Detection of Engineered T Cells in FFPE Tissue by Multiplex In Situ Hybridization and Immunohistochemistry

Abstract

Identifying engineered T cells in situ is important to understand the location, persistence, and phenotype of these cells in patients after adoptive T cell therapy. While engineered cells are routinely characterized in fresh tissue or blood from patients by flow cytometry, it is difficult to distinguish them from endogenous cells in formalin-fixed, paraffin-embedded (FFPE) tissue biopsies. To overcome this limitation, we have developed a method for characterizing engineered T cells in fixed tissue using in situ hybridization (ISH) to the woodchuck hepatitis post-transcriptional regulatory element (WPRE) common in many lentiviral vectors used to transduce chimeric antigen receptor T (CAR-T) and T cell receptor T (TCR-T) cells, coupled with alternative permeabilization conditions that allows subsequent multiplex immunohistochemical (mIHC) staining within the same image. This new method provides the ability to mark the cells by ISH, and simultaneously stain for cell-associated proteins to immunophenotype CAR/TCR modified T cells within tumors, as well as assess potential roles of these cells in on-target/off-tumor toxicity in other tissue.

1. Introduction

Engineered adoptive cell therapy is becoming widespread for treating human diseases, particularly in oncology. Chimeric antigen receptor (CAR) and T cell receptor T (TCR) genetically-modified T cells are adoptive T cell therapies with defined, pre-selected targets [1]. Chimeric antigen receptors contain antibody fragments directed against cell surface tumor antigens fused to a T-cell receptor signaling domain, and a costimulatory molecule. TCR-T cells contain native T cell receptors previously isolated from high-avidity antigen-specific T-cell clones, which can recognize peptides derived from either intracellular or surface proteins. Both types of engineered T cells can provide powerful antigen-specific immunotherapy. Three CD19-targeted (CAR)-modified T (CAR-T) cell products have been approved by the Food and Drug Administration (FDA) for the treatment of B cell malignancies, and many more engineered T cell products are in clinical trials [2, 3].

Interrogating the phenotype and location of these cells after infusion into the patient is critical for translational research to understand how and why therapies succeed in some patients and fail in others. CAR-T and TCR-T cells are routinely detected in the blood and fresh tissue biopsies using multiparameter flow cytometry, employing combinations of cell surface markers, however their detection in formalin-fixed paraffin-embedded (FFPE) biopsies are more challenging. FFPE preservation is standard practice for archiving research biopsies gathered throughout clinical trials. The ability to robustly detect CAR-T and TCR-T cells in archived samples would be extremely useful for retrospective analysis of closed and active trials. Moreover, as CAR-T and TCR-T therapies move into solid tumors where less efficacy is observed [4], there is an urgent need to assess the phenotype of these cells as well as spatial interaction with other cells in the tumor microenvironment (TME).

Identifying the CAR-T/TCR-T cells in situ in patients after infusion has proved challenging. Cell surface markers used for enrichment during manufacture (i.e., EGFRt, tCD19) are not unique in humans and are therefore not ideal for tagging the cells in situ. Detection of engineered T cells in FFPE tissue can be done using specific anti-idiotype antibodies generated against the CAR or TCR protein [5, 6] or by in situ hybridization (ISH) [5–9], probing for unique sequences within transgenes introduced during the engineering process.

Lentiviral vectors used to engineer these cells contain post-transcriptional regulatory elements; these elements were included in prototype retroviral, adenoviral and lentiviral vectors used to transduce cells to maximize accumulation of mRNA [1, 10–12]. The woodchuck hepatitis post-transcriptional regulatory element (WPRE)[13], in particular, is found in many preclinical and clinical CAR-T and TCR-T cells [14]. WPRE provides a long, conserved sequence and is unrelated to any human endogenous sequence. Probes based on WPRE DNA sequence are used to determine level of transduction of T cells during production [15, 16], and this sequence has been used extensively to identify virally infected cells in tissue by in situ hybridization in experimental models i.e. [17–19], but has not been published for use for CAR-T detection in situ previously to our knowledge. Since WPRE is highly transcribed as part of the viral promoter-driven polycistronic message that also encodes the CAR or TCR, we hypothesized that targeting WPRE by ISH would provide a useful method for detection of CAR-T/TCR-T cells in situ and obviate the need to generate specific antibody or probe reagents for each engineered cell type.

It is important to not only identify CAR-T/TCR-T cells in situ, but to also characterize them in terms of immunophenotype and spatial relationship with other cell types. Several reports have been published performing ISH to detect endogenous or viral RNA, followed by immunofluorescence (IF) using confocal microscopy for detection of cell-associated proteins [20, 21]. This method requires overlaying images generated with the different detection techniques. The alternative is to preform mIHC and ISH in adjacent tissue sections [5]. In this report, we present a novel protocol that allows for multiplex IHC together with WPRE fluorescent in situ hybridization (FISH) within the same image for simultaneous immunophenotyping of engineered CAR-T or TCR-T cells and analysis of spatial relationships with other cells within the intact tissue microenvironment.

