Dual nanobody-redirected and Bi-specific CD13/TIM3 CAR T cells eliminate AML xenografts without toxicity to human HSCs

ABSTRACT

Adoptive cell therapy including chimeric antigen receptor (CAR) T cells targeting CD19 has been approved by FDA to treat B cell-derived malignancies with remarkable success. The success has not yet been expanded to treating Acute Myeloid Leukemia (AML). We previously showed that a nanobody and single-chain fragment variable (scFv) CD13 (Nanobody)/TIM-3 (scFv) directed bispecific split CAR (bissCAR) T cells, while effective in eliminating AML in preclinical models, also caused substantial toxicity to human hematopoietic stem cells (HSCs) and other lineages. To maintain the bissCART specificity and efficacy, yet reduce toxicity to normal cells including HSCs, we generated new anti-TIM-3 nanobodies and constructed new cognate nanobodies-directed CD13/41BB and TIM3/CD3zeta nbiCARTs. The resultant nbiCARTs showed strong antitumor activity to CD13/TIM3 positive leukemic cells in vitro and in preclinical models. Importantly, the 3^rd generation of nbiCARTs had little toxicity to human bone marrow-derived colony forming progenitors ex vivo and the human HSCs in mice with a humanized immune system. Together, the current studies generated novel and 3^rd G CD13/TIM-3 nbiCARTs that displayed stronger antitumor activity yet minimal toxicity to normal tissues like HSCs that express a moderate level of CD13, paving the way to further evaluate the novel CD13/TIM-3CARTs in treating aggressive and refractory AML in clinical studies.

Introduction

Adoptive T-cell cancer therapy using chimeric antigen receptor (CAR)-expressing T cells can eradicate relapsed or refractory B-cell lymphoma or B-cell lymphocytic leukemia by targeting CD19.^1,2 This remarkable success in targeting the B cell-specific CD19 marker is yet to be translated to other types of leukemia like acute myeloid leukemia (AML), which has less than 28% of 5-year survival rate. One impediment to expanding the CAR T application to AML is often a lack of choice of tumor associated antigens (TAAs). TAAs for AML, such as CD33^3–8 and CD123,^7,9–11 while effective as targets for CART therapy in preclinical models, are often expressed in other tissues including human hematopoietic stem cells (HSCs)^7,12 thus rendering the CD33-CARTs or CD123-CART treatments toxic to normal tissues including HSCs.^7,12

CD13 is upregulated in many AML blasts and leukemia stem cells (LSCs) as well as in HSCs,^3,13–15 but TIM-3 is often expressed in AML LSCs but not HSCs.^16,17 We previously reported that bispecific and split CD13-CD3zeta/TIM-3-41BB bissCARTs effectively eradicated human AML PDX in NSG immunodeficient mice,^18,19 indicating promising efficacy of the therapy in preclinical models. However, the CD13/TIM-3 bissCARTs also exhibited substantial toxicity toward human HSCs in the humanized mice (HIS), showing loss of about 50% of HSC-enriched population and even more of the monocyte population in peripheral blood.¹⁸ In our bissCAR, CD13 nanobody binder was linked to the CD3zeta domain¹⁸ with three ITAM signaling domains that leads to immune synapse formation and target cell cytotoxicities.^20,21 On the other hand, TIM-3 binding scFv was linked to the 41BB domain, a co-stimulatory receptor that synergizes with CD3zeta domain to elevate target specific cytotoxic effects.^18,21,22

As CD3zeta domain has a more potent and lethal signaling compared to 41BB, we previously attempted to switch the intracellular domains, i.e. CD3zeta and 41BB, between the CD13 binder (Nb157) and TIM-3 scFv by separating the split CARs by a P2A that causes expression of separate CARs on the cell’s surface.²³ The hypothesis being CD13-41BB/TIM-3-CD3zeta bissCARTs will be more lethal to the TIM-3-expressing LSCs but should be less toxic to normal HSCs that have moderate levels of CD13. However, multiple attempts failed in producing the split CD13-41BB/TIM-3-CD3zeta bissCARTs, due to the lack of simultaneous expression of both these CARs on primary human T cells. The reason for this remains unclear.

