Plasma-based Endogenous Metabolomics Proling of High-Risk Human Papillomavirus and Their Emerging Roles in the Progression of Cervical Cancer

Objective: High-risk human papillomavirus (HR-HPV) is the main etiological factor for cervical cancer. Accumulating evidence has suggested that the active role of metabolites in the initiation and progression of cancers. This study was to explore the metabolic proles of HR-HPV infection and their potential functions in cervical cancer. Methods: Non-targeted metabolomics approach was used to detect metabolic alterations in the plasma obtained from HPV-16 positive (HPV16 (+)), HPV-18 positive (HPV18 (+)) and HPV negative (CTL) individuals, followed by CCK8 experiment to detect the effect of different metabolites on the proliferation of Hela and GH354. A cell migration test then veried signicant metabolites on the migration of Hela and GH354. Q RT-qPCR and western blot were used to detect malignant progression related mRNA and protein expression levels of cervical cancer. Results: HR-HPV groups shared 24 dysregulated metabolites (such as amino acids, ceramides, glycerophosphocholines). Further experiments showed ceramide species, including C8 inhibits cervical cancer cells proliferation and migration in vitro. In contrast, C12 signicantly enhanced cervical cancer cells proliferation and migration in vitro. Protein and mRNA expressions indicated C8 and C12 were related to the malignant behavior of cervical cancer in vitro. The underlying mechanism demonstrated that C8 intervention inhibited proliferation and migration in cervical cancer cells via the MAPK/JNK signaling pathway, while C12 intervention promoted proliferation and migration in cervical cancer cells via the MAPK/ERK signaling pathway. These ndings may contribute to the treatment of HR-HPV-induced cervical cancer by intervening in its initiation and progression. Conclusion: Our study shed some light on how metabolites inuenced the relationship between HR-HPV oncogenic capability and metabolic phenotype change and identify species C8 and C12 as critical lipid metabolites that modulate cervical cancer cell’s function.


Plasma-based Endogenous Metabolomics Pro ling of High-Risk Human Papillomavirus and Their
active metabolites accompanied by the concept of activity metabolomics [9]. The emerging studies have highlighted the functional role of metabolites in physiology and disease. Also, some examples support mTOR kinase6 act as active entities in cell nutrients and energy [10]. An additional study unraveled αketoglutarate active macrophage and regulated immunity [11]. Abnormal accumulations of fumarate and succinate, termed oncometabolite, causes potential transformation to malignancy [12]. Metabolites such as lipids, amino acids and bile acids play a role in regulating insulin sensitivity [13]. Lysophosphatidic acid can mediate the cell migration and metastasis of ovarian cancer cells by activating the AMPK pathway [14]. HPV infection drive metabolic modi cations and their metabolites as potential markers in predicting the risk of cervical cancer [15]. However, the role of bioactive metabolites and underlying mechanisms remain largely unknown. Metabolomics enables us to identify metabolites with the potential to modulate biological processes associated with cervical cancer caused by HR-HPV infection.
This article aims to explore the difference in the metabolite expression pro les between HR-HPV and negative groups and reveals how metabolites intervene in HR-HPV oncogenic capability. Non-targeted metabolomics was employed to show metabolite expression pro les between HRHPV and negative groups. Interestingly whereas changes observed in cervical cancer cells function due to ceramide species sampling was processed by trained gynecologists and followed the approved protocol. All the blood samples were collected in the anticoagulant tubes and centrifuged at 3000 rpm for 15 min at 4 ℃. The supernatant was then transferred into 1.5 ml tubes and stored at -80 ℃ before use.

Sample preparation
After thawing, about 100 µl of the supernatant sample was added into 10 µl of 2-chloro-L-phenylalanine (0.3 mg/mL) and Lyso PC17:0, 0.01 mg/mL diluted in methanol as internal standard, and then vortexed for 10s. Pre-cold methanol and acetonitrile (2/1, v/v) were mixed and added into the sample, and then vortexed for 1 min, samples were ultrasonically extracted in an ice water bath for 10 min, and held at -20℃ for 30 min. After centrifuging at 13000 rpm at 4°C for 10 min, 300 µL of the supernatant was transferred into the LC-MS vial and evaporated to dryness. Next, 400 µL of methanol and water (1/4, v/v) were applied to each sample, vortexed for 30 s, held at 4℃ for 2 min, and then stored at -20℃. After 30 min, samples were centrifuged at 13000 rpm at 4°C for 10 min. The supernatants were aspirated using syringes, ltered through 0.22 µm micro lters, transferred to LC vials, and nally stored at -80°C.

