Assessing SPP1/Osteopontin (OPN) Splice Variants and Their Association to Nonmelanoma Skin Cancer by Absolute Quantification: Identification of OPN-5 Subvariants and Their Protein Coding Potential

The study evaluated whether SPP1/osteopontin (OPN) splice variants are differentially expressed in nonmelanoma skin cancer compared to normal skin. The absolute number of mRNA molecules of OPN-a predominated in normal skin and nonmelanoma skin cancer compared to OPN-b, OPN-c, and OPN-5. However, mRNAs of OPN-a, OPN-b, and OPN-c were expressed in higher levels in cutaneous squamous cell carcinomas (cSCCs) and basal cell carcinomas relative to normal skin. Additionally, OPN-5 expression was higher than OPN-b and OPN-c, and OPN-c, in normal skin and nonmelanoma skin cancer, respectively. Furthermore, we identified four OPN-5 splice variants, which were cloned and analyzed for protein expression.

Keywords: Osteopontin; alternatively spliced variants; cutaneous squamous cell carcinoma; basal cell carcinoma; absolute quantification; OPN-5 subvariants

Secreted phosphoprotein 1 (SPP1), commonly known as osteopontin (OPN), is a matricellular, acidic glycoprotein. Across species, its primary sequence consists of several conserved functional motifs: 1) a string of 8–10 aspartic acid residues and a calcium-binding region, both of which bind to hydroxyapatite; 2) Arg-Gly-Asp (RGD) and the adjacent sequence Ser-Val-Val-Tyr-Gly-Leu-Arg-Ser (SVVYGLRS) bind integrin receptors; and 3) the putative sequence Asp-Lys-His-Leu-Lys-Phe-Arg-Ile (DKHLKFRI) at the carboxy terminal binds to CD44s or its variants (Figure 1(A)). OPN belongs to a family of intrinsically disordered proteins ([1]). However, it retains transient substructures when bound to cell surface receptors or ligands ([2]). The flexibility of OPN's semi-random coiled structure, coupled with various conserved functional motifs, contributes to its capacity to interact with multiple partners and, consequently, to exhibit pleiotropic functions, such as mineralization, prevention of calcification, chemotaxis, cell signaling, cell proliferation, and cell survival ([[3]]). Elevated expression of OPN is associated with various cancer types. It is hypothesized to promote tumorigenesis, tumor progression, and/or metastasis ([[6]]).

PHOTO (COLOR): Figure 1. Exon organization and schematic structure of human SPP1/OPN protein and the expression of OPN splice variants in normal skin, cSCCs, and BCCs. (A) OPN-a consists of seven exons. OPN-b, OPN-c, and OPN-4 are similar to OPN-a in exon organization, except for deleted exon 5, deleted exon 4, and deleted exons 4 and 5, respectively. In these isoforms, the translation start sites are present in exon 2. The structure of the most common human OPN protein, with 314 amino acids (aa), is indicated above the exon organization of OPN-a. Conserved motifs are indicated in red and blue boxes. OPN-5 has an extra exon between exons 3 and 4 of OPN-a. Furthermore, the translation start site is located in the extra exon. The predicted full length of the OPN-5 protein is 327 aa. Seq.: sequence; TG: transglutaminase site, Asp: aspartic acid; RGD: Arg-Gly-Asp; SVVYGLRS: Ser-Val-Val-Tyr-Gly-Leu-Arg-Ser; DKHLKFRI: Asp-Lys-His-Leu-Lys-Phe-Arg-Ile. (B) Top graph, expression of the absolute number of mRNA molecules of OPN-a, OPN-b, OPN-c, and OPN-5 normalized to β-actin. Since the number of mRNA molecules of OPN-4 is low, this variant was not included in the graph. Primer pairs used to amplify the OPN variants are included in Table 1. Lower graph, an enlarged histogram indicating the absolute number of mRNA molecules of OPN-b, OPN-c, and OPN-5. The dotted and solid lines across the histograms represent the means and the medians, respectively. Normal skin, n = 16 (OPN-a, OPN-b, and OPN-c), n = 15 (OPN-5); cSCCs, n = 11 (OPN-a, OPN-b, and OPN-c), n = 9 (OPN-5); BCCs, n = 12 (OPN-a, OPN-b, OPN-c, and OPN-5).

