Mice and housing conditions
All in vivo experiments were conducted in accordance with protocols approved by the Brigham and Women’s Hospital Institutional Animal Care and Use Committee (IACUC) and Stanford University IACUC. Mice were housed according to the standard guidelines with free access to food and water in a 12 h light:12 h dark cycle.
For brain tumour allograft experiments, NSG mice (NOD-SCID-IL2R gamma chain-deficient, the Jackson Laboratory) were used. Male and female mice were used equally. According to the IACUC guidelines, signs of morbidity rather than maximal tumour volume was used as indication for termination of brain allograft mouse experiments. Mice were euthanized if they exhibited signs of neurological disease or if they lost 15% or more of their body weight. For in vivo optogenetic stimulation of the premotor circuit (M2), Thy1-ChR2; NSG or WT; NSG mice were used.
For lung tumour experiments, mice were euthanized when they exhibited signs of sickness behaviour (such as dyspnea, abnormal gait or posturing, or ill-groomed fur) or lost >15% of body weight in accordance with IACUC guidelines. No limits were exceeded in any mouse experiments. In these experiments, Rb1fl/fl;Trp53fl/fl;p130fl/fl, luciferase-expressing (RPR2-luc) genetic mouse models were used as described previously39,44. In these mice, lung tumours and later distant metastases form spontaneously after intratracheal administration of Adeno-CMV-Cre (University of Iowa Vector Core) at 2 months of age as described44 and following a published protocol60. To study MYC-driven lung tumour models, we used RPM mice42. These mice were infected intratracheally as above with Adeno-CMV-Cre or Adeno-CGRP-Cre at 3 to 4 months of age.
Intracranial allografts
All SCLC brain allografts were performed as described2. In brief, a single-cell suspension from cells cultured from either 16T-mGFP or NCI-H446-GFP SCLC neurospheres was prepared in sterile HBSS immediately before surgery. Mice at postnatal day (P)21–35 were anaesthetized with 1–4% isoflurane and placed in a stereotactic apparatus. The cranium was exposed via midline incision under aseptic conditions. 70,000 cells in 3 µl sterile HBSS were stereotactically injected into the M2 region of the cortex through a 31-gauge burr hole using a digital pump at an infusion rate of 0.4 µl min−1 and a 31-gauge Hamilton syringe. Stereotactic coordinates for Thy1-ChR2 mouse allografts used were as follows: 0.5 mm lateral to midline, 1.0 mm anterior to bregma, −1.5 mm deep to cortical surface. At the completion of infusion, the syringe needle was allowed to remain in place for a minimum of 2 min, then withdrawn at a rate of 0.875 mm min−1 to minimize backflow of the injected cell suspension. To generate mice with interneuron-specific expression of light-sensitive ion channels, 1 µl of AAV8-Dlx-ChRmine-p2A-mCherry61 (virus titre = 6.9 × 1012) (a gift from K. Deisseroth) was unilaterally injected using a Hamilton Neurosyringe and 450 Stoelting stereotaxic injector over 5 min. These mice then received allografts of 16T-mGFP or xenografts of NCI-H446-GFP SCLC cells as described above at 0.8 mm lateral to midline, 1.0 mm anterior to bregma, −1.2 mm deep relative to the cortical surface. Allografts of 16T-mGFP-ChR2 SCLC cells were performed following the procedure as above with exception for the following alterations: 15,000 cells were injected in 1 µl sterile HBSS 0.8 mm lateral to midline, 1.0 mm anterior to bregma, −1.2 mm deep relative to cortical surface. For electrophysiology experiments and electron microscopy tissue analysis, 16T-mGFP cells were allografted into the CA1 region of the hippocampus at the following coordinates: 1.5 mm lateral to midline, 1.8 mm posterior to bregma, −1.35 mm deep to the cortical surface.
In vivo optogenetic manipulation
For in vivo optogenetic stimulation of M2 region of Thy1-ChR2; NSG or WT; NSG mice, a single stimulation paradigm was employed as previously described1. In brief, a fibre optic ferule was placed 1 week following or simultaneously with and ipsilateral to the SCLC allografts. After 1–2 weeks to allow for recovery from the procedure, the mice were connected to a 100-mW 473-nm diode-pumped solid-state laser system with a mono fibre patch cord, which freely permits wakeful behaviour of the mice. Pulses of light with approximately 4 mW measured output at tip of the patch cord were administered at a frequency of 20 Hz for periods of 30 s, followed by 90 s recovery periods, for a total session duration of 30 min. The mice were euthanized after 24 h post stimulation, and brains were collected for histological analysis. For stimulation of cortical interneurons in Dlx-ChRmine mice, 595-nm light was used at 40 Hz frequency and 10 ms width. Pulses of light with approximately 10 mW measured output at tip of the patch cord were administered.
For in vivo optogenetic depolarization of SCLC cells, ChR2–YFP (pLV-ef1-ChR2(H134R)-eYFP WPRE) construct (generated by the laboratory of K. Deisseroth and placed in the piggyback transposon system by M. Su in the laboratory of M.M.) was lentivirally transduced into 16T SCLC cells, which were then allografted into premotor cortex (M2) following the procedure described above. A fibre optic ferule was implanted during the same surgery ipsilateral to the cell injection site at following coordinates: 0.8 mm lateral to midline, 1.0 mm anterior to bregma, −0.9 mm deep relative to cortical surface. At 1 week post-allograft and for three consecutive days, all mice were connected to the laser system to receive blue light or mock stimulations at a frequency of 10 Hz for periods of 30 s, followed by 90 s recovery periods for a total session duration of 30 min. Mice were euthanized 24 h after the final (3rd) stimulation session.
Immunohistochemistry of patient tissue
Patient tissue samples were obtained with informed consent and analysed in accordance with institutional review board-approved protocols. Immunohistochemistry of patient SCLC brain metastases tissue samples was performed on formalin-fixed paraffin-embedded tissue sections per standard protocols including deparaffinization, antigen retrieval, incubation with primary antibody and detection per the manufacturers’ instructions. The following antibodies were used: mouse anti-Ki67 (Dako/Agilent), mouse anti-neurofilament (Ventana Roche; prediluted). Staining for Ki67 was performed on a Leica Bond III automated stainer. Staining for neurofilament was performed on a Ventana Ultra automated stainer. Proliferation index was determined by quantifying the fraction of Ki67+ cells out of the total number of cells in the region.