Detection of metabolic pro ling by LC-MS
Metabolic pro les in both electrospray ionization (ESI) positive and ESI negative ion modes were carried out using an ACQUITY UPLC I-Class system (Waters Corporation, Milford, USA) coupled with an AB SCIEX Triple TOF 5600 System (AB SCIEX, Framingham, MA). The binary gradient elution systems consisted of water containing 0.1% formic acid, v/v (A) and acetonitrile containing 0.1 % formic acid, v/v, (B). The separation was achieved using the following conditions: 20% B for 2 min; 60% B for 4 min; 100% B for 11 min; 100% B for 13 min; 13.5 5% B for 13.5 min and a nal 5% B for 14.5 min. All the samples were analyzed at 4°C. The injection volume was 2 µl. In full scan mode (m/z ranges from 70 to 1000) combined with information-dependent acquisition (IDA) mode, data acquisition was performed at a collision energy of 10 eV (+) and − 10 eV (-). Parameters of mass spectrometry were as follows: ion source temperature ranged from 550°C (+) to 550°C (-) throughout the acquisition; ion spray voltage was set from 5500 V (+) to 4500 V (-); curtain gas was 35 PSI; decluttering potential was set between 100 V (+) and − 100 V (-); interface heater temperature was from 550°C (+) to 600°C (-). For IDA analysis, the range of m/z was set as 25-1000, and the collision energy was 30 eV.

Data analysis
Metabolites were identi ed and analyzed by public databases (http://www.hmdb.ca/; http://www.lipidmaps.org/) and self-built databases. The distinct tendency among groups was analyzed using principal component analysis (PCA) and orthogonal partial least-squares discriminant analysis (OPLS-DA) models. The variable importance in projection (VIP) > 1 and p values < 0.05 were considered statistically signi cant. Qualitative and metabolic pathway analyses of differential metabolites were investigated with the Kyoto encyclopedia of genes and genomes (KEGG) online database. Furthermore, the human metabolome database (HMDB) IDs and KEGG IDs of the metabolites were entered into the ingenuity pathway analysis (IPA) server (IPA, Ingenuity® Systems, http://www.ingenuity.com) to analyze networks between metabolites. and were prepared in dimethyl sulfoxide (DMSO) and aliquots of this solution were added to cultures, keeping DMSO concentration below 0.1%, same DMSO was added to controls.

CCK8-assay
The effect of cervical cancer cells proliferation was assayed by CCK8 (TargetMOL, China). Cultured Hela and GH354 cells were suspended in 100 µl culture medium with 10% FBS and inoculated in a 96-well plate (3000 cells/well) along with 0, 5, 10, 20, 30 µM C8, C12 and 10 µl CCK8 solution was added to each well after incubating 24hrs, 48hrs and 72hrs respectively. The plate was incubated for an additional 1 hour before measuring the absorbance at 450 nm wavelength using a microplate reader.

Cell migration assay
Cultured cells were suspended in a 2ml culture medium with 16% FBS and inoculated in a 6-well plate (100000 cells/well) along with C8 and C12. Control is added with the same DMSO. Migration e ciency was detected at a time of 0hrs, 24hrs, 36hrs and 72hrs, respectively. . The RT-qPCR procedures were as follows: 95°C for 5 seconds, followed by 40 cycles at 95°C for 10s and 60°C for 30 s. Quanti ed mRNA was normalized to beta-actin as a control. The relative expression of mRNA was determined by the 2-ΔΔCT method.

Protein extraction and western blot analysis
Cells were prepared by homogenization in the presence of protease inhibitors (Thermo Fisher Scienti c, Waltham, MA, USA), and centrifuged to remove cell pellet. The samples were heat-denatured at 95°C for 5 minutes with 5× SDS-PAGE loading buffer and fractionated on 10% SDS-PAGE gels (Bio-Rad, Hercules, CA, USA). GAPDH (HRP-60004, Proteintech, Manchester, UK) was used as a standard control protein.

