Hypothesis

Spatial stromal multi‑omics can forecast EGFR‑TKI resistance in NSCLC by detecting exosome‑mediated metabolic reprogramming in cancer‑associated fibroblasts (CAFs) before any tumor‑genomic alteration appears.

Background

Multi‑omics integration reveals resistance mechanisms invisible to genomics, capturing transcriptomic, proteomic, metabolomic and epigenomic shifts that drive tumor adaptation [1] [2]. In EGFR‑mutant NSCLC, resistance often arises through bypass signaling such as MET activation, a process traced from mutation to phenotypic change via proteogenomic workflows [3]. Critically, the tumor microenvironment confers resistance through epigenetic remodeling of stromal cells like CAFs and tumor‑associated macrophages (TAMs) [5]. Despite these advances, clinical multi‑omics remains largely baseline and tumor‑centric, lacking prospective, spatial reads of stromal evolution under therapeutic pressure [9] [10].

Mechanistic Insight

We propose that CAFs under EGFR‑TKI stress release exosomes enriched in specific miRNAs (e.g., miR‑21) and metabolic enzymes (e.g., LDHA, GLUT1) that remodel the redox state of neighboring cancer cells. This exosomal cargo raises cytosolic NADH/NAD⁺, stabilizes HIF‑1α independently of hypoxia, and fuels a glycolytic‑dependent bypass cascade that activates MET/PI3K‑AKT signaling. Because these stromal changes precede measurable tumor‑genomic shifts, spatial multi‑omics of the CAF niche can serve as an early warning system.

Testable Prediction

In a prospective cohort of EGFR‑mutant NSCLC patients receiving first‑generation EGFR‑TKI, serial minimally invasive biopsies (or fine‑needle aspirates) will be taken from the tumor core and the invasive stromal front every 2 weeks. Each sample will undergo:

• Spatial transcriptomics (10x Visium) and proteomics (CODEX or MIBI) to map CAF activation states.
• Spatial metabolomics (MALDI‑IMS) for lactate, NAD⁺/NADH ratios.
• Isolation of extracellular vesicles from interstitial fluid for miRNA‑seq and enzyme activity assays.

Parallel plasma will be monitored for ctDNA EGFR T790M and MET amplification using ultrasensitive PCR.

We predict that a statistically significant increase in CAF‑derived exosomal miR‑21 and LDHA activity, coupled with a spatial rise in stromal NADH/NAD⁺ ratio, will occur a median of 6 weeks before ctDNA‑detectable resistance mutations. Patients lacking this stromal signature will either remain sensitive or develop resistance through alternative, genomically detectable routes.

Falsifiability

If longitudinal spatial multi‑omics fails to show any consistent stromal exosomal or metabolic shift that anticipates resistance—i.e., the timing and magnitude of CAF changes are random with respect to ctDNA emergence—the hypothesis is falsified. Conversely, observation of the predicted stromal lead‑time would support the model and justify early stromal‑targeted interventions (e.g., LDHA inhibitors or exosome release blockers) to delay or prevent resistance emergence.

Hypothesis

Engineered iPSC lines lacking HLA class I/II and expressing PD‑L1 generate cardiomyocytes that engraft in aged myocardial tissue, survive long term, and improve function without requiring chronic immunosuppression.

Mechanistic Rationale

Aged hearts exhibit heightened innate immune surveillance and reduced regenerative capacity, which contributes to poor outcomes after cell therapyiPSC dominance. HLA molecules on graft cells trigger recipient T‑cell activation, while PD‑L1 engagement delivers inhibitory signals that dampen effector responses. Removing HLA eliminates direct antigen presentation, and overexpressing PD‑L1 reinforces peripheral tolerance. In aged tissue, senescent cells secrete SASP factors that can upregulate PD‑1 on infiltrating lymphocytes, making PD‑L1 blockade a relevant checkpointiPSC immunoevasion. Combining HLA knockout with PD‑L1 overexpression should therefore create a double shield: loss of antigenic visibility plus active inhibition of residual immune activity.