Single-cell sequencing from SCLC-neuron co-culture
16T SCLC cells were cultured alone or in neuron co-cultures with or without 1 µM TTX for 24 h per protocol below and collected in PBS 0.5% BSA, 1 mM EDTA (Invitrogen), 1x DNAse (Worthington Biochemical). GFP-negative cells were collected in parallel to serve as negative control during FACS. Calcein violet (Thermo Fisher) was used to label live cells. GFP+calcein+ cells were sorted and collected then lysed and combined into droplets with barcoded beads which captured the mRNA then used for reverse transcription with the The Chromium Single Cell Gene Expression platform (10X Genomics) per the manufacturer’s instructions. We then followed the rest of the 10X standard or high-throughput protocols and used the Dual Index Kit TT Set A for library production. The experiment was repeated for a total of three biological replicates.
Processing of fastq files from monoculture and co-culture samples was performed individually using the 10X Genomics Cell Ranger 7.1.0 based on the mm10 mouse genome reference, with the incorporation of the eGFP sequence. Seurat62 (v.5.0.1) was employed for data loading at the individual sample level. Subsequently, the scCB263 (v.1.12.0) package was utilized to filter out empty droplets, employing an FDR threshold of 0.01 to identify real cells, while potential doublets were removed using the scDblFinder64 (v.1.16.0) package. Cells with no GFP expression, exhibiting a high fraction of mitochondrial molecules (>5%) and those expressing a low number of unique genes (indicating low library complexity) were excluded.
Samples from the same replicate were merged, and highly variable genes were selected using the Seurat package. The relative expression values of these highly variable genes were used for principal component analysis (PCA). The number of PCA components for each replicate was determined based on achieving a cumulative proportion greater than 80% in the PCA plot. Subsequently, UMAP embeddings were generated and cells were clustered using Seurat’s Louvain algorithm-based FindClusters function.
Differentially expressed genes were identified using the SeuratWrapper’s (v.0.3.19) RunPrestoAll function. Genes detected in a minimum of 30% of the cells within each cluster, with at least a 0.25-fold mean log difference, were subjected to statistical testing using the Wilcoxon rank sum test with Bonferroni correction for multiple testing. Genes with adjusted P value < 0.05 were retained.
GSEA was conducted using the fgsea65 (v.1.28.0) and genekitr66 (v.1.2.5) packages, exploring GO, KEGG, REACTOME, Hallmarks, Biocarta and WikiPathways databases. Finally, the GSVA67 (v.1.50.0) package was used to calculate the ssGSEA scores for synaptome, astrocytes, and cell proliferation signatures.
Single-cell sequencing from human primary or metastatic SCLC lesions
Human SCLC brain tissue transcriptomic library preparation
Frozen tissues were processed as described68,69. Tissue blocks were embedded in optimal cutting temperature (Tissue-Tek, Sankura 4583), and then sectioned on a Leica CM1950 cryostat (Leica) into 20-µm-thick curls (generating up to 20 curls for each sample). These were then placed in 5-ml tubes (Eppendorf), washed with ice-cold PBS (Thermo Fisher Scientific, 10010023), centrifuged at 400g for 2 min, and the supernatant was discarded. The tissue was resuspended in 1 ml Salt Tris (ST) buffer (146 mM NaCl, 10 mM Tris-HCL pH 7.5, 1 mM CaCl2 and 21 mM MgCl2 in ultrapure water) with 0.03% Tween-20 (Sigma Aldrich, P7949; TST buffer), supplemented with 0.1% BSA (New England Biolabs, B9000S) and optionally 40 U ml−1 RNAse inhibitor (RNAse OUT, Thermo Fisher Scientific). The suspension was mechanically dissociated by pipetting 15 times with a 1-ml pipette and incubated on ice for 5 min. Afterward, the pipetting step was repeated, and the reaction was quenched with 4 ml ST buffer, with or without RNAse inhibitor. The mixture was filtered through pre-wetted 70-µm nylon mesh filters (Thermo Fisher Scientific) into 50-ml conical tubes, washed with 5 ml ST buffer, and centrifuged at 500g for 5 min to isolate the nuclei. The nuclei pellet was resuspended in 100–400 µl ST buffer, filtered through a 40-µm mesh (Thermo Fisher Scientific), and counted using a Neubauer counting chamber (Bulldog Bio) after staining nuclear DNA with 50 µg ml−1 Hoechst 33342 (Thermo Fisher Scientific, H3570). Approximately 0.9 to 1.5 × 10³ nuclei were loaded into a Chromium Controller using ST buffer without RNAse inhibitor and processed with Chromium reagents and 5′V2 capture kits (1000006 and 1000263) from 10X Genomics. Following reverse transcription and cleanup, cDNA libraries were prepared per manufacturer protocols, including one additional cycle of amplification to account for the lower RNA content in nuclei compared with whole cells. Final sequencing libraries were created using the library construction kit (1000190) and Dual Index Kit TT Set A (1000215) and sequenced on an Illumina NovaSeq S4 platform with 2× 150 bp paired-end reads, achieving a minimum of 25,000 reads per cell.
Filtering background noise in gene expression matrices
Demultiplex FASTQ files from raw RNA-sequencing reads were aligned using CellRanger v.6.1.1 (10X Genomics) to the GRCh38 genome70. Gene counts were quantified with CellRanger’s ‘count’ function, including intronic reads. The feature_bc_matrix.h5 files generated by CellRanger were used as inputs for the ‘remove-background’ function in CellBender v.0.2.0, which removed ambient RNA gene counts and empty droplets71. The CellRanger metric ‘expected-cells’ defined the ‘Expected Number of Cells’ parameter, and the total-droplets-included parameter was set to a value between 10,000 and 40,000, chosen from the plateau region of the barcode-rank plot generated by CellRanger.
