November 5, 2026
An AI Conversation
The search context does not support the premise that polarized neutrophils directly program hematopoietic stem cells (HSCs) for dissemination; rather, polarized membrane domains are a feature of HSCs themselves that facilitate their own homing and engraftment.
- HSC Homing Mechanism: Quiescent hematopoietic stem and progenitor cells (HSPCs) display a polarized membrane domain enriched in tetraspanins (like CD82) and integrins (like VLA-4). This polarity is required for efficient homing to the bone marrow niche and subsequent engraftment, not for dissemination into the blood.
- Neutrophil Role in Dissemination: Neutrophils contribute to the mobilization (dissemination) of HSCs into the bloodstream by releasing proteolytic enzymes and influencing G-CSF levels, which disrupt the retention signals (such as CXCL12/CXCR4 interactions) in the niche.
Clearance and Regulation: Aged neutrophils home to the bone marrow where they are cleared by macrophages, which can downregulate CXCL12 and promote HSPC egress, but this is a regulatory feedback loop rather than a programming event by polarized neutrophils.
Is your source only reference regarding question? Do homing neutrophils program stem cells in bone marrow to be disseminated into the bloodstream?
Good question on both counts. My previous answer drew on general knowledge without citing specific sources — the search results above now provide well-supported references.
Short answer: No, homing neutrophils do not “program” HSCs. The word “program” implies direct instruction of the stem cell. What actually happens is that neutrophils modify the niche environment, which removes the retention signals that keep HSCs anchored in the bone marrow.[1] The stem cells then egress in response to the altered chemical landscape.
The mechanisms are:
- Protease release – G-CSF-activated neutrophils degranulate and release serine proteases (neutrophil elastase, cathepsin G) and MMP-9, which cleave CXCL12, CXCR4, and VCAM-1 — the very molecules that tether HSCs to stromal cells.[2] The anchors are destroyed; the stem cells are freed.
- ROS-mediated stromal cell apoptosis – Neutrophil-derived reactive oxygen species kill mesenchymal progenitor cells and osteoblasts, reducing production of retention factors (CXCL12, SCF, VCAM-1).[3]
- Circadian clearance pathway – Aged neutrophils (CD62L⁻ CXCR4⁺) home to the BM, are phagocytosed by resident macrophages via the LXR pathway, which downregulates CXCL12 in the stroma, promoting rhythmic HPC release into circulation (Méndez-Ferrer et al., Cell 2013).[4]
So the neutrophil’s role is indirect niche remodeling, not direct programming of the stem cell. The HSC is a passive responder to the loss of its retention cues.
Key sources: Cell 2013 – Rhythmic Modulation of the Hematopoietic Niche, PMC – Neutrophils as regulators of the hematopoietic niche, Springer 2024 – Role of immune cells in BM on HSCs.
Are you saying that Tumor-Associated Neutrophils originating from the primary tumor do not carry any genetic material related to that tumor?
That’s a fair challenge. My earlier framing was too absolute. TANs are not tumor cells — their own genome is a normal neutrophil genome. But they do carry tumor-derived genetic material internally, and this is well documented:
1. Oncogenic tumor DNA is sequestered inside neutrophils. A key study (Kisiel et al., Translational Oncology, 2018) tracked mutant HRAS and HER2 gDNA in tumor-bearing mice and found that circulating neutrophils contained the highest levels of tumor-derived oncogenic DNA — exceeding plasma, exosomes, platelets, and RBCs.[1,2] This DNA was taken up via tumor-derived extracellular vesicles (EVs) and free nucleosomes.[1,2] Tumor excision caused the WBC-associated gDNA signal to disappear within 2–9 days, consistent with neutrophil turnover.[1,2]
2. Neutrophils internalize tumor-derived exosomes carrying tumor nucleic acids. Multiple studies confirm active uptake of tumor exosomes by neutrophils (verified by fluorescent labeling). These exosomes deliver:
- Tumor mRNAs and miRNAs (e.g., miR-362-5p from HCC exosomes, which activates NF-κB in neutrophils)
- Tumor-derived fatty acids (e.g., from lung cancer exosomes, which trigger lipid accumulation → NET formation)
- Tumor stem-like cell exosomal RNAs that prime bone marrow neutrophils for pro-tumoral behavior[3]
3. The relationship is bidirectional. TANs also deliver their own genetic material to tumor cells — for example, neutrophils transfer SPI1 mRNA to cancer cells via extracellular vesicles, upregulating glycolytic gene transcription in the tumor.[4]
Bottom line: TANs don’t become tumor cells, but they are biological reservoirs of tumor-derived genetic material (oncogenic DNA, mRNAs, miRNAs) acquired from the tumor microenvironment. This has practical implications — neutrophils are now proposed as a novel liquid biopsy platform for detecting circulating tumor DNA.
Key sources: Kisiel et al., Transl Oncol 2018, Tumor-derived exosomal FFAs reprogram neutrophils, 2026, HCC exosomes reprogram TANs, 2025, CRC stem cell exosomes prime neutrophils, 2019.
Isn’t is true that these “reservoirs, tumor-associated neutrophils, are originally recruited to tumor environment by HMGB1/CXCL12. They become polarized, accumulate in the tumor and transition to a survival phenotype, to later “home” to bone? If they do not “program” HSCs with identifying genetic material from “wound” source, is there an alternative HSC pathway in the bone marrow of this process?
This is a three-part question, and the search results let me address each precisely.
1. HMGB1/CXCL12 recruitment — yes, but it’s one axis among several.[1]
The reduced form of HMGB1 forms a heterocomplex with CXCL12 that signals through CXCR4 more potently than CXCL12 alone (Schiraldi et al., JEM 2012).[1] This is confirmed in multiple reviews as a neutrophil-chemotactic mechanism in the TME. However, the primary directional axis for neutrophil migration to the tumor is the CXCR2 ligand family (CXCL1, CXCL2, CXCL5, CXCL8/IL-8), while CXCR4/CXCL12 is more involved in retention in the marrow under homeostasis.[2] The HMGB1–CXCL12 complex is a genuine but secondary recruitment mechanism.[3]
2. Polarization → survival → “homing to bone” — partially supported, but the sequential model is overstated.
- Polarization: Yes. TANs adopt N1 (anti-tumor) or N2 (pro-tumor) phenotypes depending on the local milieu.
