Comparative transcriptomic analysis of vascular endothelial cells after hypoxia/re-oxygenation induction based on microarray technology

Abstract

Objective: To provide comprehensive data to understand mechanisms of vascular endothelial cell (VEC) response to hypoxia/re-oxygenation. Methods: Human umbilical vein endothelial cells (HUVECs) were employed to construct hypoxia/re-oxygenation-induced VEC transcriptome profiling. Cells incubated under 5% O₂, 5% CO₂, and 90% N₂ for 3 h followed by 95% air and 5% CO₂ for 1 h were used in the hypoxia/re-oxygenation group. Those incubated only under 95% air and 5% CO₂ were used in the normoxia control group. Results: By using a well-established microarray chip consisting of 58 339 probes, the study identified 372 differentially expressed genes. While part of the genes are known to be VEC hypoxia/re-oxygenation-related, serving as a good control, a large number of genes related to VEC hypoxia/re-oxygenation were identified for the first time. Through bioinformatic analysis of these genes, we identified that multiple pathways were involved in the reaction. Subsequently, we applied real-time polymerase chain reaction (PCR) and western blot techniques to validate the microarray data. It was found that the expression of apoptosis-related proteins, like pleckstrin homology-like domain family A member 1 (PHLDA1), was also consistently up-regulated in the hypoxia/re-oxygenation group. STRING analysis found that significantly differentially expressed genes SLC38A3, SLC5A5, Lnc-SLC36A4-1, and Lnc-PLEKHJ1-1 may have physical or/and functional protein–protein interactions with PHLDA1. Conclusions: The data from this study have built a foundation to develop many hypotheses to further explore the hypoxia/re-oxygenation mechanisms, an area with great clinical significance for multiple diseases.

1. Introduction

Vascular endothelial cells (VECs) function as a barrier and active tissue in the inner layer of blood vessels, and play an important role in the reaction of organ ischemia, hypoxia, and inflammatory immunity. VECs participate in the regulation of many important balances by secreting a variety of active substances and regulating organ blood flow and coagulation, leukocyte and platelet activity (Nallamshetty et al., 2013; Lim To et al., 2015; Baldea et al., 2018). It has been reported that the endothelial cells seem to provide better hypoxia tolerance than many other cells (Filippi et al., 2018), and the relatively full extent VEC gene expression induced by hypoxia has been established (Urbanek et al., 2014; Sun et al., 2015; Wu et al., 2015; Tang et al., 2016; Baldea et al., 2018).

However, hypoxia/re-oxygenation is the common feature and may have more adverse consequences in many physiological and pathological processes, such as inflammation, obstructive sleep apnea, diabetes, atherosclerosis, myocardial infarction, ischemic stroke, pulmonary embolism, shock, cardiac arrest, cardiac pulmonary bypass, organ transplantation, and tumorigenesis (Atkeson et al., 2009; Shay and Celeste Simon, 2012; Bader et al., 2015; Pan J et al., 2015; Jiang et al., 2017; Xie et al., 2017; Li F et al., 2018; Li WY et al., 2018; Wang et al., 2018; Zhang et al., 2018; Pan H et al., 2019). In contrast to general belief, VEC hypoxia/re-oxygenation injury has recently been identified as a driver rather than a bystander to result in the possible development of organ dysfunction (Carden and Granger, 2000). So far, increased free radical generation and excessive activation of inflammatory responses are considered as important pathogeneses of endothelial hypoxia/re-oxygenation injury (Shay and Celeste Simon, 2012; Wu et al., 2015; Tang et al., 2016; Ferrucci et al., 2018; Filippi et al., 2018; Taylor et al., 2018; Haybar et al., 2019). It is a potentially serious problem, but the detailed physiological processes and transcriptome profiling of VECs after exposure to hypoxia/re-oxygenation have not yet been elucidated.

The present research is the first to focus on the comprehensive gene expression of hypoxia/re-oxygenation-induced human umbilical vein endothelial cells (HUVECs) using a microarray technique.

2. Materials and methods

2.1. Cell lines and cell culture

HUVECs were kindly provided by Procell Life Science & Technology Co., Ltd. (Wuhan, China). High-glucose Dulbecco’s modified Eagle medium (DMEM; Thermo Fisher Scientific, Shanghai, China) had 10% (0.1 g/mL) fetal bovine serum (Biological Industries Israel Beit Haemek Ltd., Beit Haemek, Israel), and 100 U/mL streptomycin and 100 U/mL penicillin (Genom, Hangzhou, China) added were used for cell culture. Before being induced by hypoxia/re-oxygenation, cells in all groups were replaced with fresh medium at 37 °C under 5% CO₂ and 95% air for 12 h. Then, HUVECs in the hypoxia group were incubated under 5% O₂, 5% CO₂, and 90% N₂ by 3131 Single Tri-gas 184L incubator (Thermo Fisher Scientific) for 3 h (Niu et al., 2019); HUVECs in the hypoxia/re-oxygenation group were incubated under 5% O₂, 5% CO₂, and 90% N₂ for 3 h followed by 95% air and 5% CO₂ for 1 h; HUVECs in the normoxia group were cultured under 95% air and 5% CO₂.