Quality control and normalization
Each generated matrix for was processed with R v.4.1.1 and Seurat v.4.1.0 on a per-sample basis72. Filters were applied based on the Seurat pipeline to retain only cells with 500–10,000 detected genes, 1,000–60,000 unique molecular identifiers and less than 10% mitochondrial gene content. Scrublet v.0.2.1 was used to identify and remove doublets, with the expected doublet rate set between 2.5% and 7.5%, depending on the initial loading rate73. Following Seurat’s pipeline, the data were log-normalized using the NormalizeData function. The top 2,000 variable genes from each sample were identified with the FindVariableFeatures function, and the resulting matrix was centred and scaled with the ScaleData function. All signatures were computed through entered the gene list and a merged Seurat object of all samples on a per-cohort basis using the AddModuleScore function provided by Seurat.
Integration of cohort samples
Individual Seurat samples were integrated using Seurat canonical correlation analysis (CCA) pipeline to remove batch effects from individual samples. Samples from the labelled Chan cohort from and CUIMC cohort were integrated separately72. Per the CCA pipeline, SelectIntegrationFeatures and FindIntegrationAnchors was then run to select 2,000 anchors between each sample with the top 50 dimensions from CCA to define search space for integration, using the raw RNA counts assay for each sample. IntegrateData was then run using the previously defined anchors to generate the integrated dataset. The integrated data were then scaled, and clustered using ‘FindNeighbors’ with 10 dimensions and FindClusters using a resolution of 0.5. UMAPs were calculated with the top 30 PCA dimensions, using Seurat’s RunUMAP.
Cell-type identification
Cell types were initially labelled using SingleR v.1.8.0 using the built in BlueprintEncodeData reference74. Immune cells identified through this process were used as a diploid reference for inferCNV v.1.10.1 to infer chromosomal copy number alteration (CNA) profiles for each cell. A minimum average read count threshold of 0.1 per gene was applied for reference nuclei. The ‘subcluster’ setting was used for clustering, and results were denoised with the default ‘sd_amplifier’ value of 1.5. InferCNV used a hidden Markov model to predict CNA levels, and the proportion of scaled CNAs was averaged across all chromosomes for each cell74. Malignant cells were identified using sample-specific thresholds based on these average values, which distinguished immune and non-immune CNA levels.
Non-malignant cell types were further analysed following CCA-based integration. Clustering of non-malignant cells was performed with the FindClusters function at varying resolutions, followed by differential gene expression analysis with FindAllMarkers72. Broad cell-type annotations were assigned manually based on established marker genes identified as differentially expressed in each cluster.
Non-negative matrix factorization
Non-negative matrix factorization (NMF) was employed for feature extraction and dimensionality reduction on non-negative gene expression data75. The NMF function implemented in RcppML v.0.3.7, was selected for its computational efficiency by minimizing reconstruction error and optimizations matrix factorization76. RcppML operates directly on raw count matrices and incorporates L1 regularization with reproducible factor scaling, enabling robust handling of ambiguous zeroes in single-cell data. NMF iteratively decomposing the data matrix into two lower-dimensional non-negative matrices, a basis matrix representing gene programmes and a coefficient matrix capturing the effect of each gene programmes on each cell. To incorporate prior biological knowledge, the supervised framework from Tagore et. al was adapted77. NMF was run on each sample to identify latent gene programmes, followed by rank-factor. Genes were pre-filtered to retain the top 7,000 genes based on total counts from the filtered RNA assay, serving as input.
Metaprogrammes (also referred to as consensus factors) were identified through co-correlation analysis using Spearman’s correlation applied to sample-specific factors generated by NMF. The optimal number of consensus factors was determined using silhouette and distortion scores. Ward’s clustering was then performed on individual factors, with the optimal number of consensus factors setting the clustering thresholds to define metaprogrammes77. This analysis was conducted on a per-cohort basis.
Metaprogramme correlation with annotated hallmarks of tumour heterogeneity
To annotate the biological functions underlying each metaprogramme, gene signatures were defined based on the top genes with the highest normalized contributions to each metaprogramme. Functional characterization of these gene signatures was performed through GSEA using recently described transcriptional hallmarks of tumour heterogeneity46. To assess the similarity between gene signatures, the Jaccard similarity index and a derived correlation metric were calculated. The Jaccard similarity index quantifies overlap between two sets as the size of their intersection divided by the size of their union. To facilitate interpretation in a heatmap, pairwise Jaccard similarity scores were scaled linearly to a correlation-like metric ranging from 0 to 1. The resulting heatmap highlights pairwise relationships between gene signatures, with metaprogrammes exhibiting the highest correlation with a given metaprogramme annotated as representing biologically related functions.
Normalized gene contribution towards metaprogrammes
For each metaprogramme, genes were ranked based on a weighted Stouffer-integrated expression value. The top 50 genes per metaprogramme, determined by this ranking, were selected for downstream analyses. Within each sample, cell barcodes from the H matrix generated by NMF were assigned to a specific metaprogramme based on their highest association with a corresponding factor. These assignments linked individual barcodes to metaprogrammes according to their factor membership. To evaluate the gene contributions to each metaprogramme, an expectation-maximization Gaussian mixture model (EM-GMM) was applied to the normalized gene expression matrices77. This model assessed the modality of gene expression distributions, enabling the identification of peaks that correspond to the ‘normalized gene contribution’ for each metaprogramme. These contributions were subsequently scaled between 0 and 1 across all metaprogrammes to ensure comparability.
GSEA across metaprogrammes
The top 50 ranked genes for each metaprogramme, based on the weighted Stouffer-integrated expression values, were used for functional enrichment analysis. GSEA was performed using EnrichR v.3.1, leveraging the Hallmarks of MSigDB v.7.4.1 and GO Biological Process 2021 databases78. Enrichment results were aggregated across cohorts, with the top-ranked gene sets identified for each metaprogramme. The P values and q scores returned by EnrichR were scaled and visualized to highlight significant functional associations for each metaprogramme.
