MicroRNA expression profiling of cutaneous squamous cell carcinomas and precursor lesions

Abstract Background Actinic keratoses (AK) are pre‐malignant skin lesions caused by chronic sun exposure. Progression from an AK to intraepidermal carcinoma (IEC) and a cutaneous squamous cell carcinoma (SCC) is well known but the rate of transformation to an invasive SCC is highly variable. Since no definitive biomarkers are available, treatment decisions are made ad hoc. Objectives To fully characterise our AK to SCC progression series, we performed microRNA (miRNA) microarray expression profiling of normal and photodamaged skin, as well as AKs, IEC, and invasive SCCs. Methods The study recruited 27 patients who donated fresh biopsies of normal skin, photodamaged skin, AK, IEC, and SCC (n = 67 specimens). All miRbase (v.21) miRNAs were profiled to identify miRNAs related to SCC progression. miRNAs were validated using qRT‐PCR and in vitro phenotypic assays. Results There were 234 robustly expressed miRNAs across the tissue collection, which resulted in 20 miRNA that were differentially expressed ((cor)p ≤ 0.05 and ≥ 10 fold) between normal skin and SCC. Hierarchical clustering all samples illustrated that AKs, IEC, and SCCs were largely indistinguishable, which confirms the premalignant status of an AK. A panel of miRNAs showed significant dysregulation between normal and photodamaged skin and AK. Importantly, we found miR‐34a‐5p and miR‐31‐5p had significant differential expression between AKs and IEC and IEC and SCC respectively. Phenotypic assays determined that the miR‐31 duplex had opposing effects on SCC cell lines which suggests that dysregulation of this duplex may be related to the dynamic control of progression of transformed keratinocytes. Conclusions This study confirmed the continuum of AK with IEC and SCC highlighting that miRNA expression plays a role in keratinocyte transformation. Development of our putative miRNA biomarker candidates is warranted to aid in clinical management of patients experiencing high AK load to determine the most appropriate treatment.

Cutaneous SCC is the second most common nonmelanoma skin cancer and represents a major healthcare cost due to its increasing global incidence. 1,2It is considered that the emergence of invasive SCC is part of a clinical continuum that begins with an AK that progresses into a SCC.AKs are a precancerous, atypical neoplasms of epidermal keratinocytes characterised by scaly erythematous patches and/or hyperkeratotic papules. 3AKs (or solar keratosis) frequently occur in sun-damaged sites of the scalp and face and it has been estimated that in the US alone, >40 million people have a diagnosed AK. 4 In Australia, it has been reported that >40% of people over the age of 40 have a prior diagnosis. 5Although these lesions at either end of the spectrum are relatively easy to identify clinically, differentiation becomes more difficult in the intermediate areas (i.e.Intraepidermal carcinoma) of this spectrum. 6Studies that have attempted to investigate the natural history of AK lesions, and to ascertain both the rate of malignant transformation from AK to SCC, and regression of AK, rely on clinical diagnosis and varying methods of mapping of AK lesions.The general consensus amongst clinicians is that if an AK is left untreated, it may eventually progress to the non-invasive form of SCC (IEC or in situ SCC) and ultimately progress to an invasive SCC.
The rate of malignant transformation of an AK is highly variable, with estimates ranging from 0.1% to 20% per year/per lesion. 7As such, identifying which AKs will progress into an invasive SCC is currently challenging.Instead, all AKs are treated through medical intervention, which may include the application of topical cream, cryotherapy, photodynamic therapy, as well as minimally destructive surgical procedures.This is becoming a key issue in clinical diagnostics, as the majority of patients have multiple AKs, with a recurrence rate of up to 60% within 12-24 months. 8urthermore, the treatment for AKs with intrusive medical intervention is a major financial burden to the health care system, as AKs may regress or not progress.As such, a biomarker that can differentiate between AKs and SCC would be of benefit for diagnostic purposes, and possibly a source for new tailored treatment options.
Herein we performed a comprehensive miRNA profile of a collection of normal and photo-damaged skin, AK, IEC, and SCC to identify a panel of miRNA that may offer clinical utility as predictive biomarkers.We also provide insight into the function of key miRNA and highlight the role they play in SCC progression.

