Abstract
Epigallocatechin‐3‐gallate (EGCG) is a major bioactive compound in tea
polyphenol extract. After ingestion, EGCG reaches the intestine and may
commence anti‐inflammation in the intestinal organ. Thus, in this
paper, the anti‐inflammatory effect of EGCG was studied using
lipopolysaccharide (LPS)‐induced inflammation in RAW 264.7 cells. LPS
induction instigated morphological deformation extensively which was
normalized by EGCG. In LPS‐induced macrophage cells, EGCG was found to
lower cellular nitric oxide (32% of LPS group) and intercellular ROS
level (45.4% of LPS group). It also suppressed the expression of IL‐1β
(LPS 132.6 ± 14.6, EGCG 10.67 ± 3.65), IL‐6 (LPS 2994.44 ± 178.5, EGCG
408.33 ± 52.34), TNF‐α (LPS 27.11 ± 2.84, EGCG 1.22 ± 0.03), and iNOS
(LPS 40.45 ± 11.17, EGCG 10.24 ± 0.89). The GO function analysis
identified that these differential genes involved 24 biological
processes, 18 molecular functions, and 19 cellular component‐related
processes. KEGG pathway enrichment analysis revealed that LPS
significantly affects NF‐κB, TNF, and TLR signaling pathways. Western
blotting revealed that EGCG diminished P‐IκB/IκB ratio by 75% and
p‐p65/p65 by 50% compared to the LPS group. Finally, Arg‐1 and CD‐206
mRNA expression were determined by RT‐PCR, which was consistent with
the RNA‐Seq result. These findings indicate that EGCG exerts an
anti‐inflammatory effect by reducing NO and ROS production, suppressing
TLR4 protein expression, and inhibiting IκB and p65 phosphorylation.
Keywords: anti‐inflammation, EGCG, LPS, macrophages, NF‐κB
__________________________________________________________________
Epigallocatechin‐3‐gallate (EGCG) is a major bioactive compound in tea
polyphenol extract. In LPS‐induced macrophage cells, EGCG was found to
lower cellular nitric oxide (32% of LPS group) and intercellular ROS
level (45.4% of LPS group). It also suppressed the expression of IL‐1ß
(LPS 132.6 ± 14.6, EGCG 10.67 ± 3.65), IL‐6 (LPS 2994.44 ± 178.5, EGCG
408.33 ± 52.34), TNF‐a (LPS 27.11 ± 2.84, EGCG 1.22 ± 0.03), and iNOS
(LPS 40.45 ± 11.17, EGCG 10.24 ± 0.89). Western blotting revealed that
EGCG diminished P‐IκB/IκB ratio by 75% and p‐p65/p65 by 50% compared to
the LPS group. Finally, Arg‐1 and CD‐206 mRNA expression were
determined by RT‐PCR, which was consistent with the RNA‐Seq result.
graphic file with name FSN3-11-4634-g010.jpg
1. INTRODUCTION
Inflammation, which is the first line of defense of our body against
pathogens, irritation, and injury, is a sophisticated response
regulated by cytokines and chemokines. Various immune cells including
monocytes, neutrophils, and macrophages are acquired by our innate
immune system to address the inflammation (Novilla et al., [34]2017;
Oyungerel et al., [35]^2013). Macrophage cells play a vital role in
inflammation modulation, as they are the first responders to
inflammation. Exposure to cytokines, chemokines, or bacterial
lipopolysaccharide (LPS) leads to macrophage cell activation.
Phagocytic activity of macrophage cells increases many folds during
inflammation and activated macrophage cells fight pathogens directly.
Macrophage cells indirectly maneuver inflammation by secreting
proinflammatory cytokines (IL‐1β, TNF‐α, IL‐6) and inflammatory
mediators (NO, iNOS) (Arango Duque & Descoteaux, [36]2014). Unregulated
secretion of cytokines and inflammatory mediators results in damage on
both cellular and tissue levels. Cellular damage ends up in apoptosis
and necrosis while tissue damage includes the development of many
chronic diseases, viz., rheumatoid arthritis, chronic hepatitis,
diabetes, pulmonary fibrosis, and cancer (Kim et al., [37]2016; Laveti
et al., [38]^2013; Liu et al., [39]^2017; Tsai et al., [40]^2018).
Nuclear factor kappa‐light‐chain‐enhancer of activated B cells (NF‐κB)
is one of the most important transcription factors that regulate
inflammatory response along with other physiological processes (Hossen
et al., [41]2020, [42]2021; Schulert & Grom, [43]^2015). As long as an
inhibitor of κBs (IκBs) is not phosphorylated, NF‐κB remains inactive
in the cytoplasm. Phosphorylation of IκB leads to NF‐κB nuclear
translocation and ends up in transactivation of downstream genes (Baker
et al., [44]2011; Siebenlist et al., [45]^2005). Activated downstream
target genes direct the cells to produce inflammatory cytokines and
mediators including NO, TNF‐α, and IL‐6. They also direct the
recruitment of innate immune cells to combat inflammation
(Lawrence, [46]2009; Schneider et al., [47]^2014). These make NF‐κB an
ideal candidate to study anti‐inflammatory chemicals and substances.
Green tea is very popular in East Asia mainly in China, Japan and
gaining popularity around the world (Chacko et al., [48]2010). Previous
studies have shown that the consumption of green tea may reduce risks
associated with cardiovascular disease and exert numerous health
benefits (Wang et al., [49]2011). Around 30% dry weight of green tea
are polyphenols and among them, flavonoids are most important; 80%–90%
of the flavonoids are catechins and epigallocatechin‐3‐gallate (EGCG),
which cover 59% of the total catechins present (Jigisha et
al., [50]2012; Reygaert, [51]^2017; Roowi et al., [52]^2010). EGCG has
low bioavailability (0.1%) but its content in the intestine is very
high. The bioavailability of EGCG is proportional to immune system
upregulation and improvement of health. The more EGCG reaches the
target site, the better it is for the body (Xu et al., [53]2015).