Sample preparation and image acquisition for electron microscopy
NSG mice were engrafted with either 16T-mGFP or NCI-H446-GFP cells into mouse hippocampi. Three weeks after engraftment, mice were sacrificed by transcardial perfusion with Karnovsky’s fixative: 2% glutaraldehyde (EMS 16000) and 4% paraformaldehyde (PFA) (EMS 15700) in 0.1 M sodium cacodylate (EMS 12300), pH 7.4. The samples were then post-fixed in 1% osmium tetroxide (EMS 19100) for 1 h at room temperature, washed 3 times with ultrafiltered water, then en bloc stained for 2 h at room temperature. Samples were dehydrated in graded ethanol (50%, 75% and 95%) for 15 min each at 4 °C; the samples were then allowed to equilibrate to room temperature and were rinsed in 100% ethanol 2 times, followed by acetonitrile for 15 min. Samples were infiltrated with EMbed-812 resin (EMS 14120) mixed 1:1 with acetonitrile for 2 h followed by 2:1 EMbed-812:acetonitrile for 2 h. The samples were then placed into EMbed-812 for 2 h, then placed into TAAB capsules filled with fresh resin, which were then placed into a 65 °C oven overnight. Sections were taken between 40 and 60 nm on a Leica Ultracut S (Leica) and mounted on 100-mesh Ni grids (EMS FCF100-Ni). For immunohistochemistry, microetching was done with 10% periodic acid and eluting of osmium with 10% sodium metaperiodate for 15 min at room temperature on parafilm. Grids were rinsed with water three times in between and followed by 0.5 M Glycine quench. Grids were incubated in blocking solution (0.5% BSA, 0.5% Ovalbumin in PBST) at room temperature for 20 min. Primary rabbit anti-GFP (1:300; MBL International) was diluted in the same blocking solution and incubated overnight at 4 °C. The following day, grids were rinsed in PBS three times, and incubated in secondary antibody (1:10 10 nm Gold conjugated IgG TED Pella 15732) for 1 h at room temperature and rinsed with PBST followed by water. For each staining set, samples that did not contain any GFP-expressing cells were stained simultaneously to control for any non-specific binding. Grids were contrast stained for 30 s in 3.5% uranyl acetate in 50% acetone followed by staining in 0.2% lead citrate for 90 s. Samples were imaged in the tumour mass within the CA1 region of the hippocampus or within the contralateral normal hippocampus using a JEOL JEM-1400 transmission electron microscope at 120 kV and images were collected using a Gatan Orius digital camera.
Electron microscopy data analysis
Sections from the allografted hippocampi of mice were imaged as above using transmission electron microscopy. Here, 36 sections of 16T-mGFP across 6 mice were analysed. Electron microscopy images were taken at 6,000× with a field of view of 15.75 μm2. Synapses were inspected by two individual investigators. SCLC cells were counted and analysed after unequivocal identification of immunogold particle labelling with four or more particles. For identification of synaptic structures, all three of the following criteria had to be clearly met: (1) visually apparent synaptic cleft; (2) presence of synaptic vesicle clusters in a cell on one side of the cleft; and (3) identification of clear post-synaptic density on the cell on opposite side of cleft. For identification of neuron-to-tumour synapses, the post-synaptic cell had to exhibit clear immunogold particle labelling. Tumour cells in perisynaptic positions were identified when an immunogold-particle-positive SCLC cell was seen apposed to or surrounding the synaptic structures between two other immunogold-particle-negative cells. Density of both types of synaptic relationships (neuron-to-tumour synapses, neuron-to-tumour perisynaptic connections) were quantified as the number of connections per number of SCLC cells identified within the region.
Histology
Mice with intracranial tumour allografts were anaesthetized with intraperitoneal avertin (tribromoethanol), then transcardially perfused with 20 ml of ice-cold PBS. Brains were fixed in 4% PFA overnight at 4 °C, then cryoprotected in 30% sucrose, embedded in Tissue-Tek O.C.T. (Sakura) and sectioned in the coronal plane at 40 μm using a sliding microtome. Lung tumour mice were perfused as above but also with 10 ml ice-cold 4% PFA. The lungs were inflated with 1–3 ml 2% UltraPure Low Melting Point Agarose (Invitrogen). Lungs and livers were fixed overnight at 4 °C on a shaker, then transferred to 70% ethanol and sectioned at 15 μm for H&E and immunohistochemistry, or cryoprotected in 30% sucrose and sectioned at 150 μm on a cryotome for visualizing tumour innervation.
For immunofluorescence staining, the coronal sections were incubated in blocking solution (3% normal donkey serum, 0.3% Triton X-100 in TBS) at room temperature for 45 min, followed by an overnight incubation with primary antibodies in antibody diluent solution (1% normal donkey serum in 0.3% Triton X-100 in TBS) at 4 °C. On the next day, after a 5-min rinse in TBS, sections were incubated with DAPI (1 μg ml−1 in TBS, Thermo Fisher) for 5 min, and rinsed again with TBS for 5 min. Afterwards, slices were incubated in secondary antibody solution at 4 °C overnight, then washed thrice in TBS and mounted with ProLong Gold Mounting medium (Life Technologies). All images were acquired with Zen 3.4 and analysed using Fiji ImageJ 2.1.0. For quantifying SCLC proliferation, average number of Ki67+ SCLC cells divided by total number of SCLC cells labelled by GFP was calculated within regions of interest either in axon-rich/axon-poor areas of the tumour or in response to various optogenetic manipulations. For quantification of SCLC spread, average distance from the core was measured as longest distance from the initial site of injection to outer core of the tumour. For quantification of tumour invasion, average number of SCLC cells that migrated out of the circumscribed tumour edge over ~500 μm was calculated.
For visualizing lung and liver tumour innervation with immunofluorescence, 150-μm-thick sections were processed as above, except primary antibody incubation time was extended to 72 h. Streptavidin Alexa Fluor 594 conjugate (Invitrogen) was used to visualized airway epithelium79. To quantify lung and liver tumour burden, five equidistant H&E sections from each organ were evaluated by a pathologist blinded to the experimental conditions to estimate percent of section area occupied by the tumour. Any section with <5% of tissue estimated to be occupied by tumour was given a score of 0, 5–25% was given a score of 1, 25–50% was given a score of 2, 50–75% was given a score of 3, and >75% was given a score of 4. The score for each mouse was then generated as an average across the five sections.