| Patient specimen details
This study was approved by the University of Queensland Human Research Ethics Committee (Brisbane, Australia; #2013000711) and was carried out in accordance with the Declaration of Helsinki.All patients were recruited and consented at the Warner Family Medical Practice (Brisbane, Australia).All patient samples suspected of having an AK, an IEC or a cutaneous SCC were included in the study.All specimens were diagnosed by an expert pathologist and only patient samples diagnosed to be an AK, IEC, or SCC, were included for this study.A demographic summary of all patient and samples can be found in Table 1.
At the time of tissue collection, the specimen was bisected, with one half for routine histopathological diagnosis and the other half for laboratory analysis.The laboratory tissue was preserved in RNAlater (Thermo Fisher Scientific Inc, USA) prior to storage at −80°C.The DNA and RNA extractions were performed using the AllPrep DNA/RNA/miRNA Universal Kit (QIAGEN, Hilden, Germany).Briefly, the tissue specimens were placed in Buffer RLT Plus (QIAGEN,

What is already known about this topic?
� Keratinocytic skin cancers are the most common skin cancers in Australia and other countries with high levels of UV index.Actinic keratosis (AK) is a precursor of cutaneous squamous carcinomas (SCC), but the transformation rate is highly variable.Many people experience elevated AK load due to cumulative UV exposure.

What does this study add?
� A panel of microRNAs showed significant dysregulation between normal and photodamaged skin and AK.Importantly, we found miR-34a-5p and miR-31-5p had significant differential expression between AK and intraepidermal carcinoma (IEC) and IEC and SCC respectively.In vitro phenotypic assays determined that miR-31-3p and miR-31-5p had opposing effects on SCC cell lines, suggesting that dysregulation of this miRNAduplex may be related to the dynamic control of progression of transformed keratinocytes from skin to AK/IEC/SCC.
Hilden, Germany) and β-mercaptoethanol (Invitrogen, #21985023) and then disrupted and homogenised using a Bullet Blender (Next Advance, NY, USA).The debris was pelleted, and the supernatant was collected for co-extraction as per manufacturer's instructions.All RNA samples were quantified using Qubit RNA HS Assay kit (Invitrogen, #Q32852) as per manufacturer's protocol.

| MicroRNA profiling and data analysis
A total of 500 ng Total RNA from normal skin (n = 4), perilesional (n = 4), AK (n = 10), IEC (n = 10) and SCC (n = 10) were shipped to LC Sciences (Houston, USA) to perform the miRNA microarray profiling.The profiling was performed using a custom array platform (μPar-aflo® technology) containing ~2500 miRNA probes (based on miRbase v21).The quality control, labelling (Cy-3), hybridisation, scanning, signal background subtraction and global normalisation (LOWESS) were performed by LC Sciences.Advanced data analyses were performed in Genespring GX (Agilent Technologies, Santa Clara, USA) using the LOWESS normalised signal intensity values.All signal intensity values < 30 were considered 'background expression' (personal communication with LC Sciences) and changed to 0.01 prior to log2 transformation in Genespring.
Expression of the 20-miRNA panel was assayed in each sample with 4 technical replicate Taqman assays.Real-time expression data was extracted and analysed as previously described. 9

| RNA extraction
The total RNA of transfected cells was extracted using the miRNeasy Mini Kit (QIAGEN, Hilden, Germany) as per manufacturer's protocol.Briefly, cells were harvested using 0.25% trypsin solution and centrifuged for 2 min at 1200 rcf.The supernatant was discarded, and the cells were lysed in QIAZOL solution (QIAGEN, Hilden, Germany), phase separated with chloroform, and centrifuged at 12,000 g at 4°C.Total RNA was precipitated from the upper aqueous layer with 100% ethanol prior to column purification.

| Cellular viability assay
The sulforhodamine B (SRB) assay was utilised to quantify cellular proliferation as described previously. 10riefly, ~2000 cells/well were seeded into a 96-well plate and reverse-transfected with 5 nM final concentration of each miRNA mimic and negative control (5 wells/transfection).Cells were then incubated for 72 h (Colo-16) or 96 h (SCC-9) prior to being fixed with methanol and washed according to standard SRB assay conditions.Colourimetric quantification of the SRB dye was performed at 564 nm wavelength by using a microtiter plate reader (Multiskan GO; Thermo Fisher).Each assay was repeated in triplicate.

