Why do senescent cells persist: target evasion or failure of immune clearance?

We revisit the key methods and competing primary studies and recompute co-culture, uptake, wound and systemic-activation source tables. Conflicting labels and values remain part of the conclusions; missing times and sample identities are not filled in as facts.

01Clearance involves several steps; CMA processes proteins within cells

Immune cells must first encounter and recognize targets. NK or cytotoxic T cells can induce target death; macrophages can take up apoptotic cells and debris. Engulfed material must then undergo acidification and degradation. Persistent living targets, accumulated apoptotic debris and undegraded intracellular cargo are different burdens.

CMA does not engulf a whole senescent cell into a lysosome. HSC70 recognizes suitably targeted proteins, which undergo LAMP2A-mediated translocation into lysosomes. Changing protein turnover can alter target secretion and macrophage metabolism or receptor turnover, thereby affecting clearance between cells. 5 11

The measurements also differ. A transcriptional CMA score predicts pathway state from weighted expression of network genes. KFERQ-Dendra reports protein delivery to lysosomes and related changes. Accumulation after protease inhibition or uptake by isolated lysosomes provides additional turnover evidence. None directly measures how many living targets a macrophage eliminates per day. In the 2026 paper, bleomycin reduced the transcriptional score while fibroblast reporter activity remained preserved: prediction and function need not move together. 1 7

Original reporter validation includes LAMP2A dependence, macroautophagy controls and substrate degradation. Puncta still require interpretation: reporter expression, delivery, residence and degradation may contribute. The new paper checks Dendra expression and protease inhibition, supporting impaired delivery/turnover in old cells more convincingly than puncta counts alone. 1 11

02The crossed experiment exists, but its endpoint is not a clearance rate

Palbo-induced fibroblasts with intact CMA or L2A deletion were co-cultured for 48 hours with macrophages from control or myeloid L2A-deletion mice. This is a complete target-state × effector-state 2×2, with 3, 4, 4 and 3 independent experiments. 1

The Fig.4c source table gives mean residual fibroblast counts of 8.10, 12.54, 7.51 and 11.50. Macrophage CMA deletion corresponds to an increase of 4.44 source count units with control targets (unadjusted 95% CI −1.17 to 10.04), and 3.99 with deficient targets (0.11 to 7.86). The difference between these effects is −0.45 (−6.06 to 5.16). Our independent-group, unequal-variance calculation does not identify a stable interaction; the wide interval does not establish its absence. 2

However, the workbook's macrophage-count block repeats all 14 fibroblast values exactly, while the published macrophage plot has a different pattern. We withhold macrophage-count inference and do not use those values to normalize target counts. The fibroblast block remains a source-table recomputation without original-image validation. The original six-contrast family retains three missing comparisons; neither macrophage-genotype contrast reaches 0.05 after adjustment. Clarification requires the original investigators; we do not guess corrections. 2

Residual counts also require initial target abundance, spontaneous death, proliferation, exposure per surviving macrophage and direct tracking to separate greater clearance from other explanations. Separate conditioned-medium experiments support macrophage death and impaired bead uptake, so cell number and function can both matter. Some measurements use cells or fields from three experiments without their experiment identifiers; dozens of cells cannot become dozens of independent experiments. 1 2

03Uptake of apoptotic targets cannot substitute for killing living ones

IgG bead uptake in CMA-deficient macrophages averages 0.680 of control, with six versus five experiments. The independent-group difference is −0.320 (95% CI −0.611 to −0.028); the exact mean-allocation P is 0.0303 and becomes 0.0909 after Holm correction across our three functional comparisons. The direction supports an uptake deficit, with limited precision. 2

The efferocytosis difference is larger. With previously nonsenescent targets induced apoptotic, fluorescence-positive macrophages average 53.47% in controls and 2.94% with CMA deletion. With Palbo targets induced apoptotic, the means are 45.13% and 2.58%. The crucial step is staurosporine-induced target death before macrophage incubation. The peer-review response also explicitly describes this added experiment as testing efferocytosis. 1 2 6

There are only three source experiments per group, without pairing identifiers. Assuming exchangeable independent experiments, mean reallocation has 20 possible assignments and a minimum two-sided P of 0.10. A large group difference can coexist with a coarse randomization test. This calculation does not certify whether the original design was paired, invalidate the authors' parametric test or create additional samples; effect direction, precision and extrapolation need separate judgments.