Human OPN pre-mRNA undergoes alternative splicing, generating three major forms, OPN-a (OPN-v1), OPN-b (OPN-v2), and OPN-c (OPN-v3), initially found in gliomas ([13]) (Figure 1(A)). Recently, two additional splice variants, OPN-4 (OPN-v4) and OPN-5 (OPN-v5), were found in esophageal adenocarcinomas and glioblastomas and in various tumor cell lines ([[14]]). Differential splice variants of OPN are associated with specific cancer types, suggesting differential functions of these variants ([17],[18]).

In normal skin, OPN is minimally expressed in low sun-exposed and non-inflammatory regions but highly induced in wound-healing, psoriasis, melanoma, and nonmelanoma skin cancer ([19],[20]). Our preclinical studies with two-stage skin carcinogenesis and photocarcinogenesis models support OPN's role in facilitating the development of papillomas and cutaneous squamous cell carcinomas (cSCCs) ([8],[9],[21]). Gene expression profile studies using microarray analyses show elevated transcripts of human OPN in cSCCs compared to normal skin ([22],[23]) and in cSCCs more than in actinic keratoses, precursors to cSCCs ([24]). However, these studies do not distinguish the various splice variants of human OPN. Whether skin and nonmelanoma skin cancers express alternatively spliced variants of OPN has not been reported.

We hypothesized that OPN splice variants are present in the skin and that they are differentially elevated in nonmelanoma skin cancer relative to those in normal skin. Thus, we evaluated whether OPN splice variants are differentially expressed in nonmelanoma skin cancer compared to normal skin. We found that the OPN-a transcript was the predominant splice variant expressed in the skin compared to OPN-b, OPN-c, and OPN-5 and that OPN-4 was minimally expressed. Additionally, OPN-5 expression was higher than that for OPN-b and OPN-c in normal skin and higher than OPN-c in cSCCs and BCCs. OPN-5 consisted of several subvariants. We identified and cloned four alternative spliced transcript variants of OPN-5 and examined their protein expression.

The Institutional Review Board at the University of Alabama at Birmingham approved the protocol. Remnant specimens of nonmelanoma skin cancer were obtained from debulking specimens from Mohs surgeries. Normal skin specimens were collected from standing cone redundancies generated, removed, and discarded during wound-closure procedures. The specimens were immediately flash frozen in liquid nitrogen and stored at −80 °C prior to molecular analyses. The specimens collected varied in size. Consequently, small tissue samples did not allow for all analyses. The numbers of samples used for each assay are specified in the figure legends.

cSCC cell lines, SCC-13 and A431, were kindly provided by Dr. Andrei Slominski and Dr. Natalia Keshville, respectively. Cells were maintained in DMEM with 10% FBS, 2% glutamine, and 0.5% antibiotics. For analysis of OPN splice variant transcripts, cells were seeded at equal densities in a 100-mm dish and grown to near 90% confluency prior to harvest.

Skin specimens were homogenized, and total RNA was extracted by TRIzol® reagent (Life Technology, Carlsbad, CA). The concentration and purity of total RNA were assessed spectrophotometrically at 260 and 280 nm. The first strand of cDNA was synthesized according to the manufacturer's instructions, using 2 μg of total RNA with SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA) and random hexamers.

In generating standard curves to determine the absolute numbers of mRNA molecules of OPN splice variants expressed in normal skin and nonmelanoma skin cancer, plasmids containing specific cDNA fragments that distinguish among OPN-a, OPN-b, OPN-c, OPN-4, and OPN-5 were generated. These specific cDNA fragments were obtained by PCR cloning with primers (Table 1) using patients' cDNA. Target PCR products were quantified in 2% 1X TAE agarose gel electrophoresis, cut out, and eluted. The purified PCR products were cloned into a pGEM-T vector system (Promega, Madison, WI) according to the manufacturer's instructions. The recombinant vectors were transformed into E. coli JM109 competent cells and selected with X-gal/IPTG and ampicillin. Positive colonies were expanded, followed by plasmid purification. The specific cDNA fragments corresponding to each of the OPN transcript variants were confirmed by restriction enzyme digestion and DNA sequencing. The DNA concentrations (µg/µl) of these plasmids were determined with a UV–VIS spectrophotometer, and the actual copy numbers were calculated with the following equation:

