinks not only for ordinary people but also for patients in convalescence period after major surgical treatment, but too much coffee may harm human health. However, published results are controversial with respect to its effects on cardiovascular diseases, inflammation, diabetes, Parkinson's disease, cancer, and other diseases [1][2][3]. The beneficial pharmacological effects of coffee mentioned in the literature include anti-inflammatory, anti-oxidant, anti-angiogenic, anticancer, chemoprotective, and hepatoprotective effects [4][5][6][7], and coffee has been reported to contain polyphenols, diterpenes (kahweol and cafestol), melanoidins, and trigonelline [8][9][10][11].
Epidemiological studies support associations between coffee-specific diterpenes and various beneficial health effects. Although free cafestol and kahweol (coffee-specific diterpenes) have been recently reported to have antiangiogenic properties, little data is available regarding the health effects of esterified cafestol and kahweol, in particular, of their palmitate esters, which are the main diterpene esters present in coffee. Cafestol and kahweol palmitates inhibit the angiogenesis of human microvascular endothelial cells (HMVECs), although the effect of the kahweol ester is greater [12]. Kahweol has also been reported to protect against liver inflammation by downregulating the expressions of LPS-stimulated phospho-nuclear factor kappa B (NFkB) and signal transducer and activator of transcription 3 (STAT-3) expression [13].
Polyphenols derived from coffee beans have beneficial effects on blood pressure and vascular endothelial function, improve skin condition, and participate in cutaneous blood flow regulation after cold stress [14]. Chlorogenic acid hydrogels applied topically to significantly reduced wound areas during the inflammatory phase, possibly because of the well-known antioxidant and anti-inflammatory effects of chlorogenic acid, whereas caffeine (a known anti-oxidant) impeded keratinocyte proliferation and migration, suggesting it had an inhibitory effect on wound healing and epithelialization [15].
In our previous study, dialyzed coffee extract (DCE) and artificial coffee (AC) induced protein expressions in RAW 264.7 cells were compared by IP-HPLC analysis. DCE, which contains most of the minor components of coffee (including chlorogenic acid and caffeine), induced the expressions of proteins required for essential cellular functions in RAW 264.7 cells, while AC (1 mM chlorogenic acid and 2 mM caffeine, which are the same concentrations found in DCE) induced a quite different protein expressional pattern as determined by IP-HPLC. DCE caused the upregulations of proteins associated with cellular proliferation and protection, and antioxidant-related proteins, and the downregulations of apoptosis-related, angiogenesis-related, and oncogenic proteins; and enhanced cMyc/MAX, Rb/E2F, and RAS, growth factor signaling as well as osteogenesis in RAW 264.7 cells. Actually, overall protein expressional changes after DCE treatment revealed a signaling circuit triggered by antioxidant-related proteins and genetic/epigenetic activation [16].
The present study was undertaken to examine changes in protein expressions in mouse liver after administering animals the equivalents of 2.5, 5, or 10 cups of coffee in man (DCE-2.5, DCE-5, or DCE-10, respectively; determined on a body weight basis) to determine whether coffee has a beneficial effect on the liver. Mice were injected intraperitoneally with DCE to avoid issues associated with gastrointestinal absorption variability, and global protein expressions in mouse livers obtained 24 h later were determined by IP-HPLC.

Production of dialyzed coffee extract (DCE)
Twenty cups of coffee (20 × 150 mL = 3000 mL) were prepared from medium roasted coffee beans (Coffea arabica L., Nepal, at 20 g per cup and 90-95°C. Aliquots (300 mL) of this extract were repeatedly dialyzed 10 times using a permeable cellulose bag (< 1000 Da; 131,492, Spectra, USA) in 1500 mL of double distilled water at 4°C with stirring for 2 h. The dialyzed coffee extract (DCE) was immediately stored at − 70°C until use.