- Extended survival: Yes. Tissue-resident neutrophils in the TME can survive >5 days, far exceeding the typical 6–24 h blood lifespan (Ng et al., Nature 2024).[1]
- “Homing to bone” as a programmed step: This is not well-established as a discrete sequential event. What is documented is that CXCR4 upregulation allows neutrophils to return to the BM after extravasation, and that tumor-derived G-CSF disrupts the CXCR4–CXCL12 retention axis. But the evidence does not support a fixed “tumor → bone marrow” homing itinerary. Instead, the BM is independently and simultaneously reprogrammed by systemic tumor signals (see below).
3. The alternative pathway — direct tumor-to-BM signaling, bypassing neutrophils entirely.
This is where the critical evidence lies. Tumors do not need neutrophils as intermediaries to affect HSCs. They act on the marrow directly:
| Mechanism | What happens | Key reference |
| Tumor exosomes → BM progenitors | Melanoma exosomes carrying MET “educate” BM progenitor cells toward a pro-metastatic, pro-vasculogenic phenotype | Peinado et al., Nat Med 2012 |
| Tumor exosomes → BM stromal cells | AML exosomes upregulate DKK1 in stromal cells, suppressing WNT signaling and normal hematopoiesis; downregulate SCF and CXCL12 | Gao et al., Leukemia 2017; Gao et al., Leukemia 2015 |
| Tumor exosomes → HSCs directly | AML exosomes impair HSC clonogenicity, cause loss of CXCR4/c-Kit, and repress hematopoietic transcription factors (c-Myb, Cebp-β, Hoxa-9) | Gao et al., Leukemia 2015 |
| Tumor cytokines → BM | G-CSF, GM-CSF, IL-8 from tumors drive emergency granulopoiesis systemically, expanding Ly6G⁺ neutrophils in the BM with altered retention/egress signals (↑CXCR2, ↓CXCR4) | Recent J Immunol 2026 |
| Tumor exosomes → MSCs | Tumor exosomes endow BM mesenchymal stromal cells with a pro-tumorigenic phenotype, promoting macrophage infiltration and suppressing normal HSC support | Zhang et al., Oncogene 2016 |
The key conceptual correction: The tumor’s effect on the bone marrow is systemic and direct — via circulating exosomes and cytokines that reach the marrow independently of any neutrophil intermediary. Neutrophils are not a required “messenger” carrying tumor genetic material back to the HSC niche. The HSCs and their stromal support cells are directly targeted by tumor-derived extracellular vesicles and soluble factors. The neutrophil and the HSC are parallel, not sequential, targets of the same systemic tumor signal.[4]
In other words: the tumor broadcasts. Both the neutrophil lineage and the HSC niche receive the broadcast simultaneously. There is no relay.
Key sources: Peinado et al., Nat Med 2012, Gao et al., Leukemia 2015, Gao et al., Leukemia 2017, Schiraldi et al., JEM 2012, J Immunol 2026 – emergency granulopoiesis, IJMS 2026 – neutrophil-lineage cells in skeletal colonization.
According to your response, then DTCs are only one of several pathways, mechanisms, for tumor metastasis?
Yes — but with an important clarification about terminology. DTC dissemination is not “one pathway among several” in the sense of being optional; it is the definition of metastasis (a tumor cell physically relocating to a distant site). What varies is the route and the enabling mechanisms:
Routes of DTC dissemination:
| Route | Mechanism |
| Hematogenous | Intravasation → survival in blood → extravasation at distant capillary bed (most common for solid tumors) |
| Lymphatic | Lymphatic intravasation → lymph node transit → eventual blood entry (common in breast, melanoma) |
| Transcoelomic | Shedding into body cavities (peritoneal, pleural) — e.g., ovarian, gastric |
| Perineural | Migration along nerve sheaths — e.g., pancreatic, head/neck |
| Extravascular | Direct tissue-to-tissue migration without entering a lumen (emerging evidence, particularly in melanoma and sarcoma) |
Enabling mechanisms (not “pathways” of metastasis per se, but prerequisites):[1]
- Pre-metastatic niche (PMN) formation — tumor exosomes, cytokines, and metabolites remotely prepare distant organs (via integrin-specific targeting, ECM remodeling, immune cell recruitment) before any DTC arrives.[2] This is a support mechanism, not a route of spread.
- Immune suppression — TANs, TAMs, Tregs, and MDSCs create permissive conditions both at the primary site and at distant organs.
- EMT / plasticity — intrinsic tumor-cell programs (SNAIL, TWIST, ZEB) that enable detachment, motility, and survival in circulation.
- Survival in transit — NETs, platelet cloaking, and immune evasion allow DTCs to survive shear stress and immune attack.
So the precise answer: DTC dissemination is the sine qua non of metastasis. What is “one of several” is the route (hematogenous vs. lymphatic vs. transcoelomic vs. perineural vs. extravascular) and the enabling machinery (PMN, immune evasion, EMT, etc.). The pre-metastatic niche and exosome-mediated organ conditioning are supportive, non-cellular mechanisms that increase the probability that disseminated DTCs will successfully colonize — but they are not themselves metastasis.
Key sources: Hoshino et al., Cell 2015 – exosome integrins determine organotropism, Tang et al., MedComm 2025 – mechanistic insights, Front Cell Dev Biol 2026 – intraluminal vs. extraluminal routes, PMC 2023 – cellular and molecular characteristics of PMNs.
Is this in conflict with wound healing regarding recruitment of pluripotent stem cells additional to growth factors?
Is this in conflict with wound healing regarding recruitment of pluripotent stem cells additional to growth factors?