| 2D colony-forming assay
Firstly, ~1000 cells/well of Colo-16, SCC-25 and SCC-9 were seeded in a 6-well plate for 24 h.Then, the media was removed, and cells were forward transfected as per manufacturers recommendation.All three miRNA mimics and negative control assays were performed in triplicate wells with a final concentration of 5 nM.The cells were incubated at 37°C, in a 5% CO 2 humidifier for 10 days.To visualise the colonies, the cells were fixed and stained with 4% (vol/vol) paraformaldehyde and 0.5% (vol/vol) crystal violet mixture for 30 min.Each assay was repeated in triplicate.Each plate was scanned and colonies visually inspected and counted manually.

| Statistical analysis
GraphPad Prism version 8.0.0 for Windows (San Diego, California USA) was used for all statistical analysis.Each miRNA was assessed for expression variance using a one-way analysis of variance (ANOVA) or Kruskal-Wallis test.Mann-Whitney U-test were performed for pair-wise comparisons.Predictive ability of each miRNA was evaluated using receiver operating characteristic (ROC curve) and area under the curve (AUC) or AUROC (GraphPad Prism 8).A pvalue ≤0.05 was considered statistically significant.

| MicroRNA expression levels are dysregulated in cutaneous SCCs and precursor lesions
To find miRNAs that may be involved in the transformation of AK into SCC, we performed global miRNA expression profiling of normal skin (n = 4), perilesional skin (n = 4), AK (n = 10), IEC (n = 10), and SCC (n = 10) samples (Table 1), using a microarray platform that contained ~2500 miRNAs (miRBase v21).To identify miRNAs that had robust expression levels, the microarray signal intensities were stringently filtered (Materials and Methods) which reduced the list to a total of 234 miRNAs.Next, differentially expressed (DE) miRNA that may be related to SCC progression, were identified using a volcano-plot analysis (p ≤ 0.05 and ≥10 fold; corrected for multiple testing using Benjamini-Hochberg) was performed with the group of normal skin samples (n = 4) compared with SCC (n = 10).This revealed a specific gene list of 31 miR-NAs that may be related to SCC progression.To ensure that only robustly expressed miRNAs (Materials and Methods) were present in the 31-gene list, we next performed a Venn analysis to find overlap with the prior filtered gene list (n = 234) based upon robust expression, which resulted in 20 miRNA genes (Table 2).Three miRNAs (miR-664b-3p, miR-6782-5p, miR-7975) were significantly downregulated in SCC compared to normal skin, whilst 17 (Table 2) of the miRNAs were upregulated in SCC.Interestingly, 11 of 20 miRNA have not previously been associated with SCC progression.Next, we performed a semi-supervised hierarchical clustering of the 20 DE miRNAs (Figure 1).This highlighted that there were two main clusters consisting of normal/photo-damaged skin and lesional tissue (AK, IEC, and SCC) (Figure 1).It is important to note that the lesional tissue were not clustered according to their specific category which provide further support for the AK-IEC-SCC clinical continuum.

| Quantitative real-time PCR validation of differentially expressed miRNAs
To confirm the expression of the 20-miRNA panel, we used a sensitive method of validation (Materials and Methods) in an extended collection (including samples present on the microarray) of skin and lesional tissue.The combined tissue cohort (n = 67) comprised of normal skin (n = 16), perilesional skin (n = 16), AK (n = 14), IEC (n = 11) and SCC (n = 10) samples.Expression was detected in all dilutions of a positive control for 14 of the 20 miRNAs (Table 3) except for miR-617, miR-1224, miR-1273c, miR-4298, miR-4306 and miR-6782-5p which experienced assay failure; thus, the panel herein will be referred to as 'SCC-miR-14'.The 14 expressed miRNAs along with ANOVA and multiple comparative tests are represented in Table 3 and Figure S1, with significance indicated (range, p = 0.0466 -<0.0001).Three miRNA (miR-7-5p, miR-142-5p, and miR-1233-5p) showed no significance in an ANOVA across all groups, however miR-1233-5p did show significance in normal skin versus AK (p = 0.0113) (Table 3 and Figure S1).