We then sought to generate Variable heavy domain of heavy-chain nanobodies (VHHs) against TIM3 to be able to replace the TIM3-ScFv. The new combinations of nanobody binders to CD13 and TIM3 as bispecific split CAR (nbiCAR) are likely to be more compatible for primary T cells expression. Indeed, the multiple new anti-TIM3 nanobodies were generated, seven of them were incorporated into CD13-41BB/TIM-3-CD3zeta nbiCARs and were conducive for co-expression in T cells. Importantly, 3 out of 7 nbiCARTs showed potent killing of the CD13/TIM-3 dual positive AML tumor cell-line xenografts (CDX) with no obvious toxicity to the mature and HSCs in HIS mice. Together, these results suggest that these three-top ranked CD13/TIM-3 nbiCARTs are promising candidates for further clinical evaluation.

Results

Generating phage display library and screening for VHHs that specifically bind to TIM-3

To develop VHHs against TIM3, Llamas were immunized with purified TIM3 protein, followed by phage library construction from Llama PBMCs. The library was counter-selected with human NB4 cells, followed by panning on NB4- TIM3⁽⁺⁾ cells (Figure 1a). The results from flow-cytometry analysis showed that phage library specifically bound to NB4- TIM3⁽⁺⁾ expressing cells, but not the NB4 cells (Figure 1b). Forty-five out of 50 clones bound to the NB4-TIM3 cells specifically (Figure 1c, Figure S1B). DNA sequencing revealed 19 unique clones. Phylogenetic analysis further uncovered seven clusters of VHHs (Figure S1C). A representative nanobody (VHH) from each of the seven clusters (highlighted in red boxes) were analyzed further for their potential role in directing the nbiCARs to specifically target tumor cells. The nbiCARTs used in this manuscript were initially denoted by VHH clone numbers, i.e. BiCAR-38, 12, 13, 28, 30, 32, and 33 each derived from the respective VHHs, for clarity, are now referred to as nbiCAR-7,6,5,4,3,2 and 1, respectively.

Figure 1.

Screening VHH phage display libraries and identifying TIM3 binding VHHs. (a) Illustration of immunization of llama with human TIM-3 protein, construction of the phage VHH library, TIM3-specific VHH phage screening with TIM-3 positive and TIM-3 negative NB4 cells. (b) The specific binding of phage library to TIM3+ cells. (c) Isolation and identification of individual VHH phage clones.

Generation of dual nanobody CD13/TIM-3 bispecific CAR (nbiCAR) T cells and characterization of their co-expression in primary T cells

To determine whether anti-TIM3 VHHs can direct T cells to kill tumor cells, the VHH sequences of the above described seven VHHs were used to generate the nbi-CAR, wherein Nb157 recognizing CD13 was linked to 4-1BB costimulatory domain, while anti-TIM3 VHHs recognizing TIM3 was linked to CD3z (Figure 2a–b). The resulting nbiCARs, including Bi-CAR-3, Bi-CAR-2, Bi-CAR-1, were used for lentiviral transduction in primary human T cells. Flow-cytometric assay using recombinant His tag-CD13 protein showed that each of the seven nbiCARs on the transduced T cells bound to the CD13 protein, but not to control untransduced (UTD) T cells (Figure 2c). Consistently, six out of the seven nbiCARs showed obvious binding to the second TAA target of the nbiCARs, i.e. recombinant TIM3 protein with His tag (Figure 2d), except for nbiCAR-7 with a minimal binding of 3.35% only (Figure 2D). The VHH expression was measured for three donor T cells as shown in Figure S3B. These results demonstrated that most of the anti-TIM3 VHHs, unlike the previously reported anti-TIM3 ScFv could be co-expressed along with anti-CD13 (Nb157) in the nbiCARs successfully in primary T cells.

