A 40-Marker Panel for High Dimensional Characterization of Cancer Immune Microenvironments by Imaging Mass Cytometry

Multiplex immunophenotyping technologies are indispensable for a deeper understanding of biological systems. Until recently, high-dimensional cellular analyses implied the loss of tissue context as they were mostly performed in single-cell suspensions. The advent of imaging mass cytometry introduced the possibility to simultaneously detect a multitude of cellular markers in tissue sections. This technique can be applied to various tissue sources including snap-frozen and formalin-fixed, paraffin-embedded (FFPE) tissues. However, a number of methodological challenges must be overcome when developing large antibody panels in order to preserve signal intensity and specificity of antigen detection. We report the development of a 40-marker panel for imaging mass cytometry on FFPE tissues with a particular focus on the study of cancer immune microenvironments. It comprises a variety of immune cell markers including lineage and activation markers as well as surrogates of cancer cell states and tissue-specific markers (e.g., stroma, epithelium, vessels) for cellular contextualization within the tissue. Importantly, we developed an optimized workflow for maximum antibody performance by separating antibodies into two distinct incubation steps, at different temperatures and incubation times, shown to significantly improve immunodetection. Furthermore, we provide insight into the antibody validation process and discuss why some antibodies and/or cellular markers are not compatible with the technique. This work is aimed at supporting the implementation of imaging mass cytometry in other laboratories by describing methodological procedures in detail. Furthermore, the panel described here is an excellent immune monitoring tool that can be readily applied in the context of cancer research.


INTRODUCTION
Technologies that support the high dimensional analysis of biological systems are essential in scientific research and have become increasingly relevant in clinical contexts. For instance, the advent of T cell checkpoint blockade therapies for cancer treatment has revitalized the field of cancer immunotherapy but also introduced an urgent need for the discovery of biomarkers that guide patient selection for therapies (1,2). Furthermore, recent works making use of single-cell platforms based on RNA sequencing and mass cytometry have delivered a wealth of data revealing previously unappreciated cell subsets and novel functionalities (3)(4)(5). Nevertheless, most immunophenotyping techniques are held back by the lack of spatial resolution, limitations in the number of targets that can be visualized simultaneously, or cumbersome protocols. Methodologies such as flow cytometry can be employed to analyze multiple markers but are insufficient to chart the vast spectrum of immune cells in an unbiased manner (6). Single-cell mass cytometry overcomes this limitation by currently allowing the simultaneous analysis of ∼40 cellular markers. However, it also lacks spatial information, failing to reveal tissue context and cellular interactions which are extremely relevant in physiological and disease states (7)(8)(9). Conversely, multispectral fluorescence imaging provides spatial context but is limited to few markers and is thus best suited to investigate specific research questions in large cohorts (10,11). The recent introduction of imaging mass cytometry has considerably advanced the potential to simultaneously obtain information on phenotypes, their localization within a tissue, and to map cellular interactions.
Mass cytometry makes use of metal isotopes conjugated to antibodies of interest, in contrast to flow cytometry and immunofluorescence techniques that rely on fluorescent dyes. The metal isotopes are distinguished by mass in a time-offlight mass spectrometer and, thus, the number of markers that can be detected simultaneously is not limited by spectral overlap. Since its discovery in 2009 (12), mass cytometry has been successfully applied for the immunophenotyping of cancer microenvironments. This has accelerated the discovery of new immune cell subsets, the assessment of potential biomarkers and correlation of immune-phenotypical changes to therapeutic outcomes (5,(13)(14)(15). Imaging mass cytometry makes use of a high resolution laser that is coupled to the mass cytometer (16). Successive ablations of small portions of tissue (∼1 µm 2 ) are analyzed by CyTOF (Cytometry Time-Of-Flight) thereby quantifying the presence of metal isotopes per area of tissue. This data is reconstructed into an artificial multilayer image resulting in a broad and comprehensive overview of protein expression in situ. Imaging mass cytometry can be employed for imaging up to 40 markers in different tissue sources (e.g., snap-frozen, FFPE), but the combination of a large number of antibodies in the same experiment raises methodological challenges: (1) The testing and validation of a large number of antibodies is an onerous and labor-intensive process. (2) The choice of tissue source must weigh the availability of antibodies directed against native or denatured antigen conformations. Furthermore, the use of FFPE requires that all antibodies function under the same antigen retrieval conditions. (3) The optimal immunodetection conditions are variable for different antibodies. By combining 40 antibodies into one experiment an optimized workflow must be designed in order to obtain best antibody performance.
We developed a 40 marker panel for the analysis of FFPE tissues by imaging mass cytometry. Next to a large amount of lineage and functional immune cell markers, the panel also contains surrogates of cancer cell states (e.g., proliferation, apoptosis) and structural markers (e.g., epithelium, stroma, vessels) for a comprehensive overview of cancer immune microenvironments but also to investigate cancerimmune cell interactions. Furthermore, we created an optimized immunodetection protocol in which antibodies are split into two incubation steps, thereby reducing the concentration of total antibody per working-solution and employing the optimal incubation time and temperature for each antibody. This work provides the scientific community with a ready-to-use imaging mass cytometry antibody panel and provides a blueprint for laboratories that wish to develop dedicated imaging mass cytometry panels.