Lambert et al. ([54]2003) intragastrically fed male CF‐1 mice
163.8 μmol/kg EGCG and the levels in the small intestine and colon were
45.2 ± 13.5 and 7.86 ± 2.4 nmol/g, respectively. Accumulating studies
have demonstrated that EGCG possesses numerous health benefits,
including anti‐inflammatory, antioxidant, anticancer, and antitumor
properties. EGCG lowers the pro‐inflammatory cytokine and chemokine
expression; also suppresses MAPK, STAT, TLR4, and NF‐κB signaling
pathway (Almatroodi et al., [55]2020; Cao et al., [56]^2019; Chu et
al., [57]^2017; Yahfoufi et al., [58]^2018).
Therefore, this study intends to check the effect of EGCG on
suppressing body inflammation and the possible mechanisms involved.
This is for the first time the effects of EGCG from morphological
impact to the biochemical changes, gene expression level, and protein
expression changes are combined in one single manuscript. We attempted
to compare the effects of EGCG and dexamethasone (DEX) on inflammation
in terms of cell phagocytic capacity and the inhibitory effects on
IL‐1β, IL‐6, TNF‐α, and iNOS pro‐inflammatory factors effect. We also
observed the cell morphology using an inverted microscope, measured
physiological indicators such as NO and ROS, and used reverse
transcription PCR (RT‐PCR) to determine the level of inflammatory
factors to evaluate the anti‐inflammatory effect of EGCG. Additionally,
transcriptome sequencing was used to determine the mRNA levels of
related factors, and western blotting was used to determine related
protein expression related to the possible anti‐inflammatory mechanism.
Our findings will provide a more theoretical basis for the role of EGCG
in maintaining intestinal permeability, and its later use in the
development of healthy foods.
2. MATERIALS AND METHODS
2.1. Reagents
EGCG (HPLC grade, ≥98%) (Catalogue no. HY‐N6263) and
3‐(4,5‐Dimethylthiazol‐2‐yl)‐2,5‐diphenyltetrazolium bromide (MTT)
(Catalogue no. BN30793) were purchased from Shanghai Yuanye
Biotechnology Co., Ltd. Neutral red (Catalogue no. DE‐E895A) was
purchased from Shanghai Macklin Biochemical Co., Ltd. Dulbecco's
Modified Eagle Medium (DMEM) (Catalogue no. [59]B20290), fetal bovine
serum (FBS) (Catalogue no. P08X20), and glutamine (Catalogue no.
BIO‐000001) were purchased from Gibco Corporation (Life Technologies,
Thermo Fisher Scientific). Phosphate‐buffered saline (PBS) (Catalogue
no. [60]B20719) and trypsin (Catalogue no. 84278A) were obtained from
HyClone. Dimethyl sulfoxide (DMSO) (Catalogue no. BN35879), 2′,
7’‐Dichlorofluorescin diacetate (DCFH‐DA) (Catalogue no. DE‐D1002),
dexamethasone (Dex) (Catalogue no. E120263), and LPS (Catalogue no.
S11060) were bought from Sigma‐Aldrich. Sodium pyruvate (Catalogue no.
113–24‐6) was obtained from Beijing Banxia Biological Technology Co.,
Ltd. Isopropanol (Catalogue no.67–63‐0), ethanol (Catalogue no.
[61]G00004), and glacial acetic acid (Catalogue no. A116166) were
purchased from Beijing chemical works company Ltd. RNA extraction kit
was purchased from TransGen Biotech Co., Ltd. SYBR Green PCR master mix
and reverse transcription kit were purchased from Toyobo (Japan). Total
nitric oxide (NO) assay kit was obtained from Beyotime Institute of
Biotechnology Co., Ltd.
2.2. Cell culture
Raw 264.7 macrophage cells were collected from the Stem Cell Bank,
Institute of Zoology (Chinese Academy of Sciences). Cells were
maintained in DMEM supplemented with 10% FBS, 1% glutamine, and 1%
sodium pyruvate. Cells were incubated at 37°C in a 5% CO[2] atmosphere
and sub‐cultured every 2 days (Hossen et al., [62]2021).
2.3. Cytotoxic and MTT assay
To prepare the EGCG stock solution, 1 mg of EGCG was dissolved in 1‐mL
PBS and then stored at −20°C till use. Cell viability was measured by
MTT assay following the method followed by Hossen et al. ([63]2021).
RAW 264.7 cells (1 × 10^5 cells/mL) inoculated at 100 μL in 96‐well
plates for 24 h. Later, the culture solution is replaced with a
serum‐free medium mixed with different doses of EGCG and LPS (1 μg/mL)
and Dex (25.48 μmol/L). After 24 h of incubation, the culture solution
was discarded, and then MTT stock solution was added at 37°C and left
for incubation for 4 h. The formazan crystals formed in this step were
dissolved by adding 150‐μL DMSO and optical density (OD) is measured at
570 nm using a spectrophotometer (Tecan, Männedorf, Switzerland). The
following formula is used to measure cell viability.
[MATH: cell viability%=ODContro
l−ODSample
OControl−ODBlak×10% :MATH]
2.4. Cell morphology analysis
For morphology analysis, 6.3 × 10^5 RAW264.7 cells were seeded in a
six‐well plate and incubated in a humidified incubator for 24 h at 37°C
with 5% CO[2] in it. Then, 43.6 μmol/L EGCG +1 μg/mL LPS, 25.48 μmol/L
DEX + 1 μg/mL LPS, and 1 μg/mL LPS were added and incubated for 24 h.
Later, an inverted microscope is used to randomly select three
locations in the dish for morphological recording, and calculate the
pseudo‐foot ratio (Hong et al., [64]2012).
2.5. Phagocytosis
RAW264.7 cells (1 × 10^5 cells/mL) were seeded in a 96‐well plate and
incubated for 24 h at 37°C in a 5% CO[2] incubator until full
confluency. After LPS treatment, the cells were added with EGCG
(21.8 μmol/L, 43.6 μmol/L, and 87.2 μmol/L) and DEX (25.48 μmol/L), and
incubated for 24 h. In the following, the cells were added with 100 μL
of 0.1% neutral red solution and incubated for another 4 h. Later, the
culture solution is discarded and washed thrice with PBS to remove
neutral red that has not been engulfed by the cells. Then, 100 μL of
acetic acid and ethanol solution (1:1, v/v) was added to each well, and
the plates were placed at 4°C for 4 h to allow full lysis of the cells.
