Novel fluorescent staining protocol for thick sections of human osteochondral tissues to facilitate correlation with MRI and CT

Abstract

Objective:

To describe a novel fluorescent histochemical protocol to visualize osteoclasts, vasculature, and nerves in thick sections of human osteochondral tissues and to demonstrate its feasibility for use in radiologic-pathologic research correlation studies.

Materials and Methods:

Consecutive patients scheduled for total knee arthroplasty surgeries underwent pre-operative MRI. CT imaging was performed after tissue collection and abnormal osteochondral regions were sectioned to 1-2 mm in thickness and decalcified. Fluorescent labeling of osteoclasts was performed by staining for tartrate-resistant alkaline phosphatase activity with a fluorescent substrate. Vascular structure was visualized with fluorescently labeled lectin Ulex europaeus Agglutinin I (UEA-I). Immunostaining was performed for proteins including smooth muscle actin expressed in smooth muscle cells surrounding arterioles and fibrotic myofibroblasts, as well as for neuropeptide Y expressed in sympathetic nerves. Sections were then recut at 5 μm and stained with hematoxylin and eosin (H&E).

Results:

Edema-like and cyst-like regions identified with MRI and CT were easily located in fluorescent images and appeared to have increased osteoclast activity. Fibrotic regions were identified with thickened arterioles and increased myofibroblasts. Sympathetic nerve fibers travelled alongside arborizing blood vessels. Stained sections became transparent in a water-based refractive index-matched medium, permitting deep 3D visualization of the elaborate neurovascular network in bone. Sequential staining procedures were successfully performed with the same sections, demonstrating the potential to compare multiple cellular markers from the same locations. Routine H&E staining could be performed after the fluorescent staining protocol.

Conclusion:

We have developed a multimodal framework to facilitate comparisons between histology and clinical MRI and CT.

Introduction

The importance of radiologic-pathologic correlation studies is well known and has been included as an integral component in the field of osteoarthritis research. Through these studies, an enriched understanding of pathoetiology is provided, allowing for more specific and confident diagnoses which ultimately improve patient care and treatment. Radiologic-pathologic correlation studies can be challenging, however, due to the vast differences in resolution between clinical imaging modalities and histology.

With clinical MRI, slices thicknesses are in the 2 to 3 mm range, whereas historically, application of histological techniques has been limited to thin sections (<10 μm thickness) generated from tissues embedded in plastic or paraffin after decalcification. The lengthy processing time for osteochondral tissues is often suboptimal for the preservation of enzymatic activity and antigen integrity. Moreover, direct 3D visualization of the complexity of elaborate structures such as blood vessels and peripheral nerves is not possible on thin sections and structural integrity of the tissue is often compromised during tissue slicing. With recent advances in tissue clearance, it has been possible to visualize endogenous fluorescent proteins and fluorescent immunolabeling in rodent bone without destructive sectioning [1-6]. However, there still remains a paucity of studies which utilize human tissues, which contain a higher density of collagen than several other species [7], are often are more difficult to render transparent [8, 9], and are otherwise problematic for immunostaining or enzymatic staining protocols [10].

Armed with insights gained from the abovementioned studies, we began to incorporate various advantageous aspects of different techniques with our efforts to immunolabel thick sections of human bone tissues. We reasoned that cutting tissues into sections of 1 to 2 mm thicknesses would shorten decalcification time, permitting better staining through increased preservation of enzymatic activity and antigens. Since the abnormal area of interest might also be limited in size, and different staining techniques are not always compatible, it is also desirable to stain sections sequentially, allowing comparisons between multiple biomarkers from an identical anatomic location. To be able to correlate with findings on clinical imaging modalities, it is also important to generate images in dimensions that are comparable to those of MRI and CT with reasonable experimental and imaging time. The purpose of our study is to describe a novel fluorescent histochemical protocol to visualize osteoclasts, vasculature, and nerves in thick sections of human osteochondral tissues and to demonstrate its feasibility for use in radiologic-pathologic research correlation studies.

Materials and Methods

Study population

This study was approved by our Institutional Review Board and all subjects provided signed informed consent. Consecutive participants with clinically advanced knee osteoarthritis scheduled for total knee arthroplasty were enrolled. Patients were excluded if they had contraindications to MRI.