| miRNA expression can distinguish normal and photo-damaged skin from AKs, IECs or squamous cell carcinoma
Among the SCC-miR-14, there were several miRNAs expressed in the AK, IEC, and SCC tissues, that were significantly DE compared with normal and photodamaged skin (Table 3 and Figure S1).However, only miR-34a-5p reached significance (cor.p = 0.0128) when the panels expression was compared between AK and IEC and there was no significant expression difference when AK and SCCs were compared or between IECs and SCCs.These lack of significant differences between AK/IEC/SCC is not surprising given the clinical scenario with AKs known to be a precursor for IEC and SCCs.The lack of significant differences can also be illustrated in Figure 1, with no clear delineation of AKs, IECs, or SCC.We next performed area under the receiver operator curve (AUROC) analysis to determine the sensitivity and specificity of the comparisons.It is apparent that the significant differences in expression noted in Table 3 also corresponded to high AUROC scores (Table 4) in most cases.This strongly indicates that all members of SCC-miR-14, with the exception of miR-7-5p, and miR-424-5p, contributes to AK/IEC/SCC development.
Since, the formation of AK can lead to the IEC/SCC progression, we next explored if any of the SCC-14 miRNAs had the potential to be used as a predictive biomarker of AK progression.Table 4 illustrates that most miRNA were involved early in the formation of AK/ IEC/SCC from normal/perilesional skin, whereas only miR-34a-5p and miR-31-5p have the potential utility as predictor of AK to IEC and IEC to SCC transformation respectively.

| Overexpression of miR-31-3p and miR-7150 increases cellular viability and colony formation in squamous cell carcinoma cell lines
Next, we selected three miRNAs to perform functional validation as to the role they may play in SCC progression.We selected the known miRNA duplex of miR-31 (miR-31-3p and miR-31-5p) as well as a novel miRNA (miR-7150), as they had some of highest fold changes between normal skin and AK/IEC/SCC to perform functional assays (Table 2).Importantly, neither of these miRNAs had been specifically characterised in the setting of keratinocyte skin cancer.To assess for cellular viability, miRNA mimics and negative control were transiently transfected into two cell lines (Colo-16 and SCC9).Interestingly, both miR-31-3p and miR-7150 significantly (p-value <0.05) increased cellular proliferation (Figure 2) whereas miR-31-5p did not show any significant changes to cellular proliferation in the assessed cell lines (Figure 2).Next, to Abbreviations: AK, actinic keratosis; ANOVA, analysis of variance; FDR, False Discovery Rate; IEC, intraepidermal carcinoma; SCC, cutaneous squamous cell carcinoma.
determine if miRNA expression levels effect the ability to form 2D colonies in SCC cell lines, transfection of miR-31-3p and miR-7150 led to an increase in colonies, as compared to negative control (Figure 3).In keeping with our previous observation of cell viability following overexpression of miR-31-5p, we observed a decrease of colony formation in all three SCC cell lines (Colo-16, SCC9, SCC25).

