Summary

A Rapid Method for Multispectral Fluorescence Imaging of Frozen Tissue Sections

Published: March 30, 2020
doi:

Summary

We describe a rapid staining method to perform multispectral imaging on frozen tissues.

Abstract

Multispectral fluorescence imaging on formalin-fixed paraffin-embedded (FFPE) tissues enables the detection of multiple markers in a single tissue sample that can provide information about antigen coexpression and spatial distribution of the markers. However, a lack of suitable antibodies for formalin-fixed tissues may restrict the nature of markers that can be detected. In addition, the staining method is time-consuming. Here we describe a rapid method to perform multispectral fluorescence imaging on frozen tissues. The method includes the fluorophore combinations used, detailed steps for the staining of mouse and human frozen tissues, and the scanning, acquisition, and analysis procedures. For staining analysis, a commercially available semiautomated multispectral fluorescence imaging system is used. Through this method, up to six different markers were stained and detected in a single frozen tissue section. The machine learning analysis software can phenotype cells that can be used for quantitative analysis. The method described here for frozen tissues is useful for the detection of markers that cannot be detected in FFPE tissues or for which antibodies are not available for FFPE tissues.

Introduction

Recent advances in microscopic imaging techniques have significantly improved our knowledge and understanding of biological processes and disease states. In situ detection of proteins in tissues via chromogenic immunohistochemistry (IHC) is routinely performed in pathology. However, detection of multiple markers using chromogenic IHC staining is challenging1 and newer methods to use multiplex immunofluorescence (mIF) staining approaches, wherein multiple biological markers are labeled on a single tissue sample, are being developed. The detection of multiple biological markers is useful, because information related to tissue architecture, spatial distribution of cells, and antigen co-expression are all captured in a single tissue sample2. The use of multispectral fluorescence imaging technology has made detection of multiple biological markers possible. In this technology, using specific optics the fluorescence spectra of each individual fluorophore can be separated or "unmixed", enabling the detection of multiple markers without any spectral crosstalk3. Multispectral fluorescence imaging is becoming a critical approach in cell biology, preclinical drug development, clinical pathology, and tumor immune-profiling4,5,6. Importantly, the spacial distribution of immune cells (specifically CD8 T cells) can serve as a prognostic factor for patients with existing tumors7.

Various approaches to multiplex fluorescence staining have been developed and can be performed either simultaneously or sequentially. In the simultaneous staining method, all the antibodies are added together as a cocktail in a single step to label the tissue. UltraPlex technology uses a cocktail of hapten-conjugated primary antibodies followed by a cocktail of fluorophore-conjugated anti-hapten secondary antibodies. InSituPlex technology8 uses a cocktail of unique DNA-conjugated primary antibodies that are simultaneously added to the tissue followed by an amplification step and finally fluorophore-conjugated probes that are complementary to each unique DNA sequence on the primary antibody. Both of these technologies enable the detection of four markers plus 4’,6-diamino-2-phenylindole (DAPI) for nuclear staining. Two other approaches for simultaneous multiplex staining are based on secondary ion mass spectrometry9. The Hyperion Imaging System uses imaging mass cytometry10 to detect up to 37 markers. This technology uses a cocktail of metal-conjugated antibodies to stain the tissues, and specific areas of the tissues are ablated by a laser and transferred to a mass cytometer where the metal ions are detected. Another similar technology is the IONPath, which uses multiplexed ion beam imaging technology11. This technology uses a modified mass spectrometry instrument and an oxygen ion source instead of laser to ablate the metal-conjugated antibodies. While all these simultaneous multiplex staining approaches enable the detection of multiple markers, the costs involved for conjugating DNA, haptens, or metals to antibodies, the loss of tissue due to ablation, and the extensive image processing for unmixing cannot be underestimated. Moreover, kits and staining protocols are currently available only for FFPE tissues and developing custom panels entails additional time and expenditure.

The sequential multiplex staining method, in contrast, includes labeling the tissue with an antibody to one marker, stripping to remove the antibody, followed by sequential repeats of this process to label multiple markers12. The tyramide signal amplification (TSA) is the most frequently used sequential multiplexing method. Two other multiplexing technologies use a combination of simultaneous and sequential staining methods. The CODEX platform13 employs a cocktail of antibodies conjugated to unique DNA oligonucleotide sequences that are eventually labeled with a fluorophore using an indexed polymerization step followed by imaging, stripping, and repeating the process to detect up to 50 markers. The MultiOmyx multiplex staining approach14 is an iteration of staining with a cocktail of three to four fluorophore-conjugated antibodies, imaging, quenching the fluorophores, and repeating this cycle to detect up to 60 markers on a single section. Similar to the simultaneous multiplex staining method, while a broad range of markers can be detected, the time involved in staining, image acquisition, processing, and analysis is extensive. The stripping/quenching step involves heating and/or bleaching the tissue sample and thus, the sequential multiplex staining approach is commonly performed on FFPE tissues that maintain tissue integrity upon heating or bleaching.

Formalin fixation and subsequent paraffin embedding is readily performed in a clinical setting, tissue blocks are easy to store, and several multiplex staining protocols are available. However, the processing, embedding, and deparaffinization of FFPE tissues, as well as antigen retrieval15, a process by which antibodies can better access epitopes, is time-consuming. Furthermore, the processing involved in FFPE tissues contributes to autofluorescence16 and masks target epitopes, resulting in the variability and lack of antibody clone available to detect antigens in FFPE tissues17,18,19. An example is the human leukocyte antigen (HLA) class I alleles20. In contrast, snap freezing of tissues does not involve extensive processing steps prior to or after fixing, circumventing the need for antigen retrieval21,22, and making it beneficial for detecting a wider range of targets. Therefore, using frozen tissues for multispectral fluorescence imaging can be valuable to detect targets for preclinical and clinical studies.

