Sensitive in vivo imaging of T cells using a membrane-bound Gaussia princeps luciferase

Abstract

We developed a new approach to bioluminescent T cell imaging using a membrane-anchored form of the Gaussia luciferase (GLuc) enzyme, termed extGLuc, which we could stably express in both mouse and human primary T cells. In vitro, extGLuc+ cells emitted significantly higher bioluminescent signal when compared to cells expressing GLuc, Renilla luciferase (RLuc) or membrane-anchored RLuc (extRLuc). In vivo, mouse extGLuc+ T cells showed higher bioluminescent signal when compared to GLuc+ and RLuc+ T cells. Application of this imaging approach to human T cells genetically modified to express tumor-specific chimeric antigen receptors (CARs) enabled us to show in vivo CAR-mediated T cell accumulation in tumor, T cell persistence over time and concomitant imaging of T cells and tumor cells modified to express firefly luciferase. This sensitive imaging technology has application to many in vivo cell-based studies in a wide array of mouse models.

Figure 1

(a) Schematic representations of SFG luciferase constructs containing the native secreted GLuc, extGLuc, RLuc, extRLuc, and GFP-FFLuc. All constructs other than GFP-FFLuc contained an IRES-hrGFP reporter cassette to determine transduction efficiency. Black box, human CD8 leader sequence; LTR, long terminal repeat; SD, splice donor; SA, splice acceptor; CD8, CD8 transmembrane domain; arrows, start of transcription. (b) Bioluminescent imaging of PG-13 retroviral producer cells expressing the luciferase constructs at 24 h of culture and signal from tissue culture supernatants. (c) Quantitative analysis of signal from Fig 1b demonstrate a majority of the GLuc signal (98%) is contained in the culture medium. Supernatants from PG-13 cells containing the extGLuc, RLuc, and extRLuc demonstrate the percentage of signal in the filtered supernatant (27, 48, and 57% respectively). Marks indicate a statistically significant difference between individual groups compared to extGLuc total signal (*) and extGLuc cells (●) with P < 0.01. (d) In vitro time course of bioluminescent signal generated by PG-13 T cells demonstrates a plateau signal for GLuc and extGluc⁺ PG-13 cells of 10 min following the addition of coelenterazine. (e) FACS analysis of human T cells either untransduced, transduced with GLucIRES-GFP, extGLucIRES-GFP, or extGLuc labeled with murine GLuc specific monoclonal antibody verifying expression of extGluc on the cell surface. (f) Cytospin immunohistochemistry of human T cells transduced to express either extGluc or native GLuc stained with murine GLuc specific monocolonal antibody confirms surface expression of extGLuc but not the GLuc.

Figure 2

(a) SCID-Beige mice injected with MHC mismatched C57BL6 T cells expressing extGLucIRES-hrGFP, RLucIRES-hrGFP, or GLucIRES-hrGFP, were imaged by BLI on d 1, 2, 5, and 7 following bolus intravenous injection of coelenterazine. (b) Quantitative analysis of signal at each time point. Asterisk represents a significant difference (P <0.001) when RLuc or GLuc is compared to extGLuc. Data represents one of three experiments with similar results with five mice in each cohort. (c) In vivo time course characterization of bioluminescent signal comparing the extGluc, RLuc and GLuc constructs was obtained on d 7 following allogeneic T cell infusion by obtaining 30 sec acquisitions per minute for 30 min. Once more, the extGluc⁺ T cell signal was >15 fold greater over 2–3 min following substrate infusion than that seen in mice infused with either RLuc⁺ or GLuc⁺ alloreactive C57BL6 T cells. (d) Presence of C57BL6 extGLuc T cells at sites commonly associated with GvHD, including the gut, skin (pinna), lung, liver, and spleen was confirmed by BLI. (e) Presence of T cells and the existence of graft versus host disease is evident in H&E tissue stains. Black arrows shown in the lung, liver, esophagus, and stomach photomicrographs demonstrate the presence of lymphocytes and histiocytes. Increased numbers of mitotic figures, as well as apoptotic and necrotic epithelial cells, are evident in the small intestine, stomach, and esophagus (red arrows). The spleen microsection shows highly mitotic T cells (black arrows) in the red pulp area not present in normal SCID-Beige mouse spleens (data not shown).

