Investigation of the expression level of human endogenous retrovirus E env transcript in cervical cancer

Abstract

Although HPV infection is obligatory for almost all cases of cervical cancer (CC), other risk factors can promote the progression of cervical cancer. In this context, the expression of human endogenous retroviruses (HERVs) in the development of CC has been investigated. In this study, the expression status of HERV-E env transcripts was analyzed in 111 cervical biopsies, including 35 cervical cancer samples, 20 precancerous lesions, and 56 normal samples. Real-time PCR with specific primers was used to quantify the relative expression of HERV-E env, HPV 16 and 18 E6/E7 genes, and GAPDH as a normalization control. Our results indicated an increase in the expression of HERV-E env, and the difference was statistically significant in the cancer group compared to the precancerous group (1.5-fold change) (P = 0.031). In HPV 16 or 18-infected patients, a higher mean value of HERV-E env mRNA was also found in the cancer group than in the precancerous group. ROC curve analysis showed a significant difference in env expression between precancerous and cancerous lesions in all patients analyzed (P = 0.015) and in a group of patients infected with HPV 16 or 18 genotypes (P = 0.023). In addition, there was a positive correlation between the higher expression of HERV-E env mRNA with E7 (R = 0.34, P = 0.016) and age (R = 0.35, p = 0.016) in HPV 16-infected patients. In conclusion, our study found a possible association between HERV-E env expression and cervical cancer, as HERV-E is actively transcribed during the progression of cervical lesions. Future studies on the potential interaction of HERV-E env with HPV 16 E7 oncoprotein are likely to elucidate common signaling pathways in the progression of cervical cancer and other HPV-related malignancies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12977-025-00672-3.

Introduction

Cervical cancer (CC) is the fourth most common cancer in women worldwide, which caused 604,127 new cases (6.5% of all cases) and 341,831 deaths (7.7% of all deaths) worldwide in 2020 [1]. High-risk HPV types are responsible for almost all invasive cervical cancers (ICC), with HPV 16 and 18 responsible for around 70% of ICC [2]. HPV is considered a necessary but not sufficient cause of cervical cancer. It is assumed that other co-factors play an important role in the development of cervical cancer, including immunosuppression, smoking, the use of oral contraceptives, and co-infection with additional infectious agents [3–6]. The expression of retroelements has also been associated with the development of CC [7, 8].

Human endogenous retroviruses (HERVs) infected the ancestral germline millions of years ago, accounting for approximately 8% of the human genome. Overexpression of HERV elements, probably triggered by changes in gene regulation (epigenetic changes) or viral infections, can lead to the production of HERV transcripts and proteins. These elements, particularly the env protein, are associated with disrupting normal cellular processes and sometimes even act as oncogenes [7]. HERV activation has been detected in various normal and malignant tissues due to the presence of their open reading frames from which viral proteins can be produced [9]. HERVs may contribute to different aspects of cancer development through various mechanisms. Their long terminal repeats (LTRs) could act as alternative promoters or enhancers for other genes, and their potential coding proteins could influence cell cycle regulation or immune function [10–12].

While the study of HERV expression in other cancers is a rapidly growing field, there are few studies in CC, although the initial studies are promising. Some reports have identified unique expression patterns of specific HERVs in CC that may be related to various factors such as tumor type and HPV infection status, suggesting their value as biomarkers and potential targets for immunotherapy [7, 13]. For example, a reduced gag expression was found in CC and cervical intraepithelial neoplasia 3 (CIN 3) groups, while its expression increased in CIN 1 and 2 [13]. The expression of env is also altered in CC and CIN 1–3 compared to normal tissues [13]. In addition, a previous study by our group demonstrated that HERV-K elements are overexpressed in pre-cancerous and cancerous groups compared to normal groups. This intriguing link between HERV reactivation and cancer development emphasizes their potential use as therapeutic targets and biomarkers for cancer treatment and diagnosis, respectively [14].