Given the above mentioned limitations when using FFPE tissues, we asked whether multispectral fluorescence imaging can be performed on frozen tissues. To address this question, we tested a simultaneous multiplex staining method using a panel of fluorophore-conjugated antibodies to detect multiple antigens and analyzed the staining using a semiautomated multispectral imaging system. We were able to simultaneously stain up to six markers in a single tissue section within 90 min.

Protocol

Mouse spleen and HLF16 mouse tumor tissues23 were obtained from our laboratory. Human tonsil tissue was purchased from a commercial vendor. Details are provided in the Table of Materials.

1. Tissue Embedding

  1. Embed fresh tissue in OCT (optimal cutting temperature) solution and snap freeze using either dry ice or liquid nitrogen.
  2. Store tissues at -80 °C.

2. Cryosectioning

  1. Cut 8 μm sections in a cryostat with temperatures set at -25 °C.
    NOTE: The preferred section thickness can be adjusted to generate crisper images.
  2. Place sections on charged glass slides.
  3. Air-dry the sections for 1 h at room temperature (RT) prior to fixing in histology grade ice-cold acetone for 10 min.
    NOTE: Acetone causes coagulation of water-soluble proteins and extracts lipids but does not impact carbohydrate-containing components. In contrast, formalin preserves most lipids and has little impact on carbohydrates24. The choice of fixative is important depending on the choice of marker being detected.
  4. Store the slides at -20 °C.

3. Selection of Antibodies and Fluorophores

NOTE: Before tissue staining, antibody clones that will robustly and specifically stain their antigens of interest within sequential sections from acetone fixed tissue must be validated. Some antibodies may require a different fixative, and their compatibility with other antibodies in the panel will also have to be empirically determined. The goal is also to identify fluorophores with minimal overlap that can be detected with the epifluorescence filters for DAPI, FITC, Cy3, Texas Red, and Cy5.

  1. Confirm staining by conventional IHC or immunofluorescence (IF) detection in tissue sections with known expression target antigen.
  2. Using the excitation and emission filter sets available on the semiautomated imaging system and after testing various combinations of fluorophore-conjugated primary antibodies, prepare fluorophores to be used that have minimal spectral overlap (e.g., see Table 1).

4. Staining

NOTE: The tissue rehydration and slide washes were performed in a Coplin jar. The blocking and antibody incubation steps were performed in a humidified slide box.

  1. Allow the slides to warm to RT for 5–10 min.
  2. Rehydrate in phosphate buffered saline (PBS) for 5 min.
  3. Perform a blocking step prior to staining tissues with antibodies. For mouse sections, use specialized blocking solution (see Table of Materials) for 10 min at RT. For human sections, use 10% normal pooled human serum (NHS) diluted in PBS for 15 min at RT.
    NOTE: Different blocking buffers may be tested as needed to preserve the specific properties of the sections depending on the follow-up procedures to be used.
  4. Wash the slides for 5 min in PBS after blocking.
  5. For multiplex staining, prepare a cocktail of antibodies with compatible fluorophores at predetermined optimal dilutions.
  6. Add the cocktail of fluorophore-conjugated antibodies to the slides. For single-marker staining, add only the primary-conjugated antibody to the slide.
  7. Include a control unstained slide that undergoes the same staining procedure without the addition of any primary-conjugated antibody.
  8. Incubate the slides for 1 h at RT in the dark and then wash the slides 2x with PBS for 5 min each. From here on, the resulting slide is referred to as multiplex-stained.
  9. To counterstain, add DAPI to the multiplex-stained slide, incubate for 7 min in the dark at RT, and wash the slides 2x with PBS for 5 min each. Do not counterstain single-stained and unstained slides.
  10. To coverslip, add a drop of the mounting medium, and gently place the glass coverslip over the tissue.

5. Preparing a Spectral Library

  1. Image acquisition
    1. Set the lamp power to 100%. Usually the power is set to 10% because fluorescence detection on FFPE tissues includes a signal amplification step.
    2. Begin by opening the microscope operating software (see Table of Materials).
    3. Select Edit Protocol and then New Protocol.
    4. Provide a "Protocol Name", and select Fluorescence under Imaging Mode, and provide a "Study Name".
    5. Place the single-stained slides on the stage and examine each marker in its corresponding fluorescence channel to ensure staining. Choose a region on the tissue expressing the strongest signal for the marker.
    6. Adjust the exposure times using the Autofocus and Autoexpose options.
    7. Acquire snapshots for the single-stained and unstained slides and save the protocol.
      NOTE: The following steps are performed in machine learning software (see Table of Materials), using the single stained and unstained slides to verify specific staining as well as to determine antibody cross talk.
    8. Under the Build Libraries tab in the software, load each single-stained slide image, choose the appropriate Fluor, and click Extract. The software will automatically extract the fluorescence signal chosen in the Fluor.
    9. To save the extracted color, click on Save to Store. A "New Group" may be created or the extracted color can be stored to an existing group.
  2. Verifying the spectral library
    1. Check the emission spectral curve window, located to the right of the extracted image, for each filter set.
      NOTE: The extracted signal is correct if the spectral curve is observed only in the filter sets where the fluorophore is detected. If a spectral curve is observed in the wrong filter set, it can mean that either the primary signal expected in the filter set is not strong enough, or the software is detecting another signal that is too high, possibly due to spectral overlap. In this case, first try using the Draw Processing Regions tool to draw regions around areas expressing the fluorescent marker and autofluorescence in the image. This trains the software to detect the true signal and remove any interfering signals. If this does not work, repeat the staining process for the single-stained slide to test different antibody titrations.