2.2. Construction of transcriptome profile

Agilent SurePrint G3 Human Gene Expression v3 Microarray (8×60K; Design ID 072363, Agilent, Palo Alto, CA, USA) was used for the construction of transcriptome profiling. Each of three samples from the hypoxia/re-oxygenation and normoxia groups was used for the microarray experiment. Total RNA was extracted using TRIzol (Life Technologies, Waltham, MA, USA), quantified using the NanoDrop ND-2000 (Thermo Fisher Scientific), and assessed for integrity using Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA) following the manufacturer’s instructions. Total RNA (0.2 μg) of each sample was used for library construction based on the manufacturer’s standard protocols. The Agilent Scanner G2505C (Agilent Technologies) was used for array scanning. Feature Extraction software (Version 10.7.1.1, Agilent Technologies) was used to extract raw data by reading array images.

2.3. Analysis of microarray raw data

The software GeneSpring (Version 13.1, Agilent Technologies) was employed to normalize the raw data with the quantile algorithm. The probes that at least 100% of the values in any one out of all conditions have flags in “Detected” were chosen for differentially expressed gene analysis. Fold change of ≥2.0 and P value of ≤0.05 calculated with the t-test were used as the threshold set to identify a differentially expressed gene. Afterwards, Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis were applied to explore the potential functions and pathways of these differentially expressed genes. Fold enrichment of GO/KEGG analysis=(list hits/list total)/(population hits/population total), where “list hits” is the number of differentially expressed genes annotated to each GO/KEGG entry, “list total” is the number of differentially expressed genes involved in GO/KEGG analysis, “population hits” is the number of genes in whole microarray annotated to each GO/KEGG entry, and “population total” is the number of genes in whole microarray involved in GO/KEGG analysis. P≤0.05 indicates significant enrichment. The distinguishable gene expression pattern among samples was established by hierarchical clustering. The raw and normalized data have been submitted to the National Center of Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm. nih.gov/geo), GSE144715.

2.4. Real-time PCR

The quantitative real-time polymerase chain reaction (qRT-PCR) was used for the messenger RNA (mRNA) expression assay. Total RNA was extracted from cells using the RNeasy® Plus Mini kit (Redwood City, USA). The qRT-PCR was performed using TB Green™ Premix Ex Taq™ II kit (TaKaRa, Japan). The specific primers were synthesized by Sangon Biotech Co., Ltd. (Shanghai, China). PHLDA1-forward: 5'-GG AGAGTAGCGGCTGCAAAG-3', PHLDA1-reverse: 5'-CCTCGGTGAGGATGCAACAC-3'. Biosystems™ 7500 Fast Dx Real-Time PCR system (Thermo Fisher Scientific) was used for the PCR reaction. Applied Biosystems™ 7500 Fast Real-Time PCR software (Thermo Fisher Scientific) was employed for quantitative analysis. Relative quantification was analyzed by the 2^{−ΔΔ C T} method. β-Actin was used as housekeeping control.

2.5. Western blot assay

The protein expression was detected by western blot. Extracted protein from cells was quantified by BCA Protein Assay kit (Thermo Fisher Scientific), mixed with 5×sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) sample loading buffer (Beyotime Biotechnology, Shanghai, China), boiled, separated by 12% (0.12 g/mL) SDS-PAGE (80 V in the stacking gel and 120 V in the separating gel), and transferred to a polyvinylidene fluoride (PVDF) membrane at 200 mA (Millipore, Billerica, MA, USA). The PVDF membrane was then blocked with 5% (0.05 g/mL) skimmed milk/phosphate-buffered saline with Tween 20 (PBST) for 60 min at room temperature, and incubated with rabbit monoclonal for PHLDA1 (Abcom, Shanghai, China) or β-actin (CST, Danvers, MA, USA) at 4 °C overnight, followed by being incubated in 5% (0.05 g/mL) skimmed milk/PBST containing anti-rabbit peroxidase-conjugated secondary antibodies (Proteintech, Wuhan, China) for 60 min at room temperature. Finally, the membranes were washed with PBST. The images of the western blot were acquired using Image Lab 3.0 by Bio-Rad ChemiDoc System (Bio-Rad, Hercules, California, USA).

2.6. Protein interaction analysis

The STRING database was used for protein–protein interaction analysis (http://string-db.org).

2.7. Statistical analysis

The continuous data were expressed as mean±standard deviation (SD). Statistical analysis between means was determined using the Student’s t-test. P value of <0.05 was considered statistically significant.