Research also points to a possible link between HERV-E and various types of cancer. The activation of HERV-E elements has been investigated in many tumors, such as breast, testicular, ovary, and colon [15–18], in autoimmune disorders [19], and during placental development [20]. Activation of HERV-E may produce proteins that suppress the immune system and thus contribute to cancer development [21]. HERV-E proteins have the unique ability to both modulate the immune system and potentially contribute to cancer development [9]. In ovarian cancer, antibodies have been developed against specific HERV-E proteins, e.g., against a segment of the p15 env protein (clone 4-1 peptide) [22]. These antibodies can counteract the immunosuppressive effects of HERV-E proteins and even those of other HERVs like HERV-K/HML-2 and HERV-R/ERV3 [23]. Interestingly, studies on kidney cancer have identified a specific HERV-E protein segment (CT-RCC-1) that acts as a tumor-specific antigen [24]. HERV-E might be a valuable target for the development of new cancer therapies. It also has the potential to serve as a biomarker for earlier cancer detection—a crucial step for better patient classification and more informed treatment decisions in clinical practice. This study aimed to investigate the expression status of HERV-E env in cervical cancer tissues.

Materials and methods

Sample collection and patient stratification

A total of 111 uterine cervix biopsies were obtained from female patients over 21 years old at Yas Hospitals and Imam Khomeini Hospital (Tehran, Iran). The samples were divided into three groups: 35 samples with confirmed cervical cancer (average age 45 years), 20 samples with confirmed precancerous lesions (mild to severe dysplasia based on pathological results) (average age 37 years), and 56 normal samples as a control group (no history of cervical abnormalities and histologically confirmed as normal) (average age 36 years). Clinical and demographic data were collected using questionnaires given to all participants. Written informed consent was obtained from all participants before participation in the study. The study protocol was reviewed and approved by the local TUMS ethics committee (IR.TUMS.SPH.REC.1402.054).

RNA and DNA extraction and cDNA synthesis from tissue biopsies

Tissue biopsies were first stabilized in RNA Later (Gene All Co., Seoul, South Korea) overnight at + 4 °C, followed by storage at − 80 °C. Total RNA and DNA was then extracted from the prepared tissue samples using Trizol^® total RNA isolation reagent (Gene All Co., Seoul, South Korea), according to the manufacturer’s protocol.

The extracted RNAs were treated with RNase-Free DNase I (Fermentase) according to the manufacturer’s instructions to eliminate contaminating genomic DNAs. The purity and concentration of RNAs were determined using the Nanodrop Spectrophotometer (Thermo Fisher, Waltham, MA, USA). The RNA concentration of each sample was adjusted to a final concentration similar. The cDNA was synthesized from total RNA using SinaClon First Strand cDNA synthesis kit (Cat. No. RT5201; SINACLON, Tehran, Iran) using a single cycle in a final volume of 20 µL according to the manufacturer’s instructions.

HPV DNA detection and genotyping

The HPV DNA detection and genotyping were described in our previous study [14]. Briefly, a 150 bp fragment of the L1 gene was amplified using nested-PCR with primer pairs of MY09/11 and GP+5/+6, respectively, followed by bidirectional direct sequencing to distinguish HPV genotypes.

The Env and HPV-16 and 18 E6/E7 mRNA expression analysis

Real-time PCR with specific primers (Table 1) was used to relatively quantify the expression of the HERV-E envelope (env) gene and HPV 16/18 E6/E7 genes in tissue samples, and assess the human housekeeping gene GAPDH, which served as a normalization control [25]. Real-time PCR was performed using SYBR^® Premix Ex Taq™ master mix from Takara Bio (Takara, Shija, Japan) on a StepOne Plus Real-time PCR system (Applied Biosystems). Briefly, 10 µl reaction mixtures containing cDNA template, SYBR^® Premix Ex Taq™ (2X), and forward and reverse primers (10 µM each primer) were prepared. The thermal cycling program consisted of an initial denaturation step at 95 °C for 90 s, followed by 40 cycles of denaturation at 95 °C for 10 s, annealing at 56 °C for 30 s, and extension at 60 °C for 20 s. A melting curve analysis was generated to confirm single gene-specific peaks by heating samples from 60 to 99 °C at the end of the amplification cycles.

Table 1.