DCE was subjected to non-adherent reverse phase column chromatography (YMC-Pak, Japan) using water as an eluent and a HPLC unit (1100, Agilent, USA). It was found that the primary constituents of DCE were caffeine and chlorogenic acid. HPLC analysis of DCE revealed a caffeine concentration of~2 mM, indicating that 150 mL DCE contained~60 mg of caffeine (Table 1). Because 150 mL of ordinary coffee extract contained~120 mg of caffeine, the dialysis coefficient for the caffeine of coffee was~50% and 300 mL of DCE was equivalent to one cup of coffee extract (150 mL) for a human adult (mean 60 kg, 59.4 l). Thus, 300 mL of DCE for a human adult (DCE-1) was equivalent to 0.15 mL of DCE for a mouse (mean 30 g) in animal experiment (Table 1). In order to check for lipopolysaccharide (LPS) contamination in DCE, a LPS detection assay was performed by IP-HPLC using anti-LPS antibody (Santa Cruz Biotechnology, USA). DCE (1 and 2 mL, experiments 1 and 2), 1 mL LPS solution (1 ng/mL, Sigma Aldrich, USA, positive control), and 1 mL distilled water (negative control) were separately analyzed by IP-HPLC. Peak areas of experiments 1 and 2 were similar to that of the negative control, while the peak area of the positive control (LPS solution) was predominantly increased (Additional file 1) [16]. These results indicated DCE was effectively free from LPS contamination.

Histological and immunohistochemical observation
Mouse liver microsections in 4 μm thickness were subjected to hematoxylin and eosin staining, and serial microsections were also prepared for immunohistochemical staining using representative antisera of hepatocyte growth factor (HGF), glutathione S-transferase-1 (GST-1), poly-ADP ribose polymerase (PARP), or tumor necrosis factor-α (TNFα) (Santa Cruz Biotech. USA). Immunohistochemical reaction protocols differed according to target antigens and manufacturers' protocols, but briefly, after xylene deparaffinization and rehydration in an ethanol series, sections were incubated with 0.5% hydrogen peroxide in phosphate-buffered saline for 30 min. Primary anti-human (rabbit/mouse/goat) polyclonal antibodies were applied to each section using triple sandwich indirect immunohistochemical methods [17]. Histochemical stains were observed under a light microscope, and images were captured by a digital camera (DP-73; Olympus Co., Japan).
Briefly, protein samples were mixed with 5 mL of binding buffer (150 mM NaCl, 10 mM Tris-HCl pH 7.4, 1 mM EDTA, 1 mM EGTA, 0.2 mM sodium vanadate, 0.2 mM phenylmethylsulfonyl fluoride (PMSF) and 0.5% Tergitol-type NP-40 (nonyl phenoxypolyethoxylethanol) and incubated in protein A/G agarose columns at 4°C for 1 h (columns were placed on a rotating stirrer during the incubation). After washing each column with sufficient PBS (phosphate-buffered saline), target proteins were eluted using 150 μL of IgG elution buffer (Pierce, USA). Immunoprecipitated proteins were analyzed using a HPLC unit (1100 series, Agilent, USA) equipped with a reverse phase column and a micro-analytical detector system (SG Highteco, South Korea). Elution was performed using 0.15 M NaCl containing 20% acetonitrile at 0.4 mL/min for 30 min, and detection by UV spectroscopy at 280 nm. Control and experimental samples were run sequentially to allow comparisons. For IP-HPLC, whole protein peak areas (mAU*s) were calculated by subtracting the antibody peak areas of negative controls, and experimental protein peak area square roots were compared with control one (Additional file 2) [16].
When IP-HPLC results were compared with western blot data for cytoplasmic housekeeping protein (β-actin), IP-HPLC errors were < ± 5%, whereas western blot errors exceeded ± 20% and were not suitable for statistical analysis (Additional file 3) [16]. In particular, repeat IP-HPLC runs (4-10 runs) to determine errors associated with protein expression revealed errors were ± 5% (Additional file 4) [16]. Based on these findings, IP-HPLC was used rather than western blot to analyze protein expressional changes.

Histological observations
Mouse livers treated with DCE-2.5 showed hypertrophic hepatocytes containing more eosinophilic cytoplasms than non-treated controls. Most hepatic cords were thickened with hypertrophic hepatocytes, but their sinusoidal spaces were well preserved. Many hepatocytes showed increased heterochromatic nuclei, but no necrotic hepatocytes were observed during the histological observation (Fig. 1b). Mouse livers treated with DCE-5 consistently showed hypertrophic hepatocytes and narrow sinusoidal spaces (Fig. 1c), while those treated with DCE-10 showed shrunken hepatocytes and had larger sinusoidal spaces than non-treated controls (Fig. 1d).

