CAR-NK Cell Immunotherapy

NK cell activity is completely independent of antigen and HLA; they can be obtained from unrelated allogeneic HLA-incompatible donors. This HLA incompatibility between donor NK cells and recipient tumor cells may promote NK cell antitumor alloreactivity through KIR/KIR ligand mismatch (see Section 1 ) . Interestingly, unlike allogeneic T cells, NK cells do not cause any GVHD. The reason for this phenomenon is not fully understood, but activator receptors may be key. Healthy non-stressed ( e.g., skin, gut ) host tissues lack stress-induced ligands on their surfaces, rendering them immune to NK cell toxicity. In an HLA-incompatible mouse HSCT model, donor NK cells even exerted protective effects against GVHD by destroying the recipient's antigen-presenting cells. This effect on the recipient's immune system could facilitate transplantation and make allogeneic NK cells less susceptible to immune rejection .

“Off-the-shelf” allogeneic NK cells present huge advantages in terms of efficiency from laboratory to bedside and cost  [24, 28, 29] ( Table 1 ). Despite all these advantages of CAR-NK cells, there are still some limitations.

(1) First, their survival time in the body is relatively short, about 1-4 weeks. In a clinical trial of short-term ( 12-16 hours ) IL-2 activation of CD3 - CD56 + NK cells from familial haploidentical donors in patients with refractory AML , lymphodepletion with fludarabine and cyclophosphamide was shown to have an effect on NK Expansion of cells is critical, and the presence of IL-15 is a predictor of expansion and clinical response  [30] . In vivo expansion of NK cells also requires in vivo administration of IL-2, which has significant toxic effects. Adding cytokine transgenes ( such as IL-2 and IL-15 ) to the construction of CAR-NK cells has the potential to improve expansion efficiency and persistence  [28] .

(2) Second, the limited number of NK cells in the blood requires ex vivo expansion techniques in order to obtain the necessary dose using apheresis after harvest. Different methods for ex vivo expansion and activation of NK cells have been investigated. These methods involve incubation with cytokines ( mainly IL15 and IL2 ) for several days ( usually 14 days ). IL-2 is essential and currently used to promote NK cell expansion after allogeneic NK cell delivery, while IL-15 plays a critical role in NK cell development and homeostasis. Some reports have used feeder cells such as Jurkat T lymphoblasts or Epstein-Barr virus-transformed lymphoblastoid cell lines ( EBV-LCL ) to expand NK cells on a large scale. K562 is a leukemia cell line genetically engineered to express membrane-bound forms of IL-15 and 4-1BB ( CD137L ). Recently, K562 cells have been engineered to express membrane-bound IL-21 and CD137L, showing some interesting results.

There is still controversy about which method is most suitable for efficient expansion of NK cells. Some studies have found that K562 feeder cells can induce robust and sustained proliferation ( 100 to >10,000-fold expansion after 3 to 5 weeks of culture ) [31] .

(3) Third, NK cell transformation is difficult, highly sensitive to apoptosis and low in gene expression . Currently, the most successful alternatives for introducing genes into NK cells are rapid transient expression by electroporation, or sustained but low expression by viral vectors. Therefore, the optimization of the transfection protocol remains an important factor in the development of CAR-NK therapy  [32] . Recently, lentiviral transduction of primary human natural killer cells in an ex vivo expansion setting has been improved. Optimizing genetically engineered NK cell methods will promote the widespread application of CAR-NK cells [33] .

(4) Finally, the tumor microenvironment ( TME ) reduces the metabolic activity of NK cells in hematological malignancies, constituting an important obstacle to enhancing the anti-tumor effect of NK cells [34,35] . In the same way as the incorporation of IL-15, the use of transgenes encoding chemokine receptors can facilitate CAR-NK trafficking at tumor sites. Combined use with checkpoint inhibitors such as anti-PD1/anti-PDL1 can help overcome the inhibitory effect of TME, thereby unlocking the full potential of CAR-NK cells. This effect appears to be successful in preclinical studies in mouse models but still needs to be demonstrated in clinical studies in patients  [36,37,38] . It is also important to note that most of the data on checkpoints blocked by CAR cells comes from experience with CAR-T cells and very little experience with CAR-NK cells.

