Hepatic immune signals: new T cells, or expansion of existing cells?

Immune aging involves more than a decline in T-cell counts. Reduced thymic output, changes in the proliferation, survival and movement of existing peripheral cells, and impaired antigen presentation can all weaken responses. Supplying support signals may affect several of these pathways together. s1

The study, published online in December 2025 and in a 2026 issue of Nature, calls its combination of mRNAs encoding DLL1, FLT3L and IL-7 “DFI.” Hepatic expression increased naive T cells in aged mice, strengthened vaccine-specific CD8 responses and improved control in selected tumor models. These are animal findings with functional endpoints. They do not partition the benefit into newly generated thymic cells, existing peripheral cells and other immune mechanisms. s2

01Counts, composition and origin are different questions

Naive T cells have not undergone the corresponding antigen-driven activation. They can come from the thymus or from division of naive cells already present in the periphery. A naive surface phenotype does not establish recent thymic birth. Likewise, T-cell receptor (TCR) diversity measures sequences and their frequency distribution, rather than where cells were generated.

The main aged comparison used 72-week-old mice, with 6-week-old mice as young references. After 28 days, naive CD4/CD8 frequencies and absolute counts increased. IL-7 alone expanded effector-memory cells, whereas the full combination did not show the same memory shift. The combination therefore cannot be reduced to a repetition of the single-factor result. s2

The authors report no increase in clonal concentration in bulk splenic TCR data and interpret this as favoring thymic output over peripheral expansion. Yet expansion need not concentrate the repertoire. Proportional growth doubles counts without changing relative frequencies; preferential growth of previously rare clones can increase diversity.

The counts 90, 5 and 5 are chosen examples. Proportional growth to 180, 10 and 10 leaves Shannon diversity at 0.394. Growth of the two existing rare clones to produce counts of 90, 55 and 55 increases it to 1.069 without adding a clonotype. This demonstrates one possible competing explanation. It neither fits DFI cell origins nor establishes a universal outcome of physiological expansion.

A clinical precedent makes the alternative substantive. A 2008 IL-7 study in 16 adults with refractory cancer observed vigorous proliferation of naive and recent-thymic-emigrant-like cells, increased T-cell mass and greater repertoire diversity. Considering cycling, TREC dilution and absent thymic enlargement, its authors mainly attributed short-term changes to peripheral expansion and redistribution. Different participants and treatments prevent direct falsification of DFI, but greater diversity is clearly not exclusive to new thymic production. s3

02What supports thymic involvement?

In aged mice, DFI partially restored thymic mass and cell density and increased mass-normalized early DN1–DN3 counts. That subset figure normalizes per milligram of thymus, rather than directly measuring the whole-organ absolute increment. Common lymphoid progenitors and progenitors associated with thymic homing also increased. These observations align along a developmental pathway and support thymic involvement more directly than peripheral counts alone. s2

The study also reports Rag2-eGFP developmental signals, Nur77 signals and increased peripheral TCR excision circles (TRECs). Two boundaries matter. The Reporting Summary lists the Rag2 and Nur77 reporter mice as 6–12 weeks old, creating an age bridge to the 72-week functional cohorts. These reporter results should not all be described as direct tracing in old mice. TRECs are nonreplicating DNA left by receptor rearrangement; their abundance also depends on division, survival and blood-cell composition. The reported relative TREC/Tcra measurement cannot be converted into mature cells newly exported per animal per day. s2, s4

Together, the observations support processes consistent with enhanced thymopoiesis. They do not estimate its contribution to peripheral counts or vaccine benefit: separating that contribution would require source-resolving lineage evidence or a thymus-dependency comparison that excludes simultaneous peripheral effects. Peer review discussed this distinction, but the authors' reply is an interpretation within the same study, rather than an independent replication. s5

03What can the deposited TCR files establish?

The paper reports both bulk splenic and single-cell tumor-infiltrating repertoires. The bulk comparison includes 5 young, 12 aged-control and 9 aged-DFI samples, but the downloadable files do not provide the original bulk data for reanalysis. Four tumor-infiltrating sequencing runs were available. The supplementary figure identifies two mice pooled per run, leaving two independent pools per condition. The 34,104 analyzed cells are not 34,104 treatment replicates. s2, s6, s7

AgingScope performed a descriptive audit using high-confidence, productive paired TCR cells found in each corresponding expression matrix. TRB-only and all-productive-chain definitions provide sensitivity checks. Analytic expected diversity under sampling without replacement compares libraries at equal depth. Clones are defined by amino-acid CDR3 combinations without additionally distinguishing V/J genes; this can merge some different receptors and cannot certify antigen recognition. This does not reproduce the original RNA QC, CD8 reference classification or tetramer-positive/negative subsets, and is therefore an all-eligible-TCR depth audit rather than a complete reproduction of Figure 4r.