Compare tumor signaling vs wound healing recruitment
- Reviewing tumor-derived systemic signaling to bone marrow
- Evaluating the role of CXCL12/CXCR4 axis in recruiting stem cells
- Analyzing tumor-to-BM signaling vs wound healing
- Comparing tumor vs wound CXCL12/CXCR4 signaling directions
- Analyzing tumor-induced HSC suppression vs wound homing
- wound healing recruitment bone marrow stem cells CXCL12 growth factors mechanismtumor impaired wound healing bone marrow stem cell recruitment conflictpre-metastatic niche vs wound healing stem cell homing CXCL12
No — they are not in conflict. They are parallel, non-exclusive uses of the same molecular axis (CXCL12/CXCR4), operating in opposite directions and at different scales:
Wound healing — local, paracrine, “pull”:
- Injury → hypoxia → HIF-1α → upregulation of CXCL12 at the wound site
- CXCL12 gradient (high at wound → low in BM) pulls CXCR4⁺ stem/progenitor cells out of the marrow and into the wound
- Growth factors (VEGF, PDGF, FGF, TGF-β) act as secondary signals for differentiation and proliferation after arrival
- Additional axes: SCF/c-KIT, CCL27/CCR10, CCL21/CCR7
- The signal is soluble and transient — no genetic material is transferred to the stem cell[1]
Tumor — systemic, exosome-mediated, “push + impair”:
- Tumor exosomes reach BM stromal cells → downregulate CXCL12 and SCF → HSCs are released but simultaneously functionally impaired (loss of CXCR4, c-Kit; repression of c-Myb, Cebp-β, Hoxa-9)
- The signal is genetic material in exosomes that reprograms the niche cells
Where they intersect (the actual clinical conflict):
When a patient has both a wound and an active tumor, the two systems compete:
| Wound signal | Tumor signal | |
| CXCL12 | ↑ at wound (pull) | ↓ in BM niche (push) |
| HSC function | Intact (responds to gradient) | Impaired (exosome-mediated) |
| Outcome | Recruitment to wound | Release into circulation + dysfunction |
The tumor’s systemic suppression of BM CXCL12 and direct HSC impairment overrides the wound’s local recruitment signal. This is exactly what is observed clinically: cancer patients (especially during chemotherapy) have impaired wound healing due to reduced recruitment and survival of BM-derived progenitor cells at the wound site.
So the precise answer to your question:[2,3,4,5,6]
The wound healing model is not primarily growth-factor-driven — it is chemokine-gradient-driven (CXCL12/CXCR4 is the master axis), with growth factors acting downstream. The tumor model I described does not conflict with this; it hijacks and inverts the same axis systemically. The wound says “come here” (local CXCL12 ↑). The tumor says “leave the marrow, but be broken” (systemic CXCL12 ↓ + exosomal reprogramming). When both are present, the tumor wins, and healing fails.
Key sources: MDPI 2026 – SDF-1/CXCL12 and wound healing, PMC 2026 – Injury-induced Cxcl12 and Shh/Ihh in bone regeneration, MDPI 2021 – CXCL12 Crossroads in CSCs, Nature 2012 – Epithelial stem cells, wound healing and cancer, ScienceDirect 2026 – TME mirrors wound healing.
Aside from definitions of metastasis and DTCs, have any studies identified metastatic spread without evidence of DTCs, only stem cells? And if so, how is genetic information delivered to distant sites?
No study has demonstrated a complete, overt metastatic lesion arising at a distant site with zero DTCs present. The standard model still requires a tumor cell to physically arrive. However, there are three lines of evidence that push against a strict DTC-only requirement:
1. “Educated” bone marrow progenitors (Peinado et al., Nat Med 2012)
This is the closest to your question. Melanoma exosomes carrying the MET oncoprotein permanently reprogram uncommitted HSCs in the marrow toward a pro-metastatic, pro-vasculogenic phenotype. These educated BMDCs are then recruited to distant organs before any tumor cell arrives, where they form cellular clusters and remodel the ECM. Critically, though, they do not become tumor cells — they create a permissive microenvironment that increases the probability that arriving DTCs will successfully colonize. The DTC is still the “seed”; the educated BMDC is “fertilized soil.”[1,2,3]
2. Exosome-induced malignant transformation of normal cells (in vitro)
Two studies come closest to “exosome-only” tumorigenesis:
- Pancreatic cancer exosomes in the classic NIH/3T3 two-step transformation assay: exosomes from pancreatic cancer cells (but not normal cells) initiated malignant transformation, and the transformed cells formed tumors in vivo in nude mice. This is the strongest evidence that exosomes alone can drive a normal cell toward malignancy.
- Abdouh et al. (2017): Exosomes from cancer patients’ sera transferred malignant traits to normal cells carrying a single oncosuppressor mutation, inducing phenotypic changes consistent with malignant transformation. The key caveat: the recipient cell needed a pre-existing mutation to be susceptible.[4]
Neither of these is a natural metastasis event — they are in vitro/in vivo transformation models. No one has shown that, in a living organism, exosomes alone (without any DTC) can produce a clinically recognizable metastatic deposit.
3. “Genometastasis” — a proposed but unproven concept
A 2024 Physiology Reviews article explicitly names this hypothesis: EV-mediated horizontal transfer of oncogenic DNA/RNA to normal cells at distant sites could, in theory, “induce a tumorigenic phenotype in normal cells, thereby establishing a new mechanism that drives malignant growth.” The mechanism proposed is integration of exogenous oncogenic DNA at double-strand break (DSB) sites in the recipient cell genome. This remains theoretical — no study has demonstrated de novo tumor formation at a distant site via this mechanism in vivo.