Then, the absorbance of the cell lysate was measured using a
spectrophotometer (Tecan) at 540 nm, and relative phagocytic activity
was calculated by following the method of Chen et al. ([65]2016).
2.6. Determination of NO content in cells
Griess method was applied to measure the NO content by following the
method of Joo et al. ([66]2014). Raw 264.7 cells (6.3 × 10^5) grew to
confluency in a 6‐cm cell culture dish containing a 3‐mL culture
medium. After confluency, various concentrations of EGCG (10.9 μmol/L,
21.8 μmol/L, 43.6 μmol /L, and 87.2 μmol/L) and DEX (25.48 μmol/L) were
added to cells pretreated with LPS (1 μg/mL). After 24 h of incubation,
cell supernatants were separated by centrifugation at 1500x g, then
100 μL of supernatants from each type was added to 100‐μL Griess
reagent. Absorbance was measured at 540 nm using a spectrophotometer
(Tecan, Männedorf, Switzerland) in a 96‐well microplate reader, and NO
concentration was calculated using the standard curve.
2.7. ROS level determination
DCFH‐DA fluorescent probe method was used to detect the ROS level
(Hossen et al., [67]2021). RAW264.7 cells (1 × 10^5 cells) were
cultured in a 96‐well dark plate until confluency. Then, the culture
medium was discarded and cells were washed gently with PBS buffer.
Afterward, cells were treated with LPS (1 μg/mL) and then each group
was treated with EGCG (10.9 μmol/L, 21.8 μmol/L, 43.6 μmol /L, and
87.2 μmol/L) and DEX (25.48 μmol/L). Then, cells were incubated for
24 h. DCFH‐DA was added with a final concentration of 10 μmol/L and
incubated at 37°C in the dark for 30 min. Then, the cell medium was
discarded and unbound DCFH‐DA was washed with PBS buffer. Then, 100‐μL
cell medium was added and fluorescence intensity was measured with an
excitation wavelength of 485 nm and an emission wavelength of 530 nm
using a spectrophotometer (Tecan, Männedorf, Switzerland).
2.8. .2.8 Transcriptome analysis
RAW264.7 cells (6.3 × 10^5 cells/mL) were cultured till adherence in a
6‐cm‐diameter cell culture dish. After the cells adhered to the wall,
the serum‐free medium was used to replace the culture medium and then
inoculated with LPS and EGCG + LPS for 24 h. After that, cells were
washed twice with PBS buffer and 1‐mL Transzol (TransGen Biotech,
Beijing) lysate was added to lyse the cells. After lysis, lysates were
transferred to a 1.5 mL of RNase‐free centrifuge tube and stored at
−80°C. Subsequent RNA extraction, detection, library construction,
sequencing (Illumina HiSeq platform), and preliminary analysis were
performed in Biomarker Technologies Corporation, Beijing. Sequenced
data were filtered to get clean data after primary analysis, and
compared the sequence with the mouse reference genome to get mapped
data. Subsequent evaluation of library quality included insert length
test, randomness test, and data saturation test as well as the analysis
of sequence–structure levels such as variable splicing analysis, new
gene discovery and gene structure optimization, and finally
differential expression quantification of sample genes and difference
analysis (Yu et al., [68]2018).
The expression level of the sample genes was calculated, and the FPKM
algorithm was used to normalize the expression level:
[MATH: FPKM=cDNA
fragmentsMapped fragmentsMillions×Transcript
lengthkb :MATH]
cDNA fragments indicate the number of fragments aligned to the
transcript; mapped fragments (millions) indicate the total number of
fragments aligned to the transcript, in millions; Transcript length
(kb) is the length of the transcript.
DEseq software was used for differential gene screening, and the
screening criteria were fold change ≥2.0 (Log2 fold change ≥l) and FDR
≤0.01. Among them, fold change represents the ratio of expression
between two groups of samples, q‐value is the significance of the
differential expression, and the p‐value is corrected to obtain the FDR
value of the false discovery rate. Controlling FDR below a certain
threshold can reduce the false‐positive rate differential expression of
genes.
2.9. Determination of IL‐1β, IL‐6, TNF‐α, and iNOS using RT‐PCR analysis
RT‐PCR analysis was conducted using a Bio‐Rad CFX96 touch system
following the method described by Gao et al. ([69]2020). The total RNA
from the RAW 264.7 macrophages treated with LPS in the presence or
absence of EGCG and DEX was extracted using the TransZol Up reagent
(TransGen Biotech). Complementary DNA (cDNA) was synthesized from 1 μg
of total RNA using a transcription kit (TOYOBO). NCBI blast and Primer
5 were used to design specific primer sequences for RT‐PCR. Primer
sequences used for RT‐PCR are given in Table [70]1. The reaction
conditions used for 38 cycles are as follows: predenaturation at 95°C
for 2 min, denaturation at 95°C for 30 s, annealing at 57°C for the
30s, extension at 72°C for 30s, extension at 72°C for 2 min.
TABLE 1.
Primer sequences used for RT‐PCR.
Name Sequence
β‐Actin‐F CCTAGAAGCATTTGCGGTGCACGATG
β‐Actin‐R TCATGAAGTGTGACGTTGACATCCGT
IL‐1β‐F TGCAGAGTTCCCCAACTGGTACATC
IL‐1β‐R GTGCTGCCTAATGTCCCCTTGAATC
IL‐6‐F AAGTGCATCATCATCGTTGTTCATACA
IL‐6‐R GAGGATACCACTCCCAACAGACC
TNF‐α‐F TACAGGCTTGTCACTCGAATT
TNF‐α‐R ATGAGCACAGAAAGCATGATC
iNOS‐F GCTGTGTGTCACAGAAGTCTCGAACTC
iNOS‐R AATGGCAAACATCAGGTCGGCCATCATC
Arg‐1‐F CAAGACAGGGCTCCTTTCAG
Arg‐1‐R GTAGTCAGTCCCTGGCTTATGG
CD206‐F CCTCAACCCAAGGGCTCTTCTAA
CD206‐R AAGGTGGCCTCTTGAGGTATGTG
[71]Open in a new tab
2.10. Quantification of 67LR, IκB, p65, PPARγ, and TLR4 using western
blotting
Western blot analysis was conducted for protein expression as described
by Hossen et al. ([72]2020). RAW264.7 cells (8.5 × 10^5 cells/mL) were
inoculated in a 6‐cm cell culture dish and incubated for 24 h. Later,
LPS or LPS + EGCG was added for 24 h. Then, cells were washed twice
with PBS and protein was extracted using cell lysate. The protein
concentrations were determined by the BCA method. An equal amount of
protein (50 μg) was separated by 10% SDS‐PAGE at 80 V for 120 min.