Figure 2.

Construction of nbiCARTs using Nb157 & anti-tim VHHs (a) construction of bi-CARs using newly generated anti-TIM3 VHHs. (b) Expression of the Bi-CARs on primary human T cells as detected by flow assay for binding of ectopic CD13 protein (left panel) or TIM3 protein (right panel).

New CD13/TIM3 nbiCARTs specifically and potently eradicate CD13⁺TIM3⁺ tumor cells

To test the efficacy of the nbiCARTs cells we used the NB4 cell line that expresses CD13 protein. NB4 stable cells were created with CD13 knockout denoted as NB4-CD13 (KO). In addition, both the NB4 and NB4-CD13(KO) cells were used to overexpress TIM3 denoted as NB4-TIM3 and NB4-TIM3-CD13(KO), respectively. These cell lines were verified using western blots and flow cytometric analysis as shown in Figure 3a,b respectively. The aforementioned cell lines with the expression of either CD13 or TIM3 or both CD13/TIM3 were used to test the cytotoxicity effects of the seven nbiCARTs.

Figure 3.

Cytotoxicity assay by the three nbiCARTs on killing the various NB4 cells CD13-TIM-3 null or CD13 and/or TIM-3 expression. (a) Western blot for detecting target expression of TIM3, CD13 using antibodies. Histone 4 was used as loading control (b) flow-cytometry assay detecting the target expression of various NB4 stable cell lines. Graphical representation of LDH assays with bi-CARTs co-cultured with NB4 (c), NB4-CD13KO (d), NB4-TIM3 (e) and NB4-TIM3-CD13KO (f) NB4-CD13KO cells n = 4, *P  < 0.05 **P  < 0.003 ***P  < 0.0001 ns means there is no difference. A two-way anova was used to compare means of the two groups E: T 1:1 and E: T 3:1 with their respective UTD.

In vitro cytotoxicity assay, using LDH release assay showed that each of the seven nbiCARs killed NB4 (CD13+TIM-) target cells (Figure 3c). However, the NB4-CD13 (KO) cells were resistant to the nbiCARs-induced cytotoxicity as expected (Figure 3d), indicating a CD13-dependent killing by most of the tested nbiCARTs. Coculture of the 7 nbiCARTs with NB4-TIM3 cells led to significant killing of the dual positive target cells, except for nbiCART-7, in a dose-dependent manner (Figure 3e).

Of note, co-culture of each of the seven nbiCARTs with NB4-TIM3-CD13KO cells also showed that these nbiCARTs, except for nbiCART-7, displayed toxicity to the TIM3 single positive target cells, in a dose-dependent manner (Figure 3f). Un-transduced T (UTD) cells were used as negative control in co-culture with the tumor cells (Figure S3A & Figures 3c–f)). It is likely that nbiCAR-7 CARTs had lower or no toxicity to the target cells because of lower nbiCAR expression on the T cells surface, as detected in CD13 or TIM3 protein-binding assay (Figure 2c,d). These results indicate that all the six distinct nbiCARTs can kill the tumor cells in a CD13-dependent manner, but when TIM3 is expressed, even without CD13 target expression, 6 out of 7 nbiCARTs (except for nbiCAR-7 biCARTs) retain the killing potential.

CD13/TIM3 nbiCAR T cells eliminated AML tumor xenografts (CDX) in preclinical models

Furthermore, to determine the efficacy of nbiCAR T cells in AML CDX-xenografts in vivo, NB4 or NB4-TIM3 cells were subcutaneously transplanted into flanks of the NSG mice, respectively (Figure S2A). As expected, out of the six nbiCARTs injected three nbiCAR 4, 5 and 6 showed only a modest response whereas the other three CARTs namely nbiCAR 1, 2 and 3 had impressive tumor suppression as demonstrated in the NB4-TIM3 tumor bearing mice (Figure S2C). The corresponding CDR amino acid sequences for the same are shown in Figure S3C. Also, the NB4 cells without TIM3 did not show tumor suppression (Figure S2B). As such, the three nbiCARTs 1,2 and 3 were chosen for further investigation.

