Transcriptional Regulation of Human Arylamine N-Acetyltransferase 2 Gene by Glucose and Insulin in Liver Cancer Cell Lines

Abstract

Arylamine N-acetyltransferase 2 (NAT2) is well-known for its role in phase II metabolism of xenobiotics and drugs. More recently, genome wide association studies and murine models implicated NAT2 in regulation of insulin sensitivity and plasma lipid levels. However, the mechanism remains unknown. Transcript levels of human NAT2 varied dynamically in HepG2 (hepatocellular) cells, depending on the nutrient status of the culture media. Culturing the cells in the presence of glucose induced NAT2 mRNA expression as well as its N-acetyltransferase activity significantly. In addition, insulin or acetate treatment also significantly induced NAT2 mRNA. We examined and compared the glucose- and acetate-dependent changes in NAT2 expression to those of genes involved in glucose and lipid metabolism, including FABP1, CPT1A, ACACA, SCD, CD36, FASN, ACLY, G6PC, and PCK1. Genes that are involved in fatty acid transport and lipogenesis, such as FABP1 and CD36, shared a similar pattern of expression with NAT2. In silico analysis of genes co-expressed with NAT2 revealed an enrichment of biological processes involved in lipid and cholesterol biosynthesis and transport. Among these, A1CF (APOBEC1 complementation factor) showed the highest correlation with NAT2 in terms of its expression in normal human tissues. The current study shows, for the first time, that human NAT2 is transcriptionally regulated by glucose and insulin in liver cancer cell lines and that the gene expression pattern of NAT2 is similar to that of genes involved in lipid metabolism and transport.

Arylamine N-acetyltransferase 2 (NAT2) and its isozyme, arylamine N-acetyltransferase 1 (NAT1), play key roles in the detoxification of carcinogenic arylamines and xenobiotics by catalyzing acetyl-coenzyme A (CoA)-dependent biotransformation of arylamines to N-arylacetamides (Sim et al., 2014). NAT2 expresses a well-defined genetic polymorphism in humans that translates into differential xenobiotic metabolism (Fathzadeh et al., 2017). Based on the combination of alleles, individuals can be categorized into 3 distinct acetylator phenotypes (ie, rapid, intermediate and slow acetylators) and exhibit clinically meaningful differences in ability to metabolize NAT2 substrates, including drugs (eg, isoniazid) and carcinogens (eg, 4-aminobiphenyl) (Hein, 2002; Sim et al., 2014).

More recently, a novel and unexpected role of NAT2 has been reported. Genome-wide association studies (GWAS) have identified links between polymorphisms of human NAT2 gene and insulin resistance, high serum triglyceride, and coronary artery disease as well as high fasting plasma glucose level (Fathzadeh et al., 2017; Knowles et al., 2015). Analysis of Nat1 knockout (KO) mice revealed that NAT1 enzyme deficiency leads to development of insulin resistance and metabolic defects at the systemic level, further supporting the previously unrecognized role of human homologue NAT2 in regulating metabolism and insulin sensitivity (Camporez et al., 2017; Chennamsetty et al., 2016). Moreover, the follow-up studies have attributed insulin resistance in Nat1 KO mice to mitochondrial dysfunction (Camporez et al., 2017; Chennamsetty et al., 2016). However, it is currently unknown how human NAT2 influences cellular metabolism and insulin sensitivity, or if NAT2 expression or activity is reciprocally regulated by the energy or metabolic state.

Endogenous NAT2 expression is limited to the liver and small and large intestines (Sim et al., 2014), suggesting that NAT2 plays organ-specific roles. For this reason, we have been working with human hepatocellular carcinoma cell lines, including HepG2, to study the role of human NAT2 in the liver, where its expression is the highest. In our preliminary experiments, we cultured HepG2 cells in culture media containing different nutrients (eg, glucose, pyruvate, and l-glutamine, etc.) and discovered that NAT2 mRNA level fluctuated dynamically in HepG2 cells depending on the availability of certain nutrients, suggesting that NAT2 expression is differentially regulated by energy or metabolic state of the cells. To our knowledge, no previous studies have reported such finding. In the present study, we report the effects of glucose and insulin on NAT2 expression in HepG2 cells. In addition, the glucose-dependent expression pattern of NAT2 was compared with those of genes involved in glucose and lipid metabolism. Finally, to better understand the transcriptional regulation and novel cellular roles of NAT2, we performed an in silico analysis of genes co-expressed with NAT2 in human tissues.

MATERIALS AND METHODS

 

 

Cell culture

HepG2 and Hep3B cells were obtained from ATCC and cultured in DMEM (Cytiva) containing d-glucose (5.5 mM), sodium pyruvate (1 mM), l-glutamine (4 mM), fetal bovine serum (FBS; 10%), and penicillin-streptomycin (1%) at 37°C with 5% CO₂ in humidified air.

Reagents and treatments

d-glucose, sodium pyruvate, l-glutamine, dexamethasone, sodium acetate, and insulin (bovine pancreas; 10 mg/ml solution) were purchased from Sigma-Aldrich. Cell membrane-permeable, 8-bromo-cAMP (sodium salt) was purchased from Tocris. The final treatment concentrations of individual reagents used in the study were the following: d-glucose (5 mM), pyruvate (1 mM), l-glutamine (4 mM), 8-bromo-cAMP (500 µM), dexamethasone (1 µM), acetate (5 mM), insulin (100 nM), and FBS (10%). For treatments, cells were plated on either 12-well or 24-well plates the day before the treatment. For the treatment with selected nutrients (eg, d-glucose), the nutrients (either individually or in combination) were added to a control DMEM (“Control”) which was free of d-glucose, pyruvate, l-glutamine, FBS, and phenol red (Cytiva), and the cells were treated for 24 h. For one of the experiments, HepG2 cells were treated in the absence or presence of insulin (100 nM final) for 6 h in regular, serum-free DMEM which contained 5.5 mM d-glucose, 1 mM sodium pyruvate, and 4 mM l-glutamine. Bovine serum albumin (BSA, heat shock fraction, protease-free, fatty acid-free, and essentially globulin-free, pH 7, ≥98%) and sodium palmitate were purchased from Sigma-Aldrich. BSA (10%)-Palmitic acid (5 mM) (BSA-PA) complex stock solution was prepared as described previously (Cousin et al., 2001). Fatty acid-free BSA solution (10%) was used for vehicle control. For the palmitate treatment, HepG2 cells were incubated with BSA (0.5%)-PA (0.25 mM) or 0.5% BSA alone in Control DMEM for 24 h.

RNA isolation and RT-qPCR

Cells were approximately 50%–70% confluent at the time of harvest. For isolation of total RNA, E.Z.N.A. Total RNA Kit I (Omega Bio-Tek) was used according to the manufacturer’s instructions with an added, in-column DNase digestion step. RNA concentration was measured based on absorbance at 260 nm using NanoDrop spectrophotometer (Thermo Scientific). For cDNA synthesis, 500 ng total RNA was processed in a 10-µl reaction using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s instructions. For qPCR, 1 µl of cDNA, the gene-specific primer set, and iTaq Universal SYBR Green Supermix (Bio-Rad) were used in a 20-µl reaction. For PCR amplification and quantification of SYBR Green signal, StepOne Real-Time PCR System (Applied Biosystems) was used. Samples underwent 40 cycles of 2-step PCR (denaturation at 96°C [15 s] followed by annealing and extension at 60°C [1 min]). For calculation of relative levels of the gene of interest, delta-delta Ct method (a.k.a., the 2^–ΔΔCt method) was used using 18S rRNA as an internal control. The sequences of the PCR primers used in the current study are available in Table 1.

Table 1.