How genetic information is delivered (in the cases that exist):
| Cargo | Example | Effect in recipient |
| miRNAs | miR-105, miR-122, miR-362-5p | Transient: vascular permeability, metabolic reprogramming, NF-κB activation |
| mRNAs | MET, hTERT, SPI1 | Protein expression in recipient; semi-stable |
| Oncogenic DNA | Free gDNA, nucleosomes, retrotransposon elements | Theoretical: integration at DSB sites → stable mutation |
| Proteins | MET oncoprotein, mutant p53 | Signaling activation; not heritable |
| circRNAs | Various | Induce uncontrolled proliferation in normal cells |
The critical limitation: RNA and protein effects are self-limiting (degraded, diluted with cell division). For a stable, heritable malignant transformation, you would need DNA integration — and that has not been demonstrated in a natural metastasis context.
Bottom line: The field has strong evidence that exosomes and educated BMDCs are necessary enablers of metastasis (without them, DTCs fail to colonize), but no evidence that they are sufficient to produce a metastatic lesion in the absence of any tumor cell. The “genometastasis” hypothesis is the logical next step, but it remains unproven.
Key sources: Peinado et al., Nat Med 2012, Abdouh et al., 2017, Pancreatic cancer exosome transformation, 2019, Physiology Reviews 2024 – genometastasis, Frontiers Oncol 2026 – sEV DNA horizontal transfer, Bioengineering 2022 – EV-mediated malignant transformation.[5,6]
Since tumor-programmed stem cells from bone marrow already carry the identity of the tumor cells, doesn’t if follow that the purpose of this process, though dysfunctional, is to establish itself in the pre-cancerous niche and proliferate, alternatively from converting existing cells?
The logic is internally valid if the premise holds — and the premise is where it breaks. Let me separate the two issues:
“Carry the identity of the tumor cells” is an overstatement.
What the exosome actually transfers is a signaling perturbation, not a genomic identity:
| What is transferred | Nature | Duration | Sufficiency for malignancy |
| MET oncoprotein | Receptor tyrosine kinase (protein) | Hours–days (protein turnover) | No — single pathway activation |
| Tumor mRNAs (e.g., hTERT, MET) | Transient transcripts | Hours (no polyA tail protection in cytosol) | No — not heritable |
| Tumor miRNAs (e.g., miR-362-5p) | Post-transcriptional regulators | Days | No — modulates, doesn’t reconstitute |
| Oncogenic gDNA (free, nucleosomal) | Fragments, not a genome | Unknown; theoretical integration | Theoretically sufficient if integrated — but never demonstrated in vivo |
The educated BMDC acquires a pro-vasculogenic signaling state (MET→Tie2→VEGF-A axis). This is a phenotypic shift, not a genomic reconstitution. Cancer requires multiple cooperating, stably integrated mutations (e.g., RAS + TP53 + RB1 + telomere maintenance). A single oncoprotein delivered in a lipid vesicle does not meet this bar.
However — your intuition is not wrong, and there is one study that supports it:[1]
The 2014 prostate cancer exosome study (PubMed 24715691) is the strongest counter-evidence to the “soil only” model:
Patient-derived adipose stem cells (pASCs) treated with prostate cancer exosomes acquired cytogenetic aberrations, underwent mesenchymal-to-epithelial transition, expressed epithelial/neoplastic/vasculogenic markers “reminiscent of molecular features of PC tumor xenografts,” and PC cell-derived exosomes were sufficient to recapitulate formation of prostate tumorigenic mimicry in vivo.[2]
This is not “soil.” This is a stem cell acquiring a neoplastic phenotype from exosome cargo alone, in the absence of any DTC. The caveat: it was a specific cell type (adipose stem cells), a specific cancer (prostate), and a specific exosome population. It has not been generalized.
The 2026 Springer review on the pre-metastatic niche adds a second data point:
“Resident stromal stem-like cells within the PMN can be activated by tumor-derived signals. These cells may then undergo malignant transformation, driven by epigenetic modifications, metabolic adaptation, and other mechanisms, acquiring tumorigenic properties.”[3]
The word “may” is doing heavy lifting — this is a hypothesis, not a demonstrated mechanism.
On “purpose”:
These processes have no purpose in a teleological sense. Tumor exosome secretion is a byproduct of the tumor’s metabolic and signaling state. The “education” of BMDCs is an unintended consequence of systemic exosome dissemination that happens to increase metastatic efficiency. From an evolutionary (Darwinian) perspective, tumors that secrete more effective exosomes outcompete those that don’t — but the exosomes aren’t aimed at the BMDC. They are broadcast systemically, and the BMDC is one of many recipients.
So the precise answer to your question:
- In the canonical model (Peinado 2012, Kaplan 2005): No. Educated BMDCs are non-malignant, terminally differentiated toward a pro-vasculogenic state. They do not proliferate autonomously. They are soil.
- In the emerging evidence (2014 prostate pASC study; 2026 stromal transformation hypothesis): The logic can hold under specific conditions — if the exosome cargo is sufficiently oncogenic AND the recipient cell is a stem/progenitor cell with the plasticity to undergo transformation. But this has not been demonstrated as a general mechanism, and it remains the exception rather than the rule.
The gap between “signaling perturbation” and “malignant transformation” is the multi-hit requirement. Exosomes bridge part of that gap (one or two hits, delivered as protein/RNA), but not all of it. The DTC still carries the full mutational load. The educated BMDC carries a partial, transient, non-heritable version of the tumor’s program.
Key sources: Peinado et al., Nat Med 2012, eLife 2018 replication, Prostate cancer exosomes reprogram pASCs, 2014, Springer 2026 – PMN stromal transformation
For clarification, are you saying that bone marrow-derived stem cells carrying primary tumor lineage do NOT “become” cancer cells at distant sites?
