Spanning-tree Progression Analysis Of Density-normalized Events Research Articles

New Concepts of Flow Cytometry Analysis in Oncohematology: Application to Diagnosis and Follow up (Minimal Residual Disease) in AML, ALL and MDS

IntroductionThe development of mass cytometry (CyTOF), using up to 40 concomitant tags for a given sample, has led to the generation of large databases. In order to properly analyze these extensive files, new software solutions have been devised using either mathematical reduction of multidimensional spaces or clustering methods allowing for two-dimensional representations. Such tools as SPADE (SPanning-tree progression Analysis of Density-Normalized events), t-SNE (t-Stochastic Neighborhood Embedding) and FlowSOM (Flow Self Organizing Map) therefore were published with CyTOF data. They have however seldom been used to analyze classical flow cytometry (FCM) data, although current instruments generate 10-12 parameters analyzes that would be good candidates for such analyzes. Here we report on the new paradigm of using such tools to interpret harmonized classical FCM data in particular for minimal residual disease (MRD).Materials and MethodsMultiparametric FCM was performed using panels and recommendations proposed by the Harmonemia group (Lacombe F et al. Leukemia 2016), on a Navios® instrument (Beckman Coulter, Miami, FL) since 2013, yielding a cohort of diagnosis (D) and follow-up (FU) samples of 460 AML (160 D, 300 FU), 695 ALL (129 D, 566 FU), and 150 MDS (100 D, 50 FU). In parallel, 45 normal bone marrow (NBM) samples (respectively 22 for AML and MDS cases and 23 for ALL cases) were processed in the same conditions as pathological samples with the relevant immunophenotyping panels (three combinations for AML/MDS and 3 others for ALL). All files were first processed using specific R-software scripts and Bioconduct or CRAN packages to normalize fluorescence and light diffusion intensity including shift intensity, compensation and biexponential/logical adjustment. Harmonized flow cytometry files (HF) ([1305 pathological + 45 NBM files] x 3 panels corresponding to a total of 4050 HF files) were then generated and used for further analysis.NBM files were adjusted and merged in a single, events-adjusted, representative file of NBM (R-NBM). For each case, merges of this R-NBM, D and FU files were processed using 3 different analysis ways 1) principal component analysis (PCA), 2) t-SNE and 3) FlowSOM. New resulting parameters were generated for each combination of R-NBM, D and FU merged files, i.e. PCA, t-SNE and FlowSOM. These parameters could then be analyzed using the Kaluza® software (Beckman Coulter). Graphical representations of all analyses were then performed in an automated unsupervised manner with Kaluza®, allowing for a direct comparison of NBM and pathological samples for each case. Interestingly, HF files generated by the R software, being integrated in the system, could also be classically analyzed by Kaluza®, in a supervised manner, by drawing gates around PCA/t-SNE clusters or FlowSOM nodes of interest. Such gates allowed to further analyze data by backgating on classical biparametric representations of antibody combinations as well as on the classical CD45/SSC bone marrow cartography.ResultsBoth t-SNE and FlowSOM provide a clear identification of normal and abnormal subsets, while PCA is less discriminative. Comparative simultaneous representations of the three types of samples - (D), (FU) and (NBM)- allow for both a rapid visualization and accurate quantification of normal and abnormal cell subpopulations. The FlowSOM solution further allows to identify and visually characterize specific subsets both within the diagnostic clone and along normal hematopoietic differentiation. FU samples can then be easily analyzed for the persistence of abnormal immunophenotypic characteristics, both in AML and ALL. Even unsupervised results allow for a rapid appreciation and quantification of MRD. Besides, such comparisons, between NBM and D samples provide a new approach of the complexity of MDS. Moreover, both supervised and unsupervised results were comparable.ConclusionThe combination of automated sophisticated multidimensional representation of numerous subsets and their in-depth analysis by classical manual tools of FCM analysis is a truly revolutionary approach. The strength of FCM, which retains and analyses morphological characteristics of the cells, is thus combined to the precision of mathematical analyzes developed for CyTOF. These versatile new tools open the way for a new era of classical FCM analysis in hematological malignancies. [Display omitted] DisclosuresDupont:Beckman Coulter: Employment.