2. Materials & methods/Experimental

2.1. Method Summary

Formalin-fixed paraffin-embedded tissue slides are processed using standard protocols, but in place of the protease-pretreatment step used in standard RNAscope (Advanced Cell Diagnostics), saponin is used to permeabilize tissue prior to hybridization with RNAscope probes targeting a WPRE sequence common in many CAR-T and TCR-T lentiviral vectors. Following the RNAscope step, samples are moved directly into multiplex immunohistochemistry (mIHC) using standard fluorescent tyramide-based staining and detection methods. The entire procedure is run together on a Leica BOND RX autostainer (Leica, Buffalo Grove, IL). Visualization and quantification are done using HALO software (Indica labs, Corrales, NM).

2.2. Cell pellet preparation

Pre-infusion CAR-T and TCR-T cells were thawed, assessed for viability and counted. Defined admixtures were created by mixing these cells with human peripheral blood mononuclear cells (PBMCs). Cells were fixed in 10% neutral buffered formalin for 48 hours. Cells were collected by centrifugation and re-suspended in 70% ethanol for 30 minutes, then in 100% ethanol for 30 minutes. Cells were collected by centrifugation and supernatants decanted. A 1:1 volume of pre-warmed HistoGel was added to cell pellets and mixed by vortex and placed on ice to solidify for 1 hour. Cell pellets were then wrapped in Bio-Wrap (Leica # 3801090) saturated with 80% alcohol and placed in a cassette. The cell pellet was then embedded in paraffin using standard procedures.

2.3. TCR-T cell quantification by flow cytometry and mCherry IHC

The lentiviral construct (pRLLSIN-mcherry-P2A-TCR.WPRE) used to express the transgenic TCR in these studies also contains a mCherry gene segment separated by a self-cleaving 2A element from the porcine teschovirus (P2A) to ensure coordinated gene expression of both TCR and mCherry transgenes [22, 23]. Defined admixtures were washed with FACS buffer (PBS +1%FBS) and run on a BD FACS Celesta flow cytometer. TCR-T quantification in the defined admixtures was determined by the percentage of mCherry positive cells in the sample. After fixation and embedding, mCherry-positive cells were stained with standard immunohistochemistry using an anti-mCherry mouse monoclonal antibody, clone G6, generated in-house and used at 1:15,000 dilution [24].

2.4. Humanized mouse model treated by TCR-T infusion

The MISTRG (M-CSF^h/hIL-3/GM-CSF^h/hSIRPa^h/hTPO^h/hRAG2^−/−IL2Rg^−/−) humanized mouse model supports hematopoiesis after transplantation of human CD34+ stem and progenitor cells resulting in functional human innate and adaptive immune cells [25]. CD8 and CD4 T cells transduced with a MAGEA1-directed TCR (MAGEA1-TCR-T) were adoptively transferred into MISTRG mice. Briefly, CD8⁺ and CD4⁺ T cells were enriched by magnetic bead selection from the spleen of MISTRG mice followed by stimulation with CD3/CD28 coated microbeads (1:1 ratio) for 24 hours before viral transduction with a lentiviral construct containing a MAGEA1 specific TCR. Six days post transduction cells were FACS sorted on TCR expression using MAGEA1 peptide loaded tetramers and put into a T cell rapid expansion protocol as previously described [26]. TCR-T were harvested one week post expansion and transferred into MISTRG littermates harboring MAGE-A1+ Me275 xenograft tumors (subcutaneous, left flank) [25, 27, 28]. Lung and spleen tissue from these mice were harvested 28 days post TCR-T transfer and fixed in neutral buffered formalin, and paraffin-embedded for histological analysis. All mouse experiments were performed under the FHCRC IACUC #50941.

