Microarray expression profile of mRNAs and long noncoding RNAs and the potential role of PFK-1 in infantile hemangioma

Background Infantile hemangioma (IH) is the most common benign tumor in children. Long noncoding RNAs (lncRNAs) play a critical role in tumorigenesis. However, the expression levels and biological functions of lncRNAs in IH have not been well-studied. This study aimed to analyze the expression profile of lncRNAs and mRNAs in proliferating and involuting IHs. Methods The expression profiles of lncRNAs and mRNAs in proliferating and involuting IHs were identified by microarray analysis. Subsequently, detailed bioinformatics analyses were performed. Finally, quantitative real-time polymerase chain reaction (qRT-PCR) and immunohistochemistry (IHC) analyses were conducted to validate the microarray results. Results In total, 146 differentially expressed (DE) lncRNAs and 374 DE mRNAs were identified. The DE mRNAs were enriched mostly in angiogenesis-related biological processes (BPs) and pathways by bioinformatics analysis. In addition, metabolism-related BPs (e.g., “glycogen biosynthetic process” and “metabolic process”) and pathways (e.g., “oxidative phosphorylation”) were identified. A lncRNA-mRNA co-expression network was constructed from 42 DE lncRNAs and 217 DE mRNAs. Twelve lncRNAs were predicted to have cis-regulated target genes. The microarray analysis results were validated by qRT-PCR using 5 randomly selected lncRNAs and 13 mRNAs. The IHC results revealed that both LOXL2 and FPK-1 exhibited higher protein expression levels in proliferating IH than in involuting IH. Moreover, inhibition of PFK-1 could suppress hemangioma-derived endothelial cell proliferation and migration, induce cell arrest, and reduce glucose uptake and lactate and ATP production. Conclusions The findings suggest that the identified DE lncRNAs and mRNAs may be associated with the pathogenesis of IH. The data presented herein can improve our understanding of IH development and provide direction for further studies investigating the mechanism underlying IH.


Introduction
Infantile hemangioma (IH) is the most common benign tumor in children, with a prevalence of 4-5% [1]. IH is predominantly found in female children; the female:male ratio ranges from 1.4:1 to 3:1 [2]. IH is usually absent at birth and exhibits a characteristic growth pattern with a proliferating phase lasting for 1 year after birth, with the most rapid growth at 5-8 weeks, followed by spontaneous involution lasting up to 5 years [3]. In the

Open Access
Cell Division proliferating phase, the IH will enlarge, become more elevated and develop a rubbery consistency; in contrast, in the involuting phase, the IH will flatten and shrink from the center outward [4]. IHs are most frequently found in the head and neck, followed by the trunk and extremities. Indications for treatment include disfigurement, impaired function and/or a threat to life [5].
Both vasculogenesis and angiogenesis are widely accepted to contribute to the mechanism of IH development [6]. Several signaling pathways that regulate vasculogenesis and angiogenesis have been demonstrated to be associated with the pathogenesis of IH [7,8] primarily including the vascular endothelial growth factor (VEGF) and VEGF receptor pathways, notch pathway, mammalian target of rapamycin pathway, β-adrenergic signaling pathway and angiopoietin and Tie2 signaling pathway [7] [8],. However, the exact pathogenesis of IH has not been fully elucidated.
Long noncoding RNAs (lncRNAs) are most commonly defined as transcripts longer than 200 nucleotides without protein-coding function [9]. LncRNAs were previously considered transcriptional "noise". However, accumulated studies indicate that lncRNAs play a critical role in multiple biological processes (BPs), including the regulation of gene expression at the chromatin modification level, transcriptional and posttranscriptional processing, and cell differentiation and development [9,10]. Recently, lncRNAs have attracted increasing attention in tumorigenesis for their important roles in stemness acquisition and maintenance, growth and metastasis of cancer cells, as well as in vascular diseases [11]. Moreover, lncRNAs have been demonstrated to play a critical role in endothelial cell (EC) differentiation and angiogenesis [12].
Several studies have reported that lncRNAs are implicated in IH [13][14][15]. However, the association between IH and the expression levels and biological functions of lncRNAs remains unclear. This study aimed to analyze the expression profile of lncRNAs and mRNAs in proliferating and involuting IH by microarray analysis. Additionally, using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, we identified the clinical significance of differentially expressed (DE) lncRNAs and mRNAs and validated their expression using quantitative real-time polymerase chain reaction (qRT-PCR) and immunohistochemistry (IHC) staining.