MRI

Pre-operative imaging of the knee was performed within one week of the total knee arthroplasty on a 3T clinical MRI scanner (MR750, GE Healthcare, Milwaukee, WI) using an eight-channel knee coil. MRI protocols incorporated the following sequences: axial fast spin echo (FSE) intermediate-weighted with fat suppression (2,350/30; echo-train length of 8; 3-mm slice thickness; 0.5-mm interslice gap; 320×320 matrix; 15-cm field of view (FOV); and two signal averages), sagittal FSE T2-weighted with fat suppression (4,300/70; echo-train length of 18; 3-mm slice thickness; 0.5-mm interslice gap; 320×256 matrix; 15-cm FOV; and one signal average), sagittal FSE T1-weighted (480/8; echo-train length of 3; 3-mm slice thickness; 0.5-mm interslice gap; 320×320 matrix; 15-cm FOV; and one signal average), coronal FSE T2-weighted with fat suppression (4,500/70, echo-train length of 8, 3-mm slice thickness, 0.5-mm interslice gap, 320×256 matrix, 15-cm FOV, and one signal average), and coronal FSE T1-weighted (700/8; echo-train length of 2; 3-mm slice thickness, 0.5-mm interslice gap, 320×320 matrix, 15-cm FOV, and one signal average).

Tissue collection, CT imaging and analysis

Excised osteochondral tissues from the arthroplasty were collected, immediately fixed with 2% paraformaldehyde/12.5% of saturated picric acid solution in 0.1 M phosphate buffer (pH 7.0) at 4°C for 24 hours, then rinsed with saline. Tissues were placed flat and scanned on a Siemens SOMATOM Force (Siemens Healthcare AG, Erlangen, Germany) at isocenter using the ultra-high-resolution (UHR) mode, 120 kVp, 90 mA, and 0.3 pitch. Images were reconstructed using the Ur77u kernel with Edge Technology with 0.1 mm isotropic voxels. Using a DICOM viewer (RadiAnt 2020.1, Medixant, Poznan, Poland), MRI and CT images were evaluated by a fellowship-trained musculoskeletal radiologist (A.F.L., with 7 years of experience) for bone marrow abnormalities, including edema-like, cyst-like, and sclerotic lesions [11].

Tissue pretreatment and thick sectioning

After CT imaging, tissues were rinsed with saline and soaked in an antifreeze solution containing 30% ethylene glycol, 30% sucrose, and 1% polyvinylpyrrolidone at 4°C for at least three days, then stored at −20°C until further processing. Osteochondral pieces containing abnormalities identified by MRI and CT imaging were cut into 1-2 mm thick sections using a precision sectioning low-speed saw (IsoMet 1000; Buehler, Illinois, USA). Sections were decalcified at 4°C for up to ten days with 10% ethylenediaminetetraacetic acid (pH 7.0), then stored in antifreeze at −20°C until staining.

Fluorescent staining of thick osteochondral sections

Staining of osteoclasts.

To detect tartrate-resistant alkaline phosphatase (TRAP) activity from osteoclasts [12], decalcified sections were washed with saline, then incubated at 4°C overnight with 1 mM each of Ca²⁺, Mg²⁺, and Mn²⁺ in tris(hydroxymethyl)aminomethane buffered saline (TBS) (pH 7.4). Sections were then rinsed with saline, incubated at 4°C for 30 minutes in cold 200 mM acetate, 50 mM tartrate (pH 5.0), then at 4°C for 30 minutes in the same solution but with the addition of 5 μM ELF-97 (ThermoFisher Scientific, Waltham, MA) [12] to allow the substrate to diffuse into tissues. To initiate enzymatic reaction, tissues were brought to room temperature (RT) and incubated for an additional 60 minutes. The enzymatic reaction was stopped with a saline wash.

Staining of endothelial cells of blood vessels.

Sections were washed with TBS (pH 7.4) and incubated at 4°C for 1-2 days with 4 μg/ml Dylight 649-conjugated Ulex europaeus Agglutinin I (UEA-I 649) (Vector Laboratories, Burlingame, CA), and 0.2 μg/ml 4′,6-diamidino-2-phenylindole (DAPI) in TBS supplemented with 2 mM Ca²⁺ and Mg²⁺. Sections were washed with saline and stored in antifreeze at −20°C until imaging.

Immunostaining of thick osteochondral sections

Antigen Retrieval.

After confocal imaging of ELF-97 and UEA-I fluorescence, antigen retrieval was performed by heating sections to 70°C for 60 minutes in 10 mM citrate buffer (pH 6.0), cooling to RT, then rinsing with saline until the next day. This treatment also unbinds and removes UEA-I from the sections.