The following primary antibodies were used: chicken anti-GFP (Aves Labs, 1:500), rabbit anti-MAP2 (EMD Millipore, 1:500), mouse anti-NeuN (EMD Millipore, 1:500), rabbit anti-Ki67 (Abcam, 1:500), guinea pig anti-synapsin (Synaptic Systems, 1:500), rabbit anti-HOMER1 (Synaptic Systems, 1:500), mouse anti-neurofilament (Abcam, 1:500), mouse anti-nestin (Abcam, 1:1,000), guinea pig anti-VChAT (Synaptic Systems, 1:200), mouse anti-TH (Abcam, 1:200), and rat anti-MBP (Abcam, 1:200). The following secondary antibodies were used (all Jackson Immuno Research, 1:500): DyLight 405 Donkey Anti-Mouse IgG, Alexa Fluor 488 Donkey Anti-Chicken IgG, Alexa Fluor 488 Donkey Anti-Guinea Pig IgG, Alexa Fluor 488 Donkey Anti-Rabbit IgG, Alexa Fluor 594 Donkey Anti-Rabbit IgG, Alexa Fluor 647 Donkey Anti-Mouse IgG, Alexa Fluor 647 Donkey Anti-Rabbit IgG, Alexa Fluor 647 Donkey Anti-Rat IgG.
Cell culture
The mouse 16T SCLC line was derived from a primary tumour from the lungs of an Rb/p53 mutant mouse44. 16T-GFP cells were generated by transducing 16T cells with pLV-CMV-GFP followed by FACS selection28. These cells are grown as neurospheres (unless otherwise stated) in 10% FBS medium consisting of DMEM (Invitrogen) and 1× liquid antibiotic-antimycotic (Invitrogen). The spheres were dissociated using TrypLE (Gibco) for seeding of in vitro experiments. For human cell lines, SCLC22H, H69, CORL47, and H526 were generously provided by M. Oser. SCLC22H was cultured in DMEM plus 10% FBS (Cytiva) and 1× Glutamax. H69, CORL47, H446 and H526 were cultured in RPMI with 10% FBS and 1× Glutamax. H1048 was cultured in RPMI plus 10% FBS, 1× Glutamax, and 1× Insulin transferrin selenium (Fisher Scientific). Primary small airway epithelial cells (HSAECs) were purchased from ATCC and grown in airway epithelial cell basal media supplemented with bronchial epithelial cell growth components (ATCC). All cultures were monitored by short tandem repeat fingerprinting for authenticity throughout the culture period and mycoplasma testing was routinely performed.
Co-culture of SCLC cells with primary mouse neurons
Neurons were isolated from the brains of CD1 mice using the Neural Tissue Dissociation Kit – Postnatal Neurons (Miltenyi), followed by the Neuron Isolation Kit, Mouse (Miltenyi) per the manufacturer’s instructions. After isolation, 300,000 neurons were plated onto circular glass coverslips (Electron Microscopy Services) pre-treated for 20 min at 37 °C with poly-l-lysine (Sigma) and then 3 h at 37 °C with 5 μg ml−1 mouse laminin (Thermo Fisher). Neurons were cultured in BrainPhys neuronal medium (Stemcell Technologies) supplemented with 1× Glutamax (Invitrogen), pen/strep (Invitrogen), B27 supplement (Invitrogen), BDNF (10 ng ml−1; Shenandoah), and GDNF (5 ng ml−1; Shenandoah), TRO19622 (5 μM; Tocris), β-mercaptoethanol (1×, Gibco) and 2% fetal bovine serum. Half of the medium was replenished on day in vitro (DIV) 1 and 5-fluoro-2′-deoxyuridine (UFDU) was added at 1 μM. This was repeated at DIV 3. On DIV 5, half of the medium was replaced with serum-free medium in the morning. In the afternoon, the medium was again replaced with half serum-free medium containing 75,000 SCLC cells dissociated from neurospheres or attached cultures with TrypLE. Tumour cells were cultured with neurons for 24 h and then fixed with 4% PFA for 20 min at room temperature and stained for immunofluorescence analysis as described below.
Co-culture of SCLC cells with human iPS cell-derived glutamatergic neurons
iPS cell lines were obtained from the Brigham and Women’s Hospital NeuroHub Core Facility and all permissions were received for use (from BWH NeuroHub Core and Rush Alzheimer’s Disease Center’s Biospecimen Distribution Committee). Induced neurons were generated from BR33 iPS cells as described80. In brief, iPS cells were plated in mTeSR1 medium at a density of 95,000 cells per cm2 on Matrigel-coated plates for viral transduction. Viral plasmids were obtained from Addgene (plasmids #19780, #52047 and #30130). FUdeltaGW-rtTA was a gift from K. Hochedlinger (Addgene plasmid #19780; http://n2t.net/addgene:19780; RRID: Addgene_19780). TetO-FUW-EGFP was a gift from M. Wernig (Addgene plasmid #30130; http://n2t.net/addgene:30130; RRID: Addgene_30130). pTet-O-Ngn2-puro was a gift from M. Wernig (Addgene plasmid #52047; http://n2t.net/addgene:52047; RRID: Addgene_52047). Lentiviruses were obtained from Alstem with ultrahigh titres (~1 × 109) and used at the following concentrations: pTet-ONGN2-puro: 0.13 µl, 50,000 cells; TetO-FUW-eGFP: 0.13 µl, 50,000 cells; FUdelta GW-rtTA: 0.13 µl, 50,000 cells. Transduced cells were dissociated with 3:1 DPBS: Accutase (Stemcell Technologies) + ROCKi (10 µM, Stemcell Technologies) and plated onto Matrigel-coated plates at 200,000 cells per cm2 in StemFlex medium + ROCKi (10 µM, Stemcell Technologies) (day 0). On day 1, medium was changed to KSR medium with doxycycline (2 µg ml−1, Sigma). Doxycycline was maintained in the medium for the remainder of the differentiation. On day 2, medium was changed to 1:1 KSR: N2B medium with puromycin (5 µg ml−1, GIBCO). Puromycin was maintained in the medium throughout the differentiation. On day 3, medium was changed to N2B medium + 1:100 B27 supplement (Life Technologies). From day 4 on, cells were cultured in NBM medium + 1:50 B27 + puromycin (5 µg ml−1) + BDNF, GDNF, CNTF (10 ng ml−1, Peprotech). Half medium changes were performed every 2–3 days. Around D10, puromycin was removed from medium. Around D12-D-15, SCLC cells were added at a ratio of 1:3 (cancer cells:neurons). Cultures were monitored over the next ten days in the case of MEA recordings. For histological analyses, SCLC cells were added in the presence of 1 μM TTX (Tocris), 50 μM MK801 (Selleck Chemicals), 50 μM CNQX (Tocris), or vehicle. Co-cultures were then fixed and analysed 24 h later in the case of proliferation assays, and 5 days after co-culture in the case of synapse quantification.