The sequence of primers used in this study

Primer’s name	The sequence of primer	Amplicon size	References
HERV-E-env-F	GGAGTAATAACAGTATTAGGAACCTGCT
HERV-E-env-R	CTTGTGCTGAACTATTTTGGTGAATT	124 bp	[24]
GAPDH- F	ACCAGGGCTGCTTTTAACTC
GAPDH- R	TGACAAGCTTCCCGTTCTCAG	162 bp	[15]
HPV-16-E6-F	AATGCTTCAGGACCCTTACG
HPV-16-E6-R	GTTGCTTACAGCACACTCAATC	107 bp	This study
HPV-16-E7-F	ACCGGACACAGCCGATTACA
HPV-16-E7-R	GTACGCACTACCGATGCGTA	71 b p	This study
HPV-18-E6-F	CGCTATGAGGATCCATGACG
HPV-18-E6-R	GCAGTGTAGTGTTCAGTACCG	70 bp	This study
HPV-18-E7-F	ACCAGAACGTCACACATTGTAG
HPV-18-E7-R	GGAATGCACGAAGGTCAGCT	95 bp	This study

A standard curve was also generated using serially diluted pooled cDNA samples (a 10-fold dilution series) amplified for GAPDH to evaluate the assay’s performance. A linear relationship between Ct values and cDNA concentrations confirms reliable amplification. Finally, the 2^−ΔΔCt method was used to analyze HERV-E env gene expression. This method normalizes target gene expression to the reference gene (GAPDH) and allows comparison of relative expression among groups. The relative values of env were expressed as the fold-change to compare mRNA levels among different groups [26].

Statistical analysis

The data were analyzed using GraphPad Prism 9.0 (GraphPad Software). Normally distributed data were assessed with t-tests and ANOVA, while non-normally distributed data were analyzed with Mann-Whitney U-tests and Kruskal-Wallis ANOVA with post-hoc tests. We used R software (R- Version 4.3.1) to generate receiver operating characteristic (ROC) curves and to calculate the area under the ROC curves (AUC) to assess the diagnostic potential of the test. Statistical significance was set at *P < 0.05, **P < 0.01, and ***P < 0.001.

Result

Characteristics of examined samples

In the present study, 111 cervical biopsies were examined, including 56 normal, 20 precancerous, and 35 cancerous samples. The median age of women was 36, 37, and 45 years in the normal, precancer, and cancer groups, respectively. The median age was statistically significant different concerning stage (P < 0.001), as it was higher in the cancer group in comparison to the normal and the precancer groups. However, no statistically significant differences were observed for the median age by the cancer type (P > 0.9) (Table 2).

Table 2.

The demographic and virological characteristics among different pathological groups

Variable	Stage			P-value	Cancer Type		P-value
	Normal N = 56	Pre Cancer N = 20	Cancer N = 35		AdCa N = 8	SCC N = 22
Median age	36	37	45	< 0.001	45	45	> 0.9
HPV positive	55 (98.2)	19 (95.0)	34 (97.1)		8 (100)	21 (95.5)	–
HPV type
16 or 18	28 (51%)	14 (74%)	28 (82%)		5 (63%)	20 (95%)
Other	27 (49%)	5 (26%)	6 (18%)	0.007	3 (37%)	1 (5%)	0.052
Total	55 (100)	19 (100)	34 (100)		8 (100)	21 (100)

Pathological examination confirmed squamous cell carcinoma (SCC) in 22 (63%) and adenocarcinoma (AdCa) in 8 (23%) of the cancer samples. In the remaining 5 (14%) of the cancer samples, the pathological examination was indeterminate. HPV was detected in 108 (97%) of the samples. The infection rate of HPV16 or HPV18 in normal individuals, in the pre-cancer and the cancer group was 51%, 74%, and 82%, respectively. This finding indicated that the HPV 16 or 18 were more prevalent among precancer and cancer groups than normal group and a statistically significant difference was found in this regard (P < 0007). Also, stratification of cancer type was shown that the prevalence of HPV 16 or 18 was higher in the SCC (95%) than the AdCa (63%) group (Table 2).

HERV-E Env mRNA expression analysis

Higher levels of HERV-E env transcripts were significantly found in the cancer group than in the precancerous groups (1.5-fold change) (P = 0.031) (Fig. 1). However, no statistically significant differences were found in the expression levels of env between the precancerous and normal groups (P = 0.25) or the cancer and normal groups (P = 0.25). As shown in Fig. 2, there were no significant changes in the stratification of CC patients into SCC and AdCa groups (P = 0.56).

Fig. 1.