Delipidation and bleaching.

We used a methanol dehydration-based protocol known as immunolabeling-enabled three-dimensional imaging of solvent-cleared organs (iDISCO+) [13]. Fixed samples were dehydrated in progressively increasing concentrations of methanol diluted in phosphate buffered saline (PBS) of 40%, 80%, and 2x 100% for 60 minutes each, then incubated overnight in a solution of methanol/dichloromethane (1:2). After two 60-minute washes with 100% methanol, samples were bleached with 5% H₂O₂ in methanol (1 volume 30% aqueous H₂O₂/5 volume methanol) at 4°C overnight. Sections were then rehydrated in progressively decreasing concentrations of methanol diluted in PBS: 80%, 40%, 20%, and pure PBS.

Acid treatment of sections.

Successful immunostaining with intact or halved rodent bone tissues was reported with an iDISCO protocol that added a decalcification step using Morse solution (22.5% formic acid/22.5% sodium citrate, pH between 2.0 and 3.0) after delipidation [4]. We reasoned that acid treatment not only decalcified bone tissues, but also degraded proteins, increasing tissue permeability and antigen exposure. Anticipating that this milder acid treatment would still preserve structural integrity and retain protein, we incubated sections at RT for 60 minutes with Morse solution. Tissues were then thoroughly rinsed with PBS before immunostaining.

Staining of sympathetic nerves with neurotransmitter neuropeptide Y (NPY).

Pre-treated samples were incubated in a mixture containing PBS with 0.2% Triton X-100 (PBST), 20% dimethylsulfoxide (DMSO), 0.3M glycine, 3% hydroxyl acetyl-proline, 1 mM β and γ cyclodextrans (pH ≥ 5.0) at 37 °C for 24-72 hours. Hydroxyl acetyl proline was used to loosen collagen and cyclodextrans were used to extract cholesterols [14]. Samples were rinsed with PBS, then blocked in PBST supplemented with 10% DMSO and 6% horse serum at 37 °C for 24 hours. Samples were incubated with rabbit anti-NPY monoclonal antibody (Cell Signaling, Danvers, MA) diluted 1:1,000 in PBS with 0.2% Tween-20 (PBSTw) supplemented with 10 μg/ml heparin, 5% DMSO, and 3% horse serum at 37 °C for 3 days. After five 1-hour washes with PBSTw, sections were incubated with 2 μg/ml Dylight 647 anti-Rabbit secondary antibodies, 10 μg/ml fluorescein-conjugated UEA-I (Vector Laboratories, Burlingame, CA), and 0.2 μg/ml DAPI in PBSTw supplemented with 10 μg/ml heparin and 3% horse serum at 37 °C for 3 days. Sections underwent five final 1-hour PBSTw washes at RT, were soaked in the antifreeze solution for at least one day at 4 °C , then stored at −20 °C until imaging. Some sections were also made transparent with an optical clearing solution (PROTOS [15]) before imaging.

Staining of smooth muscle actin (sma) in arterioles and myofibroblasts.

For some sections, immunostaining for sma was carried out after TRAP staining without delipidation. Briefly, sections were permeabilized and blocked with PBST and 2% bovine serum albumin (BSA) for at least 2 hours, then incubated at 4 °C for 2 days with mouse antibody against sma (Sigma, Saint Louis, MO) diluted 1:2,000 in the blocking solution. Sections underwent four 1-hour PBS washes, then were incubated at 4 °C overnight with Alexa-488 conjugated anti-mouse antibodies (ThermoFisher, Carlsbad, CA) diluted 1:1,000 in PBST plus 2% BSA. Samples underwent four final 1-hour PBST washes before being soaked and stored in antifreeze.

Stripping of antibodies and repeat round of immunostaining.

To test whether previously stained sections could be stained again, immunolabeling was stripped by heating sections in 1% sodium dodecyl sulfate and 20 mM glycine (pH 2) at 50 °C for 60 minutes, then incubating in the same solution at RT overnight to allow unbinding of antibodies. Imaging was performed to confirm stripped immunolabeling from previous rounds, and NPY staining was repeated as described above beginning with blocking in PBST, 10% DMSO, and 6% horse serum.

