Multiplexed barcoding image analysis has recently improved the characterization of the tumor microenvironment, permitting comprehensive studies of cell composition, functional state, and cell-cell interactions. Herein, we describe a staining and imaging protocol using the barcoding of oligonucleotide-conjugated antibodies and cycle imaging, which allows for the use of a high-dimensional image analysis technique.
Multiplexed imaging technology using antibody barcoding with oligonucleotides, which sequentially detects multiple epitopes in the same tissue section, is an effective methodology for tumor evaluation that improves the understanding of the tumor microenvironment. The visualization of protein expression in formalin-fixed, paraffin-embedded tissues is achieved when a specific fluorophore is annealed to an antibody-bound barcode via complementary oligonucleotides and then sample imaging is performed; indeed, this method allows for the use of customizable panels of more than 40 antibodies in a single tissue staining reaction. This method is compatible with fresh frozen tissue, formalin-fixed, paraffin-embedded tissue, cultured cells, and peripheral blood mononuclear cells, meaning that researchers can use this technology to view a variety of sample types at single-cell resolution. This method starts with a manual staining and fixing protocol, and all the antibody barcodes are applied using an antibody cocktail. The staining fluidics instrument is fully automated and performs iterative cycles of labeling, imaging, and removing spectrally distinct fluorophores until all the biomarkers have been imaged using a standard fluorescence microscope. The images are then collected and compiled across all the imaging cycles to achieve single-cell resolution for all the markers. The single-step staining and gentle fluorophore removal not only allow for highly multiplexed biomarker analysis but also preserve the sample for additional downstream analysis if desired (e.g., hematoxylin and eosin staining). Furthermore, the image analysis software enables image processing-drift compensation, background subtraction, cell segmentation, and clustering-as well as the visualization and analysis of the images and cell phenotypes for the generation of spatial network maps. In summary, this technology employs a computerized microfluidics system and fluorescence microscope to iteratively hybridize, image, and strip fluorescently labeled DNA probes that are complementary to tissue-bound, oligonucleotide-conjugated antibodies.
The tumor microenvironment (TME) is extremely heterogeneous, consisting of tumor cells, tumor stromal cells, immune cells, noncellular components of the extracellular matrix, and numerous abundant molecules produced and released by tumor, stromal, and immune cells1,2. Accumulating evidence demonstrates that the TME has a pivotal role in reprogramming tumor differentiation, growth, invasion, metastasis, and response to therapies3.
Understanding how different cell types in the TME interact and communicate with each other through signaling networks is essential for improving cancer diagnosis, optimizing immunotherapy, and developing new treatments4. Traditional tissue microscopy techniques, including immunohistochemistry (IHC) and immunofluorescence (IF), have been used for decades to study the cell types, abundance, and communications in tumor samples. Unfortunately, these techniques typically can evaluate only one or two protein markers in a tissue section and cannot reveal the complex spatial and structural relationships among these cells5,8,7.
Over the past two decades, several multiplexed imaging technologies have been established8. These technologies provide much improved views of the composition, function, and location of immune cells within the TME, leading to rapid advancements in the ability to identify and spatially profile complex TMEs at the single-cell level9,10. The spatial and structural relationships of various tumor and immune cells in the TME are now at the forefront of biological and clinical studies using these multiplexed imaging technologies11,12.
The recently developed multiplexed imaging technology using oligonucleotide-conjugated antibody barcoding is an influential single-cell biological research platform based on the detection of oligonucleotide-conjugated antibodies in formalin-fixed, paraffin-embedded (FFPE) samples13,14. Currently, this multiplexed imaging technology allows for the simultaneous imaging of more than 100 markers in a single tissue section15, which has increased the number of cell types that are distinguishable in situ. This enables a level of spatial analysis of tumor and immune cells that is not possible using traditional immunophenotyping approaches16.
Herein, we describe an optimized protocol for conjugating purified antibodies to oligonucleotides and validating this conjugation using the multiplexed imaging platform and a multicycle imaging procedure with FFPE tissue. In addition, we describe the basic image processing and data analysis procedures used with this technology.
