Several commercially available anti-SNAI1 antibodies were evaluated for their applicability to paraffin-IHC (P-IHC). Further specificity testing then focused on SC10432 (goat polyclonal, [E-18]; St. Cruz Biotechnology, Santa Cruz, CA) and AF3639 (goat polyclonal; R&D Systems, Minneapolis, MN). Xenografts of two previously studied cell lines SiHa and ME180 obtained from the American Type Culture Collection (ATCC, Manassas, VA) were employed as controls for all staining procedures . Although AE1/AE3-positive, both cell lines represent opposite ends of the epithelial-mesenchymal spectrum with an E-cadherin(-), p63(-), SNAI1(+), FAK(+) phenotype in SiHa and an E-cadherin(+), p63(+), SNAI1(-), FAK(+/-) phenotype in ME180 (Additional File 1: A). In addition, three different FFPE tissue samples were used for initial antibody testing: (i) 1st trimester human placenta, (ii) SiHa xenograft tissue derived from SCID mice after s.c. suspension injection and (iii) tissue from a spindle cell carcinoma specimen of the piriform sinus with vimentin positivity and foci of osteosarcoma. To validate the nuclear SNAI1 staining seen with those tissues two tests were performed. First, SC10432 was tested in all specimens after prior incubation with the corresponding blocking peptide SC10432P at 1:10 protein ratio versus PBS (Additional File 1: B). Second, immunoblotting was performed on SiHa cell lysates treated with SNAI1 siRNA versus control. Immunoblots revealed a singular band at ~30 kDa with a clear decrease after treatment (Additional File 1: C). By IHC, an essentially identical nuclear reactivity was found for both SC10432 and AF3639. However, AF3639 in our hands produced less cytoplasmic staining and a stronger nuclear signal than SC10432 (Additional File 1: D).
Clinical FFPE material was cut at 4 μm thickness and processed with standard procedures. Microwave antigen retrieval was performed in 10 mM Citrate buffer, pH 6.0, for 10 min. The following antibodies and conditions were applied: SC10432 (1:200, overnight) and AF3639 (1:1000, overnight), rabbit polyclonal FAK (#3285, 1:200, overnight; Cell Signaling Technology, Danvers, MA), mouse monoclonal HMWK 34βE12 (M0630, 1:100, 1 hr; Dako, Mississauga, ON), mouse monoclonal Vimentin (Vim3B4, 1:200, 1 hr; American Research Products, Belmont, MA), mouse monoclonal E-cadherin (36B5, 1:100, overnight; Vector Labs, Burlington, ON), mouse monoclonal p63(7JUL, 1:50, 2 hrs; Vector Labs) and human cytokeratin cocktail (AE1/AE3, 1:200, 1 hr; Dako). Secondary antibodies were biotin-labeled IgG (Vector Labs). Linking reagents were Streptavidin-HRP (ID Labs, London, ON) and/or Streptavidin-AP (Dako). Labeling reagents were NovaRed (SK-4800, Vector Labs), DAB+AP double-labeling: DAB (K3466, Dako) & AP VectorRed (SK-5100, Vector Labs). Sections were counterstained with Gill modified hematoxylin and coverslipped with Permount* Mounting Medium (Fisher Scientific, Nepean, ON). First-trimester placenta and xenograft tissue were carried with every staining series as controls.
Scoring and Statistical Methods
IHC scoring was done by a pathologist-in-training (JS) and a board-certified pathologist (WRG). Scoring was performed on full sections to account for rare and localized events such as EMT at the tumor invasion front. A histology scoring system (H-score) was used to account for area and staining intensity since considerable heterogeneity was observed in individual sections:
E-cadherin, FAK, p63:
Intensity Category 0
Intensity Category 1 × Area of Category 1 (%) +
Intensity Category 2 × Area of Category 2 (%) +
Intensity Category 3 × Area of Category 3 (%) = H-Score
FAK scoring was done for the predominant cytoplasmic compartment; however occasional nuclear staining was also seen. E-cadherin was scored with intensity 2 and 3 for weak and strong membrane staining, respectively. Cytoplasmic staining of any intensity without membrane labeling was considered intensity 1 . P63 was scored for nuclear staining only, and no significant labeling of any other cell compartments was seen. SNAI1 staining was categorized as follows:
Category 0: no tumor cells with nuclear staining
Category 1: rare tumor cells with nuclear staining, less than 5%
Category 2: ≥ 5% tumor cells with nuclear staining (= SNAI1-positive)
SNAI1 in tumor-surrounding stroma was assessed using the following categories: absent (no SNAI1 even at 400×), rare/occasional cells (single cells can be found at 400×), frequent (SNAI1 positive stroma cells are readily detectable), abundant (positivity of a majority of stroma cells easily discernable at 100×).
Statistical analysis of categorical data was performed using Fisher's exact test. Spearman nonparametric correlation was used to assess the relationship between variables. Kaplan-Meier survival data were compared using the log-rank test. Two-tailed p values < 0.05 were considered statistically significant. Statistical software was GraphPad Prism, Version 5.01 (GraphPad Software, La Jolla, CA).