In further studies, NB4-CD13 (KO) cells, NB4 cells, and NB4-TIM3 cells were subcutaneously transplanted into NSG mice, respectively, followed by treatment with the three selected nbiCARTs (Figure 4a). NB4-CD13 (KO) tumor CDX xenografts that lacked both CD13 and TIM3 indicated no tumor suppression by the nbiCAR T cells as expected (Figure 4c). Similarly, NB4 tumor CDX xenografts that only express CD13 alone also showed exponential growth (Figure 4b). Notably, in dual positive NB4-TIM3 tumor xenograft model, injection of each of the three nbiCARTs, i.e. nbiCARs-1,2,3 but not control UTD T cells, significantly suppressed the tumor growth and led to tumor eradication by the end of study (Figure 4d). There is a notable difference in the in vitro and in vivo cytotoxicity profiles displayed by these CARTs in terms of targeting the TAA wherein the presence of CD13 alone elicits modest killing in vitro. However, in vivo data show that the nbiCARTs are highly efficacious in eradication of CD13 and TIM3 double positive tumor CDX xenografts but not CD13 single TAA displaying xenografts.

Figure 4.

The specific effects of the various nbiCARTs in tumor suppression in vivo. (a) Flowchart of establishing xenograft model in NSG mice and transfusion with UTD T cells or biCarts. The anti-tumor effects of Bi-CARs T cells against NB4-CD13KO (b), NB4 (c) and NB4-TIM3 (d) tumors were evaluated. Various NB4 cells were injected subcutaneously into each flank of a NSG mice (n = 3, 6 tumors), and the indicated CARTs (5 × 10⁶) were injected into each mouse via tail vein at 7 days after tumor cell injection. (**P  < 0.01 versus UTD group, two-way ANOVA)). The number of human T cells in peripheral blood of mice bearing NB4-CD13KO, NB4 and NB4-TIM3 tumors was measured at 7 days (e) and 14 days (f) after T cells infusion by flow cytometry. (**P  < 0.01 versus UTD group, ns means there is no difference, two-tailed unpaired Student’s t-test).

Consistently, the nbiCAR T-cell numbers in peripheral blood of dual positive NB4-TIM3 tumor-bearing mice were significantly higher than in NB4-CD13(KO) or NB4 tumor-bearing mice at 7 days (Figure 4e) and 14 days (Figure 4f) after T cells infusion. These results suggest that nbiCARTs were activated and amplified upon interaction with the dual TAA positive tumor cells. Together, these results indicate that the three nbiCARTs were each capable of eliminating the AML tumor xenografts that expressed CD13/TIM3. This data is promising as clinical data analysis of 173 AML samples from TCGA database confirms the dual expression of both CD13 and TIM3 in many subjects (Figure S1A), thereby qualifying nbiCARTs as candidates for ‘AND’ gate bispecific CARTs.

To further confirm our observations, we investigated the cytotoxic effects of the nbiCART-3 using Kasumi cells that naturally express both CD13/TIM3 proteins. The in vitro cytotoxicity assay was performed by analyzing the numbers of effector and target cells using CD3 and CD33 antibodies respectively using flow cytometry. The results showed that the Kasumi 6 tumor cells were unaffected by the negative control UTD cells but the nbiCART-3 significantly killed the target cells as shown in Figure 5a. Consistent with this data, we also found that the nbiCART-3 cells showed a significant target cell-dependent increase in T cell numbers as compared to the UTD cells (Figure 5b). The cytokine release results further showed an enhanced IFNγ secretion in a dose-dependent manner with the increasing effector to target cell ratios (Figure 5c).

Figure 5.

