Can anti-CD28 CAR T cells target T lineage malignancies with less lymphodepletion?

Why CD28 offers a different trade-off than CD7

T-ALL has no approved immunotherapy, and the most advanced CAR T approach targets CD7. That target works early in T cell development, so anti-CD7 CAR T cells also eliminate healthy T cells and their precursors, producing profound T cell aplasia; CD7-negative relapse adds a second escape route. The paper's premise is that a target appearing later in T cell development, or on a narrower set of mature cells, could preserve more of the immune system while still hitting the leukemia.

CD28 is a costimulatory receptor that normally amplifies T cell activation. The study first asked whether it is present on malignant T cells at levels that a CAR could exploit. In a cohort of 54 pediatric T-ALL patients at diagnosis and 14 healthy bone marrow donors, 81.5% of patients expressed CD28, with a mean of 67.5% of T-ALL blasts positive versus 3.7% of healthy lymphoid progenitors (p = 0.0002). Among CD28-positive cases, the intensity of CD28 on blasts was comparable to that on physiological CD4 T cells, measured as median fluorescence intensity.

The authors also report uniformly increased CD28 in nodal T-follicular helper cell lymphomas of angioimmunoblastic type (3/3 samples) and follicular type (2/2 samples), while other T cell lymphomas tested did not express it. That broader expression pattern matters for interpreting the target's scope: the paper frames CD28 as actionable in pediatric T-ALL, these two lymphoma entities, and multiple myeloma, not across T cell malignancies generally.

Knocking CD28 out of the CAR T cells prevents fratricide

A CD28-directed CAR faces an immediate practical problem: the therapeutic T cells themselves express CD28, so they can recognize and kill one another, a phenomenon called fratricide. The authors solved this with CRISPR/Cas9-mediated knockout of CD28 in primary human T cells before generating the CAR. The paper reports that CD28-knockout T cells expanded normally after activation and retained effector functionality, which is the evidence that removing the receptor did not cripple the cell product.

This design detail is central to reading the potency results. The anti-CD28 CAR T cells are not simply T cells with a new receptor; they are CD28-deficient T cells carrying a CD28-specific receptor. Any comparison with anti-CD7 CAR T cells therefore compares two engineered products, each with its own target-driven biology, rather than a single cell background with swapped binders.

Potency matches anti-CD7 in cells and mice

Two lead anti-CD28 CAR constructs were tested. In vitro, they conferred high and specific cytotoxicity against CD28-positive T-ALL cell lines. In vivo, the comparison used T-ALL xenograft mouse models, including the CCRF-CEM and Jurkat cell lines and the ALL-230 patient-derived xenograft. In the CCRF-CEM experiment, mice treated with anti-CD7 or anti-CD28 CAR T cells survived significantly longer than anti-CD19 CAR T cell controls, with median survival of 34 days for the CD7 CAR, 38 days for the lead CD28 CAR, and 22 days for the CD19 control.

A pooled survival analysis across costimulatory domains gave median survival of 36 days for the 28CAR_2 construct and 33 days for 28CAR_14, compared with 35 days for the CD7 CAR, 27 days for the CD19 CAR, and 28 days for untransduced T cells. The authors conclude that the lead anti-CD28 CAR matches anti-CD7 CAR T cells in these models, and that 28CAR_2 outperformed 28CAR_14. These are mouse survival and tumor-burden readouts, not clinical response rates.

Less lymphodepletion in autologous immune cultures

The central safety-relevant comparison came from co-culturing each CAR T cell product with the same donor's peripheral blood mononuclear cells, then measuring which immune subsets disappeared. Anti-CD7 CAR T cells depleted CD4 T cells, CD8 T cells, and NK cells broadly. Anti-CD28 CAR T cells instead preferentially depleted CD4 T cells, largely sparing CD8 T cells and NK cells. B cells were depleted by the anti-CD19 control but not by the anti-CD28 product, which is consistent with CD28 being absent from immature B cell malignancies such as B-ALL.

