Imaging the mammary gland and mammary tumours in 3D: optical tissue clearing and immunofluorescence methods

The following reagents were purchased from Sigma Aldrich: neutral buffered formalin (NBF), dimethyl sulfoxide (DMSO), tetrahydrofuran (THF), dichloromethane (DCM), dibenzyl ether (DBE), benzyl alcohol, benzyl benzoate, urea, N,N,N’,N’-tetrakis(2-hydroxypropyl)ethylenediamine, 2,2′,2″-nitrilotriethanol, fructose, α-thioglycerol, D-sorbitol and 4′,6-diamidino-2-phenylindole (DAPI) dilactate. RapiClear-CLARITY Specific (RC-CS) Solution and Mounting Medium and iSpacer imaging chambers were purchased from the SunJin Lab: 2,2′-Azobis[2-(2-imidazolin-2-yl)propane] dihydrochloride was purchased from Wako Pure Chemical Industries. Imaging dishes were purchased from Ibidi (81158). Acrylamide (40%) was purchased from Bio-Rad Laboratories. Sodium dodecyl sulphate (SDS) was purchased from Melford Laboratories. Triton-X100 was purchased from VWR International. The following primary antibodies were used for immunostaining: rabbit anti-α-smooth muscle actin (SMA) (Abcam, ab5694; 1:200-1:300 for 2D and 3D studies), rabbit anti-keratin 5 (BioLegend, 905501; 1:100 (3D)), rat anti-cytokeratin 8 (Developmental Studies Hybridoma Bank, TROMA-I; 1:50 (3D) or 1:150-200 (2D)), rabbit anti-E-cadherin (Cell Signaling, 3195; 1:50 (3D) or 1:200 (2D)), mouse anti-E-cadherin (BD Transduction Laboratories, 610182; 1:300 (2D)), rabbit anti-cleaved caspase 3 (Cell Signaling, 9661S; 1:200 (2D)) and rabbit anti-human epidermal growth factor receptor 2 (HER2) (DAKO, A0485; 1:300 (3D) or 1:500 (2D)). The following Alexa Fluor conjugated secondary antibodies were purchased from Life Technologies and diluted 1:500 (2D and 3D studies) in blocking buffer: goat anti-rat Cy3 (A10522), goat anti-rabbit 647 (A21245) and chicken anti-rabbit 647 (A21443). Anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibody was purchased from DAKO (P0448; 1:500).

Syngeneic mammary tumours were established by orthotopically implanting 5 × 10³ TUBO cells [17], into the abdominal (fourth) mammary gland. This cloned cell line was established from a mammary carcinoma that spontaneously arose in a BALB-neuT mouse and therefore carries the Her-2/neu oncogene driven by the MMTV promoter. Mice were monitored regularly and tumours were harvested before exceeding humane endpoints (approx. 4–5 weeks).

3DISCO was performed as previously described [8]. Solvent immersion times were adjusted for mammary tissue pieces (approx. 10 × 10 × 1 mm) as follows: 50% (v/v) THF in H₂O (40 min), 70% (v/v) THF in H₂O (40 minutes), 80% (v/v) THF in H₂O (40 minutes), 100% (v/v) THF (3 × 40 minutes), and DCM (15 minutes). All solvent immersion steps were performed in glass vials. Although DBE is reported to be a superior optical clearing agent vs. BABB [8], this solvent requires specialised imaging chambers and solvent-resistant adhesive (e.g. dental cement), and can severely damage objectives if the chamber fails. Thus, we used BABB as the final clearing agent for transmission and confocal imaging in this study. BABB was prepared as a mixture of benzyl alcohol and benzyl benzoate (1:2). Immunostaining was performed as per the iDISCO protocol [14], with a methanol pre-treatment. Following the iDISCO immunolabelling protocol, samples were washed in PBS and incubated with DAPI (10 μM) for 2–3 h, cleared using the 3DISCO protocol and imaged the same day.