| DISCUSSION
It is evident that miRNAs are dysregulated in the cancer setting and their regulation of important signalling pathways play an important role in the transformation of normal cells into the cancerous state. 19However, there have been limited studies performed in the setting of SCC or their precursor lesions such as AK and IEC. 20,21oreover, we still have very limited understanding on natural history of AK formation.Here, we performed a global miRNA expression profiling in SCC and its associated precursor lesions with the goal to find potential biomarkers and the role of miRNA in SCC development.
Initially, we identified 20 miRNA that were DE between normal skin and SCC via microarray profiling.Hierarchical clustering of these miRNA highlighted that AK, IEC, and SCC were essentially indistinguishable from one another, and that they are indeed highly related.A panel of 14 (SCC-miR-14) were subsequently determined to have detectable expression in an extended tissue collection.Upon statistical analysis, several of the SCC-miR-14 panel had significant dysregulation in their expression between normal (and photodamaged) skin and SCC which highlights that normal skin and SCC have distinct miRNA expression signatures.AUROC assessment of these miRNAs revealed a high AUC score indicating that these miR-NAs could easily be detected in patient samples with high specificity and sensitivity.Similarly, a panel of miRNAs showed significant dysregulation between normal skin and AK.Importantly, we found miR-34a-5p and miR-31-5p had dysregulation between AKs and IEC and IEC and SCC respectively.This has important clinical relevance, as currently no biomarkers are currently routinely available that can distinguish AKs from SCC, as such the differential expression of these miRNA may aid in histopathological diagnosis of AK highlighting their malignant potential.This corroborates previous report that AKs and SCC have distinct molecular signature based on cDNA microarray studies. 22,23emarkably, we found several miRNAs that show significant dysregulation between normal skin and photodamaged skin.This suggests that miRNA dysregulations occur very early on in the transformation of normal keratinocytes, leading to the formation of AK lesions and later into SCC.To further elucidate the role miRNA in SCC, we performed phenotypic assays to observe the effect of the overexpression of miR-31-5p, miR-31-3p and miR-7150 in three SCC cell lines (Colo-16, SCC-9 and SCC-25).MiR-31 functional assays have been previously performed in the SCC context, 24 however the individual members of the duplex (miR-31-5p and miR-31-3p) had not been studied separately.In a study by Wang et al, 24 the miR-31 precursor (pre-miR-31) was used which contains both strands (5 0 and 3 0 ) of the miRNA duplex.Wang et al 24 found that the overexpression of pre-miR-31 resulted in increased proliferation, colony formation, and cell migration and invasion.Increasingly, it has now been shown that for many miRNAs, both strands of the miRNA duplex have functional activity and in some instances, they may have an antagonist function. 25verexpression of miR-31-3p resulted in increased cellular proliferation and increased 2D colony formation in the SCC cell lines, whereas miR-31-5p overexpression did not alter cell proliferation and reduced colony formation was shown.Mir-31-5p is known to have the higher expression, compared with miR-31-3p across many different tissue types, which is also observed in our study.Whilst we didn't assess the specific effect of miR-31-5p on invasion and migration, Wang et al 24 confirmed that the expression of pre-miR-31 leads to increased invasion.Based upon our observations, the opposing effects of the miR-31 duplex suggests that their expression may be important in fine tuning the growth of transformed keratinocytes.This suggests the expression of miR-31 may be enriched in keratinocytes and such would be suitable as a keratinocytic cancer biomarker.
Our study has a number of limitations including small sample size and lack of specimens with histopathologic evidence of SCC arising from an AK/IEC.This study has however confirmed the clinical continuum of the premalignant AK with IEC and SCC and we have highlighted that miRNA expression plays a role in keratinocyte transformation.Further development of our putative miRNA biomarker candidates is warranted to aid in clinical management of patients experiencing high AK load to determine the most appropriate treatment.This may be in the form of tape stripping 26

F I G U R E 1 A
semi supervised hierarchical clustering of 20 significantly (p(corr)≤0.05)differentially expressed (DE) microRNA (miRNA) between squamous cell carcinoma (SCC) and normal skin.T A B L E 3 Differential expression of candidate miRNAs quantified in the extended validation patient samples using RT-qPCR.Statistically significant FDR corrected p-value (≤0.05) has been highlighted.

T A B L E 4
Abbreviations: AK, actinic keratosis; AUROC, area under the receiver operating characteristics; IEC, intraepidermal carcinoma; SCC, cutaneous squamous cell carcinoma. or

F I G U R E 3
The 2D colony formation of Colo-16, SCC-9 and SCC-25 cell lines transfected with 5 nM of miR-31-5p, miR-31-3p and miR-7150 after 10 days and stained with crystal violet.The representative image is derived from triplicate experiments.

miRNA Fold change Adjusted p-value Alteration in SCC relative to normal skin Known/ Novel
The list of 20 differentially expressed (DE) miRNAs in squamous cell carcinoma (SCC) from global microRNA (miRNA) expression profiling.
T A B L E 2