Testable Predictions

1. iPSC‑derived cardiomyocytes with HLA‑KO/PD‑L1 will show <5 % T‑cell mediated lysis in vitro compared with wild‑type iPSC‑cardiomyocytes when co‑cultured with aged human peripheral blood mononuclear cells.
2. In immunodeficient mice reconstituted with aged human immune cells, grafts of HLA‑KO/PD‑L1 cardiomyocytes will retain >70 % engraftment at 8 weeks, whereas wild‑type grafts will drop below 30 %.
3. Functional improvement (ejection fraction increase ≥10 %) will be observed in aged murine myocardial infarction models receiving HLA‑KO/PD‑L1 cardiomyocytes without any postoperative immunosuppressants.
4. Adding a PD‑1 blocking antibody to the circulation will abolish the survival advantage of HLA‑KO/PD‑L1 grafts, confirming PD‑L1 dependence.

Experimental Design

• Cell preparation: Generate iPSCs from a healthy donor, edit with CRISPR‑Cas9 to knockout HLA‑A, HLA‑B, HLA‑C, HLA‑DR, HLA‑DQ, and HLA‑DP loci, then insert a constitutive PD‑L1 cassette. Differentiate to cardiomyocytes using established protocols (80‑90 % efficiency)iPSC differentiation. Validate HLA surface loss by flow cytometry and PD‑L1 overexpression by western blot.
• In vitro assay: Co‑culture 1 × 10⁴ cardiomyocytes with 5 × 10⁴ PBMCs from donors >65 years old for 48 h. Measure lactate dehydrogenase release and Annexin V/PI staining.
• In vivo model: Induce myocardial infarction in 12‑month‑old NSG mice, then reconstitute with irradiated aged human PBMCs (1 × 10⁶ cells via tail vein). Inject 1 × 10⁶ cardiomyocytes into the infarct border zone. Groups: (1) HLA‑KO/PD‑L1, (2) wild‑type iPSC‑cardiomyocytes, (3) vehicle. No cyclosporine or tacrolimus administered.
• Readouts: Bioluminescence imaging for graft survival weekly, MRI for ejection fraction at 4 and 8 weeks, histology for CD45⁺ infiltration and caspase‑3 apoptosis at endpoint.
• Checkpoint test: In a subset of HLA‑KO/PD‑L1 recipients, administer anti‑PD‑1 (200 µg i.p. twice weekly) starting day 0.

Potential Outcomes and Falsifiability

If the data confirm predictions 1‑3 and show that prediction 4 abolishes graft benefit, the hypothesis is supported: HLA knockout plus PD‑L1 overexpression creates an immunoprivileged iPSC‑derived graft suitable for aged hosts without chronic drugs. Conversely, if HLA‑KO/PD‑L1 cardiomyocytes exhibit lysis rates comparable to wild‑type, fail to engraft beyond 30 %, or show no functional improvement despite lack of immunosuppression, the hypothesis is falsified. Likewise, if PD‑1 blockade does not reduce graft survival, the claimed mechanism would be incorrect, prompting re‑evaluation of other immune pathways.

This framework directly tests whether engineering universal iPSC lines can overcome the age‑related immune barrier that has limited ESC and unedited iPSC therapies, offering a clear, falsifiable path toward clinical translation.

Hypothesis

Expressing a newly engineered polychromatic opsin that responds to broad-spectrum ambient light in surviving retinal bipolar cells will restore intracellular calcium homeostasis, thereby re-establishing proper signal transduction to retinal ganglion cells and improving visual perception in advanced retinal degeneration.

Rationale

• Current optogenetic approaches target retinal ganglion cells (RGCs) with microbial opsins that require high-intensity artificial light and lack adaptation 1, 4, which limits natural vision and may exacerbate excitotoxic stress.
• In late-stage degeneration, bipolar cells undergo remodeling that alters calcium channel expression and disrupts the outer plexiform layer, impairing their ability to buffer calcium signals from photoreceptors 2.
• Mammalian cone opsins (e.g., MW-opsin) expressed in RGCs confer light adaptation and sensitivity to natural light 1, suggesting that the opsin’s native signaling cascade can modulate intracellular second messengers beyond simple depolarization.
• Engineering a polychromatic opsin (inspired by recent ambient-light-sensitive designs 4) to couple to G-protein pathways that activate plasma-membrane Ca2+-ATPase (PMCA) or sodium-calcium exchanger (NCX) would provide a light-driven calcium extrusion mechanism.