Evaluation of the efficacy of nbiCART-3 with the lowest potential cytotoxicity using in vitro assays. (a) Graphical representation of flow cytometry analysis of number of kasumi-6 tumor cells counts when co-cultured with nbiCART-3 cells at the indicated time points. (n = 3, donors; 3 replicates/group) two-way anova test was used to test statistical significance ****p < 0.0001; *p = 0.03. Error bars indicate mean SD. (b) Graph of the nbiCART-3 in co-culture with the kasumi 6 tumor cells quantitated using flow cytometry at the indicated time points to show increase in the T cells. UTD (untransduced cells) alone or UTD with kasumi 6 cells serve as controls. Two-way anova test was used to test the statistical significance p < 0.0001 for the indicated groups. (n = 2 donors;3 replicates/group). Error bars indicate mean-sd. (b) (c) cytokine release of nbiCART-3 upon stimulation with Kasumi6 cells. Graphical representation of duo-elisa showing IFNg release in the co-culture of the nbiCART-3 with kasumi-6 tumor cells at the indicated ratios. Test of significance between the pairs were done using two-way anova on three independent (n = 3 donors; quadruplicate/group) experiments. ns-non-significant; *p = 0.0021; **p < 0.0001.

CD13/TIM3 nbiCARTs show no obvious toxicity to bone marrow progenitors or HSCs ex vivo

Colony-forming assay was performed to determine the potential toxicity of the nbiCARTs toward normal human hematopoietic stem cells (HSCs) and progenitors. Human bone marrow-derived CD34⁺ cells were cocultured with control UTD T cells, or single NbCD13-CAR, or each of the nbiCARTs ex vivo, and then plated on methocult semi-solid medium, to quantify the colony formation units (CFUs) (Figure 6a). Compared to the UTD T cells, CD13 targeting single NbCD13CAR significantly reduced the number of CFU in a dose-dependent manner (Figure 6b), as expected and as we previously reported.¹⁸ nbiCART-1 and nbiCART-2 cells moderately, yet significantly reduced the clone formation in an effector/target cell-ratio dependent manner (Figure 6b–c). Importantly, nbiCART-3 cells showed no obvious cytotoxicity to bone marrow CD34⁺ CFUs in the Figures 6(b–c). These results based on the ex vivo assay suggest that nbiCART-3 cells are a good candidate for further investigation in the category of reduced toxicity to HSCs.

Figure 6.

Evaluation of the potential toxicity of the nbiCAR T cells against human bone marrow CD34+ cells. (a) BM CD34+ cells (5000 per well) co-cultured with the indicated T cells (0.3:1, 1:1, 3:1) for 4 hours. Then the cells were transferred to 12-well plates and cultured in MethoCult™ H4435 enriched medium. Two weeks later, the number of clones was measured. (b) The number of the colonies from control (UTD) or the biCART treated plates. (c) A graph showing the dose-dependent effect of the CARTs on the colony number. (n = 3, **P  < 0.01 versus UTD group, ns means there is no difference. two-tailed unpaired student’s t-test).

CD13/TIM3 nbiCARTs show no obvious toxicity to human HSCs and differentiated hematopoietic lineages in humanized mice in vivo

Next, we investigated the impact of the three nbiCARTs on normal hematopoietic differentiated lineages and HSC-enriched populations in mice with humanized immune system (HIS), to assess potential toxicities (Figure 7a). NSG mice were conditioned with busulfan and engrafted with human bone marrow CD34⁺ cells from a normal adult donor, followed by treatment with each of the three nbiCARTs. The binding of the nbiCARs to both CD13 and TIM3 his tagged proteins is shown in Figure 7b. Peripheral blood (PB) and bone marrow from these mice were analyzed 3 weeks post-treatment. nbiCARTs were detected in the peripheral blood PB as compared to the UTD cells (Figure 7c–d).

Figure 7.