Fluorescence positivity indicates acquisition of target-derived material, rather than kinetics of its complete subsequent degradation. Testing that step requires separating attachment, internalization, acidification and final cargo disappearance over time. Evidence for CMA degradation of intracellular proteins cannot replace evidence for processing engulfed whole-cell material.

04Target evasion can involve secretion and direct contact

A 2022 primary study used live imaging in human and mouse co-cultures: macrophages moved around senescent cells without killing or engulfing them under those conditions. Senescent cells also suppressed uptake of nearby apoptotic corpses. Conditioned-medium transfer and transwell separation did not reproduce this suppression; direct contact mattered. CD47/QPCT/L and CD24 in some human epithelial cells contributed. 8

That study separated uptake from acidification using dual-labelled apoptotic targets. Uptake decreased, whereas the acidification proportion of acquired material did not show the same reduction. Acidification itself does not establish final degradation. Blocking or deleting CD47 restored some corpse processing without making macrophages engulf the living senescent cells: a strong counterexample to removing one inhibitory signal being sufficient. 8

In contrast, conditioned medium in the new CMA paper reduces bead uptake and affects macrophage CMA and survival; an activator partially improves uptake. Targets, media, exposure and outcomes differ between studies. The 2022 null result does not exclude all paracrine effects, and the 2026 positive result does not make SASP the explanation for every form of evasion. Direct contact, transwells, matched medium exposure and receptor blockade need comparison within one system. 1 8

Other immune effectors provide another part of the chain. A 2019 autologous human dermal-fibroblast/NK/CD8 co-culture study supports HLA-E/NKG2A inhibition as a target-evasion mechanism. Target apoptosis and effector degranulation were measured separately; blockade or knockdown enhanced some responses. This supports inhibition of the killing step, but a few culture donors do not constitute a treatment trial in healthy older people. 10

A 2025 RvD2 study found that senescent macrophages themselves were poorly taken up. Reducing CD24/CD47 improved uptake without reaching apoptotic controls. RvD2 increased an apoptosis-related caspase measurement, and its combination with inhibitory-signal interventions produced more uptake. This positive evidence supports distinct constraints from readiness for death and permission for uptake; it does not prescribe one-sided or two-sided restoration for every target. 9

05CMA and receptor turnover are not direct translocation of every membrane protein

The new paper finds higher SIRPα levels and lower lysosome-associated turnover in CMA-deficient macrophages. Blocking endocytosis reduces the earlier between-group turnover difference; SHP-1 and cytoskeletal measurements also change. This supports a connection between CMA state, inhibitory-receptor endocytosis/turnover and uptake. 1

SIRPα is a membrane receptor. We do not depict the whole membrane protein passing directly through the CMA translocation pore. CMA processing of intracellular proteins and receptor transport through endocytosis require distinction. Associations and inhibition controls do not replace receptor-specific rescue, surface-residence tracking and complete mediation tests under common conditions. Improved bead uptake also does not demonstrate complete restoration of post-engulfment processing.