6. Multispectral Imaging

NOTE: Once the spectral library is created and verified, perform the following steps for the multiplex-stained slide.

  1. Whole slide scan
    1. Adjust focus and exposure times on the multiplex-stained slide as mentioned under in section 5.1.
    2. Under Scan Slides, create a New Task and choose the protocol saved above.
    3. Perform a whole slide scan on the multiplex-stained slide.
    4. Using the whole slide scan software, open the whole slide scan image. This image has not been spectrally unmixed.
    5. Select regions of interest (ROI) across the whole slide scan image using the Stamp or ROI tool. These ROIs will be scanned using the exposure times set in 6.1.1 to be used for spectral unmixing and analysis.
    6. Click Process Slide to acquire ROIs at 20x magnification
  2. Spectral unmixing
    1. Once the ROIs have been acquired, in the machine learning software, under the Manual Analysis tab, load the multiplex stained images by clicking Open under File.
    2. In the Spectral Library Source dropdown menu click Select Fluors.
    3. A new window will open. Here, choose the spectral library or group created above.
    4. Load the unstained slide image. Click the AF ink marker icon located above the selected spectral library and draw a line or region on the unstained slide to identify autofluorescence in the tissue.
    5. Under the Edit Markers and Colors tab, assign names for each marker. Pseudo colors can be assigned at this step.
      NOTE: The color for the Autofluorescence image defaults to DarkSlateGray. Change this to Black.
    6. Click Prepare All.
  3. Verifying spectrally unmixed images
    NOTE:
    When the spectral unmixing step is completed, a composite image consisting of all the colors is created.
    1. Click the Edit the View eye icon. Here, each color in the "Component Display" can be turned off or on to view the staining of each individual marker.
    2. Visually inspect the staining and the morphology of the cells to ensure that there is no overlapping of marker, unless it is biologically relevant. A pathologist can help verify the staining as well.
      NOTE: The staining on the multiplexed slide should be validated by leaving out one fluorophore at a time and reviewing the staining pattern. In addition, the validation will also help to identify strong fluorophores that appear in adjacent spectra due to antibody cross talk or bleed-through.
    3. For Pathology Views, which simulate brightfield images for each fluorescent marker, click the Select a Component Image button. Here, choose a marker to view a simulated brightfield image.

7. Analyzing Multispectral Images via Cell Segmentation and Phenotyping

NOTE: After verifying the spectrally unmixed image, cell segmentation can be performed using the machine learning software, which will provide step-by-step instructions. Tissue segmentation was not performed here. If the panel includes one or more tissue specific marker and especially if the tissue is messy, tissue segmentation should be performed.

  1. Select "Cytoplasm" and "Membrane" under the 'Segment' option.
    NOTE: "Nuclei" is chosen by default.
  2. Select a marker from the panel. Configure the marker to detect either nuclei, cytoplasm, or membrane. For example, DAPI can be selected to detect nuclei and CD3 for membrane.
  3. Click on the ellipsis button ('…') to select an option under "use this signal to find". For example, 'nuclei' for DAPI. Multiple markers from the panel can be selected for segmentation.
    NOTE: For cytoplasmic or membrane markers, select the "Use this signal to assist in nuclear segmentation" option.
  4. The software automatically detects and creates a mask each for the nucleus, cytoplasm, and membrane in the image.
  5. Ensure all the cells are 'masked' for segmentation. To adjust, switch to the Pathology View for the specific marker chosen in 7.3. and use the configuration options in the software.
  6. Click "Segment All" to segment cells.
  7. After cell segmentation, proceed with phenotyping cells. In this step, choose the markers needed for phenotyping and manually select at least five cells that are brightly stained with the chosen marker. This trains the software to then automatically detect all cells stained with the chosen marker in the image.
  8. A phenotype map is created. Analyze to ensure that the cell stained with the marker is correctly phenotyped.
    NOTE: Cell phenotyping can be an iterative process. If the software is unable to phenotype the cells correctly, it means that the training is inadequate or incorrect. In this case, the user has to manually select more cells and retrain the software and repeat this step until the user is satisfied with the training.
  9. Create a group named "Others" and include cells that are not stained for any of the markers.
    NOTE: This step is important to train the software to exclude all the unstained cells from phenotyping.

8. Exporting Images and Analysis Tables

  1. Click the Export button to view the Export Settings panel.
  2. In the "Export Directory", click Navegar to select a location to export the images.
  3. In the "Image Export Options", choose the Image Output Format.
  4. In the "Images to Export" list, select the images to be exported. "Composite image" is the final pseudocolored unmixed image. "Pathology Views" are the individual simulated brightfield images and the "Component Images (multi-image TIFF)" is a multi-image TIFF file of component data that can be used by third party analysis software.
  5. Click "Export for All" button to export the images.
    NOTE: Tables from analysis can also be selected and exported at this step.

Representative Results

Detection of single-stained markers on frozen spleen sections
As the semiautomated imaging system uses a liquid crystal tunable filter (LCTF) system that allows for a wider range of wavelength detection25, and because no signal amplification steps were performed here, we first optimized the detection of our primary-conjugated antibodies for each marker on the microscope. An example is shown in Figure 1, where each single-stained marker is pseudo-colored red. The Alexa Fluor conjugated antibodies used here have been validated by the companies for immunofluorescence and flow cytometry. However, the Per-CP-Cy5.5 fluorophore is only validated for flow cytometry. We were able to detect this color in our single-stained slides, supporting the suitability of flow cytometry validated antibodies for use in multispectral imaging and providing the benefit of using such antibodies to validate observations using two different techniques (i.e., flow cytometry and multispectral imaging). We then proceeded to perform multispectral imaging on frozen tissues.