3. Results

3.1. Reliability and repeatability analyses of microarray-based hypoxia/re-oxygenation-induced HUVEC transcriptome profiling

HUVECs were employed to construct hypoxia/re-oxygenation-induced VEC transcriptome profiling using an Agilent SurePrint G3 Human Gene Expression v3 Microarray. HUVECs incubated under 5% O₂, 5% CO₂, and 90% N₂ for 3 h followed by 95% air and 5% CO₂ for 1 h were used as the hypoxia/re-oxygenation group. HUVECs incubated under 95% air and 5% CO₂ were used as the normoxia control group. Rigorous data analysis was applied to ensure that genes identified by microarray analysis were truly differentially expressed. After exclusion of absent calls and probes that were expressed in fewer than half of experimental samples, 29 986 out of an initial 58 339 probes (51.4%) were considered for analysis (Fig. 1a). Principal component analysis (PCA) was used to evaluate the distribution of samples in the hypoxia/re-oxygenation and normoxic control groups. The results showed that the biological repeatability and uniformity of the experimental samples were both high, which means that the experimental design was reasonable (Fig. 1b).

Fig. 1.

Characteristics of microarray-based HUVEC hypoxia/re-oxygenation expression profile

(a) Composition analysis of microarray-based HUVEC hypoxia/re-oxygenation expression profile. A total of 58 339 probes were used for microarray; 28 353 probes, which were not detected in more than half of samples, were invalid probes (48.6%); 29 986 probes, which were detected in more than three samples, were used for transcriptome profiling analysis (51.4%). (b) Principal component analysis (PCA). The closer the samples of the same group were in the two-dimensional space, the more representative these samples are and the better the biological repetition was. HUVEC: human umbilical vein endothelial cell; H/R_HUVEC: hypoxia/re-oxygenated group; CG_HUVEC: normoxia control group

3.2. Hypoxia/re-oxygenation-related differentially expressed gene probes

In this study, a statistically significant up-regulation or down-regulation of differentially expressed genes was screened according to fold change of ≥2.0 and P value of ≤0.05. This is represented by a volcano plot (Fig. 2a). A total of 372 differentially expressed gene probes, including 106 up-regulated and 266 down-regulated gene probes, were identified. These differentially expressed gene probes were represented by a heat map of a hierarchical clustering based on HUVEC O₂ condition (Fig. 2b); the figure demonstrates very low inter-sample variation in gene expression between replicates. The top 30 differentially expressed gene probes with Entrez Gene ID that are most strongly up-regulated and down-regulated following exposure to hypoxia/re-oxygenation are listed in Tables 1 and 2. The full list of differentially regulated gene probes is shown in Table S1) and is available at http://www. ncbi.nlm.nih.gov/geo (GSE144715).

Fig. 2.

Hypoxia/re-oxygenation-induced differentially regulated probes

(a) Volcano plot for differential expression screening. To analyze the significantly differently expressed genes between samples in the hypoxia/re-oxygenated and normoxia groups using the t-test, the log₂ value of fold change is used as the abscissa, and the negative log₁₀ value of the P value of the t-test is used as the ordinate to obtain the volcano plot. Gray dots: genes with the P value of >0.05; Green dots: genes with absolute fold change of <2 and P value of ≤0.05; Red dots: genes with fold change of ≥2 and P value ≤0.05; These are significantly up-regulated differentially expressed genes. Blue dots: genes with fold change of ≤−2 and P value of ≤0.05; These are significantly down-regulated differentially expressed genes. (b) Heat map of a hierarchical clustering based on HUVEC O₂ condition. Gene expression profiling of hypoxia/re-oxygenation and normoxia HUVECs demonstrated that a total of 372 probes were differentially expressed (fold change of <2, P value of ≤0.05). A heat map is generated where each row representes one probe set and each column representes one experimental sample. The probes are ordered by correlation and significance. The intensity of color in a cell represents the normalized expression of the probe, where red and green colorings depict high and low expression, respectively, in the hypoxia/re-oxygenation condition compared to the normoxia control

Table 1.

Thirty most highly up-regulated genes with Entrez Gene ID of HUVECs induced by hypoxia/re-oxygenation

