Identifying intracellular organelles within the 3D label-free tomograms of cells’ refractive indexes recorded in flow cytometry is challenging. Here we present a method for the 3D statistical segmentation of nuclei and vacuoles in flowing cells.

Background: Lung cancer (LC), including both non-small (NSCLC) and small (SCLC) subtypes, is currently treated with a combination of chemo- and immunotherapy. However, predictive biomarkers to identify high-risk patients are needed. Here, we explore the role of peripheral blood mononuclear cells (PBMCs) as a tool for novel biomarkers searching. Methods: We analyzed the expression of the cGAS-STING pathway, a key DNA sensor that activates during chemotherapy, in PBMCs from LC patients divided into best responders (BR), responders (R) and non-responders (NR). The PBMCs were whole exome sequenced (WES). Results: PBMCs from BR and R patients of LC cohorts showed the highest levels of STING (p < 0.0001) and CXCL10 (p < 0.0001). From WES, each subject had at least 1 germline/somatic alteration in a DDR gene and the presence of more DDR gene mutations correlated with clinical responses, suggesting novel biomarker implications. Thus, we tested the effect of the pharmacological DDR inhibitor (DDRi) in PBMCs and in three-dimensional spheroid co-culture of PBMCs and LC cell lines; we found that DDRi strongly increased cGAS-STING expression and tumor infiltration ability of immune cells in NR and R patients. Furthermore, we performed FACS analysis of PBMCs derived from LC patients from the BR, R and NR cohorts and we found that cytotoxic T cell subpopulations displayed the highest STING expression. Conclusions: cGAS-STING signaling activation in PBMCs may be a novel potential predictive biomarker for the response to immunotherapy and high levels are correlated with a better response to treatment along with an overall increased antitumor immune injury.

The actual gap of the label-free quantitative phase microscopy in respect to fluorescence microscopy, that allows the subcellular characterization by using exogenous markers, is the lack of intracellular specificity. Recently, computational methods based on artificial intelligence have been demonstrated, which allow a virtual staining of single cells in both 2D and 3D, but they require co-registration systems able to collect simultaneously both fluorescence and quantitative phase information. However, a real limitation exists, i.e. these approaches cannot be used in flow cytometry condition. In this paper, we discuss a new methodology for adding the intracellular specificity analysis to tomographic phase microscopy in flow cytometry. The proposed strategy is based on the statistical clustering of tomograms voxels, thus allowing the segmentation of cell’s organelles. Here we report the results of nuclear region identification for cancer cells.

In recent years, label-free microscopy has gained momentum over the well-established fluorescence microscopy, as it allows overcoming many important drawbacks related to the staining process. Among the label-free imaging techniques, Quantitative Phase Imaging (QPI) has emerged since biophysical properties of cells and tissues are measured. The latest development of QPI is Tomographic Phase Microscopy (TPM), which allows reconstructing the 3D volumetric distribution of the Refractive Indices (RIs) at the single-cell level by combining multiple phase-contrast maps recorded all around the sample. Very recently, the TPM paradigm has been even demonstrated working in Flow Cytometry (FC) modality, thus opening the route to the label-free, 3D, quantitative and high-throughput recording of living suspended cells. Nevertheless, the several advantages of QPI and TPM over fluorescence microscopy are counterbalanced by the lack of intracellular specificity due to the stain-free imaging modality. In fact, the inner cell contrast usually is not enough to properly recognize the several organelles, thus preventing intracellular studies. In QPI and in static TPM, virtual staining has been proposed as a solution, based on the training of deep learning strategies to numerically emulate the chemical staining process. However, the virtual staining approach cannot be replicated in the TPM-FC technique since a dataset of paired 3D RI and fluorescent tomograms of cells cannot be created. Here we show a computational method for the stain-free segmentation of the nucleus in 3D inside the TPM-FC tomograms of flowing cells based on an ad hoc clustering of the intracellular voxels according to their statistical similarities.

In vivo imaging of SK-N-BE(2) tumor growth in NB orthotopic mice.

In vivo imaging of IMR-32 tumor growth in NB orthotopic mice.

Supplementary Figures. Figure S1: Percentage of DHS regions shared by NB cell lines and other Encode cell lines; Figure S2: Descriptive statistics of Whole Genome Sequencing (WGS) of 14 normal-primary NB sample pairs; Figure S3: Somatic variants filtering summary; Figure S4: Results from mutational enrichment analysis; Figure S5: K-means clustering of NB samples; Figure S6: ADPRHL1 and MCFL1 genomic interactions with the mutated TFBS; Figure S7: TH deregulation by somatic regulatory variants.

Supplementary Tables. Table S1. Clinical features of In-house collected NB samples; Table S2. Results of mutational enrichment analysis; Table S3. Results of Cox regression analysis; Table S4. Variant clusterization and expression levels of candidate target genes.