Correct — in the well-supported canonical model, they do not. The educated BMDCs remain host cells at distant sites. They never acquire the tumor’s full mutational genome, and they do not form the metastatic lesion itself. Here is what the evidence consistently shows:
What educated BMDCs actually do at distant sites:[1]
| Cell type | Fate at PMN | Malignant? |
| VEGFR1⁺ HPCs (Kaplan 2005) | Form cellular clusters, secrete SDF-1 to attract arriving DTCs | No — they are CD45⁻, c-Kit⁺, Tie2⁺ progenitors |
| Educated HSPCs (2016) | Differentiate into MDSCs (immunosuppressive myeloid cells) | No — immunosuppressive, not neoplastic |
| Cancer-educated BMSCs (2019) | Expand PMN-MDSCs via CXCL5, iNOS, GM-CSF | No — they support cancer cells, don’t become them |
| Prostate exosome-educated BMSCs (2021) | Secrete CXCL12 to promote cancer cell seeding in bone | No — stromal support role |
The critical test: blocking VEGFR1 or depleting these BMDCs prevents metastasis — not because the tumor cells are destroyed, but because the niche that would have supported them is absent. The DTC still arrives; it just fails to colonize.[2]
Where the distinction matters most:
The educated BMDC is to the metastatic lesion what a transplanted organ is to the donor: it is structurally and functionally remodeled by the tumor’s signals, but it retains its own genome. It is a modified host cell, not a new tumor cell.
The two exceptions I flagged before still apply, with important caveats:
- 2014 prostate pASC study — exosomes drove a specific stem cell type (adipose) toward a neoplastic phenotype. This was a transformation model, not a natural metastasis event, and has not been generalized.
- 2006 chemical carcinogen study (PubMed 16984729) — BMDCs treated with 3-methylcholanthrene formed multiple tumor types. This is chemical carcinogenesis, not exosome-mediated tumor-to-stem-cell transfer.[3]
So the precise answer: No, tumor-educated BMDCs do not become cancer cells at distant sites in any well-replicated, natural metastasis model. They are permissive stromal/immune cells that increase the probability that arriving DTCs will successfully colonize. The DTC — carrying the full mutational genome — remains the obligate source of the metastatic lesion. The BMDC is the soil; the DTC is the seed. The soil is modified, but it does not become the plant.
Key sources: Kaplan et al., Cell 2005, Peinado et al., Nat Med 2012, Cancer-educated BMSCs, Cell Death Dis 2019, HSPCs → MDSCs, 2016, Prostate exosomes → BMSCs, 2021.
Is the burden of proof backwards? How do researchers determine DTCs from tumor-programmed, BMDCs?
Your instinct is right — and there is a specific study that directly exposes the problem.
The “burden of proof” problem is real and documented.
The traditional DTC identification pipeline is:
- Collect bone marrow aspirate
- Enrich for CK⁺/EpCAM⁺/CD45⁻ nucleated cells
- Call them “DTCs”
This assumes epithelial marker expression = tumor origin. But the Demeulemeester et al.[1] (Genome Biology, 2016) single-cell sequencing study shattered that assumption:
They sequenced 63 single cells from bone marrow of 6 non-metastatic breast cancer patients that were morphologically classified as “tumor cells.” By comparing each cell’s copy number aberration (CNA) landscape and somatic SNVs to the matched primary tumor, they found:
- 53% were true DTCs (shared CNAs + tumor-specific mutations)
- ~30% were “aberrant cells of unknown origin” — they had CNA profiles that did NOT match the tumor and lacked tumor-specific mutations
- ~17% were normal cells with copy-neutral genomes[2]
The “aberrant cells of unknown origin” (AUs) are almost certainly bone marrow-derived cells that acquired independent genomic aberrations (age-related, or possibly tumor-exosome-mediated). They were misclassified as DTCs by the standard CK/EpCAM/CD45 criteria.
So how are DTCs actually confirmed today?
| Level | Method | What it proves |
| 1. Phenotypic (traditional) | CK⁺/EpCAM⁺/CD45⁻, nucleated, >5 μm | Epithelial-like, non-leukocyte — necessary but not sufficient |
| 2. Molecular (RT-PCR) | CK19, mammaglobin, TWIST1 mRNA | Tissue-of-origin marker — still not tumor-specific |
| 3. Genomic (definitive) | Single-cell DNA sequencing: CNA + SNV matching to primary tumor | Clonal relationship — the gold standard |
| 4. Clonal (metastasis confirmation) | Matching driver mutations (e.g., BRAF V600E, TP53 specific mutation) between primary and metastatic lesion | Proves the metastasis is a clonal descendant of the primary |
The critical point: Only level 3 (genomic matching) can definitively distinguish a true DTC from a tumor-educated BMDC. The phenotypic markers (level 1) are insufficient because:
- Educated BMDCs can express low levels of epithelial markers
- Normal bone marrow contains rare CK⁺ cells (the 2023 JID study confirmed EpCAM⁺/CK⁺ cells in healthy human and murine bone marrow)
- EMT causes DTCs to lose epithelial markers, making them harder to detect by the same criteria [3,4]
On “burden of proof backwards”:
Yes — the field historically operated on the assumption that “epithelial marker in bone marrow = tumor cell.” The genomic data shows this is wrong ~47% of the time. The burden of proof should be on genomic clonality, not phenotypic similarity. The Demeulemeester study effectively reversed the logic: instead of “looks like a tumor cell → is a tumor cell,” the correct logic is “shares the tumor’s specific mutational genome → is a tumor cell.”
The metastatic lesion itself is confirmed as DTC-derived by clonal matching: the metastasis must share the primary tumor’s trunk mutations (e.g., the same TP53 mutation, the same BRAF V600E). A BMDC, no matter how “educated,” cannot acquire the full multi-hit mutational genome of the primary tumor. It can acquire a few exosome-delivered mRNAs or a single oncoprotein, but not a coordinated set of chromosomal aberrations and point mutations that define the tumor’s clonal identity.
In short: The phenotypic criteria are a screening tool, not a diagnostic one. The genomic criteria are the diagnostic. And the field’s historical reliance on phenotypic criteria alone means that a substantial fraction of cells called “DTCs” in the literature may actually be aberrant BMDCs.
Key sources: Demeulemeester et al., Genome Biology 2016, JID 2024 – EpCAM/CK in normal bone marrow, PMC 2023 – EpCAM/CK in normal BM, DTC-Flow, Nat Commun 2025.
