The Aliquot

PTPN2 and PTPN1 are negative regulators of the JAK-STAT signaling axis, the intracellular relay that translates cytokine signals into gene expression changes. Both phosphatases dampen inflammation by stripping phosphate groups from JAK and STAT family members, blunting responses to interferon-gamma (IFNγ), interleukin-2 (IL-2), and IL-15. PTPN2 also targets proximal T cell receptor (TCR) signaling molecules LCK and FYN, raising the activation threshold for T cells. Inhibiting both simultaneously, the authors reasoned, should amplify immune signaling through multiple parallel pathways.

The structural challenge was formidable. The active site of PTPN2 is lined with basic residues that demand highly polar inhibitors for enzymatic potency, but polar compounds typically fail to cross cell membranes at useful concentrations. The team's solution was a zwitterionic scaffold built around a thiadiazolidinone dioxide core. The crystal structure of AC484 bound to PTPN2 reveals nine interactions with residues in the active site Cys216 region, including hydrogen bonds with Cys216, Arg222, Asp182, Ser217, and Ile220. An isopentyl-amine group extends to engage Asp50 and Met256. The resulting compound carries a biochemical IC50 of 1.8 nM against PTPN2 and 2.5 nM against PTPN1, with a cellular EC50 of 176 nM for IFNγ-mediated STAT1 phosphorylation in B16 melanoma cells. Its precursor, A-650, achieved a cellular EC50 of 2,651 nM under the same conditions. That 15-fold improvement in cellular potency, achieved without sacrificing biochemical affinity, is the central medicinal chemistry achievement of the paper.

The cellular consequences of this inhibition are broad. In B16 tumor cells, AC484 dose-dependently enhanced IFNγ-driven growth arrest, matching the effect of genetic deletion of both Ptpn2 and Ptpn1. Transcriptomic profiling showed that AC484-treated cells and Ptpn2/n1-null cells had nearly identical global responses to IFNγ, with enrichment of hallmark IFNγ response, IFNα response, and TNF-NF-κB gene signatures. Critically, AC484 alone, without IFNγ, did not spontaneously activate interferon-stimulated gene (ISG) expression, suggesting the drug amplifies existing signals rather than generating noise.

In primary mouse T cells, AC484 increased the frequency of activated CD69+ and CD25+ cells and boosted production of IFNγ and TNF in a dose-dependent manner. The effect was additive in Ptpn2-null or Ptpn1-null T cells, confirming that the two phosphatases have redundant roles and that dual inhibition is more potent than targeting either alone. In human whole blood from both healthy donors and patients with lung cancer, kidney cancer, and melanoma, AC484 produced comparable dose-dependent increases in STAT5 phosphorylation, a finding with direct relevance to clinical translation.

The in vivo data are the paper's most provocative contribution. Across four syngeneic tumor models, B16 melanoma, KPC pancreatic adenocarcinoma, 4T1 mammary carcinoma, and EMT-6 breast carcinoma, oral AC484 monotherapy induced tumor regression and extended survival. In each case, efficacy was comparable to or greater than anti-PD-1 antibody treatment. The gap was widest in the 4T1 and EMT-6 models, both of which are well-established as resistant to PD-1 blockade. AC484 controlled tumor growth in these resistant models where anti-PD-1 showed minimal effect. In the CT26 colon cancer model, combining AC484 with anti-PD-1 further enhanced efficacy beyond either agent alone, pointing toward a rational combination strategy.

The metastasis data are particularly striking. In the B16 pulmonary metastasis model, luciferase-expressing tumor cells were injected intravenously to seed the lungs. By day 10, untreated mice and anti-PD-1-treated mice had developed detectable lung metastases. AC484-treated mice had no detectable disease, and the survival rate at study end was 100%, compared with 0% in both control groups. In the 4T1 orthotopic model, where metastases arise spontaneously from a primary mammary fat pad tumor, AC484 reduced the number of lung metastases.

Immune dependency was confirmed by testing AC484 in immunodeficient NSG mice bearing KPC tumors. The drug had no effect in the absence of a functional immune system, establishing that its anti-tumor activity is entirely immune-mediated. Rechallenge experiments in MC38-cured mice showed resistance to tumor regrowth, consistent with the formation of immune memory.

