Efficient in vivo generation of CAR T cells using a retargeted fourth-generation lentiviral vector

Abstract

Chimeric antigen receptor (CAR) T cell therapy has proved remarkably successful for the treatment of hematological malignancies. However, the bespoke manufacturing of autologous CAR T cells is complex and expensive. The development of methods for in vivo engineering of T cells will enable generation of CAR T cells directly within the patient, bypassing the need for ex vivo manufacturing and thereby enabling greater access for patients. Here, we describe development of an improved retargeted Nipah envelope system paired with a fourth-generation lentiviral vector capable of specifically targeting T cells with increased efficiency, which generates high levels of functional CAR T cells in vivo. The retargeted vectors exhibited greater specificity to T cells compared to the VSV-G pseudotyped vector. Vectors targeted to either CD3 or CD8 similarly generated high levels of CAR T cells, which rapidly eradicated B cells, suggesting that T cell receptor (TCR) engagement is not required for lentiviral vectors to efficiently transduce T cells in vivo. Furthermore, the fourth-generation lentiviral vector platform (referred to as the TetraVecta system) employs the TRiP system to prevent incorporation of CAR protein into the vector particles, minimizing the risk of inadvertent transduction of tumor cells.

Keywords: CAR T cells; gene therapy; immunotherapy; in vivo CAR T; lentiviral vectors.

Graphical abstract

Figure 1

Use of a DARPin and the mixed envelope composition improved the transduction efficiency of retargeted LV vectors pseudotyped with a modified NiV envelope (A) Schematic representation depicting the overall domain organization of the modified Nipah G proteins used for retargeting the LV vectors in combination with a modified Nipah FΔ22 protein (F, not shown). The mixed envelope is composed of a mixture of both (i) the retargeted NiV GmΔ34 protein including a targeting moiety to a cell-specific receptor fused to the carboxy terminus and (ii) the non-targeted NiV GmΔ34 without a targeting moiety. Both versions contain four point mutations to prevent binding to the native targets (E501A, W504A, Q530A, and E533A, indicated with asterisks) and a truncation of the cytoplasmic tail to enhance incorporation into the vector particle, as described previously by Bender et al. CT, cytoplasmic tail; TM, transmembrane domain; ED, ectodomain; L, linker; BD, binding domain; His, His-tag. (B) Schematic representation of the composition of the retargeted NiV envelope (i) and the mixed envelope (ii) composed of a mixture of retargeted NiV GmΔ34 (fused to a targeting moiety) and non-targeted NiV GmΔ34. (C) Impact of the mixed envelope composition on the transduction efficiency of GFP vectors pseudotyped with a retargeted NiV envelope composed of a retargeted CD8-NiV GmΔ34 attachment protein displaying either scFv or DARPin binders for CD8 (CD8^scFv-NiV and CD8^DARPin-NiV GmΔ34, respectively). The composition of the G protein component was either 100% retargeted CD8-NiV GmΔ34 or a mixed envelope with 75% retargeted CD8-NiV GmΔ34 and 25% non-targeted NiV GmΔ34 proteins. Transduction efficiency was evaluated by flow cytometry on CD8⁺ T cells 12 days after transduction (n = 2). Relative efficiency is shown compared to the CD8^scFv-NiV 100% vector. (D) Impact of different proportions of retargeted CD8^DARPin-NiV GmΔ34 and non-targeted NiV GmΔ34 on the transduction efficiency of a GFP LV vector. The proportion of retargeted CD8^DARPin-NiV GmΔ34 and non-targeted NiV GmΔ34 proteins was varied from 0% to 100% and 100% to 0%, respectively. Flow cytometric analysis of GFP expression in CD8⁺ and CD8⁻ negative populations 12 days after transduction (n = 2). Axis labels refer to the proportion of G protein composed of the retargeted CD8^DARPin-NiV GmΔ34 with the remaining fraction made up of non-targeted NiV GmΔ34. Data shown as mean ± SD. Statistical significance calculated by unpaired two-tailed t test and shown as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. See also Figure S1.

Figure 2

Specific transduction of CD4⁺, CD8⁺, and CD3⁺ T cells with retargeted LV vectors composed of a mixed NiV envelope using scFv, DARPin, or VHH targeting moieties Flow cytometry analysis of T cells transduced with CD19 CAR LV vectors using a mixed envelope with a ratio of 75% retargeted NiV GmΔ34 and 25% non-targeted NiV GmΔ34 incorporating a variety of different targeting moieties. Flow cytometry plots of CAR expression and target receptor expression for T cells transduced with vectors pseudotyped with the following envelopes: (A) CD8^scFv-NiV, (B) CD8^DARPin-NiV, (C) CD8^VHH1-NiV, (D) CD8^VHH2-NiV (humanized VHH), (E) CD3^scFv-NiV LV, (F) CD3^DARPin-NiV LV, (G) CD4^DARPin-NiV, and (H) VSV-G. Flow cytometry analysis performed 12 days after transduction with matched particle numbers (1.5E+10 RNA copies). (I) Impact of different proportions of retargeted CD3^DARPin-NiV GmΔ34 and non-targeted NiV GmΔ34 proteins on the transduction efficiency of a CD19 CAR LV vector with a mixed envelope. Axis labels refer to the proportion of retargeted CD3^DARPin-NiV GmΔ34 as percentage of the overall amount of G protein, with the remaining fraction made up of non-targeted NiV GmΔ34. Flow cytometric analysis of CAR expression in the CD3⁺ population 12 days after transduction (n = 2). Data shown as mean ± SD. Statistical significance calculated by unpaired two-tailed t test and shown as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Figure 3