2.5. Preparation and co-cultures of human skin explants with T-cell clones

CD8⁺ T cell clones specific for the HLA-A*0201-restricted minor histocompatibility (H) antigen HA-1 were generated as previously described [29]. Sequencing of TCR alpha and beta chains, reconstruction and cloning of the TCR into a lentiviral vector, transduction, and expansion of CD8⁺ T cells was performed as reported [30]. Skin biopsies and PBMC were obtained from a male donor that provided written informed consent in a protocol approved by the Fred Hutchinson Cancer Research Center Institutional Review Board (IRB 3895). The preparation of human skin explants was carried out as previously described [31]. Briefly, standard 4-mm punch biopsy specimens were collected in sterile Phosphate Buffered Saline (PBS) (Gibco) and divided into 4 sections of equal size under sterile environment. Each section was cultured on separate 6 well plates in complete cytotoxic T lymphocyte (CTL) medium composed of RMPI-1640 medium supplemented with 10% human serum, L-glutamine and penicillin/streptomycin. As the donor was HA-1 genotypically negative HA-1⁻(RS_1801284, G/G), HA-1 antigen was first introduced into the biopsy using peptide-pulsed monocytes. Specifically, monocytes were isolated using a CD14⁺ cell isolation kit (Miltenyi Biotec) and pulsed with 1 µg/ml HA-1 peptide (VLHDDLLEA) for 3 hours at 37°C. A total of 20 x 10⁴ cells pulsed-monocytes were injected into skin explants at utilizing a 27 G x ½” syringe and were incubated 1 hour at 37°C. After incubation, 1 x 10⁶ T cells HA-1 TCR-T cells were injected per skin explant; the negative control was skin explant incubated with medium alone. After 3 days of co-culture explants were washed 3x with PBS and were fixed in 10% buffered formalin for 7 days and embedded in paraffin for multiplex ISH/IHC procedures.

2.6. Human lymph node biopsy

An FFPE tissue block from a patient treated on a phase 1b dose-finding study (NCT02706405) of the safety and feasibility of combination therapy with JCAR014 CD19-specific 4–1BB-costimulated CAR-T cells and escalating doses of durvalumab, a PD-L1 monoclonal antibody, in adults with relapsed/refractory B-cell non-Hodgkin lymphoma [32] was used for validation of the ISH assay. The study was conducted according to the principles of the Declaration of Helsinki with informed consent and approval by the Fred Hutchinson Cancer Research Center Institutional Review Board. The lymph node biopsy was obtained 28 days after CAR-T cell infusion.

2.7. WPRE ISH Staining with DAB Chromogen on FFPE Tissue

Formalin-fixed paraffin-embedded (FFPE) tissues were sectioned at 4 µm onto positive-charged slides, baked for 1 hour at 60°C and loaded onto the Leica BOND RX Autostainer platform (Leica, Buffalo Grove, IL). The automated procedure is outlined below: slides were baked and dewaxed using Leica Dewax Solution (Leica Biosystems AR9222). Antigen retrieval was performed at 95°C for 15 minutes using Epitope Retrieval Solution 2 (Leica Biosystems AR9640) followed with permeabilization solution (1% Saponin, 1 mM EGTA in PBS) pretreatment at 40°C for 60 minutes or Advanced Cell Diagnostics (ACD) Enzyme (Proteinase III) at 40°C for 15 minutes. The slides were then blocked with hydrogen peroxide for 10 minutes and incubated with ACD target probe, WPRE (ACD #517728), positive control probes for mouse (ACD #313918) and human peptidyl-prolyl cis-trans isomerase B (PPIB; ACD #322100) or negative control probe 4-hydroxy-tetrahydrodipicolinate reductase (DapB; ACD #312038) at 42°C for 120 minutes. After the probe incubation, the staining continued with the RNAscope 2.5 LS Reagent Kit-Brown (ACD #322100) for the amplification, then detected by 3,3’-diaminobenzidine (BOND Polymer Refine DAB Detection, Leica DS9800) for 20 minutes. After counterstaining with hematoxylin and bluing, the slides were dehydrated and cleared through alcohol and xylene and coverslips placed and sealed using Cytoseal XYL mounting media (Thermo, #8312–4).

2.8. WPRE (FISH) & IHC Staining of FFPE tissue

FFPE tissues were sectioned at 4 µm onto positive-charged slides and baked for 1 hour at 60°C and loaded onto the Leica Bond Rx Autostainer platform (Leica, Buffalo Grove, IL). The automated procedure is outlined below: slides were baked and dewaxed using Leica Dewax Solution (Leica Biosystems AR9222). Antigen retrieval was performed at 95°C for 15 minutes using Epitope Retrieval Solution 2 (Leica Biosystems AR9640) followed with 1% Saponin (purified from Quillaja Bark (Sigma S-4521) in EGTA-PBS solution pretreatment at 40°C for 60 minutes or ACD Enzyme (Proteinase III) at 40°C for 15 minutes. The slides were then blocked with hydrogen peroxide for 10 minutes and incubated with ACD target probe, WPRE (ACD #517728), positive control probe human or mouse PPIB (ACD #322100, ACD #313918) or negative control probe DapB (ACD #312038) at 42°C for 120 minutes. After hybridization, the RNAscope 2.5 LS Reagent Kit-Brown (ACD #322100) is applied for the amplification, then detected by Opal fluor 570 (Perkin Elmer, #FP1488001KT) at 1:500 for 10 minutes. To perform the CD3 immunofluorescence (IF) staining, the probe was stripped from the tissue using Epitope Retrieval Solution 2 (Leica Biosystems AR9640) at 100°C for 20 mins. Endogenous peroxidase was blocked with 3% H₂O₂ (Fisher Scientific S25359) for 8 minutes followed by protein blocking with TCT buffer (0.05M Tris, 0.15M NaCl, 0.25% Casein, 0.1% Tween 20, pH 7.6 +/− 0.1) for 30 minutes. The CD3 (Thermo #RM9107) primary antibody at 1:400 was applied for 60 minutes followed by the secondary antibody application (PowerVision poly-HRP anti Rabbit IgG; #PV6119) for 10 minutes and the application of the tertiary TSA-amplification reagent Opal fluor 650 (Perkin Elmer, #FP1496001KT) at 1:100 for 10 minutes. Note: substitution of 0.5% Triton X-100 for 20 minutes at room temperature for the 1% saponin pretreatment step also worked to permeabilize cells prior to WPRE FISH/CD3 IHC however we used saponin for the general method for the duplex as well as higher order multiplex staining.