Comparison of the advantages and disadvantages of autologous CAR-T cells and allogeneic CAR-NK cells

Source of CAR-NK cells

A major advantage of allogeneic CAR-NK cells is their ready availability, which relies on a good cryopreservation process before it is possible to generate large homogenous batches from a specific cell source. Many NK cell sources have been used to produce allogeneic CAR-NK cells.

NK-92 cell line

NK-92 is an interleukin-2-dependent, EBV-positive natural killer cell line derived from the peripheral blood mononuclear cells of a 50-year-old white man with rapidly progressive non-Hodgkin lymphoma. Cytotoxicity. Their phenotype includes activation of the receptors NKp30, NKp46 and NKG2D. Interestingly, the only inhibitory receptors on the surface of NK-92 are represented by ILT2, CD94-NKG2A ( which recognizes HLA-E ), and a lesser expressed inhibitory KIR receptor, KIR2DL4. They are very convenient to use because CAR-NK cells can theoretically be mass-produced and "infinitely" generated, with huge gains in cell number, time, and cost . However, due to their tumorous nature, these CAR-NK cells derived from the NK-92 cell line need to be irradiated before injection, which alters their lifespan and division. Although irradiated, they appeared to retain their cytotoxic capacity, but their in vivo proliferative capacity was drastically altered, disappearing within 7 days of injection, requiring multiple injections.

Some evidence for the use of CAR-NK cells in human patients with CD33-CAR-NK cells derived from the NK-92 cell line has been reported in three patients with relapsed and refractory AML after lymphodepleting chemotherapy [ 39 ] . These three patients received salvage chemotherapy followed by infusion of anti-CD33 CAR-NK cells in increasing doses ( three increasing doses of irradiated CAR-NK-92 cells per patient ). Up to 5×10 9 CD33-CAR-NK cells were safely administered per patient with no significant side effects or grade 3-4 toxicity; only transient fever was reported. However, even when the treatment was well tolerated and the safety of repeated infusions was demonstrated, the efficacy response obtained was very short-lived , ranging from 10 days in one patient to up to 4 months in another [39] .

In vitro, the efficiency of lentiviral CD33-CAR vectors in NK-92 cells was much higher than 90%, but cytotoxicity experiments showed that it was less effective against human HL-60 promyelocytic leukemia cell line than parental NK-92 cells. The cytotoxicity is moderately enhanced [39] . CAR-NK-92 cells have also been explored against multiple myeloma cells. Jiang et al. reported that NK-92 cells transduced with anti-CD38 had strong antitumor activity against myeloma cell lines and primary myeloma cells in vitro and in a xenograft NOD-SCID mouse model [ 40 ] , similar results were observed with NK-92 cells expressing anti-CS1/SLAMF7 CAR [41] .

In order to increase the efficiency of CAR-NK cells derived from the NK-92 cell line, the 4th generation CAR originally intended for T cells was produced on a platform called UniCAR. The UniCAR system consists of two elements:

(i) CAR-NKG2 targeting the peptide epitope E5B9, an antigen not naturally expressed on the cell surface;

(ii) A bispecific component called a targeting module ( TM ). This bispecific module expresses the E5B9 antigen on one side and an antibody specific for the tumor antigen on the other side ( Figure 2 ). This construct enables contact between CAR-NKG2 cells ( via E5B9 antigen ) and target cells ( via specific antibodies ). Besides that, the effect can be stopped immediately ( on/off effect ), because the TM has a very short lifespan. Once the steady ( continuous ) infusion of TM is stopped, TM is eliminated and UniCAR is inactivated. In fact, this system enables targeted selection of a large number of tumor antibodies by changing TMs that can be administered simultaneously or in a sequential manner [42] .