Graph

$$\text{Number}\mathrm{}\text{of}\mathrm{}\text{copies}/\text{µl}=\frac{6.02\text{×}{10}^{23}\mathrm{}\mathrm{(}\text{molecules}/\text{mole}\text{)×}\text{DNA}\mathrm{}\text{concentrations}\text{(g}/\text{µl)}}{\text{Number}\mathrm{}\text{of}\mathrm{}\text{base}\mathrm{}\text{pairs}\text{×}660\mathrm{}\text{daltons}}$$

Table 1. Primers for real-time qPCR.

1 aOPN-a and OPN-5 share the same reverse primer; bbp: nucleotide base pairs.

For generating standard curves, each plasmid containing a specific cDNA fragment of an OPN splice variant was serially diluted. QPCR analyses were performed with their corresponding primer sets (Table 1) on each of the serially diluted plasmids to obtain their Cp values using a Roche LC480 Light Cycler system (Roche, Mannheim, Germany). The Cp values were plotted against the log of their initial template copy number. The absolute numbers of molecules of each OPN isoform mRNA were determined by extrapolating the Cp value to its standard curve followed by normalization with β-actin mRNA.

qPCR amplification was performed using PowerUp SYBR Green PCR Master Mix according to the manufacturer's specifications (Applied Biosystems Incorporated, Foster City, CA) and analyzed with a Roche LC480 Light Cycler system.

To determine subvariants of OPN-5, the protein-coding cDNAs of OPN-5 and its subvariants were cloned using the cDNA derived from the total mRNA of normal skin and cSCCs as the template. Briefly, full-length PCR cloning was performed with Q5 DNA polymerase (New England Biolabs Inc., Ipswich, MA) using primer set complementary to the putative ATG site of exon 4 of OPN-5 (Figure 3(A)) and the end of exon 8 (forward primer: GCCAAAGGATCCATGGGCATTGTCCCCAGA and reverse primer: TTTTTTCTCGAGTTAATTGACCTCAGAAGA). Approximately 1 kilobase pairs of PCR products were isolated, eluted, and digested with BamHI and XhoI. The enzyme-digested PCR products were cloned into the modified pcDNA version 3.1 (Invitrogen, Carlsbad, CA, USA) Flag-tagged expression vectors ([25]). These vectors were transformed into E. coli JM109 competent cells as described above. Positive colonies containing plasmids with OPN-5 or its subvariants were confirmed by restriction enzyme digestion and DNA sequencing.

OPN-5a and its subvariants (OPN-5b, OPN-5c, OPN-5d, and OPN-5e) in modified pcDNA 3.1-Flag-tagged expression vectors ([25]) were transfected into SCC-13 cells with Lipofectamine 2000 according to the protocol from the vendor (Invitrogen, Fisher Scientific, Waltham, MA). After 48 h, the media were collected and centrifuged, and the supernatants were concentrated 40-fold. Cells were lysed with RIPA buffer containing protease inhibitors and phosphatase inhibitors (Roche, Indianapolis, IN). Protein from 30 µl of the concentrates and 30 µg of cell lysates were separated on 10% SDS-polyacrylamide gels and transferred to PVDF membranes for Western blot analysis, as previously described ([9]).

Skin specimens were homogenized in RIPA buffer containing protease inhibitors (Roche, Indianapolis, IN) and centrifuged. The protein concentrations in the supernatants were determined by PierceTM bicinchoninic acid assays (Thermo Fisher Scientific Co., Waltham, MA). Equal amounts of protein lysates (40 µg/lane) or pre-stained molecular weight standards (Precision Plus Protein™ Dual Color Standards, Bio-Rad, Hercules, CA) were separated by 8.5% or 10% SDS-polyacrylamide gel electrophoresis, transferred onto polyvinylidene fluoride (PVDF) membranes (EMD Millipore), and probed with a monoclonal antibody to human OPN (mouse IgG2b,κ, BioLegend, San Diego, CA) with β-actin (mouse IgG1,κ, Santa Cruz Biotechnology, Inc., Dallas, TX) as an internal control. For detection of recombinant Flag-tagged OPN-5 variants, the PVDF membranes were probed with antibody to flag (rat IgG2a,γ, Biolegend, San Diego, CA). Protein bands were visualized with chemiluminescence substrate (Amersham ECL, GE Healthcare, Pittsburgh, PA) after incubation with secondary antibodies coupled to horseradish peroxidase and exposure to blue X-ray films. Densitometric analyses of the protein bands were performed with ImageJ Software.