Four-plex staining follows the same automated procedure above except using Opal fluor 520 (Perkin Elmer, # FP1487001KT) for CD3, and continuing with the primary and secondary antibody stripping with Epitope Retrieval Solution 2 for 20 minutes before repeating the process with the second primary antibody Ki-67 (Dako #M7240) at 1:100 (position 2) with Opal fluor 620 (Perkin Elmer, # FP1495001KT) or the third primary antibody CD163 (BioSB #BSB327) at 1:800 and CD68 (Dako #M0876) at 1:800 cocktail (position 3) with Opal fluor 650 (Perkin Elmer, # FP1496001KT) starting with a new application of 3% H₂O₂. The process was repeated until all three positions were completed; however, there was no stripping step after the third position.

Slides were removed from the BOND Autostainer and stained with Spectral DAPI (Sigma 08417) for 5 minutes, rinsed, and coverslips placed on stained sample with Prolong Gold Antifade reagent (Invitrogen/Life Technologies, Grand Island, NY). The slides were cured for 24 hours at room temperature in the dark and images were acquired on an Aperio ScanScope FL Imaging System, Perkin Elmer Vectra 3.0 or Vectra Polaris (Indica).

2.9. HALO digital image analysis

Images were analyzed across the entire tissue biopsy with HALO image analysis software (Indica labs) using the Indica-Labs_ISH-v3.4.3.0, Indica-Labs_Cytonuclear v2.0 and Indica-Labs_ISH-IHC-v1.3.2.0 modules. The percentage of WPRE, DapB, PPIB, CD3, or mCherry positive cells were calculated by total brown chromogen or fluorescence positive cell counts, divided by total Hematoxylin or DAPI positive nuclei, respectively. The number of cells quantified and listed in the figure legends. Percent positive cell values were imported into Prism (GraphPad) for graphing and statistical analysis.

3. Results & Discussion

3.1. WPRE as a general target for probe design to detect CAR and TCR T cells

Detection of engineered T cells in FFPE tissue can be done using ISH, probing for expressed transgenes introduced during the engineering process [5–7]. In order to reliably distinguish these cells from endogenous T cells in tissue, this sequence needs to be unique to the vector, long enough for generation of multiple probes to allow adequate amplification, and absent in the host genome, generally requiring a specific probe for each type of CAR or TCR T cell type. We sought to find a common sequence shared by different CAR or TCR-T vectors to use as a general reagent. Because WPRE was present in optimized, prototype lentiviral vectors [12], this element was found in all of CAR and TCR lentiviral vectors used in preclinical and clinical studies that we have tested. The WPRE element is co-transcribed as part of the polycistronic mRNA that also encodes the CAR or TCR proteins, and therefore it should be present if the cells are still expressing the TCR, at least at the RNA level. This approximately 600 bp sequence is unique, sharing little or no homology with sequences in the human or mouse genome and is available commercially as an RNAscope probe (ACD; Advanced Cell Diagnostics, Inc.).

3.2. Detection of TCR-T cells by ISH in cell pellet admixtures; comparison with IHC

As a test for sensitivity and specificity of detection of CAR-T and TCR-T cells using WPRE ISH, we created admixtures of known proportions of pre-infusion human TCR-T or CAR-T cells with human PBMCs. Cells were fixed in formalin and embedded in paraffin using similar methods used for tissue samples. Sections of embedded cell pellets containing these admixtures were prepared and hybridized with WPRE RNAscope ISH probes (ACD) using standard methods for automated staining. Figure 1 illustrates specific staining of MAGEA1-TCR-T cells with the WPRE probes, showing an increased proportion of positive cells in admixtures with higher percentage of TCR-T cells (Figure 1A, compare middle and right panels). No staining was seen in the PBMC only cell pellet (0%, left panel). Similar trends were observed with staining of admixtures containing four other TCR-T or CAR-T cell types (data not shown). The relative percentage of WPRE positive cells seen by ISH was quantified in HALO (Figure 1B; black bars). This TCR-T construct also contained an mCherry cell surface protein marker. Quantification of TCR-T cells in cell pellet admixtures stained by standard immunohistochemistry using an mCherry monoclonal antibody was performed in parallel (Figure 1B; grey bars) and a good correlation was seen between these two quantification methods.