This retrospective study was approved by the Institutional Review Board of the University of Texas MD Anderson Cancer Center. The FFPE tissue samples were collected from patients at MD Anderson as part of routine standard care. No diagnostic or therapeutic interventions were performed. Informed consent was obtained from the patients for using the samples collected for research and publication.
1. Antibody sources used for the antibody panel design
2. Before antibody conjugation
3. Antibody conjugation
4. Conjugation confirmation
NOTE: Before performing staining experiments with a user-conjugated antibody using the multiplexed imaging technology, the conjugation should be confirmed by using gel electrophoresis with 5 μL of the conjugated antibody (see step 3.2.6) along with 1 μg of an unconjugated antibody (usually in 2 μL of the mixture) as a control. A successful antibody conjugation will be demonstrated by an increase in the molecular weight of the antibody. However, this confirmation protocol only assesses the success of the chemical reaction for conjugation and does not address the antibody validation used for multiplexed imaging.
5. Oligonucleotide-conjugated antibody staining
6. Multiplexed imaging reporter plate
NOTE: A 96-well plate, referred to as a reporter plate, containing barcoded fluorophores in individual wells is prepared according to custom-designed multiplexed imaging experiments and correlates with each stained coverslip sample. The following steps are for the preparation of the reporter plate.
7. Calibrating and running the multiplexed imaging machine
NOTE: The high-resolution imaging fluorescence microscope captures four different fluorescence channels in each multiplexed imaging cycle at 20x, 100% excitation light, and with low photobleaching.
8. Image collection
NOTE: Multiplexed images can be collected using any adapted inverted fluorescence microscope configured with four fluorescence channels (DAPI, Cy3, Cy5, and Cy7) and equipped with a Plan Fluor 20x lens. Imaging and washing of the coverslip samples are iteratively performed automatically using a specially developed fluidics setup. The images are acquired using Processor software (v1.8.0.7) in QPTIFF format.
9. Image analysis
NOTE: The acquired images can be uploaded to a patented automated image analysis software or open-source software program (Figure 5) for downstream analysis.
We employed FFPE tonsil samples to develop a 26 marker immune oncology panel to illustrate the immune status of FFPE tissue using a barcoding image analysis system. Overall, 19 antibodies are currently used in other multiplexed imaging studies in our lab. All of the markers have been tested using FFPE tissue with chromogenic IHC. All the antibodies were conjugated to unique DNA oligonucleotides. When setting up the coverslips using the web-based instrument manager (Figure 4) for this barcoding image analysis technology, it should be noted that the first and last cycles are always "blank" (Figure 2 and Figure 4), which provides the background fluorescence signals to be subtracted from the specific signals from the antibodies. After images in QPTIFF format have been collected with the fluorescence microscope, they can be visualized using several patented automated image analysis software or open-source software programs. Composite images can show all the markers or selected markers for a better view of the signals (Figure 5). Moreover, each antibody can be visually evaluated for nuclear, cytoplasmic, or membranous localization. Immune, tumor, and stromal cells can be easily identified. Subsequently, image analysis can provide information on the signal intensity, dynamic range, and spatial distribution of all the markers (Figure 6). This technique allowed us to analyze all 26 markers at the subcellular level in a single tissue section (Figure 7). By analyzing the markers' co-localization, we could identify the cellular phenotypes, localize the spatial cell position, calculate the distance between cells, and find the distribution of the cells. The crucial impact of this technology is the presentation of a robust 26 marker panel focused on the immune status of the tissue microenvironment.
Figure 1: Image of custom-conjugated antibody validation using a Bis-Tris protein gel. Lane 1 of the gel shows the protein standard. Lane 2 and Lane 4 show barcode-conjugated antibodies (arrows). Lane 3 and Lane 5 show the heavy and light chain bands from an unconjugated antibody (arrowheads). Please click here to view a larger version of this figure.