The expression levels of HERV-E env were shown in different pathological groups including normal, precancer, and cancer groups. The mean expression level of the HERV‐E env transcript was analyzed by the ANOVA test and the Kruskal − Wallis post hoc test assessment

Fig. 2.

The expression levels of HERV-E env were compared among CC patients stratified by cancer type (SCC vs. AdCa) by the nonparametric t‐test. AdCa adenocarcinoma; SCC squamous cell carcinoma

In addition, all HPV-positive samples were divided into two groups: the HPV 16 or 18-positive samples and those infected with other HPV types. Although higher levels of HERV-E env were found in the HPV 16 or 18-infected group compared to the groups infected with other HPV types, the difference did not reach a statistically significant level (P = 0.26) (Fig. 3). When HPV 16 or 18-infected patients were stratified into three pathological groups (normal, precancerous and cancerous groups), it was found that the higher mean value of HERV-E env mRNA differed significantly only between the cancerous and pre-cancerous groups (Fig. 4).

Fig. 3.

The expression level of HERV-E env mRNAs regarding HPV genotypes by the nonparametric test

Fig. 4.

The expression levels of HERV-E env were indicated among HPV 16 or 18-infected patients stratified into three pathological groups, including normal, precancer, and cancer groups. The mean expression level of the HERV‐E env transcript was analyzed by the ANOVA test and the Kruskal − Wallis post hoc test assessment

ROC curve analysis and accuracy of HERV-E Env mRNA expression

ROC curve analysis was used to evaluate the diagnostic accuracy of increased HERV-E env expression in differentiating precancerous and cancerous groups from controls. The AUC values showed a significant predictive value for the differentiation of precancerous and cancerous groups (Fig. 5A) (AUC = 0.68; CI 0.52–0.83; 57% sensitivity, 75% specificity, P = 0.015). However, the ROC curve analysis did not show a clear distinction between women with normal cervix in contrast to the precancerous and cancerous groups (AUC = 0.59, p = 0.12; and AUC = 0.57, P = 0.88, respectively) (data were not shown).

Fig. 5.

ROC curves of HERV-E env were described between precancerous and cancerous groups in all study patients (A) and HPV 16 or 18 infected patients (B). ROC curve analysis was also depicted to differentiate the pre-cancer group from healthy controls in patients infected with other types of HPV than HPV 16 or 18 (C). AUC, the area under the ROC curve; env, envelope; HERV, human endogenous retroviruses; ROC, receiver operating characteristic

Stratification of HPV type 16 or 18-infected patients into three different pathological groups (normal, precancerous, and cancerous) yielded similar results and showed that a significant diagnostic value for HERV-E env mRNAs between precancerous lesions and cervical cancer groups (Fig. 5B), (AUC = 0.69; 95% CI 0.52–0.86; 57% sensitivity and 78% specificity, P = 0.023). The AUC values showed no significant predictive value in separating HPV 16 or 18-infected women with normal cervix in contrast to the two groups with precancerous and cancerous lesions (AUC = 0.54, P = 0.33; and AUC = 0.6, P = 0.91, respectively) (data were not shown). Further stratification based on single and multi-type co-infections and stratification of HPV-positive samples by genotype (16/or 18 vs. other types) also showed no significant predictive values based on AUC in the ROC curve analysis (data were not shown). In addition, ROC curve analysis showed that the AUC value for HERV-E env mRNAs (AUC = 0.79, 95% CI 0.58–1 with 48% specificity and 100% sensitivity, P = 0.021) showed excellent distinction between women infected with other HPV types (not HPV 16 or 18) in the pre-cancer group and healthy controls (Fig. 5C).

Correlation analysis

In a series of patients with normal cervix and CC biopsies, it was investigated whether HERV-E env transcripts were associated with the expression level of HPV E6 and E7 mRNAs, depending on HPV 16 and HPV 18 genotype, and with the average age of the patients. As shown in Fig. 6A, although there was a trend toward a positive correlation between higher expressions of HPV E7 mRNA with HERV-E env mRNA in patients with precancerous lesions, no significant correlations were observed between the expression of HPV E6 mRNA and HERV-E env mRNA in normal, precancerous, and cancer groups (data not shown).

Fig. 6.