Which immune cells are doing the work depends on context. In MHC class I-proficient KPC and B16 tumors, depletion of CD8+ T cells abolished AC484 efficacy, while NK cell depletion had no significant effect. The picture inverted in Jak1-deficient KPC tumors, which express low levels of MHC class I: there, NK cell depletion abrogated the response while CD8+ T cell depletion did not. In the 4T1 metastasis model, NK cells were again the critical effector population. AC484 simultaneously enhances multiple cytotoxic immune populations, with the dominant effector determined by the tumor's immune evasion strategy. This context-specificity is mechanistically coherent and clinically relevant: tumors that have lost HLA expression to escape T cell killing remain vulnerable to NK cell surveillance, and AC484 enhances both arms.

To understand how AC484 reshapes the tumor microenvironment (TME), the authors performed single-cell RNA sequencing on 68,060 CD45+ tumor-infiltrating immune cells from B16 and KPC tumors across untreated, AC484-treated, and anti-PD-1-treated conditions. The results reveal a broad remodeling that goes well beyond what anti-PD-1 achieves.

AC484 treatment increased the overall ratio of lymphoid to myeloid cells, driven by expansion of CD8+ T cells and NK cells. Among myeloid cells, the ratio of M1 to M2 macrophages shifted toward the pro-inflammatory M1 phenotype, and the ratio of monocytes to myeloid-derived suppressor cells (MDSCs) increased. A distinct population of ISG-high monocytes appeared specifically in AC484-treated tumors. Immunofluorescence microscopy confirmed that CD45+ and CD8+ cells not only increased in frequency but also infiltrated deeper into tumor tissue compared with untreated controls. The chemokine environment shifted accordingly: expression of Ifng, Tnf, Il15, Cxcl9, and Cxcl10 increased, while Ccl22 and Ccl17, chemokines that recruit immunosuppressive regulatory T cells, fell.

The T cell story is where the mechanistic depth of the paper becomes most apparent. Re-clustering of the lymphoid compartment revealed that AC484 induced a distinct effector CD8+ T cell population expressing high levels of Gzmb, Prf1, and Ifng. This population was absent in untreated and anti-PD-1-treated tumors but was the most abundant CD8+ T cell cluster in AC484-treated tumors. A proliferating variant of the same population, marked by elevated Mki67, was also AC484-specific.

To probe the epigenetic basis of this state, the authors sorted progenitor-exhausted (SLAMF6+) and terminally exhausted (TIM-3+) CD8+ T cells from treated and untreated tumors and performed bulk ATAC-seq alongside RNA-seq. Unsupervised clustering of differential open chromatin regions defined four modules. The module associated with exhaustion-related genes, including Tox, Itgav, and Epas1, was reduced in AC484-treated progenitor and terminal subsets. A separate module containing effector and memory genes, including Gzma, Il7r, Sell, and Tcf7, was enriched specifically in AC484-treated conditions. The ATAC-seq tracks at the Tox locus show visibly reduced chromatin accessibility in AC484-treated TIM-3+ cells, providing direct epigenetic evidence that the drug counteracts the exhaustion program at the chromatin level.

The transcription factor motif enrichment analysis of differential open chromatin regions is particularly informative. Nine of the top ten most differentially enriched motifs in AC484-treated TIM-3+ cells, including those for BATF, JUNB, STAT3, STAT1, and STAT5, match those previously reported in T cells treated with a combination of IL-2 and anti-PD-L1. A gene set enrichment analysis confirmed this overlap directly: the IL-2 plus anti-PD-L1 combination gene signature was highly enriched in AC484-treated TIM-3+ cells, while the anti-PD-L1-alone signature was enriched in anti-PD-1-treated cells. AC484 appears to phenocopy the transcriptional and epigenetic state induced by combined IL-2 and checkpoint blockade, a combination that has shown promise in reversing T cell exhaustion.

Flow cytometry on TILs from B16 tumors confirmed the transcriptional findings at the protein level. AC484-treated mice showed lower TIM-3 and TOX expression on CD8+ T cells and a reduced frequency of TIM-3+TOX+ double-positive cells compared with both untreated and anti-PD-1-treated mice. Phosphorylated STAT5 (pSTAT5) was elevated in TILs from AC484-treated but not anti-PD-1-treated mice, confirming active JAK-STAT pathway engagement within the tumor.

The metabolic consequences of this signaling shift are measurable. AC484 increased both maximal oxygen consumption rate and extracellular acidification rate in activated primary mouse T cells, indicating enhanced mitochondrial respiration and glycolysis. Total mitochondrial content, measured by MitoTracker staining, also increased. These metabolic changes are consistent with the known effects of IL-2 and IL-7 signaling through STAT5 on T cell bioenergetics.