Figure 2: Staining plate configuration map for two coverslip samples. Abbreviations: HB = hydration buffer; B = antibody diluent/block; PSFS = post-staining fixative solution; PBS = phosphate-buffered saline; MeOH = methanol; S = storage solution. Please click here to view a larger version of this figure.
Figure 3: Reporter plate configuration. Each conjugated antibody has a barcode that is complementary to a specific reporter. To set up the reporter plate, every conjugated antibody and its corresponding reporter should be listed. Next, each antibody is assigned a cycle number. The performance of two blank cycles (C1 and C18) is used to evaluate the level of autofluorescence in the three fluorescence channels and for post-imaging background subtraction using the image acquisition control software. A software wizard will check the instrument at this stage to ensure all the settings are correct (Figure 4). Please click here to view a larger version of this figure.
Figure 4: Image acquisition using the control software for the experiment setup. (A) Select the Experiment tab in the left-bottom corner of the control software to prepare and start the setup. (B,C) Select New Template to input the experimental settings with a new project and experiment name. (D) Change the start cycle well and the number of cycles to reflect the reporter location in the 96-well reporter plate. (E) Assign the proper fluorescence channels to the four channels designated for the experimental run. Please click here to view a larger version of this figure.
Figure 5: Image visualization using web-based software (QuPath). The viewer window shows 26 markers in the stained tonsil tissue FFPE section. The Brightness & Contrast window shows the markers with the checkmarks. Finally, the viewer window shows the FFPE sample with the selected markers. Please click here to view a larger version of this figure.
Figure 6: Image visualization using web-based software. (A) Tonsil tissue sections were stained for the 26 markers, and the images of the sections in QPTIFF format were visualized using commercial digital slide viewer software or open-source software (QuPath) for annotation and review. (B–F) Six markers were displayed in the same annotation for a better view of the signals. Please click here to view a larger version of this figure.
Figure 7: Views of the 26 individual markers used with web-based software. The marker expression in the tonsil tissue is shown via immunofluorescence staining with an immune oncology panel (top left). Individual markers in two small areas (red rectangles) are shown. The zoomed-in insert shows cells positive for these markers (white arrows). Please click here to view a larger version of this figure.
Figure 8: Summary of the multiplexed image acquisition workflow. FFPE tissue sections were stained using the 26 antibody panel followed by a multicycle reaction. Raw images of the stained sections were computationally processed, and a cell density and spatial analysis was performed using the composite images. Please click here to view a larger version of this figure.
Supplementary Figure 1: IHC validation in tonsil tissue. FFPE tissue sections were stained using an individual antibody. The marker expression in the tonsil tissue is shown at a low magnification, and the zoomed-in insert shows cells positive for the marker (red rectangles). Please click here to download this File.
The TME plays an essential role in cancer development, progression, and treatment responses. Additionally, the density of specific tumor-infiltrating lymphocyte subsets in the TME can serve as a prognostic biomarker for certain types of cancer. Remarkably, in addition to the TME's cellular composition, the spatial characteristics of a tumor can provide an outline for understanding the tumor's biology and identifying potential prognostic biomarkers12,17.
As numerous immune cell populations are involved in procancer or anticancer responses, a better understanding of these cells and their spatial relationships with each other and with cancer cells will help guide the identification of new immunotherapeutic strategies. Previous studies have stratified the location and spatial distribution of TME cells based on the tissue structure in the intratumoral and peritumoral areas and the invasive margins of the tumor cells18,19. Over the past 15 years, technological advancements have made the phenotypic analysis of individual cells based on their spatial dispersal a novel, influential tool for studying the TME and categorizing potential biomarkers for tumor immunotherapy. Multiplex IF histochemistry can concurrently estimate multiple biological markers20.