The Correlation of the HERV-E env gene expression to multiple variables was shown according to Pearson’s correlation coefficient (R) and P‐value (P). Scatter plots show HERV‐E env transcription relation with HPV E7 mRNA in different pathological groups (A) and HPV 16 or HPV 18 genotypes (B). Scatter plots show the correlation of HERV‐E env transcription with the average age of patients in HPV 16 or HPV 18 genotype groups (C)

In patients infected with HPV 16, a significant positive correlation (R = 0.34, P = 0.04) was found between higher expression of HPV 16-E7 and HERV-E env mRNAs (Fig. 6B). Interestingly, a correlation analysis showed a significant positive correlation between older age (> 40 years) and the expression of HERV-E env mRNA in HPV 16 (R = 0.35, P = 0.016). However, a negative correlation between older age and the expression of HERV-E env mRNA in HPV 18 (R=-0.52, P = 0.048) (Fig. 6C). For HPV 16- and 18-positive samples, correlation analysis showed an inverse pattern for HERV-E env expression and HPV E6 mRNAs or patient age (data not shown). When analyzing disease stages, no significant correlation was found between HERV-E env expression and two forms of cancer, including SCC and AdCa (data not shown).

Discussion

HPV malignancies such as cervical cancer represent a major public health problem, and their progression depends on many cofactors, including environmental and host cofactors [27]. Among the various cancer hallmarks, hyperactivation of specific HERV gene products, such as HERV-E env, could be a promising additional hallmark [14]. The present study is the first to investigate HERV-E env expression in 111 cervical biopsies, including precancerous, cancerous lesions, and normal biopsies. HERV-E env RT-PCR assays showed a statistically significant increase in HERV-E env expression in the cancer group compared to the precancerous group. ROC curve analysis allowed a partial differentiation of env mRNA expression between precancerous and cancerous lesions. Similarly, a significant diagnostic value of HERV-E env expression was demonstrated between precancerous and cancerous groups of patients infected with HPV 16 or 18 genotypes. In patients infected with HPV 16, there was also a positive correlation between a higher expression of HERV-E env mRNA and a higher level of HPV 16 E7 oncogene mRNA.

Interests in HERV-E studies focus on their ability to induce (auto) antibodies as they can produce retroviral antigens, on the immunosuppressive properties of the HERV-E env protein, and on the ability of HERV-E proviruses to act as alternative, tissue-specific promoters or enhancers [28]. Our results are consistent with previous studies showing higher expression of HERV-E transcripts in lung cancer [29] and ovarian cancer compared to non-malignant tissues [30]. Johanning et al. investigated HERV-E env expression in prostate cancer by RT-PCR and reported the presence of several HERVs, including HERV-E, in cancer tissues [22]. The study of HERV-E transcripts in renal cell carcinoma (RCC) was performed by Takahashi et al. [31], who demonstrated overexpression of 10-mer highly immunogenic peptide encoded by the HERV-E env gene (CT-RCC). This CT-RCC tumor-specific antigen was frequently expressed in RCC tumors, but not in normal tissues, at levels that can stimulate reactive T-CD8⁺ and cytotoxic RCC cells [31]. In another study, different transcriptional patterns of HERV-E sequences were found in hematopoietic cell lines and in several epithelial cell lines, including SiHa (infected with HPV-16) and HeLa (infected with HPV-18), suggesting that HERV-E may be selectively expressed in different cancer types [9].

There are few reports on the expression analysis of HERVs in CC, especially the association of HERV-E with CC has not been sufficiently investigated. A previous study by our group conducted by Soleimani et al. showed higher expression levels of HERV-K (HML‐2) transcripts (env, Np9, and Rec) in CC lesions, and high levels of env and Np9 mRNAs were shown in squamous cell carcinoma type of CC than adenocarcinoma form [14]. Similar to Soleimani’s study, we observed a significant increase of HERV-E env in CC biopsies compared to pre-cancer biopsies (P = 0.031). We did not find significant expression of HERV-E env in SCC and AdCa groups, which is likely due to our small sample size in the cancer group. A study by Curty et al. went deeper and investigated the expression of retroelements in CC and their correlation with HPV infection and the activity of neighboring genes. Their analysis of 40 CC biopsies revealed 8060 expressed retroelements, including HERV-K, HERV-H, HERV-E, HERV-I, and HERV-L, whose expression depends on the type of tumor and the presence of HPV [8].