Co-culture of SCLC cells with human iPS cell-derived GABAergic neurons
Induced pluripotent stem cell lines were obtained from the Brigham and Women’s Hospital NeuroHub Core Facility and all permissions were received for use (from BWH NeuroHub Core and Rush Alzheimer’s Disease Center’s Biospecimen Distribution Committee). In brief, iPS cells were plated in mTeSR1 medium at a density of 95,000 cells per cm2 on Matrigel-coated plates for viral transduction. Viral plasmids were obtained from Addgene (plasmids #19780, #97329 and #97330). FUdeltaGW-rtTA was a gift from K. Hochedlinger (Addgene plasmid #19780). TetO-Ascl1-puro was a gift from M. Wernig (Addgene plasmid #97329; http://n2t.net/addgene:97329; RRID: Addgene_97329). DLX2-hygro was a gift from M. Wernig (Addgene plasmid #97330; http://n2t.net/addgene:97330; RRID: Addgene_97330). Lentiviruses were obtained from Alstem with ultrahigh titres (~1 × 109) and used at the following concentrations: TetO-Ascl1-T2A-Puro: 0.13 µl, 50,000 cells; DLX2-hygro: 0.13 µl, 50,000 cells: FUdelta GW-rtTA: 0.13 µl, 50,000 cells. Transduced cells were dissociated with 3:1 DPBS: Accutase + ROCKi (10 µM) and plated onto Matrigel-coated plates at 200,000 cells per cm2 in StemFlex medium + ROCKi (10 µM) (day 0). On day 1, medium was changed to N2B medium with doxycycline (2 µg ml−1) and forskolin (10 µM, Sigma). Doxycycline and forskolin were maintained in the medium for the remainder of the differentiation. On day 2, medium was replaced with fresh N2B medium. On day 3, medium was changed to N2B medium + puromycin (10 µg ml−1, GIBCO). On day 4 onwards, cells were cultured in N2B medium + puromycin (8 µg ml−1). From day 5 onwards, cells were cultured with NBM medium + 1:50 B27 + puromycin (5 µg ml−1) + BDNF, GDNF, CNTF (10 ng ml−1) + doxycycline (2 µg ml−1) + forskolin (10 µM) + AraC (2 µM, Sigma). Half medium changes were performed every 2–3 days. Around day 14, puromycin, doxycycline, AraC, and forskolin were removed from medium. Around day 20 to day 25, SCLC cells were added at a ratio of 1:3 (cancer cells:neurons). For histological analyses, SCLC cells were added in the presence of 1 μM TTX (Tocris), 20 μM gabazine (Tocris) or vehicle. Co-cultures were then fixed and analysed 24 h later in the case of proliferation assays, and 5 days after co-culture in the case of synapse quantification.
Induced neuron protocol medium
KSR medium: Knockout DMEM, 15% KOSR, 1× MEM-NEAA, 55 µM β-mercaptoethanol, 1× Glutamax (Life Technologies). N2B medium: DMEM/F12, 1× GlutaMAX (Life Technologies), 1× N2 supplement B (StemCell Technologies), 0.3% dextrose (d-(+)-glucose, Sigma). NBM medium: neurobasal medium, 0.5× MEM-NEAA, 1× GlutaMAX (Life Technologies), 0.3% dextrose (d-(+)-glucose, Sigma).
Conditioned media assays
For conditioned media assays, either glutamatergic or GABAergic iPS cell-derived neurons were cultured as above. On day 17–18, neurons were replenished with a full medium change with or without the presence of 1 μM TTX. Conditioned medium was collected 24 h later for immediate use. Mouse 16T cells were then plated (30,000 cells in a 48 well plate) with 500 μL neuronal conditioned medium (again in the absence or presence of TTX) with the addition of 10 μM EdU to each well. After 24 h, cells were fixed cells with 4% PFA and stained using Click-iT 594 EdU kit and protocol (Invitrogen). All conditioned media assays were performed alongside direct co-culture assays described above.
EdU incorporation assay
EdU staining was performed on glass coverslips in 24-well plates which were precoated with poly-l-lysine (Sigma) and 5 μg ml−1 mouse laminin (Thermo Fisher). Neurosphere cultures were dissociated with TrypLE and plated onto coated slides with 10 μM of EdU. After 24 h the cells were fixed with 4% PFA in PBS for 20 min and then stained using the Click-iT 594 EdU kit and protocol (Invitrogen) with or without additional antibody staining and mounted using Prolong Gold mounting medium (Life Technologies). Proliferation index was determined by quantifying the fraction of EdU-labelled cells/GFP-labelled cells using confocal microscopy at 40× magnification.
Synaptic puncta staining and visualization
For immunohistochemistry, fixed coverslips were incubated in blocking solution (3% normal donkey serum, 0.3% Triton X-100 in TBS) at room temperature for 1 h. Primary antibodies guinea pig anti-synapsin1/2 (1:500; Synaptic Systems), rabbit anti-HOMER1 (1:500; Synaptic Systems), rabbit anti-gephyrin (1:300, Cell Signaling Technologies), or mouse anti-neurofilament (1:500; Abcam) in 0.3% Triton X-100 in TBS and incubated overnight at 4 °C. Samples were then rinsed 3 times in TBS and incubated in secondary antibody solution (Alexa 488 donkey anti-guinea pig IgG; Alexa 594 donkey anti-rabbit IgG, and Alexa 647 donkey anti-mouse IgG, all at 1:500 (Jackson Immuno Research)) in antibody diluent solution at 4 °C overnight. Coverslips were rinsed three times in TBS and mounted with ProLong Gold Mounting medium (Life Technologies). Images were collected using a 63× oil-immersion objective on a Zeiss LSM800 confocal microscope and processed with Airyscan. Colocalization of puncta was quantified as described2.