All statistical analyses were performed with SigmaPlot version 11.0 (Systat Software, Inc. San Jose, CA, USA). Paired t-tests were used for comparisons between groups. Normality tests by Shapiro–Wilk were performed. For those groups that did not pass the normality test, the Mann–Whitney Rank Sum Test was performed. p values <.05 were statistically significant.

Skin specimens were obtained from 39 subjects. These de-identified specimens included 16 normal skin; 11 cSCCs, mostly of the well-differentiated type; and 12 BCCs of basaloid tumor that extended from the epidermis to the dermis and beyond. Most of the specimens were from sun-exposed skin of Caucasian males (Table 2, detailed in Table S1). The average ages of patients providing specimens of normal skin, cSCCs, and BCCs were 65 ± 8, 79 ± 10, and 70 ± 11 years old, respectively.

Table 2. Summary of skin samples from subjects with Mohs surgery.

• 2 aMean ± SD; bNormal skin from standing cone redundancies, see Supplementary Materials and Methods.
• 3 cSCC: cutaneous squamous cell carcinoma; BCC: basal cell carcinoma; W: Caucasian; A: Asian; M: male; F: female.

To evaluate the expression of OPN splice variants, we determined the absolute numbers of mRNA molecules of OPN-a, OPN-b, OPN-c, OPN-4, and OPN-5 in normal skin and nonmelanoma skin cancer. Among the splice variants analyzed, OPN-4 showed minimal expression in normal skin, cSCCs, and BCCs; the number of mRNA molecules ranged from 2 to 261. Comparison between OPN splice variants indicated that OPN-a transcript expression predominates (p <.001) in the skin, cSCCs, and BCCs (Figure 1(B), top panel). In normal skin, comparing the average number of mRNA molecules of OPN-a to those of OPN-b, OPN-c, and OPN-5 showed levels of 227-, 268-, and 42-fold, respectively.

Relative to OPN splice variants in normal skin, the expression of mRNA molecules of OPN-a, OPN-b, and OPN-c were elevated in nonmelanoma skin cancer (Figure 1(B)). For OPN-a, the average numbers of mRNA molecules in cSCCs and BCCs were 4-fold higher than that of normal skin. For OPN-b, the average numbers of mRNA molecules in cSCCs and BCCs were 13-fold and 8-fold higher, respectively, than that of normal skin. For OPN-c, the average numbers of mRNA molecules in cSCCs and BCCs were 3-fold higher than that of normal skin. Additionally, the transcript levels of OPN-b in cSCCs (p <.022) and BCCs (p <.007) were higher than that of OPN-c in cSCCs and BCCs.

Compared to OPN-b and OPN-c, OPN-5 mRNA molecules were abundantly expressed, implicating their functional importance in skin biology. In normal skin, there were more mRNA molecules of OPN-5 compared to OPN-b (p <.003) and OPN-c (p <.002). Levels of OPN-5 transcripts were also higher in cSCCs and BCCs compared with those of OPN-c (Figure 1(B), lower panel). In the real-time quantitative PCR products of OPN-5, there were additional bands below the predicted OPN-5 at 324 bp (Figure 2(A)) while only single bands were present for OPN-a, OPN-b, and OPN-c (Figures 2(B–D), respectively), suggesting the possibility of subvariants of OPN-5. Subsequent cloning analyses of protein-coding cDNAs from normal and cSCC samples followed by sequencing confirmed the identification of four alternative splice variants of OPN-5, which were named OPN-5b, OPN-5c, OPN-5d, and OPN-5e (Figure 3(A)). We designated OPN-5a as consisting of the full eight exons with the translation start site in the additional exon "E4" in Figure 3 (stated as "extra" in Figure 1(A)) not present in OPN-a, OPN-b, and OPN-c. OPN-5b consists of truncated E4. OPN-5c and OPN-5d consist of an additional 9 bp of intron at the 3′ end of E4 with OPN-5d also lacking E6 (Figure 3, designated as E5 in Figure 1(A)). OPN-5e lacks E6 (Figure 3). For these four subvariants of OPN-5, we speculated that the number of mRNA molecules of OPN-5 in normal skin, cSCCs, and BCCs shown in Figure 1(B) based on the designated PCR primer set used (see Table 1 and Figure 3(A)) likely included OPN-5b and OPN-5c.