Spheroid image analysis, 3D invasion image analysis and antibodies list.

Cell cycle and clonogenic profiles after treatments with ponatinib and axitinib.

Supplementary Table 1. Functional prioritization of SNPs in linkage disequilibrium (r2=0.6) with significant rs4673067 at SCG2 locus. Supplementary Table 2. SNPs in LD (r2>0.60) with the SNP rs118727. Supplementary Table 3. SNPs in LD (r2>0.60) with the SNP rs196830. Supplementary Table 4. SNPs in LD (r2>0.60) with the SNP rs169061. Supplementary Table 5. SNPs in LD (r2>0.60) with the SNP rs11994014. Supplementary Table 6. SNPs in LD (r2>0.60) with the SNP rs17830286. Supplementary Table 7. Functional prioritization of SNPs in 3'UTR region of NEFL and in linkage disequilibrium (r2=0.6) with significant typed SNPs rs118727, rs196830, rs17830286, rs11994014. Supplementary Table 8. Prediction of mircoRNAs whose binding is affected by the SNPs rs1059111, rs2979704, rs3761 by mrSNP web tool. Supplementary Table 9. Prediction of transcriptional factors whose binding is affected by the SNPs rs1059111, rs2979704, rs3761 by HaploReg V2. Supplementary Table 10. List of UTR motifs analyzed for the SNPs rs3761, rs2979704, and rs1059111. Supplementary Table 11. Imputation results of SNPs at SCG2 and NEFL loci (+/- 1Kb) by using 1000 Genomes data. Supplementary Table 12. Association of NEFL SNP genotypes with pathologic characteristics of neuroblastoma in European American cohort. Supplementary Table 13. Association of NEFL SNP genotypes with pathologic characteristics of neuroblastoma in Italian cohort. Supplementary Figure 1. Linkage disequilibrium (LD) plot of NEFL gene (chr8: positions 24,845,004 to 24,887,674) by Haploview 4.2 for HapMap CEU subjects. Supplementary Figure 2. REST gene silencing in neuroblastoma cells. The efficiency of gene silencing mediated by lentiviral delivery of hairpin RNA directed against REST (shREST) in SH-SY5Y and SK-N-BE2c cell lines was assessed by western blotting (a-b). The bar graphs show integral optical density (OD) value for each band, normalized respect to β-Actin expression. The results are shown as mean of three experiments and are represented as fold respect to shCTR cells which are infected by lentivirus-mediated delivery of non-silencing hairpin RNA. Induction of NEFL and REST gene expression levels after REST silencing in SH-SY5Y (c) and SK-N-BE2c (d). *P<0.05 Supplementary Figure 3. (a) Association between NEFL SNP genotype and gene expression in 16 neuroblastoma cell lines. The gene expression measure of the NB1 cell line was excluded from this analysis as affected the normal distribution of data (data not shown). The analysis was performed using the SNPs rs12545967 and rs2976427 in complete linkage disequilibrium (r2=1) with rs11994014 and rs1059111, respectively. Conventionally, we report the allele code of rs11994014 and rs1059111 in the plot. (b) Association between NEFL SNP genotype and gene expression in 198 LCLs. The analysis for rs1059111 (missing in the dataset) was performed using the SNP rs2979704 in complete linkage disequilibrium (r2=1). Supplementary Figure 4. NEFL gene silencing in neuroblastoma cells with rs1059111 TA genotype. Evaluation of cell growth (a) and invasion (b) after NEFL silencing in SH-SY5Y and SK-N-BE2c cell lines. The efficiency of gene silencing mediated by lentiviral delivery of hairpin RNA directed against NEFL (shNEFL) in SH-SY5Y and SK-N-BE2c cell lines was assessed by western blotting (c). The bar graphs show integral optical density (OD) value for each band, normalized respect to β-Actin expression. The same results were observed for mRNA measurements (data not shown). The results are shown as mean of three experiments and are represented as fold respect to shCTR cells which are infected by lentivirus-mediated delivery of non-silencing hairpin RNA. *P<0.05 Supplementary Figure 5. Association between NEFL SNP genotype and neuronal marker genes in neuroblastoma cell lines. Eight cell lines (IMR-32, SK-N-DZ, SK-N-AS, KP-N-SI9s with rs1059111 TT genotype and SIMA, SK-N-BE(2), SH-SY5Y, SK-N-FI with rs1059111 TA protective genotype) with the extremes of NEFL mRNA expression were selected based on the analysis shows in the Supplementary Figure 2a. The analysis was performed using the SNPs rs2976427 in complete linkage disequilibrium (r2=1) with and rs1059111. Conventionally, we report the allele code of rs1059111 in the plot.