Samples
This study was approved by the Institutional Review Board of the West China Hospital of Sichuan University. Written informed consent was obtained from the parents of all patients. All IH tissues were obtained surgically at the West China Hospital of Sichuan University. In total, tissues from six proliferating IHs and four involuting IHs were collected, and all specimens were stored at − 80 °C after excision. Detailed information on the ten patients is presented in Additional file 1: Table S1.

RNA extraction
Total RNA was extracted from frozen tissues using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and was further purified with a miRNeasy Mini Kit (QIAGEN, Valencia, CA, USA) according to the manufacturer's instructions. RNA quantity and quality were measured with a NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA), and RNA purity and integrity were assessed by agarose gel electrophoresis. Then, RNA samples were stored at -80 °C until further use.

Microarray analysis
LncRNA and mRNA expression profiling were performed using the Affymetrix GeneChip Human Transcriptome Array 2.0 platform (Affymetrix, Inc., Santa Clara, CA, USA). This high-resolution array contains more than 6 million distinct probes, including over 245,000 mRNA and 40,000 lncRNA probes. Raw microarray data were extracted by using the Agilent Feature Extraction (v. 10.7), summarized, normalized and quality-controlled using the GeneSpring GX program software (v. 12.6.1) package (Agilent Technologies) and R program. Then microarray analysis was performed via the Gene-Cloud of Biotechnology Information (GCBI, Shanghai, China; https ://www.gcbi.com.cn) platform. Briefly, the DE lncRNAs and mRNAs between the proliferating group and the involuting group were first identified. Second, GO enrichment and KEGG pathway analyses were performed with the DE genes. Finally, the DE lncRNAs and mRNAs were used to construct a lncRNA-mRNA co-expression network. LncRNAs and mRNAs significantly DE between the two groups were identified by filtering with an adjusted P value < 0.05 and fold change > 1.5 using SAM analysis. A volcano plot was generated to distinguish the significantly DE lncRNAs and mRNAs. Hierarchical clustering was applied to display the expression profiles of significantly DE lncRNAs and mRNAs between the two groups.

GO and KEGG pathway analyses
GO analysis was used to categorize and describe the biological functions of the significantly DE genes using the cellular component (CC), molecular function (MF) and BP categories. Based on the KEGG database, pathway analysis was used to predict the main pathways enriched by the DE genes. Differentially expressed mRNAs were selected and uploaded into the Database for Annotation, Visualization and Integrated Discovery (DAVID, https :// david .ncifc rf.gov/) for annotation and functional analysis, including gene set enrichment analysis and mapping gene sets to the KEGG pathway [19,20]. Both GO enrichment analysis (P < 0.05 and FDR < 0.05) and KEGG pathway analysis (P < 0.05 and FDR < 0.05) of DE mRNAs were performed using the GCBI platform.

Protein-protein interaction (PPI) network analysis
PPI networks are an important tool for the system-level understanding of cellular processes [21]. To identify hub DE mRNAs related to the development of IH, we performed PPI network analysis automatically via STRING version 11.0 online software (https ://strin g-db.org/) [22]. Then, the PPI network was drawn with the Cytoscape software 3.6.1 (The Cytoscape Consortium, San Diego, CA, USA). Moreover, the top 10 hub genes were selected for further functional analysis using Metascape (http:// metas cape.org) [23].