Within CD4 T cells, the loss was not uniform. After 48 hours of co-culture, the effector-memory CD4 subset (CD45RO+CD197-) was reduced to 10% with the anti-CD28 CAR, versus 33% with the anti-CD7 CAR, 23% with the anti-CD19 CAR, 22% with untransduced T cells, and 20% in PBMC-only cultures (p = 0.0058). Frequencies of CCR4+ CD4 T cells fell to 4% with anti-CD28 CAR T cells, compared with 31% for anti-CD7, 13% for anti-CD19, 16% for untransduced, and 14% for PBMC-only (p = 0.0077). Anti-CD28 CAR T cells preferentially depleted TH2 cells (p = 0.028) and TH17 cells (p = 0.040), while the Th1-correlate CXCR3+CD4+ frequency was unchanged.

The authors connect this pattern to CD28 surface levels, which were highest on CD45RO+ CD4 subsets including central and effector memory cells. That is a plausible explanation for why memory CD4 cells are hit hardest, but the depletion pattern itself is the measured result; the link to expression level is the paper's interpretation.

CD28 loss appears in one model but not others

Antigen loss is the escape mechanism that limits anti-CD7 therapy, so the paper looked for it directly. In CCRF-CEM cells recovered from mouse spleen or bone marrow after CD28-targeting CAR T cell treatment, CD28 surface expression was reduced. In independent experiments using Jurkat cells or the ALL-230 patient-derived xenograft, CD28 expression on relapsed cells from bone marrow was unchanged. The authors therefore describe CD28 loss under selective pressure as something that can happen in some T-ALL cells, not a uniform escape route.

They also distinguish this from a short-term effect: CD28 internalization after soluble antibody exposure was more pronounced in Jurkat than in CCRF-CEM cells in vitro, which the authors read as evidence that the observed CD28 downregulation in vivo occurs through a distinct mechanism rather than simple receptor internalization. This remains a model-dependent observation, and the paper does not establish how often CD28-negative escape would occur in patients.

CD28 signaling may support leukemia survival

Beyond serving as a target, CD28 appears to have functional relevance in T-ALL cells. Co-culture of CCRF-CEM cells with CD86-expressing monocyte-derived dendritic cells increased proliferation under low-serum conditions, but blocking CD80/CD86 with the CTLA4-Fc fusion protein abatacept did not reduce that proliferation, indicating the effect was independent of CD28 ligation. Under JAK inhibition with ruxolitinib, however, CD28 ligation by monoclonal antibodies increased proliferation of both CCRF-CEM and Jurkat cells in an abatacept-sensitive manner.

The authors interpret this as a latent, JAK-independent pro-survival program downstream of CD28 that becomes visible when JAK signaling is blocked. This is a mechanistic hypothesis supported by cell-line experiments, not a demonstrated driver of clinical resistance, and it does not by itself change the targeting rationale.

Where the less-lymphodepletion claim holds

The claim that anti-CD28 CAR T cells cause markedly less lymphodepletion rests on autologous co-culture experiments with cells from a small number of healthy donors, analyzed at 24 and 48 hours, plus mouse xenograft survival data. It is a short-term, in vitro measure of subset depletion, not a clinical measure of immune reconstitution. The sparing of CD8 T cells and NK cells is relative to anti-CD7 CAR T cells in the same assay, not an absolute guarantee of preserved immunity.

The target-expression data come from pediatric and adolescent T-ALL samples and a small set of lymphoma samples, and the functional and escape experiments rely on T-ALL cell lines and xenografts. The paper's own framing is that CD28 is a promising target for a potentially less immunosuppressive approach; whether the CD4-skewed depletion profile translates into a clinical advantage, and whether CD28-negative escape emerges under treatment, are questions these models cannot answer.

The priority is sparing CD8 T cells and NK cells

For readers weighing T-ALL targets, the paper's most decision-relevant result is not that anti-CD28 CAR T cells kill leukemia, since anti-CD7 CAR T cells already do that in these models. It is the split in collateral damage: comparable tumor control alongside relative preservation of CD8 T cells and NK cells, with the cost concentrated in CD4 cells, particularly TH2 and TH17 subsets. That profile is what would distinguish a CD28-directed product from a CD7-directed one if it holds in patients.

The practical threshold the paper sets is therefore a trade-off, not a single efficacy number: a target must be present on enough T-ALL blasts (81.5% of patients in this cohort expressed it) and absent or lower on the immune subsets whose preservation matters most. CD28 satisfies that in these preclinical systems, but the CD4 depletion it does cause, and the model-dependent CD28 loss seen in CCRF-CEM, are the specific findings to watch as this target moves toward clinical testing.

Sources used in this answer