Effects of DCE on the expressions of cMyc/MAX/MAD network proteins in mouse livers
Mouse livers treated with DCE showed expression changes of cMyc, MAX, and MAD of < ± 5% as was observed for control housekeeping proteins (Fig. 3B1, B2), indicating DCE did not activate cMyc/MAX signaling in murine hepatocytes. Effects of DCE on the expressions of p53/Rb/E2F signaling proteins in mouse livers DCE-2.5 or DCE-5 altered the expressions of p53, MDM2, Rb-1, E2F-1, and CDK4 by < ± 5%, which was similar to those observed for control housekeeping proteins, but DCE-10 increased the expression of CDK4 (105.4%) and reduced the expression of Rb-1 (94%) as compared with non-treated control livers (Fig. 3C1, C2). These results suggest that DCE-2.5 and DCE-5 did not activate p53/Rb/E2F signaling in murine hepatocytes.

Effects of DCE on the expression of Wnt1/β-catenin signaling proteins in mouse livers
Mouse livers treated with DCE showed expression changes in Wnt1, β-catenin, APC, and TCF-1 of < ± 5%, which was similar to that observed for control housekeeping proteins, but had reduced snail levels (93.9%) (Fig. 3D1, D2). These results suggested DCE did not activate Wnt/β-catenin signaling in murine hepatocytes.

Effects of DCE on the expressions of protein translationrelated proteins in mouse livers
Mouse livers treated with DCE showed expression changes of DHS, eIF5A-1, eIF5A-2, and eIF2AK3 of < ± 5%, like control housekeeping proteins, but showed slight decreases in the levels of DOHH (91.6%) after treatment with DCE-5 or DCE-10, and a slight increase in eIF2AK3 (107.9%) after treatment with DCE-10 ( Fig. 4B1, B2). These results suggested DCE slightly inactivated protein translation by downregulating DOHH and upregulating eIF2AK3.

Effects of DCE on FAS-mediated apoptosis-related protein levels in mouse livers
Mouse livers treated with DCE-10 had higher FASL (106.1%), FAS (105.6%), and FLIP (106%) levels than non-treated controls. The levels of other FAS-mediated apoptosis-related proteins, that is, FADD, caspase 8, caspase 3, c-caspase 3, and BID, changed by < ± 5% in response to DCE, as was observed for control housekeeping proteins (Fig. 6D1, D2). These results suggested that DCE-2.5 and DCE-5 did not induce FAS-mediated apoptosis-related protein expressions in murine hepatocytes.

Global protein expressions in mouse livers treated with DCE-5
Global protein expressions induced by DCE in mouse liver changed by < ± 10%, that is, they probably remained in the physiological homeostatic range. The levels of many proteins essential for molecular signaling were altered by ± 5% as was observed for housekeeping controls (β-actin, α-tubulin, and glyceraldehyde-3-phosphate dehydrogenase, GAPDH). These results indicated DCE treatment caused minimal cellular stress and did not induce inflammatory or chemical stress or oncogenic injury. Generally, mouse livers treated with DCE-5 showed characteristic changes in functional protein levels (Fig. 8). Increases in the 10% range were observed for proliferation-related proteins (MPM2, CDK4, and Ki-67), growth factors (HGF-1, IGF-1, HER2, GH, GHRH, insulin, KRAS, RAF-B, ERK-1, pAKT1/2/3, and JNK-1), cellular adaptation-related proteins (leptin, PKC, AKAP, and HXK II), cellular differentiation-related proteins (Jagged 2, Notch 1, GLI-1, Muc1, Muc4, and SP-1), inflammatory proteins (IL-6, IL-8, IL-12, TGF-β1, COX-1, and LL-37), p53-mediated apoptosis-related proteins (BAX, APAF-1, c-caspase 9, and c-PARP), angiogenesis-related proteins (VEGF-C, CMG2, LYVE-1, FGF-2, and VCAM), and antioxidant and protection-related proteins (HO-1, HSP-70, GST-1, leptin, and hepcidin). On the other hand, decreases in the 10% range were observed for NFkB signaling proteins (mTOR, LC3, and GADD45), inflammatory proteins (TNFα, COX-2, CD68, and cathepsin G), and oncogenic proteins (TERT, MBD4, and YAP1). cMyc/MAX/MAD network proteins, p53/Rb/E2F signaling proteins, Wnt/β-catenin signaling proteins, epigenetic modification-related proteins, protein translation-related proteins, and FAS-mediated apoptosis-related proteins were rarely affected by DCE treatment (Fig. 8). Discussion 2.5, 5, or 10 cups of coffee are slowly absorbed through the gastrointestinal tract, whereas in the present