A particularly clean experiment tested whether transient AC484 exposure could durably improve T cell function in vivo. OT-I transgenic CD8+ T cells specific for the SIINFEKL peptide were expanded in vitro with AC484 or vehicle for four days, washed free of drug, and transferred into mice bearing EL4-OVA tumors. Recipients of AC484-treated T cells suppressed tumor growth, and four of ten mice achieved complete responses. Recipients of vehicle-treated T cells did not control tumors. The drug was gone; the functional improvement persisted.

In a chronic antigen stimulation assay designed to model T cell exhaustion, primary mouse CD8+ T cells were stimulated five times with SIINFEKL peptide over five days with or without AC484. Chronic stimulation alone produced the expected exhausted phenotype: high TOX and PD-1 expression, reduced cytokine production. AC484 treatment during chronic stimulation significantly reduced the frequency of PD-1+TOX+ cells and produced a roughly threefold increase in T cells producing both IFNγ and TNF simultaneously, a hallmark of polyfunctional, non-exhausted T cells. When these chronically stimulated cells were tested in a co-culture killing assay against B16-OVA tumor cells, AC484-treated cells showed significantly restored cytotoxic activity compared with vehicle-treated chronically stimulated cells.

The breadth of this dataset is genuinely impressive. The authors move from crystal structure to in vivo efficacy to single-cell epigenomics in a single paper, and the mechanistic story is internally consistent throughout. The observation that AC484 efficacy in Jak1-null tumors depends on NK cells rather than CD8+ T cells is a particularly elegant piece of mechanistic dissection, and it has direct implications for patient selection in the clinic.

A few questions are worth holding onto. The mechanism of STAT5 activation in the TME is not fully resolved. In vitro, AC484 augments IL-2-induced pSTAT5 but does not activate STAT5 in the absence of cytokine. The authors show that the combination of AC484 and IFNγ can induce pSTAT5 even when common gamma-chain receptor signaling is blocked, suggesting a non-canonical pathway. Whether this mechanism operates in vivo, where IFNγ concentrations are high in inflamed tumors, remains an open question. The authors are appropriately cautious here, and the data they present are suggestive rather than definitive.

The reversibility of immune-related adverse effects in the rat toxicology study is reassuring. Dose-dependent immune cell infiltrates in kidneys, joints, and livers resolved within 28 days of treatment cessation. The short half-life of a small molecule, compared with an antibody, is a genuine pharmacological advantage when managing inflammatory toxicity. That said, the translation from rat to human immune activation thresholds will require careful monitoring in the ongoing Phase I trial.

The selectivity profile is strong: AC484 shows 6-8-fold weaker activity on PTPN9 and no detectable activity on SHP-1 or SHP-2 in biochemical screens, with no off-target activity detected across a broad panel of kinases and receptors. For a compound targeting a highly conserved active-site motif, that selectivity is notable and will need to be confirmed in human tissues.

AC484 is, to the authors' knowledge, the first active-site phosphatase inhibitor to enter clinical evaluation for cancer immunotherapy. That distinction matters beyond the compound itself. The field has long assumed that phosphatase active sites are off-limits for small-molecule drugs. This work shows the assumption was wrong, at least for PTPN2 and PTPN1, and the structural and chemical logic used to solve the problem is transferable to other phosphatase targets.

For patients, the most immediate implication is the potential to treat cancers that do not respond to existing checkpoint inhibitors. The 4T1 and EMT-6 breast cancer models, both PD-1-resistant, responded to AC484 monotherapy. Tumors with JAK1 or JAK2 mutations, which confer resistance to T cell-mediated killing by disabling IFNγ sensing, remained sensitive to AC484 through NK cell-mediated mechanisms. Tumors with B2m loss, which eliminates MHC class I surface expression and renders cells invisible to CD8+ T cells, also responded. The breadth of coverage across immune evasion strategies is the compound's most clinically compelling feature.

AC484 is currently being evaluated in Phase I clinical studies as monotherapy and in combination with anti-PD-1 in patients with advanced solid tumors (ClinicalTrials.gov identifier NCT04777994). The preclinical data presented here provide a detailed mechanistic rationale for both arms of that trial. Whether the distinct T cell state AC484 induces in mice, one that resembles the product of combined IL-2 and checkpoint blockade, translates to human tumors is the question the trial will begin to answer.