Separated proteins are then transferred to a PVDF membrane (Millipore),
after blocking the membrane with 5% nonfat milk prepared in Tris‐buffer
saline mixed with 0.1% Tween 20 (TBST) for 2 h at room temperature.
Primary and secondary antibodies for β‐actin, IκB, p IκB, p65, p‐p65,
and PPARγ were obtained from Cell Signaling Technology. TLR4 and 67LR
were purchased from Santa Cruz Biotechnology, Inc. Primary antibodies
were used at 1:1000 dilution and secondary antibodies at 1:2000
dilution. The membrane was incubated with the primary antibodies at 4°C
overnight. Later, the membrane was washed twice with TBST and incubated
with the secondary antibodies at 37°C for 1 h. The protein bands were
visualized using the ECL detection kit (Beijing BioDee Biotechnology
Company Ltd) with a chemiluminescence system.
2.11. Data processing
Screening and mapping of differential genes of transcriptome sequencing
data were completed using various data processing tools provided by the
Bimaike cloud platform. Excel 2010 was used to process the data, and
then SPSS 17.0 was used for one‐way analysis of variance. Finally,
GraphPad 7 was used for plotting. Fisher's LSD test was used to
corroborate the differences that occur among groups.
3. RESULTS AND ANALYSIS
3.1. Cytotoxicity and MTT assay
Before experimenting with EGCG, we attempted to evaluate the cytotoxic
effects of EGCG with different doses. RAW264.7 cell survival rate was
determined by MTT assay. After 24 h of incubation of cells with EGCG
(Figure [73]1a), DEX + LPS, and EGCG + LPS (Figure [74]1b) at 90% and
above, it showed no obvious toxic effect on cells. After treating cells
with 10.9 μmol/L EGCG for 24 h, the cell survival rate reached about
109%, which was significantly higher than that of the control group
(p < .05). It is evident that the concentrations of EGCG
(10.9–87.3 μmol/L) used in the subsequent experiments are within the
safe range and have no effect on cell survival.
FIGURE 1.
FIGURE 1
[75]Open in a new tab
RAW264.7 macrophage cells (1 × 10^5 cells) cultured and treated with
EGCG and EGCG+LPS and DEX for 24 h. Effect of EGCG and LPS on the
survival rate of RAW264.7 macrophage cells (^# p < .05 vs. control).
Each data point represents the mean ± SD (n = 3).
3.2. Cell morphology of raw 264.7
Figure [76]2 represents the cell morphology of the macrophage RAW 264.7
cells under EGCG and DEX treatment in the presence or absence of LPS
(1 μg/mL). Cells were then observed and photographed with an inverted
microscope after 24 h of incubation and then the pseudo‐foot ratio was
calculated. The cells in the control group were mostly round and bright
(Figure [77]2a), and the LPS‐treated cells formed long and slender
pseudopods (Figure [78]2b), which were different from the normal cells.
After EGCG (Figure [79]2c) and DEX (Figure [80]2d) treatment, most of
the cells were still round, significantly inhibiting LPS‐induced
morphological changes.
FIGURE 2.
FIGURE 2
[81]Open in a new tab
RAW264.7 macrophage cells (1 × 10^5 cells) cultured and treated with
EGCG and EGCG+LPS and DEX for 24 h. Morphological changes in RAW264.
7 cells were observed in an inverted microscope. (a). control, (b).
LPS, (c). EGCG, (d). DEX.
The number of pseudopods in the LPS group was about four times higher
than that of the control group (p < .01). EGCG significantly inhibited
the generation of pseudopods caused by LPS, the number of pseudopods in
the EGCG group was about 110% (p < .01). The number of pseudopods in
the DEX group was about 220%, which was significantly lower than that
in the LPS group, but it was still about twice that of EGCG (p < .05)
(Figure [82]3).
FIGURE 3.
FIGURE 3
[83]Open in a new tab
RAW264.7 macrophage cells (1 × 10^5 cells) cultured and treated with
EGCG and EGCG+LPS and DEX for 24 h. Percentage of cells with pseudopods
was calculated using an inverted microscope. ^## p < .01 versus control
group; *p < .05 versus LPS group, **p < .01 versus LPS group. Each data
point represents the mean ± SD (n = 3).
3.3. Effects of EGCG on LPS‐induced phagocytosis
Although the LPS group could promote the phagocytic capacity of cells,
there was no significant difference from the control group
(Figure [84]4). Compared with the LPS group, DEX stimulated the
phagocytic capacity of the cells, which were approximately 120% and
EGCG significantly inhibited the phagocytic capacity of the cells at
21.8 μmol/L, 43.6 μmol/L, and 87.2 μmol/L. The phagocytic capacity was
95%, 89%, and 85% (p < .05), respectively, and the inhibitory effect of
EGCG was significantly higher than the DEX group (Figure [85]4).
FIGURE 4.
FIGURE 4
[86]Open in a new tab
RAW264.7 macrophage cells (1 × 10^5 cells) cultured and treated with
EGCG and EGCG+LPS and DEX for 24 h. The effects of EGCG on LPS‐induced
phagocytosis of RAW264.7 cells. *p < .05 versus LPS group. Each data
point represents the mean ± SD (n = 3).
3.4. Effect of EGCG on LPS‐induced NO
After incubating RAW264.7 cells with LPS or LPS + EGCG for 24 h, the
nitrite concentration in the culture solution was measured by the
Griess method to characterize the content of NO. Compared with the
control group, LPS can significantly increase the NO content by 55.55%.
