SMARCD1 is progressively silenced in ccRCC
The paper starts from a survival comparison among core SWI/SNF subunits in the GEO dataset GSE22541 (n = 68). SMARCD1 expression had the strongest association with favorable outcome among those subunits (HR = 0.695, 95% CI 0.528–0.916, P = 0.00974), which made it the candidate worth following.
The expression gradient is the more striking observation: SMARCD1 declined from normal kidney to primary tumor and further in metastatic lesions, and low expression in the same GSE22541 cohort corresponded to shorter disease-free survival. At the protein level, paired tumor and adjacent normal tissue from 20 patients showed an average 50% reduction in tumors (paired t-test, P < 0.001), and an independent 63-patient immunohistochemistry cohort showed strong nuclear SMARCD1 in normal renal epithelium that was reduced or absent in tumors.
The authors then asked how the gene is switched off. Global 5-methylcytosine levels did not differ significantly between tumors that later recurred or metastasized and matched normal tissue (P = 0.2208), so global methylation was not the explanation. Bisulfite sequencing instead found elevated methylation at CpG island I of the SMARCD1 promoter (mean difference 11.09%, P = 0.0009), and droplet digital PCR in eight tumor–normal pairs showed copy number near 1:1, arguing against genomic deletion as the main cause.
DNMT3B is the methyltransferase that represses SMARCD1
To connect the methylation mark to actual repression, the study overexpressed three DNA methyltransferases in HEK-293 cells. Only DNMT3B overexpression significantly repressed SMARCD1 (P = 0.0074); DNMT1 and DNMT3A had no significant effect. That comparison is what supports the paper's claim that a specific de novo methyltransferase, rather than methylation in general, silences this subunit.
The pharmacological test runs in the other direction. Treating renal cancer cells with the methyltransferase inhibitor decitabine restored SMARCD1 protein in a dose-dependent manner in 786-O and Caki-1 cells (dose effect F(5,24) = 4.36, P < 0.05), with similar upregulation in A-498, OS-RC-2 and Renca cells. ACHN cells responded minimally, a heterogeneity the authors note rather than explain. Decitabine also attenuated the increased proliferation, clonogenicity, migration and invasion produced by SMARCD1 knockdown, tying the methylation state to the tumor phenotypes.
Losing SMARCD1 makes renal cancer cells more aggressive
The functional case rests on knockdown plus rescue. Two independent siRNAs targeting SMARCD1 were validated, and the more effective one was used throughout. Knockdown increased proliferation in CCK-8 and EdU assays, raised colony counts, and increased migration and invasion in Transwell assays across Renca, 786-O and Caki-1 cells (n = 5 independent experiments). Re-expressing an siRNA-resistant SMARCD1 cDNA reversed those gains, which addresses the concern that the phenotype came from off-target siRNA effects.
Knockdown did not change the paralogs SMARCD2 and SMARCD3, so the effect is not simply loss of the SMARCD family. In vivo, mice injected with shSMARCD1 Renca cells showed reduced survival (log-rank P < 0.001) and increased lung weight, and metastatic nodules appeared even at a low inoculation dose of 2 × 10^5 cells, whereas control mice showed no evidence of metastasis. The authors frame this aggressive phenotype as proceeding through epithelial–mesenchymal transition.
DDX3X, MAFF and MFN1 form the mechanistic chain
The mechanism the paper proposes is a relay. SMARCD1 normally acts as a nuclear gatekeeper that restricts DDX3X; when SMARCD1 is lost, DDX3X binds the MAFF promoter and raises transcription of MAFF, a stress-responsive bZIP transcription factor. MAFF then promotes ubiquitin-mediated degradation of MFN1, a mitochondrial fusion protein. The result is fragmented mitochondria rather than an elongated fusion network.
Two features of this chain matter for interpretation. First, the degradation is selective: MFN2, OPA1 and DRP1 were spared, which the authors take to mean MAFF works through specific E3 ligase complexes or adapters that remain unidentified. Second, the metabolic consequence runs against the usual picture of ccRCC. Rather than relying on glycolysis, SMARCD1-deficient cells shift toward oxidative phosphorylation, and the paper reports increased ATP production and altered redox balance as advantages for metastatic dissemination.