So, concluding, no metastatic disease sites have been found that carry the genomic identity of the primary tumor, but no disseminated tumor cells? If so, does any mechanism exist to verify this?
The question contains a hidden contradiction that is worth making explicit:
“Genomic identity of the primary tumor” and “DTC origin” are the same thing.[1]
If a cell at a distant site carries the primary tumor’s trunk mutations, its specific CNA landscape, and its driver mutations, then by definition it is a descendant of a DTC. The genomic identity is the proof of DTC origin. There is no third category: a cell is either clonally related to the primary tumor (→ DTC-derived) or it is not.
What “genometastasis” would actually produce (if it ever worked):
The 2024 Physiology Reviews framework distinguishes two mechanisms:
| T1: DTC-derived metastasis | T2: “Genometastasis” | |
| Genetic content | Full clonal identity: trunk mutations + CNA landscape + driver mutations | Partial: one or a few oncogenic mutations delivered via exosomes |
| DTC present? | Yes (at some point) | No |
| Distinguishable? | Yes — shares trunk mutations | Yes — lacks trunk mutations; only shares the specific oncogene(s) delivered |
A genometastasis site would look like a de novo primary tumor with one or two shared mutations, not like a metastasis. It would fail the clonal-matching test. You could verify its non-DTC origin by showing:
- Absence of trunk mutations present in the primary tumor
- Absence of the primary tumor’s CNA landscape
- Presence of only the specific oncogenic cargo (e.g., a single RAS mutation, a single HER2 amplification)
- Normal bone marrow or stromal cell lineage markers (CD45⁺, or MSC markers) rather than epithelial markers[2]
The Demeulemeester data confirms the distinction is detectable:
The “aberrant cells of unknown origin” (AUs) in their study had CNA profiles discordant from the tumor and lacked all tumor-specific trunk mutations. They were genomically distinguishable from true DTCs.[2] If a genometastasis event ever occurred, it would look like an AU with one or two shared oncogenic mutations — not like a true DTC.
So the precise answer:
No such site has been found, because the premise is self-contradictory. A site that “carries the genomic identity of the primary tumor” is, by definition, a DTC-derived metastasis. The verification mechanism is the genomic matching itself: if it matches, it’s a DTC; if it doesn’t match (but shares one or two mutations), it’s a genometastasis candidate — and that is a fundamentally different thing from a metastasis carrying the primary’s identity.
The two are mutually exclusive by the logic of clonal genetics.
Key sources: Demeulemeester et al., Genome Biology 2016[2], PMC 2024 – ctDNA and CTCs in metastasis[3], Hallmarks of Metastasis, Cancer Res 2019[4], Science 2020 – Origins of metastases in CRC.[5]
Does this explanation run counter to wound-healing role of BMDCs, where “pluripotent” role is not exactly pluripotent?
No — it does not run counter. If anything, the wound-healing data reinforces the same principle: the cells are multipotent, not pluripotent, and in both contexts their primary contribution is paracrine/immunomodulatory signaling, not transdifferentiation.
The “pluripotent” label is a misnomer in both contexts.
BMDCs (MSCs, HSPCs, VEGFR1⁺ HPCs) are multipotent — they can differentiate within a limited set of lineages (mesenchymal: osteoblasts, chondrocytes, adipocytes, myofibroblasts; hematopoietic: all blood lineages). They cannot generate all three germ layers. The term “pluripotent” appears in older wound-healing literature as a loose synonym for “multipotent,” but it is technically incorrect.
The wound-healing role is predominantly paracrine, not differentiation:
The most rigorous evidence comes from the Keio University chimeric mouse study (Blood 2011): using GFP-labeled BM chimeras and high-resolution confocal microscopy, they showed:
- ~80% of recruited BMDCs at the wound were macrophages (depleted by CSF-1 inhibition)
- Zero endothelial differentiation observed in any of 50+ fields across three wound models
- The proangiogenic effect was entirely macrophage-mediated (VEGF, TGF-β, PDGF secretion), not transdifferentiation[1]
A 2026 review on MSC therapy for chronic wounds states it plainly:
“While MSCs could change into fibroblasts, keratinocytes, or endothelial-like cells, the chance of stable long-term integration and functional differentiation occurring in chronic microenvironments is not high, and clinicians rely more on paracrine and immunomodulatory effects.”[2]
The parallel is exact:
| Wound | Tumor PMN | |
| Cell recruited | BMDC (multipotent) | BMDC (multipotent) |
| Primary fate | Macrophage (CSF-1 dependent) | MDSC (immunosuppressive) |
| Primary function | Paracrine: VEGF, TGF-β, PDGF → angiogenesis, fibroblast recruitment | Paracrine: CXCL5, iNOS, GM-CSF → immunosuppression, DTC support |
| Transdifferentiation? | Minimal (keratinocytes: 0.0001–0.0003% of epidermis) | Not observed |
| “Pluripotent”? | No — multipotent, paracrine-dominant | No — multipotent, paracrine-dominant |
The unifying principle: In both contexts, the BMDC is a multipotent signaling cell whose fate is dictated by the niche. The wound niche says “become a macrophage, secrete growth factors.” The tumor niche says “become an MDSC, suppress immunity.” The cell’s intrinsic capacity is the same; only the external signal changes. Neither context requires or demonstrates pluripotency.
Key sources: Blood 2011 – BMDCs as proangiogenic macrophages, PMC 2026 – MSC therapy for chronic wounds, Frontiers 2024[3] – BMDCs in wound healing, PMC 2012 [4] – Mobilised BMDCs and wound healing.[5]
Regarding, “The unifying principle: In both contexts, the BMDC is a multipotent signaling cell whose fate is dictated by the niche,” if so, then why are BMDCs “programmed” at all?
The question exposes a genuine tension in the language, and the recent literature resolves it precisely.
“Programmed” and “dictated by the niche” are not contradictory — they describe different levels of the same event.
