CAR-T cell therapy is a standard treatment for relapsed diffuse large B cell lymphoma (DLBCL); however, many patients fail to achieve a durable response. Poor T cell quality, including differentiated or exhausted phenotypes, is a mechanism of resistance to CAR-T (Locke, Blood Adv. 2020). Tumor microenvironment (TME) factors such as hypoxia, nutrient deprivation, and immunosuppressive oncometabolites may impact both T cells made into autologous CAR-T and their function after infusion. Single cell RNASeq of DLBCL patients' axicabtagene ciloleucel (axi-cel) CAR-T infusion products (n=39) revealed MTRNR2L8, encoding a 24-amino-acid peptide called Humanin-like 8 (HN8), upregulation in products of patients that achieved a durable response of >9 months (Yu, JITC 2025). The HN8 amino acid sequence is almost identical to Humanin, encoded by a mitochondrial gene MT-RNR2, known to promote mitochondrial functions including glycolysis and oxidative phosphorylation, and reduce apoptosis.

We manufactured CD19 CAR-T with CD3z/CD28 co-stimulatory domains similar to axi-cel, which also overexpressed HN8 (19.28z.HN8). HN8 overexpression was confirmed by western blot using Humanin antibody mixture (Sigma H2414, Invitrogen PA1-41325 and PA1-41610) and Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS).

We performed bulk RNA sequencing on 19.28z.HN8, and 19.28z CAR-T not expressing HN8, manufactured from healthy donor PBMCs (n=11). Differential gene expression analyses were done with DESeq2 analyzed via paired sample Wilcoxon tests of rlog normalized gene expression values, reporting FDR adjusted p-values. Pathway analyses were conducted via single sample gene set enrichment analysis (ssGSEA). 19.28z.HN8 CAR-T had downregulation of exhaustion-associated transcripts including TOX (p=6.1×10⁻4), TOX2 (p=0.012), PRDM1encoding BLIMP-1(p=0.003), TIGIT (p=8.2×10⁻9), ENTPD1-encoding CD39 (p=3.7×10⁻4), NT5E encoding CD73 (p=6.4×10⁻5), IKZF2 encoding HELIOS (p=2.5×10⁻5), and immune checkpoint CTLA4 (p=1.9×10⁻12) and upregulation of cytotoxicity gene GZMB (p=5.4×10⁻13) compared to 19.28z control. Next, for functional validation, T cells from two non-responding DLBCL patients leukapheresis products used to manufacture axi-cel, were flow cytometry sorted into unexhausted CD39- (UE) and pre-exhausted CD39+ (PE) fractions prior to 19.28z and 19.28z.HN8 CAR manufacture. Following manufacture from PE fractions, 19.28z.HN8 CD8+ and CD4+ had significantly lower percentages of PD-1 and TIGIT double positive CAR-T cells. CAR-T manufactured from UE fractions had fewer apoptotic cells upon activation. Notably, 19.28z.HN8 CAR-T were significantly more cytotoxic, and secreted more IFNg, than control in both UE and PE fractions.

We hypothesized that decreased exhaustion and improved effector function was due to HN8 impact on metabolic and anti-apoptotic programs similar to the function of humanin. RNA Seq demonstrated 19.28z.HN8, vs 19.28z control, mediates upregulation of HIF1A(p=1.2×10⁻3) and STAT3 (p=8.4×10⁻7), which promote glycolysis, and downregulation of glycolysis-suppressing transcription factor KLF2 (p= 3.5×10⁻4) and to lesser degree anti-apoptotic genes, specifically increased BFL-1 (p =0.04) and the NF kappa B signaling pathway (p=0.014). To validate, functional metabolic assays mito stress (n=9) and glycolysis stress (n=7) tests were performed using Agilent Seahorse XF at baseline and after stimulation for 15 hours with plate bound CD19 peptide. Consistent with transcriptomic data, 19.28z.HN8 exhibited enhanced basal glycolysis (n=9, p=0.008) and compensatory glycolysis (p=0.027). No differences were observed after stimulation. Upon stimulation 19.28z.HN8 had enhanced OCR (p=0.027) and ECAR (p=0.002), ATP production (p=0.023), maximal respiration (p=0.039), and basal respiration (p=0.039).

To evaluate apoptotic pathways, we performed BH3 profiling, revealing lower cytochrome C release in 19.28z.HN8 vs 19.28z. in a dose dependent fashion with FS1 mimetic (Biosynth), which inhibits BFL-1, an antiapoptotic member of the BCL2 family. Taken together with RNA seq this suggests higher BFL1 in 19.28z.HN8 contributes to prevent intrinsic apoptosis.In conclusion, HN8 overexpression in CAR-T cells drives a less exhausted and more functional phenotype, likely informed by HN8 enhancement of glycolysis and resistance to apoptosis. Ongoing studies aim to uncover how HN8 boosts CAR-T cell function within TME like conditions, and in vivo.

This content is only available as a PDF.
Sign in via your Institution