Multicolor fluorescence detection on frozen mouse spleen
To test the multiplex staining method and spectral imaging, we used mouse frozen spleen tissue, which has an abundance of immune cells. Figure 2 shows a spectrally unmixed image of different markers in a section of frozen mouse spleen. The antibodies, clones, and concentrations used for staining are described in Table 2. The captured image shows a portion of the T cell zone identified by the presence of CD3, CD4, and CD8 markers, surrounded by myeloid cells expressing CD11b mostly observed in the marginal zone. Regions of proliferating cells expressing Ki67 are observed primarily in germinal centers. The "Pathology Views" option for each marker showed its individual staining pattern within the tissue. Distinct staining patterns for CD11b, Ki67 and the CD3, CD4, and CD8 markers together were observed, suggesting that the multiplex staining method and spectral imaging worked on the frozen tissue.

Multi-color fluorescence detection on frozen human tonsil
After testing a staining panel suited for mouse tissue, we next assessed a separate panel for frozen human tonsil tissue. Figure 3 shows a spectrally unmixed image of the different markers used. The antibodies, clones, and concentrations used for staining are described in Table 2. The captured image shows a follicular germinal center26 expressing B-cells identified by the CD20 marker. Proliferating cells in the follicular germinal center were identified by Ki67. Some of these proliferating cells costained with CD20 and may be centroblasts27, naïve B-cells that undergo active somatic hypermutation. The follicular germinal centers were surrounded by the interfollicular T cell region identified by the expression of CD3, CD4, and CD8. Again, distinct staining patterns were observed here, confirming that the methodology worked on a frozen tissue.

Application of multispectral fluorescence imaging on frozen mouse tumor tissue
Multispectral imaging is a useful tool for monitoring immune cell infiltration in tumors as a prognosis for immunotherapies. To this end, we set out to stain a frozen mouse tumor tissue sample and detect immune cell infiltrates. The HLF16 cell line is used as a human papillomavirus (HPV)+ tumor model of cervical cancer in HLA-A*0201 transgenic mice28. The transgenic cell line was developed by transfecting heart lung fibroblasts from HLA-A*0201 with HPV16 E6 and E7 oncogenes and H-Ras V1228. T cells and tumor associated macrophages are the most common immune infiltrates present in tumors29. Figure 4 shows a spectrally unmixed image on a frozen HLF16 tumor section along with the pathology views; the individual staining patterns are shown in Supplementary Figure 1. The antibodies, clones, and concentrations used for staining are described in Table 2. The captured image shows regions of tumor-infiltrating T cells identified by the CD3 and CD8 markers along with the presence of other myeloid cell lineages identified by the CD11b marker. The captured region also shows tumor associated macrophages (TAMs), possibly of the M2 phenotype, detected by the CD206 marker30, which is closely associated to several proliferating cells detected as Ki67+.

The machine learning software2 comes with features like tissue and cell analyses. These analyses are commonly performed on FFPE tissues stained with signal amplification. As our methodology does not use signal amplification, we wanted to test if the software could be used to analyze staining on frozen tissues. Using the adaptive feature of the software, we were able to segment cells based on a nuclear and membrane marker and to phenotype cells based on the markers used for staining. Figure 5 shows the cell segmentation and phenotype maps and the number of stained cells for each marker analyzed by the software, demonstrating that the software can be used for quantification of multispectral staining on frozen tissues.

Figure 1
Figure 1: Detecting primary-conjugated antibodies using a liquid crystal tunable filter microscope. Primary-conjugated antibodies to the indicated markers were stained on a frozen mouse spleen and detected using the Vectra 3.0 multispectral imaging system under 20x objectives. CD3 on Alexa Fluor 488, CD8 on Alexa Fluor 594, CD11b on Per-CP Cy5.5, CD206 on Alexa Fluor 647, and Ki67 on Alexa Fluor 555 were used. Each marker is pseudo-colored red. No counterstain was used on these slides. Scale bar = 20 μm. Please click here to view a larger version of this figure.

Figure 2
Figure 2: Multispectral imaging on a frozen mouse spleen. (A) Whole slide scan of the multiplexed stained slide taken under 4x objectives. A 2 x 2 stamp across different regions of the tissue were chosen for multispectral imaging. Scale bar = 100 μm. (B) Composite image taken under a 20x objective after spectral unmixing. The pseudo-colored markers are indicated and a magnified image within the red box is shown next to the image. Scale bar = 20 μm. (C) Pathology views for each individual marker. A magnified image for each marker within the red box is shown next to the image. Scale bar = 20 μm. Please click here to view a larger version of this figure.

Figure 3
Figure 3: Multispectral imaging on a frozen human tonsil. (A) Whole slide scan of the multiplexed stained slide taken under a 4x objective. A 1 x 1 stamp (669 μm x 500 μm) across different regions of the tissue was chosen for multispectral imaging. Scale bar = 100 μm. (B) Composite image after spectral unmixing taken under a 20x objective. The pseudo-colored markers are indicated and a magnified image within the red box is shown next to the image. Scale bar = 20 μm. (C) Pathology views for each individual marker. A magnified image for each marker within the red box is shown next to the image. Scale bar = 20 μm. Please click here to view a larger version of this figure.