Recently, in another relevant study, Alldredge et al. [32] investigated numerous HERV transcript expressions in CC tissues using RNAseq analysis, focusing on variables such as tumor HPV status, stage, and patient ethnicity. Remarkably, their research showed that HERV transcripts are upregulated in locally advanced cancer tissue in contrast to early-stage tumors, in patients with active HPV infection and depending on the patient’s ethnicity. Alldredge et al. also showed increased HERV expression in CC biopsies from patients with a higher mean age (53 years old) compared to a lower mean age (45 years old). In the present study, a significant positive correlation was found between older age and the expression of HERV-E mRNA in patients infected with HPV 16 (R = 0.35; P = 0.016). This could be the result of higher hypomethylation levels in HERV-E LTR sequences due to older age or abnormal demethylation in tumor-specific microenvironments [33]. These results suggest that examination of the HERV expression panel in combination with clinical factors such as HPV status and age may predict the outcome of CC [32].

ROC curve analysis revealed that HERV-E env expression had significant diagnostic value in all cancer patients in the study (AUC = 0.68; P = 0.015) and in cancer groups infected with HPV 16 or 18 genotypes (AUC = 0.69; P = 0.023) or with types other than HPV 16 or 18 (AUC = 0.79; P = 0.021). However, ROC curve analysis showed that the ability to distinguish precancerous lesions from healthy controls is limited. The study by Soleimani et al. found that ROC curve analysis of HERV-K env and Np9 expression had an excellent potential value for discriminating between precancerous and cancerous biopsies, and they suggested that these expressions could be a good potential biomarker for predicting CC progression [14]. In agreement with our study, one study investigated ERV-3 protein expression by the IHC technique in 19 locally advanced CC versus 28 early-stage CC and showed higher ERV-3 protein expression in early-stage CC patients [34]. Further stratification within the pre-cancer and cancer groups based on HPV type or co-infection could ensure the ability of HERV as a potential biomarker for the development of CC.

We were able to demonstrate that higher expression of HERV-E env in the correlation analysis (R = 034; P = 0.016) shows some agreement with high levels of HPV-16 E7 transcripts. Interestingly, HERV-E env transcripts are much more strongly associated with some HPVs. This explanation was demonstrated in the study by Allderdge et al., which showed a higher correlation of HERV transcriptional activity with active HPV infection compared to HPV-negative infection [32]. In addition, a significant positive correlation between HERV-K Np9 and HPV oncoprotein was recently found, suggesting that these viruses likely share common molecular mechanisms in carcinogenesis [14]. Therefore, further studies focusing on these promising interactions will clarify the complex interplay between HERV-E env expression, HPV infection, and CC development.

The most important limitations of the present study were the small sample size and the lack of functional validation of our results. We are suggesting the use of siRNAs against E7 of HPV 16/18 in HPV 16 (CasKi) and HPV 18 (HeLa) cell lines as well as human normal cervical fibroblast-like (HNCF) as a negative control, in future studies to clarify whether these oncoproteins activate HERV-E or not in cervical cancer.

Conclusion

In conclusion, our study showed that HERV-E is actively transcribed during the progression of cervical lesions. ROC curve analysis allowed a partial differentiation of env mRNA expression between precancerous and cancerous lesions. The positive correlation between a high level of HERV-E env transcripts and a high level of HPV 16 E7 mRNA suggests a possible role of HERV-E env in the transcription of HPV oncoprotein and vice versa. Future challenges lie in characterizing the possible interaction of the HERV-E env protein with the HPV oncoprotein to target the likely common signaling pathways and define a new therapeutic strategy for CC progression and other HPV-related malignancies.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

SMM and SJ have contributed to the design of the study; ShJ and RS have done all tests, MF, ZS, and AA have contributed to the analysis and interpretation of data; SJ and SMM have contributed to drafting the article.

Funding

This study has been funded and supported by Tehran University of Medical Sciences (TUMS), Grant no. (66589).

Data availability

Data is available on request from the authors.

Declarations

Ethics approval and consent to participate

Our research was conducted ethically by the World Medical Association Declaration of Helsinki. We declare that informed consent was obtained from all study subjects and the study was approved by the local ethical committee of Tehran University of Medical Sciences (IR.TUMS.SPH.REC.1402.054).

Competing interests

The authors declare no competing interests.