In vivo humanized mice assay of the potential toxicity of three nbiCAR T cells. (a) Schematic diagram of HIS mice for evaluation of human HSC toxicity. A total of 1.5 million normal donor bone marrow (BM) CD34+ cells was injected into each NSG mouse. Nine weeks later, 3 million Bi-car T cells or UTD T cells were injected via tail vein, followed by flow cytometry analysis of peripheral blood and bone. Total of 23 mice were used 5 mice per group for all except the CD13 CAR group with 3 mice. (b) Expression of the nbiCARs on primary human T cells. (c and d) the percentage of CAR-T cells in total T cells in PB.

In parallel studies, we also analyzed changes in human CD19-expressing B-cell lineage and CD33+ myeloid lineage (using human CD45+ hematopoietic cells) from the PB of mice, using a flow cytometric assay. While the absolute percentages of CD19+ and CD33+ lineages were relatively low, partly reflecting marked expansion of injected human CD45+ T cells and relatively moderate differentiation of the mature lineages. These results indicate three nbiCARTs showed no reduction in the percentage of human CD19⁺ B cells (Figure 8a–b) and CD33⁺ peripheral monocytes (Figure 8a–b). We further detected CD34⁺CD38⁺ progenitors and CD34⁺CD38⁻ HSCs from the bone marrow of these mice by using flow-cytometry assays. The results showed that none of the three nbiCARTs, reduced the percentage of HSCs-enriched CD34⁺CD38⁻ cells and CD34⁺CD38⁺ myeloid progenitors (Figure 8c–d) as compared to the UTD control. In summary, these findings indicated that the three selected nbiCARTs did not show any obvious toxicity to the detectable B-lineage and CD33+ myeloid lineage, nor CD34+CD38-enriched HSCs in HIS mice. As the percentage of CD33+ cells in peripheral blood are relatively low, a thorough evaluation of the percentage in bone marrow is desirable in future investigation.

Figure 8.

In vivo humanized mice assay of the potential toxicity of three nbiCAR T cells. (a) Peripheral blood of HIS mice, which were treated with T cells for 3 weeks, was analyzed by flow cytometry after staining with CD33/CD19/7-AAD. (b) Representative fluorescence-activate cell sorting plots were used to identify B cells (CD19+CD33-) and monocyte (CD19-CD33+). (c) Bone marrow of HIS mice treated with T cells for 3 weeks were analyzed by flow cytometry after staining with CD45/Lin/CD34/CD38/7-aad. (d) Representative fluorescence-activate cell sorting plots were used to identify myeloid progenitors (CD34+CD38+) and HSC (CD34+CD38-).

Discussion

Bispecific and split CARTs (nbiCARTs) that target two distinct TAAs may broaden the spectrum of the tumor target cells. nbiCARTs have enhanced efficacy to eradicate the target cells with dual TAA expression, yet with no obvious toxicity to normal cells with low to moderate level of one of the TAA (CD13). This scenario results from a delicate goldilocks balance of nbiCARTs in adequately suppressing the target tumor cells while sparing the normal tissues. Hirabayashi et al. reported that bissCAR targeting TAAs GD2 and B7-H3 in solid tumors enhances antitumor activity in preclinical models, even when one of the two TAAs lost its expression in target tumor cells.²⁴ Moreover, co-targeting CD33 and CD123 in AML by biCARTs showed obvious anti-tumor activity with no off-tumor effects observed in mouse model-related preclinical studies,²⁵ even though both CD33 and CD123 are expressed in certain normal cells like HSCs. It is therefore possible to have a biCAR with lower toxicities to the normal cells while retaining the potential to kill the tumor cells that express both the TAAs.

We previously reported that CD13-CD3zeta/TIM3scFv-CD28-41BB bissCARTs eradicated AML in mouse models,^18,19 yet caused substantial toxicity to normal human hematopoietic cells as compared to single TAA targeting CD13CARTs.^18,19 CD13 is found on the human myeloid cells as well as HSCs and hence our previous bissCART with CD13 binder expressed with CD3z caused toxicity to the normal tissues. Multiple attempts were made constructing a TIM3-CD3zeta/CD13-CD28-41BB bissCARs, which may reduce toxicity due to the particular ordering of these domains. However, we failed at this attempt due to lack of co-expression of these domains on T cells possibly owing to improper protein folding or unstable biCAR expression.