Figure 4
Figure 4: Multispectral imaging on a frozen mouse tumor. Multispectral imaging was performed on a 1 x 1 region of a frozen mouse HLF16 tumor taken under a 20x objective. The composite image (above). The pseudo-colored markers are indicated and the pathology views (below) for each individual marker is depicted. A magnified image for each marker within the red box is shown next to the original image. Scale bar = 20 μm Please click here to view a larger version of this figure.

Figure 5
Figure 5: Analysis of a frozen mouse tumor section. (A) Cell segmentation map. Adaptive cell segmentation using inForm was performed. The software was trained to identify nuclei (green) and membrane (red). (B) Phenotype maps for each stained marker. The software was trained to identify phenotypes based on the staining. Colored dot represents the indicated marker. "Other" (black) refers to cells that were not stained for the indicated marker. (C) Bar plot (mean ± STDEV) depicting the number of stained cells for each marker in two multispectral images. The number indicates cells phenotyped in each image but does not indicate if the markers are coexpressed. Please click here to view a larger version of this figure.

Fluorophore Excitation maximum (nm) Emission maximum (nm) Expected Detection in Filter Set (name)
Alexa Fluor 488 488 519 FITC
Alexa Fluor 555 555 580 Cy3 and Texas Red
Alexa Fluor 594 590 617 Texas Red
Alexa Fluor 647 650 668 Texas Red and Cy5
DAPI 350 470 DAPI
PerCP-Cy 5.5 482 690 Cy3, Texas Red and Cy5

Table 1: List of fluorophores with their maximum excitation and emission wavelengths and expected detection in appropriate filter sets.

Frozen Mouse Spleen
Antibody/Dye Clone Concentration (μg/mL)
Alexa Fluor 594 anti-mouse CD8a 53-6.7 10
Alexa Fluor 488 anti-mouse CD3 17A2 20
Alexa Fluor 647 anti-mouse CD4 GK1.5 10
PerCP-Cy 5.5 Rat Anti-CD11b M1/70 2
Alexa Fluor 555 Mouse anti-Ki-67 B56 10
DAPI 0.1
Frozen Mouse Tumor
Antibody/Dye Clone Concentration (μg/mL)
Alexa Fluor 594 anti-mouse CD8a 53-6.7 20
Alexa Fluor 488 anti-mouse CD3 17A2 20
Alexa Fluor 647 anti-mouse CD206 (MMR) C068C2 5
PerCP-Cy5.5 Rat Anti-CD11b M1/70 1
Alexa Fluor 555 Mouse anti-Ki-67 B56 0.25
DAPI 0.1
Frozen Human Tonsil
Antibody/Dye Clone Concentration (μg/mL)
Alexa Fluor 594 anti-human CD3 UCHT1 10
PerCP/Cyanine5.5 anti-human CD4 RPA-T4 4
Alexa Fluor 647 anti-human CD8a C8/144B 10
Alexa Fluor 488, eBioscience anti-human CD20 L26 10
Alexa Fluor 555 Mouse anti-Ki-67 B56 1
DAPI 0.1

Table 2: List of antibodies, clones, and concentrations used.

Supplementary Figure 1: Individual staining patterns of the frozen HLF16 tumor after spectral unmixing. Scale bar = 20 μm Please click here to download this figure.

Discussion

Frozen tissues have extensively been used for mIF imaging to traditionally detect three to four markers31 on a tissue using the direct and indirect method32. In the direct method, antibodies are conjugated to fluorescing dyes or quantum dots33 to label the tissue, whereas in the indirect method, an unconjugated primary antibody is used to label the tissue followed by a fluorophore-conjugated secondary antibody that specifically recognizes the primary antibody. Some of the recent simultaneous multiplex staining approaches discussed earlier can also be used to stain frozen tissues and detect more than four markers. But the cost for the reagents and the time taken for staining become intensive depending on the number of markers being detected. Another multiplex technique for frozen tissues is the multi-epitope-ligand-cartography (MELC)34. The technique involves staining the sample with fluorophore-conjugated antibodies, imaging, and photobleaching of the fluorophore. A major caveat of this technique is that only one field or region of the tissue can be multiplexed. In order to multiplex other fields or regions the technique needs to be performed manually, which is time-consuming, or requires automation, which is costly. One group35 was able to detect six colors on frozen tissues using a combination of direct, indirect, and TSA staining. However, the tissue staining took 2 days. In comparison, our methodology uses a simultaneous multiplex staining method involving the application of a cocktail of five directly conjugated antibodies plus DAPI to stain frozen tissues within 90 min. Furthermore, using a semiautomated multispectral fluorescence imaging system, we were able to spectrally separate and detect six markers in frozen spleen, tonsil, and tumor tissues using this simplified multiplex staining technique. The Cy3, Texas Red, and the Cy5 filter sets available on the microscope provide opportunities for detecting additional fluorophores, thereby potentially increasing the number of markers that can be detected simultaneously in frozen tissues.