Statistical analysis
Univariate analysis of variance (ANOVA) quanti ed the differences between the HR-HPV infected and the CTL groups with GraphPad Prism 6.0. Data were presented as the mean ± standard error (SE). P-values were determined using a two-tailed Student's t-test or one-way analysis of variance (ANOVA) with a Tukey's post hoc correction for multiple group comparisons. All data analyses were processed using GraphPad Prism, version 5.0 (GraphPad Software, San Diego, CA). A two-sided P-value of < 0.05 was considered statistically signi cant. Signi cance was set as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

HR-HPV infection accompanied by distinct metabolome pro les
To examine the differences in CTL and HR-HPV infection group metabolites, we conducted a multivariate assessment and OPLS-DA analysis. After the PCA model was established, they plot a separation tendency between the CTL group and the HPV16 (+) (Fig. 1A) or HPV18 (+) group (Fig. 1B). OPLS-DA models were obtained with principal predictive components and principal orthogonal components. OPLS-DA removed unassociated data from the dataset and then veri ed the metabolic pro les' dissolution between groups (Fig. 1C-D). We next performed a permutation examination of the OPLS-DA model. The R2 and Q2 intercept values were (0.0,0.893) and (0.0, -0.285) respectively between CTL and HPV16 (+) infection group (Fig. 1E), and (0.0,0.862), (0.0, -0.321) respectively between CTL and HPV18 (+) infection group (Fig. 1F).

Identi cation of differential metabolites in HR-HPV infection and CTL plasma
A total of 88 differential metabolites in the HPV16 (+) group (VIP > 1, p < 0.05) were identi ed, among which 31 metabolites were signi cantly up-regulated and 57 metabolites were signi cantly downregulated ( Fig. 2A). In addition, a total of differentially expressed 101 metabolites were identi ed in the HPV18 (+) group, among which 26 metabolites were signi cantly up-regulated and 75 metabolites were signi cantly down-regulated (Fig. 2B). Further pathway analysis (p < 0.05) revealed that most enriched metabolic pathways were almost in the high-risk groups (Fig. 2C-D), indicating that HPV16 and HPV18 may have similar metabolic functions in the initiation and progression of cervical cancer. To evaluate similarities and dissimilarities in the datasets, we compared the number of shared and unique metabolites among the groups. Venn diagram showed 24 identi ed metabolites (Fig. 2E). Details of the metabolites with the highest difference were summarized in Table 1. Compared to the CTL group, the HR-HPV groups shared 12 up-regulated metabolites and 12 down-regulated metabolites respectively. Metabolites were roughly categorized into twelve distinct classes (Fig. 2F). Amino acids and peptides (Fig. 3A) and monosaccharides (Fig. 3B) were reduced in HRHPV groups compared to the CTL group, while glycerophosphocholines were increased and D-Mannitol was decreased in HR-HPV groups (Fig. 3C).
Steroid conjugates and benzene derivatives were increased, while ceramides were reduced in HR-HPV groups (Fig. 3D-F). Additionally, two acids with six and ve carbon atoms in their fatty acid moieties were detected.

C8 and C12 regulate the proliferation and migration of CC cell lines and malignant behavior related molecules at mRNA and protein levels
We conducted the assays to examine the proliferation and migration potency of the ceramides compounds on cervical cancer cell lines Hela and GH354, it turned out the proliferation rate of C8 intervention on HeLa and GH354 cells was declined while C12 intervention on HeLa and GH354 cells was restored based on CCK8 assay results ( Fig. 4A-D). Comparing the migration index of C8 and C12 intervention cell lines on 0 hour and 48 hours revealed that migration index was signi cantly diminished in the C8 group but increased for C12 intervention (Fig. 4E-F). To verify the observed effects of ceramide species intervention in vivo, we analyzed the expression levels of several molecules closely related to malignant behavior by western blotting and Q PCR experiments. Q PCR analysis revealed the changes in E-cadherin, N-cadherin, Vimentin and MMP9 expression on C8 (concentration is 30µM) and C12 (concentration is 10µM) intervention on HeLa and GH354 cells. Speci cally, E-cadherin was up-regulated after C8 intervention on Hela and GH354. Conversely, N-cadherin, Vimentin were down-regulated after C8 intervention on Hela (Fig. 4G, I). N-cadherin and MMP9 were up-regulated after C12 intervention in Hela and GH354 (Fig. 4H, J). Western blot veri ed that signi cant upregulated E-cadherin and downregulated N-cadherin after C8 intervention on Hela and GH354 at the protein level, and opposite trend after C12 intervention ( Fig. 5B-C).