Graph: Figure 2. PCR products of OPN-a, OPN-b, OPN-c, and OPN-5 in normal skin, cSCCs, and BCCs. (A) Real-time qPCR using primers (Table 1 and location of forward and reverse primers are indicated in Figure 3(A)) for OPN-5 showed the generation of several PCR products, including the expected 324 bp in normal skin, cSCCs, and BCCs. Zoom images of PCR products below 0.4 kbp are shown for normal skin, cSCCs, and BCCs. Ten specimens ([[1]]). (B) A single PCR product of OPN-a at 159 bp in normal skin, cSCCs, and BCCs. (C) A single PCR product of OPN-b at 100 bp in normal skin, cSCCs, and BCCs. (D) A single PCR product of OPN-c at 149 bp in normal skin, cSCCs, and BCCs. Six specimens ([[1]]) in (B), (C), and (D). D: DNA ladder; kbp: kilo base pair; NC: no template.

PHOTO (COLOR): Figure 3. Exon organization and protein expression of OPN-5a and its subvariants. (A) Schematic diagram of exon organization of OPN-5a and its splice variants. OPN-5a consists of eight exons. Note that exon 4 (E4) is designated as an "Extra" exon in Figure 1(A), top diagram. The newly identified splice variants of OPN-5a are named OPN-5b, OPN-5c, OPN-5d, and OPN-5e. The exon diagrams below the top diagram indicate the cloned protein-coding cDNA. The corresponding predicted numbers of amino acids (aa) of the cloned OPN-5 splice variants are specified at the end of each exon diagram. Please note that for absolute quantification of OPN-5 mRNA molecules by the standard curve method the primer pair (indicated by the forward and reverse arrows below E4 and E7, see Table 1 for sequences of primer pairs) used to amplify specific fragments of OPN-5a can also amplify additional PCR products of 230 and 333 bp, which correspond to OPN-5b and OPN-5c, respectively. (B) Recombinant protein expression of OPN-5a and its subvariants in media and cell lysates of SCC-13 cells. Expression of OPN-5a and its subvariants were detected using an antibody to Flag-tag. Top panel, OPN-5a, OPN-5b, and OPN-5e are highly expressed in the media. Lower panels, OPN-5b is also expressed intracellularly. The lower intense band (∼40 kDa, top panel) in the media and the two intense bands (∼40 and ∼34 kDa, lower panels) in cell lysates are non-specific proteins that cross-react with antibody to Flag-tag. The arrows indicate the bands that correspond to various OPN-5 variants. The number of aa stated corresponds to the sum of the protein-coding cDNA plus eight aa of the Flag-tag. Ct: cells only; v: cells transfected with vector lacking recombinant OPN-5 cDNA; a: cells with OPN-5a vector; b: cells with OPN-5b vector; c: cells with OPN-5c vector; d: cells with OPN-5d vector; e: cells with OPN-5e vector. MW std: molecular weight standard (Precision Plus Protein™ Dual Color Standards, Bio-Rad).

To determine if subvariants of OPN-5a have protein-coding potential, OPN-5a and its four subvariant's protein-coding cDNAs in Flag-tagged mammalian expression vectors were transfected into a cSCC cell line, SCC-13, which expresses minimal endogenous OPN (Table S2). Using an anti-Flag antibody to detect the OPN-5a and its subvariants, we found that OPN-5a, OPN-5b, and OPN-5e were expressed as secreted proteins. Although, the secreted OPN-5b has lower number of amino acids (aa) (302 aa, which consists of 294 aa plus 8 aa from Flag-tag) compared to OPN-5a (335 aa) the apparent high molecular weight is due to its abundance of expression resulting in slower migration (Figure 3(B), top panel). In addition to secretion of OPN-5b, it was also present intracellularly (Figure 3(B), lower panels). In contrast to the other OPN-5 variants, there were minimal expressions of OPN-5c and OPN-5d as indicated by very faint bands in the media and no visible intracellular bands (Figure 3(B), top and low panels, respectively).