Multiplex staining using the TSA approach on FFPE tissues requires an antigen retrieval step followed by sequential labeling, washing, and stripping steps. Performing the procedure ranges from 1–2 days depending on the incubation times used for the antibodies6. Recently, using a microfluidic tissue processor, a four-plex staining using the TSA approach was performed on FFPE tissues under 90 min. Other multiplex techniques for FFPE tissues can be performed under 4–5 h using an automated stainer. However, the appropriate antigen retrieval steps need to be optimized to ensure epitope availability for the antibodies36, which in turn relies on the careful consideration of the antibody clones as well as the order of markers being detected. For example, some antibody clones of CD3 cannot be used, because subsequently CD4 and CD8 antibodies are not detected37. Similarly, there is variability of antigen detection among the available antibody clones17,18. Therefore, multiplex staining of FFPE tissues requires optimization of antibody clones, their concentrations, and the order in which they are stained, all of which is time-consuming. In contrast, the lack of extensive tissue processing of frozen tissues allows for the use of various antibody clones with high specificity. Moreover, antibody clones available for frozen tissues can also be used in flow cytometry and ELISA, allowing simultaneous validation across various assays. Tissue architecture is also a concern in frozen tissues38. The multiplex staining shown here on frozen tissues is significantly faster than the TSA approach. The fluorophore combinations require a careful selection of markers to ensure that antibodies do not sterically hinder each other, especially when different antigens expressed in the same cellular location are detected. We chose markers that stain different cells on fluorophores that are spectrally separate, enabling better detection. The antibody concentrations also require optimization, but because the staining protocol is quick, the overall time taken for optimization is not time-consuming. A caveat to our method may be the inability to detect low expressing markers in tissues. Some of the multiplex staining approaches on FFPE tissues involve a signal amplification step that is useful to detect low expressing markers. However, the use of secondary and tertiary antibodies can be employed to boost the signal. In such cases, cross reactivity to antibodies in the panel should be avoided to prevent incorrect interpretation of the results.

In addition to mouse spleen and human tonsil tissues, which are rich in immune cells, we have used tumor tissue as an example for the proposed multispectral fluorescence imaging of frozen tissues. Multispectral fluorescence imaging in tumors has provided valuable information, such as characterizing the tumor microenvironment (TME)39, predicting the success of adoptive T cell transfers in malignant melanoma40, and characterizing proteins in tumor signal transduction pathways41. Using markers specific to immune cells commonly found in tumors, we were able to detect infiltrating CD8 T cells and TAMs in the HLF16 tumor tissue used in this study. We used the machine learning software to successfully segment and phenotype cells. Multiplex staining for FFPE tissues employs a signal amplification step to enhance signal-to-noise (SNR) ratio that helps image processing and quantification42,43. The machine learning software has been used for analyses on FFPE tissues stained with signal amplification2. The methodology present here does not use signal amplification, but we were able to successfully segment and phenotype cells using the software. For further analysis (e.g., phenotype coexpression and spatial relationships) and accurate interpretation of the quantitation, the software training requires validation using a training set, test set, and validation set. In an iterative process, one set of images (i.e., the training set) is used to train the machine learning software to identify the phenotypes until the model’s predictions for a separate set of images (i.e., the test set) are accurate. After training is complete, a final set-aside batch of images (i.e., the validation set) is analyzed to see if there has been over-fitting.

In conclusion, the methodology presented here is a rapid way of performing multispectral fluorescence imaging using frozen tissues. The method is useful to detect markers to which either antibodies are not available or cannot be detected in FFPE tissues. In conjunction with the machine learning software, the time taken to perform quantitative analyses can significantly facilitate rapid preclinical and clinical diagnoses and may be applied in the field of high-resolution spatial transcriptomics that utilizes frozen tissues44.

Divulgaciones

The authors have nothing to disclose.

Acknowledgements

Imaging and analysis guidance was provided by the Research Resources Center – Research Histology and Tissue Imaging Core at the University of Illinois at Chicago established with the support from the office of the Vice Chancellor for Research. The work was supported by NIH/NCI RO1CA191317 to CLP, by NIH/NIAMS (SBDRC grant 1P30AR075049-01) to Dr. A. Paller, and by support of the Robert H. Lurie Comprehensive Cancer Center to the Immunotherapy Assessment Core at Northwestern University.

Materials

Acetone (histological grade) Fisher Scientific A16F-1GAL Fixing tissues
Alexa Fluor 488 anti-mouse CD3 BioLegend 100212 Clone – 17A2; primary conjugated antibody
Alexa Fluor 488, eBioscience anti-human CD20 ThermoFisher Scientific 53-0202-82 Clone – L26; primary conjugated antibody
Alexa Fluor 555 Mouse anti-Ki-67 BD Biosciences 558617 Primary conjugated antibody
Alexa Fluor 594 anti-human CD3 BioLegend 300446 Clone – UCHT1; primary conjugated antibody
Alexa Fluor 594 anti-mouse CD8a BioLegend 100758 Clone – 53-6.7; primary conjugated antibody
Alexa Fluor 647 anti-human CD8a BioLegend 372906 Clone – C8/144B; primary conjugated antibody
Alexa Fluor 647 anti-mouse CD206 (MMR) BioLegend 141711 Clone – C068C2; primary conjugated antibody
Alexa Fluor 647 anti-mouse CD4 Antibody BioLegend 100426 Clone – GK1.5; primary conjugated antibody
C57BL/6 Mouse Charles River Laboratories 27 Mouse frozen tissues used for multispectral training
Coplin Jar Sigma Aldrich S6016-6EA Rehydrating and washing slides
DAPI Solution BD Biosciences 564907 Nucleic Acid stain
Diamond White Glass Charged Slides DOT Scientific DW7590W Adhering tissue sections
Dulbecco's Phosphate Buffered Saline 1x (without Ca and Mg) Fisher Scientific MT21031CV Washing and diluent
Gold Seal Cover Slips ThermoFisher Scientific 3306 Protecting stained tissues
Human Normal Tonsil OCT frozen tissue block AMSBio AMS6023 Human frozen tissue used for multispectral staining
Human Serum 1X Gemini Bio-Products 100-512 Blocking and diluent for human tissues
inForm Akoya Biosciences Version 2.4.1 Machine learning software
PerCP/Cyanine5.5 anti-human CD4 BioLegend 300529 Clone – RPA-T4; primary conjugated antibody
PerCP-Cy 5.5 Rat Anti-CD11b BD Biosciences 550993 Clone – M1/70; primary conjugated antibody
Phenochart Akoya Biosciences Version 1.0.8 Whole slide scan software
ProLong Diamond Antifade Mountant ThermoFisher Scientific P36965 Mounting medium
Research Cryostat Leica Biosystems CM3050 S Sectioning tissues
Superblock 1X ThermoFisher Scientific 37515 Blocking mouse tissues
Tissue-Tek O.C.T Solution Sakura Finetek 4583 Embedding tissues
Vectra 3.0 Automated Quantitative Pathology Imaging System, 6 Slide Akoya Biosciences CLS142568 Semi-automated multispectral imaging system
Vectra Software Akoya Biosciences Version 3.0.5 Software to operate microscope