List of qPCR Primer Sequences Used in the Study

Gene (Human)	Description		Sequence
18S rRNA	Internal control	Forward	GGAAGGGCACCACCAGGAGT
		Reverse	TGCAGCCCCGGACATCTAAG
ACACA	Acetyl-CoA carboxylase alpha	Forward	CTGGCTGTCTTACATGCCCA
		Reverse	TAGCCTGGCTCTACCAACCA
ACLY	ATP citrate lyase	Forward	GGTGCTCCGGATTTTGC
		Reverse	ACATGGCTGCAGAGAGACCT
ACOX1	Acyl-CoA oxidase 1	Forward	GTAGCAGTCTGGCCAACCAT
		Reverse	GCTCCCCTGAAGGAAATCCC
ACSS1	Acyl-CoA synthetase short chain family member 1	Forward	ACCAAGATCGCCAAATATGC
		Reverse	TGCTTGTCCTTGCACTTCTG
ACSS2	Acyl-CoA synthetase short chain family member 2	Forward	GGATTCCAGCTGCAGTCTTC
		Reverse	CATGCCACCACAAGTCAATC
ACTB	Actin beta; β-actin	Forward	GCAGTCGGTTGGAGCGAGCA
		Reverse	ATCACCTCCCCTGTGTGGACTTGG
CD36	Fatty acid translocase (FAT)	Forward	CAGCCTCATTTCCACCTTTTGT
		Reverse	GCAAAGGCCTTGGATGGAAGA
CPT1A	Carnitine palmitoyltransferase 1A; CPT1	Forward	GATGAGTCGTGCCACCAAGA
		Reverse	GCTCTTGCTGCCTGAATGTG
FABP1	Fatty acid binding protein 1	Forward	GGGAAGGGAGCCCCCTATAA
		Reverse	TGGATCACTTTGGACCCAGC
FASN	Fatty acid synthase (FAS)	Forward	CACACACGATGGACCCTCAG
		Reverse	GAAGAAGGAGAGCCGGTTGG
NR1H4	Nuclear receptor subfamily 1 group H member 4; Farnesoid X-activated receptor (FXR)	Forward	GACCTCGACAACAAAGTCATGC
		Reverse	ATAGCTTCAACCGCAGACCC
NR1I2	Nuclear receptor subfamily 1 group I member 2; pregnane X receptor (PXR)	Forward	ATGGGCCATCTGGGGTCTAT
		Reverse	ATGATGTGGCCGGAATGAGG
G6PC	Glucose-6-phosphatase catalytic subunit 1	Forward	ACGAATCTACCTTGCTGCTCA
		Reverse	AAAATCCGATGGCGAAGCTG
GAPDH	Glyceraldehyde-3-phosphate dehydrogenase	Forward	GGTGAAGCAGGCGTCGGAGG
		Reverse	GAGGGCAATGCCAGCCCCAG
NAT1	N-Acetyltransferase 1	Forward	CACTGTTTGGTGGGCTTCAC
		Reverse	GCACAAGCTTTCTCTGCAAGG
NAT2	N-Acetyltransferase 2	Forward	TGGACCAAATCAGGAGAGAGC
		Reverse	GCCCACCAAACAGTAAACCC
PCK1	Phosphoenolpyruvate carboxykinase 1; PEPCK-C	Forward	TGATGAGCCGCTAGCTTCAG
		Reverse	GCCTTTATGTTCTGCAGCCG
PPARGC1A	PPARG coactivator 1 alpha; PGC1α	Forward	CACGGACAGAACTGAGGGAC
		Reverse	TTCGTTTGACCTGCGCAAAG
SCD	Stearoyl-CoA desaturase; SCD1	Forward	GCAGCCGAGCTTTGTAAGAG
		Reverse	GTTCTACACCTGGCTTTGGG

Note: The table shows the complete list of all RT-qPCR primers used in the current study. All primer sets target corresponding genes in human. Primers were synthesized at Integrated DNA Technologies.

Western blot

The cells were lysed in Laemmli buffer (50 mM Tris-Cl [pH 6.8], 2% sodium dodecyl sulfate [SDS; w/v], 0.1% bromophenol blue, 10% [v/v] glycerol) and boiled for 10 min. Protein concentrations were determined using Pierce BCA Assay kit (Thermo Scientific) per manufacturer’s instructions. β-mercaptoethanol was then added to each sample and boiled for additional 5 min. Fifty micrograms of protein per sample was loaded and separated on a 4%–12% gradient Bis-Tris Plus polyacrylamide gel (Invitrogen). The gel was transferred to a PVDF membrane and blocked in 5% (w/v) skim milk in tris-buffered saline containing 0.1% Tween 20 (TBST) for 30 min. Membranes were incubated with primary antibody at 4°C overnight. Membranes were washed with TBST and then were incubated for 1 h at room temperature with HRP-conjugated secondary antibodies (1:5000 in TBST). Membranes were washed and the protein-antibody complex was detected using Clarity MAX ECL substrate (BioRad). Monoclonal rabbit anti-NAT2 antibody (ab194114) was purchased from Abcam and used at 1:1000 dilution. The antibody against GAPDH was purchased from Cell Signaling Technology and used at 1:5000 dilution.

In vitro NAT2 assay

The N-acetylated product of the NAT2-selective substrate sulfamethazine (SMZ) was quantified by separation of substrate and acetylated product by reverse phase high-performance liquid chromatography (HPLC) as described (Hein et al., 2006). Briefly, treated HepG2 cells were lysed in a lysis buffer containing 20 mM sodium phosphate pH 7.4, 1 mM EDTA, 0.2% Triton X-100, 1 mM DTT, 100 µM PMSF, 1 μg/ml aprotinin, and 2 µM pepstatin A (MilliporeSigma). Cell lysate was centrifuged at 15 000 × g for 20 min in the cold room, and the resulting supernatant was used for the assay. NAT2 enzymatic assay reactions containing 50 µl of cell lysate, 300 µM SMZ, and acetyl-CoA (1 mM), were incubated at 37°C for 10 min. Reactions were terminated by the addition of 10 µl of 1 M acetic acid, then proteins were precipitated by centrifugation (15 000 × g for 10 min) and supernatant was injected onto a C-18 (250 mm × 4 mm; 5 µm) reverse phase column using an Agilent 1260 Infinity II system (Agilent Technologies). Separation of N-acetyl-SMZ was achieved using a gradient of 100:0 of 55 mM sodium phosphate pH 4.0: Methanol to 0:100 55 mM sodium phosphate: Methanol over 20 min. Absorbance was recorded at 260 nm. In vitro NAT2 activity was normalized to total protein (and to incubation time) and expressed as N-acetylated-SMZ nmol/min/mg protein. Each value was then normalized to the control (no glucose) group and expressed as a relative level.

NAT2 genotyping of HepG2

Genomic DNA was isolated from pelleted HepG2 cells by using the QIAamp DNA Mini Kit (QIAGEN) according to the manufacturer’s instructions. NAT2 genotype and deduced phenotype were determined as described previously (Doll and Hein, 2001). The analysis showed that HepG2 has NAT2*6A and NAT2*14C alleles, and thus is a slow acetylator.

Data analysis and presentation

Graphical presentation and statistical analysis of the data was performed with Prism (GraphPad). One-way ANOVA with Bonferroni post hoc test was used for multiple group comparisons. For studies involving comparison of 2 experimental groups, unpaired t test was used. Inter-group differences with p < .05 were considered statistically significant.

In silico analysis of genes co-expressed with NAT2

The following 4 different, online analytical engines were used to identify genes whose expression patterns were similar to that of human NAT2 (ie, co-expressed genes): GEPIA (http://gepia.cancer-pku.cn/; last accessed August 2022) (Tang et al., 2017), COXPRESdb (https://coxpresdb.jp/; last accessed August 2022) (Obayashi et al., 2019), SEEK (https://seek.princeton.edu/seek/; last accessed August 2022) (Zhu et al., 2015), and Archs⁴ (https://maayanlab.cloud/archs4; last accessed August 2022) (Lachmann et al., 2018). The list of top 50 genes identified by each engine is available in Table 2. Based on these lists, we compiled a combined gene list by identifying genes that appear more than once in the lists, which resulted in a total of 40 genes (see Table 3). The list of co-expressed genes were then analyzed for enrichment of biological processes they are associated with using GOnet (https://tools.dice-database.org/GOnet/; last accessed August 2022) (Pomaznoy et al., 2018) and PANTHER (http://www.pantherdb.org/; last accessed August 2022) (Mi et al., 2013). The top 50 Gene Ontology (GO) terms/biological processes significantly enriched in the list (q < 0.001) are shown in Table 4. For analyzing and graphing correlations between NAT2 and individual gene expression (shown in Figure 5) in multiple human tissues, the graphing and statistical tools available at GEPIA (http://gepia.cancer-pku.cn/; last accessed August 2022) (Tang et al., 2017) were employed using Genotype-Tissue Expression (GETx) database on normal human tissues.

Table 2.