Figure 1.

Detection of MAGEA1-TCR-T cells by ISH with WPRE probes and comparison to quantification by mCherry IHC (A) Images of cell pellet admixtures are shown stained with WPRE ISH RNAscope. Mixtures tested include human PBMCs only (left panel), 82% PBMCs + 18% MAGE-A1-TCR-T (middle panel), and 55% PBMCs + 45% MAGE-A1-TCR-T cells (right panel), as quantified by flow cytometry prior to fixation. WPRE ISH is visualized as brown stain, with all cells counterstained with hematoxylin (blue). (B) FFPE samples stained for WPRE as shown in (A) were quantified in HALO (Indica-Labs_ISH-v3.4.3.0 module) and the percentage of ISH+ cells are shown in the black bars.. The percentage of WPRE positive cells in pellets were compared with the number of mCherry positive cells as measured by standard immunohistochemistry staining and Indica-Labs_Cytonuclear v2.0 module for quantification (grey bars). A total of 10,000–20,000 cells were scored per sample; this staining and analysis was done twice, counting all cells in the samples. The errors bars indicate the range between the biological replicates. A one-tailed Pearson’s correlation calculated in Prism (GraphPad) for the mean values indicated that these two detection methods gave similar quantification (R²=0.9999; p=0.038).

3.3. Specific detection of TCR-T cells in human skin explants

We next tested the WPRE RNAscope detection on human skin explants with and without injection of HA-1-TCR-T cells into the skin sample ex vivo, followed by co-culturing. As a positive control, slides were hybridized with probes directed against the human housekeeping gene peptidyl-prolyl cis-trans isomerase B (mPPIB+), and as a negative control, with probes directed against the bacterial gene, 4-hydroxy-tetrahydrodipicolinate reductase (DapB-). WPRE positive cells were detected only in skin samples treated with the TCR-T cells, while human PPIB staining was present in both, and DapB staining was negligible (Figure 2). Representative images of stained tissue are shown. These data indicated that WPRE could detect TCR-T cells within FFPE-fixed human tissue.

Figure 2.

WPRE ISH staining of human skin explants with and without infusion of HA-1-TCR-T cells show TCR-T specific staining within human tissue specimens: (A) (left panels) WPRE probe; (middle panels) DapB⁻ control probe; and (right panels) human PPIB⁺ control probe. ISH signal is visualized as brown stain, and cells are counterstained with hematoxylin (blue).

3.4. Using saponin as an alternative to protease for cell permeabilization

Our overarching goal was to characterize CAR-T and TCR-T cells in terms of immunophenotype and spatial relationship to other cells in the tissue. Thus, we sought to develop methods to multiplex the WPRE ISH with immunocytochemistry, staining for multiple cell protein markers. We found that results using standard ACD ISH-IHC protocols [33] were variable as cell and tissue integrity was compromised by protease pre-treatment followed by multiple rounds of antigen retrieval needed to stain more than one cellular antigen. A previous report used the surfactant, saponin, in place of protease to permeabilize fixed tissue and allow DNA probes access to hybridize with target RNA, suggesting an alternative protocol for WPRE RNA ISH [34]. Saponin has been used historically in immunohistochemistry, immune-electron microscopy [35, 36], and flow cytometry [37] to permeabilize cells and allow access of antibodies to intracellular antigens while preserving membrane organelles. Saponin is purported to interact with cholesterol and create pores in the plasma membrane of cells [38, 39]. Unlike other non-ionic detergents that also make the plasma membrane permeable, saponin does not interact appreciably with proteins and so is likely to preserve cell surface membrane protein epitopes better [40]. 0.5% Triton X-100 was also tested in parallel and worked in this protocol, but we chose to continue with saponin given the reduced potential to interact with protein antigens for subsequent mIHC.