MEA recordings
All MEA recordings were taken and analysed using the Axion Biosystems platform. Prior to culturing, 6-well Axion plates were coated with poly-l-lysine and laminin. Day 4 iNs (created as described80) were then thawed and plated in neurobasal medium, at 100,000 cells per well. For the initial week prior to co-culture, iNs were subjected to a half medium change every 3–4 days. Doxycycline and puromycin treatment were stopped after day 10 to allow for the addition of tumour cells. On day 14–15, SCLC cells were added at 30,000 cells per well MEA plates. MEA plates were recorded every day for 10 min for up to 2 weeks after co-culture. All spike numbers and amplitude were assessed using proprietary Axion Software.
Electrophysiology
For all electrophysiology experiments, 35,000 16T-GFP cells were allografted into the CA1 hippocampus of 4–6 week old NSG mice. Three weeks after allograft, brain slices were obtained using standard techniques. Mice were anaesthetized by isoflurane inhalation and perfused transcardially with ice-cold ACSF containing (in mM) 125 NaCl, 2.5 KCl, 25 NaHCO3, 2 CaCl2, 1 MgCl2, 1.25 NaH2PO4 and 25 glucose (295 mOsm kg−1). Brains were blocked and transferred into a slicing chamber containing ice-cold ACSF. Coronal slices of hippocampus were cut at 300-μm thickness with a Leica VT1000s vibratome in ice-cold ACSF, transferred for 10 min to a holding chamber containing choline-based solution consisting of (in mM) 110 choline chloride, 25 NaHCO3, 2.5 KCl, 7 MgCl2, 0.5 CaCl2, 1.25 NaH2PO4, 25 glucose, 11.6 ascorbic acid, and 3.1 pyruvic acid at 34 °C then transferred to a secondary holding chamber containing ACSF at 34 °C for 30 min and subsequently maintained at room temperature (20–22 °C) until use. All recordings were obtained within 5 h of slicing. Both choline solution and ACSF were constantly bubbled with 95% O2/5% CO2.
Individual brain slices were transferred into a recording chamber, mounted on an upright microscope (Olympus BX51WI or Scientifica SliceScope Pro 1000) and continuously superfused (2–3 ml min−1) with ACSF and bubbled with 95% O2/5% CO2 warmed to 32–34 °C by passing it through a feedback-controlled in-line heater (SH-27B; Warner Instruments). Cells were visualized through 40× or 60× water immersion objectives with either infrared differential interference contrast optics or epifluorescence to identify GFP+ cells. For whole-cell voltage clamp recording, patch pipettes (2–4 MΩ) pulled from borosilicate glass (Sutter Instruments) were filled with internal solution containing (in mM) 135 CsMeSO3, 10 HEPES, 1 EGTA, 3.3 QX-314 (Cl− salt), 4 Mg-ATP, 0.3 Na-GTP, 8 sodium phosphocreatine (pH 7.3 adjusted with CsOH; 295 mOsm·kg−1). sEPSCs and spontaneous inhibitory post-synaptic currents (sIPSCs) of GFP+ cells were recorded for 5 min at –70 or 0 mV holding potential. For membrane current variance, 10 consecutive sweeps, each 10 s long, recorded in voltage clamp at −70 mV holding potential were analysed by calculating the s.d. of the current after applying a 3-point median filter. To record evoked excitatory post-synaptic currents (eEPSCs) and evoked inhibitory post-synaptic currents (eIPSCs), the membrane voltages were clamped at –70 mV or 0 mV. For perforated-patch recording eIPSCs (gramicidin D, 40–60 μg ml−1) with access resistance 40–60 MΩ, variable holding membrane voltages were applied to generate an I–V curve and identify chloride’s equilibrium potential. Extracellular stimulation was performed with a stimulus isolation unit (MicroProbes, ISO-Flex), bipolar electrodes (75 μm apart, PlasticOne or MicroProbes) which were placed away 100–400 μm along the CA1–CA2 axis from the recorded cells. Stimuli were delivered at 10 s intervals with 0.1 ms duration and 100–200 μA amplitude. In some recordings, to confirm whether sEPSCs or eIPSCs were glutamatergic or GABAergic, 10 μM NBQX and CPP (Sigma) or 10 μM gabazine (Tocris) were added to bath ACSF. For perforated patched recordings, 10 μM NBQX and CPP were in bath continuously. Pyramidal cells in allografted and contralateral sites of hippocampus were recorded in whole-cell current clamp to study intrinsic properties. Patch pipettes were filled with internal solution containing (in mM) 135 potassium methanesulfonate, 10 HEPES, 1 EGTA, 4 Mg-ATP, 0.4 Na-GTP, 8 sodium phosphocreatine (pH 7.3 adjusted with KOH; 295 mOsm kg−1). From a resting potential of −70 mV, currents were injected for 1,000 ms at 50 pA steps from −100 to 1,000 pA. Intrinsic properties of hippocampus pyramidal cells were analysed with software in Matlab.
Extracellular local field potentials (fEPSPs) were recorded from hippocampal allograft and contralateral sites under current clamp (I = 0 mode). A bipolar electrode (75 μm spacing, MicroProbes) was placed in stratum radiatum near the allograft or in the correspondent contralateral site. Recording glass electrodes (2–3 MΩ) filled ACSF were place 100–200 μm perpendicularly away from the stimulating bipolar electrode at a depth of 50 100 μm. The stimulation currents were delivered at 20 s intervals with 0.1 ms duration and variable intensities from 20 μA to 140 μA using 20 μA steps and repeated 10 times. The average slopes of fEPSPs for each current intensity were measured.
The resting membrane potential of SCLC was assessed through cell-attached configuration (Rseal > 1 GΩ) using the procedure previously described in Antel et al.81. Glass pipettes were filled with a solution containing (in mM) 145 KCL, 2 MgCl2, 5 HEPES (pH was adjusted to 7.3 with KOH). A voltage protocol consisting of hold at 0 mV, stepping to 100 mV and ramping back to −120 mV over 137.5 ms was applied to GFP+ cells. Five to ten consecutive traces were recorded and averaged. All data were acquired with a MultiClamp 700B amplifier (Molecular Devices) and digitized at 3 kHz with a National Instruments data acquisition device (NI USB- 6343).