List of Top 50 Genes Co-Expressed With Human NAT2

	Sources
Rank	Archs4	COXPRESdb	SEEK	GEPIA
1	FLJ22763	MOGAT2	A1CF	CYP2C8
2	OTC	C3orf85	TM4SF5	NR1I2
3	TMEM82	IYD	SLC39A5	MOGAT2
4	FABP1	OTC	LEAP2	KDM8
5	MOGAT2	SLC17A4	SLC10A1	C8A
6	CLRN3	CYP2C9	OTC	SLC25A47
7	PLA2G12B	ADGRG7	FABP1	AFM
8	MTND4P20	FABP1	NR1I2	PLGLA
9	TM4SF5	CLRN3	MTTP	LINC01093
10	GUCA2B	AGXT2	APOA5	RDH16
11	CYP2C19	NR1I2	SLC17A4	MAT1A
12	ADGRG7	NAT1	HPX	F9
13	ABCG8	ALDOB	AFM	C3P1
14	HMGCS2	GLOD5	CYP4F2	CYP2C9
15	NR1H4	APOB	RDH16	GDF2
16	SLC17A4	CYP2C19	CYP3A4	ADH4
17	APOC3	A1CF	ADH4	SMUG1P1
18	CYP2C18	HSD3B2	PRAP1	APOF
19	ALDOB	HMGCS2	C8A	SAA4
20	GBA3	TAT	NPC1L1	ABCG5
21	APOA4	CDH17	CYP3A43	GYS2
22	UGT2A3	GC	ABCG8	CYP4A22
23	PRAP1	LOC285733	APOC3	RP11-622A1.2
24	NR1I2	NR1H4	SLC30A10	PLGLB2
25	SLC2A2	MALRD1	GLYCTK	HP
26	UGT1A4	IGSF5	KHK	CYP8B1
27	TM6SF2	SI	CYP2B6	TTC36
28	LRRC31	FMO4	CYP2C9	PLG
29	SLC28A1	ZG16	APOB	PROZ
30	SLC39A5	GBA3	SLC2A2	APOA5
31	CCL15-CCL14	PPP1R1C	ADH6	F11
32	CYP3A4	ANKS4B	GBA3	CYP4F2
33	ACE2	GLYATL1	CPN2	AGXT
34	AGXT2	ABCC6P1	AGXT	C9
35	ADH1C	LECT2	CYP2C19	HAO1
36	RP11-400G3.5	ZNF503-AS1	C8G	HPX
37	APCS	KNG1	AKR1C4	C6
38	F11	CFHR5	ABCB11	UROC1
39	CFHR2	SLC26A3	HAO1	MBL2
40	ADH6	LOC100509856	ALDOB	GCKR
41	GSTA2	CYP2C18	SERPINA4	KLKB1
42	OIT3	SLC2A2	GYS2	RP11-290F5.1
43	F9	LRRC31	NR1I3	SERPINA10
44	GOLT1A	MS4A12	HRG	RP11-116D2.1
45	GSTA1	SULT1B1	PLA2G12B	C8B
46	RNF186	MEP1B	HNF4A	SLC22A1
47	RP4-608O15.3	F9	MOGAT3	ADH6
48	UGT2B10	UGT2B15	SULT2A1	ACOT12
49	CFHR5	SLC15A1	OSTALPHA	ITIH4
50	AFM	BCO1	CYP2C18	ORM2

Note: Four separate analytical engines (ie, Archs⁴, COXPRESdb, SEEK, and GEPIA) were used to identify top 50 genes that share a similar expression pattern (ie, co-expressed) with human NAT2. Only the gene symbols are shown here. The genes that appear more than once in the lists are shown in Table 3.

Table 3.

List of Genes Co-Expressed With Human NAT2 (in Alphabetical Order)

	Gene ID	Gene Name	Protein Class/Type	No. Hits (out of 4)a
1	A1CF	APOBEC1 complementation factor	RNA metabolism	2
2	ABCG8	ATP-binding cassette sub-family G member 8	ATP-binding cassette (ABC) transporter	2
3	ADGRG7	Adhesion G-protein coupled receptor G7; ADGRG7	G-protein coupled receptor	2
4	ADH4	All-trans-retinol dehydrogenase [NAD(+)] ADH4	Dehydrogenase	2
5	ADH6	Alcohol dehydrogenase 6	Dehydrogenase	3
6	AFM	Afamin	Transfer/carrier protein	3
7	AGXT	Alanine-glyoxylate and serine-pyruvate aminotransferase	Transaminase	2
8	AGXT2	Alanine-glyoxylate aminotransferase 2	Transaminase	2
9	ALDOB	Fructose-bisphosphate aldolase B	Aldolase	3
10	APOA5	Apolipoprotein A-V	Apolipoprotein	2
11	APOB	Apolipoprotein B-100	Apolipoprotein	2
12	APOC3	Apolipoprotein C-III	Apolipoprotein	2
13	C8A	Complement component C8 alpha chain	Complement component	2
14	CFHR5	Complement factor H-related protein 5	Complement component	2
15	CLRN3	Clarin-3; transmembrane Protein 12; TMEM12; USH3AL1	–	2
16	CYP2C18	Cytochrome P450 2C18	Oxygenase	3
17	CYP2C19	Cytochrome P450 2C19	Oxygenase	3
18	CYP2C9	Cytochrome P450 2C9	Oxygenase	3
19	CYP3A4	Cytochrome P450 3A4	Oxygenase	2
20	CYP4F2	Cytochrome P450 4F2	Oxygenase	2
21	F11	Coagulation factor XI	Serine protease	2
22	F9	Coagulation factor IX	Serine protease	3
23	FABP1	Fatty acid-binding protein, liver	Transfer/carrier protein	3
24	FLJ22763	Chromosome 3 open reading frame 85 (c3orf85)	–	2
25	GBA3	Cytosolic beta-glucosidase	Glycosidase	3
26	GYS2	Glycogen [starch] synthase, liver	–	2
27	HAO1	Hydroxyacid oxidase 1	Oxidoreductase	2
28	HMGCS2	Hydroxymethylglutaryl-CoA synthase, mitochondrial	HMG-CoA synthase; ketogenesis	2
29	HPX	Hemopexin; HPX	Metalloprotease	2
30	LRRC31	Leucine-rich repeat-containing protein 31	–	2
31	MOGAT2	2-acylglycerol O-acyltransferase 2	Acyltransferase	3
32	NR1H4	Bile acid receptor; FXR	C4 zinc finger nuclear receptor	2
33	NR1I2	Nuclear receptor subfamily 1 group I member 2; PXR	C4 zinc finger nuclear receptor	4
34	OTC	Ornithine transcarbamylase, mitochondrial	Transferase	3
35	PLA2G12B	Group XIIB secretory phospholipase A2-like protein	Phospholipase	2
36	PRAP1	Proline-rich acidic protein 1	–	2
37	SLC17A4	Probable small intestine urate exporter	Secondary carrier transporter	3
38	SLC2A2	Solute carrier family 2, facilitated glucose transporter member 2; GLUT2	Secondary carrier transporter	3
39	SLC39A5	Zinc transporter ZIP5	Secondary carrier transporter	2
40	TM4SF5	Transmembrane 4 L6 family member 5	–	2

Note: These genes were identified to be co-expressed with NAT2 by more than one of 4 analytical engines (ie, Archs⁴, COXPRESdb, SEEK, and GEPIA). The original lists of co-expressed genes are shown in Table 2.

^aNumber of times the gene appeared in 4 separate analyses of co-expressed genes.

Table 4.

Gene Ontology Biological Processes Enriched Among Genes Co-Expressed With NAT2 (Selected)