We varied saponin concentration, time and temperature to get optimal detection of WPRE probe. Using this optimized protocol, we compared the staining to the ACD protease protocol in various preclinical samples. Figure 3 shows a comparison of staining for WPRE positive cells in human skin explants pre-treated with TCR-T cells using the protease vs. the optimized saponin protocol. The relative signal was similar, but with far greater resolution and tissue integrity remaining using saponin (compare Figure 3, panels A and B). When the number of WPRE positive cells were quantified in HALO from the two images, slightly more WPRE positive cells were identified with the saponin protocol, due to better resolution of individual cells. These results suggested that using saponin permeabilization would allow for multiplex staining as it gives similar sensitivity as protease, while better preserving tissue structure. Intact cellular and nuclear structure are important to allow for accurate quantitative analysis of images in HALO (Indica Labs).

Figure 3.

Images showing comparison of WPRE ISH performed using standard RNAscope procedure (A) vs. alternative saponin procedure (B) on human skin explants infused with HA-1-TCR-T cells. ISH signal is visualized as brown stain, and cells are counterstained with hematoxylin (blue). (C) Quantitative comparison of ISH⁺ cells with the two procedures performed using the Indica Labs ISHv3.4.3 module is shown. Quantification was done only once but data represent the entire tissue sample. For saponin protocol, a total of 1590 cells were counted in tissue explant; For protease protocol, a total of 845 cells were counted. The fewer number of cells enumerated was due to loss of tissue integrity.

3.5. WPRE probe and CD3 antibody duplex staining on humanized mouse tissue containing TCR-T cells

To optimize our multiplex ISH/IHC protocol, we used lung and spleen tissues isolated from humanized mice [25] with and without treatment with systemic TCR-T infusion [27]. Initial single stain tests were performed on these tissues, hybridizing with WPRE probes for detection of TCR-T cells, as well as with mouse PPIB positive, and DapB negative control probes. Mouse PPIB stained the majority of the lung cells in both treated and untreated mice, while WPRE only stained cells in mice previously infused with TCR-T cells, and DapB staining was essentially negative (Supplemental Figure 1). The percent of ISH+ cells were quantified in HALO (Indica Labs) and shown in Supplemental Figure 1B.

We next performed a multiplex staining with WPRE vs. DapB control probes by fluorescent in situ hybridization (FISH) using the saponin protocol, followed by IHC using anti-human CD3 antibody. Images of lung (Figure 4A, B and E) and spleen tissue (Figure 4D) are shown. WPRE⁺CD3⁺ cells were visualized as a subset of the CD3⁺ cells in the lung and spleen of these mice. The nuclei and cell membranes were nicely preserved in structure as seen with DAPI and the CD3 cell surface marker staining, respectively. A multiplex done using the standard RNAscope protease pre-treatment is shown for comparison (Figure 4C). While both WPRE and CD3 staining are visualized, the cell and nuclear integrity are reduced significantly even after only one round of antigen retrieval prior to antibody staining (Fig 4C), as compared to the saponin protocol (Fig 4A).

Figure 4.

A subset of T cells stained strongly with WPRE in lung and spleen tissue from humanized mice with prior infusion of MAGEA1-TCR-T cells. FISH/IHC duplex images of lung tissue stained with WPRE/CD3 (A) and DapB/CD3 (B) using the saponin permeabilization are shown with CD3 (red), WPRE (green), and nuclei (DAPI, blue) visualized. A WPRE/CD3 duplex staining done using standard RNAscope protease permeabilization protocol and is shown in (C) for comparison. Additional WPRE/CD3 stained images of spleen (D) and lung (E) tissue from TCR-T treated mice at both 10x and 40x magnification are shown with CD3 (green), WPRE (red), and DAPI (blue) visualized. (F) Three sections of the same humanized mouse lung tissue with TCR-T cells were stained with WPRE/CD3 and quantified in HALO using the Indica-Labs_ISH-IHC-v1.3.2.0 module. The number of total human T cells (CD3⁺), WPRE⁺, and CD3⁺WPRE⁺ double positive (TCR-T) cells are shown for each tissue slice. We present the data as separate quantifications as the sections were not adjacent and not the same size. The total number of cell counted by DAPI are section 1: 18,467 cells; section 2: 50,602 cells, section 3: 125,600 cells.

Three separate sections of the humanized mouse lung tissue were stained independently for WPRE and CD3 using the saponin protocol, and CD3⁺, WPRE⁺, and CD3⁺WPRE⁺ double positive cells were quantified in HALO (Indica Labs). Graphs of these data are shown in Figure 4F. In all three sections, similar levels of WPRE⁺ and CD3⁺WPRE⁺ cells indicated strong co-localization of these markers. In addition, the percentage of T cells that were WPRE⁺ was also similar. In the section containing fewer T cells, fewer WPRE⁺ and WPRE⁺CD3⁺ cells were seen.