This challenge was overcome by the development of nanobodies for TIM3 to replace the TIM3ScFv from the bissCARTs to create nbiCARs with nanobody binders for CD13/TIM3. This suggests nanobodies or distinct VHH binders have increased chances of co-expression as bi-split CARs in primary T cells.

It is noteworthy that most of our tested CD13/TIM3 nbiCARTs, except for nbiCAR7 are potent against the dual positive target cells (Figure 3). The three distinct nbiCARTs, i.e. nbiCART-1, nbiCART-2 and nbiCART-3 potently eradicated CD13/TIM3 dual positive tumor cells in vivo (Figure 4). This correlated with increased CAR T cell numbers in PB of the dual positive tumor bearing mice, as compared to the single CD13 positive tumor-bearing mice. These results suggest that the dual positive AML are most susceptible to the three selected nbiCARTs. Nevertheless, it cannot be ruled out that a higher dose of the nbiCARTs can further suppress or eliminate even the single TAA containing tumor xenografts.

Importantly, injection of each of the three nbiCARTs into the HIS mice did not reduce the human B-lineage and myeloid cells in the peripheral blood. Consistently, our results also demonstrate that the three distinct CD13/TIM3 nbiCARTs did not reduce the number of the human HSC-enriched population in vivo (Figure 8), even though nbiCART-1, nbiCART-2 and nbiCART-3 showed moderate toxicity in ex vivo CFU assay (Figure 6). These results indicate that ex vivo assay may be too sensitive in evaluating the potential toxicity of the nbiCARTs. However, a thorough investigation of the impact of the CAR T cells on the bone marrow CD33+ cells is important in future studies to determine the potential toxicity to the myeloid lineage cells. As each of the three distinct CD13/TIM3 nbiCARTs showed the effective antitumor activity in vivo, but none of them elicited obvious deleterious effects on the hematopoietic progenitors and HSC-enriched populations at the CART dose used, the three CD13/TIM3 nbiCARs are deemed viable candidates for further clinical studies.

Materials and methods

Nanobody phage library construction from the TIM3-immunized llama

A llama was immunized with TIM3 protein 3 times (Capralogics, Hardwick, MA). Peripheral blood mononuclear cells (PBMCs) isolation, RNA extraction and cDNA synthesis were performed as described previously.²⁶ The phage library was prepared as we described earlier using the pComb3× vector.¹⁸ The list of the antibodies used in this study is included in Table 1. The sequences of primers used for PCR and sgRNAs are listed in Table 2.

Table 1.

List of reagents.

Name		Source	Product
anti-M13/fd/F1 Filamentous Phages Biotin mouse monoclonal, B62-FE2		PROGEN	Cat# 61597
APC Mouse Anti-Human CD3		BD Biosciences	Cat# 555335
anti-CD13		Abcam	Cat# ab227111
Anti-Human CD33 APC		eBioscience	Cat# 17-0338-41
Alexa Fluor® 647 AffiniPure Goat Anti-Alpaca IgG, VHH domain		Jackson ImmunoResearch Laboratories	Cat# 128-605-230
Alexa Fluor 488 Goat anti-mouse		Thermo Fisher	Cat# A-11001

Table 2.

List of primers used.

Primers for 1st round PCR of llama VHH amplification	FORWARD:
	GTCCTGGCTGCTCTTCTACAAGG
	REVERSE:
	GGTACGTGCTGTTGAACTGTTCC
Primers for 2nd round PCR of llama VHH amplification	FORWARD:
	GAGGAGGAGGAGGAGGAGGCGGGGCCCAGGCGGCCCAGGTGCAGCTGCAGGAGTCTGGRGGAGG
	REVERSE:
	GAGGAGGAGGAGGAGGAGCCTGGCCGGCCTGGCCACTAGTGGCGGCCGCTGAGGAGACGGTGACCTGGGT
sgRNA targeting sequences: hCD13:	ATGGCCGGCTCATCGAAGCA, CTTCCCATGCTTCGATGAGC, CTTCATGGGGCCATAGACCT
sgRNA targeting sequence: hTIM3:	ATGAGAATACCCTAGTAAGGGGG