Referencias

  1. van der Loos, C. M. Chromogens in Multiple Immunohistochemical Staining Used for Visual Assessment and Spectral Imaging: The Colorful Future. Journal of Histotechnology. 33 (1), 31-40 (2010).
  2. Stack, E. C., Wang, C., Roman, K. A., Hoyt, C. C. Multiplexed immunohistochemistry, imaging, and quantitation: A review, with an assessment of Tyramide signal amplification, multispectral imaging and multiplex analysis. Methods. 70 (1), 46-58 (2014).
  3. Bian, L., et al. Multispectral imaging using a single bucket detector. Scientific Reports. 6, 24752 (2016).
  4. Zhou, L., El-Deiry, W. S. Multispectral fluorescence imaging. Journal of Nuclear Medicine. 50 (10), 1563-1566 (2009).
  5. Parra, E. R., et al. Validation of multiplex immunofluorescence panels using multispectral microscopy for immune-profiling of formalin-fixed and paraffin-embedded human tumor tissues. Scientific Reports. 7 (1), 13380 (2017).
  6. Wickenhauser, C., Pico de Coaña, Y., et al. . Immune Checkpoint Blockade: Methods and Protocols. , 13-31 (2019).
  7. Feng, Z., et al. Multispectral imaging of formalin-fixed tissue predicts ability to generate tumor-infiltrating lymphocytes from melanoma. Journal for ImmunoTherapy of Cancer. 3 (1), 47 (2015).
  8. Manesse, M., Patel, K. K., Bobrow, M., Downing, S. R. The InSituPlex((R)) Staining Method for Multiplexed Immunofluorescence Cell Phenotyping and Spatial Profiling of Tumor FFPE Samples. Methods in Molecular Biology. 2055, 585-592 (2020).
  9. Gamble, L. J., Anderton, C. R. Secondary Ion Mass Spectrometry Imaging of Tissues, Cells, and Microbial Systems. Microscopy Today. 24 (2), 24-31 (2016).
  10. Giesen, C., et al. Highly multiplexed imaging of tumor tissues with subcellular resolution by mass cytometry. Nature Methods. 11 (4), 417-422 (2014).
  11. Angelo, M., et al. Multiplexed ion beam imaging of human breast tumors. Nature Medicine. 20 (4), 436-442 (2014).
  12. Tsujikawa, T., et al. Quantitative Multiplex Immunohistochemistry Reveals Myeloid-Inflamed Tumor-Immune Complexity Associated with Poor Prognosis. Cell Reports. 19 (1), 203-217 (2017).
  13. Goltsev, Y., et al. Deep Profiling of Mouse Splenic Architecture with CODEX Multiplexed Imaging. Cell. 174 (4), 968-981 (2018).
  14. Gerdes, M. J., et al. Highly multiplexed single-cell analysis of formalin-fixed, paraffin-embedded cancer tissue. Proceedings of the National Academy of Sciences of the United States of America. 110 (29), 11982-11987 (2013).
  15. Shi, S. R., Key, M. E., Kalra, K. L. Antigen retrieval in formalin-fixed, paraffin-embedded tissues: an enhancement method for immunohistochemical staining based on microwave oven heating of tissue sections. Journal of Histochemistry & Cytochemistry. 39 (6), 741-748 (1991).
  16. Viegas, M. S., Martins, T. C., Seco, F., do Carmo, A. An improved and cost-effective methodology for the reduction of autofluorescence in direct immunofluorescence studies on formalin-fixed paraffin-embedded tissues. European Journal of Histochemistry. 51 (1), 59-66 (2007).
  17. Sorensen, I. V., et al. Characterization of anti-TIMP-1 monoclonal antibodies for immunohistochemical localization in formalin-fixed, paraffin-embedded tissue. Journal of Histochemistry & Cytochemistry. 54 (10), 1075-1086 (2006).
  18. Parra, E. R., Villalobos, P., Mino, B., Rodriguez-Canales, J. Comparison of Different Antibody Clones for Immunohistochemistry Detection of Programmed Cell Death Ligand 1 (PD-L1) on Non-Small Cell Lung Carcinoma. Applied Immunohistochemistry & Molecular Morphology. 26 (2), 83-93 (2018).
  19. Boger, C., Kalthoff, H., Goodman, S. L., Rocken, C. Validation and comparison of anti-alphavbeta3 and anti-alphavbeta5 rabbit monoclonal versus murine monoclonal antibodies in four different tumor entities. Applied Immunohistochemistry & Molecular Morphology. 21 (6), 553-560 (2013).
  20. Torigoe, T., et al. Establishment of a monoclonal anti-pan HLA class I antibody suitable for immunostaining of formalin-fixed tissue: unusually high frequency of down-regulation in breast cancer tissues. Pathology International. 62 (5), 303-308 (2012).
  21. Dapson, R. W. Macromolecular changes caused by formalin fixation and antigen retrieval. Biotechnic & Histochemistry. 82 (3), 133-140 (2007).