3.6. WPRE ISH staining in combination with higher order multiplex immunohistochemical staining

In a recent report, Dikshit et al. performed multiplex FISH/IHC, in which they used protease to permeabilize the cells, and omitted the antigen retrieval/stripping step prior to antibody staining which allowed better preservation of cell and nuclear structures but this procedure would only allow for one round of antibody staining [33]. We tested our saponin permeabilization procedure with a higher order multiplex staining, starting with WPRE vs. DapB FISH, followed by three rounds of IHC, staining for T cells (CD3), macrophages (CD163/CD68), and a proliferation marker (Ki67). As shown in Figure 5, we saw significant staining with all four markers. WPRE was localized inside CD3⁺ cells as expected (Figure 5A), which were surrounded by macrophages in the humanized mouse lung biopsy (Figure 5B). Figure 5 C–E show a different set of images from the same slide, illustrating co-localization of the Ki67 marker inside a CD3⁺WPRE⁺ TCR-T cell.

Figure 5.

WPRE ISH can be multiplexed, staining with three antibodies as a four-plex, using WPRE FISH, followed by CD3/CD163-CD68/Ki67 IHC staining in series: (A) and (B) show visualization of CD3/WPRE and WPRE/CD3/CD68-CD163 stained images of lung tissue from humanized mice with prior infusion with MAGEA1-TCR-T cells in the same field of view; (C-E) show visualization of CD3/WPRE (C), CD3/WPRE/Ki67 (D), and WPRE/Ki67 (E) in the same field of view to identify a proliferating TCR-T cell in situ. Proliferating TCR-T cell is indicated by the white circle. All images were captured from the same 4-plex stained sample.

3.7. Detection of CAR-T cells by WPRE ISH/mIHC staining in a clinical biopsy

We also tested our multiplex staining protocol on an archived research biopsy of a cervical lymph node from a patient treated with CD19 CAR-T cells on a phase 1b clinical trial (NCT02706405), collected 28 days post-infusion. We performed initial characterization of this tissue with single staining by RNAscope ISH with human PPIB, DabB and WPRE probes (Supplemental Figure 2). It is important to note that PPIB housekeeping control gene should be done in parallel to ensure sufficient RNA quality in the regions of the tissue to be analyzed, especially for archived tissue samples. We selected a region of this biopsy that had strong PPIB staining to proceed with analysis (image shown in Supplemental Figure 2A). WPRE⁺ cells were readily detectable as a single stain (Supplemental Figure 2C) so next we tested staining this sample with WPRE vs. DapB FISH, followed by CD3 IHC. As shown in Figure 6A–B, we saw a strong signal for WPRE in a subset of CD3⁺ cells, with little background staining in the DapB- control. The relative percentage of WPRE+ CD3 vs. DapB+ CD3 cells throughout the tissue were quantified in HALO (Figure 6C). Colocalization of WPRE FISH stain with CD3 cell surface protein allows one to identify intact CAR-T or TCR-T cells, distinguishing them from genetic debris in dying cells or cells phagocytosed by macrophages.

Figure 6.

Robust detection of CD19 CAR-T cells in situ in a patient’s lymph node biopsy. Duplex FISH/IHC images with DapB⁻/CD3 (A) and WPRE/CD3 (B) are shown with CD3 visualized in red, WPRE in green, and DAPI in blue. (C) The number of CD3+FISH+ cells were quantified in HALO using the Indica-Labs_ISH-IHC-v1.3.2.0 module and the relative percentage of CD3/FISH double positive cells are shown. The percentages presented are based on counting all cells in the tissue biopsy (between 59,300–136,00) and dividing CD3⁺FISH⁺ cells by the total cells detected in each tissue biopsy as defined by DAPI counterstain. WPRE/CD3 and DapB/CD3 staining and quantification was done only once on this biopsy due to limited material.

Additional staining with higher order multiplex was done on this biopsy, staining for WPRE, followed by CD3, CD163/CD68, and Ki67. As with the humanized mouse tissue, all four markers stained well, and clear cell co-localization of WPRE and CD3 can be visualized (Figure 7A). Figure 7B shows the same fields of view as in 7A, with visualization of staining with CD163/CD68 antibody cocktail included, showing relative position of T cells, CAR-T cells, and macrophages. Figure 7C–F show higher magnification images from a different field of view of this 4-plex staining, illustrating CAR-T cells that were also positive for Ki67, suggesting that these cells were proliferating.

Figure 7.

Multispectral images of a cervical lymph node biopsy from a patient after infusion with CD19 CAR-T cells stained in a four-plex using WPRE FISH, followed by CD3/CD163-CD68/Ki67 IHC staining in series: (A) Images showing staining with CD3 (green), WPRE (red), and DAPI (blue); (B) the same fields of view as in (A) showing CD3 (green), WPRE (red), CD163/CD68 (yellow) and DAPI (blue). Images in (C-F) show four higher magnification images of the same field of view, focusing on proliferating CAR-T cells in situ: (C) WPRE (red) and CD3 (green); (D) WPRE (red), CD3 (green), and DAPI (blue); (E) WPRE (red), Ki67 (teal); (F) WPRE (red), CD3 (green) and Ki67 (teal), CD163/CD68 (yellow). Proliferating cells are indicated with white arrows. All images were captured from the same 4-plex stained sample.