Since mRNA molecules of OPN splice variants with OPN-a as the predominant form are elevated in nonmelanoma skin cancer compared to those of normal skin, we anticipated that OPN protein expression is also increased in cSCCs and BCCs. Western blot analyses were performed for selected specimens with sufficient tissue. Three major forms of OPN protein (45, 52, and 62 kDa) were highly expressed in cSCCs compared with normal skin (Figure 4). In BCCs, except for the 62 kDa protein, the 45 and 52 kDa OPN proteins were not elevated compared with normal skin.

PHOTO (COLOR): Figure 4. Protein expression of OPN in normal skin, cSCCs, and BCCs. (A) Western blot analysis and (B) histogram of densitometric analysis of OPN at 45, 52, and 62 kDa normalized to β–actin. MW: relative molecular weight of protein standards. Normal, n = 5; cSCCs, n = 4; BCCs, n = 4. kDa: kilodalton.

To evaluate and compare the different amounts of OPN splice variant transcripts expressed in normal skin and nonmelanoma skin cancer, we resolved to use the standard curve method for qPCR as opposed to the relative qPCR method reported by others. We believe that determining the absolute numbers of mRNA molecules expressed by each of the OPN splice variants better reflect subsequent data generated from RNA sequencing (RNA-seq) analyses as they become more cost-effective to perform. RNA-seq directly measures transcript abundance and therefore aligns with our determination of the absolute numbers of mRNA molecules. In contrast, relative qPCR only shows the relative fold differences between two groups of transcripts.

The numbers of mRNA molecules of OPN-a are the predominant splice variant expressed in both normal skin and nonmelanoma skin cancer, and its average expressions in cSCCs and BCCs were 4-fold higher than in normal skin. Additionally, even though there were fewer mRNA molecules of OPN-b and OPN-c than of OPN-a, they were both elevated in cSCCs and BCCs relative to those of normal skin. This suggested the importance of cSCCs and BCCs expressing more active and/or different splicing factors compared to normal skin and, consequently, contributing to an increase in alternatively spliced variants of SPP1 transcripts.

In addition, OPN-5 mRNA molecules were markedly expressed in normal skin and nonmelanoma skin cancers. qPCR analysis indicated subvariants of OPN-5, implying that OPN-5 and its subvariants likely have functional importance in skin biology and pathology. We identified and cloned four new alternative splice subvariants of OPN-5a. We designated OPN-5a as the transcript encoding the full-length protein and its subvariant transcripts as OPN-5b to OPN-5e. Therefore, OPN-5 (in Figure 1(B)), consisting of both OPN-5b and OPN-5c, showed elevated levels in normal skin compared to OPN-b and OPN-c, and their levels in cSCCs and BCCS were higher than that of OPN-c, indicating that their presence in normal skin and nonmelanoma skin cancer may be more relevant in its function than OPN-c.

At the protein level, only OPN-5a, OPN-5b, and OPN-5e RNAs are protein-coding variants. Our preliminary studies indicate that they are expressed as secreted proteins and that OPN-5b containing the truncated "E4" exon (Figure 3(A)) is also expressed intracellularly. In contrast to the three protein-coding variants, the E4 introgenic-retaining transcript variants, OPN-5c and OPN-5d, showed minimal expression of secreted protein. Additional studies are necessary to determine the functional significance of OPN-5a and its subvariants at both the transcript and protein levels in normal skin and nonmelanoma skin cancer. The current dogma on alternative splice variants relates to the increase of diverse protein functions. However, there is evidence suggesting that non-coding spliced RNA and even bifunctional coding RNA contribute to regulatory functions on gene expression ([26]).