Interaction network and downstream effects of differential metabolites
Dysregulated metabolites were imported into IPA software for further biological pathway prediction to reveal potential targets and mechanisms, and its results indicated a close association with PI3K/AKT signaling, mTOR signaling and PTEN signaling, and speci c lipids were associated with ERK/MAPK signaling, TGF-β signaling and PLA2G2A regulation (Fig. 5A). Thus, whether C8 and C12 intervention modi ed the expression of molecules in the signaling pathway was then explored with Western blot analysis, and it turned out that the expression levels of JNK remained the same regardless of C8 intervention, but the expression level of Bax was up-regulated whereas the expression level of P-JNK and Bcl2 were down-regulated after C8 intervention, which thereby inhibited the activity of MAPK (Fig. 5B). The expression levels of ERK remained the same regardless of C12 intervention, and the expression level of Bax was down-regulated, whereas the expression levels of P-ERK and Blc2 were up-regulated after C12 intervention (Fig. 5C).

Discussion
HR-HPV is one of the potent human carcinogens. Persistent HRHPV infection is a necessary risk factor for the cervix and cervical cancer [16]. It induces epithelial cells' malignant transformation and suppresses the immune response via encoding oncoproteins [17]. For instance, E6 and E7 proteins encoded by HRHPV could promote cervical cancer [18,19], E6 protein encourages the growth of cervical cancer cells by targeting P53 protein [20], and E7 protein immortalizes human epithelial cells by targeting pRb protein [21]. Rapid proliferation is a driving force for the massive energy required by malignant cells to adapt to metabolic modi cations [22]. Driven by oncoproteins, metabolites delineate a comprehensive characterization of HR-HPV's molecular mechanisms [22]. Besides, the abnormal expression or activation of metabolic pathway-related enzymes is tightly associated with the occurrence of cancers [23]. Previous studies revealed that E6/E7 could also regulate the glycolytic pathway via elevated expression of hexokinase-II, acting as a promotor in HPV-associated cervical lesions in serum [24]. However, the metabolic pro le in response to HPV16/18 infection in plasma has not yet been elucidated. Thus, metabolomics analysis may be helpful to explore HRHPV infection and its carcinogenic effects.
In our study, metabolomics analysis was applied to HPV16 (+), HPV18 (+) and CTL groups. The two HR-HPV groups shared 12 up-regulated metabolites and 12 down-regulated metabolites compared to the control. These metabolites were mainly divided into 6 categories, and they are amino acids and peptides, ceramides, fatty acids, glycerophosphocholines, monosaccharides and steroid conjugates. Notably, LysoPC (P-16:0) and PC (O-10:1(9E)/0:0) were increased in both HPV16(+) and HPV18(+) groups in our study indicating that lysoPC and PC might play a role in HR-HPV-induced oncogenesis. Phosphatidylcholine (PC) and lysophosphatidylcholine (LysoPC) were previously considered as potential biomarkers for cervical cancer [25]. LysoPC originates from the cleavage of PC, the main component of oxidatively damaged low-density lipoprotein, which was reported can aggregate in ammation and plays an important role in the invasion, metastasis and prognosis of tumors [16]. In addition, an increasing body of evience has indicated that enzymes that participate in key lipid metabolism are potential therapeutic targets, as they either inhibit their synthesis or stimulate their degradation [9]. Lactate dehydrogenase A is a key enzyme for lactic acid synthesis, and it also could promote apoptosis and leads to a decrease in the cell cycle when it is inhibited [26]. Phosphoglycerate dehydrogenase, the key enzyme of serine biosynthesis, its inhibition has potential therapeutic value in lung adenocarcinoma [12]. Interestingly, we observed an opposite trend of Ceramide 1-Phosphate (CerP) (d18:1/18:0), it was signi cantly decreased in both HPV16(+) and HPV18(+) groups, Ceramide is a component of eukaryotic cell membranes that acts as a bioactive lipid in apoptosis, in ammation, cell cycle arrest and the heat shock response [13]. Recent papers found that Ceramide synthase 2-C -ceramide axis limits the metastatic potential of ovarian cancer cells [27], and ceramide in breast cancer patients was correlated with prognostic [28]. Studies have reported that CerP lipidoids could regulate cell proliferation, apoptosis and migration [29], implying that CerP might play a role in HRHPV-induced oncogenesis, however, CerP lipidoids function has not been reported yet in CC, and our results suggested that C8 and C12 have an important in uence on cervical cancer cell proliferation and migration in vitro. The mRNA and protein expression levels of the malignant indicators for cervical cancer helped to verify the function of ceramide in vitro.
Metabolites serve as controllers of biological processes and phenotype [9], and tumorigenesis may change the overall metabolism of the human body. Therefore, the changes in these biological processes are re ected in the metabolomics, including amino acids and lipid metabolites [30]. Bioactive metabolome drives phenotype modulation by participating in life activities' regulation process and exerting their biological activities through various pathways such as competitive inhibition, post-translational modi cations, and signal transduction [31]. Accumulation of oncometabolite is a causal process in malignant conversion that contribute to propagating cancer.
Our study revealed that the dysregulated metabolites were closely associated with PI3K/AKT signaling, mTOR signaling and speci c lipids were associated with ERK/MAPK signaling, TGF-β signaling and PLA2G2A regulation. The PI3K/AKT/mTOR network plays a key role in HRHPV oncogenes in the host cell [32], and the abnormal activation of PI3K/mTORC2/ AKT signaling pathway could increase migration of keratinocytes and eventually lead to carcinogenesis [33,34]. It is suggested that ERK-mediated phosphorylation is related to sphingosine metabolism and subsequent in ammation [35], which is an important target for cervical cancer [36]. Besides, the AMPK signaling pathway also plays a vital role in cervical cancer and is closely related to the malignant progression of cervical cancer [37]. The IPA analysis indicated that HRHPV-induced dysregulated metabolites were closely associated with cervical cancer risk. To directly examine the possible mechanism underlying the observed phenotypic changes, we analyzed the expression levels of molecules in the signaling pathway and found that C8 intervention inhibited proliferation and migration in cervical cancer cells via the MAPK/JNK signaling pathway, while C12 intervention promoted proliferation and migration in cervical cancer cells via the MAPK/ERK signaling pathway. Our study provides preliminary evidence that metabolites link the relation between tumorigenesis and metabolic phenotype changes.
In summary, HR-HPV infection-induced distinct metabolites pro le changes in humans. The altered metabolites may contribute to the process of HPV16/18 infection and cervical cancer. This paper is the rst to identify C8 and C12 as critical lipid metabolites that modulate cervical cancer cell's function. Our study provides novel insights into the mechanism of the oncogenic process of HPV16/18 infection.