Fluorescence imaging

Imaging was performed on an inverted confocal laser scanning microscope (LSM880, Zeiss, Germany) using a 10x objective or a fluorescence microscope (BZ-X710, Keyence, Itasca, IL) using a 4x objective. Stained sections were placed in antifreeze on a #1 coverslip glued across a window cut into a plastic dish. The following excitation lasers were used: DAPI, 405 nm; ELF-97, 405 nm; Fluorescein or Alexa 488, 488 nm; Cy 3 or Alexa 568, 568; Dylight 649, 633 nm. Fluorescence was collected based on spectra of fluorophores: DAPI, 415-460 nm; ELF-97, 541-552 nm; Alexa 488, 495-560 nm; Cy3 or Alexa 568, 580-630 nm; Dylight 649, >645 nm. For tissues not treated to transparency, multiple tile scan images along the z-plane were generated to cover a section with area up to 4 cm², depth up to 250 μm, and at resolution of 2.77 x 2.77 μm/pixel. Imaging times per section were typically less than 60 minutes. Images were processed, and stacks of images were projected using Fiji ImageJ (https://imagej.net/Fiji).

Thin resectioning

After imaging of the thick sections, tissue was dehydrated through graded alcohols and Pro-Par xylene substitute (Anatech, Battle Creek MI), and infiltrated with Paraplast wax medium (Leica Microsystems, Buffalo Grove IL). Tissue was embedded in EM-400 wax (Leica Microsystems, Buffalo Grove IL) and sectioned at 5 μm for hematoxylin and eosin (H&E) staining. Stained sections were digitally scanned at 10x (Zeiss Axio Scan Z1, Carl Zeiss Microscopy GmbH, Jena, Germany).

Results

Four participants underwent pre-operative MRI and provided their excised tissues for this study (four males, mean age of 71.0 ± 8.8). All had advanced osteoarthritis with multiple bone marrow abnormalities, including edema-like, cyst-like, and sclerotic lesions. After identification and localization of the abnormalities on MRI and CT, serial sectioning of the areas of interest was performed to produce samples between 1-2 mm in thickness.

Correlation of fluorescent and H&E images with MRI and CT

Figure 1 demonstrates typical results from a 1 mm thick osteochondral section of a cyst-like lesion with surrounding bone marrow edema-like signal in the weightbearing aspect of the medial femoral condyle. In this example the majority of the cyst-like lesion lacked vascularity. The surrounding bone marrow edema-like regions stained positively for TRAP activity, indicative of osteoclasts, and for sma in branching and regional distributions, consistent with arterioles and myofibroblasts in fibrovascular regions. Following fluorescent staining and imaging, this section was embedded in paraffin and sectioned for H&E staining. Among structures identified in fluorescent images, the same cyst-like lesion and sma-labeled arterioles were easily observed, while a small patch of sma-positive myofibroblasts were not pronounced in H&E image of comparable magnification. At higher magnification, eroded bone surface was observed near area with TRAP fluorescent staining. Therefore, fluorescent staining of thick section can be used to study the molecular and cellular changes of structural abnormality identified by MRI and CT, and is compatible with subsequent standard histology.

Fig. 1.

Correlation among MR, CT, confocal, and H&E images in the medial femoral condyle of an 84-year-old man. (a) T2-weighted fat-suppressed MR image shows cyst-like lesions with surrounding bone marrow edema-like signal. (b) Ex vivo CT image corresponding to the boxed area in a confirms the cyst-like lesions. (c) Confocal image in the boxed area in b after staining for TRAP activity (osteoclasts, yellow), UEA-I (vasculature, red), sma (arterioles and fibrotic myofibroblasts, green) and DAPI (nuclei, blue). Autofluorescence from cartilage and trabecular bone is also shown in the blue channel. The cyst-like lesion can clearly be observed on the confocal image (asterisk). (d) H&E staining of a 5 μm paraffin section generated from the thick section shown in c post imaging. (e) Enlarged view of the boxed area in c with only yellow, red, and blue channels shows TRAP activity (yellow) and branching vascular structures (red). (f) The same region as e with only green and blue channels highlights arterioles (thin arrows) and myofibroblasts in a small fibrotic area (arrowhead). (g) Enlarged view of the boxed area in d shows arterioles (thin arrows) corresponding to those in e and f. The inset image shows an enlarged view of a remodeling surface (open arrow), near TRAP activity showed in e.