The niche is the input signal. “Programming” is the cell’s molecular response to that signal — specifically, chromatin and transcriptional remodeling. The cell is not a passive receiver; it actively reorganizes its own gene accessibility in response to the niche. But the content of that reorganization is determined entirely by the external signal.
The Cancer Cell 2025 study on mammary tumor-driven myelopoiesis is the cleanest demonstration:
- Tumor → IL-1β → HSPC chromatin remodeling: 19,117 sites gained accessibility, 9,064 lost
- Result: myeloid-biased transcriptional program, immunosuppressive neutrophil output
- Anti-IL-1β treatment → chromatin landscape reverts (GATA1 sites re-opened, myeloid program collapsed, balanced hematopoiesis restored)[1]
The “program” was written by the niche and erased when the niche signal was removed. The cell was never “reprogrammed” in the sense of acquiring a new identity — it was signaled into a different chromatin state, and it reverted when the signal stopped.
But there is a spectrum, and some “programming” is more durable:
| Durability | Mechanism | Evidence | Reversibility |
| Transient (hours–days) | Transcription factor activation, cytokine signaling | Anti-IL-1β restores HSPC chromatin (Cancer Cell 2025) | Fully reversible |
| Semi-stable (weeks) | Histone modifications, enhancer priming | Tumor-induced marrow remodeling persists after tumor removal (Frontiers 2026) | Reversible with sustained opposing signal |
| Durable (months+) | DNA methylation, 3D chromatin architecture | MPP3 myeloid bias persists in treatment-naïve breast cancer patients (WashU 2024) | Requires active epigenetic intervention |
| Irreversible | DNA sequence change | Not observed in BMDC “education” by tumor signals | N/A |
The WashU dissertation on breast cancer HSPC reprogramming is important here: it shows that the myeloid bias in MPP3 cells is “durable” — it persists in treatment-naïve patients, not just during active tumor burden. This is the closest to true “programming” in the BMDC context. The mechanism is Dkk1 from bone stromal cells (which the tumor induces), which acts on MPP3s to lock in a myeloid transcriptional state.[2]
So the precise answer to “why are BMDCs programmed at all?”:
They are programmed in the narrow, molecular sense: the niche signal induces real, measurable changes in chromatin accessibility and transcription factor binding that alter the cell’s behavior. This is not mere “response” — it is a state change with variable durability. But it is not programming in the genomic sense (no DNA sequence change) or in the identity sense (no transdifferentiation into a new lineage). The cell’s genome is unchanged; its chromatin landscape is modified by the niche, and the modification’s durability ranges from fully reversible (transient signaling) to semi-permanent (epigenetic memory).
The “unifying principle” holds: the niche determines the outcome. What was imprecise was the implication that “dictated by the niche” means the cell is a passive, stateless receiver. It is not. It remembers — to a degree that depends on the strength, duration, and type of signal. But the memory is epigenetic, not genetic, and in most documented cases, reversible.[3]
Key sources: Cancer Cell 2025 – reversible tumor-driven myelopoiesis reprogramming[1], WashU 2024 – durable HSPC reprogramming in breast cancer, Frontiers 2026 – persistent post-tumor marrow remodeling[5], MDPI 2026 – epigenetic memory in disease[3].
Additional reference
Neutrophils home to bone marrow?
Senescent neutrophils home to the bone marrow for clearance via the CXCR4-CXCL12 chemokine axis.
- Mechanism: As neutrophils age in circulation, they upregulate CXCR4 expression, which binds to CXCL12 (SDF-1α) produced by bone marrow stromal cells (e.g., LepR+ stromal cells).
- Retention and Clearance: This interaction retains mature neutrophils in the marrow reserve under homeostatic conditions and facilitates their return from tissues or circulation to be cleared by efferocytosis (phagocytosis) by resident macrophages.
- Reverse Transmigration: Neutrophils can exit injured sites, re-enter the bloodstream, upregulate CXCR4, and traffic back to the bone marrow for elimination, a process that avoids local tissue damage.
Finally, in tumor immunology, many examples of both protumorigenic and antitumorigenic properties of neutrophils have been described. This has important consequences for cancer therapy, as targeting neutrophils can lead to either suppressed or stimulated antitumor responses.
Furthermore, S100A8 and S100A9 were shown to be involved in neutrophil and tumor cell motility as well as tumor cell survival [217], making them interesting targets for preventing metastasis. As the CXCR4/CXCL12 axis is also crucially involved in the bone marrow kinetics of neutrophils, this target is likely too drastic to be a viable option.
Neutrophils in cancer: from biology to therapy, Cellular & Molecular Immunology, 2024
https://www.nature.com/articles/s41423-024-01244-9
The spleen, for instance, holds neutrophil subpopulations at different anatomical locations with distinct functions important for, e.g., bacterial eradication, and the lung was recently shown to re-educate neutrophils that had trafficked from a site of sterile injury to home back to bone marrow for elimination. Further, recent reports demonstrate subpopulations of neutrophils with different actions during homeostasis, infection, tissue restitution and cancer. It is becoming increasingly clear that this cannot be due to different stages of neutrophil activation during their life span but instead points towards distinct subpopulations of neutrophils with different effector functions.
The neutrophil: one cell on many missions or many cells with different agendas?, Cell and Tissue Research, 2018
https://link.springer.com/article/10.1007/s00441-017-2780-z
Neutrophil-lineage cells occupy a unique position in this setting because they are generated, retained, mobilized, aged, and reprogrammed within the same bone marrow niches that disseminated tumor cells exploit for homing and survival. This review examines how neutrophils, tumor-associated neutrophils, immature neutrophils, low-density neutrophils, and PMN-MDSCs shape skeletal colonization. We discuss tumor-to-marrow signaling, CXCR2-dependent recruitment, CXCR4/CXCL12-mediated marrow retention, neutrophil–circulating tumor cell interactions, vascular arrest, dormancy escape, NET-mediated matrix remodeling, immune suppression, and effects on osteoclast–osteoblast coupling. Evidence is strongest in breast and prostate cancer models, where pathways such as CXCL5/CXCR2, CTNND1– CXCR4/CXCL12, PR3–RAGE, and DKK1–CKAP4–STAT6–CHI3L3 link neutrophil-lineage cells to skeletal progression and immunotherapy resistance. However, several mechanisms, including CTC–neutrophil clustering and NET-driven dormancy awakening, remain partly extrapolated from non-skeletal models.