LncRNA-mRNA co-expression network
Based on the correlation analysis between the significantly DE lncRNAs and mRNAs, a lncRNA-mRNA co-expression network was generated to associate the lncRNAs with their potential target mRNAs using Pearson correlation. Pearson correlation coefficients > 0.97 or < − 0.97 with P-values < 0.05 were used as the standard threshold values for identifying the lncRNAs and protein-coding mRNAs. The co-expression network was first constructed via the GCBI platform and was then drawn with the Cytoscape 3.6.1 software.

Prediction of cis-regulated targets
The prediction of cis-regulated target genes was performed based on the results of the lncRNA-mRNA co-expression analysis. mRNAs were considered cis-regulated target genes when the Pearson correlation coefficient was > 0.97 or < − 0.97 and the mRNA locus was within 10 kb of each given lncRNA.

qRT-PCR validation
qRT-PCR analysis was performed using SYBR qPCR Master Mix (Vazyme, Nanjing, China) according to the manufacturer's protocol. The qRT-PCR conditions were as follows: 95 °C for 30 s, followed by 40 cycles at 95 °C for 15 s and 60 °C for 30 s. All experiments were performed in triplicate. The expression levels of lncRNAs and mRNAs were quantified using the 2 −ΔΔCt method and normalized to GAPDH expression. All the primers used in this study are listed in Additional file 2: Table S2.

IHC staining
To assess the protein expression and cellular localization of LOXL2 and PFK-1, fifty pairs of proliferating and involuting IH tissues were selected for IHC analysis. The 5 μm tissue sections were cut and deparaffinized by heating to 60 °C for 1 h, washed three times with xylene for 15 min each, and rehydrated by consecutive washes in 100%, 95%, and 70% ethanol followed by a 5-min wash in water. Then, sections were incubated with 3% hydrogen peroxide for 30 min to inhibit endogenous peroxidase activity and subsequently underwent consecutive washes with PBS. Antigen retrieval was carried out by heating sections twice for 20 min each in EDTA antigen retrieval buffer. Sections were blocked for 30 min in 5% serum at room temperature. Primary antibodies against LOXL2 (1:100, GTX105085, GeneTex) and PFK-1 (1:50, sc-377346, Santa Cruz) were added for incubation overnight at 4 °C. Slides were washed and incubated with secondary antibody at room temperature for 30 min. Images were acquired using a Leica microscope camera (Leica Microsystems, Wetzlar, Germany).

Western blot analysis
Briefly, the protein concentration was determined using the Bradford protein assay kit (Bio-Rad). Then, the protein samples were separated by sodium dodecy1 sulphate-polyacrylamide gel electrophoresis, followed by electrophoretic transfer onto a nitrocellulose membrane. The membrane was incubated with primary antibody in TBST at 4 °C overnight. Next, the membrane was washed three times and incubated with the appropriate secondary antibody. The protein bands were visualized using enhanced ECL-associated fluorography.

Lentiviral vector construction and transfection
Briefly, two specific shRNAs for PFK-1 (shRNA-1 and shRNA-2) were purchased form GeneChem Co., Ltd. (Shanghai, China) and Lipofectamine 2000 (Invitrogen, USA) was applied for transfection according to the manufacturer's protocol. After transfection with 48 h, the cells were collected for various experiments.

CCK-8 assay
Briefly, 5,000 transfected cells were seeded into 96-well plates for 24 h. Then, 10 μl per well of CCK-8 kit reagents were added and incubated for 2 h at 37 °C. Finally, the absorbance of each well was read at 450 nm on a microplate reader. All experiments were independently repeated at least three times.

Migration assay
The transfected cells were plated in serum-free medium in the top chamber. The membrane without a coat (24well insert; 8 μm pore size; Millipore, USA) and medium supplemented with 10% serum was in the lower chamber. After 24 h, the bottom of the chamber insert was stained with methanol and 0.1% crystal violet and then imaged. Each migration assay was conducted at least 3 replicates.