Figure 2 shows representative results from a 2 mm thick osteochondral section of a cyst-like lesion in another patient with surrounding bone marrow edema-like signal and sclerosis. On confocal images, massive TRAP (i.e., osteoclast) activity can be seen in this region. Sympathetic nerve fibers were observed in the area surrounding, but not within the cyst-like lesion.

Fig. 2.

Bone remodeling and neurovascular invasion in the medial femoral condyle of a 65-year-old man. (a) T2-weighted fat-suppressed MR image shows both marginated and ill-defined altered marrow signal, consistent with a cyst-like lesion, bone marrow edema-like lesion, and sclerosis. (b) Ex vivo CT image corresponding to the boxed area in a confirms the cyst-like lesion and sclerosis. (c and d) Confocal images in the boxed area in b shows massive regional TRAP activity with surrounding sympathetic nerve fibers (right inset in d).

Figure 3 demonstrates results from a 1-mm thick osteochondral section with bone marrow edema-like signal and sclerosis imaged with a fluorescence microscope. Areas with edema-like signal were enriched with vascularity, and extensive sma staining was seen where the thickening of arterioles and presence of myofibroblasts indicated fibrosis. Imaging time for a single planar section covering 4 x 1 cm of tissue was less than 15 minutes using the fluorescence microscope, and about one hour on the confocal microscope when a stack of 10 planes of section was performed covering the same size. These wide FOV fluorescent images could be reliably compared with MRI and CT, facilitating co-localization of diseased regions.

Fig. 3.

Correlation among MR, CT, and fluorescence microscope images in the medial femoral condyle of an 84-year-old man. (a) T2-weighted fat-suppressed MR image shows bone marrow edema-like lesions. (b) Ex vivo CT image of the boxed area in a highlights area of osteosclerosis. (c) Fluorescence microscope image shows that the bone marrow edema-like regions are enriched with vascular structures stained with UEA-I and sma. (d) Enlarged view of boxed area in c shows that sma stains both arterioles, which line endothelial-stained UEA-I regions, and myofibroblasts.

Visualization of the 3D neurovascular network in trabecular bone

Without tissue clearance, it was possible to obtain 4-5 optical sections of 14 μm thickness in the marrow region. Due to uneven sample surface and placement, it often required 10 or more confocal planes along the z-axis to capture the entire large tissue section in one imaging projection. Although the section could be imaged on both sides to improve coverage, most of the center regions of the 1-2 mm thick slices could not be visualized. However, after overnight soaking in PROTOS solution, the stained sections became transparent and could be imaged several hundreds of micrometers deep into the tissue slice, particularly for fluorophores with long wavelength emission. This deeper imaging into the tissue facilitated the avoidance of potential artifacts on the cutting surface. From the many imaged slices, a thick z-stack could be generated to visualize complex structures such as vessels and nerves (Figure 4). This was in stark contrast to what was possible with a single optical section of 14 μm in thickness, which despite being twice as thick as sections for traditional histology, was still unable to capture the elaborate arborization of nerve fibers and vascular structures.

Fig 4.

Optically-cleared 1.5 mm osteochondral section from the anterior chamfer piece (inferior trochlea) from a 67-year-old man. (a) Projection of 34 x 14 μm optical sections along the z-axis shows the elaborate vasculature (UEA-I, green) and associated sympathetic nerves (NPY, red). Inset gross image shows the transparent section wetted with a small amount of clearing solution. (b) Enlarged view of the boxed area in a. (c) On a single 14 μm optical section, the neurovascular network is less well appreciated, highlighting the limitations of thin sectioning.

Repeated staining of samples

Figure 5 shows the results of a 1.5 mm thick osteochondral section that was stained sequentially with TRAP/UEA-I, NPY, and NPY again. The same region can be identified among the three rounds of staining, highlighting the potential to compare multiple biomarkers from identical anatomic locations.

Fig 5.

Repeated staining of the same section from the anterior chamfer piece (inferior trochlea) from a 67-year-old man. (a) Confocal image shows positive TRAP and UEA-I staining with abundant osteochondral channels extending into the subchondral bone plate. (b) Enlarged view of the boxed area in a shows vascular network with scattered osteoclasts. (c) The same region as b after NPY staining shows sympathetic nerves. (d) Confocal image after NPY staining was stripped and re-stained shows persistent strong immunofluorescence. Inset image shows the typical appearance of sympathetic nerves.