study, DCE-2.5, DCE-5, and DCE-10 were injected intraperitoneally, which is likely to have a far greater effect on liver. However, the DCE-induced hypertrophic effect on murine hepatocytes observed was more marked than was expected, although hypertrophic hepatocytes were well aligned in hepatic cords and sinusoidal spaces were well maintained. Furthermore, no necrotic hepatocytes recruiting inflammatory cells were observed histologically, and thus, we considered the pharmacological effects of DCE-10 still lay in the homeostatic range of hepatocyte metabolism, and that DCE-2.5 which induced mild hypertrophic effect on murine hepatocytes would be safer than DCE-5 which induced marked hypertrophic effect. Nevertheless, mice treated with DCE-10 showed marked hepatocyte shrinkage and greater sinusoidal spaces, indicating that DCE-10 caused a certain amount of cellular damage. Although no necrotic hepatocytes were observed in DCE-10-treated mouse livers, it is possible that the metabolic statuses of hepatocytes may have been diminished due to smaller amounts of hepatocyte cytoplasm observed in DCE-10 than in DCE-2.5and DCE-5-treated mouse livers. These observations suggest that mouse livers were over-stimulated by DCE-10, and that hepatocytes underwent transient retrogressive degeneration. The protective and antioxidant effects of DCE observed in mouse livers were similar to those reported for kahweol (a coffee-specific diterpene) in SH-SY5Y cells, which upregulated HO-1 and p38 levels [21,22]. It has been suggested the anti-apoptotic effect of DCE in mouse liver might play a role in the radiation-protective effect of caffeine via the downregulation of BAX protein [23]. In other studies, the anti-inflammatory effects of DCE-2.5 and DCE-5 were found to be closely related to their downregulation of NFkB signaling [8,24]. Furthermore, kahweol has been reported to suppress the proliferation and induce the apoptosis of human colorectal cancer cells, and head and neck squamous cell carcinoma cells [25][26][27]. However, in the present study, DCE slightly induced the proliferation of murine hepatocytes and simultaneously reduced the expressions of inflammatory proteins (TNFα, COX-2, IL-6, and CD68) and of oncogenic proteins (YAP1 and TERT).
The liver is for detoxifying various metabolites and producing biochemicals required for digestion, and its roles also include the regulation of glycogen storage, the decomposition of red blood cells, and the production of hormones. The liver is the only visceral organ that possesses the capacity to regenerate, for example, after surgical removal or chemical injury, in fact, as little as 25% of original liver mass can regenerate a full-sized liver [28,29]. In the present study, DCE slightly increased the expressions of p53-mediated apoptosis-related proteins, but did not affect the expressions of FAS-mediated apoptosis-related proteins, and these protein expressional changes occurred in parallel with increases in the expressions of MMPs and enhanced hepatocyte proliferation. These observations suggest that DCE induced the apoptosis of old hepatocytes and enhanced scavenging of resulting debris by MMPs and hepatocyte regeneration by activating proliferation-related proteins, and followed by reactive de novo angiogenesis for sinusoidal vasculature.
The anti-inflammatory effects of DCE observed in mouse liver in the present study were consistent with its strong antioxidant effect, its inactivation of NFkB signaling, and its promotion of anti-oncogenic signaling, but the observed increase in the hepatocyte proliferation may not have been related to cMyc/MAX signaling, Rb/ E2F signaling, or epigenetic modification, but rather to slight increases in the expressions of growth factors, such as HGF, IGF-1, GH, GHRH, and CTGF; and slight activation of RAS signaling involving RAF-B, ERK-1, PI3K, JNK-1, and pAKT1/2/3. In particular, DCE downregulated FGF-1, TGF-β1, TGF-β2, and SMAD4 levels (fibrosis-related proteins), but upregulated levels of Erβ and insulin (anti-fibrosis proteins) in mouse liver. These findings suggest that mouse livers treated with DCE might progressively regenerate via growth factor/RAS signaling and anti-inflammation, low cellular stress, and anti-oncogenic signaling.