Category	Rank	GO Term ID	GO Term	p Value	P_FDR_adja	No. Genes	Genes
Xenobiotic metabolism	5	GO : 0017144	Drug metabolic process	1.00E-10	2.47E-07	12	ADH4|ADH6|AGXT|AGXT2|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|HAO1|HMGCS2|NR1I2
	31	GO : 0042737	Drug catabolic process	3.74E-07	1.33E-04	6	CYP2C18|CYP2C19|CYP2C9|CYP3A4|HAO1|NR1I2
	6	GO : 0042738	Exogenous drug catabolic process	3.10E-09	5.77E-06	5	CYP2C18|CYP2C19|CYP2C9|CYP3A4|NR1I2
		GO : 0097267	Omega-hydroxylase P450 pathway	8.43E-07	2.51E-04	3	CYP2C19|CYP2C9|CYP4F2
	17	GO : 0019373	Epoxygenase P450 pathway	7.54E-08	4.89E-05	4	CYP2C18|CYP2C19|CYP2C9|CYP4F2
Lipid; fatty acid; cholesterol	21	GO : 0033344	Cholesterol efflux	1.11E-07	5.84E-05	4	ABCG8|APOA5|APOB|APOC3
	50	GO : 0030301	Cholesterol transport	3.80E-06	8.38E-04	4	ABCG8|APOA5|APOB|APOC3
	42	GO : 0006869	Lipid transport	1.64E-06	4.29E-04	7	ABCG8|APOA5|APOB|APOC3|FABP1|NR1H4|PLA2G12B
	44	GO : 0010876	Lipid localization	2.63E-06	6.58E-04	7	ABCG8|APOA5|APOB|APOC3|FABP1|NR1H4|PLA2G12B
	10	GO : 0042632	Cholesterol homeostasis	9.80E-09	1.06E-05	6	ABCG8|APOA5|APOB|APOC3|NR1H4|PLA2G12B
	26	GO : 0055088	Lipid homeostasis	2.15E-07	8.95E-05	6	ABCG8|APOA5|APOB|APOC3|NR1H4|PLA2G12B
	3	GO : 0006629	Lipid metabolic process	0.00E + 00	1.13E-08	18	ADH4|ADH6|APOA5|APOB|APOC3|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|FABP1|GBA3|HAO1|HMGCS2|MOGAT2|NR1H4|NR1I2|PLA2G12B
	4	GO : 0044255	Cellular lipid metabolic process	0.00E + 00	2.73E-08	16	ADH4|ADH6|APOA5|APOB|APOC3|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|FABP1|GBA3|HAO1|HMGCS2|MOGAT2|PLA2G12B
	43	GO : 0034370	Triglyceride-rich lipoprotein particle remodeling	2.54E-06	6.52E-04	3	APOA5|APOB|APOC3
	8	GO : 0016042	Lipid catabolic process	4.60E-09	6.39E-06	9	APOA5|APOB|APOC3|CYP3A4|CYP4F2|FABP1|GBA3|HAO1|PLA2G12B
	18	GO : 0044242	Cellular lipid catabolic process	8.62E-08	5.03E-05	7	APOA5|APOB|APOC3|CYP4F2|FABP1|GBA3|HAO1
	27	GO : 0019433	Triglyceride catabolic process	2.19E-07	8.95E-05	4	APOA5|APOB|APOC3|FABP1
	35	GO : 0046464	Acylglycerol catabolic process	6.47E-07	1.98E-04	4	APOA5|APOB|APOC3|FABP1
	47	GO : 0046503	Glycerolipid catabolic process	3.53E-06	8.27E-04	4	APOA5|APOB|APOC3|FABP1
	29	GO : 0006641	Triglyceride metabolic process	3.34E-07	1.24E-04	5	APOA5|APOB|APOC3|FABP1|MOGAT2
	39	GO : 0006639	Acylglycerol metabolic process	1.36E-06	3.84E-04	5	APOA5|APOB|APOC3|FABP1|MOGAT2
	34	GO : 0070328	Triglyceride homeostasis	6.47E-07	1.98E-04	4	APOA5|APOC3|NR1H4|PLA2G12B
	36	GO : 0055090	Acylglycerol homeostasis	6.47E-07	1.98E-04	4	APOA5|APOC3|NR1H4|PLA2G12B
	46	GO : 0001676	Long-chain fatty acid metabolic process	3.13E-06	7.49E-04	5	CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2
	33	GO : 0046461	Neutral lipid catabolic process	6.47E-07	1.98E-04	4	APOA5|APOB|APOC3|FABP1
	41	GO : 0006638	Neutral lipid metabolic process	1.43E-06	3.84E-04	5	APOA5|APOB|APOC3|FABP1|MOGAT2
Sterols; Steroids	11	GO : 0055092	Sterol homeostasis	1.06E-08	1.06E-05	6	ABCG8|APOA5|APOB|APOC3|NR1H4|PLA2G12B
	14	GO : 0008202	Steroid metabolic process	3.17E-08	2.50E-05	8	APOA5|APOB|CYP2C19|CYP2C9|CYP3A4|HMGCS2|NR1H4|NR1I2
	1	GO : 0006721	Terpenoid metabolic process	0.00E + 00	1.13E-08	9	ADH4|ADH6|APOB|APOC3|CYP2C18|CYP2C19|CYP2C9|CYP3A4|HMGCS2
	2	GO : 0006720	Isoprenoid metabolic process	0.00E + 00	1.84E-08	9	ADH4|ADH6|APOB|APOC3|CYP2C18|CYP2C19|CYP2C9|CYP3A4|HMGCS2
	15	GO : 0001523	Retinoid metabolic process	4.89E-08	3.59E-05	6	ADH4|ADH6|APOB|APOC3|CYP2C18|CYP3A4
	16	GO : 0016101	Diterpenoid metabolic process	6.87E-08	4.73E-05	6	ADH4|ADH6|APOB|APOC3|CYP2C18|CYP3A4
	48	GO : 0034754	Cellular hormone metabolic process	3.69E-06	8.38E-04	5	ADH4|ADH6|CYP2C18|CYP2C9|CYP3A4
	24	GO : 0016098	Monoterpenoid metabolic process	1.41E-07	6.49E-05	3	CYP2C19|CYP2C9|CYP3A4
Other	22	GO : 0065008	Regulation of biological quality	1.32E-07	6.49E-05	23	A1CF|ABCG8|ADH4|ADH6|AFM|AGXT|AGXT2|APOA5|APOB|APOC3|CYP2C18|CYP2C9|CYP3A4|CYP4F2|F11|F9|GBA3|HPX|NR1H4|OTC|PLA2G12B|SLC2A2|SLC39A5
	25	GO : 0015711	Organic anion transport	1.88E-07	8.30E-05	9	ABCG8|AGXT|APOA5|APOC3|FABP1|NR1H4|PLA2G12B|SLC17A4|SLC2A2
	40	GO : 0006820	Anion transport	1.42E-06	3.84E-04	9	ABCG8|AGXT|APOA5|APOC3|FABP1|NR1H4|PLA2G12B|SLC17A4|SLC2A2
	32	GO : 0050892	Intestinal absorption	4.46E-07	1.54E-04	4	ABCG8|FABP1|MOGAT2|SLC2A2
	9	GO : 0044281	Small-molecule metabolic process	7.90E-09	9.68E-06	17	ADH4|ADH6|AGXT|AGXT2|ALDOB|APOA5|APOB|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|HAO1|HMGCS2|MOGAT2|NR1H4|OTC
	12	GO : 1901615	Organic hydroxy compound metabolic process	1.16E-08	1.07E-05	10	ADH4|ADH6|APOA5|APOB|CYP3A4|CYP4F2|HAO1|HMGCS2|MOGAT2|NR1H4
	23	GO : 0006066	Alcohol metabolic process	1.38E-07	6.49E-05	8	ADH4|ADH6|APOA5|APOB|CYP3A4|HAO1|HMGCS2|MOGAT2
	38	GO : 0046185	Aldehyde catabolic process	1.16E-06	3.36E-04	3	ADH4|AGXT|AGXT2
	49	GO : 0006081	Cellular aldehyde metabolic process	3.80E-06	8.38E-04	4	ADH4|AGXT|AGXT2|ALDOB
	13	GO : 0009056	Catabolic process	1.28E-08	1.09E-05	18	ADH4|AGXT|AGXT2|ALDOB|APOA5|APOB|APOC3|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|FABP1|GBA3|HAO1|NR1I2|OTC|PLA2G12B
	45	GO : 1901575	Organic substance catabolic process	3.11E-06	7.49E-04	14	ADH4|AGXT|AGXT2|ALDOB|APOA5|APOB|APOC3|CYP3A4|CYP4F2|FABP1|GBA3|HAO1|OTC|PLA2G12B
	19	GO : 0044248	Cellular catabolic process	8.67E-08	5.03E-05	16	ADH4|AGXT|AGXT2|APOA5|APOB|APOC3|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|FABP1|GBA3|HAO1|NR1I2|OTC
	7	GO : 0032787	Monocarboxylic acid metabolic process	3.70E-09	5.83E-06	11	ADH6|AGXT|AGXT2|ALDOB|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|HAO1|NR1H4
	20	GO : 0019752	Carboxylic acid metabolic process	9.66E-08	5.32E-05	12	ADH6|AGXT|AGXT2|ALDOB|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|HAO1|NR1H4|OTC
	28	GO : 0043436	Oxoacid metabolic process	2.70E-07	1.06E-04	12	ADH6|AGXT|AGXT2|ALDOB|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|HAO1|NR1H4|OTC
	30	GO : 0006082	Organic acid metabolic process	3.38E-07	1.24E-04	12	ADH6|AGXT|AGXT2|ALDOB|CYP2C18|CYP2C19|CYP2C9|CYP3A4|CYP4F2|HAO1|NR1H4|OTC

Note: The list of genes that are co-expressed with NAT2 (listed in Table 2) were analyzed using GOnet (Pomaznoy et al. 2018) to identify Gene Ontology biological processes (GO terms and IDs) that are enriched by the group. Some GO terms were categorized according to common biological processes they belong to (eg, Xenobiotic Metabolism). Among the co-expressed genes in Table 2, the genes that contribute to each GO term are shown under the column, “Genes.” Only top 50 GO biological processes are shown here.

^aFalse discovery rate-adjusted p value (or q value).

Figure 5.

Representative genes which are co-expressed with NAT2 among nondiseased human tissues. The dot plots show the mRNA levels (expressed in log₂ of transcripts per kilobase million [TPM]) of NAT2 (x-axis) and the corresponding mRNA level of the gene of interest (eg, A1CF) (y-axis) in the same individual tissue samples. R, Pearson correlation coefficient. The analysis was done using tools available at GEPIA (Tang et al., 2017).

RESULTS

 

Culture Condition-Dependent Changes in NAT2 mRNA in Liver Cancer Cell Lines

While culturing HepG2 cells, we observed that NAT2 mRNA levels fluctuated substantially depending on the culture conditions. Because no previous reports have described such phenomenon, we investigated this further. Starting with DMEM media devoid of glucose, pyruvate, l-glutamine, and serum (“Control”), we cultured HepG2 cells in media which contained cell culture components/nutrients either individually or in various combinations for 24 h and measured NAT2 mRNA via reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR). Compared with Complete Media (containing 5.5 mM d-glucose, 1 mM pyruvate, 4 mM l-glutamine, and 10% FBS), there was a significant and marked reduction in NAT2 mRNA level in cells cultured in Control DMEM (Figure 1A). Addition of d-glucose (5 mM) or serum (10% FBS) alone was able to recover the NAT2 mRNA level (13.5-fold increase by glucose and 18.2-fold increase by FBS). Although pyruvate (1 mM) or l-glutamine (4 mM) alone was also able to moderately increase the NAT2 mRNA level (2.1- and 6.4-fold, respectively), the induction was not found to be statistically significant (Figure 1A). Combining glucose with pyruvate or combining glucose with pyruvate and l-glutamine did not significantly increase the NAT2 mRNA further, compared with the glucose treatment alone (Figure 1A). However, the NAT2 mRNA level in Complete Media (23.9-fold increase) was significantly higher than that in glucose treatment alone (13.5-fold increase), suggesting a moderately additive effect of serum and other components on NAT2 expression (Figure 1A). Taken together, these results suggest that NAT2 expression is influenced by the presence of glucose and serum in HepG2 cells.