EGCG has a dose‐dependent effect on LPS‐induced NO production. EGCG at
10.9 μmol/L and 21.8 μmol/L failed to inhibit the production of NO, and
there was no significant difference from the LPS group. 43.6 μmol/L and
87.2 μmol/L EGCG inhibited the production of NO induced by LPS, the
content was almost reduced to the control group, which was
significantly lower than that of the LPS group, and there was no
significant difference from the control group. At the same time, the
positive control (DEX) also decrease the NO content by about 21 μM,
which is 25% lower than the LPS group (Figure [87]5).
FIGURE 5.
FIGURE 5
[88]Open in a new tab
RAW264.7 macrophage cells (1 × 10^5 cells) cultured and treated with
EGCG and EGCG+LPS and DEX for 24 h. The effect of EGCG on LPS‐induced
NO expression in RAW264.7 cells. Each data point represents the
mean ± SD (n = 3).
3.5. EGCG inhibits ROS production in RAW264.7 macrophages
As can be seen from Figure [89]6, EGCG can inhibit the increase of ROS
induced by LPS, and there is a certain dose–effect relationship. The
content of ROS in the LPS group increased twice that of the control
group, which significantly increased the expression of ROS (p < .05).
DEX significantly inhibited the production of ROS induced by LPS with a
content of about 100%, which was not significant. The relative ROS
levels of 10.9 μmol/L and 21.8 μmol/L in the EGCG group were about 200%
and 170%, respectively, and there was no significant difference from
the LPS group. The relative ROS levels of the 43.6 μmol/L and
87.2 μmol/L EGCG groups were about 150% and 100%, respectively, which
were significantly lower than those of the LPS group (p < .05).
FIGURE 6.
FIGURE 6
[90]Open in a new tab
RAW264.7 macrophage cells (1 × 10^5 cells) cultured and treated with
EGCG and EGCG+LPS and DEX for 24 h. The effect of EGCG on LPS‐induced
ROS expression in RAW264.7 cells. ^# p < .05 versus control group;
*p < .05 versus LPS group. Each data point represents the mean ± SD
(n = 3).
3.6. Transcriptome analysis
A total of 23,939 genes were detected in this experiment, including
7346 differential genes. Principal component analysis (PCA) of
differential genes found that the three groups are distributed in
different areas, and each group does not interfere with each other,
which can clearly distinguish the control group, LPS group, and
LPS + EGCG group (Figure [91]7).
FIGURE 7.
FIGURE 7
[92]Open in a new tab
Principal component analysis of differential genes of control, LPS, and
LPS + EGCG group.
At the same time, the correlation analysis revealed that the
correlation between the control group and the LPS group was only about
0.7, the correlation between the LPS group and the LPS + EGCG group was
about 0.78, and the correlation between the control group and the
LPS + EGCG group was as high as about 0.91. This indicates that after
EGCG treatment, the overall gene expression trend of cells tends to be
that of normal cells (Figure [93]8).
FIGURE 8.
FIGURE 8
[94]Open in a new tab
Correlation analysis of differential genes of control, LPS, and
LPS + EGCG group.
In the differential gene expression volcano diagram, there are 3677
differential genes between the control group and the LPS group,
including 1703 upregulated genes (red dots) and 1974 downregulated
genes (green dots); there are 4094 differential genes in the LPS group
and the EGCG + LPS group. Among them, there are 1860 upregulated genes
and 2234 downregulated genes. The control group and the LPS + EGCG
group have a total of 5686 differential genes, 2556 upregulated genes,
and 3130 downregulated genes (Figure [95]9).
FIGURE 9.
FIGURE 9
[96]Open in a new tab
Volcano plot. Distinct transcriptome profile obtained after treating
RAW 264.7 macrophages with EGCG, determined by RNA‐seq. Horizontal
coordinates represent variations in mRNA expression levels.
Longitudinal coordinates represent significant changes in mRNA
expression levels. Red and green dots in the Volcano plot indicate
mRNAs with increased and decreased expression levels.
To study the effect of EGCG on the biological process (BP), cellular
component (CC), and molecular function (MF) of RAW264.7 cells, the GO
secondary function was obtained using the GO database Gene enrichment.
Figure [97]9 shows that 24 BP, 18 MF, and 19 CC‐related processes are
involved. In BP, we found that differential genes differ from all genes
in terms of metabolic processes, multicellular biological processes,
signals, immune system processes, biological stages, detoxification,
and cell killing. In terms of CC, there are differences in gene
enrichment in nucleoids, viruses, organelle parts, macromolecular
complexes, extracellular regions, and luminal parts enclosed by
membranes. In MF, the difference in gene transduction, molecular
transformation, translation regulation, antioxidant activity, catalytic
activity, and other genes are significantly different from those of all
genes (Figure [98]10).
FIGURE 10.
FIGURE 10
[99]Open in a new tab
Gene ontology (GO) classifications of DEGs across three comparisons.
The Y‐axis represents the number of DEGs in a category. The results are
divided into three main categories: biological process (BP), cellular
component (CC), and molecular function (MF).
Analysis of differentially expressed genes (DEGs) revealed that the
control group and LPS group had 452 differential genes, the control
group and LPS + EGCG group had 1073 differential genes, and the LPS
group and LPS + EGCG group had 568 differential genes. The three groups
involved 858 common differential genes, indicating that LPS induces
changes in these genes, and EGCG will also regulate gene expression
changes caused by LPS. After that, we further analyzed the differential
genes shared by the three groups, which laid the foundation for
studying the possible ways of EGCG (Figure [100]11).
FIGURE 11.
FIGURE 11
[101]Open in a new tab
Venn diagram of differential genes. Venn diagram representing total
number of genes identified among the groups.
GO enrichment analysis showed that the three groups of differential
genes were enriched in the GO function annotations for the 20 most
significant related functions. Among them, 115 differential genes are
involved in the Adenosine triphosphate (ATP) binding process. Secondly,
the innate immune response involves 20 differential genes, the cell
response to LPS involves 17 differential genes, the cell response to
interferon‐β involves 11 differential genes, and the cell response to
interferon‐α involves five differential genes, the regulation of
phagocytosis involves four differential genes. These processes are
closely associated with the inflammatory immune response of macrophage
cells (Figure [102]12).
FIGURE 12.