The A4P0 hydrogel formulation was selected for PACT-based clearing of the mouse mammary gland [11]. A4P0 was prepared to contain acrylamide (4% (v/v)), 2,2′-Azobis[2-(2-imidazolin-2-yl)propane] dihydrochloride (0.25% (w/v)) in PBS. PACT-clearing solution consisted of SDS (8% (w/v)) in distilled water, pH 7.5. Mammary tissue pieces (approx. 10 × 10 × 1 mm) were incubated in A4P0 hydrogel monomer for 4 days at 4 °C and heated to 37 °C in a water bath for 4–6 h. Excess gel was carefully removed from the tissue and samples were immersed in PACT clearing solution for 24 h at room temperature. Samples were immersed in fresh clearing solution, incubated at 37 °C for 4 days (with replenishment every second day), and finally washed with PBS containing triton-X100 (0.1% (w/v)) for 24 h. For immunostaining, samples were blocked in PBS containing triton-X100 (0.5% (w/v)) with goat serum (10% (v/v)) overnight at 4 °C. Primary antibodies were diluted in blocking buffer at 4 °C for 4 days with agitation, tissue was washed (3 × 1 h) in PBS and incubated with Alexa Fluor conjugated secondary antibodies for 2 days. Samples were washed in PBS and incubated with DAPI (10 μM) for 2–3 h. PACT-sRIMS samples were incubated in sRIMS for 4 days or until imaging. sRIMS was prepared by combining sorbitol (70% (w/v)) in 0.02 M phosphate buffer [11, 18]. PACT-RC samples were incubated in Rapiclear CS for 4 h and mounted between two coverslips using RC-CS Mounting Medium and iSpacers for image acquisition and long-term storage.

CUBIC-based tissue clearing was performed as previously described [10], with minor modifications. CUBIC Reagent 1 was prepared as a mixture of urea (25% (w/w)), N,N,N’,N’-tetrakis(2-hydroxypropyl)ethylenediamine (25% (w/w)), triton-X100 (15% (w/w)) in distilled water. CUBIC Reagent 2 was prepared using sucrose (44% (w/w)), urea (22% (w/w)), 2,2′,2″-nitrilotriethanol (9% (w/w)), triton-X100 (0.1% (w/w)) in distilled water. CUBIC Reagent 1A was prepared using urea (10% (w/w))), N,N,N’,N’-tetrakis(2-hydroxypropyl)ethylenediamine (5% (w/w)), triton-X100 (10% (w/w)) and NaCl (25 mM) in distilled water (unpublished, protocol available at http://​cubic.​riken.​jp/​). Tissue pieces (approx. 10 × 10 × 1 mm) were immersed in CUBIC Reagent 1 or 1A at 37 °C for 2–3 days, depending on the size of the tissue piece. For immunostaining samples were washed and subsequently blocked in PBS containing triton-X100 (0.5% (w/v)) and goat serum (10% (v/v)) overnight at 4 °C. Primary antibodies were diluted in blocking buffer at 4 °C for 4 days with gentle rocking. Tissue was washed (3 × 1 h) and incubated with Alexa Fluor conjugated secondary antibodies for 2 days, washed in PBS and incubated with DAPI (10 μM) for 2–3 h. Samples were transferred to CUBIC Reagent 2 at 37 °C for at least 1 day for refractive index matching. Samples were immersed in CUBIC Reagent 2 for imaging and were imaged within 1 week. Diffuse, non-specific fluorescence was observed using the CUBIC protocol in the absence of the primary antibodies (Additional file 1: Figure S1, top panel).