Mechanistic Insight

We hypothesize that light activation of this opsin triggers a Gq/11-mediated phospholipase C pathway, producing IP3 that stimulates Ca2+ release from internal stores, followed by activation of store-operated calcium entry (SOCE) and subsequent activation of PMCA/NCX via calcium-calmodulin kinase II (CaMKII). The net effect is a light-dependent increase in calcium clearance, reducing resting cytosolic Ca2+ levels that are pathologically elevated in remodeled bipolar cells. Normalized calcium levels would restore the voltage-dependent release of glutamate at bipolar cell terminals, re-establishing faithful transmission to downstream RGCs.

Testable Predictions

1. In vitro, bipolar cells from rd10 mice expressing the polychromatic opsin will show a significant decrease in baseline Fluo-4 fluorescence after 5 min of ambient white light (~10 lx) compared with cells expressing ChrimsonR or GFP controls (p < 0.01).
2. Patch-clamp recordings will reveal restored light-evoked inhibitory postsynaptic currents in RGCs that correlate with the opsin-mediated calcium decrease, indicating recovered synaptic transmission.
3. In vivo, subretinal AAV delivery of the opsin to bipolar cells in late-stage rd10 mice will improve optomotor response thresholds under natural daylight conditions (~100 lx) by at least 0.5 log units relative to ChrimsonR-treated mice, without requiring external goggles.
4. Calcium imaging of retinal slices will demonstrate that light-evoked calcium spikes in bipolar cells are shorter in duration and amplitude, reflecting enhanced clearance.

Falsifiability

If ambient light exposure fails to lower basal calcium in opsin-expressing bipolar cells, or if RGC light responses do not improve despite opsin expression, the hypothesis is refuted. Likewise, if the opsin’s effect is abolished by pharmacological inhibition of PMCA/NCX but not by blocking phospholipase C, the proposed calcium-extrusion mechanism would be invalidated.

Implications

Confirming this mechanism would shift optogenetic design from merely driving spikes to actively repairing ionic homeostasis in remodeled inner retinal neurons, potentially extending therapeutic efficacy to patients with advanced retinal degeneration and reducing reliance on bulky illumination devices.

Hypothesis

Expressing a low‑conductance, red‑shifted human opsin (e.g., melanopsin variant) directly in surviving rods during the early phase of retinal degeneration provides tonic, sub‑threshold light‑driven signaling that activates Ca2+-dependent survival pathways, thereby slowing photoreceptor loss and preserving retinal thickness.

Rationale

Recent work shows that red‑shifted microbial channelrhodopsins enable daylight‑compatible activation while reducing blue‑light phototoxicity Red‑shifted variants for safer activation. Human opsins, when expressed in inner retinal neurons, require lower irradiance, exhibit built‑in adaptation, and better preserve circuit architecture Human opsins require lower light intensity and better emulate natural adaptation. Clinical data from MCO‑010 trials demonstrate durable visual acuity gains in late‑stage RP without goggles MCO‑010 shows durable BCVA gains in late‑stage RP and mutation‑agnostic bypass of degenerated photoreceptors Eight of 11 trials target RP with mutation‑agnostic approach. Moreover, MCO‑010 exhibits neuroprotective effects in rd1/rd10 models when delivered to inner retinal neurons after photoreceptor death MCO‑010 demonstrates neuroprotective effects in mouse models. These findings suggest that optogenetic activity itself can convey protective cues, yet the potential of delivering such activity directly to stressed photoreceptors remains unexplored.

We propose that a weakly conducting human opsin expressed in rods will generate a steady, low‑level depolarization in response to ambient light. This modest Ca2+ influx can activate calmodulin‑dependent kinase II (CaMKII) and the cAMP response element‑binding protein (CREB) pathway, which are known to upregulate antioxidant genes (e.g., Nrf2 targets) and inhibit apoptosis. Importantly, the opsin’s low conductance avoids excitotoxic overload, while its red‑shifted spectrum minimizes additional phototoxic stress.