The 10q24.33 locus is known to be associated with susceptibility to cutaneous malignant melanoma (CMM), but the mechanisms underlying this association have been not extensively investigated. We carried out an integrative genomic analysis of 10q24.33 using epigenomic annotations and in vitro reporter gene assays to identify regulatory variants. We found two putative functional single nucleotide polymorphisms (SNPs) in an enhancer and in the promoter of OBFC1, respectively, in neural crest and CMM cells, one, rs2995264, altering enhancer activity. The minor allele G of rs2995264 correlated with lower OBFC1 expression in 470 CMM tumors and was confirmed to increase the CMM risk in a cohort of 484 CMM cases and 1801 controls of Italian origin. Hi-C and chromosome conformation capture (3C) experiments showed the interaction between the enhancer-SNP region and the promoter of OBFC1 and an isogenic model characterized by CRISPR-Cas9 deletion of the enhancer-SNP region confirmed the potential regulatory effect of rs2995264 on OBFC1 transcription. Moreover, the presence of G-rs2995264 risk allele reduced the binding affinity of the transcription factor MEOX2. Biologic investigations showed significant cell viability upon depletion of OBFC1, specifically in CMM cells that were homozygous for the protective allele. Clinically, high levels of OBFC1 expression associated with histologically favorable CMM tumors. Finally, preliminary results suggested the potential effect of decreased OBFC1 expression on telomerase activity in tumorigenic conditions. Our results support the hypothesis that reduced expression of OBFC1 gene through functional heritable DNA variation can contribute to malignant transformation of normal melanocytes.

BACKGROUND:FGFR1 regulates cell-cell adhesion and extracellular matrix architecture and acts as oncogene in several cancers. Potential cancer driver mutations of FGFR1 occur in neuroblastoma (NB), a neural crest-derived pediatric tumor arising in sympathetic nervous system, but so far they have not been studied experimentally. We investigated the driver-oncogene role of FGFR1 and the implication of N546K mutation in therapy-resistance in NB cells.METHODS:Public datasets were used to predict the correlation of FGFR1 expression with NB clinical outcomes. Whole genome sequencing data of 19 paired diagnostic and relapse NB samples were used to find somatic mutations. In NB cell lines, silencing by short hairpin RNA and transient overexpression of FGFR1 were performed to evaluate the effect of the identified mutation by cell growth, invasion and cologenicity assays. HEK293, SHSY5Y and SKNBE2 were selected to investigate subcellular wild-type and mutated protein localization. FGFR1 inhibitor (AZD4547), alone or in combination with PI3K inhibitor (GDC0941), was used to rescue malignant phenotypes induced by overexpression of FGFR1 wild-type and mutated protein.RESULTS:High FGFR1 expression correlated with low relapse-free survival in two independent NB gene expression datasets. In addition, we found the somatic mutation N546K, the most recurrent point mutation of FGFR1 in all cancers and already reported in NB, in one out of 19 matched primary and recurrent tumors. Loss of FGFR1 function attenuated invasion and cologenicity in NB cells, whereas FGFR1 overexpression enhanced oncogenicity. The overexpression of FGFR1N546K protein showed a higher nuclear localization compared to wild-type protein and increased cellular invasion and cologenicity. Moreover, N546K mutation caused the failure in response to treatment with FGFR1 inhibitor by activation of ERK, STAT3 and AKT pathways. The combination of FGFR1 and PI3K pathway inhibitors was effective in reducing the invasive and colonigenic ability of cells overexpressing FGFR1 mutated protein.CONCLUSIONS:FGFR1 is an actionable driver oncogene in NB and a promising therapy may consist in targeting FGFR1 mutations in patients with therapy-resistant NB.

Background We recently conducted Cetuximab-AVElumab-Lung (CAVE-Lung), a proof-of-concept, translational and clinical trial, to evaluate the combination of two IgG1 monoclonal antibodies (mAb): avelumab, an anti-PD-L1 drug, and cetuximab, an anti-epidermal growth factor receptor (EGFR) drug, as second- or third-line treatment in non-small cell lung cancer (NSCLC) patients. We have reported clinically relevant anti-tumor activity in 6/16 patients. Clinical benefit was accompanied by Natural Killer (NK) cell-mediated antibody-dependent cell cytotoxicity (ADCC). Among the 6 responding patients, 3 had progressed after initial response to a previous treatment with single agent anti-PD-1, nivolumab or pembrolizumab. Methods We report long-term clinical follow-up and additional findings on the anti-tumor activity and on the immune effects of cetuximab plus avelumab treatment for these 3 patients. Results As of November 30, 2021, 2/3 patients were alive. One patient was still on treatment from 34 months, while the other two patients had progression free survival (PFS) of 15 and 19 months, respectively. Analysis of serially collected peripheral blood mononuclear cells (PBMC) revealed long-term activation of NK cell-mediated ADCC. Comprehensive genomic profile analysis found somatic mutations and germline rare variants in DNA damage response (DDR) genes. Furthermore, by transcriptomic analysis of The Cancer Genome Atlas (TCGA) dataset we found that DDR mutant NSCLC displayed high STING pathway gene expression. In NSCLC patient-derived three-dimensional in vitro spheroid cultures, cetuximab plus avelumab treatment induced additive cancer cell growth inhibition as compared to single agent treatment. This effect was partially blocked by treatment with an anti-CD16 mAb, suggesting a direct involvement of NK cell activation. Furthermore, cetuximab plus avelumab treatment induced 10-, 20-, and 20-fold increase, respectively, in the gene expression of CCL5 and CXCL10 , two STING downstream effector cytokines, and of interferon β , as compared to untreated control samples. Conclusions DDR mutations may contribute to DDR-induced STING pathway with sustained innate immunity activation following cetuximab plus avelumab combination in previously treated, PD-1 inhibitor responsive NSCLC patients.