Similar to the oligonucleotide-conjugated antibody strategy, four types of protein-based multiplex platforms are used to study the TME: chromogen-, fluorescence-, DNA barcode-, and metal isotope-labeled antibody detection systems. The cost-effective chromogenic IHC platforms enable whole-slide visualization and pathological assessment by using conventional bright-field microscopy. In multiplexed IF and IHC, antibodies conjugated with fluorophores are used. The multiplex IF/IHC platform detects antibodies with high specificity and can quantify targeted antibodies even at the subcellular level6,21. In addition, owing to the nature of chromogens and fluorophores, the use of one antibody panel can capture the expression of up to 10 biomarkers on a single slide. On metal isotope-based platforms, metal-tagged antibodies are used to perform multiplexed imaging with single-cell and spatial resolution, and high sensitivity for individual tissue sections22. Theoretically, these metal-conjugated antibody approaches enable the simultaneous detection of more than 100 biomarkers on a single tissue section. One challenge of the isotope-labeling technique is isobaric interference, which prevents 100% purity of enrichment from being reached23. Moreover, the interference increases as the number of markers increases. DNA-conjugated antibody detection platforms recognize antibodies labeled with unique DNA barcodes. More than 40 biomarkers can be simultaneously captured with high specificity on these platforms6.
Multiplexed imaging is a commercially available DNA barcode-labeled antibody detection platform for applying DNA-conjugated antibodies to a single tissue slide in one step (Figure 8). For the tissue preparation stage, unlike the multiplexed ion beam imaging platform, which requires the use of gold-coated slides obtained from manufacturers, the multiplexed imaging platform requires only regular coverslips or slides coated with 0.1% poly-L-lysine to help the tissue adhere to it and keep tissue intact during the staining and imaging process. The use of tissue sections on coverslips within 4 weeks after sectioning is recommended, as the prolonged storage of unstained slides results in antigenicity reduction. A stained coverslip sample can be maintained in storage buffer at 4 °C for up to 2 weeks without losing its staining signal. No special equipment is required for the storage of the coverslip samples. The multiplexed imaging system has been upgraded to use regular slides instead of coverslips, which enables the staining of larger tissues and easy handling. When using a reduction solution for antibody conjugation (step 3.2.3), the reaction should be limited to no more than 30 min to prevent damage to antibodies. The blocking buffers in step 5.6.6 should be freshly prepared, and the blocking buffers must not be reused.
Compared with chromogen-, fluorescence-, and metal isotope-labeled multiplex antibody detection platforms, the multiplexed imaging technology has certain advantages. For example, more than 60 predesigned antibody panels for multiplexed imaging are commercially available, which helps save time and costs in antibody conjugation and validation, and the number of predesigned antibody panels is growing. These antibodies, which include the carcinoma marker pan-cytokeratin, the melanoma marker SOX10, the vascular marker CD31, the stromal marker SMA, and numerous immune cell markers, are validated and experiment-ready. For antibodies that are not predesigned, the commercially available conjugation kit designed for use with multiplexed imaging is straightforward and user-friendly. Customer-conjugated antibodies are good for 1 year when stored at 4 °C. Additionally, machine warm-up is not required for capturing the images. In this multiplexed imaging technology, the iterative washing, hybridization, and stripping steps in the image acquisition rarely result in decreased marker intensity or degraded tissue morphology5,24,25. Furthermore, composite images are captured in QPTIFF format with a simple three-color fluorescence microscope and can be uploaded and analyzed using third-party digital analysis software. The staining markers can be visualized at single-cell resolution, and cell phenotypes can be characterized via the co-localization of the markers (Figure 6 and Figure 7). The comprehensive analysis of a multiplexed image further reveals the tissue compartments, single-cell marker quantification, and nearest neighbor and proximity data (Figure 8).
A challenge in multiplexed image analysis is cell-type identification. Usually, when more single-object classifiers are applied to an image, more uncommon phenotypes will be annotated. Therefore, using known markers that are not co-expressed in the same classifier and applying only the phenotype-related classifier to the annotation of single cells are recommended. Variations in cell-type annotation will result in substantially different spatial results, such asdifferences in cell spatial distribution and cellular neighborhood analysis26,27.
Multiplexed image analysis has proven to be successful in staining and imaging many sample types, including FFPE tissue, fresh frozen tissue, archived whole slides, and tissue microarrays. Multiplexed images of breast, brain, lung, spleen, kidney, lymph node, and skin tissue sections can be acquired with deep single-cell spatial phenotyping data5,16,25,28.