Flow cytometry cell cycle analysis
The cell cycle detection kit purchased from 4A Biotech Co., Ltd. (Beijing, China) was used to detect the cell cycle. HemECs (4.0 × 10 5 /well) were plated into 6-well plates and cultured for 24 h. Then, the cells were collected by trypsinization and washed with cold PBS. Subsequently, 95% cold ethanol was used to immobilize the cells at 4 °C overnight. After washing again with PBS, the cells were incubated with RNase and then labelled with propidium iodide (PI) according to the manufacturer's protocol. A CytoFLEX flow cytometer (Becton-Dickinson, USA) was used to detect the cell cycle. The cell cycle distribution was analyzed with the ModFIT software (BD Biosciences).

Glucose uptake assay
The glucose uptake of HemECs was estimated using the Glucose Uptake Cell-Based Assay Kit (Cayman Chemical, USA). Briefly, HemECs (5 × 10 4 /well) were seeded in 96-well plates and then incubated at 37 °C overnight. On the next day, the cells were treated for 1 h with glucosefree medium and fluorescent 2-NBDG at a concentration in the glucose-free medium. The plate was centrifuged at room temperature for five minutes at 400 g followed by aspirating the supernatant. Then, 200 μl of Cell-Based Assay Buffer was added to each well, and the cells were analyzed immediately using a CytoFLEX flow cytometer (Becton-Dickinson, USA).

Lactate production
Lactate levels was measured using the L-Lactate Assay Kit (Cayman Chemical, USA) according to the manufacturer's protocol. Briefly, 1 × 10 4 cells/well in 120 μl of culture medium were seeded in a 96-well plate for 24 h. Then, 20 µl of medium was collected into a new 96 well plate for colorimetric detection at 490 nm with a microplate reader. All experiments were performed in triplicate.

ATP production
ATP levels was measured using an ATP Colorimetric/ Fluorometric Assay kit (BioVision, USA) according to the manufacturer's instructions. Briefly, lysates of 1 × 10 6 cells were collected to detect the ATP concentrations using a microplate reader at 490 nm.

Statistical analysis
Data were analyzed using SPSS version 23.0 (SPSS, Chicago, IL, USA). Continuous variables are presented as the means ± standard deviations, and the Student's t-test was used to assess the differences between two groups. Correlation analysis was utilized for the expression levels of lncRNAs and mRNAs. P values < 0.05 were considered statistically significant.

LncRNA and mRNA expression profiles in IH
To explore the expression level profiles of lncRNAs and mRNAs in IH, lncRNA and mRNA microarray analyses were performed with six proliferating IH samples and four involuting IH samples (Fig. 1). A volcano plot was generated to provide an overview of the DE lncRNAs and mRNAs in our microarray data (Fig. 2a, b). In total, 146 DE lncRNAs-97 upregulated lncRNAs and 49 downregulated lncRNAs-were identified (Additional file 3: Table S3). Of the 374 DE mRNAs, 115 were upregulated and 259 were downregulated (Additional file 4: Table S4). Hierarchical clustering analyses were used to display the lncRNA and mRNA expression patterns (Fig. 2c,  d). These data implied that lncRNA and mRNA expression levels differed between proliferating and involuting IHs. The top 10 DE lncRNAs and mRNAs are shown in Tables 1 and 2, respectively.

