New anti-CD22 CAR designs keep pressure on B-cell tumors after CD19 escape
Relapse after CD19 CAR T-cell therapy remains one of the most stubborn failures in B-cell malignancy care. Tumors can evade recognition by losing or downregulating CD19, leaving clinicians with a disease that still depends on antigen targeting but no longer carries the original target on its surface. The work from Ruengdej and colleagues moves directly at that problem by building new CAR T cells around CD22, a B-cell antigen that often persists even after CD19-directed treatment has failed.
Two newly generated monoclonal antibody clones, 24A7G6 and 29H9C8, emerged as the most useful starting points after initial screening of four candidates. Both clones showed favorable binding and surface expression profiles, which made them suitable for conversion into second-generation CAR constructs. In practical terms, the study is asking a simple but clinically important question: can a fresh antibody scaffold produce a CAR that still recognizes malignant B cells when the more familiar CD19 route has already been exhausted?
The answer in these experiments was yes, at least in vitro.
Binding behavior and CAR architecture shaped the lead constructs
The investigators first screened antibody clones against CD22 and then moved the best performers into CAR design. Structural modeling with AlphaFold was used to predict how the single-chain variable fragments might interact with the CD22 extracellular domain, which contains seven immunoglobulin-like regions. That modeling step matters because CAR efficacy depends not only on whether a target exists, but on whether the engineered receptor can engage it with enough stability and geometry to trigger T-cell activation.
Both 24A7G6 and 29H9C8 were carried forward into second-generation CAR backbones. Each construct incorporated an IgG2 hinge, a CD28 transmembrane region, and either a 4-1BB or CD28 intracellular costimulatory domain fused to CD3ζ. This allowed a direct comparison between two common signaling strategies that often shape T-cell behavior in distinct ways: 4-1BB is generally associated with persistence and memory-like features, while CD28 tends to drive brisker activation and more immediate effector function.
The choice is not cosmetic. CAR signaling domains help determine whether the engineered cells act like short-lived fighters or longer-haul patrollers.
Engineered T cells expanded well and kept a memory-skewed phenotype
Primary human T cells from healthy donors were activated, transduced with the retroviral constructs, and then tracked for expression and function. Both 24A7G6-derived and 29H9C8-derived CAR T cells expanded efficiently after transduction. Their phenotype leaned heavily toward CD8-positive populations, with naïve and central memory features predominating rather than a terminally differentiated effector profile.
That pattern is encouraging for a therapy intended to persist after infusion. A naïve or central memory-skewed product can be better positioned to expand on antigen encounter and maintain activity over time, especially in a setting where tumor burden may be variable and residual disease may continue to present antigen for weeks or longer. The study also assessed activation and inhibitory markers after antigen exposure. CAR T cells increased activation markers and showed modest rises in inhibitory receptor expression, a familiar sign of T-cell engagement under repeated stimulation rather than an immediate collapse into dysfunction.
The cells activated, but they did not appear to burn out at once.
Antigen-specific killing separated CD22-positive cells from CD22-negative controls
Functional testing showed clear antigen specificity. The engineered CAR T cells mediated potent cytotoxicity against CD22-positive Raji and Nalm-6 cells, while showing no activity against CD22-negative K562 cells. That selectivity is one of the basic requirements for moving a CAR candidate forward, since off-target killing would raise obvious safety concerns and blur interpretation of the signal.
Across short-term assays, the activity was broadly comparable among constructs. Both antibody-derived CARs worked, and the main distinction came later, when the cells were forced to operate under prolonged antigen pressure. Cytokine testing after co-culture also showed functional response, with the assay measuring IFN-γ, TNF-α, IL-2, IL-4, IL-6, and IL-10. Although the paper does not provide a full cytokine table in the abstracted material here, the design itself reflects a standard attempt to capture both effector function and inflammatory tone after target recognition.
In CAR biology, killing and signaling usually travel together. A receptor that cannot drive cytokine release often cannot drive tumor control for long.
Prolonged co-culture separated the better sustained construct from the rest
The deeper signal emerged in long-term co-culture and sequential rechallenge assays. Under repeated antigen exposure, all constructs retained antitumor activity, showing that the newly generated scFvs could support more than a brief burst of cytotoxicity. The most notable divergence came from the 24A7G6-CD28 construct, which demonstrated enhanced sustained tumor control compared with the paired 4-1BB version in extended assays.
That result is worth careful reading. It does not overturn the usual expectation that 4-1BB can support longer persistence, but it does show that the best signaling configuration can depend on the antibody scaffold, the target epitope, and the functional context of the assay. A CAR that performs well in the first 24 hours may not be the same CAR that continues to suppress tumor regrowth after repeated challenge. In a disease where antigen escape and residual disease can co-exist, the ability to keep pressure on tumor cells across multiple exposures may matter as much as the initial burst of lysis.
Long-term control is where many CAR concepts separate from truly useful constructs.
CD22 remains a rational escape target after CD19-directed failure
The biological rationale for CD22 is straightforward. CD22 is a transmembrane sialoglycoprotein on B cells that participates in B-cell receptor regulation, and its expression often persists in malignant cells that have already escaped CD19-directed therapy. Because it is also internalized efficiently, it has long been attractive for antibody-based and CAR-based targeting. The study’s preclinical data fit that logic well: the new clones recognized CD22-positive cells, ignored CD22-negative cells, and preserved function after repeated stimulation.
For patients with relapsed or refractory B-cell disease, especially those who have already lost CD19 expression, a CD22-directed strategy offers a way to sidestep one of the most common antigen-escape routes. The current work does not claim clinical efficacy, and it does not reach beyond the bench. What it does provide is a clean in vitro foundation for choosing which antibody-derived CARs deserve the next round of preclinical testing.
The remaining hurdles sit outside the dish
This is still an early-stage platform. The study was conducted entirely in vitro, using healthy donor T cells and established cell lines rather than patient samples or in vivo tumor models. It does not answer how these CARs will behave in circulation, how long they will persist after infusion, or whether the antigen density seen in cell line systems matches what real relapsed leukemia presents at the bedside. It also does not resolve how these constructs will handle the wider immunologic stresses of the tumor microenvironment, where exhaustion, trafficking barriers, and antigen heterogeneity all shape outcome.
Even so, the comparison is useful because it narrows the field. Two new anti-CD22 monoclonal antibody clones, 24A7G6 and 29H9C8, produced CAR T cells that expanded well, remained antigen-specific, and kept killing under continued tumor challenge. The 24A7G6-CD28 construct stood out for stronger sustained control in extended assays, making it a plausible lead for additional preclinical development. The next step is not a broad claim of readiness, but a more demanding test of durability, safety, and function in systems that resemble human disease more closely than a culture plate can.