Construction of the nanobody CAR vector

Generation of the Nanobody CAR (Nb CAR) constructs in lentiviral vector is shown in Figure 2a. pHIV, a third-generation self-inactivating lentiviral vector plasmid was used. Nb CAR constructs were custom synthesized by IDT.

In vivo studies performed using mice models

NOD/Shi-scid/IL-2 Rγnull (NSG) mice, 8–12 weeks old (Jackson Laboratories) were injected with 1 × 10⁷ cells of NB4 subcutaneously. When tumor volume reached 100 mm³ nbiCAR T cells or untransduced (UTD) human T cells (5 × 10⁶ cells) were administered via tail vein.

All experiments on mice in our research protocol were approved by IACUC at the University of Pennsylvania abiding by relevant institutional and national guidelines and regulations.

Cytotoxicity and proliferation assays

The co-culture of Kasumi6 tumor cells with CARTs was performed using 50,000 cells each in a 48-well assay plate. The cells were seeded for the time points that included Day 0, 2, 5 & 7 and a flow cytometric analysis was performed to quantitate both the tumor cells and T cells using CD33-APC antibody (Biolegend-Cat No 303,408) and CD3-Alexa 488(BD Pharmingen Cat No 557,694) antibody respectively. The total volume of cell suspension was counted using the Accuri6 flow cytometer to a volume of 150 μl. The exact volumes were then used to express the cells per μl volume for ease of representation.

Colony formation assay

Donor derived CD34+ Bone Marrow cells were seeded at 5000 per well and co-cultured with the indicated T cells (0.3:1, 1:1, 3:1) for 4 h. Then the cells were transferred to 12-well plates and cultured in MethoCult™ H4435 enriched medium. The CFUs were counted after 2 weeks to compare the CART mediated toxicity. The bone marrow cells were obtained from a donor from the Penn Human Immunology Core (HIC).

Humanized immunodeficient NSG mice and nbiCART toxicity studies in the HIS mice

BM sample from a different donor (HIC core facility) was collected to create HIS mice as described earlier.¹⁸ 1.5 × 10⁶ of bone marrow-derived CD34+ cells were used. Nine weeks later, 3 million CAR T cells or UTD T cells were transduced into the mice. Each group namely UTD, nbiCAR-3, nbiCAR-2, and nbiCAR-1included (5 mice/group). The mice were sacrificed 3 weeks after initial treatment, peripheral blood and femur were collected for analysis.

Statistical analysis

Microsoft Excel and GraphPad Prism software were used for statistical analysis. Student’s t test or ANOVA were used to determine the significance of the results unless otherwise indicated. All error bars are represented in SD.

Funding Statement

The work was supported by the National Cancer Institute and Harrington Discovery Institute Takeda Rare Disease Scholar award; Funding agency NIH- award no. 5-R01-CA-259998-03.

Disclosure statement

Hua, X and Zhang, XY are the inventors for a patent application for the CD13/TIM3 CARs. Hua, X is a consultant for Chimeric Therapeutics.

Author contributions

X. Y. Zhang and X. Hua designed most experiments and generated the TIM3 VHHs and the respective CAR constructs. Z.J. Feng designed and performed the experiments on evaluating the impact of the CAR T cells on human hematopoietic cells. X. Y. Z., Z. J. F., A. Pranatharthi-Haran, X. H. designed and performed experiments analyzed data and generated figures. X. Z., Z. F. and A. P-H., X. Hua prepared the manuscript. Hua conceived and supervised the project. All authors commented and revised on the manuscript and approved the paper. A. P-H. submitted the manuscript.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/2162402X.2025.2458843