GO and KEGG pathway analysis of DE mRNAs
Analysis of the DE mRNAs via GO and KEGG pathway analysis could provide a clue to the role of DE mRNAs in the IH disease process. All DE mRNAs were used in GO analysis, and the top 10 GO terms, including those in the BP, CC and MF categories, are shown in Fig. 3a-c. A total of 139 BPs were identified (Additional file 5: Table S5). The most significantly enriched GO term was "angiogenesis", followed by "small molecule metabolic process", "cell adhesion" and "positive regulation of cell migration". Moreover, in addition to classical BPs associated with cell proliferation and migration in IH, interesting metabolism-related BPs were also found, including "glycogen biosynthetic process", "branched-chain amino acid catabolic process" and "mitochondrial electron transport, NADH to ubiquinone". In total, 19 (13.7%) metabolic BPs were identified and are shown in Table 3.
KEGG pathway analysis identified 64 pathways (Additional file 6: Table S6). The DE mRNAs were mostly enriched in the "focal adhesion", "metabolic pathways" and "PI3K-Akt signaling pathway" KEGG pathways. The top 10 significant pathways are shown in Fig. 3d. In addition, 11 (17.2%) metabolism-related pathways were found and are shown in Table 4. Then, we conducted pathway network analysis using all the significant pathways to illustrate the critical pathways in the process of hemangioma neovascularization (Fig. 3e). The pathway network comprised 34 network nodes and 74 edges. In the network, the "MAPK pathway", "pathway in cancer" and "Citrate cycle (TCA cycle)" KEGG pathways were considered the most relevant pathways because they had the highest degree values, indicating that these pathways may play the most important role in IH.

PPI network analysis
Genes mutually affect the expression of each other. To investigate the functions of DE genes at the protein level and determine the core genes in IH, protein-protein network analysis was performed using STRING. The resulting PPI network contained 374 nodes and 935 edges (Additional file 7: Fig. S1). After degree calculation, 31 hub genes were identified with a degree ≥ 15 (Table 5).

Construction of the lncRNA-mRNA co-expression network
To identify hub regulatory factors associated with IH, lncRNA-mRNA co-expression network analysis was performed with 146 DE lncRNAs and 374 DE mRNAs. In total, 42 lncRNAs and 217 mRNAs were included in the co-expression network, which comprised 259 network nodes and 915 connections (Fig. 5). Among these included mRNAs and lncRNAs, 28 mRNAs and 15 lncR-NAs were upregulated and 189 mRNAs and 27 lncRNAs were downregulated. In the co-expression network, many mRNAs were correlated with a single lncRNA and vice versa. We then calculated the core degree to estimate the relative significance of an individual lncRNA or mRNA in the co-expression network. The top twelve DE lncRNAs with a relatively high core degree (≥ 5) may be critical to the regulation of the pathogenetic mechanism underlying IH (Fig. 6a). However, to date, the functions of most lncRNAs have not been well-annotated. Therefore, we can predict the potential role of lncRNAs by analyzing the function of DE mRNAs via GO and KEGG pathway analyses [24]. GO and KEGG pathway analysis of the Fig. 1 Hematoxylin-eosin-stained infantile hemangioma histopathological specimens (original magnification: × 100). The proliferating phase is characterized by densely packed tumor cells that form immature vessels. In the involuting phase, the disorganized vasculature consists of a flat endothelium and pericytes co-expressed mRNAs via Metascape indicated that the top 12 lncRNAs were mostly involved in cell metabolismassociated BPs and pathways (Fig. 6b).

Prediction of cis-regulated targets of lncRNAs
To explore the functions of lncRNAs in IH, we predicted the cis-regulated genes of the DE lncRNAs using co-expression network analysis. Altogether, 15 lncRNAs were predicted to have cis-regulated target genes. These lncRNAs and their target genes are shown in Table 6.

Validation of differential lncRNA and mRNA expression
To validate our microarray results, we randomly selected 5 lncRNAs (3 upregulated and 2 downregulated) and 12 mRNAs (10 upregulated and 2 downregulated) and measured their expression levels via qRT-PCR. The qRT-PCR results showed that the expression of lncRNAs n334063, n334214 and ENST00000417970 was upregulated, whereas that of lncRNAs n333411 and n335635 was downregulated. In addition, two mRNAs, AIMP1 and CUL5, were downregulated, and the remaining mRNAs, including NOTCH3, PFK-1, LOXL2, RHOB, KDR, COL4A2, COL18A1, ACVRL1, THY1 and MCAM, were upregulated. The qRT-PCR results were thus consistent with the microarray analysis results (Fig. 7a). Moreover, we further selected two of the most significantly DE mRNAs, in which we were the most interested, and assessed their protein expression levels by IHC analysis and western blot (Fig. 7b, c). The IHC staining and western blot results showed that both LOXL2 and FPK-1 exhibited higher protein expression in proliferating IH than in involuting IH. Moreover, both LOXL2 and FPK-1 were located mainly in the cytoplasm (Fig. 7b).