Quantitative Phase Imaging (QPI) has gained popularity in bioimaging because it can avoid the need for cell staining, which in some cases is difficult or impossible. However, as a result, QPI does not provide labelling of various specific intracellular structures. Here we show a novel computational segmentation method based on statistical inference that makes it possible for QPI techniques to identify the cell nucleus. We demonstrate the approach with refractive index tomograms of stain-free cells reconstructed through the tomographic phase microscopy in flow cytometry mode. In particular, by means of numerical simulations and two cancer cell lines, we demonstrate that the nucleus can be accurately distinguished within the stain-free tomograms. We show that our experimental results are consistent with confocal fluorescence microscopy (FM) data and microfluidic cytofluorimeter outputs. This is a significant step towards extracting specific three-dimensional intracellular structures directly from the phase-contrast data in a typical flow cytometry configuration.

Liquid biopsy (LB) is a promising oncological tool that aims to the early diagnosis of tumors and to improve the therapeutical efficacy by following patient treatment. It is based on the detection and analysis of circulating tumor cells (CTCs) released into the bloodstream from primary or metastatic tumors. However, a reliable label-free LB tool has not yet been developed. Therefore, here we discuss the advantages derived from combining microfluidics and label-free quantitative imaging, to access the full three-dimensional (3D) information of a biological specimen by performing the tomographic reconstruction at single-cell level in high throughput modality. Experimental results on some CTCs are reported.

We investigate the ability of machine learning to provide an accurate classification of cancer cell in microfluidics when only raw digital holograms are used as input data. Comparison among different learning strategies is addressed.

The label-free single cell analysis by machine and Deep Learning, in combination with digital holography in transmission microscope configuration, is becoming a powerful framework exploited for phenotyping biological samples. Usually, quantitative phase images of cells are retrieved from the reconstructed complex diffraction patterns and used as inputs of a deep neural network. However, the phase retrieval process can be very time consuming and prone to errors. Here we address the classification of cells by using learning strategies with images coming directly from the raw recorded digital holograms, i.e. without any data processing or refocusing involved. Indeed, in the raw digital hologram the entire complex amplitude information of the sample is intrinsically embedded in the form of modulated fringes. We develop a training strategy, based on deep and feature based machine learning models, in order extract such information by skipping the classical reconstruction process for classifying different neuroblastoma cells. We provided an experimental validation by using the proposed strategy to classify two neuroblastoma cell lines.

In last years, introduction of immunotherapeutic agents, such as immune checkpoint inhibitors (ICIs) and chimeric antigen receptor T-cell (CAR-T), are changing the clinical scenario of anti-cancer treatment with great results in multiple cancer types [1].Especially in thoracic malignancies, including non-small cell lung cancer (NSCLC) and also small cell lung cancer (SCLC) and malignant pleural mesothelioma (MPM), the addition of ICI anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), programmed cell death protein 1 (PD-1), or programmed cell death ligand 1 (PD-L1), are currently used worldwide [2-4].However, despite the general improvement in patients' prognosis, only a subgroup of patients achieve a long-term clinical and survival benefit and there is a big gap of knowledge regarding biomarkers of response.Moreover, monitoring anti-cancer immune response in patients maybe not easy, due to the dynamic changes during times, and also preclinical studies on immunotherapy drugs need often specific models, such as syngeneic murine models [5] or ex vivo models [6] that preserve immune components, and a great researchers' expertise.

A lab-on-chip platform for blood cells analysis will be presented that combine phase-contrast label-free imaging, microfluidics and artificial intelligence. Such platform will have important impact in the aerospace field quantifying the effect on the blood cells of astronauts stresses. Furthermore, smart and innovative platform for blood analysis could be the key element for innovative biomedicine and telemedicine solutions in aerospace applications.