From Bone Marrow Reserve to Metastatic Niche: How Neutrophil-Lineage Cells Shape Skeletal Colonization, International Journal of Molecular Science, 2026
https://pubmed.ncbi.nlm.nih.gov/42450242
Nevertheless, extended follow-up studies have unveiled a notable increase in the risk of locoregional recurrence associated with breast conserving therapy in contrast to mastectomy. This article introduces a novel hypothesis rooted in the biological phenomenon of homing to elucidate this intriguing clinical observation. We postulate that a breast homing mechanism of reactivated circulating and disseminated tumor cells (OR PRIMED NEUTROPHILS recruited to this axis, ie HMGB1?) mediated by chemotaxis involving at least the CXCR4-SDF-1 axis may provide a biological rationale for this clinical phenomenon.
Unlocking the Power of the Homing Phenomenon: Why Breast Conserving Surgery Outshines Mastectomy in Overall, Survival, 2023
https://www.clinical-breast-cancer.com/article/S1526-8209(23)00263-X/pdf
In this study we provide evidence that the SDF-1/CXCR4 chemokine axis is involved in both the retention of neutrophils within the bone marrow and the homing of senescent neutrophils back to the bone marrow.
It can be seen from these sections that the CXCR4(high) neutrophils that home to the bone marrow are observed in the hematopoietic compartment of the bone marrow; i.e., these cells are able to attach to and migrate through the sinus endothelium into the stroma.
These data indicate that the homing of senescent neutrophils back to the bone marrow is mediated via CXCR4.
We propose that the gradient of chemokine created across the sinus endothelium stimulates chemotaxis of neutrophils present in the hematopoietic compartment and induces their migration across the bone marrow sinusoidal endothelium, such that they can be liberated into the systemic circulation.
Chemokines acting via CXCR2 and CXCR4 control the release of neutrophils from the bone marrow and their return following senescence, Immunity, 2003
https://www.cell.com/immunity/fulltext/S1074-7613(03)00263-2
Neutrophils accumulate in many types of human and murine tumors and represent a significant portion of tumor-infiltrating myeloid cells.
The phenotype and function of tumor-associated neutrophils (TANs) in humans, particularly in the early stages of tumor development, have not been extensively investigated.
Phenotype and function of tumor-associated neutrophils and their subsets in early-stage human lung cancer
Cancer Immunology & Immunotherapy, 2017
https://pmc.ncbi.nlm.nih.gov/articles/PMC5522629
Pubmed hmgb1-cxcl12 complex recruits primed neutrophils?
Research indicates that HMGB1 and CXCL12 form a heterocomplex that recruits inflammatory cells, including neutrophils, to damaged tissues by signaling exclusively through the CXCR4 receptor. While early studies established that this complex enhances the migration of monocytes and hematopoietic stem cells at suboptimal chemokine concentrations, more recent findings highlight a distinct, atypical mechanism where CXCL12 signaling specifically enhances neutrophil migration by modulating nuclear deformability.
- The recruitment process involves the following key mechanisms:
- Heterocomplex Formation: HMGB1 (specifically in its reduced state) binds with CXCL12 to form a complex that activates CXCR4 dimers, promoting the migration of leukocytes such as monocytes and B cells into injured areas.
- Neutrophil Modulation: CXCL12 alone or in conjunction with this pathway has been shown to enhance neutrophil migration in 3D environments by increasing nuclear pliability through ACKR3 (CXCR7) receptor signaling and PKA-dependent chromatin compaction.
- Signal Specificity: The recruitment of inflammatory cells via the HMGB1/CXCL12 complex is independent of TLR4 or RAGE receptors, distinguishing it from the pro-inflammatory cytokine release mediated by oxidized HMGB1.
Brave Search 07-08-2026
Pubmed: Neutrophils accumulate in tumor and take-on survival phenotype?”
Tumor-derived cytokines, specifically IL-1β and G-CSF, significantly increase the survival of neutrophils, thereby promoting their accumulation in the tumor microenvironment.
- Mechanism: These cytokines inhibit neutrophil apoptosis, while migration inhibitory factor further prevents programmed cell death.
- Result: This prolonged survival leads to a substantial accumulation of tumor-associated neutrophils (TANs), particularly those adopting pro-tumoral, immunosuppressive phenotypes.
Brave Search 07-08-2026
Pubmed: Recruited by HMGB1 neutrophils accumulate in tumor cell?
HMGB1 (High Mobility Group Box 1) acts as a damage-associated molecular pattern (DAMP) that recruits neutrophils and other leukocytes to the tumor microenvironment, facilitating cancer progression and metastasis.
- Recruitment Mechanism: Tumor-derived HMGB1 attracts TLR4+ neutrophils to primary tumor sites, as demonstrated in melanoma and breast cancer models, where it induces CD62Ldim neutrophil polarization via the TLR2 or TLR4 pathways.
- Accumulation and Function: These recruited neutrophils accumulate in the tumor and surrounding tissues, where they release Neutrophil Extracellular Traps (NETs) containing HMGB1. This accumulation promotes tumor cell intravasation, migration, and lung metastasis by activating signaling pathways such as NF-κB and increasing IL-8 secretion.
- Clinical Relevance: Elevated serum and tissue levels of HMGB1 are correlated with increased neutrophil infiltration, aggressive tumor phenotypes, and poorer survival outcomes in cancers including triple-negative breast cancer, glioma, and colorectal cancer
Brave Search 07-08-2026