FIGURE 12
[103]Open in a new tab
Analysis of GO pathway enrichment of differential genes. Gene ontology
(GO) analysis of significant genes. Bar plots displaying enriched
biological processes, cellular components, and molecular function. The
plots show significantly enriched GO terms.
We conducted KEGG enrichment analysis on the three groups of
differential genes, and then evaluated the enrichment degree of KEGG by
enrichment factor (rich factor), q‐value, and gene quantity and
displayed the top 20 most enriched signal pathways. The greater the
enrichment factor, the more significant the enrichment level of
differential genes in this pathway. Among them, the classic
inflammation pathway NF‐κB is the most significant among the
inflammation‐related pathways, and its enrichment degree is about 3.1.
Secondly, there are the TNF‐alpha signaling pathway, Toll‐like receptor
signaling pathway, etc. (Figure [104]13).
FIGURE 13.
FIGURE 13
[105]Open in a new tab
Analysis of KEGG pathway enrichment of differential genes. The color
depth of nodes refers to the p‐value. The size of nodes refers to the
number of genes.
Heatmap analysis of the NF‐κB pathway shows that EGCG inhibits
LPS‐induced upregulation of Ccl4, Cxcl2, Ptgs2, TNF, Bcl21, Card11,
Ltb, and Plcg1. Among them, Ptgs2 is a promoter of tumor and cancer
formation, and Cxcl2 is a chemokine, which is closely related to
inflammation. CD40 can activate PI3K, Rel/NF‐κB transcription factors,
induce proteins such as Bcl‐xL and Cdk4, and inhibit Pidd1. p53 induces
death domain proteins, causing inflammation, etc. In the TNF signaling
pathway, LPS induces upregulation of Ptgs2, Cxcl2, TNF, Tnfsf13b,
Socs3, lfi47, Mapk11, Edn1, Lif, IL‐6, and Cxcl3, and the expression of
these genes is suppressed after EGCG intervention. In the Toll receptor
signaling pathway, EGCG inhibits LPS‐induced upregulation of Ccl3,
Spp1, Ccl4, TNF, STAT1, Irf7, Trl3, Mapk11, IIl12b, and IL‐6, thereby
regulating the Toll‐like receptor signaling pathway to suppress
inflammation. Tollip is a negative regulator of TLR4. EGCG can
significantly increase its expression and inhibit the overexpression of
TLR4 (Figure [106]14).
FIGURE 14.
FIGURE 14
[107]Open in a new tab
Heatmap analysis of differential genes involved in NF‐κB, TNF, and
Toll‐like receptor signaling pathways.
3.7. EGCG inhibited LPS‐induced cytokine expression
Raw 264.7 macrophage cells were stimulated with LPS and treated with
various concentrations of EGCG; after which IL‐1β, IL‐6, TNF‐α, and
iNOS expression were measured using RT‐PCR. Compared with the control
group, LPS‐induced RAW264.7 cells exhibited IL‐1β mRNA expression over
100 times in 24 h. EGCG and DEX at various concentrations significantly
inhibited IL‐1β mRNA expression (p < .01), the expression level is
below 50 times (Figure [108]15a). The expression of IL‐6 in the LPS
group was as high as 16,000 times, and EGCG at various concentrations
significantly inhibits the expression of IL‐6. However, DEX treatment
is not effective in inhibiting IL‐6 expression (Figure [109]15b). LPS
treatment increased the TNF‐α level by 2700 times compared to the
control group. The inhibitory effect of EGCG at various concentrations
on LPS‐induced TNF‐α mRNA expression was very significant, it almost
normalized the level (p < .001) (Figure [110]15c). iNOS level increased
by 4000% in the LPS treatment group compared to normal control.
21.8 μmol/L EGCG rather increased iNOS level instead of lowering,
43.6 μmol/L, and 87.2 μmol/L EGCG and DEX significantly inhibited iNOS
mRNA expression, which is significantly different from the LPS group
(p < 0.05) (Figure [111]15d).
FIGURE 15.
FIGURE 15
[112]Open in a new tab
EGCG reduces cytokine expression. (a). IL‐1β, (b). IL‐6, (c). TNF‐α,
and (d). iNOS. Data are represented as the mean ± SD (n = 3). # p < .01
versus control group, ^## p < .01 versus control group, *p < .05 versus
LPS group, **p < .01 versus LPS group, ***p < .001 versus LPS group.
By measuring the mRNA expression of Arg‐1 and CD206 (these are key
markers of M2 macrophages) to study whether EGCG can regulate the
transformation of RAW264.7 cells to M2 type. Under LPS stimulation,
43.6 μmol/L EGCG elevate CD206 mRNA expression by seven folds and
87.2 μmol/L by nine folds. Positive control, DEX (25.48 μmol/L)
significantly increase CD206 expression levels by two folds. For Arg‐1,
43.6 μmol/L EGCG has no significant effect after LPS stimulation but
87.2 μmol/L EGCG significantly increased the mRNA expression of Arg‐1
by about five times (p < .05) (Figure [113]16).
FIGURE 16.
FIGURE 16
[114]Open in a new tab
EGCG alters mRNA expression of cell surface markers. (a). CD206, and
(b). Arg‐1. Data are represented as the mean ± SD (n = 3). ^# p < .01
versus control group, *p < .05 versus LPS group, **p < .01 versus LPS
group, ***p < .001 versus LPS group.
3.8. Effect of EGCG on LPS‐induced NF‐κB signaling pathway
The expression of IκB in the LPS group was significantly lower than
that in the EGCG+LPS group, while the expression of activated P‐IκB was
significantly higher than that in the EGCG+LPS group (P < .05). At the
same time, the ratio of P‐IκB/IκB in the EGCG+LPS group was 75% lower
than that of the LPS group, indicating that EGCG could significantly
inhibit LPS‐induced activation of IκB, thereby inhibiting the
activation of NF‐κB (p < .05) (Figure [115]17).
FIGURE 17.
FIGURE 17
[116]Open in a new tab
EGCG significantly inhibits LPS‐induced activation of IκB. Data are
represented as the mean ± SD (n = 3). *p < .05 versus LPS group. The
beta Actin band in this figure is the same band in Figure [117]20a
because we have separated them in the same membrane.