Excised and fixed mammary glands were immersed in reagent 1 for 2–3 days. Glands were removed and stained with methyl green (0.5% (w/v)), Harris haematoxylin (10%) or carmine for 1.5–2 h at room temperature with gentle agitation. After staining, tissues were rinsed twice in tap water and once in distilled water before de-staining with acid alcohol (50% ethanol with hydrochloric acid (25 mM)) for 20 minutes and immersion in reagent 2. For detection of β-glucosidase expression (magenta histochemical stain) [19], mammary glands were excised and fixed for 4 h at room temperature. Endogenous β-glucosidase activity was heat inactivated at 65 °C for 15 minutes in PBS. Whole mammary glands were incubated for 48 h at 50 °C in a solution containing 1 part Solution A (5-Bromo-6-chloro-3-indolyl- β-D-glucopyranoside (1% (w/v)) in DMSO) and 25 parts solution B (magnesium chloride (0.02% (w/v)), potassium ferricyanide (0.096% (w/v)) and potassium ferrocyanide (0.13% (w/v)) in PBS). Mammary glands were post-fixed in 10% NBF overnight at 4 °C and cleared using the standard CUBIC-clearing protocol.

For whole-mount immunohistochemical analysis, samples were dehydrated by a methanol series and incubated overnight in methanol containing DMSO (20%) and H₂O₂ (3%) to quench endogenous peroxidase activity [20]. Samples were rehydrated by methanol series and blocked and permeabilised in PBS containing BSA (10% (w/v)) and triton-X100 (1% (w/v)). Samples were incubated with rabbit anti-SMA antibody (1:200) for 4 days at 4 °C with gentle agitation. After washing, samples were incubated with anti-rabbit HRP-conjugated secondary antibody (1:500) for a further 2 days, before immersion in reagent 2. Alternatively, quenching, blocking and antibody steps can be performed after immersion in reagent 1 with a similar outcome.

SeeDB-based clearing was performed as previously described [9], with minor modifications. Briefly, mammary tissue pieces (approx. 10 × 10 × 1 mm) were blocked and permeabilised overnight at 4 °C in PBS with triton-X100 (1% (w/v)) and BSA (10% (w/v)). Primary antibodies were diluted in blocking buffer at 4 °C for 4 days with gentle rocking. Tissue was washed (3 × 1 h) and incubated with secondary antibodies for 2 days before further washing in PBS and incubation with DAPI (10 μM) for 2–3 h. Samples were serially incubated for 8–16 h (twice daily changes) in 2–3 mL of 20%, 40%, 60% and 80% (w/v) fructose in distilled water, and subsequently transferred to 100% (w/v) fructose solution (24 h) and 115% (w/v) fructose solution for 24 h or until imaging. All fructose solutions contained α-thioglycerol (0.5% (v/v)) to inhibit the Maillard reaction [9, 21] and incubations were performed with gentle agitation. For optimal performance, samples were imaged within 2 weeks of clearing; however, staining was still observed up to 6 months after clearing. Diffuse, non-specific fluorescence was observed using the SeeDB protocol in the absence of the primary antibodies (Additional file 1: Figure S1, bottom panel).

Samples were immersed in PBS for approx. 3 days for passive rehydration and removal of the clearing agent [9]. Standard protocols for paraffin processing and embedding using alcohol and xylene were employed. Paraffin-embedded sections (4–6 μm) were de-waxed in xylene and antigen retrieval was performed by boiling in a pressure cooker in tri-sodium citrate buffer (10 mM, pH 6.0), for 11 minutes [22]. Sections were blocked in goat serum (5% (v/v)) in PBS supplemented with triton-X100 (0.05% (w/v)) for 1 h in a humidified chamber at room temperature. Sections were incubated with primary antibodies overnight at 4 °C. Primary antibodies used were: rat anti-K8, rabbit anti-SMA, rabbit anti-E-cadherin, mouse anti-E-cadherin and rabbit anti-cleaved caspase 3. Alexa Fluor conjugated secondary antibodies were diluted 1:500. Nuclei were counterstained with DAPI (1–5 μM).