In the future, more predesigned antibodiesfor multiplexed imaging are expected. Additionally, the development of specific software for multiplexed image analysis is greatly needed. Currently, many commercially available and open-source software programs for Hi-Plex image analysis exist29, but scientists still need help in creating a standard workflow for these analyses30,31. Although the composite images captured using this protocol are compatible with third-party software, this may result in extra costs for the user. Another disadvantage of the multiplexed imaging technology is the signal reduction in nuclear protein detection after iterative washing, hybridization, and stripping with large panels of antibodies. Fortunately, this can be minimized by retrieving the barcoded fluorophores at early cycles when designing the reporter plates. Recently, this platform was upgraded with a new high-speed scanning system, which has dramatically reduced the time to obtain composite images32. Additionally, a new strategy using tyramide-conjugated barcodes has been reported to enhance the oligonucleotide-conjugated antibody barcoding-based imaging. This technology aims to amplify staining signals for which barcode-conjugated antibodies are difficult to obtain33.
The authors have nothing to disclose.
The authors thank Donald R. Norwood of Editing Services, Research Medical Library at MD Anderson for editing this article and the multiplex IF and image analysis laboratory in the Department of Translational Molecular Pathology at MD Anderson. This project was supported in part by the Translational Molecular Pathology-Immunoprofiling laboratory (TMP-IL) Moonshots Platform at the Department of Translational Molecular Pathology, The University of Texas MD Anderson Cancer Center and the NCI Cooperative Agreement U24CA224285 (to the MD Anderson Cancer Center CIMAC).
10x AR9 Buffer | Akoya Biosciences | AR900250ML | |
10x Buffer | Akoya Biosciences | 7000001 | |
16% paraformaldehyde | Thermo Fisher Scientific | 28906 | |
1X Antibody Diluent/Block | Akoya Biosciences | ARD1001EA | |
Antibody Conjugation Kit | Akoya Biosciences | 7000009 | Contains Filter Blocking Solution, Antibody Reduction Solution 1, Antibody Reduction Solution 2, Conjugation Solution, Purification Solution, Antibody Storage Solution |
Assay Reagent | Akoya Biosciences | 7000002 | |
Diethyl pyrocarbonate | Sigma-Aldrich | 40718-25ML | |
Dimethyl sulfoxide | Avantor/VWR | BDH1115-4LP | |
Ethanol, 200 proof | |||
G Blocker V2 | Akoya Biosciences | 240199 | |
Histoclear | Thermo Fisher Scientific | 50-329-50 | |
Methanol | Sigma-Aldrich | 34860-1L-R | |
Milli-Q Integral 10 | Millipore | ZRXQ010WW | |
Niknon Fluorescence microscope | Keyence Corp. of America | BZ-X810 | |
Nuclear Stain | Akoya Biosciences | 7000003 | |
Nuclease-free water | Thermo Fisher Scientific | AM9938 | |
NuPAGE | Thermo Fisher Scientific | NP0008 | |
PBS | Thermo Fisher Scientific | 14190136 | |
PhenoCycler Barcodes/Reporters Combination | Akoya Biosciences | 5450004 (BX025/RX025) | |
5450003 (BX022/RX022) | |||
5450023 (BX002/RX002) | |||
5250002 (BX020/RX020) | |||
2520003 (BX023/RX023) | |||
5250005 (BX029/RX029) | |||
5250007 (BX035/RX035) | |||
5250012 (BX052/RX052) | |||
5550012 (BX030/RX030) | |||
5550015 (BX042/RX042) | |||
5550014 (BX036/RX036) | |||
QuPath | Open-Source | https://qupath.github.io/ | |
SimplyBlue SafeStain | Thermo Fisher Scientific | LC6065 | |
Staining Kit | (Akoya Biosciences | 7000008 | Contains Hydration Buffer, N Blocker, J Blocker, S Blocker, Fixative Reagent, Storage Buffer |