Suppression of PFK-1 inhibits HemEC growth and migration and induces cell cycle arrest
To investigate the biological roles of PFKFB3 in the progression of HemECs, we constructed lentiviral vectors expressing shRNAs targeting PFK-1, and then infected the HemECs with these two shRNAs lentivirus (shPFK-1). The mRNA and protein expression levels of PFK-1 were significantly inhibited by transfection with shRNAs ( Fig. 8a, b). Then, a CCK-8 assay was performed to assess the effects of PFK-1 on HemEC proliferation. Compared with the negative control group, cell proliferation was significantly decreased in HemECs transfected with shPFK-1 (Fig. 8c). Furthermore, a cell cycle assay was performed to confirm the effect of PFK-1 on cell proliferation by flow cytometry. As shown in Fig. 8d-g, cell cycle was arrested in the G1 phase in HemECs transfected with the shPFK-1 lentivirus. The effect of PFK-1 knockdown on cell migration was evaluated by a Transwell assay in vitro. HemEC migration was significantly suppressed   (Fig. 8h-k). The above results suggest that inhibition of PFK-1 suppresses HemEC proliferation and migration and induces cell cycle arrest.

Suppression of PFK-1 reduces glucose uptake, lactate secretion and ATP production
To investigate whether PFK-1 knockdown affects glycolysis in HemECs, glucose uptake and lactate production assays were performed. As shown in Fig. 9a, b, both glucose uptake and lactate secretion were markedly decreased in PFK-1-shRNA-treated HemECs. ATP production was also significantly lower in PFK-1-shRNAinfected HemECs when compared with the control group. These results revealed that inhibition of PFK-1 could affect glycolysis through by decreasing glycolytic flux.

