Moderately differentiated oral tongue squamous cell carcinoma samples from four male and five female patients, aged 30–73 (mean, 58.2) years, were sourced from the Gillies McIndoe Research Institute Tissue Bank for this study, which was approved by the Central Regional Health and Disability Ethics Committee (ref. no 12/CEN/74). Written consent was obtained from all participants.

Hematoxylin and eosin (H&E) staining was used to confirm the presence and appropriate histological grading on 4 μm thick formalin-fixed paraffin-embedded of MDOTSCC sections from nine patients by an anatomical pathologist (HDB). These MDOTSCC sections were then used for DAB IHC staining, as previously described (6, 13), using primary antibodies for cathepsin B (1:1,000; cat# sc-6490-R, Santa Cruz, CA, USA), cathepsin D (1:200; cat# NCL-CDm, Leica, Newcastle upon Tyne, UK), cathepsin G (1:200; cat# sc-33206, Santa Cruz, CA, USA), EMA (ready-to-use, cat# PA0035, Leica), OCT4 (1:30, cat# MRQ-10, Cell Marque, Rocklin, CA, USA), and tryptase (1:300, cat# NCL-MCTRYP-428, Leica). All DAB IHC-stained slides were mounted in Surgipath Micromount (Leica).

Immunofluorescent (IF) IHC staining was performed to determine co-expression of two proteins on two samples of MDOTSCC from the original cohort of nine patients used for DAB IHC staining. Vectafluor Excel anti-mouse 488 (ready-to-use; cat# VEDK2488, Vector Laboratories, Burlingame, CA, USA) and Alexa Fluor anti-rabbit 594 (1:500; cat# A21207, Life Technologies, Carlsbad, CA, USA) were utilized to detect the combinations. All IF IHC-stained slides were mounted in Vecta Shield Hardset mounting medium with 4′,6′-diamino-2-phenylindole (Vector Laboratories).

Positive control tissues used for the primary antibodies were human placenta for cathepsin B; human breast cancer for cathepsin D; mouse bone marrow for cathepsin G; and human seminoma for OCT4 and SALL4. A negative MDOTSCC control sample was prepared for DAB IHC staining by using an IgG isotype control (ready-to-use; cat# IR600, Dako, Santa Clara, CA, USA). For IF IHC staining, a negative control was performed using a section of MDOTSCC tissue with the combined use of primary isotype mouse (ready-to-use; cat# IR750, Dako, Copenhagen, Denmark) and rabbit (read-to-use; cat# IR600, Dako) antibodies.

All antibodies were diluted with Bond primary antibody diluent (cat# AR9352, Leica), and DAB and IF IHC staining was carried out on the Leica Bond Rx autostainer, as previously described (16).

Six snap-frozen samples of MDOTSCC from the original cohort of nine patients used for DAB IHC staining were used for isolation of total mRNA for NanoString nCounter™ Gene Expression Assay (Nanostring Technologies, Seattle, WA, USA), as previously described (6, 13). Probes for the genes encoding for cathepsin B (NM_001908.2), cathepsin D (NM_001909.3), cathepsin G (NM_001911.2), and the housekeeping gene PGK1 (NM_000291.3) were used in the analysis. Raw data were analyzed by nSolver™ software (NanoString Technologies). Results were normalized against the housekeeping gene, graphed using Excel (Microsoft Office 2013), and subjected to t-tests for related samples, to compare the relative abundance of each cathepsin.

4 μm thick formalin-fixed paraffin-embedded sections of six samples of MDOTSCC from the original cohort of nine patients used for DAB IHC staining were used for CISH. Staining was carried out on the Leica Bond Rx auto-stainer and detected using the ViewRNA red stain kit (Affymetrix, Santa Clara, CA, USA), as previously described (16). The probes used for cathepsin B (cat# VA1-12282), cathepsin D (cat# VA1-12281), cathepsin G (cat# VA1-21016), and Bacillus (cat# VF1-11712, Affymetrix) as a negative control. Positive controls used were the same for DAB IHC staining.

Cell counting was performed on six fields of view of the DAB IHC-stained slides of the nine MDOTSCC samples, including cells within the TNs and those within the peri-tumoral stroma, at 400× magnification. Fields of views were selected on regions that exhibited the highest density of staining. The proportion of cells within the TNs and the peri-tumoral stroma stained positively in each field of view was calculated using Excel (Microsoft Office, 2013), and results were subjected to t-tests for related samples, using SPSS v.22 statistical package. Total cell counts were also subjected to χ² statistical analysis, comparing relative abundance of each cathepsin.

Enzyme activity assays for cathepsin B (cat# ab65300, Abcam), cathepsin D (cat# ab65302, Abcam), and cathepsin G (cat# ab126780, Abcam) were performed on three snap-frozen MDOTSCC samples of the cohort of six patients used for NanoString mRNA analysis, according to the manufacturer’s protocol. A snap-frozen tonsil sample was used as an appropriate positive control for the cathepsins B (17) and D (18) assays, and a denatured sample of the same tonsil was used as an appropriate negative control. The capthepsin G assay kit was supplied with its positive and negative controls. The results were obtained using the Variskan Flash plate reader (Thermo Fisher Scientific). Experiments were performed in duplicates with averages taken for each.