Figure 1.

Effect of different culture components on the mRNA levels of NAT2 and other genes assayed for comparison. HepG2 cells were cultured in the presence of the indicated components/nutrients (either individually or in combinations) for 24 h. DMEM free of d-glucose, pyruvate, l-glutamine, and FBS (Control) served as a treatment control. Gluc, d-glucose (5.5 mM). Pyr, pyruvate (1 mM). Glut, l-glutamine (4 mM). “Complete Media” contained all components listed here, including FBS (10%). A–F, The relative mRNA level of the indicated gene was measured by RT-qPCR using 18S ribosomal RNA as an internal control. Bar graphs represent mean ± SD (standard deviation) (n = 4 per group). *p < .05; **p < .01; ***p < .001. The asterisks located above the bars are for the comparison with the control DMEM treatment group (ie, Control), unless otherwise specified. The asterisks above the bracketed lines are for the comparison between 2 treatment groups.

We also examined the mRNA levels of its closely related isozyme, NAT1, and additional genes in the same set of samples (Figure 1). The NAT1 transcript did not share the same culture condition-dependent gene expression pattern with NAT2 and remained relatively steady. For instance, the presence or absence of glucose or serum did not affect the NAT1 transcript level (Figure 1B). Interestingly, there was a modest, yet significant, increase (1.7-fold) in NAT1 mRNA when cells were cultured in l-glutamine alone. However, such effect on NAT1 expression was gone when l-glutamine was combined with other components, such as glucose and pyruvate (Figure 1B). For comparison, we also analyzed the same set of samples for G6PC (glucose-6-phosphatase catalytic subunit 1) mRNA, which is involved in gluconeogenesis and is one of the insulin receptor target genes (Barthel and Schmoll, 2003). Its expression pattern largely contrasted with that of NAT2. Although there was a significant, yet marginal, increase (1.4-fold) in G6PC mRNA in pyruvate-treated cells, glucose failed to induce its level (Figure 1C). Compared with Control DMEM, G6PC transcript was significantly and markedly downregulated when the cells were cultured in complete media (0.11-fold) as well as in media containing serum (0.22-fold) or the combination of components (0.36-fold) (Figure 1C). We additionally examined the expression of GLUT2 (SLC2A2), the major glucose transporter in hepatocytes. Its expression was significantly induced (approximately 2-fold) by glucose, pyruvate, and FBS (Figure 1D). Such induction of GLUT2 was not observed when the cells were treated with l-glutamine alone, and, interestingly, addition of l-glutamine prevented the gene induction by other components (ie, glucose, pyruvate, or FBS). Transcript levels of commonly used housekeeping genes, GAPDH and ACTB (β-actin), remained largely unchanged between culture conditions tested (when normalized to 18S rRNA), although ACTB (β-actin) expression was more variable than GAPDH, questioning its validity as an internal control (Figs. 1E and 1F). Taken together, these results indicated that the culture condition-dependent expression of NAT2 is not a global but a gene-specific phenomenon.

We also tested an additional liver cancer cell line, Hep3B, to see if glucose or pyruvate produced similar effects on NAT2 expression in multiple cell lines. In Hep3B cells, glucose (5 mM) alone significantly increased the relative NAT2 mRNA level (9.7-fold) (Figure 2A). Interestingly, treatment with pyruvate (1 mM) alone also significantly induced NAT2 mRNA (6.7-fold) in Hep3B cells (Figure 2A). The effects of pyruvate on NAT2 mRNA was not as robust with HepG2 cells (2.3-fold increase) in the same experiment (Figure 2A). These results indicate that induction of NAT2 mRNA by glucose is not a cell line-specific effect.

Figure 2.

Induction of NAT2 mRNA and NAT2 N-acetyltransferase activity by glucose in liver cancer cell lines. HepG2 or Hep3B cells were cultured in the presence of the indicated components/nutrients for 24 h. DMEM free of d-glucose, pyruvate, l-glutamine, and FBS served as a treatment control (Control). A, HepG2 (left panel) and Hep3B (right panel) were incubated with 5 mM d-glucose (+Glucose) or 1 mM sodium pyruvate (+Pyruvate) in Control DMEM. The relative NAT2 mRNA level was measured by RT-qPCR using 18S ribosomal RNA as an internal control. B, Concentration-dependent increases in NAT2 mRNA in HepG2 cells by glucose. HepG2 cells was incubated in the indicated concentration of d-glucose in Control DMEM for 24 h. The relative NAT2 mRNA level was measured by RT-qPCR. C, HepG2 cells were incubated in the indicated concentrations of d-glucose (in Control DMEM) for 24 h and analyzed for NAT2 protein level using Western blot (left panel). NAT2 band intensity was quantified and normalized to that of GAPDH and expressed as a relative level (panel D). D, relative level of NAT2 protein and enzymatic activity. HepG2 cells were incubated in the indicated concentrations of d-glucose (in Control DMEM) for 24 h, lysed, and measured for NAT2 protein (white bars) by Western blot or NAT2 activity (gray bars) using sulfamethazine (SMZ) as its substrate. The level of N-acetylated-SMZ (Ac-SMZ) was measured using HPLC and normalized to the time of incubation (min) and total protein amount (mg protein). The protein and activity measurements were expressed as relative values to the untreated control. According to 2-way ANOVA analysis, glucose concentration significantly contributed to variation (p = .0005). Bar graphs represent mean ± SD (n = 4 per group). *, p < .05; **, p < .01; ***, p < .001. ns, not statistically significant. The asterisks located above the bars are for the comparison with the control DMEM treatment group (ie, Control), unless otherwise specified. The asterisks above the bracketed lines are for the comparison between 2 treatment groups.

Concentration-Dependent Induction of NAT2 mRNA by Glucose

To investigate if induction of NAT2 mRNA by glucose was concentration-dependent, we exposed HepG2 cells to different concentrations of glucose (0, 0.2, 1, 5, and 25 mM) in Control DMEM for 24 h. Between 0 and 5 mM glucose, the relative level of NAT2 mRNA increased in a concentration-dependent manner (Figure 2B). However, the mRNA level peaked at 5 mM glucose, and did not increase further at 25 mM treatment (Figure 2B).

Changes in NAT2 Protein Level and N-Acetyltransferase Activity by Glucose

HepG2 cells were incubated in Control DMEM containing 0, 2.5, or 25 mM of d-glucose for 24 h, and the treated cells were analyzed by Western blot using an antibody against human NAT2 (Figure 2C). Following the glucose treatment, NAT2 activity was measured in vitro in the presence of a NAT2-selective substrate, SMZ and acetyl-CoA. There was a significant increase (1.4-fold) in in vitro NAT2 activity following treatment with 25 mM glucose, compared with the no-glucose treatment group (Figure 2D), and there was significant variation in both protein and activity with respect to concentration (p = .005). Glucose treatment marginally increased NAT2 protein to an extent that did not differ significantly from its increase in NAT2 catalytic activity.

Induction of NAT2 by Insulin

Previous studies on mouse NAT1 showed that it plays an important role in maintaining insulin sensitivity in mice (Camporez et al., 2017; Chennamsetty et al., 2016; Knowles et al., 2015). We questioned if NAT2 mRNA expression was controlled by insulin or a fasting condition in HepG2 cells. Insulin (100 nM) treatment for 24 h significantly induced NAT2 mRNA in HepG2 cells (Figure 3A). The level of induction (1.4-fold), however, was not as robust as that observed with glucose treatments (13.5-fold; see Figure 1A). We also subjected the cells to treatment with 8-bromo-cAMP and dexamethasone (cAMP/Dex), which mimics the effect of glucagon and “starvation” condition in vitro. cAMP/Dex did not alter or further decrease the NAT2 mRNA level (Figure 3A). G6PC served as the treatment control. As expected, insulin treatment significantly reduced G6PC mRNA (0.36-fold), whereas cAMP/Dex treatment significantly induced its expression (13.3-fold) (Figure 3B). Next, we examined if insulin can induce NAT2 expression even in the presence of normal, serum-free, culture media containing glucose, pyruvate, and l-glutamine. Insulin treatment (100 nM for 6 h) induced NAT2 mRNA expression (approximately 2.5-fold) even in the presence of glucose (Figure 3C). This finding indicates that insulin induces NAT2 expression in HepG2 cells and that NAT2 may be a novel insulin receptor target gene.

Figure 3.