Apart from the finding of increased alternative splicing of OPN transcripts in cSCCs and BCCS compared to normal skin, the markedly elevated expression of OPN-a in cSCCs suggests its potential role in contributing to human cSCC development. This is supported by our studies with the photocarcinogenesis model indicating that chronic UVB-induction of mouse OPN (homologous to OPN-a) expression for 44 weeks facilitates the development of cSCCs in wild-type female mice; no cSCC was evident in OPN-ablated mice ([9]). Based on results from short-term UVB-induced cell apoptosis and UVB-induced epidermal hyperplasia experiments, we hypothesized that, instead of promoting cell proliferation, induced OPN secreted into the microenvironment facilitates the survival of normal basal keratinocytes and, concomitantly, prolongs the survival of initiated cells, which subsequently transform to a malignant phenotype ([9]).

The elevated mRNA expression of OPN-a and, to a lesser extent, of OPN-b and OPN-c in cSCCs and BCCs over that of normal skin and of OPN-5 over that of OPN-c in nonmelanoma skin cancers, reflect the increase of OPN protein in cSCCs and BCCs previously found by immunohistochemical analyses ([20]) and by the present Western blot data. Immunohistochemical analyses indicated that, compared to normal skin with minimal sun exposure, cSCCs, and actinic keratoses, precursors of cSCCs, express elevated OPN protein. In contrast, OPN is minimally expressed in solid or undifferentiated basal cell epitheliomas, but it is elevated in differentiated forms of basal cell carcinomas (BCCs), such as keratotic BCCs ([20]). The present Western blot data validated the increase of OPN protein expressed in cSCCs and BCCs. Additionally, the multiple molecular weights of OPN protein were expected, as OPN synthesis undergoes posttranslational modification resulting in phosphorylation, glycosylation, and sulfation prior to secretion. Further, OPN can be cleaved by metalloproteinases, thrombin, plasmin, cathepsin-D, and proprotein convertase 5/6 ([[27]]). Notably, proprotein convertase 5/6 is expressed in the epidermis ([30]), and MMP9 is expressed in nonmelanoma skin cancers ([31]). We speculate that the three forms of OPN protein in normal skin, cSCCs, and BCCs are largely derived from OPN-a transcripts since its number of mRNA molecules is substantially greater than those for the other OPN splice variants with the 62 and 45 kDa bands as the post-translationally modified and pre-secreted or non-post-translationally modified forms of OPN, respectively. Information is not available from Biolegend as to which epitope of OPN the monoclonal antibody binds to. Whether any of the multiple forms of OPN protein are derived from alternatively spliced coding RNA of OPN will require the preparation of new monoclonal antibodies to specific epitopes that can distinguish each of these OPN protein products.

This work has some limitations. We lack access to specimens of actinic keratoses and to information on the various stages of cSCC and BCC specimens, which precludes us from analyzing the association of spliced variant RNA of OPN to different stages of nonmelanoma skin cancer. Because we did not use the microlaser capture method to select for only epidermal cells in normal skin or malignant cells in cSCC and BCC specimens, it is possible that some of the absolute mRNA molecules of OPN splice variants are contributed, in part, by the minor presence of fibroblasts and/or resident immune cells. However, analyses of cSCC cell lines, consisting largely of homogeneous malignant cells, such as A431 and SCC-13, indicated that OPN-a is the predominant transcript compared to other OPN splice variants, supporting our finding that OPN-a is the major mRNA expressed in cSCCs (Table S2).

In summary, we report the absolute numbers of mRNA molecules of OPN splice variants in normal skin and nonmelanoma skin cancer as opposed to their relative transcript expression. Compared with other OPN splice variants, OPN-a is the major mRNA in skin, with high expressions in cSCCs and BCCs compared to normal skin. Additionally, OPN-5 and its subvariants are markedly expressed in normal skin over that of OPN-b and OPN-c and over that of OPN-c in nonmelanoma skin cancer.

The authors greatly appreciate Dr. Donald Hill's and Ms. Rebecca Lipscomb's assistance in reviewing and editing the manuscript.

PLC conceived the study and drafted the manuscript; CCH provided material support. PLC and CFC designed and analyzed the study; CFC and NBD performed the experiments. All authors reviewed and approved the final version of the manuscript.

The authors declare no conflict of interest.