Immunohistochemistry
Antibody specificity prior and post metal-conjugation as well as optimal antigen retrieval conditions were assessed by chromogenic immunohistochemistry (IHC). Tissue sections were deparaffinized and rehydrated with xylene and decreasing concentrations of ethanol, respectively, followed by endogenous peroxidase blockade using a 0.3% hydrogen peroxidase/methanol solution (Merck Millipore, Burlington, MA, USA). Sections were boiled in either Tris-EDTA (10 mM/1 mM, pH 9) or citrate (10 mM, pH 6) buffers for antigen retrieval and were allowed to cool down to room temperature. To decrease non-specific antibody binding, tissue sections were blocked with Superblock solution (Thermo Fisher Scientific, Waltham, MA, USA) and incubated overnight at 4 • C with a primary antibody ( Table 1). Following washes in PBS, the tissues were incubated with Polyhorseradish peroxidase solution (Immunologic, Duiven, The Netherlands) for 1 h at room temperature. Antibody binding was detected with DAB+ chromogen (DAKO, Agilent technologies, Santa Clara, Ca, USA) and the sections were counterstained with hematoxylin (Thermo Fisher Scientific).

Antibodies and Metal Conjugation
Carrier-free IgG antibodies (concentrations between 0.5 and 1 mg/mL) were conjugated to purified lanthanide metals (Fluidigm, San Francisco, CA, USA) ( Table 1) using the MaxPar antibody labeling kit and protocol (Fluidigm). After conjugation, all antibodies were eluted in 50 µl W-buffer (Fluidigm) and 50 µl antibody stabilizer (Candor Bioscience, Wangen im Allgäu, Germany) supplemented with 0.05% sodium azide. To conjugate anti-CD45 (clone D9M8I) to 89 Y, Yttrium(III) chloride (Sigma-aldrich, Saint Louis, MO, USA) was dissolved *These dilutions were applied in already diluted stock solutions as described in the methods section. **CD39 is detected indirectly with a Qdot800 secondary antibody.
in L-buffer (Fluidigm) to 1 M. Five microliters of a 50 mM working solution were used for conjugation as described in the MaxPar antibody labeling protocol. Conjugation of anti-Vimentin and anti-Keratin antibodies to 194 Pt and 198 Pt (Fluidigm), respectively, was performed as described previously by Mei et al. (17). In order to exclude that the labeling process substantially affected the performance of the antibodies, these were tested by IHC and immunodetection patterns were compared to their non-conjugated counterparts. Stock solutions of antibodies with a strong signal were further diluted in antibody stabilizer supplemented with 0.05% sodium azide. Antibodies were stored at 4 • C and remained stable for at least 6 months.

Imaging Mass Cytometry Acquisition
Prior to acquisition, the Hyperion mass cytometry system (Fluidigm) was autotuned using a 3-element tuning slide (Fluidigm) according to the tuning protocol provided by the Hyperion imaging system user guide (Fluidigm). As an extra threshold for successful tuning, a detection of at least 1,500 mean duals of 175 Lu was used. Regions of interest were selected based on hematoxylin and eosin stains performed on consecutive tissue sections after which areas of 1,000 × 1,000 µm were ablated and acquired at 200 Hz. Ablation of one area took ∼2 h. Data was exported as MCD files and visualized using the Fluidigm MCD TM viewer. In order to better separate antibody signal and noise, each marker was visually inspected and a minimum signal threshold of 1 or 2 dual counts was set in the Fluidigm MCD TM viewer.

Notes
-All steps should be performed using plastic containers and without the use of glass to reduce metal binding to glass and metal contamination. -Small differences exist in metal concentrations between batches. Thus, it should be taken into account that with a new metal-conjugation, optimal antibody dilutions can vary. Therefore, it is advised to validate the antibody performance by imaging mass cytometry after every conjugation.