Probe name	Gene name	Gene symbol	Gene identifier	Entrez Gene ID	Fold change*	P value
A_23_P21057	Septin 1	SEPT1	NM_052838	1731	9.36	0.03
A_32_P211248	Long intergenic non-protein coding RNA 1405	LINC01405	NR_036513	100131138	9.23	0.01
A_21_P0012449	Long intergenic non-protein coding RNA 971	LINC00971	NR_033860	440970	5.77	0.03
A_21_P0000019	Lysine demethylase 4C	KDM4C	NM_001146695	23081	5.70	0.03
A_24_P366122	Acyl-CoA binding domain containing 4	ACBD4	NM_024722	79777	4.95	0.01
A_24_P160380	PDZ and LIM domain 2	PDLIM2	NM_176871	64236	4.84	0.01
A_22_P00002166	Family with sequence similarity 229, member A	FAM229A	NM_001167676	100128071	4.39	0.03
A_33_P3285271	Flavin containing dimethylaniline monooxygenase 6 pseudogene	FMO6P	NR_002601	388714	3.88	0.02
A_33_P3422085	SPANX family member N2	SPANXN2	NM_001009615	494119	3.73	0.02
A_23_P159893	Chordin-like 1	CHRDL1	NM_145234	91851	3.64	0.04
A_22_P00014410	SACS antisense RNA 1	SACS-AS1	NR_103450	100506680	3.51	0.05
A_33_P3423425	Zinc finger protein 770	ZNF770	NM_014106	54989	3.49	0.00
A_24_P322229	RAS like family 10 member B	RASL10B	NM_033315	91608	3.39	0.02
A_33_P3211078	High mobility group AT-hook 2	HMGA2	NM_001300918	8091	3.33	0.04
A_21_P0001900	Gastric cancer associated transcript 1	GACAT1	NR_126369	104326057	3.22	0.01
A_33_P3384710	Family with sequence similarity 222 member B	FAM222B	NM_001288631	55731	3.19	0.01
A_23_P420692	Protein tyrosine phosphatase receptor type F (PTPRF) interacting protein α 4	PPFIA4	XM_006711588	8497	3.04	0.02
A_33_P3286953	ADAM metallopeptidase with thrombospondin type 1 motif 6	ADAMTS6	NM_197941	11174	2.96	0.02
A_23_P102000	C-X-C motif chemokine receptor 4	CXCR4	NM_001008540	7852	2.95	0.05
A_33_P3294302	Pleckstrin homology-like domainfamily A member 1	PHLDA1	NM_007350	22822	2.87	0.02
A_21_P0006019	Uncharacterized LOC101927358	LOC101927358	NR_121182	101927358	2.84	0.03
A_22_P00009701	Uncharacterized LOC100131655	LOC100131655	NR_040024	100131655	2.81	0.05
A_24_P362904	6-Phosphofructo-2-kinase/fructose-2,6-biphosphatase 4	PFKFB4	NM_004567	5210	2.81	0.05
A_33_P3771067	Uncharacterized LOC283140	LOC283140	NR_126004	283140	2.79	0.02
A_23_P46131	Family with sequence similarity 110 member D	FAM110D	NM_024869	79927	2.67	0.04
A_23_P45751	Chloride channel accessory 4	CLCA4	NM_012128	22802	2.66	0.02
A_23_P16415	Low density lipoprotein (LDL) receptor-related protein 3	LRP3	NM_002333	4037	2.66	0.04
A_23_P62647	Signaling lymphocytic activation molecule family member 1	SLAMF1	NM_003037	6504	2.64	0.02
A_24_P221575	RUN and FYVE domain containing 3	RUFY3	NM_001037442	22902	2.63	0.03
A_33_P3222018	Chromobox 3	CBX3	NM_007276	11335	2.62	0.03

^*The fold change is (hypoxia/re-oxygenation group)/(normoxia group)

Table 2.

Thirty most highly down-regulated genes with Entrez Gene ID of HUVECs induced by hypoxia/re-oxygenation

Probe name	Gene name	Gene symbol	Gene identifier	Entrez Gene ID	Fold change*	P value
A_21_P0007488	Long intergenic non-protein coding RNA 1490	LINC01490	NR_120468	101928420	33.12	0.00
A_22_P00018708	Uncharacterized LOC101927913	LOC101927913	XR_241962	101927913	7.43	0.03
A_23_P389500	Regenerating family member 1 β	REG1B	NM_006507	5968	6.02	0.00
A_22_P00020802	Uncharacterized LOC101927694	LOC101927694	NR_104648	101927694	5.67	0.04
A_33_P3268035	Lysozyme-like 1	LYZL1	NM_032517	84569	5.35	0.03
A_24_P327084	Serine protease 33	PRSS33	NM_152891	260429	5.02	0.00
A_23_P86411	Myosin IIIA	MYO3A	NM_017433	53904	4.47	0.01
A_23_P111978	Potassium two-pore-domain channel subfamily K member 9	KCNK9	NM_001282534	51305	4.46	0.01
A_21_P0009060	Uncharacterized LOC101928035	LOC101928035	NR_104657	101928035	4.31	0.01
A_23_P36050	Olfactory receptor family 8 subfamily K member 3 (gene/pseudogene)	OR8K3	NM_001005202	219473	4.26	0.03
A_23_P502590	Killer cell immunoglobulin-like receptor, two Ig domains and short cytoplasmic tail 4	KIR2DS4	NM_012314	3809	4.17	0.05
A_33_P3238182	Long intergenic non-protein coding RNA 588	LINC00588	NR_026772	26138	4.14	0.03
A_19_P00809368	Long intergenic non-protein coding RNA 1215	LINC01215	NR_110028	101929623	4.06	0.01
A_32_P234926	BANF family member 2	BANF2	NM_001014977	140836	4.04	0.04
A_32_P155666	Endothelin converting enzyme-like 1	ECEL1	NM_004826	9427	4.03	0.00
A_21_P0001945	Uncharacterized LOC101927156	LOC101927156		101927156	4.00	0.03
A_24_P361427	TBC1 domain family member 29	TBC1D29	NM_015594	26083	3.76	0.04
A_22_P00010840	Uncharacterized LOC101927272	LOC101927272	NR_110908	101927272	3.74	0.02
A_21_P0010721	Long intergenic non-protein coding RNA 869	LINC00869	XR_426752	57234	3.62	0.04
A_23_P419156	Opsin 1, short wave sensitive	OPN1SW	NM_001708	611	3.55	0.02
A_22_P00014526	CFAP44 antisense RNA 1	CFAP44-AS1	NR_046728	100874029	3.49	0.01
A_21_P0004097	Uncharacterized LOC101929261	LOC101929261	XR_241733	101929261	3.47	0.02
A_33_P3235721	Chromosome 11 open reading frame 87	C11orf87	NM_207645	399947	3.42	0.04
A_22_P00002954	Uncharacterized LOC101929261	LOC101929261	XR_241733	101929261	3.38	0.04
A_33_P3882624	Protection of telomeres 1	POT1	NM_015450	25913	3.36	0.00
A_33_P3262635	Cat eye syndrome chromosome region, candidate 1	CECR1	NM_001282225	51816	3.30	0.05
A_21_P0004219	Long intergenic non-protein coding RNA 1018	LINC01018	NR_024424	255167	3.30	0.01
A_33_P3307795	Family with sequence similarity 124 member B	FAM124B	NM_024785	79843	3.28	0.00
A_22_P00008580	Long intergenic non-protein coding RNA 1231	LINC01231	NR_121585	101929247	3.26	0.03
A_23_P150064	Multimerin 2	MMRN2	NM_024756	79812	3.19	0.04