At the common depth of 3,499 cells, the mean expected Shannon difference, DFI minus control, is +1.277. Other specified depths and both alternative chain definitions retain the positive direction. Thus, the broader distribution cannot be explained entirely by one condition sampling more cells. Pool variation remains substantial: inverse Simpson diversity, which emphasizes common clones, is lower in DFI pool RW than in control pool RV. The conditions are not completely separated by every metric.

With two pools per condition, only six two-versus-two label allocations exist and the minimum two-sided P is 1/3. This audit reaches that minimum. It diagnoses limited small-sample resolution, rather than testing or refuting the paper's other animal experiments. It provides neither an animal-level treatment interval nor a fraction of newly generated cells. Pairing and subset quality, pooling and absent baseline repertoires further limit interpretation.

The separate download labeled “circulating T cells scRNA:TCR” contains six age-specific RNA matrices, totaling 78,636 columns, without TCR contigs or per-animal metadata. It belongs to age discovery rather than the DFI comparison and cannot reproduce the complete reported 21-age, 96,683-cell blood analysis. Two spatial-data registrations are also distinct source datasets, rather than extra independent treatment replications. s7, s8, s9, s13

04Functional improvement does not partition its causes

The vaccine experiment used four mice per group. The full combination increased antigen-specific CD8 frequencies and counts, whereas individual factors did not show the same benefit. “Approximately 24 weeks younger” maps a vaccine readout onto its experimental age curve; it does not mean that whole-body biological age fell by 24 weeks. Healthspan and lifespan were not measured. s2

Tumor experiments include preconditioning and combination treatment after tumors were established. In the B16-OVA preconditioning cohort, three of eight DFI mice developed no measurable tumor, rounded to 40% in the text. The established-tumor combination with anti-PDL1 used ten mice per group and also showed delayed growth and improved survival. Transplanted tumors bearing a model antigen and specific immunotherapy constrain generalization. The 72-hour washout in the MC38 experiment does not establish equivalent function weeks after withdrawal. s2, s6

These functional findings should be retained. DFI also changes dendritic and B-cell compartments, however, so antigen presentation, survival and tumor recruitment of existing T cells could contribute. Main splenic spatial counts compare hundreds of fields from one animal per condition. Fields describe within-organ variation and cannot replace animal replication; the Methods' biological-replicate description differs from the main counting unit. The functional results support an effect of the combination without identifying each pathway's share. s2

Withdrawal gives a clear temporal boundary. After 28 days of treatment followed by 28 days off, TRECs returned to baseline and some splenic T-cell and dendritic-cell counts remained modestly elevated. Yet vaccination four weeks after withdrawal no longer showed the earlier antigen-specific benefit. Persistent counts and durable function are different endpoints. The paper accordingly confines most enhancement to the dosing window. s6

05Safety remains conditional

Selected autoimmunity and tolerance models showed no clear worsening. NOD cohorts used nine mice per treatment, with six months of monitoring after four weeks of exposure; Act-mOVA tolerance experiments used five per group. EAE exposure ended at symptom onset rather than covering the entire chronic disease course. NOD and EAE experiments were not independently repeated. Short-term hepatic tests and histology support tolerability under these conditions. s2, s4

They cannot exclude uncommon events or establish the safety of long-term repeated exposure in older humans. The change history also records replacement in April 2026 of Supplementary Table 3, originally an erroneous duplicate of Table 4. This audit uses the current Tables 2/3/4, retaining below-detection entries as censored values rather than zero. Comparisons with recombinant factors differ in exposure curves and detection constraints; selected inflammatory markers cannot establish a universal safety advantage. s2, s10, s11, s12

The evidence supports a limited but useful answer: aged mouse immunity retains capacity for improvement, and DFI affects both thymus-related processes and the peripheral immune environment. The contribution of newly generated mature T cells remains unidentified. Repertoire diversity and functional improvement cannot substitute for source attribution. Human infection outcomes, vaccine protection, cancer net benefit, long-term safety and healthspan require direct human evidence. This article describes evidence and supplies no medication or gene-intervention regimen.

Download summary data and reproducible calculations. The package distinguishes source-derived summaries from chosen examples; cell sequences, original documents and private working tables are excluded.