Similarly, the expression level of P65 in the EGCG+LPS group was about
twice that of the LPS group, while the expression level of P‐P65 was
only about 2/3 of that in the LPS group. The expression of P‐P65/P65 in
the EGCG+LPS group was significantly lower than that in the LPS group,
about 1/2 times that of the LPS group, indicating that EGCG can
significantly inhibit the activation of P65 induced by LPS, thereby
exerting an inflammatory protective effect (p < .05) (Figure [118]18).
However, EGCG could not significantly increase the expression of PPARγ
(Figure [119]19).
FIGURE 18.
FIGURE 18
[120]Open in a new tab
EGCG significantly inhibits the activation of P65 induced by LPS. Data
are represented as the mean ± SD (n = 3). *p < .05 versus LPS group.
The beta Actin band in this figure is the same band in Figure [121]19b
because we have separated them in the same membrane.
FIGURE 19.
FIGURE 19
[122]Open in a new tab
Effects of EGCG on PPARγ. Data are represented as the mean ± SD
(n = 3).
Compared with the control group, LPS did not significantly affect the
protein expression of 67LR. Similarly, there was no significant
difference between the LPS group and the LPS + EGCG group as well
(Figure [123]20a). LPS treatment elevated TLR4 expression, but it was
also not statistically significant. Compared with the LPS group, EGCG
downregulated the expression of TLR4 significantly (p < .05), and the
expression level is about 70% of the LPS group, indicating that EGCG
can exert an inflammatory protective effect by inhibiting the
expression of TLR4 (Figure [124]20b).
FIGURE 20.
FIGURE 20
[125]Open in a new tab
Effects of EGCG on 67LR (a) and TLR4 (b). Data are represented as the
mean ± SD (n = 3). *p < .05 versus LPS group. The beta Actin band in
this figure is the same band in Figure [126]18. The beta Actin band in
this figure is the same band in Figure [127]19 because we have
separated them in the same membrane.
4. DISCUSSION AND CONCLUSION
EGCG is the major part of green tea polyphenol that suppresses
inflammation, oxidation, tumor, and apoptosis. In the present study, we
demonstrated that EGCG, a bioactive polyphenol in green tea, suppressed
the expression of LPS‐induced inflammatory cytokines in Raw 264.7
macrophage cells by mediating TLR4 and NF‐κB signaling pathways.
Neutrophils are the key player in the body's defense to combat
inflammation. These neutrophils extrude pseudopods while performing
their innate task. Pseudopods have many forms and their size and shape
depend on the degree of polymerization of actin filaments (Rocheleau et
al., [128]2016). EGCG significantly inhibited the formation of cell
pseudopods, and the effect was better than DEX. Cui et al. ([129]2019)
also observed similar morphological changes in macrophages. The cells
in the control group aggregated and showed a round shape. After LPS
treatment, the cell adhesion increased and the body shape increased,
forming a long and slender pseudopod protrusion.
Macrophages participate actively in the immune response to fight
foreign stimuli, pathogens, and damaged cells by engulfing them. In
this experiment, EGCG can significantly reduce the phagocytic capacity
of RAW264.7 cells after LPS stimulation. This may be because, after
EGCG intervention, it significantly reduced inflammatory factors such
as IL‐1β and IL‐6, reducing the damage of the inflammatory response to
the body by the ability of macrophages to engulf harmful stimuli also
decreases. When Bougainvillea xbuttiana extract inhibited LPS‐induced
macrophage phenotypic transformation, it also showed that Bougainvillea
xbuttiana extract had a similar effect on the phagocytic capacity of
RAW264.7 cells (Arteaga Figueroa et al., [130]2017).
At the same time, EGCG can significantly inhibit the abnormal increase
of ROS induced by LPS. In our experiment, the inhibitory effect of EGCG
at various concentrations on NO content was consistent with iNOS mRNA
expression. This experiment also confirmed that it can reduce body
inflammation and reduce the expression of most inflammatory factors.
However, the results of this experiment show that DEX does not inhibit
the expression of IL‐6 mRNA induced by LPS. It is undeniable that its
anti‐inflammatory effect may be because DEX is not sensitive to IL‐6,
mainly by inhibiting other inflammatory factors to produce an
anti‐inflammatory effect. Similarly, a large number of studies have
also used DEX as a positive control. Al‐Harbi et al. ([131]2016) also
found that DEX could not downregulate the increase of IL‐6 content in
Balb/c mice induced by LPS. Muniandy et al. ([132]2018) also found that
DEX intervention increases the content of IL‐6 in RAW264.7 cells, which
is about four times that of the LPS group which is similar to our
findings (Muniandy et al., [133]2018).
We performed transcriptome sequencing analysis on three sets of cell
samples (control, LPS, and LPS + EGCG), and 7346 differential genes
were obtained. The three groups of differential genes were screened and
analyzed based on GO function enrichment terms. The results show that
the ATP binding pathway is significantly affected. In mitochondria,
adenosine diphosphate (ADP) is consumed to produce ATP and oxygen
consumption is blocked to produce O[2−], and superoxide anions play a
central role in ROS production (Piechota‐Polanczyk &
Fichna, [134]2014). About 90% of ROS in cells are produced by
mitochondrial oxidative phosphorylation and imbalance of these leads to
the release of inflammatory factors such as TNF‐α and IL‐1β, triggering
innate immunity, eventually causing immune responses (Gao et
al., [135]2008; Yamauchi et al., [136]^2008). The activation of the
TLR‐mediated signaling pathway is also closely related to ROS
generation. In LPS/TLR4‐mediated inflammation, inhibition of ROS
production helps reduce LPS‐induced NF‐κB activation (Ryan et
al., [137]2004). Therefore, a reasonable adjustment of the relationship
between ROS and ATP, and the balance of ROS production can help
suppress inflammation and maintain the health of the body.
KEGG enrichment analysis revealed that the NF‐κB pathway was
significantly affected in the common differential genes. KEGG
enrichment analysis found that NF‐κB, a classic inflammation‐related
pathway, is significantly higher (involved total of 15 differential
genes), and the TNF signaling pathway and TLR signaling pathway
involved 15 and 14 differential genes, respectively. He et al.