Neoantigens in MDS Are Associated with Two Novel CD4+ T Cell Subsets and Improved Overall Survival

BackgroundThe immune system has been shown to be important in the pathophysiology of MDS. Adaptive immunity relies on T cells ability to discriminate between self and non-self antigens in the context of a surface bound MHC protein structure. Genomic changes occurring in acquired somatic mutations can encode novel amino acid sequences, which if translated and presented by the MHC are termed a neoantigen.To establish the impact of neoantigens in MDS we have combined a predictive algorithm to screen cases for the presence of neoantigens together with mass cytometry by time of flight (CyTOF) to identify neoantigen-related immune signatures.ResultsMutation screening was performed on 204 MDS patients using our established gene panel targeting mutations commonly occurring in MDS (Mohemdali, Leukaemia 2015). Neoantigen-MHC affinity was predicted using NetMHCpan3.0 with neoantigen affinities ranked against 400,000 naturally occurring peptides. Those in the top 2% of peptides were classed as predicted neoantigens.In a multi-variate analysis of overall survival, higher numbers of mutations were associated with a poorer outcome (HR 2.71 95%CI 1.28-5.73, p <0.01). In a subset of 72 low risk MDS cases (by IPSS) the presence of neoantigen was protective (HR 0.21 95% CI 0.05-0.84, p <0.05) (fig 1). There was no survival benefit in the presence of neoantigens in high risk MDS.To look for an immune signature in MDS, we performed deep immunophenotyping on 44 MDS cases and 8 age-matched healthy donors (HDs). Peripheral blood mononuclear cells (PBMCs) were stained with 35 metal-tagged antibodies and analysed using CyTOF. Automated clustering methods (t-distributed stochastic neighbour embedding (t-SNE), spanning-tree progression analysis of density-normalized events (SPADE) and in-house clustering) were used to identify T cell subsets. As expected, T regulatory cells were increased in MDS cases when compared to HDs (7.2% vs 4.9% p <0.00001). However, within MDS there was an increase in the recently described Treg B (Kordasti, Blood 2016) in RAEB (4.9% vs 3.6%, p <0.05). These Tregs have been shown to be activated, more suppressive, quicker in responding to IL-2 by phosphorylating STAT5 and Treg specific demethylated region (TSDR) is unmethylated. Of note, there was no difference in abundance of Tregs or other T cell subsets based on presence or absence of mutation or neoantigen.To assess other features of immune suppression we analysed expression of immune checkpoints. While there was no difference in percentage of PD1 positive cells as a proportion of T cells between MDS and HD, neoantigen positive patients had higher PD1+ T cells compared to negative cases (0.77 vs 1.69% p <0.005). To explore the phenotype of PD1 positive PBMCs further we used unsupervised clustering and identified 23 phenotypically distinct clusters of PD1 high cells (fig 2). Neoantigen positive cases had significantly lower numbers of two distinct PD1hi CD4+ T cell subsets (p<0.05). Using Marker Enrichment Modelling (MEM) to better characterise these clusters, we have shown one cluster to be part of the Treg B subpopulation and the other within T effectors (fig 3) (Diggins, Nature Methods 2017). No significant differences in numbers of these cells was seen between other subtypes of MDS, including presence or absence of mutations, or between MDS and HD.DiscussionUsing multi-dimensional analysis we have shown that increased blast counts are associated with an increase in more suppressive and proliferating Tregs (Treg B). The findings of increased numbers of PD1hi T-cells and also two specific CD4+ subsets implies a more immunosenescent phenotype in neoantigen negative MDS. The lack of this association when comparing mutation positive and negative cases suggests that this immune signature is dependent on the patient-specific interaction of the mutation with HLA type rather than driven by the presence of a mutation per se.This work demonstrates clearly for the first time the protective role of neoantigens in low risk MDS. By combining knowledge of host immune factors and tumour characteristics we have identified two novel subsets of PD1 upregulated CD4+ T cells that may play an important role in tumour surveillance and identifies cells likely to be susceptible to PD1 blockade. Further work to understand the mechanistic effects of PD1 blockade on these subsets will be crucial to identifying patients most likely to respond to checkpoint inhibition. [Display omitted] DisclosuresKulasekararaj:Akari Therapeutics Plc: Consultancy, Honoraria; Alexion Pharmaceuticals: Consultancy, Honoraria; Ra Pharmaceuticals: Consultancy, Honoraria; Achillion pharmaceuticals: Consultancy, Honoraria. Irish:Cytobank Inc: Equity Ownership, Membership on an entity's Board of Directors or advisory committees; Janssen: Research Funding; Incyte: Research Funding. Kordasti:Celgene: Research Funding.