Experimental Design

1. Vector construction – AAV2‑shH10‑driven expression of a red‑shifted melanopsin variant (peak sensitivity ~620 nm) fused to a fluorescent tag; include a control AAV expressing GFP alone.
2. Animal model – rd1 mice at postnatal day 10 (early degeneration, before significant outer nuclear layer loss). Sub‑retinal injection of AAV (1 × 10^9 vg per eye).
3. Light exposure – Cyclic 12 h light/12 h dim red light (λ > 600 nm, ≤5 µW/cm²) to mimic physiological illumination without adding blue‑light stress.
4. Readouts (at 4 weeks post‑injection)
   • Outer nuclear layer thickness via OCT histology.
   • Full‑field ERG amplitudes (a‑wave and b‑wave).
   • TUNEL assay for apoptotic photoreceptors.
   • Western blot for p‑CaMKII, p‑CREB, and Nrf2 target proteins (HO‑1, NQO1).
   • Behavioral optomotor response to assess functional vision.
5. Controls – (a) GFP‑only AAV, (b) wild‑type mice with same vector, (c) rd1 mice treated with MCO‑010 AAV targeting bipolar cells (to compare inner‑vs‑outer retinal protection).

Predicted Outcomes

• Eyes receiving the photoreceptor‑targeted opsin will show significantly greater outer nuclear layer preservation (≥30 % thicker) and higher ERG a‑wave amplitudes compared with GFP controls.
• Increased p‑CaMKII/p‑CREB and Nrf2‑dependent antioxidant expression will be detected only in opsin‑treated retinas.
• TUNEL‑positive photoreceptor numbers will be reduced, correlating with improved optomotor tracking.
• No exacerbation of phototoxic damage will be observed under the red‑light regimen, confirming safety of the red‑shifted opsin.

Potential Pitfalls and Mitigations

• Over‑expression leading to calcium overload – titrate viral dose and use a weaker promoter (e.g., human synapsin minimal) to limit opsin levels.
• Off‑target transduction of neighboring cells – incorporate a photoreceptor‑specific miRNA target site (miR‑124) in the vector backbone to suppress expression in inner retinal neurons.
• Compensatory homeostatic downregulation – include a destabilizing domain (DD) fused to the opsin that can be stabilized with a low dose of Shield‑1 ligand if needed.

This hypothesis is directly testable: if tonic, low‑level optogenetic signaling in photoreceptors activates endogenous survival cascades and slows degeneration, the predicted structural, molecular, and functional improvements should be observed. Failure to detect these changes would falsify the idea that photoreceptor‑targeted optogenetic activity confers neuroprotection in early retinal disease.

Hypothesis

Aged human β‑cells lose identity transcription factors (PDX1, MAFA, INS) and exhibit chronic ER stress that uncouples ATF6α signaling from XBP1‑mediated transcription, while simultaneously increasing O‑GlcNAcylation of the prohormone convertases PC1/3 and PC2 due to heightened flux through the arginine biosynthesis and pentose phosphate pathways. This post‑translational modification impairs proIAPP processing, leading to accumulation of incompletely processed proIAPP that binds heparan sulfate proteoglycans (HSPGs) on the ER lumen, seeding amyloid nucleation. The resulting IAPP oligomers exacerbate ER stress, suppress autophagy via ATF4‑dependent transcriptional repression, and create a vicious cycle that drives β‑cell death and T2D progression.

Mechanistic Rationale

• ATF6α–XBP1 uncoupling: In young β‑cells, ATF6α cleavage and XBP1 splicing act cooperatively to expand the ERAD and lysosomal biogenesis programs ([3],[4]). Aging shifts ATF6α to smaller complexes and diminishes IRE1α‑dependent XBP1 splicing, breaking this coordination.
• Metabolic rewiring: Proteomics of aged islets shows up‑regulated arginine biosynthesis and pentose phosphate pathway metabolites ([2]), providing substrates for the hexosamine pathway and increased O‑GlcNAc transferase (OGT) activity.
• O‑GlcNAcylation of convertases: O‑GlcNAc modification of PC1/3 and PC2 reduces their catalytic efficiency, a mechanism previously observed in neuronal secretory pathways and here proposed for β‑cells.
• ProIAPP‑HSPG nucleation: Incompletely processed proIAPP exposes hydrophobic regions that bind HSPG heparan sulfate chains, a step accelerated by heparanase inhibition and blocked by heparanase overexpression ([6],[7]).
• Feedback to UPR: IAPP oligomers provoke PERK‑eIF2α‑ATF4 signaling, which up‑regulates CHOP and induces transcription of autophagy repressors (e.g., Rubicon), further suppressing autophagic clearance of amyloid.