NK cells derived from umbilical cord blood

Umbilical cord blood is a rich and interesting source of NK cells, as hundreds of thousands of umbilical cord blood ( UCB ) units are available from international blood banks. NK cells account for 15% to 30% of umbilical cord lymphocytes. Their phenotype is more immature than that of adult peripheral blood cells and has very specific characteristics, including low CD16 ( low ADCC capacity ), CD56 bright ( more proliferative than cytotoxic ), NKG2A+ ( an inhibitory receptor , recognizes HLA-E ) and low KIR ( less functional than its counterpart ) [43] . However, these limitations can be overcome by ex vivo expansion and activation using cytokines and transfection, and in the author's experience, CAR-NK cells derived from cord blood perform better on flow cytometry than CAR-NK derived from blood cells. The same activation profile was observed intraoperatively . Even though NK cells are more or less rare in UCB, they possess a very important proliferative capacity and are very sensitive to cytokine stimulation and can be expanded ex vivo.

The MD Anderson team led by Tezvani et al. A very encouraging phase I/II trial was recently published in 11 patients with relapsed and refractory B-cell malignancies ranging from non-Hodgkin lymphoma to chronic lymphocytic leukemia Inject CAR-NK-CD19 cells ( ClinicalTrials.gov, NCT03056339 ) from fresh UCB ( Figure 3 ). Following lymphodepleting chemotherapy ( fludarabine and cyclophosphamide ) at one of three doses used in the study ( 1×10 5 , 1×10 6 , or 1×10 7 cells per kilogram ) .

Notably, no adverse effects or toxicity were noted after UCB NK cell infusion, notably CRS or neurotoxicity, and the maximum tolerated dose was not reached in the trial. Response occurred rapidly within the first month after injection, with 7 of 11 subjects experiencing complete remission and return to CLL status in Richter transformation. Nonetheless, it is important to demonstrate the fact that half of the patients received maintenance therapy after NK cell injection, a combination regimen consisting of lenalidomide, rituximab, venetoclax, or HSCT. NK cells were detected up to 12 months after injection . Most likely, adding the IL-15 gene to the CAR construct contributes to the long-term survival of CAR-NK cells, although the mechanism of persistence has not yet been elucidated.

Furthermore, a possible and very important technical setback is the production of CAR-NK cells directly from fresh UCB, as well as their injection immediately after production, without any cryopreservation step. In fact, the use of cryopreserved cord blood from international banks for the production of CAR-NK cells should probably not be recommended due to the low efficiency of NK cell selection from cryopreserved cells. In the experience of the MD Anderson group, expansion from fresh cord blood may be an option—or rather, establishing a bank of selected cryopreserved NK cells that can be thawed and expanded for transduction. It should also be noted that the activity of CAR-NK cells decreases after cryopreservation, and under current technical conditions, CAR-NK cells should be administered immediately after culture . Even though the specific long-term efficacy of CAR-NK cells is difficult to assess, and the feasibility of this process on frozen cells has yet to be determined, the results are very promising [44] .

Figure 3. CAR-NK cells from umbilical cord blood: Structure of CAR-NK cells [45], from MD Anderson. CAR-NK cells were derived from a classic CD19 CAR with a costimulatory CD28 domain to which suicide and IL-15 genes were added.

NK cells from peripheral blood

As mentioned previously, most studies employing adoptive NK therapies ( excluding CAR-NK cells ) have used cells from haplotype-related or unrelated donors and have shown antitumor activity in children and adults with AML response [29,46] . The drawback, however, is that cells need to be collected from healthy adult volunteers, making the process slightly more complex than using UCB. In the presence of cytokines, some simple, GMP-compliant culture conditions can increase their expansion by 25-55 times [49] . If NK cells from PB are highly cytotoxic compared to CB NK cells, it remains to be demonstrated whether donor selection with respect to HLA incompatibility should differ in the context of CAR-NK therapy .

The group of Leivas et al. described a method that can be used and derived from PBMC [53] . Select effector cells between activated and expanded NK cells ( NKAE ) and memory T cells ( CD45RA-T cells ). NK or T cells were transduced after culture with the NKG2D-4-1BB-CD3z-CAR vector, using the NKG2D activator receptor expressed on NK and T cells, and 4-1BB for transduction (see above ). NKG2D has specific ligands, including MICA, MICB, and ULBP, which are expressed on stressed cells and many tumor cells, particularly in multiple myeloma. In vitro, memory T cells are more stably transduced than their NK counterparts. However, CAR-NKAE cells exhibited stronger cytotoxicity against MM cells in vitro while protecting healthy cells from damage. Furthermore, in their mouse xenograft myeloma model, CAR-NKAE cells exhibited efficient cytotoxicity and enhanced anti-myeloma activity compared with T cells.