Production of retargeted lentiviral vectors using the SupA2KO-LV genome and TRiP system reduced vRNA splicing and prevented CAR incorporation into the vector particles Vectors were generated using the retargeted NiV mixed envelope with 75% retargeted CD8^DARPin-LV NiV GmΔ34 and 25% non-targeted NiV GmΔ34 (CD8^DARPin-LV (75/25)) and the following CD19-41BB-CD3ζ CAR-expressing LV genomes: third-generation HIV sin-LTR genome without tbs (3^rd Gen), or with tbs (3^rd Gen + tbs), or using fourth-generation TetraVecta system with 2KO genome, SupA-LTR, and tbs (SupA 2KO). Vectors were produced in either 1.65S cells (without TRAP) or in NTRP10 cells (with TRAP). (A and B) RT-PCR analysis of splicing of viral genomic RNA (vRNA) transcripts in (A) end of production cells (EOPC) and (B) vector particles. Black arrowhead indicates full-length vRNA; white arrowhead indicates aberrantly spliced vRNA. (C) Western blot analysis of CAR and p24 incorporation into vector particles assessed using the Jess system.

Figure 4

Rapid and efficient generation of CAR T cells in vivo by intravenous injection of fourth-generation SupA2KO-LV vectors targeted to CD3⁺ or CD8⁺ T cells (A) Outline of the in vivo gene-transfer study. Humanized mice (NCG mice engrafted with human CD34⁺ hematopoietic stem cells) were injected subcutaneously (s.c.) with IL-7 on day (D) −4 and D−1 prior to intravenous (i.v.) injection of 200 μL of LV vectors or TSSM vehicle on D0. Peripheral blood (PB) samples were taken at D−7 to assess humanization and at weekly time points (D7, D14, D21, and D28) after treatment for flow cytometry analysis. Study was terminated at D28 and bone marrow (BM), spleen, and liver were taken for analysis. (B and C) Flow cytometric analysis of proportion of CAR⁺ T cells over time. (B) CAR⁺ % of CD3⁺ cells; (C) CAR⁺ % of CD8⁺ cells. Statistical significance compared to TSSM vehicle control assessed by mixed-effect model with Geisser-Greenhouse correction and Tukey’s multiple comparison test. (D) Peak levels of CAR⁺ T cells generated by each vector (average of the maximum CAR⁺ percentage of CD3⁺ cells for each mouse). (E and F) Proportion of CAR⁺ T cells (% of CD3⁺ cells) in BM (E) and spleen (F) at termination (D28). Data shown as mean ± SEM (n = 6–12 per group). Individual data points also shown for (D)–(F). Significant results indicated (Kruskal Wallis test with Dunn’s multiple comparisons). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. See also Figure S2.

Figure 5

CAR T cells generated in vivo using fourth-generation retargeted SupA2KO-LV vectors induce rapid and sustained B cell aplasia (A) B cell levels in the blood over time. For simplicity, only significant results in comparison to vehicle/TSSM are indicated, colors correspond to the groups as shown in the legend (mixed-effects analysis). (B and C) B cell levels in BM (B) and spleen (C) analyzed after termination (D28). Significant results shown for comparison to vehicle/TSSM group (Kruskal-Wallis with Dunn’s multiple comparisons test). Data shown as percentage of CD20⁺ B cells in the hCD45⁺ cell population; mean ± SEM (n = 6–12 per group). Individual data points also shown for (B) and (C). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. See also Figure S3.

Figure 6

Specific targeting of T cells in vivo with fourth-generation retargeted SupA2KO-LV vectors (A) Proportion of CAR⁺ CD8⁺ and CD4⁺ T cells in the spleen. (B–D) Assessment of CAR expression in other immune cells in the spleens of the animals. Proportion of CAR⁺ (B) CD56⁺ NK cells, (C) CD14⁺ monocytes, and (D) mCD45⁺ cells. (E) Vector copies per μg of DNA in the liver assessed by qPCR. Data shown as mean ± SEM (n = 6–12 per group); individual data points also shown for (B)–(E). Statistical analysis performed using Kruskal-Wallis with Dunn’s multiple comparison test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. See also Figure S4.