Effect of insulin, cAMP/Dex, palmitate, or acetate on mRNA levels of NAT2, G6PC, and NAT1. A and B, HepG2 cells were cultured in the presence of insulin (100 nM) (+ Insulin) or 8-br-cAMP (500 µM) + dexamethasone (1 µM) (+ cAMP/Dex). DMEM free of d-glucose, pyruvate, l-glutamine, and FBS served as a treatment control (Control). C, HepG2 cells were cultured in serum-free DMEM containing d-glucose, pyruvate, and l-glutamine, plus insulin (100 nM) (+ Insulin) for 6 h. D, HepG2 cells were cultured in Control DMEM containing 0.5% BSA or 0.5% BSA-250 µM palmitate (BSA-PA) for 24 h. E, HepG2 cells were cultured in Control DMEM containing 5 mM d-glucose or 5 mM acetate for 24 h. Following the treatments, the relative mRNA level of the indicated gene (NAT2, G6PC, or NAT1) was measured by RT-qPCR using 18S ribosomal RNA as an internal control. Bar graphs represent mean ± SD (n = 4 per group). *, p < .05; **, p < .01; ***, p < .001. ns, not statistically significant. The asterisks located above the bars are for the comparison with the control DMEM treatment group (ie, Control), unless otherwise specified. The asterisks above the bracketed lines are for the comparison between 2 treatment groups.

We next questioned if NAT2 expression is influenced by free fatty acids, an alternative energy source. HepG2 cells were treated with BSA (control) or BSA-conjugated palmitic acid (BSA-PA; 250 µM) in Control DMEM for 24 h. In contrast to the glucose treatment, BSA-PA treatment did not modulate the NAT2 mRNA level (Figure 3D).

Induction of NAT2 Expression by Acetate

Our group has demonstrated previously that NAT1 hydrolyzes acetyl-CoA in the presence of folate (Stepp et al., 2015), and CRISPR/Cas9-mediated deletion of NAT1 in MDA-MB-231 breast cancer cells leads to a significant increase in cellular acetyl-CoA levels (Stepp et al., 2019). These findings suggest that NAT1 regulates cellular acetyl-CoA level. Based on the link between NAT1 and acetyl-CoA, we questioned if NAT2 expression can be regulated reciprocally by the cellular acetyl-CoA level in HepG2 cells. Acetyl-CoA can be generated from acetate in various types of cancers (Mashimo et al., 2014), which is catalyzed by acetyl-CoA synthetase 1 and 2 (ACSS1 and ACSS2) in mammals (Fujino et al., 2001; Luong et al., 2000). To test this, we treated HepG2 cells with 5 mM glucose or 5 mM acetate in Control DMEM for 24 h and measured the changes in the NAT2 mRNA level. The acetate concentration (5 mM) used in the study was based on previous studies (Gao et al., 2016). Acetate treatment led to a significant increase in NAT2 mRNA (4.0-fold) (Figure 3E), although the level of induction was not as high as the one in the glucose (5 mM) treatment group (10.7-fold) (Figure 3E). In contrast, NAT1 transcript was not induced by acetate (Figure 3E).

Genes Involved in Glucose and Lipid Metabolism and Transport

To explore the mechanism of glucose-mediated transcriptional regulation of NAT2, we sought to compare the pattern of NAT2 mRNA expression to the expression of genes involved in glucose and lipid metabolism and transport. Gene polymorphisms in human NAT2 have been previously linked to insulin resistance and differential plasma lipid levels (Hoffmann et al., 2018; Knowles et al., 2015; Willer et al., 2013), which raises the possibility that NAT2 expression and/or activity may be regulated by energy or metabolic state of the cell. To explore this possibility, we screened a set of genes implicated in cellular metabolism to see if their gene expression patterns resembled that of NAT2 in HepG2 cells. For this, we characterized glucose- and acetate-dependent changes in the transcript levels of multiple genes.

PPARGC1A and CPT1A

PGC1α (a.k.a., PPARGC1A) regulates the expression of key mitochondrial genes and is an important regulator of mitochondrial biogenesis, and CPT1A (carnitine palmitoyl transferase 1A; CPT1) catalyzes an essential step for the mitochondrial uptake of long-chain fatty acids and their subsequent β-oxidation. Glucose-dependent changes in the expression of PGC1α and CPT1A contrasted sharply with that of NAT2. PGC1α and CPT1A were significantly downregulated in the presence of glucose (Figure 4). This likely reflects the switch to glycolysis (vs oxidative phosphorylation) and a reduction in utilization of β-oxidation of fatty acids upon introduction of glucose in the culture media, and suggests that the cellular roles of NAT2 are different from the ones played by PGC1α and CPT1A.

Figure 4.

Glucose and acetate-dependent expression patterns of genes involved in glucose and lipid metabolism. HepG2 cells were cultured in the indicated concentration of d-glucose or 5 mM acetate (Acetate) for 24 h. The relative mRNA level of the indicated gene (eg, NAT2) was measured by RT-qPCR using 18S ribosomal RNA as an internal control. The horizontal lines represent mean ± SD (n = 4 per group). *, p < .05; **, p < .01; ***, p < .001. ns, not statistically significant. The asterisks or any marks located above the lines are for the comparison with the control DMEM treatment group (ie, Control).

FXR and PXR

Although NAT1 is ubiquitously expressed in various organs and tissues, expression of NAT2 is limited to specific organs, including liver and intestines (Sim et al., 2014), suggesting NAT2 expression is likely regulated by tissue-specific transcription factors. We examined the glucose-dependent mRNA expression patterns of 2 transcription factors that are known to mediate tissue-specific (ie, liver and intestines) gene expression, FXR (farnesoid X receptor; NR1H4), and PXR (pregnane X receptor; NR1I2). FXR, which is a regulator of bile acid synthesis and also regulates lipid and glucose homeostasis (Preidis et al., 2017), exhibited an interesting biphasic pattern where its expression was upregulated at low concentrations of glucose (0.2 and 1 mM) but downregulated at higher concentrations (5 and 25 mM) (Figure 4). PXR is known to regulate genes involved in xenobiotic metabolism and secretion, but also has been implicated in energy metabolism (Wada et al., 2009). There was a concentration-dependent increase in PXR mRNA by glucose, the pattern of which was similar to that of NAT2 (Figure 4).

Genes in fatty acid metabolism and transport

Gene polymorphisms in human NAT2 have been previously linked to differential plasma lipid levels (Hoffmann et al., 2018; Willer et al., 2013), which led to the question if NAT2 expression and/or activity is co-regulated along with genes involved in fatty acid metabolism and transport. We observed glucose- and acetate-dependent changes in the transcript levels of multiple genes involved in this category. These included CD36 (fatty acid transporter; FAT), SCD (acyl-CoA desaturase), ACACA (acetyl-CoA carboxylase alpha; ACC), FASN (fatty acid synthase; FAS), FABP1 (fatty acid binding protein 1), and ACOX1 (acyl-CoA oxidase 1). The transcript levels of SCD, ACACA, and FASN, which are involved in fatty acid synthesis, did not change dynamically in response to glucose or acetate treatment in HepG2 cells (Figure 4). Among the genes tested, FABP1 and CD36 (FAT), which are involved in fatty acid transport, showed glucose concentration-dependent induction, and their expression patterns largely coincided with that of NAT2 (Figure 4). ACOX1 which mediates the first step in β-oxidation of fatty acids also showed a similar pattern of expression in response to glucose and acetate (Figure 4).

Genes in acetyl-CoA synthesis and gluconeogenesis

Our analysis included genes involved in acetyl-CoA synthesis. ACLY (ATP Citrate Lyase) catalyzes the cleavage of citrate into oxaloacetate and acetyl-CoA, and ACSS1 and ACSS2 are known to generate acetyl-CoA from free acetate (Fujino et al., 2001; Luong et al., 2000). The glucose and acetate-dependent pattern of the genes encoding them was similar to that of NAT2 (Figure 4). G6PC and PCK1 represented gluconeogenic genes. G6PC induction by glucose was biphasic in which there was a significant induction at 0.2, 5.0, or 25 mM glucose but not at 1 mM, and acetate failed to induce G6PC (Figure 4). In contrast, PCK1 showed concentration-dependent increases in mRNA by glucose and also was induced by acetate alone. Interestingly, the level of PCK1 induction by acetate was comparable with that by 25 mM glucose. Hence, the expression patterns of G6PC and PCK1 were not only dissimilar to each other’s but also to that of NAT2.

Genes co-expressed with NAT2: In silico analysis

In order to better understand the transcriptional regulation of NAT2 and its energy metabolism-related roles (eg, in insulin sensitivity), we analyzed genes whose expression patterns in various human tissues resembled that of NAT2 (ie, “co-expressed” genes), using publicly available gene expression databases and analytical tools available at GEPIA (Tang et al., 2017), COXPRESdb (Obayashi et al., 2019), SEEK (Zhu et al., 2015), and Archs⁴ (Lachmann et al., 2018). The list of co-expressed genes differed depending on the algorithm (Table 2). We compared the lists of the top 50 co-expressed genes from all 4 engines and selected the ones that appeared more than once. This retrieved a total of 40 genes (Table 3). The only gene that appeared on all 4 lists was NR1I2 which encodes pregnane X receptor (PXR). There were 13 genes that appeared on 3 out of 4 lists, and they included FABP1, MOGAT2 (monoacylglycerol O-acyltransferase 2), and GLUT2 (SLC2A2) (Table 3).