06Tissue and functional outcomes are positive, but sustained clearance remains unidentified

Myeloid L2A-deletion mice have higher levels of some senescence-related measurements in aged tissues, slower wound closure and higher local p21 and macrophage measurements. More macrophages can coexist with poorer repair, so abundance is not function. LysM-Cre also affects other myeloid cells; the animal outcome cannot be assigned exclusively to one macrophage mechanism. 1

The wound figure measures diameter in millimetres. One source baseline is 0.6 while the others are 6; days 12 and 14 contain unexplained blanks, and the final source time differs from the caption's 14/15-day description. We preserve the entries without treating blanks as closure. For the complete 0–8-day interval, the per-mouse diameter integral is 6.68 mm-days greater with deletion (exploratory 95% CI 1.63–11.72); day 8 diameter is 1.61 mm greater (0.42–2.80). Hypothetically replacing the anomalous baseline with 6 changes the integral difference to 7.06, preserving direction. This assumption does not alter the original table or certify later closure rates. 3

Systemic CA77.1 from 18 months of age for five months provides positive evidence for reductions in some tissue markers and fibrosis measurements. Compound design and earlier primary experiments support RARα–NCoR1 regulation, CMA activation and some selectivity controls. It still affects multiple cell types and is not a macrophage-only restoration experiment. 1 5 12

We directly compare old treated animals with old vehicle animals and retain all six SA-β-gal source blocks. Mean ratios are 0.85,0.85,0.57,0.46,0.32 and 0.37. Magnitude varies by tissue. None of the first five testable blocks reaches 0.05 after Holm correction across six blocks; the second lung block has an additional sample-count conflict and remains descriptive. Workbook male/female headings are opposite to figure/legend sample counts. Identity information is insufficient to certify sex, so we do not infer a sex advantage. 4

Bead uptake is also higher on average after 48-hour treatment of isolated old macrophages. The young-control source has nine values versus eight experiments in the caption, so we add no significance or complete-restoration claim. Earlier treatment performs better in the lung-injury model, but early and late groups differ in cumulative exposure. These are young male bleomycin-injured mice, not a pure timing comparison in naturally aged lungs or patients with IPF. 1 4

07What would establish interruption of a feedback loop?

Three balances organize the problem. Living senescent targets S increase through new production and propagation and decrease through death. Apoptotic debris D increases through death and decreases through uptake. Intramacrophage target material C increases through uptake and decreases through degradation:

G denotes new production, B propagation such as paracrine effects, H target death, a conversion from dead-target counts to material, Q other dead material, U uptake and V post-engulfment degradation. This is an accounting framework, without fitted biological rates, bistability or thresholds. For dimensional consistency, D,C and material flows use a common material unit; a and conversions for cell sizes, fragments and degradation products require further measurement and are not estimated here. Fluorescence-positive proportions also depend on distributions and detection thresholds, rather than directly measuring continuous cargo C or degradation flow V. Even an accurate terminal S constrains only the accumulated net balance G+B−H. A marker-positive proportion adds changes in expression and tissue composition.

Consequently, lower markers can reflect less production, less propagation, more death, altered markers or composition. More intracellular fluorescence can reflect higher U or lower V. Wound diameter validates a functional direction but cannot substitute for lineage tracking of S,D and C. Secretion-to-uptake/survival experiments support part of a feedback mechanism, without quantifying its strength, persistence threshold or stable decline after withdrawal.

A more discriminating experiment would restore target and effector CMA separately, fix initial counts, track living-target death and debris entry/disappearance simultaneously, and continue after drug withdrawal. If effector restoration alone suffices, dual restoration should not be presumed necessary. If markers decline while debris increases, complete clearance has not been established.

08The scope of this answer

Experimental evidence supports both target evasion and impaired immune function, which can influence one another. CMA research connects protein turnover, secretion, macrophage function and repair; competing primary studies show that direct contact, readiness for killing and different immune effectors also constrain clearance.

Our contribution is a mechanism assessment and source-table recomputation. Conflicts prevent certification of some counts, sex identities and later wound trajectories. We did not reprocess PXD071137 mass spectra, fit feedback parameters or obtain human CMA-treatment efficacy and long-term safety results. Official registry searches for CA77.1 and CA77 returned no matches; name searches do not cover every development code or global trial. Mouse markers, repair and pathology do not establish human healthspan or lifespan benefits. 14 15

Download the reproducibility package. It contains code, protocol changes, source-cell locations and derived outputs, excluding complete third-party papers, figures and workbooks. The same Codex agent conducted research, calculation, self-review and translation.