RESULTS
To develop the 40 marker imaging mass cytometry antibody panel, antibody performance was assessed by IHC. Initially, 52 cellular targets, of interest for the field of cancer immunology, were selected for potential implementation in the panel.
Antibody selection was based on in-house knowledge of antibody performance in IHC or, when unavailable, manufacturer's datasheets. All antibodies were tested with either low (pH 6) or high (pH 9) pH antigen retrieval buffers, with the former resulting in optimal antigen detection for the majority of antibodies. In total, 65 antibodies were tested for implementation into the panel of which 58 performed well in low pH antigen retrieval conditions (Supplementary Tables 1, 2).
After antibody conjugation to their respective metals, imaging mass cytometry was performed on tissue sections that were incubated overnight at 4 • C with an antibody mix containing all 40 antibodies ( Table 1). It was observed that the immunodetection patterns were comparable between IHC and imaging mass cytometry confirming that IHC is a useful tool for the low cost validation of imaging mass cytometry antibodies. However, low signal intensity and/or high background were observed for a relevant proportion of the tested antibodies, when compared to IHC (e.g., anti-CD163, clone EPR14643-36, Figure 1). We reasoned that this was potentially due to the lack of a signal amplification step in the procedure or the excessive complexity of the antibody mix. Therefore, we tested different conditions for each antibody. For this, each antibody was incubated either overnight at 4 • C or at room temperature for 5 h after which signal intensity and specificity were assessed by imaging mass cytometry and immunodetection patterns were compared to IHC (Figure 1, Supplementary Figure 1). For the majority of antibodies, striking differences in antibody performance (e.g., intensity, background) were observed between conditions. For instance, anti-CD163 (clone EPR14643-36) performed best after a 5 h incubation at room temperature, while anti-CD3 (clone D7A6E) performed optimally when incubated overnight at 4 • C, as shown by its specific signal and low background when compared to a 5 h incubation at room temperature (Figure 1). In general, lowly abundant antigens were difficult to detect when their respective antibodies were incubated at 4 • C and were best assessed by a room temperature incubation. In contrast, antibodies targeting abundant proteins lost specificity when incubated at room temperature which was resolved by their incubation at 4 • C. However, there were exceptions to this pattern as demonstrated for the anti-CD163 antibody (clone EPR14643-36). After determining the optimal conditions for each antibody, they were assigned into one of two consecutive incubation steps as described by the stepwise protocol in the previous section. Furthermore, after determining the optimal dilution for each antibody, a number of antibody stock solutions (directly derived from the metallabeling protocol) had to be diluted to allow their incorporation in the antibody mix volume. Specifically, anti-CD15, anti-CD31, anti-CD38, anti-CD45RO, and anti-Granzyme B were diluted 2 times, while anti-CD57, anti-D2-40, anti-HLA-DR, and anti-Ki-67 were diluted 3 times, and anti-CD68 was diluted 6 times.
A number of antibodies performed well in IHC and were, therefore, deemed good candidates for imaging mass cytometry but had to be excluded due to various reasons (Supplementary Table 2). As discussed, and in contrast to IHC, signal amplification is absent in imaging mass cytometry which limits the detection of low abundant markers. Thus, of the tested and conjugated antibodies, eight had to be excluded due to dim signal despite their good performance in IHC.
Conjugation of antibodies in house enabled the use of the 89 Y, 115 In, 194 Pt, and 198 Pt isotopes and allowed for easy implementation of new markers and clones. Of note, not all antibodies performed well after the conjugation protocol. This was observed for four antibodies that did not function in IHC after the conjugation procedure. Consequently, it was hypothesized that the binding domains of some antibodies or the antibody structure can be affected during the conjugation procedure which involves the partial reduction of antibodies. By making use of indirect antibody detection with a Qdot800-labeled secondary antibody containing 112 Cd/ 114 Cd isotopes, an additional marker could be included in the panel, resulting in a total of 40 markers.
The here described 40 marker panel for imaging mass cytometry comprised tissue structural markers, myeloid and lymphoid lineage markers (Figure 2), but also proteins involved in immune activation, immune checkpoints, and surrogates of cellular states (Supplementary Figure 1). To demonstrate the applicability of the panel for extensive immunophenotyping of tissue, imaging mass cytometry was performed on colorectal cancer tissues and evaluated for a plethora of immune cells. In a single region of interest, among other cell types, tumor cells (Keratin + ), T-helper cells (CD3 + CD4 + ), cytotoxic T cells (CD3 + CD8 + ), regulatory T cells (CD3 + FOXP3 + , Figure 3A), HLA-DR + macrophages (CD68 + HLA-DR + ), CD163 + macrophages (CD68 + CD163 + ), and CD11c + macrophages (CD68 + CD11c + , Figure 3B), were identified. Moreover, the panel allows for the visualization of additional features which are essential for the study of cancer immunology such as tissue residency-like (e.g., CD103 + T cells) and activation phenotypes (granzyme B + T cells, Figure 3C).