  22. Sompuram, S. R., Vani, K., Hafer, L. J., Bogen, S. A. Antibodies Immunoreactive With Formalin-Fixed Tissue Antigens Recognize Linear Protein Epitopes. American Journal of Clinical Pathology. 125 (1), 82-90 (2006).
  23. Cassetti, M. C., et al. Antitumor efficacy of Venezuelan equine encephalitis virus replicon particles encoding mutated HPV16 E6 and E7 genes. Vaccine. 22 (3-4), 520-527 (2004).
  24. Kiernan, J. A. Histological and Histochemical Methods: Theory and Practice, 3rd Edition. Shock. 12 (6), 479 (1999).
  25. Favreau, P., et al. Thin-film tunable filters for hyperspectral fluorescence microscopy. Journal of Biomedical Optics. 19 (1), 011017 (2014).
  26. van Kempen, M. J., Rijkers, G. T., Van Cauwenberge, P. B. The immune response in adenoids and tonsils. International Archives of Allergy and Immunology. 122 (1), 8-19 (2000).
  27. Klein, U., Dalla-Favera, R. Germinal centres: role in B-cell physiology and malignancy. Nature Reviews Immunology. 8 (1), 22-33 (2008).
  28. Eiben, G. L., et al. Establishment of an HLA-A*0201 Human Papillomavirus Type 16 Tumor Model to Determine the Efficacy of Vaccination Strategies in HLA-A*0201 Transgenic Mice. Investigación sobre el cáncer. 62, 5792-5799 (2002).
  29. Gonzalez, H., Hagerling, C., Werb, Z. Roles of the immune system in cancer: from tumor initiation to metastatic progression. Genes & Development. 32 (19-20), 1267-1284 (2018).
  30. Sica, A., Schioppa, T., Mantovani, A., Allavena, P. Tumour-associated macrophages are a distinct M2 polarised population promoting tumour progression: potential targets of anti-cancer therapy. European Jorunal of Cancer. 42 (6), 717-727 (2006).
  31. Au-Granier, C., et al. Multiplexed Immunofluorescence Analysis and Quantification of Intratumoral PD-1+ Tim-3+ CD8+ T Cells. Journal of Visualized Experiments. (132), e56606 (2018).
  32. Odell, I. D., Cook, D. Immunofluorescence Techniques. Journal of Investigative Dermatology. 133 (1), 1-4 (2013).
  33. Xing, Y., et al. Bioconjugated quantum dots for multiplexed and quantitative immunohistochemistry. Nature Protocols. 2 (5), 1152-1165 (2007).
  34. Schubert, W., et al. Analyzing proteome topology and function by automated multidimensional fluorescence microscopy. Nature Biotechnology. 24 (10), 1270-1278 (2006).
  35. de Vries, N. L., et al. High-dimensional cytometric analysis of colorectal cancer reveals novel mediators of antitumour immunity. Gut. , 03 (2019).
  36. Scalia, C. R., et al. Antigen Masking During Fixation and Embedding, Dissected. The Journal of Histochemistry and Cytochemistry : Official Journal of the Histochemistry Society. 65 (1), 5-20 (2017).
  37. Sorrelle, N., et al. Improved Multiplex Immunohistochemistry for Immune Microenvironment Evaluation of Mouse Formalin-Fixed, Paraffin-Embedded Tissues. Journal of Immunology. 202 (1), 292-299 (2019).
  38. Ackerman, L. V., Ramirez, G. A. The indications for and limitations of frozen section diagnosis; a review of 1269 consecutive frozen section diagnoses. British Journal of Surgery. 46 (198), 336-350 (1959).
  39. Mezheyeuski, A., et al. Multispectral imaging for quantitative and compartment-specific immune infiltrates reveals distinct immune profiles that classify lung cancer patients. The Journal of Pathology. 244 (4), 421-431 (2018).
  40. Feng, Z., et al. Multiparametric immune profiling in HPV- oral squamous cell cancer. JCI insight. 2 (14), 93652 (2017).
  41. Yang, L., Liu, Z., Tan, J., Dong, H., Zhang, X. Multispectral imaging reveals hyper active TGF-β signaling in colorectal cancer. Cancer Biology & Therapy. 19 (2), 105-112 (2018).
  42. Blom, S., et al. Systems pathology by multiplexed immunohistochemistry and whole-slide digital image analysis. Scientific Reports. 7 (1), 15580-15580 (2017).
  43. Roy, S., Axelrod, H. D., Valkenburg, K. C., Amend, S., Pienta, K. J. Optimization of prostate cancer cell detection using multiplex tyramide signal amplification. Journal of Cellular Biochemistry. 120 (4), 4804-4812 (2019).
  44. Vickovic, S., et al. High-density spatial transcriptomics arrays for in situ tissue profiling. bioRxiv. , 563338 (2019).

Play Video

Citar este artículo
Jaishankar, D., Cosgrove, C., Deaton, R. J., Le Poole, I. C. A Rapid Method for Multispectral Fluorescence Imaging of Frozen Tissue Sections. J. Vis. Exp. (157), e60806, doi:10.3791/60806 (2020).

View Video