The list of co-expressed genes was then analyzed for biological pathways (Gene Ontology [GO] terms) that are significantly enriched, using GONet (Pomaznoy et al., 2018). Genes encoding xenobiotic metabolizing enzymes, such as ADH4, ADH6, CYP2C18, CYP2C19, CYP2C9, CYP3A4, CYP4F2, HAO1, and NR1I2 (PXR) contributed to the enrichment of biological processes related to xenobiotic or drug metabolism (Tables 3 and 4). The GO terms in this category included “drug metabolic process” (GO: 0017144), “epoxygenase P450 pathway” (GO: 0019373), and “xenobiotic metabolic process” (GO: 0006805), and “cellular response to xenobiotic stimulus” (GO: 0071466) (Table 4). Similarly, among the co-expressed genes which were not included in the final list, there were also multiple genes encoding phase II metabolic enzymes including sulfotransferases (SULT1B1 and SULT2A1) and UDP-glucuronosyltransferases (UGT1A4, UGT2A3, UGT2B10, and UGT2B15) (Table 2), suggesting that NAT2 expression is co-regulated along with other genes involved in xenobiotic, phase II metabolism.

Interestingly, GO terms that are related to cholesterol, triglyceride, lipid, and lipoprotein synthesis and transport were overwhelmingly enriched among the selected 40 genes (Table 4). These included “cholesterol homeostasis” (GO: 0042632), “lipid homeostasis” (GO: 0055088), “triglyceride homeostasis” (GO: 0070328), “regulation of cholesterol transport” (GO: 0032374), “plasma lipoprotein particle assembly” (GO: 0034377), and “plasma lipoprotein particle remodeling” (GO: 0034369) (Table 4). Among the co-expressed genes, APOA5, APOB, APOC2, APOC3, ABCG8, ANGPTL3, FABP1, MOGAT2, and PLA2G12B contributed to the enrichment of these biological processes (Tables 3 and 4).

Next, we examined the correlation between the transcript levels of NAT2 and those of individual genes from the list using the GTEx (Genotype-Tissue Expression) dataset (non-diseased human tissues) and graphing and statistical tools available at GEPIA (Tang et al., 2017). Among the genes whose expression highly correlated with that of NAT2 (Pearson correlation coefficient, R > 0.8), A1CF (APOBEC1 complementation factor) showed the highest correlation with NAT2 with a R value of 0.9 (Figure 5). Others included ABCG8 (ATP-binding cassette sub-family G member 8) (R = 0.86), GLUT2 (glucose transporter 2; SLC2A2) (R = 0.84), CYP2C9 (R = 0.83), SLC27A2 (a.k.a., FATP2, fatty acid transport protein 2) (R = 0.72) and FABP1 (R = 0.67), (Figure 5). ABCG8 is involved in transport of cholesterol and sterols (Kerr et al., 2021). CYP2C9 is a member of the cytochrome P450 superfamily of enzymes and the most abundantly expressed human CYP2C isoform in the liver (Miners and Birkett, 1998). SLC27A2 (FATP2) activates long-chain fatty acids as a very long-chain acyl-CoA synthetase and also transports long-chain fatty acids as a fatty acid transporter (Qiu et al., 2020). Expression of NR1I2 (PXR), which was the only gene that was identified to be co-expressed with NAT2 by all 4 engines (Table 2), also correlated with that of NAT2 (R = 0.77). In contrast, NAT1, whose expression is normally found in multiple tissues and organs, did not correlate well with NAT2 (R = 0.39) whose expression is largely restricted to liver and intestines (Figure 5).

DISCUSSION

The current study shows, for the first time, that human NAT2 expression at the transcript level and its enzymatic activity is regulated by glucose in hepatocellular carcinoma cell lines. It has not been determined, however, if NAT2 is transcriptionally regulated similarly in normal human hepatocytes or liver. It is possible that liver cancer cells (eg, HepG2) are more sensitive to changes in the glucose level due to their dependency on glycolysis for energy production (ie, Warburg effect). We also speculate that it is not glucose per se, but rather its metabolism or metabolites that are responsible for the changes in the NAT2 expression. This is supported by the current finding that reduced NAT2 expression in glucose-depleted media is recovered, in part, by pyruvate, a metabolite in the glycolytic pathway, (see Figure 2A) or acetate supplementation (see Figure 3E). Notably, pyruvate can be decarboxylated and converted to acetyl-CoA by the mitochondrial pyruvate dehydrogenase complex, and free acetate can be converted to acetyl-CoA by acyl-CoA short-chain synthetases, ACSS1 and ACSS2 (Moffett et al., 2020). It would require additional studies to determine if glucose induces NAT2 expression in normal human hepatocytes and if such transcriptional regulation is mediated by its downstream metabolites, such as acetyl-CoA.

Glucose plays a central role in metabolism, serving as the preferred source of energy and carbon. Glucose (and its metabolites) induces changes in metabolism by affecting gene expression in multiple tissues and cell types, including pancreatic β islet cells, adipose tissue, and liver. One mechanism by which glucose influences the expression of metabolic genes is through carbohydrate response element binding protein (ChREBP; MLXIPL), an important regulator of an intracellular carbohydrate-sensing mechanism (Ortega-Prieto and Postic, 2019). ChREBP expression is present in multiple organs in the gut and adipose tissues, but most abundant in active sites of de novo lipogenesis including liver and adipose tissues (Iizuka et al., 2004). In response to glucose and its metabolites (eg, glucose-6-phosphate), ChREBP undergoes post-translational modifications that modulate its subcellular localization, protein stability and/or its transcriptional activity (Ortega-Prieto and Postic, 2019). Once activated, it binds to a cis-acting regulatory element called “carbohydrate response element” and transactivates the target genes. Its target genes encode key enzymes of de novo lipogenesis, such as fatty acid synthase (FAS; FASN), acetyl-CoA carboxylase (ACC; ACACA) and stearoyl-CoA desaturase (SCD) (Kawaguchi et al., 2001). In the liver, it is also a key regulator of glycolysis and VLDL secretion by modulating expression of l-pyruvate kinase (PKLR), a rate-limiting enzyme in glycolysis, and microsomal triglyceride transfer protein (MTTP), respectively (Kawaguchi et al., 2001). In the intestine, ChREBP stimulates expression of sucrase-isomaltase (SI), GLUT5 (SLC2A5), GLUT2 (SLC2A2) and ketohexokinase (KHK), which improves sucrose tolerance and fructose absorption (Kato et al., 2018). Thus, one of the possible mechanisms by which glucose regulates NAT2 expression in HepG2 cells is via ChREBP. However, it should be noted that in the current study, glucose did not affect or only had marginal effects on mRNA expression of known ChREBP target genes, such as FASN, ACACA, and SCD (see Figure 4), suggesting the experimental conditions used in the study might have been insufficient to activate ChREBP in HepG2 cells.

Another mechanism by which glucose influences gene expression and cellular metabolism is through insulin. Elevated blood glucose induces secretion of insulin which suppresses gluconeogenesis and promotes lipogenesis in the liver. Genes that are involved in lipogenesis are induced by insulin in hepatocytes, and this is mediated through the transcription factor, sterol regulatory element-binding protein 1 (SREBF1) (Horton et al., 2002). Activation of SREBF1 promotes fatty acid and cholesterol biosynthesis, for its target genes include the rate-limiting lipogenic and cholesterol biosynthetic genes, such as fatty acid synthase (FASN; FAS), acetyl-CoA carboxylase (ACC1; ACACA), HMG-CoA reductase, and the LDL receptor (Osborne and Espenshade, 2009). Another major effect of insulin on liver is inhibition of hepatic gluconeogenesis (ie, de novo synthesis of glucose), and it does so by suppressing expression of genes involved in this process, such as G6PC (Streeper et al., 1997) and PCK1 (phosphoenolpyruvate carboxykinase 1; PEPCK) (Granner et al., 1983). Conversely, these gluconeogenic genes are induced by glucagon and glucocorticoids during fasting and stress, respectively (Lamers et al., 1982; Schmoll et al., 1999). In accordance to this, we observed that (overnight) insulin treatment significantly reduced the G6PC mRNA level in HepG2 cells, whereas treatment with cAMP/Dex (which is commonly used to mimic the effects of glucagon and glucocorticoids) markedly upregulated it (see Figure 3B). In contrast to G6PC, insulin induced a marginal, yet significant, increase in NAT2 mRNA while cAMP/Dex did not alter NAT2 expression (see Figure 3A). Moreover, NAT2 transcript was significantly induced following only a 6-h treatment with insulin in the presence of glucose (5 mM) (see Figure 3C), indicating that NAT2 may be a novel insulin receptor target gene in hepatocytes. A recent study reported that reduction in mouse NAT1 level contributes to cerebral endothelial necroptosis and Aβ accumulation in Alzheimer’s disease (Zou et al., 2020). The authors showed that insulin induces mouse Nat1 expression in a concentration-dependent manner in bEnd.3, a mouse endothelioma cell line (Zou et al., 2020). In addition, cerebro-microvessels isolated from endothelial cell-specific insulin receptor KO mice express reduced levels of both mRNA and protein of mouse Nat1, suggesting that insulin regulates expression of mouse Nat1 in endothelial cells in vivo (Zou et al., 2020). These previous and current findings collectively suggest that expression of human NAT2 (and mouse Nat1) is regulated by insulin in multiple cell types, including hepatocytes and endothelial cells.