Discussion

In this study we have described and employed a novel fluorescent histochemical protocol for use on relatively thick sections of human osteochondral tissues. It is well known that fluorescence imaging, and in particular immunofluorescence, is a powerful technique that can be used to study the localization, relative expression, and activation states of target proteins in biological tissues. Unfortunately, fluorescence imaging of human tissues is often challenging, particularly for sections which contain bone [7-10]. Human tissues often have extended intervals from harvest to fixation (including uncontrolled postmortem intervals for cadaveric studies), uneven fixation due to lack of perfusion, and requirements for long fixation time related to immersion-fixation. Histological evaluations of human osteochondral tissues also require prolonged decalcification for paraffin sectioning or lengthy processing for plastic embedding. While satisfactory routine histological results may be seen, there is often suboptimal preservation of enzymatic activity and protein antigens. It has been documented that even mild EDTA decalcification of mouse bone tissues impaired immunofluorescent detection of certain antigens [16]. Decalcification of human bone tissues requires much longer time or harsher conditions than that of mouse tissues, which could partially explain the subpar results to date for immunostaining of human osteochondral tissues. To compensate for loss of antigens during decalcification, enzyme-linked amplification of immunostaining signal is often employed [17]. We reasoned that shortening decalcification time would minimize antigen loss, making the technique compatible with fixation protocols that work well with mouse bone and other soft tissues. By cutting osteochondral tissues to sections of 1-2 mm in thickness, we greatly shortened the time necessary for EDTA decalcification. Typically, this improved step takes fewer than 10 days, but it could likely be reduced in future applications since complete decalcification may not be required in all cases. Indeed, TRAP activity was well preserved in our preparations. As TRAP activity has also been reported to be preserved well with routine paraffin processing and sectioning of bone tissues [18], it remains to be determined whether our procedure would also work well with more labile enzymes such as proteases.

One challenge associated with immunolabeling of large tissues is antibody penetration. While we were able to obtain excellent staining with UEA-I and antibody against sma with a regular immunostaining procedure, in an effort to obtain the same level of results with NPY, we adopted a modified iDISCO procedure that was previously reported for adult mouse spinal vertebrae [3, 4]. This procedure employed long incubation with antibodies, as well as conditions which would facilitate antibody diffusion, namely DMSO and high temperature. Our results demonstrate that marrow associated structures can be accessible to large molecules such as lectins and antibodies, but further fine tuning may be necessary for different antigen/antibody combinations.

We also demonstrated the feasibility of our protocol to facilitate comparison with images obtained from clinical imaging systems. Pathological areas identified by MRI and CT were easily located on the fluorescent image generated by maximal-intensity projection of stacks of tiled confocal images spanning an entire section with centimeter dimensions. We noted decreased vascularity in a cyst-like lesion, while the neighboring edema-like regions were enriched with vascular structures, had higher bone turnover as indicated by higher TRAP activity, and at times showed fibrosis. These findings are consistent with histological findings previously reported by Zanetti et al [19]. There is interest as to whether injury induced nerve sprouting precedes and leads to angiogenesis and osteogenesis [20, 21] or whether osteoclast activity precedes nerve sprouting [22]. We believe our protocol, when applied to more samples, may be helpful in answering these questions.

Our protocol poses several additional advantages in terms of methodological convenience. First, fixed tissues can be kept at subzero temperatures in an antifreeze solution for many years [23]. We adapted the method to store all osteochondral tissues and cut sections and found that it did not compromise histological structural integrity. Second, stained sections can also be stored in this medium with persistence of fluorescent staining for several weeks. Finally, when stained sections are directly transferred from this antifreeze to a water-based refractive index-matched media (such as PROTOS), they become transparent within a day. These sections can be imaged several hundreds of micrometers from both sides, potentially covering an entire 1-2 mm thick section in depth if necessary with a regular, widely available confocal microscope.

In summary, we have developed a procedure that can be further optimized to provide invaluable information about abnormalities identified by MRI and CT. Our pipeline avoids plastic and paraffin embedding and greatly shortens bone decalcification time (which was permissible at 4°C), all of which optimize preservation of enzymatic activity and antigen integrity. The procedure herein described is also compatible with subsequent immunostaining as well as tissue-clearing protocols. Finally, our procedure allows repeated fluorescent staining of the same sections as well as a final round of routine thin sectioning and H&E staining, making it possible to compare multiple markers at identical anatomic locations. We believe that use of this protocol or variations of it may improve the quality of future radiologic-pathologic research correlation studies focused on osteoarthritis.