Testable Predictions

1. In islets from donors >60 yr, O‑GlcNAc levels on PC1/3 and PC2 will be significantly higher than in <40 yr islets, correlating with reduced C‑peptide/IAPP ratio.
2. Pharmacological inhibition of OGT (e.g., with OSMI‑1) will restore proIAPP processing, decrease intracellular amyloid load, and increase PDX1/MAFA expression without affecting ATF6α cleavage.
3. Forced expression of a constitutively active XBP1s (spliced) in aged β‑cells will rescue autophagy flux (LC3‑II turnover) and reduce IAPP oligomer formation, even when ATF6α signaling remains uncoupled.
4. Heparanase overexpression will attenuate amyloid nucleation only when OGT activity is normalized, indicating that HSPG binding is downstream of proIAPP misprocessing.

Experimental Approach

• Obtain human pancreatic islets from organ donors stratified by age (20‑40, 40‑60, >60 yr). Measure O‑GlcNAc on PC1/3/PC2 via immunoprecipitation followed by Western blot with RL2 antibody; quantify proIAPP/IAPP ratios by ELISA.
• Treat cultured islets with OSMI‑1 (OGT inhibitor) or vehicle for 48 h; assess autophagy flux (bafilomycin A1 chase, LC3‑II/p62), ER stress markers (HSPA5, phospho‑eIF2α, ATF4), and identity genes (PDX1, MAFA, INS) by qPCR and immunofluorescence.
• Use adenoviral vectors to express XBP1s or a cleavage‑resistant ATF6α construct; evaluate amyloid deposition with Thioflavin T staining and confocal microscopy.
• Modulate heparanase levels (overexpression or siRNA) and examine amyloid formation in the presence/absence of OGT inhibition.
• Statistical analysis: ANOVA with post‑hoc Tukey; n≥6 donors per age group.

If OGT inhibition restores proIAPP processing and reduces amyloid despite persistent ATF6α–XBP1 uncoupling, the hypothesis is supported; failure to rescue processing would falsify the proposed mechanistic link.

Hypothesis

We hypothesize that integrating spatially resolved metabolomics, proteomics, and transcriptomics with AI‑driven network inference reveals metabolite‑protein hubs that drive epigenetic reprogramming preceding drug resistance in solid tumors.

Mechanistic Rationale

Multi‑omics integration already improves disease classification and causal target identification [1][2]. Methods like sparse PLS‑DA and variational autoencoders now handle high dimensionality and missing data, allowing true integration rather than simple concatenation [4]. It's becoming clear that metabolite‑enzyme contacts can act as rapid signaling nodes. Adding spatial information lets us see where metabolites accumulate relative to protein expression and gene activity within tumor niches. Recent work shows that combining proteogenomics with Mendelian randomization and AlphaFold3 pinpoints druggable proteins [5]. We extend this by proposing that specific oncometabolites (e.g., 2‑hydroxyglutarate, succinate) physically interact with chromatin‑modifying enzymes, altering their activity and thus reshaping the epigenetic landscape. These interactions are missed when omics layers are averaged across bulk tissue.

Testable Predictions

1. In pretreatment biopsies, spatial colocalization of high 2‑hydroxyglutarate levels with increased DNMT3A protein predicts subsequent resistance to EGFR inhibitors.
2. We don't expect these hubs to be evident in normal tissue.
3. Disrupting the metabolite‑enzyme interaction (using a small‑molecule blocker) will reduce histone methylation marks and restore drug sensitivity in patient‑derived organoids.
4. Multi‑omic spatial models will outperform bulk multi‑omic models in predicting progression‑free survival, with an expected increase in median MCC of at least 0.15.