Local delivery of FTY720 induces neutrophil activation through chemokine signaling in an oronasal fistula model

PurposeCleft palate repair surgeries lack a regenerative reconstructive option and, in many cases, develop complications including oronasal fistula (ONF). Our group has developed a novel murine phenocopy of ONF to study the oral cavity wound healing program. Using this model, our team previously identified that delivery of FTY720 on a nanofiber scaffold had a unique immunomodulatory effect directing macrophages and monocytes into a pro-regenerative state during ONF healing. Here, the objective of this study was to determine the effects of local biomaterial-based FTY720 delivery in the ONF model on the early bulk gene expression and neutrophil phenotypic response within the regenerating tissue.MethodsUsing a mouse model of ONF formation, a palate defect was created and was treated with FTY720 nanofiber scaffolds or (blank) vehicle control nanofibers. At 1 and 3 days post-implantation, ONF oral mucosal tissue from the defect region was collected for RNA sequencing analysis or flow cytometry. For the RNA-seq expression profiling, intracellular pathways were assessed using the KEGG Pathway database and Gene Ontology (GO) Terms enrichment interactive graph. To assess the effects of FTY720 on different neutrophil subpopulations, flow cytometry data was analyzed using pseudotime analysis based on Spanning-tree Progression Analysis of Density-normalized Events (SPADE).ResultsRNA sequencing analysis of palate mucosa injured tissue identified 669 genes that were differentially expressed (DE) during the first 3 days of ONF wound healing after local delivery of FTY720, including multiple genes in the sphingolipid signaling pathway. Evaluation of the DE genes at the KEGG Pathway database also identified the inflammatory immune response pathways (chemokine signaling, cytokine-cytokine receptor interaction, and leukocyte transendothelial migration), and the Gene Ontology enrichment analysis identified neutrophil chemotaxis and migration terms. SPADE dendrograms of CD11b+Ly6G+ neutrophils at both day 1 and day 3 post-injury showed significantly distinct subpopulations of neutrophils in oral mucosal defect tissue from the FTY720 scaffold treatment group compared to the vehicle control group (blank). Increased expression of CD88 and Vav1, among other genes, were found and staining of the ONF area demonstrated increased VAV1 staining in FTY720‐treated healing oral mucosa.ConclusionTreatment of oral mucosal defects using FTY720 scaffolds is a promising new immunotherapy to improve healing outcomes and reducing ONF formation during cleft palate surgical repair. Local delivery of FTY720 nanofiber scaffolds during ONF healing significantly shifted early gene transcription associated with immune cell recruitment and modulation of the immune microenvironment results in distinct neutrophil subpopulations in the oral mucosal defect tissue that provides a critical shift toward pro-regenerative immune signaling.

Computational flow cytometric analysis to detect epidermal subpopulations in human skin

BackgroundThe detection and dissection of epidermal subgroups could lead to an improved understanding of skin homeostasis and wound healing. Flow cytometric analysis provides an effective method to detect the surface markers of epidermal cells while producing high-dimensional data files.MethodsA 9-color flow cytometric panel was optimized to reveal the heterogeneous subgroups in the epidermis of human skin. The subsets of epidermal cells were characterized using automated methods based on dimensional reduction approaches (viSNE) and clustering with Spanning-tree Progression Analysis of Density-normalized Events (SPADE).ResultsThe manual analysis revealed differences in epidermal distribution between body sites based on a series biaxial gating starting with the expression of CD49f and CD29. The computational analysis divided the whole epidermal cell population into 25 clusters according to the surface marker phenotype with SPADE. This automatic analysis delineated the differences between body sites. The consistency of the results was confirmed with PhenoGraph.ConclusionA multicolor flow cytometry panel with a streamlined computational analysis pipeline is a feasible approach to delineate the heterogeneity of the epidermis in human skin.