The causal claim for invasion is tested genetically: overexpressing MFN1 in SMARCD1-depleted cells reversed both mitochondrial fusion and the invasive cell morphology, including lamellipodia and invadopodia formation. That rescue is the strongest evidence that mitochondrial fragmentation is doing work in the invasive phenotype rather than merely accompanying it.
Leflunomide works only through MFN1 restoration
The therapeutic candidate is leflunomide, an FDA-approved immunosuppressant that induces MFN1 expression. In syngeneic subcutaneous and metastasis models, mice bearing shSMARCD1 or control Renca cells received 20 mg/kg daily by oral gavage. Leflunomide reversed mitochondrial fragmentation, increased mitochondrial length, raised oxygen consumption rate in isolated tumor cells, and suppressed tumor volume and weight relative to vehicle.
The specificity test is the key result: genetic ablation of MFN1 completely abolished leflunomide's antitumor effects, indicating that the drug acts through MFN1-dependent restoration of mitochondrial fusion. The authors also address why the drug can still work when MAFF accelerates MFN1 protein turnover: MFN1 transcription is preserved in this setting, so leflunomide can raise residual protein enough to restore dynamics.
Combination benefit appears in patient-derived xenografts
The most clinically relevant experiment uses TJH-cohort-P4 patient-derived xenografts in NSG mice reconstituted with human peripheral blood mononuclear cells. Mice received 5 mg/kg axitinib intraperitoneally, 10 mg/kg toripalimab, 20 mg/kg leflunomide by oral gavage, or combinations, for 21 days (n = 8 mice per group). Adding leflunomide to axitinib plus anti-PD-1 improved antitumor efficacy compared with axitinib plus anti-PD-1 alone, measured as tumor volume, tumor weight and ΔAUC from an initial tumor volume of 300 mm³.
This is a preclinical result in a small number of xenograft groups, not evidence of benefit in patients. The paper's own proposal is a biomarker-enriched phase I/II trial selecting SMARCD1-low/MAFF-high tumors, with metabolic imaging and biomarker response correlation as endpoints. The rationale for that selection comes from the inverse SMARCD1–MAFF expression pattern and the progressive rise of MAFF in recurrent tumors described in the discussion.
What the axis does and does not establish
The study's central claim is a chromatin-to-mitochondria link: a core SWI/SNF subunit, silenced epigenetically, changes mitochondrial dynamics and metabolism in a way that promotes metastasis. The authors contrast this with the usual focus on accessory subunits such as PBRM1, arguing that core subunit loss impairs chromatin remodeling across all mSWI/SNF assemblies and may therefore have broader phenotypic consequences.
Several boundaries limit how far that claim travels. The human cohorts are modest (89 patients enrolled overall, with 20 paired samples for Western blot, 63 for immunohistochemistry and 8 pairs for copy number), and the authors state that sex- and gender-disaggregated analyses were not performed because of limited subgroup sizes. All animal work used male mice except the patient-derived xenografts, where mouse sex was matched to donor sex. The E3 ligase linking MAFF to MFN1 ubiquitination is unidentified, and the therapeutic evidence comes from preclinical models only.
The practical priority the paper establishes is therefore specific: SMARCD1-deficient, MAFF-high ccRCC is the setting where restoring MFN1 with leflunomide is proposed to matter, and where the combination with tyrosine kinase inhibition and immune checkpoint blockade was tested. The decisive experiment for that priority is the MFN1 ablation that erased leflunomide's effect, because it makes MFN1 restoration, not leflunomide itself, the mechanism worth measuring in any follow-up.
Sources used in this answer
SWI/SNF subunit SMARCD1 deficiency promotes clear cell renal cell carcinoma progression via mitochondrial reprogramming
SMARCD1 is silenced by DNMT3B-mediated promoter hypermethylation; its loss lets DDX3X activate MAFF, which drives ubiquitin-mediated MFN1 degradation, mitochondrial fragmentation, oxidative phosphorylation and metastasis, and leflunomide reverses this in an MFN1-dependent manner in preclinical models.