Figure 3

(a) SCID-Beige mice bearing A20(OVA) and A20(GFP) subcutaneous tumors infused with extGLuc⁺ DO11.10 OVA specific T cells were imaged by BLI over 13 days demonstrating initial T cell retention in the lung followed by increasing infiltration of T cells in the A20(OVA) tumor but not the A20(GFP) tumor over time. Data is representative two similar experiments with five treated mice in each experiment with similar results. (b) PET images visualizing A20(OVA) and A20(GFP) tumors at d 0 show a larger A20(OVA) tumor when compared to A20(GFP) tumor. By d 7, the A20(OVA) tumor appears decreased in size in contrast to the A20(GFP) tumor which has progressed. (c) Quantitative signal of A20(GFP) and A20(OVA) tumor as assessed by PET demonstrates stable signal in the DO11.10 T cell infiltrated A20(OVA) tumor consistent with anti-tumor efficacy of infused T cells, while the A20(GFP) tumor demonstrates statistically significant progressive signal by PET consistent with unabated tumor growth (asterisk represents P <0.01). (d) Immunohistochemistry of A20(GFP) and A20(OVA) tumors at d 13 following adoptive therapy, demonstrating DO11.10 T cells infiltrating the A20(OVA) tumor, with no evidence of T cell infiltration in A20(GFP) tumors. (e) Balb/c mice bearing subcutaneous A20(OVA) and wild type A20 tumors were injected by tail vein with extGLuc⁺ DO11.10 OVA specific T cells, and monitored by BLI. At d 2 and 3, extGluc⁺ DO11.10 T cell infiltration is apparent in the A20(OVA) but not the wild type A20 tumor. Data represents one of five treated mice with similar results.

Figure 4

(a) BLI of SCID-Beige mice bearing NALM-6 subcutaneous tumors following infusion of human 19z1⁺exGLuc⁺ or Pz1⁺exGLuc⁺ T cells. At 24 and 48 h following T cell injection, mice treated with 19z1⁺exGLuc⁺ T cells demonstrated bioluminescent signal in the NALM-6 tumor, in contrast to mice treated with Pz1⁺exGLuc⁺ T cells. Five mice were treated in each cohort with similar results. (b) Immunohistochemistry of tumor at 48 h demonstrates the presence of human CD8⁺ T cells in 19z1⁺exGLuc⁺ T cell treated mice, but not in Pz1⁺exGLuc⁺ T cell treated mice, consistent with the bioluminescent images.

Figure 5

(a) Bicistronic retroviral vectors 1928zIRES-extGLuc and Pz1IRES-extGLuc. Black box, human CD8 leader sequence; LTR, long terminal repeat; SD, splice donor; SA, splice acceptor; CD8, CD8 transmembrane domain; arrows, start of transcription. (b) 1928zIRES-extGLuc⁺ T cells demonstrate similar cytotoxic potential when compared to 1928z⁺ T cells as assessed by a standard 4 h ⁵¹Cr release assay targeting NALM-6 tumor cells. Effector cell number represents the CD8⁺ transduced T cell fraction. Assay was performed in triplicate wells with percent cytotoxicity varying <10% within each data set. Infused T cell populations in both treatment groups retained a central memory phenotype (CD62L^hiCCR7⁺) (data not shown). (c) SCID-Beige mice bearing systemic NALM-6(GFP-FFLuc) tumor cells demonstrate diffuse tumor involvement by BLI on d 0. Following T cell infusion, T cells were imaged serially by BLI. 1928zIRES-extGLuc⁺ T cell signal is noted at tumor sites in the bone marrow, liver, spleen and lymphnodes on d 1 and 3. Reduced signal is noted at these sites on d 7 and 10, likely due to tumor cell eradication as evidenced by tumor cell BLI at d 11. Persistent signal appreciated in the abdomen of 1928zIRES-extGLuc⁺ T cell treated mice represents residual T cells in the liver. These data are consistent with the notion that the 1928z CAR specifically mediates T cell accumulation at sites of CD19⁺ NALM-6 tumor. Data represents one of three experiments with similar results. (d) Immunohistochemistry of mouse tissues d 4 and d 11 following T cell infusion.