A New Mechanism for T-ALL Leukemogenesis: Early Expression of Alpha-Beta TCR Occurs in a Natural Subset of CD4-CD8- Double Negative (DN) Thymocytes Where MHC Engagement May Drive NOTCH Mutation and Tumor Initiation

T cell lymphoblastic leukemia (T-ALL) is an aggressive cancer arising from transformed thymocytes. Most human T-ALL involves hyperactive NOTCH signaling that is often caused by activating NOTCH mutations. However, the identification of specific molecular signals that might induce or select for mutation and transformation are incompletely understood. We report that an understudied low-frequency, natural thymocyte subset expresses αβ T cell antigen receptor (TCR) earlier than most cells in mice and humans; engagement of the early αβTCR by major histocompatibility complexes (MHC) can cause outgrowth of NOTCH1 mutant clones and T-ALL leukemogenesis in a mouse model of T-ALL. Assessment of 5 recent human T-ALL cases found one to present this unique CD4-CD8- double- negative (DN) stage as the earliest identifiable developmental stage. These studies present a model of T-ALL leukemogenesis that identifies (i) a natural cell stage of origin susceptible to transformation, (ii) a matching mouse model showing that a signaling receptor (αβTCR) and its ligand (MHC) drive leukemogeneis and outgrowth of tumors bearing activating NOTCH1 mutations, and (iii) a human case that presents with a tumor consistent with this model and mechanism. In past work, the pre-TCR has been shown to impact T-ALL development in mice (Campese et al, Blood 2006), but an oncogenic role for the mature αβTCR is less well characterized and somewhat surprising. This is because, although T-ALL tumor cells may express variable levels of surface αβTCR/CD3, the earliest cell stages that are thought to transform are also thought to precede stages with αβTCR expression. Most conventional αβ thymocytes rearrange TCRβ and TCRα loci in separate, ordered developmental stages. However, some thymocytes in the conventional pathway rearrange both at DN stage thus exhibiting 'precocious' αβTCR (PAT) expression. Importantly, these PAT cells are indeed part of the conventional αβ lineage, being a 'subset' only due to early αβTCR expression but without known distinction in ultimate immune function (Aifantis et al, JEM 2006). We found that ~0.01% of mouse and human thymocytes are such PAT cells at steady state. To interrogate the PAT thymocyte surface phenotype, we performed multi-parametric flow cytometry with Spanning-tree Progression Analysis of Density-normalized Events (SPADE). This revealed that PAT thymocytes constitute a DN subset that is not associated with other well-described unconventional DN thymocytes known to express αβTCR, consistent with as the expectation that PATs are part of the conventional developmental pathway. We observed that the OT1 TCR transgene is expressed in mice with parallel timing and level to the natural PAT subset, allowing use of this model to study antigen-dependent signaling and oncogenesis. In a cohort study, no T-ALL was observed in wild-type C57BL/6 or OT-1.β2M-/- mice (deficient in endogenous antigen presentation), but MHC-sufficient OT-1 mice developed PAT-stage-specific T-ALL with activating NOTCH1 mutations. Transplant experiments corroborated a requirement for antigen presentation and TCR signaling for tumor maintenance as transplanted tumors grew in MHC+ but not MHC-deficient mice. This predicted that PAT thymocytes might have an unusual ability to signal through αβTCR even without coreceptor expression. When cultured in the presence of either exogenously added β2M or antigen presenting cells, both untransformed and neoplastic PAT cells upregulated CD69 in response to the OT-1 antigenic peptide, OVA. Furthermore, ex vivo analysis of PAT cells from polyclonal C57BL/6 versus MHC-deficient mice showed intrinsic upregulation of TCR-signaling-dependent Nur77 in an MHC-dependent manner. These data revealed a unique ability of PAT cells to engage in co-receptor independent but antigen-dependent signaling. Microarray analysis showed that the gene expression profile of neoplastic PAT cells from OT-1 T-ALL most closely resembled that of conventional post β-selection DN thymocytes, in agreement with the natural PAT stage during normal T cell development. These data support a model in which transformation occurred in the naturally occurring αβ PAT thymocyte subset as cell-of-origin. Collectively, our data suggest that precocious αβTCR expression and coreceptor-independent antigen engagement can cause activating NOTCH mutation and T-ALL development. Disclosures No relevant conflicts of interest to declare.

