Case Selection, Histopathological Analysis, and Overall Survival
Four cases of IMPC and one control of canine mammary gland were selected at the Laboratory of Comparative Pathology of the Institute of Biological Sciences, "Universidade Federal de Minas Gerais" (UFMG), after approval by Animal Experimentation Ethics Committee (CEUA protocol number:362/2016).
For histopathological analysis (grading), primary tumor specimens were fixed in 10% neutral buffered formalin, paraffin-embedded (FFPE), and 4 µm thick histological sections were cut and stained with hematoxylin and eosin. All cases were reviewed and re-classified independently by two pathologists (GDC and TS). In brief, carcinomas with cystic formations containing nests of epithelial cells with a moruliform appearance (infiltrating micropapillary pattern) were diagnosed as IMPC, associated or not with in situ micropapillary areas  (Figure 1). The invasive areas of canine IMPC were graded according to the Nottingham grading system [17, 18]. The overall survival time ranged from 8 to 150 days (median 71 days).
Immunohistochemistry was performed as previously described with minor modifications [7, 19]. Sections (4 μm) of primary tumors were mounted on silanized slides, and a peroxidase-based detection system, Novolink™ Polymer, was applied (Novolink™ Polymer Detection System, Leica Biosystems Newcastle Ltd., Newcastle, UK). The slides were dewaxed in xylene, and endogenous peroxidase activity was blocked with 3% H2O2 in methanol. The reagents were applied manually, and immunoreactivity was visualized by incubating the slides with 3,3'-diaminobenzidine (Lab Vision DAB substrate system; Lab Vision, Fremont, California, USA) for 5 min. The antibodies used were mouse monoclonal anti-cytokeratin (clone 34bE12, Dako, 1:100) and anti CD31, p63, and anti-vimentin, as described above. Negative controls were performed using a normal serum (Lab Vision Ultra V Block) in place of the primary antibody. For all markers, the immunohistochemical analysis was performed on in situ and invasive areas.
Phenotypic markers immunofluorescence and imaging by super-resolution microscopy
Immunofluorescence was performed for aSMA, Vimentin, CD31, Von Willebrand Factor, p63, S100A4, Lamin B1 or B2, and MUC1, as previously described in Rodrigues M.A. et al. (2016) [20–23]. Brief, FFPE tissue sections were dewaxed, rehydrated, and unmasked in trilogy solution (Cell Marque, Koclin, CA, USA) in pressurized heating (125ºC) for 20 minutes according to manufacturer's instructions. Next, samples were rinsed in Phosphate Buffered Saline (PBS, 137 mM NaCl, 2.7 mM KCl and 10 mM phosphate buffer solution, pH 7.4) (Sigma-Aldrich, Carlsbad, CA, USA) and incubated in PBS containing 0.2% Triton X-100 (Sigma-Aldrich) for another 20 minutes, then blocked in PBS containing 1% Bovine Serum Albumin (BSA, Sigma-Aldrich) for 30 minutes. Next, The sections were incubated with a rabbit polyclonal antibody against the nuclear envelope marker Lamin B1 (1:1000, Abcam, Cambridge, MA, USA) or with a mouse monoclonal antibody against Lamin B2 (1:100, Life Technologies) overnight at 4ºC, and with one of the following antibodies: a mouse monoclonal antibody against aSMA (1:100, clone 1A4, Dako, Denmark); a mouse monoclonal antibody against Vimentin (1:100, clone Vim 3B4, Dako, Denmark); a mouse monoclonal antibody against CD31 (1:100, clone JC70A, Dako, Denmark); a rabbit polyclonal antibody against Von Willebrand Factor (1:800, Dako, Denmark); a mouse monoclonal antibody against p63 (1:100, clone 4A4, NeoMarkers, CA, USA); a rabbit polyclonal antibody against S100A4 (1:100, Life Technologies, Carlsbad, CA, USA); and or with a rabbit polyclonal antibody against MUC1 (clone EP1024Y) (1:100, Abcam, USA). They were then rinsed three times for 5 minutes in PBS. Subsequently, sections were incubated with Alexa Fluor® 488 Goat Anti-Rabbit IgG antibody (1:1000, Life Technologies), Alexa Fluor® 555 Goat anti-mouse IgG antibody (1:1000, Life Technologies) and Hoechst 33258 (1 mg/mL, Life Technologies) for 1 hour at room temperature. Next, samples were washed 3x in PBS for 10 minutes and then mounted in Prolong Gold Antifade reagent (Life Technologies). The negative control was included in all reactions by omitting primary antibodies. Images were collected using a Zeiss LSM 880 with Airyscan (Carl Zeiss, Jena, Germany) using an oil 40x 1.3 NA objective lens. Samples were excited at 405 nm and observed at 420-480 nm to detect Hoechst, at 488 nm and observed at 500-525 nm to detect Alexa Fluor 488, at 543 nm and observed using LP 570 nm to detect Alexa Fluor 555 signal. The software Zeiss Efficient Navigation (ZEN) was used for orthogonal projections (XY, XZ, YZ). The fluorescence microscopy results were evaluated in 10 invasive areas of IMPC and were collected ten images from each case (n = 100).
Tissue processing for ultrastructural evaluation
For transmission electron microscopy (TEM), one health and tumor biopsies (it was selected five fields of each) fixed by 10% neutral buffered formalin were cut into approximately 2 mm (length x width) and subsequently post-fixed in 5% glutaraldehyde (biological grade; Electron Microscopy Sciences, Hatfield, PA, USA) in 0.05M phosphate buffer pH 7.3 for 24 h. After that, the fragments were post-fixed in reduced osmium (osmium tetroxide 1% and potassium ferrocyanide in distilled water) for 90 min, dehydrated in ethanol and acetone before embedding in epoxy Araldite resin (Electron Microscopy Sciences, Hatfield, PA, USA). From ultrathin sections of 60 nm of thickness were obtained images using a Tecnai G-12 FEI – 120 Kv microscope. The images were adjusted for resolution, sharpness, and contrast using Adobe Photoshop (Adobe System, Inc, Mountain View, CA, USA).