DISCUSSION
We report a 40-antibody panel for imaging mass cytometry of FFPE tissues with focus on cancer immunology. Furthermore, we optimized a protocol for optimal performance of antibodies by making use of consecutive incubation steps at different duration and temperature.
Imaging mass cytometry is a low throughput technology that generates high dimensional data, applicable on both FFPE and snap-frozen tissue. FFPE tissue is readily available at the departments of Pathology as it constitutes the standard method for archiving tissues in most medical centers. Furthermore, FFPE tissue allows for the use of tissue micro arrays (TMA) which, as compared to whole slides, increases the throughput and reduces costs associated with imaging mass cytometry.
For single-cell mass cytometry users, expertise in flow cytometry is extremely useful for supporting the development of antibody panels. However, imaging mass cytometry on tissue has a closer resemblance to IHC or immunofluorescence detection methods as shown by the high comparability in antibody performance between these techniques. It is important to note that the lack of a signal amplification step in imaging mass cytometry procedures, as compared to IHC, can hamper the visualization of low abundant proteins, as it is the case for some immune checkpoint molecules. Therefore, it should be taken into account that cell populations with a dim expression of those markers are potentially not observed by imaging mass cytometry. A major prerequisite for analysis of FFPE tissues is the need of antigen retrieval, where the crosslinks created by formalin fixation are broken to make the epitopes accessible for antibody binding. Antigen retrieval is commonly done by boiling tissues in a buffer with a pH ranging from 3 to 10, which can greatly influence antibody performance (18). In an imaging mass cytometry panel all antibodies should perform with the same antigen retrieval protocol. In the design of the current panel it was observed that the majority of antibodies performed best when employing citrate (pH 6) as antigen retrieval buffer, in line with the manufacturers' recommendations for the majority of antibodies.
Thus, for implementation in an IMC panel, it is recommended to select antibodies that are known to perform in IHC on FFPE tissue and to consider the recommended antigen retrieval conditions.
To allow the detection of 40 markers and ensure optimal performance of all antibodies, tissue incubation with the former was separated into three steps. First, tissues were incubated overnight at 4 • C with a primary anti-CD39 antibody that was detected by a Qdot800-labeled, secondary antibody. Quantumdots contain the cadmium isotopes 112 Cd and 114 Cd, which can be detected by CyTOF (14,19). Importantly, this indirect detection step can be employed for primary antibodies that are sensitive to labeling procedures. Subsequently, tissues were incubated with approximately half of the antibodies for 5 h at room temperature, followed by an overnight incubation with the remainder of the antibodies at 4 • C. Because optimal antibody performance can vary greatly between conditions we propose that antibody performance is tested in both settings upon design of an imaging mass cytometry panel.
In the development of the described panel all antibodies were conjugated in house. With the limited number of commercially available FFPE validated conjugated antibodies, in house conjugation allows for easy adaptations to the panel. Furthermore, it enabled the use of the 89 Y, 115 In, 194 Pt, and 198 Pt isotopes and together with the use of Qdot800 ( 112 Cd and 114 Cd), 40 markers can be detected simultaneously.
Sixty-five antibodies/clones were initially tested to be employed in the current imaging approach. Four of those did not perform with low pH antigen retrieval buffer, four were destroyed upon metal conjugation and eight antibodies failed to detect low abundant markers. Two antibodies, anti-Histone H3 (clone D1H2) and anti-pSMAD2 (clone 138D4), were conjugated and detectable by imaging mass cytometry and can potentially be implemented in future panels.
Immunological approaches to cancer therapy have been on the rise and with this, multiplex immunophenotyping techniques have become essential in contexts of research and immunomonitoring. Previous techniques were limited in either the number of markers that can be analyzed simultaneously or by the lack of spatial information. This was overcome with the development of technologies like imaging mass cytometry. The panel and methodology described here can support the extensive immunophenotyping of cancer FFPE tissues. It was designed to provide a comprehensive characterization of the major immune cell subsets present in tissues, in relation to cancer cells. The in-depth study of cellular interactions and investigation of multicellular contexts of antitumor immune responses is supported by the inclusion of tissue structural markers, immune lineage markers but also markers that inform about the functional state of the different cell types. To our knowledge, few labs are currently operating with a 40 marker panel in imaging mass cytometry and, therefore, this work supports the maximization of the potential of imaging mass cytometry across research groups.