Mammary tissue pieces were processed using 3DISCO, PACT-RC, PACT-sRIMS, CUBIC or SeeDB-based tissue clearing protocols, and images were acquired on a dissecting microscope (Leica MZ75) with constant exposure, gain and magnification. For quantification of sample size changes, image thresholding was performed using ImageJ (v1.49p, National Institutes of Health) and the pixel area was measured [9]. Volume changes were calculated as the ratio of the pixel area before and after tissue clearing.

Tissues cleared by 3DISCO, PACT-sRIMS, CUBIC and SeeDB were imaged in their respective RI matching solutions in Ibidi μ-Dishes. PACT-RC-cleared tissues were mounted using iSpacer chambers in RC-CS Mounting Medium. Images were acquired using a Leica TCS SP8 inverted confocal microscope with 10×/0.4 or 20×/0.75 HC PL APO objective lenses. Laser power and gain were adjusted manually to give optimal fluorescence intensity for each fluorophore with minimal photobleaching. Step size and line averaging were kept constant for all main figures (line averaging, 16; step size 1–2), excluding CUBIC and SeeDB depth cueing examples. Imaging depths were recorded from the top of the epithelial structure being imaged (typically 350 μm through the native fat pad for the CUBIC and SeeDB protocols). Image reconstructions were generated using Imaris image management software (v8.0, Bitplane) or ImageJ (v1.50c, National Institutes of Health) [23, 24]. Depth coding was performed using the Temporal Colour Code plugin with the spectrum LUT. De-noising of 3D image sequences was performed in MATLAB (R2014a, The Mathworks Inc.) [25].

Two-photon imaging was performed on a LaVision BioTec TriM Scope II with a 25×/1.05 water dipping lens and an insight DeepSee dual-line laser (tuneable 710–1010 nm and fixed 1040 nm lines), with 810 nm wavelength used to excite DAPI and HER2-AF647. Tiled images, having a 20% overlap, were stitched together using the Grid Collection/Stitching plugin in ImageJ [26].

For LSFM, samples were immunostained and cleared according to the CUBIC protocol [10]. After clearing, samples were embedded within an agarose (1% (w/v) in H₂O) tube, prepared by aspirating agarose (37–38 °C) into a pre-warmed 1-mL syringe in which the syringe neck had been cut off. Mammary tissue strips were quickly placed within the agarose tube using forceps and centred by rolling the syringe between the palms. After setting, the plunger was removed and the entire syringe was submerged in CUBIC reagent 2. Samples were imaged in reagent 2 or glycerol in H₂O (34% (w/w)).

The light sheet system was a home-built modified version of the OpenSPIM system [27]. The microscope was built and operated in the T-SPIM layout, whereby illumination happens from two sides simultaneously by overlapping two individual sheets to allow a more even illumination and to reduce artefacts, such as striping. We used two Olympus 5×/0.15 air lenses to generate the light sheet. The higher refractive indices, long working distance (20 mm) and the fact that the lenses were on threaded mounts allowed us to adjust the point of focus accordingly. The imaging light path was equipped with a Nikon 16×/0.8 water dipping lens. We imaged onto an Andor Neo 5.5 (ANDOR) or a Hamamatsu ORCA-Flash4 V2 (Hamamatsu) with 6.5 um pixels. For excitation a home-built laser combiner was used, bundling 405 nm, 488 nm, 561 nm and 640 (Coherent Cube 405 and 640, Coherent Sapphire 488 and 561) into a single-mode fibre. Channels were acquired sequentially and emission was filtered by suitable band-pass or long-pass filters (DAPI: 447/60; AF647: 705/72; both AHF Analysentechnik). The sample was mounted in a 4D (xyzθ) stage (Picard Industries) allowing optimal positioning of the sample in the light sheet. During imaging the sample was moved through the light sheet with a step size of 1.5 μm and the light sheet thickness was adjusted to be ca. 6 μm to warrant an even thickness of the sheet across the entire sample width. Exposure times were between 15 and 150 ms.