Tea and coffee have been associated, both positively and negatively, with the risk of cardiovascular disease (CVD). The effects of coffee remain controversial and concerns have been expressed regarding associations between its consumption and hypercholesterolemia, hypertension, and myocardial infarction [30]. Caffeine and kahweol are known anti-angiogenic compounds [31,32] that may function as anti-tumor and anti-myocardial infarct agents. In the present study, DCE slightly upregulated angiogenesis-related protein levels (FLT-4, COX-1, leptin, VCAM, and PAI-1) in mouse livers, which contrasts with its antiangiogenic effect in RAW 264.7 cells [16]. However, we observed these expressions of angiogenesis stimulating proteins were not associated with major angiogenesis signaling involving HIF, VEGF-A, VEGF-C, angiogenin, vWF, CD31, and PDGF-A, and thus, it would appear that the angiogenic effect of DCE in mouse livers was a transient phenomenon induced to counter vascular sinusoidal structure disruption during rapid hepatocyte regeneration triggered by DCE.
Caffeine and chlorogenic acid are the predominant polyphenol derivatives in coffee and their biological functions have been well investigated [33,34], but many other constituents have not been characterized or clearly identified. Kahweol as a coffee-specific diterpene that has been reported to have anti-cancer properties. Kahweol-mediated cyclin D1 degradation may contribute to the inhibition of human colorectal cancer cell proliferation [35], and kahweol was observed to significantly decrease TGF-β (transforming growth factor beta) stimulated expressions of type I collagen and CTGF in vitro. In addition, in hepatocytes, kahweol significantly decreased the expressions of Smad3, STAT3, ERK, and JNK, which are involved in the induction of CTGF expression by TGF-β [36]. In the present study, DCE-2.5 reduced the protein levels of TGF-β1, TGF-β2, CTGF, SMAD4, STAT3, and ERK-1 by < ± 5%, but increased the expression of JNK-1 to 108.1%. Furthermore, this increase in JNK-1 co-occurred with increases in the levels of MMP-9, PI3K, and pAKT1/2/ 3, might suggest MMP-9 stimulated PI3K/Akt/JNK signaling was induced by DCE in mouse livers to support hepatocyte regeneration.
In the present study, DCE was found to increase levels of proliferation-related proteins, growth factors, cellular adaptation-related proteins, cellular differentiation-related proteins, inflammatory proteins, p53-mediated apoptosisrelated proteins, angiogenesis-related proteins, and antioxidant and protection-related proteins; to reduce levels of NFkB signaling proteins, inflammatory proteins, and oncogenic proteins; but not to substantially effect levels of cMyc/MAX/MAD network proteins, p53/Rb/E2F signaling proteins, Wnt/β-catenin signaling proteins, epigenetic modification-related proteins, protein translation-related proteins, or FAS-mediated apoptosis-related proteins in mouse livers. Furthermore, global protein expression induced by DCE-5 was changed by < ± 10%, and thus, probably remained in the physiological homeostatic range. In addition, the levels of many essential proteins relevant to molecular signaling were increased or decreased by around ± 5% by DCE, which was similar to that observed for housekeeping proteins (β-actin, α-tubulin, and GAPDH). These changes suggest DCE-2.5-, DCE-5-, and DCE-10treated mouse livers exhibited minimal cellular stress, inflammation, or chemical or oncogenic injury, and that after the rapid removal of senile or degenerated hepatocytes through p53-mediated apoptosis they underwent hepatocyte regeneration and sinusoidal vasculature recovery similar to wound healing mechanism.
Conclusions DCE upregulated the expressions of cellular proliferation, growth factor/RAS signaling, cellular protection, p53-mediated apoptosis, angiogenesis, and antioxidantrelated proteins, but downregulated the expressions of NFkB signaling, inflammatory, and oncogenic proteins in mouse livers. These protein level changes usually fell in the range ± 10%, which suggested that murine hepatocytes reacted to DCE within thresholds of physiological homeostasis. DCE-2.5 and DCE-5 induced relatively mild dose-dependent changes in protein expressions as compared with non-treated controls, whereas DCE-10 induced fluctuations in protein expressions, which