Experimental Design

• Collect matched pretreatment and post‑progression samples from 60 patients with non‑small‑cell lung cancer receiving EGFR‑TKI therapy.
• We're performing imaging mass spectrometry for metabolites, multiplexed ion beam imaging for proteins, and spatial transcriptomics (10x Visium) on adjacent sections.
• Use a variational autoencoder framework to fuse the three modalities, generating a joint latent space that captures cross‑layer interactions.
• Apply causal network inference (e.g., ICM) to identify metabolite‑protein edges that correlate with changes in DNA‑methylation signatures measured by bisulfite sequencing.
• Validate top hubs by CRISPR‑knockout of the protein or chemical inhibition of the metabolite in organoid cultures, measuring drug response and epigenetic marks.
• Statistical analysis: compare prediction accuracy of spatial vs bulk models using DeLong test for AUC; require p<0.05 to reject null hypothesis of no improvement.

Potential Impact

If confirmed, this approach would provide a mechanistic link between microenvironmental metabolism and epigenetic drug resistance, enabling patient‑specific combinatorial therapies that target the metabolite‑protein axis before resistance emerges.

Hypothesis

A computable splicing entropy score (SES) that integrates intron retention, cryptic splice‑site usage, and exon skipping across the transcriptome can distinguish adaptive splicing diversity from deleterious transcriptomic entropy in SF3B1‑mutant cancers. Tumors with SES above a defined threshold will exhibit synthetic lethality when treated with spliceosome‑targeting agents such as SF3B1 inhibitors or CLK2/3 inhibitors, whereas tumors below the threshold will be resistant.

Mechanistic basis

Heterozygous spliceosome mutations reprogram splice site recognition, generating a mixture of functional and aberrant isoforms. Low‑level isoform variation can support tumor adaptability (e.g., drug‑resistant NT5C2 variants) while excessive mis‑splicing overwhelms RNA‑processing capacity, leading to nuclear stress, R‑loop accumulation, and activation of the p53‑dependent checkpoint. The SES captures the burden of GC‑rich intron retention and cryptic 3′ splice‑site activation that directly correlates with splicing‑factor condensate gelation under ATP‑limiting conditions. When SES exceeds the cellular buffering capacity, further pharmacological inhibition of spliceosome kinetics pushes the system past a catastrophic entropy point, causing lethal transcriptomic collapse.

Testable predictions

1. In a panel of SF3B1‑mutant cell lines, SES calculated from RNA‑seq will inversely correlate with IC50 values of H3B‑8800 and E7107.
2. CRISPR‑mediated reduction of intron‑retention–prone genes (e.g., BRD4, MAPK1) will lower SES and confer resistance to spliceosome inhibitors without altering mutant allele frequency.
3. Elevating cellular ATP via overexpression of mitochondrial uncoupling protein 2 will decrease SES‑dependent sensitivity, linking energy status to the entropy threshold.
4. Patient‑derived xenografts with pretreatment SES above the 75th percentile will show tumor regression upon CLK2/3 inhibitor treatment, while those below will exhibit stable disease.

Experimental design

• Generate isogenic SF3B1‑WT and SF3B1‑K700E lines in hematopoietic progenitors. Perform deep RNA‑seq, compute SES as (weighted sum of retained intron reads + cryptic 3′ splice‑site junction reads) / total mapped reads, weighting GC‑rich introns higher.
• Treat cells with dose‑response curves of H3B‑8800, E7107, and a CLK2/3 inhibitor. Fit IC50 and compare to SES using Spearman correlation.
• Use lentiviral shRNA to knock down selected retained‑intron genes; re‑measure SES and drug sensitivity.
• Modulate ATP levels with oligomycin or ATP‑boosting compounds; reassess SES and drug response.
• Implant SF3B1‑mutant patient‑derived xenografts in NSG mice, baseline RNA‑seq to assign SES, randomize to CLK2/3 inhibitor or vehicle, monitor tumor volume and perform endpoint RNA‑seq to verify SES changes. Statistical significance set at p<0.05, with power analysis targeting 80% detection of a 30% IC50 shift.

References: Spliceosome mutations reprogram splice site recognition, Pan-cancer analyses identify >15,000 cancer-specific splice variants, Emerging therapies include PRMT5, CLK2/3 inhibitors, and ASOs, Synthetic introns target SF3B1-mutant cells