Declarations
Con ict of Interest The authors declare that they have no competing interests.
Author Contribution QW and MX have drafted the manuscript and revised it critically for important intellectual content, and they both conducted the experiments in this study.
TC and JC have made substantial contribution to acquisition and analysis of data.
RZ and JQ have designed and funded this study. Also, they have given nal approval of this study.  Analysis of HRHPV infection plasma reveals HPV-related differences of metabolites. A. Heatmap of differential metabolites with intensities determined by LC-MS. Hierarchical clustering analysis was used to assess signi cantly up-regulated and downregulated metabolites between HPV16 (+) and CTL plasma.
B. Hierarchical clustering analysis was used to assess signi cantly up regulated and downregulated metabolites between HPV18 (+) and CTL plasma. Increased and decreased metabolite levels are depicted by red and blue colors, respectively. VIP, variable importance in projection.C. Pathway clustering analysis of HPV16 (+) samples, signi cantly enriched pathway clusters were identi ed based on p-values <0.05 and depicted in logarithmic scale (log2). D. Pathway clustering for metabolites enriched in HPV18 (+) samples by metabolomics analysis. E. Number of shared metabolites among the high-risk positive groups visualized on a Venn diagram. F. Number of shared metabolites that belong to glycerophosphocholines, amino acids and peptides, monosaccharides classes, etc.

Figure 3
Differences in the intensity of signi cantly different metabolites species between CTL and HRHPV infection groups. A. Amino acids and peptides were reduced in HRHPV groups. B. Monosaccharides were signi cantly reduced in HRHPV groups compared to CTL group. C. Glycerophosphocholines were enriched in HRHPV groups and. D-Mannitol was decreased in HRHPV groups. D-F. Steroid conjugates and benzene derivatives were increased, while ceramides were reduced in HRHPV groups  IPA analysis of metabolites related to biological network and canonical pathways and functions. Dysregulated metabolites were closely associated with PI3K/AKT signaling, mTOR signaling and PTEN signaling, and speci c lipids were associated with ERK/MAPK signaling, TGF-β signaling and PLA2G2A regulation. Lines represent the biological relationship between two nodes. Red symbols represent up-