Discussion
Aberrant expression of lncRNAs is involved in the pathogenesis and progression of many diseases by regulating gene expression profiles. However, the expression patterns and potential functions of lncRNAs in IH development and pathogenesis are not completely clear. Several previous studies have identified lncRNA and mRNA expression in IH and normal tissues [13,25]. Liu and his colleague found 2116 DE lncRNAs and 2653 DE mRNAs in IH compared with adjacent normal tissues [13]. In these studies, several lncRNAs, including MEG3, Linc0152 and MALAT1, were identified to play a critical role in the regulation of tumor angiogenesis in IH [13] and demonstrated to be associated with the development of IH [14,[26][27][28]. However, these lncRNAs were not identified in the study by Li et al., which identified 144 DE mRNAs and 256 DE lncRNAs in IH compared with matched normal tissues [25]. To further investigate the role of lncRNAs and mRNAs in proliferating and involuting IH specimens, we analyzed the expression profile of lncRNAs and mRNAs using microarray technology. The microarray data identified 146 lncRNAs and 374 mRNAs as significantly DE between the two groups, and these findings were then validated by qRT-PCR. Although the DE lncRNAs identified in this study were not found in the previous study, many DE mRNAs including COL18A1,  PDGFRB, COL4A2, THY1, ACVRL1, ANGPL1 and IFI6 were identified in earlier studies [13,25,29]. Collectively, these results indicated that these DE lncRNAs and mRNAs may play considerable roles in the development of IH.
To assess the biological functions enriched by the DE mRNAs, we further performed GO and KEGG pathway analyses. The GO analysis results showed that angiogenesis-related BPs, including "angiogenesis", "cell adhesion" and "positive regulation of cell migration", were markedly involved in the pathogenesis of IH. In addition, KEGG pathway analysis indicated that the DE mRNAs were mainly involved in angiogenesis-related pathways, including "focal adhesion", "regulation of actin cytoskeleton" and "PI3K-Akt signaling pathway". These findings further confirmed those of previous studies [7,29]. Combined with the GO and KEGG pathway analysis results, our data confirmed that angiogenesis plays a critical role in the pathogenesis of IH.
The cellular components of IH include ECs, pericytes and other cells (i.e., mast cells and stem cells) [7,8]. During the proliferating phase of IH, ECs are the predominant cells. During angiogenesis, ECs can rapidly switch from a quiescent state to an active, proliferative, migratory state in response to growth factor stimulation, primarily through VEGF [30]. The changed EC state requires vast amounts of energy to meet the bioenergetic and biomass demands of cell proliferation and migration [31]. Hence, ECs must increase their metabolic activity to quickly increase their energy generation. ECs are involved in several metabolic pathways, including the glycolysis, hexosamine biosynthesis, polyol, oxidative metabolism, amino acid metabolism, and fatty acid metabolism pathways  [30,32]. These metabolic pathways play distinct and essential roles during vessel formation. Recent studies have highlighted the importance of EC metabolism as a driving force of angiogenesis [31]. Therefore, targeting EC metabolism could offer new therapeutic opportunities to combat angiogenesis [33]. Interestingly, in the present study, some metabolism-related BPs (e.g., "glycogen biosynthetic process", "metabolic process", "cellular lipid metabolic process", and "respiratory electron transport chain") and pathways (e.g., "oxidative phosphorylation", "pyruvate metabolism", "citrate cycle", and "fatty acid degradation") were also found in IH, suggesting that cell metabolism participates and plays a vital role in the development of IH.
In tumor ECs, glycolysis is the predominant method of energy generation even in the presence of O 2 , although glycolysis is much more inefficient than oxidative phosphorylation (OXPHOS) [31]. During angiogenesis, glycolysis is further accelerated [30]. This phenomenon has several explanations, as follows: 1) it leads to environmental acidosis, which is toxic to normal cells but harmless to cancer cells; 2) it provides more carbon skeletons for biosynthesis than OXPHOS; 3) it produces ATP much faster than OXPHOS; and 4) it allows adaptation to hypoxia [34,35]. In 2013, Bock et al. demonstrated that both human umbilical vein ECs (HUVECs) and hemangioma ECs (Hem-ECs) were highly glycolytic [36]. Moreover, this group found that glycolysis was even higher in Hem-ECs than in other types of ECs [36]. PFK-1, one of the most important rate-limiting enzymes in glycolysis, converts fructose-6-phosphate (F6P) to fructose-1,6-bisphosphate (F1,6P2). Fructose-2,6-bisphosphate (F2,6P2), synthesized by phosphofructokinase-2/fructose-2,6-bisphosphatase (PFKFB) enzymes, is an allosteric activator LncRNA-mRNA co-expression network. The circles and diamonds represent mRNAs and lncRNAs, respectively. Red and green indicate upregulation and downregulation, respectively. The node size represents the degree Fig. 6 Co-expression network of twelve lncRNAs with their potential target mRNAs (Fig. 5a). The circles and diamonds represent mRNAs and lncRNAs, respectively. Red and green indicate upregulation and downregulation, respectively. The node size represents the degree. GO and KEGG terms enriched in the top 12 lncRNAs in the co-expression network (Fig. 5b). Black and gray indicate BPs and pathways, respectively of PFK-1 and the most potent stimulator of glycolysis [37]. Among PFKFB isoenzymes, PFKFB3 is not only the most abundant isoenzyme in ECs [36], but also has much more efficient kinase activity than bisphosphatase activity [38]. A previous study reported that PFKFB3 can regulate EC proliferation and stimulate vessel sprouting, implying that glycolysis can regulate angiogenesis [36]. Indeed, inhibition of PFKFB3 reduced pathological angiogenesis [39], induced tumor vessel normalization, downregulated glycolytic activity in pericytes, tightened the EC barrier, impaired metastasis and improved chemotherapy [40]. In addition, PFK-1 can promote tumorigenesis by activating the AKT pathway [41]. However, the cyclin D3-cyclin-dependent kinase 6 (CDK6) complex can inhibit the catalytic activity of PFK-1 in the glycolytic pathway, suggesting a direct link between the cell cycle and cell metabolism [42]. The PI3K-Akt pathway and cyclins, with their associated kinases, CDKs, were demonstrated to participate in the development of IH in our previous study [7,16]. Recently, glycolysis-associated molecules, including GLUT-1, HK2, PFKFB3, PKM2 and LDHA, had higher expression in HemECs compared with HUVECs [43]. In the present study, although no lncRNA was found to regulate its expression, PFK-1 was one of the most significantly DE mRNAs and was strongly expressed in proliferative IH tissues, as shown by western blot and IHC staining. Besides, we further investigated the roles of PFK-1 in cell proliferation and migration and the cell cycle. Our results revealed that silencing PFK-1 with shRNA significantly inhibited HemEC proliferation and migration and induced cell cycle arrest. In addition, we founded that PFK-1 knockdown caused a marked decrease in glucose uptake and lactate secretion in HemECs, which could affect ATP production from glycolysis by inhibiting glycolysis flux. Therefore, based on these studies, we inferred that the glycolytic activator PFK-1 also plays an important role in angiogenesis in IH. Further studies on the role of PFK-1 in the pathogenesis of IH are warranted. LOXL2, a member of the lysyl oxidase family, was reported to have an important role in promoting angiogenesis [44]. However, inhibition of LOXL2 enhanced antiangiogenic effects in angiogenic tumors [45]. Several mechanisms explain the role of LOXL2 in angiogenesis, including activating focal adhesion kinase, increasing the expression of VEGF, promoting epithelial-mesenchymal transition, and inducing hypoxia-inducible factor-1α [46]. These studies indicate that LOXL2 has a promising role in vascularized tumors. Therefore, more studies are needed to further investigate the roles of LOXL2 in IH.