([138]2017) and Wang et al. ([139]2014) reviewed the signal pathways
involved in macrophage polarization, including STAT, NF‐κB, PPARγ, IRF,
etc. (He et al., [140]2017; Wang et al., [141]^2014), which is
consistent with our sequencing results. At the same time, it also has a
significant impact on the TNF signaling pathway and the TLR signaling
pathway. TLRs are pattern recognition receptors distributed on the
surface of B‐lymphocytes, macrophages, and other cells, and exclusively
expressed in the natural immune system. The activation of TLRs can
induce an immune response, which is beneficial to the body's resistance
to pathogenic microorganisms, but excessive activation induces the
overexpression of inflammatory factors and aggravates the development
of immune diseases. At the same time, the combination of LPS and TLR4
causes the expression of TNF‐α, which also causes the activation of the
TNF signaling pathway.
In this experiment, there was no significant difference between the
67LR protein expression of the LPS group and the LPS + EGCG group. At
the same time, EGCG significantly inhibited LPS‐induced TLR4 expression
(p < .05). Similar to our findings, Bao et al. ([142]2015) showed that
in LPS‐induced 3 T3‐L1 adipocytes, 67LR protein expression remains
unaffected after EGCG treatment (Bao et al., [143]2015). Baek et
al. ([144]2019) found that EGCG can significantly downregulate the
expression of TLR4 in human aortic endothelial cells (HAECs) induced by
LPS (Baek et al., [145]2019). Byun et al. ([146]2010) showed in
macrophages that EGCG alone could downregulate the protein expression
of TLR4 for 24 h, and EGCG did not inhibit the expression of TNF‐α and
other inflammatory factors after incubating the cells with 67LR (Byun
et al., [147]2010). This indicates that LPS does not stimulate the
cells to cause inflammation by reducing the expression of 67LR but it
may combine with TLR4 to produce more factors that are inflammatory and
induce inflammation. At the same time, EGCG may play an
anti‐inflammatory role by reducing the expression of TLR4, accelerating
the degradation of TLR4, and affecting the downstream signal pathway of
TLR4 (Kumazoe et al., [148]2017). NF‐κB can induce the transformation
of macrophages to M1 type, produce a large number of inflammatory
factors, and promote the body's inflammatory response. The experimental
results show that EGCG can significantly inhibit LPS‐induced IκB
activation, and the ratio of P‐IκB/IκB is 1/4 times that of the LPS
group (p < .05); at the same time, EGCG can also significantly inhibit
LPS‐induced P65 activation, the expression level of P‐P65/P65 was about
1/2 times that of the LPS group (p < .05). Li et al. ([149]2017) also
observed that EGCG can inhibit the expression of NF‐κB in macrophages,
thereby inhibiting the expression of matrix metalloproteinase‐9 and
monocyte chemoattractant protein‐1 (Li et al., [150]2017). Similarly,
Joo et al. ([151]2012) also reported that EGCG can inhibit the
expression of nuclear protein NF‐κB P65 induced by LPS in
bone‐marrow‐derived macrophages (BMMs) (Joo et al., [152]2012). This
indicates that EGCG can significantly inhibit the activation of NF‐κB
and thus suppress cell inflammation. PPARγ is closely associated with
inflammation, and it can exert anti‐inflammatory effects by inhibiting
inflammatory signaling pathways such as NF‐κB, activator protein‐1
(AP‐1), JAK–STAT, etc. (Bright et al., [153]2008; Dana et
al., [154]^2019). PPARγ can interact with P65 to prevent protein
activation by inhibiting the activation of NF‐κB (Yin et
al., [155]2014). In this experiment, the EGCG+LPS group could not
significantly increase the expression of PPARγ. The difference is that
Jin et al. ([156]2020) found that fucoxanthinol can significantly
upregulate the downregulated PPARγ expression in LPS‐induced macrophage
cells, thereby, directly and indirectly, downregulating NF‐κB to
suppress inflammation (Jin et al., [157]2020). This shows that there
may be differences in the action pathways of EGCG and fucoxanthinol on
macrophages. In macrophages, EGCG may not indirectly inhibit NF‐κB
activation by upregulating PPARγ and regulating the conversion of
macrophages between M1 and M2 types.
In conclusion, the present study demonstrated that EGCG administration
protected against LPS‐induced inflammation in Raw 264.7 macrophage
cells. EGCG inhibited LPS‐induced enhancement of cell phagocytosis and
the upregulation of pro‐inflammatory factors. EGCG can significantly
inhibit LPS‐induced upregulation of P‐IκB and P‐P65, and inhibit the
activation of the NF‐κB pathway. Additionally, EGCG does not upregulate
the expression of PPARγ, indicating that EGCG may not promote the
transformation of macrophages to M2 type by increasing the expression
of PPARγ. The action mechanism of EGCG is yet to be fully elucidated
and needs more exploration in both in vivo and in vitro models. We
conclude that EGCG may not promote the transformation of M1‐type
macrophages to M2 type by increasing the expression of PPARγ. Combined
with transcriptome data, further study is needed to grasp what
substances and signal pathways EGCG regulates to play a role. At the
same time, in vivo experiments are needed to further study the function
and mechanism of EGCG.
AUTHOR CONTRIBUTIONS
Imam Hossen: Conceptualization (equal); formal analysis (equal);
investigation (equal); methodology (equal); validation (equal); writing
– review and editing (equal). Zhang Kaiqi: Conceptualization (equal);
data curation (equal); formal analysis (equal); investigation (equal);
methodology (equal); software (equal); validation (equal). Wu Hua:
Conceptualization (equal); funding acquisition (equal); investigation
(equal); methodology (equal); resources (equal); software (equal);
supervision (equal); visualization (equal); writing – review and
editing (equal). Junsong Xiao: Conceptualization (equal); data curation
(equal); formal analysis (equal); funding acquisition (equal);
investigation (equal); methodology (equal); project administration
(equal); resources (equal); supervision (equal); writing – original
draft (equal); writing – review and editing (equal). Mingquan Huang:
Methodology (equal); resources (equal); software (equal); validation
(equal). Yanping Cao: Funding acquisition (equal); project
administration (equal); resources (equal); supervision (equal).
Acknowledgements