Although the biological significance of the current findings would require additional studies, human NAT2 may be involved in cellular and physiological processes brought by elevated glucose level and insulin actions. Based on previous studies performed in murine Nat1 KO mice (Camporez et al., 2017; Chennamsetty et al., 2016; Knowles et al., 2015), such regulation of NAT2 expression by insulin appears to be important in maintaining insulin sensitivity. One possibility is that human NAT2 may be involved in lipid and/or cholesterol biosynthesis and transport. First, its expression is induced by insulin and glucose along with other lipogenic genes (see discussions above). Second, a group of genes that are co-expressed with human NAT2 (eg, APOA5, APOB, APOC2, APOC3, ABCG8, ANGPTL3, FABP1, MOGAT2, and PLA2G12B) (see Tables 3 and 4) is involved in lipid and cholesterol biosynthesis and transport. For example, MOGAT2 (2-acylglycerol O-acyltransferase 2; MGAT2) catalyzes the synthesis of diacylglycerol from 2-monoacylglycerol and fatty acyl-CoA as a part of the triglyceride biosynthetic process. It also plays a role in uptake of dietary fat by the small intestine (Cao et al., 2004). As a result, the enrichment analysis revealed that the GO terms that describe cholesterol, triglyceride, lipid, and lipoprotein synthesis and transport are overwhelmingly enriched among co-expressed genes (see Table 4).

Third, the novel link between human NAT2 and lipid and cholesterol metabolism is also supported by multiple GWAS. In particular, one of NAT2 single-nucleotide polymorphisms (SNPs), rs1495741, has been linked to differential plasma lipid and (LDL and total) cholesterol, as well as apolipoprotein B levels by independent GWAS (Hoffmann et al., 2018; Richardson et al., 2020; Ripatti et al., 2020; Willer et al., 2013). SNP rs1495741 is located approximately 14 kb downstream of human NAT2 coding region and thus represents a non-coding, inter-genic SNP. We have previously reported that rs1495741 (a.k.a., “tag SNP”) genotype is a good predictor of the NAT2 acetylator phenotype (García-Closas et al., 2011). In that study, the N-acetyltransferase activity of NAT2 was measured in human hepatocytes from 154 individuals and compared between genotypes at rs1495741 (ie, GG, GA, vs AA). Hepatocytes from individuals that carry the A allele exhibit significantly lower NAT2 activity compared with those with the G allele, and thus GG, GA, and AA alleles of rs1495741 predicts rapid, intermediate, and slow acetylator phenotypes, respectively (García-Closas et al., 2011). The aforementioned GWAS investigating the genetic variants associated with differential serum lipid levels reported that the G (ie, rapid) allele of rs1495741 is associated with increased serum triglyceride and LDL and total cholesterol levels (Hoffmann et al., 2018; Ripatti et al., 2020). Collectively, these findings suggest that individuals carrying rapid alleles of NAT2 (eg, G allele at rs1495741) are likely to produce higher serum lipid levels and imply that higher NAT2 activity may contribute to the development of dyslipidemia. Although there are differences between human and rat NAT2, nevertheless, rapid acetylator rats congenic for Nat2 exhibit greater dyslipidemia than slow acetylator rats (Hong et al., 2020). The GWAS and rat study findings support a hypothesis that human NAT2 is involved in metabolism and transport of lipid and cholesterol. Although the mechanism is unknown, the inter-genic NAT2 SNP (ie, rs1495741) may represent an important cis-acting element (eg, enhancer) that regulates NAT2 expression and NAT2 activity is a novel determinant of the serum lipid levels.

According to the analysis of the GTEx dataset (see Figure 5), the gene that shows the highest correlation with NAT2 in terms of its transcript level in various human tissues is A1CF (APOBEC1 complementation factor). Of note, genetic polymorphisms in A1CF also have been implicated in differential serum triglyceride and cholesterol levels by GWAS (Liu et al., 2017; Sinnott-Armstrong et al., 2021). A1CF is expressed only in liver, intestine, and kidney, with the highest levels in the liver, and its hepatic expression in humans is decreased in obese subjects compared with that in lean donors (Nikolaou et al., 2019). A1CF was originally identified as an AU-rich RNA-binding protein that regulates RNA editing activity of the cytidine deaminase APOBEC1 and posttranscriptional editing of ApoB mRNA (Mehta et al., 2000). However, studies in A1cf KO mice revealed that A1CF is dispensable for hepatic ApoB mRNA editing and expression (Snyder et al., 2017). For this reason, the cellular functions of A1CF had remained elusive until recently. In a recent study (Nikolaou et al., 2019), investigators reported that A1CF functions as a hepatocyte-specific regulator of alternative splicing. The authors showed that it regulates alternative splicing of specific genes that are involved in fructose, lipid, and glycerophospholipid metabolism, such as ketohexokinase (Khk) and glycerol kinase (Gk) genes (Nikolaou et al., 2019). Interestingly, liver-specific A1cf KO mice showed a slight increase in serum triglyceride under fed state and improvement in glucose homeostasis (Nikolaou et al., 2019), suggesting A1CF is involved in regulation of lipid and glucose homeostasis via controlling liver-specific alternative splicing of its target genes. In another recent paper, Blanc et al. studied transgenic mice that overexpress A1cf under the liver-specific ApoE enhancer (Blanc et al., 2021). The authors observed that hepatic overexpression of A1cf resulted in spontaneous hepatic steatosis in young mice, and this was concomitant with increases in expression of genes involved in fatty acid uptake and lipogenesis, including Cidea, Mogat1, Mogat2, and Cd36 (Blanc et al., 2021). Because the expression patterns of NAT2 and A1CF are highly correlative, it would be of interest to examine the role of NAT2 in the context of regulation of lipid metabolism by A1CF.

The glucose treatment significantly increased NAT2 catalytic activity. It marginally increased NAT2 protein to an extent that did not differ significantly from its increase in NAT2 catalytic activity (see Figs. 2C and 2D), suggesting that the increase in catalytic activity is secondary to the increase in protein. This contrasted with the robust increase in NAT2 mRNA by glucose. Previous studies have reported similar cases in which changes in NAT2 mRNA did not correlate with changes NAT2 protein/activity. A transcriptomics study by Carlisle et al. (2021) reported that one of the transcripts that are highly and significantly upregulated in NAT1 KO MDA-MB-231 cells (compared with the parental cells) is NAT2, and yet, the authors were not able to detect NAT2 enzyme activity in NAT1 KO cells. In a study by Salazar-González et al. (2020), an in-cell Western technique was used to measure NAT2 protein levels in cryopreserved human hepatocytes with distinctive acetylator genotypes (ie, rapid, intermediate, and slow acetylators) and found that NAT2 protein level correlated well with the acetylator genotype. The NAT2 mRNA level, however, did not correlate with either NAT2 protein level or the acetylator genotype, indicating that NAT2 mRNA level is a poor predictor of the NAT2 protein or activity levels in human hepatocytes. An additional factor that needs to be considered is the acetylator genotype of HepG2 cells used in the current study. Our analysis showed that HepG2 has NAT2*6A and NAT2*14C alleles, and thus is a slow acetylator. Given that HepG2 carries 2 slow acetylator alleles, it is not surprising that the NAT2 transcriptional upregulation conferred by glucose did not directly translate into a similar increase in protein and/or enzymatic activity. Future investigations are necessary to understand the actual phenotypic effects of this association. The underlying mechanism for such disconnects between NAT2 mRNA and protein levels observed in the previous and present studies is unknown. It is possible that NAT2 is regulated at the post-transcriptional level, for recent studies have identified microRNAs that target NAT2 (Yang et al., 2019).

In summary, expression of human NAT2 was upregulated by glucose and insulin in liver cancer cell lines, indicating that NAT2 expression can change dynamically depending on the energy and metabolic status of the cells. The pattern of culture condition-dependent NAT2 expression as well as its expression in human tissues suggests a novel role of NAT2 in lipid and cholesterol metabolism and transport. We only examined transcriptional regulation of NAT2 in cancer cells of hepatocyte origin. It is currently unknown if the same mechanism of regulation is present in normal human hepatocytes or other normal human cells. Additional studies are required to (1) validate the current findings in normal human hepatocytes; (2) explore possible mechanisms of transcriptional regulation of NAT2 by glucose and insulin; and (3) test if NAT2 is involved in lipid and cholesterol metabolism.

FUNDING

National Institutes of Health (NIEHS [National Institute of Environmental Health Sciences] T32-ES011564, NIEHS P30-ES030283, and NIGMS [National Institute of General Medical Sciences] P20-GM113226).

AUTHORSHIP CONTRIBUTIONS

K.U.H.: Conceptualization, project administration, investigation, formal analysis, validation, visualization, writing—original draft preparation. R.A.S.-G.: Investigation, formal analysis, visualization, writing—review and editing. K.M.W.: Investigation, formal analysis, visualization, writing—review and editing. D.W.H.: Conceptualization, supervision, project administration, funding acquisition, writing—review and editing.

DECLARATION OF CONFLICTING INTERESTS

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.