4. Discussion

4.1. Summary conclusions from this study

FISH using WPRE probes is demonstrated as a sensitive and specific method for CAR-T/TCR-T detection using cell pellet admixtures as well as in biopsies from both humanized mouse and human samples. Three different TCR-T and CAR-T cell types were detected with minimal background (Figures 2, 4 and 6). The use of saponin permeabilization for the RNAscope ISH protocol allowed further staining of the same tissue samples with 3 rounds of tyramide-based fluorescent detection of cell-associated proteins. In this study we demonstrated identification of proliferating TCR-T and CAR-T cells (Figures 5 and 7), as well as location relative to macrophages in situ. This method is useful to ‘tag’ CAR-T and TCR-T cell to allow qualitative assessment (i.e., immunophenotyping or spatial relationship to other cells) but it remains to be demonstrated if this method is robust enough to quantify CAR-T/TCR-T cells within preserved tissue.

4.2. Advantages of this new protocol

As compared to previous methods, this protocol enables combination of ISH with multiplexed and high-resolution IHC as a result of using the alternative tissue permeabilization conditions. In addition, use of probes directed against the WPRE present in many lentiviral vectors allows for a versatile protocol, obviating the need to design individual CAR or TCR specific probes. The fact that the WPRE element is part of the same polycistronic transcript that encodes the TCR or CAR makes it a possible surrogate marker for engineered T cells that are still expressing the TCR or CAR, at least at the RNA level. The entire process is automated as a single protocol on an autostainer making this procedure fast enough to analyze multiple samples for rapid characterization of CAR-T or TCR-T cells in situ and study other cell markers for activation, proliferation, exhaustion, and proximity to other cells in the tissue. A disadvantage of this method is that is relies on RNA stability within the tissue. However, using this assay, we detected CAR-T cells in archived biopsies up to two years-old making this assay useful for CAR-T detection and characterization in situ during ongoing trials as well as for retrospective analysis of specimens from completed trials.

4.3. Future perspective

While analysis of fresh specimens is ideal for absolute quantification of engineered cell number within tissues, this new protocol provides a rapid method for qualitative assessment of location, immunophenotype, and spatial orientation relative to other tumor and immune cells in the TME in situ, providing valuable information. Future experiments include comparison of detection with anti-idiotype antibodies as compared to WPRE ISH to assess if detection of the WPRE element at the end of the TCR transcript is a suitable surrogate marker for expression of the cell surface TCR. Expansion of this protocol to detect multiple gene targets via ISH, together with multiplex IHC to allow simultaneous analysis of targets at both the RNA and protein level is being explored. This method would complement analyses with disruptive but more in-depth transcriptomic profiling such as NanoString nCounter® gene expression system and CITE-seq analyses of biopsies from clinical trials to understand effects of the tumor microenvironment on TCR-T/CAR-T efficacy. This assay could also be used to verify suspected mechanism-based (CAR-T mediated) toxicities in archived tissue biopsies. Other applications of this assay being explored include characterization of HIV viral reservoirs in tumor samples from infected patients.

Supplementary Material

1

Supplemental Figure 1. Lung tissue from humanized mice with and without prior infusion with MAGEA1-TCR-T cells are shown in (A): (bottom panels) Mouse PPIB+ control probe; (middle panels) DapB- control probe; and (top panels) WPRE probe. All images are counterstained with hematoxylin (blue); (B) Quantification of ISH positive cells in HALO using the Indica-Labs_ISH-v3.4.3.0 module. The percentages presented are based on counting all cells in the tissue biopsy (222,000 for untreated, and 428,000 for TCR-T treated), and dividing ISH⁺ cells by the total cells detected by hematoxylin counterstain. This brightfield staining and quantification was done once before moving on to validating duplex assay for TCR-T/CAR-T detection.

Supplemental Figure 2. Images of a patient’s lymph node biopsy after infusion with CD19-CAR-T cells stained using RNAscope ISH with probes for human PPIB⁺ (A), DapB⁻ (B) and WPRE (C) are shown. ISH signal is visualized as brown stain, and cells are counterstained with hematoxylin (blue).

Abbreviations:

CAR-T

Chimeric antigen receptor T cell

TCR-T

T cell receptor T cell

FFPE

Formalin fixed paraffin embedded

WPRE

Woodchuck Hepatitis Virus Posttranscriptional Regulatory Element

TME

tumor microenvironment

FISH

Fluorescent in situ hybridization