Conclusion
This study provided a comprehensive bioinformatics analysis of lncRNA and mRNA expression profiles in proliferating and involuting IH. The identified DE Fig. 7 qRT-PCR validation of 5 randomly selected DE lncRNAs and 13 DE mRNAs in the microarray analysis (Fig. 6a). The heights of the columns represent the mean values of log-transformed fold changes in expression. IHC staining (b) and western blot (c) analysis of two of the most significantly DE genes-LOXL2 and PFK-1. Both LOXL2 and PFK-1 were more highly expressed in proliferating IH than in involuting IH (original magnification: × 400) lncRNAs and mRNAs may be associated with the pathogenesis of IH. Moreover, inhibition of the glycolytic activator PFK-1 could suppress HemEC proliferation and migration, induce cell arrest, and reduce glucose uptake and lactate and ATP production. The data presented here could improve our understanding of IH development and provide new directions for further studies investigating the mechanism underlying IH. Additional studies are Fig. 8 Suppression of PFK-1 inhibits HemEC growth and migration and induces cell cycle arrest. a Quantitative PCR and b western blot analysis was used to confirm the suppression rate of PFK-1 with shRNAs in HemECs. c CCK-8 assay was used to determine the effects of shPFK-1 on HemEC proliferation. (d-f) Inhibition of PFKFB3 with shRNA induces HemEC cell cycle arrest in the G1 phase. (g-i) Transwell assays were performed to measure the effects of PFK-1 on HemEC migration (Magnification × 100). *P < 0.05, **P < 0.01, ***P < 0.001 Fig. 9 Suppression of PFK-1 reduces glucose uptake, lactate secretion and ATP production. a Glucose uptake, b lactate production and c ATP production were examined to investigate the effects of shPFK-1 on glycolysis flux in HemECs. The same experiments were performed in triplicate. Data are the mean ± SE. **P < 0.01, ***P < 0.001