3,3-Diaminobenzidine IHC-stained and CISH-stained slides were viewed and imaged on Olympus BX53 light microscope (Olympus). IF IHC-stained slides were viewed and imaged using Olympus FV1200 confocal laser-scanning microscope and processed with cellSens Dimension 1.11 software using 2D deconvolution algorithm (Olympus).

Hematoxylin and eosin staining of 4 μm-thick, formalin-fixed, paraffin-embedded sections of all nine MDOTSCC samples confirm the presence and appropriate histological grading. 3,3-Diaminobenzidine IHC staining demonstrated cytoplasmic expression of cathepsin B (Figure 1A, brown) by cells predominantly within the TNs (Figure 1A, brown, arrows) and cells within the peri-tumoral stroma (Figure 1A, brown, arrowheads). Granular cytoplasmic staining of cathepsin D (Figure 1B, brown) was localized to cells within the TNs (Figure 1B, brown, arrows) and some cells within the peri-tumoral stroma (Figure 1B, brown, arrowheads). Cathepsin G was expressed in the cytoplasm of only some cells within the peri-tumoral stroma (Figure 1C, brown, arrowheads).

Positive controls for cathepsins B (Figure S1A in Supplementary Material, brown), D (Figure S1B in Supplementary Material, brown), and G (Figure S1C in Supplementary Material, brown) demonstrated expected staining patterns in human placenta (19) and breast cancer (20), and mouse bone marrow (21), respectively. The negative control showed minimal staining (Figure S1D in Supplementary Material, brown).

To localize cathepsins B, D, and G in relation to the CSC sub-populations, IF IHC staining was performed on two representative MDOTSCC samples from the original cohort of nine patients used for DAB IHC staining.

The EMA⁺ cells (Figures 2A–C, green) within the TNs (6) expressed both cathepsin B (Figure 2A, red) and cathepsin D (Figure 2B, red), with no expression of cathepsin G (Figure 2C, red). Consistent with the results of DAB IHC staining, IF IHC staining demonstrated immunoreactivity (IR) for cathepsin B (Figure 2A, red, arrows) and cathepsin D (Figure 2B, red, arrows) by cells within the peri-tumoral stroma. The CSC subpopulation within the peri-tumoral stroma of MDOTSCC which expresses OCT4 (6) (Figures 2D–F, green) demonstrated IR for cathepsin B (Figure 2D, red), but not cathepsin D (Figure 2E, red) or cathepsin G (Figure 2F, red).

To further characterize the cathepsin D⁺ (Figure 2G, red) and cathepsin G⁺ (Figure 2H, red) cells, we performed co-staining with tryptase, as we have reported in infantile hemangioma (14). This confirmed that tryptase (Figures 2G,H, green) was not expressed by the cathepsin D⁺ (Figure 2G, red) cells, but was expressed by the cathepsin G⁺ (Figure 2H, red) cells.

Images illustrating the individual stains demonstrated in Figure 2 are presented in Figure S2 in Supplementary Material. Minimal staining was present on the negative control (Figure S2Q in Supplementary Material), confirming the specificity of the primary antibodies used.

NanoString mRNA analysis for cathepsins B, D, and G was normalized against the housekeeping gene, PGK1, confirming transcriptional activation for both cathepsins B and D in all six MDOTSCC samples, whereas cathepsin G was detected at low levels in four of the six samples studied (Figure 3).

Statistical analysis of the gene transcripts confirmed significantly greater presence of cathepsin B than cathepsin D (t = 3.073, p < 0.05), and that both cathepsins B and D were significantly more abundant than cathepsin G (t = 4.701 and t = 4.885, respectively, p < 0.01).

Colorimetric in situ hybridization confirmed the presence of mRNA for cathepsin B (Figure 4A, pink, arrows), cathepsin D (Figure 4B, pink, arrows), and cathepsin G (Figure 4C, pink, arrows) in cells within all six MDOTSCC samples examined.

Appropriate staining was seen on positive controls for cathepsin B (Figure S3A in Supplementary Material, pink, arrows), cathepsin D (Figure S3B in Supplementary Material, pink, arrows), and cathepsin G (Figure S3C in Supplementary Material, pink, arrows), indicating the presence of mRNA transcripts in the controls. The negative control performed on each run showed the absence of mRNA transcripts for any marker (Figure S3D in Supplementary Material).

Cell counting for cathepsins B, D, and G (Figure 5) in all nine samples demonstrated significantly more cells staining positively for cathepsin B within the TNs, than those within the peri-tumoral stroma (91% vs. 77%, t = 10.281, p < 0.000). Analysis using χ² method showed that there were significantly more cathepsin B⁺ cells than cathepsin D⁺ cells (χ² = 409.9.261, p < 0.0001), and that there was significantly more cathepsin D⁺ cells relative to the cathepsin G⁺ cells (χ² = 3356.0, p < 0.0001). Consequently, we conclude that there was significantly more cathepsin B than cathepsin G.

To determine the functionality of the cathepsins B, D, and G within MDOTSCC, we performed enzymatic activity assays, which confirmed the functional activity for cathepsin B (Figure 6A) and cathepsin D (Figure 6B), but not cathepsin G (Figure 6C) within all three MDOTSCC samples, as compared with the expected activity for the positive and negative controls (Figures 6A–C).