^*The fold change is (normoxia group)/(hypoxia/re-oxygenation group)

3.3. Pathways and functions mapping with hypoxia/re-oxygenation-sensitive gene probes

In order to better understand the genomic response to hypoxia/re-oxygenation of HUVECs, the 372 differentially expressed gene probes were functionally annotated and clustered using GO analysis and KEGG analysis. A total of 803 GO IDs were matched, of which 486 GO IDs were related to a biological process, 149 GO IDs were related to a cellular component, and 168 GO IDs were related to molecular function. Based on the GO analysis results, the molecular and cellular functions related to production of oxygen free radicals, calcium overload, excessive activation of inflammation, and cell injury, including endothelial cell proliferation, apoptosis, angiogenesis, vasoconstriction, and coagulation are listed in Table 3.

Table 3.

Pathways and functions associated with hypoxia/re-oxygenation analyzed by GO

Hypoxia/re-oxygenation-related GO	GO ID	Related differentially regulated gene	Fold enrichment#	P value*
Production of oxygen free radicals
Positive regulation of phospholipase activity	GO:0010518	FGFR2	22.13	0.05
Negative regulation of nitric-oxide synthase activity	GO:0051001	CNR2	22.13	0.05
Neutrophil activation	GO:0042119	CXCR4	17.70	0.06
Nitric-oxide synthase binding	GO:0050998	SCN5A	11.94	0.09
Dioxygenase activity	GO:0051213	KDM4C	9.83	0.10
Response to hypoxia	GO:0001666	CXCR4, ALDH3A1	2.23	0.23
Calcium overload
G-protein-coupled receptor signaling pathway	GO:0007186	GPR31, OPN1SW, OR13H1, OR2A2, NPBWR1, NPY6R, OR10G8, OR10J3, OR8K3, CCL3L3, CXCR4, ADGRG3, OR10J5, FFAR2	3.11	0.00
Sodium ion transmembrane transport	GO:0035725	SLC5A5, ASIC2, SCN5A	6.43	0.01
Lipid digestion	GO:0044241	CEL	29.51	0.04
Sodium ion transport	GO:0006814	SLC38A3, SCN5A	4.82	0.07
Phospholipase binding	GO:0043274	WAS	9.28	0.11
Calcium channel regulator activity	GO:0005246	NPY	5.96	0.16
Phosphatidylinositol phospholipase C activity	GO:0004435	PLCH2	5.96	0.16
Excessive activation of inflammation
Negative regulation of interferon-γ biosynthetic process	GO:0045077	INHA	35.42	0.03
Neutrophil chemotaxis	GO:0030593	ITGA9, CCL3L3	5.57	0.05
Chemokine-mediated signaling pathway	GO:0070098	CCL3L3, CXCR4	5.09	0.06
Complement activation	GO:0006956	C1QA	3.68	0.24
Positive regulation of inflammatory response	GO:0050729	CCL3L3	3.10	0.28
Cell injury
Regulation of vascular endothelial growth factor receptor signaling pathway	GO:0030947	TMEM204	35.42	0.03
Positive regulation of epithelial cell proliferation	GO:0050679	FGFR2, SCN5A	6.05	0.05
Negative regulation of cell migration involved in sprouting angiogenesis	GO:0090051	MMRN2	22.13	0.05
Blood coagulation, intrinsic pathway	GO:0007597	VWF	9.83	0.10
Positive regulation of cell proliferation	GO:0008284	SLAMF1, VWCE, FGFR2, KDM4C, ALDH3A1	2.03	0.11
Regulation of vasoconstriction	GO:0019229	ASIC2	8.42	0.12
Angiogenesis	GO:0001525	MMRN2, FGFR2, NRCAM	2.47	0.13
Positive regulation of apoptotic process	GO:0043065	PHLDA1, HMGA2, NEURL1	1.89	0.22
DNA damage response, signal transduction by p53 class mediator resulting in cell cycle arrest	GO:0006977	PSMB11	2.67	0.32
Apoptotic process	GO:0006915	PSMB11, PHLDA1, FGFR2, CXCR4	1.17	0.45
Negative regulation of cell proliferation	GO:0008285	CCL3L3, NEURL1	0.92	0.64