Another approach is CRISPR-targeted CAR gene insertion using Cas9/RNP and adeno-associated virus gene delivery. Using this technique, CD33-targeted CAR-NKs with distinct transmembrane and signaling domains ( CD4/4-1BB+CD3ζ and NKG2D/2B4+CD3ζ ) were generated and appeared to show enhanced activation in primary NK cells. Anti-AML activity [54] .

iPSC-derived NK cells

Induced pluripotent stem cells ( iPSC ) are differentiated cells that have been genetically reprogrammed to become pluripotent stem cells. The process of reprogramming mature differentiated cells into iPSCs involves genetically modifying the cells to express four genes: Klf4, C-myc, Oct3/4, and Sox2, thereby giving them the same characteristics as embryonic stem cells. Thus, differentiated adult reprogrammed cells acquire the potential to proliferate indefinitely and can differentiate into any cell type upon addition of specific growth factors, making it useful in many fields of medicine—particularly regenerative medicine, cell therapy or Modeling of genetic diseases . This discovery won the 2012 Nobel Prize in Medicine, awarded to Shinya Yamanaka, a researcher at Kobe University ( Japan ) [55] .

Due to their immortality, a single iPSC is sufficient to produce and generate a general CAR-NK "off-the-shelf" product. NK cells generated from iPSCs have an immature phenotype with low CD16, high NKG2A, and low KIR, with poor cytotoxicity but a very important proliferative capacity. To enhance their cytotoxicity and make them persistent, genes for expression of CD16 and IL-15 or IL-2 can be added [56, 57] .

The FT596 product is the first “ready-to-use”, universal and allogeneic CAR-NK cell product derived from iPSC technology and has been approved for clinical research in the United States [58] . It consists of an anti-CD19 CAR optimized for NK cells with a transmembrane domain of the activator receptor NKG2D, a 2B4 co-stimulatory domain, and a CD3ζ signaling domain. Two key ingredients are added:

(i) A novel, high-affinity, non-cleavable CD16 Fc receptor ( hnCD16 ) that enables tumor targeting and enhanced antibody-dependent cellular cytotoxicity without negative regulation, combined with a targeted therapeutic monoclonal antibody tumor cells;

(ii) IL-15/IL-15 receptor fusion protein ( IL-15RF ) that promotes cytokine-independent persistence ( Figure 4 ). When used in combination with monoclonal antibodies such as rituximab (anti-CD20), the hnCD16 Fc receptor of FT596 binds to the Fc portion of the monoclonal antibody covering tumor cells, activates NK cells, secretes cytokines, and enhances ADCC . IL-15RF promotes cytotoxicity of NK cells and activated antitumor T cells.

A communication during ASH 2020 reported the case of a 76-year-old woman with relapsed and refractory DLBCL after eight lines of therapy, including ASCT, autologous adoptive T-cell therapy, and activated haploidentical NK cells [59] .

This patient received a unique dose of 30 × 10 6 FT596 cells following lymphodepletion monotherapy . No side effects related to FT596 cells have been reported, except for neutropenia. Tumor response assessment at the 1-month threshold showed a partial response based on 2014 Lugano criteria, with a greater than 70% reduction in 18 F-Glu uptake and a greater than 50% reduction in tumor size. The patient's second dose of FT596 is ongoing. A Phase I study is also ongoing with an estimated 285 patients enrolled ( clinicaltrials.gov: NCT04245722 ).

Conclusion

Allogeneic CAR-NK cells offer many interesting perspectives that may address some of the issues faced with the use of autologous CAR-T cells. The major setbacks faced by allogeneic NK cells to date have been their relatively short duration of action and the lack of standardized processes. Compared with CAR-T cells, CAR-NK cell therapy is largely in its infancy; as of May 1, 2022, a total of 35 trials involving CAR-NK cells, compared with the clinical trials website involving CAR-T cells Compared with 1,181 trials of this new treatment, there is less data confirming this innovative treatment. CAR-NK cell therapy has broad prospects in fighting cancer, but more research is needed to demonstrate it.