Calcium imaging
For calcium imaging, the genetically encoded calcium indicator GCaMP6s was lentivirally transduced into mouse SCLC 16T (pLV-ef1-GCaMP6s-P2A-nls-tdTomato). In this case, SCLC cells containing the GCaMP6s reporter can be identified using the tdTomato nuclear tag. These cells were isolated and grafted into the CA1 region of the hippocampus as described above. Two-photon calcium imaging experiments were performed using Prairie Ultima XY upright two-photon microscope for tissue slices equipped with an Olympus LUM Plan FI W/IR-2 40× water immersion objective. The temperature of the perfusion medium, ACSF as described above, was kept at 30 °C, and perfused through the system at rate of 2 ml min−1. Excitation light was provided at a wavelength of 920 nm through a tunable Ti:Sapphire laser (Spectra Physics Mai Tai DeepSee) to allow for excitation of both tdTomato and GCaMP6s. The actual laser power reaching the scanhead for each scope is dynamically controlled by Pockels cells via software interface. Pockels cell were set at 10 for all experiments, and photomultiplier tubes (PMTs) were set at 800 for each channel. For these settings, power at back aperture of the objective was approximately 30 mW at 920 nm. The wavelength ranges for the emission filters were PMT1: 607 nm centre wavelength with 45 nm bandpass (full-width at half-maximum) and PMT2: 525-nm centre wavelength with 70-nm bandpass (full-width at half-maximum). Recordings were made at 0.65 frames per second (~1.5 Hz) for about 30 min in the case of spontaneous activity and 10 min in the case of response to periodic electrical stimulation. Cells were identified via the expression for the nuclear tdTomato tag and were only imaged in the area of interest, specifically in the CA1 region of the hippocampus. Similar to the electrophysiology paradigm, for neuronal stimulation experiments, the electrode was placed in the hippocampus to stimulate the neuronal inputs originating in CA3. For electrical stimulation, approximately 20 µA over 200 µs was delivered to local axons using a stimulating bipolar microelectrode. For inhibitor experiments, TTX was directly diluted into the ACSF perfusion medium at 500 nM, oxygenated, and delivered to the slices through the perfusion system.
Calcium imaging analysis
Quantitative fluorescence intensity analysis was done on calcium transients that were reliably evoked by axonal stimulation. To determine the effect of TTX on the calcium transients in response to electrical stimulation of the CA3 Schaffer collaterals, the field of cells were stimulated three times in 1-min intervals to ensure synaptic connectivity. TTX (500 nM) was then perfused into the slices and the stim was repeated on the same field of cells to gauge direct effect of TTX on stimulation response. For analysis, regions of interest of each responding nucleus were set and ΔFmax/F0 (maximum difference in fluorescence intensity normalized to background fluorescence) measurements were determined before and after TTX treatment.
Visualization of human SCLC lung tumour gene expression
Expression data from 81 patients with SCLC primary tumours38 as FPKM values was accessed and plotted for genes of interest.
Unilateral cervical vagotomy
Adult RPR2-luc mice (4–7 months old, 2–5 months from virus administration) weighing more than 17 g were anaesthetized with 1–4% isoflurane through a nose cone in a supine position. Skin of the ventral surface of the neck was shaved and aseptically prepared according to IACUC guidelines. Under a dissection microscope, a 1 cm midline skin incision was made, and salivary glands revealed were separated with blunt dissection to expose the airways. The right vagal nerve was then carefully dissected from the carotid sheath and cut at the cervical level posterior to the pharyngeal branch. Sham mice underwent the same surgical procedure for blunt dissection of the vagus nerve but the latter was left intact. The mice were monitored until recovery from anaesthesia for changes in heart rate and respiration and after that monitored bi-weekly for changes in weight, eating, drinking, and general activity. RPM mice were handled the same way, except vagotomy was performed within 1 week of virus administration.
Bioluminescence imaging
For in vivo monitoring of tumour growth, bioluminescence imaging was performed using an IVIS imaging system (Xenogen). Mice were placed under 1–4% isofluorane anaesthesia and injected with luciferin substrate, then imaged in pronated position for intracranial allograft experiments or supine position for lung tumour mice. Baseline bioluminescence was used to randomize mice by a blinded investigator so that experimental groups contained an equivalent range of tumour sizes. For vagotomy experiments, each cohort of mice was imaged weekly for a total of 6 months.
Survival studies
For survival studies, morbidity criteria used were either reduction of weight by 15% of initial weight, sickness behaviour (such as dyspnea, abnormal gait or posturing, or ill-groomed fur), or severe neurological motor deficits consistent with brainstem dysfunction (that is, hemiplegia or an incessant stereotyped circling behaviour seen with ventral midbrain dysfunction). Kaplan–Meier survival analysis using log-rank testing was performed to determine statistical significance.
Mouse drug treatment studies
For all drug studies, NSG mice were xenografted as above with either 16T-GFP or NCI-H446-GFP cells and randomized to treatment group by a blinded investigator. One week post-engraftment, tumour-bearing mice were treated with systemic administration of levetiracetam (20 mg kg−1; Selleck Chemicals; formulated in a sterile saline solution) via intraperitoneal injection daily for 2 weeks. Controls were treated with an identical volume of vehicle.
Statistical analyses
Statistical tests were conducted using Prism v.9.3.1 (GraphPad) software unless otherwise indicated. Gaussian distribution was confirmed by the Shapiro–Wilk normality test. For parametric data, unpaired two-tailed Student’s t-test or one-way ANOVA with Tukey’s post hoc tests to examine pairwise differences were used as indicated. Paired two-tailed Student’s t-tests were used in the case of same-cell or same-animal experiments (as in electrophysiological recordings and vagotomy experiments). For non-parametric data, a two-sided unpaired Mann–Whitney test was used as indicated, or a one-tailed Wilcoxon matched pairs signed rank test was used for same-cell experiments. Two-tailed log-rank analyses were used to analyse statistical significance of Kaplan–Meier survival curves. In all box-and-whiskers plots, whiskers indicate minimum and maximum values, the box extends to the 25th and 75th percentile, and the centre line is plotted at the median. In all violin plots, lines are drawn at the median and quartiles.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Source link