^#Fold enrichmen=(list hits/list total)/(population hits/population total), where “list hits” is the number of differentially expressed genes annotated to each Gene Ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) entry, “list total” is the number of differentially expressed genes involved in GO/KEGG analysis, “population hits” is the number of genes in whole microarray annotated to each GO/KEGG entry, and “population total” is the number of genes in whole microarray involved in GO/KEGG analysis.

^* P value of fold enrichment, and P≤0.05 indicates significant enrichment

A total of 86 KEGG IDs were matched based on the known information, in which 42 differentially expressed genes were involved. Eleven differentially expressed genes participate in no less than five KEGG pathways, including fibroblast growth factor receptor 2 (FGFR2; 10 KEGG pathways), Wiskott-Aldrich syndrome (WAS; 10 KEGG pathways), aldehyde dehydrogenase 3 family, member A1 (ALDH3A1; 9 KEGG pathways), integrin α9 (ITGA9; 8 KEGG pathways), cytochrome c oxidase subunit VIb polypeptide 2 (testis) (COX6B2; 7 KEGG pathways), complement component 1, q subcomponent, A chain (C1QA; 6 KEGG pathways), chemokine (C-C motif) ligand 3-like 3 (CCL3L3; 6 KEGG pathways), chemokine (C-C motif) ligand 4-like 2 (CCL4L2; 6 KEGG pathways), carboxyl ester lipase (CEL; 5 KEGG pathways), chemokine (C-X-C motif) receptor 4 (CXCR4; 7 KEGG pathways), and von Willebrand factor (VWF; 5 KEGG pathways). Among these functionally active differentially expressed genes, VWF is an important endothelial cell injury marker. CCL3L3, CCL4L2, CXCR4, and WAS all participate in the chemokine signaling pathway. Therefore, we showed these two annotated KEGG pathway maps, path: hsa04610 and path:hsa04062, to indicate that these differentially expressed genes may play important roles in hypoxia/re-oxygenation, especially in inflammatory response, endothelial cytoskeleton regulation, permeability, cellular growth and differentiation, cell lysis, apoptosis, reactive oxygen species (ROS) production, and nitric oxide (NO) production (Fig. 3).

Fig. 3.

Hypoxia/re-oxygenation-related KEGG pathway and differentially expressed genes

(a) von Willebrand factor (VWF), complement component 1, q subcomponent, A chain (C1QA), and complement and coagulation cascades pathway. On the map of the complement and coagulation cascades pathway, endothelial cell injury marker VWF, which is significantly down-expressed in hypoxia/re-oxygenation group in our database, plays an important role in thrombogenesis, endothelial permeability, and anti-inflammatory responses. C1QA, which is significantly up-expressed in the hypoxia/re-oxygenation group in our database, initiates the complement cascade and is involved in cell lysis, chemotaxis, and phagocytosis. The migration, aggregation, and phagocytosis of neutrophils are the important mechanisms leading to the large release of reactive oxygen species (ROS) during ischemia-reperfusion injury. (b) Chemokine (C-C motif) ligand 3-like 3 (CCL3L3), chemokine (C-C motif) ligand 4-like 2 (CCL4L2), chemokine (C-X-C motif) receptor 4 (CXCR4), Wiskott-Aldrich syndrome (WAS), and chemokine signaling pathway. Significantly down-expressed chemokines CCL3L3 and CCL4L2 and significantly up-expressed chemokine receptor CXCR4 are the initiating molecules of the chemokine signaling pathway. This pathway is crucial in cellular growth and differentiation, cell survival and apoptosis, ROS and nitric oxide (NO) production. WAS, which is on one branch involved in regulation of actin cytoskeleton, is significantly down-expressed in the hypoxia/re-oxygenation group in our database

3.4. Expression verification of differentially expressed gene PHLDA1 and exploration of related function based on HUVEC hypoxia/re-oxygenation transcriptome profiling

PHLDA1 was first identified as a potential transcription factor required for Fas expression and activation-induced apoptosis in mouse T cell hybridomas (Moad et al., 2013). However, its expression is induced by a variety of external stimuli, and it participates in apoptosis, autophagy, cell proliferation and differentiation, inflammation, and fat metabolism (Johnson et al., 2011; Sellheyer and Nelson, 2011). Recently, PHLDA1 has received increased attention because of its association with cancer (Nagai, 2016; Fearon et al., 2018). In our microarray database, PHLDA1 was significantly up-expressed in the hypoxia/re-oxygenation group. Using real-time PCR and western blot assay, we validated that the mRNA and protein were both significantly highly expressed in the hypoxia/re-oxygenation group compared to the normoxia group (Figs. 4a–4c). Based on the STRING search results of PHLDA1 protein–protein interactions (Fig. 4d), we found many potentially related proteins SLC38A3, SLC5A5, Lnc-SLC36A4-1, and Lnc-PLEKHJ1-1 in our differentially expressed database.

Fig. 4.

Expression verification of differentially expressed gene PHLDA1 and exploration of function-related genes based on HUVEC microarray database

(a, c) Western blot measurement of PHLDA1 expression in HUVECs influenced by hypoxia/re-oxygenation. (b) Effects of hypoxia/re-oxygenation on the mRNA expression of PHLDA1 in HUVECs. Results are normalized to β-actin. Data are expressed as mean±standard deviation (SD), n=3. Normoxia group: HUVECs incubated under 95% air and 5% CO₂; Hypoxia group: HUVECs incubated under 5% O₂, 5% CO₂ and 90% N₂ for 3 h; Hypoxia/re-oxygenation group: HUVECs incubated under hypoxia for 3 h followed by 95% air and 5% CO₂ for 1 h. ^* P<0.05, compared with the hypoxia/re-oxygenation group. (d) PHLDA1-related protein–protein interaction analysis using STRING database. The interactions include both the direct physical interactions between proteins and the indirect functional correlations between proteins. Related proteins SLC38A3, SLC5A5, Lnc-SLC36A4-1, and Lnc-PLEKHJ1-1 were significantly differentially expressed in our database

4. Discussion

The known VEC hypoxia/re-oxygenation injury mechanisms consist of a complex pathophysiology involving activation of cell death programs, endothelial dysfunction, transcriptional reprogramming, and activation of the innate and adaptive immune system. Numerous pathways and signaling cascades are implicated (Salvadori et al., 2015). VECs undergo swelling, loss of pinocytotic vesicles, degradation of the cytoskeleton, and lifting from the underlying basement membrane. As a consequence, intercellular contact of endothelial cells is lost, increasing vascular permeability and fluid loss to the interstitial space (Basile et al., 2011). In addition, there is impaired endothelial function by directly reducing endothelial NO production at the transcriptional and posttranscriptional levels, leading to vasoconstriction and aggravating tissue ischemia despite reperfusion (McQuillan et al., 1994; Liao et al., 1995). Another important feature of VEC hypoxia/re-oxygenation injury is the chemotaxis of leukocytes, the accumulation and adhesion of circulating leukocytes (primarily neutrophils) to the endothelial cell surface, and enhanced oxidative stress, leading to vessel inflammation and progression (Eltzschig and Collard, 2004). As shown by research, this process is initiated by increased expression of P-selectin on the VECs and interaction of P-selectin with P-selectin glycoprotein 1 (PSGL-1) expressed on the leukocytes. Subsequently, firm adherence of the leucocytes to the endothelium is achieved by the interaction of the β2-integrins lymphocyte function-associated antigen I (LFA-I) and macrophage-l antigen (MAC-I or complement receptor 3 (CR3)) on the leukocyte and the intracellular adhesion molecule 1 (ICAM-I) on the VEC. Platelet endothelial cell adhesion molecule 1 (PECAM-I) facilitates leukocyte transmigration into the interstitial space (Carden and Granger, 2000).

The present study used whole-transcriptome microarray to explore the genomic response of HUVECs to hypoxia/re-oxygenation. The microarray analysis identified that approximately 29 986 gene probes were involved in regulation under hypoxia/re-oxygenation exposure. Among them, 106 up-regulated gene probes and 266 down-regulated gene probes were identified as the differentially expressed gene probes. These encompass gene probes were involved in a multitude of pathways and functions, including production of oxygen free radicals, calcium overload, excessive activation of inflammation, glucose and lipid metabolism, endothelial cell proliferation, differentiation, cytoskeleton regulation, permeability, cell lysis, apoptosis, and angiogenesis. Many of the aforementioned VEC hypoxia/re-oxygenation-related genes (such as: regulation of actin cytoskeleton-related WAS; NO synthase-related CNR2, SCN5A; regulation of vasoconstriction-related ASIC2; neutrophil chemotaxis and adhesion-related CCL3L3, CCL4L2, CXCR4, ITGA9; neutrophil activation and ROS-related CXCR4, KDM4C; regulation of cell proliferation and apoptosis-related SLAMF1, VWCE, FGFR2, KDM4C, ALDH3A1, NEURL, PHLDA1, HMGA2, PSMB11) were all found differentially expressed in the hypoxia/re-oxygenation group in our database. However, many VECs functionally related important genes (such as: the endothelial cell injury marker and coagulation-related VWF; regulation of vascular endothelial growth factor (VEGF) receptor signaling pathway-related TMEM204; angiogenesis-related MMRN2, FGFR2, NRCAM) and ischaemia and reperfusion injury (IRI)-related genes (such as: phosphatidylinositol phospholipase C (PLC) activity-related PLCH2; ion transport-related SLC38A3, SCN5A, NPY; complement activation-related C1QA; lipid digestion-related CEL), which were found as the differentially expressed genes, may also play a crucial role in VEC hypoxia/re-oxygenation injury.

In fact, many meaningful hypoxia genes, such as hypoxia-inducible factor 1-α (HIF1A) and VEGF-related genes, were also contained in the microarray, including probes A_24_P56388 (HIF1A), A_33_P3231277 (HIF1A), A_22_P00016429 (HIF1A antisense RNA 1 (HIF1A-AS1)), A_22_P00020043 (HIF1A-AS1), A_23_P46964 (HIF1AN), A_33_P3338152 (HIF3A), A_23_P142187 (HIF3A), A_23_P338534 (HIF3A), A_24_P12401 (VEGFA), A_23_P70398 (VEGFA), A_23_P1594 (VEGFB), A_23_P167096 (VEGFC), and A_21_P0003801 (Lnc-VEGFC-1). However, the expression levels of these genes were not significantly different between the hypoxic/re-oxygenated group and the normoxia group. On the one hand, it may be because these genes are indeed not significantly different in this model; on the other hand, it may be because they are more active under the condition of hypoxia and are relatively silent after re-oxygenation, while other genes are more active.

In addition, it has been found that endothelial cells are particularly resistant to hypoxic stress, and can maintain proliferation even after having been cultured in mild hypoxia in a time-dependent manner (Bader et al., 2015; Sun et al., 2015; Baldea et al., 2018). Our results are also consistent with the above research. As mentioned earlier, PHLDA1 acts as a mediator of apoptosis (Moad et al., 2013), plays an important role in cancer (Nagai, 2016), and is annotated to “positive regulation of the apoptotic process” GO ID (Table 3). Expression verification of PHLDA1 mRNA or protein in our study had no obvious change after only exposure to hypoxia for 3 h (5% O₂), but was significantly up-expressed after re-oxygenation. This is consistent with our microarray results (Fig. 4). Recent research has also shown that PHLDA1 was highly expressed in myocardial ischemia-reperfusion injuries, and knockdown of PHLDA1 can attenuate oxidative stress-induced ROS production and apoptosis in cardiomyocytes (Guo et al., 2020). PHLDA1 may promote oxidative stress-related apoptosis by binding to BCL2-associated X, apoptosis regulator (BAX). This research further validates our research in which PHLDA1 may play an important role in VEC hypoxia/re-oxygenation injury.

Long non-coding RNAs (lncRNAs) in the nucleus have functions in histone modification and block the binding of transcription factors to their promoters or direct transcriptional regulation (E et al., 2018; Mansoori et al., 2018). More than 200 diseases are related to dysregulated or dysfunctional lncRNAs (Zampetaki et al., 2018; Fu et al., 2019; Gudenas et al., 2019). Thus, exploration and identification of hypoxia/re-oxygenation-associated lncRNAs and their functions may provide a new insight into the understanding of VEC hypoxia/re-oxygenation pathogenesis and define novel therapeutic targets. Recent research reported that lncRNA HIF1A-AS2 can recruit lysine-specific demethylase 1 (LSD1) and epigenetically repress PHLDA1 (Mineo et al., 2016). STRING analysis results showed that significantly differentially expressed genes SLC38A3, SLC5A5, Lnc-SLC36A4-1, and Lnc-PLEKHJ1-1 in our database may have physical or/and functional protein–protein interactions with PHLDA1 (Fig. 4d). The interconnectedness of these genes is expected to reveal some potential hypoxia/re-oxygenation mechanisms related to lncRNAs. However, the validation of PHLDA1 and the analysis of interacting proteins were just one of many new discoveries in this study, and more validation and mechanism studies of the hypoxia/re-oxygenation genes found here will be further revealed.

5. Conclusions

In summary, our microarray-based HUVEC hypoxia/re-oxygenation transcriptome profiling research was reasonable in design, with high reliability and repeatability. The differentially expressed genes revealed by the microarray largely compensate for the existing VEC hypoxia/re-oxygenation-related databases. Through GO, KEGG, and STRING analyses, we explored the new VEC hypoxia/re-oxygenation-related functions and protein–protein interactions, and put forward hypotheses for possible mechanisms of some lncRNAs. We believe that the results of this study have significance for understanding the mechanism of VEC hypoxia/re-oxygenation injury.

List of electronic supplementary materials

Table S1

Full list of microarray-based HUVEC hypoxia/re-oxygenation-induced differentially regulated gene probes