Spanning tree progression analysis of density normalized events (SPADE) identification of novel myeloid derived suppressor cells (MDSC) subsets

Abstract MDSC’s are a heterogeneous population of myeloid cells at various differentiation stages. The human MDSC phenotypes, are controversial and subdivided into macrophage (M-), granulocytic (G-) and immature (i-) MDSC’s. Our studies used a single staining tube with antibodies to linage markers (CD3, CD19 and CD56), HLA-DR, CD11b, CD14, CD15, CD16, CD33, CD45, PD-L1 and LOX-1. Mobilized lymphoma patient apharesis (CA) products provided a product with a high frequency of MDSCs compared to normal donor blood samples (PB). SSCA and FSA identified lymphocytes, granulocytes and monocyte populations. A total population (CD45+), MDSCs (Lin-HLA-DR-CD11b+CD14+, Lin-HLA-DR-CD11b+CD14-CD33+ and Lin-HLA-DR−CD11b+ CD15+) and non MDSC myeloid populations (Lin-HLA-DR+CD14+, Lin-HLA-DR+CD33+ and Lin-HLA-DR+ CD15+Lin-HLA-DR-11b-) were exported from FlowJo for Spade analysis. This identified 11 nodes with 4 MDSC sub-populations. Impressively, cells from the non-MDSC populations, including monocytes and granulocytes did not occur within these nodes. SPADE identified an M-MDSC, G-MDSC and two iMDSC subsets with an increased frequency in the CA samples. Increased expression of LOX-1 and CD16 occurred in CA MDSCs vs. PB. Intracellular staining for expression of NOS-2 and Arg-1 were consistent with the M and G-MDSC subsets. Node 7 mainly expressed M-MDSC’s with a low frequency of CD15 dull G-MDSC’s. The cell frequency for G-MDSC’s in node 7 was 1,334 for CA and 0 for PB with M-MDSC’s 1,079 and 43 respectively. Node 8 had a frequency (3,319) of G-MDSC’s for CA and 2,286 for PB with M-MDSC’s 188 and 15 respectively. In summary SPADE analysis can be used to cleanly identify MDSCs from monocytes and granulocytes and provide insight into novel MDSC subsets.

Pre-existing malignancy results in increased prevalence of distinct populations of CD4+ T cells during sepsis.

The presence of pre-existing malignancy in murine hosts results in increased immune dysregulation and risk of mortality following a septic insult. Based on the known systemic immunologic changes that occur in cancer hosts, we hypothesized that the presence of pre-existing malignancy would result in phenotypic and functional changes in CD4+ T cell responses following sepsis. In order to conduct a non-biased, unsupervised analysis of phenotypic differences between CD4+ T cell compartments, cohorts of mice were injected with LLC1 tumor cells and tumors were allowed to grow for 3 weeks. These cancer hosts and age-matched non-cancer controls were then subjected to CLP. Splenocytes were harvested at 24h post CLP and flow cytometry and SPADE (Spanning-tree Progression Analysis of Density-normalized Events) were used to analyze populations of CD4+ cells most different between the two groups. Results indicated that relative to non-cancer controls, cancer mice contained more resting memory CD4+ T cells, more activated CD4+ effectors, and fewer naïve CD4+ T cells during sepsis, suggesting that the CD4+ T cell compartment in cancer septic hosts is one of increased activation and differentiation. Moreover, cancer septic animals exhibited expansion of two distinct subsets of CD4+ T cells relative to previously healthy septic controls. Specifically, we identified increases in both a PD-1hi population and a distinct 2B4hi BTLAhi LAG-3hi population in cancer septic animals. By combining phenotypic analysis of exhaustion markers with functional analysis of cytokine production, we found that PD-1+ CD4+ cells in cancer hosts failed to make any cytokines following CLP, while the 2B4+ PD-1lo cells in cancer mice secreted increased TNF during sepsis. In sum, the immunophenotypic landscape of cancer septic animals is characterized by both increased CD4+ T cell activation and exhaustion, findings that may underlie the observed increased mortality in mice with pre-existing malignancy following sepsis.

In depth comparative phenotyping of blood innate myeloid leukocytes from healthy humans and macaques using mass cytometry.

Comparative immune-profiling of innate responses in humans and non-human primates is important to understand the pathogenesis of infectious and chronic inflammatory diseases as well as for the preclinical development of vaccines and immune therapies. However, direct comparisons of the two species are rare and were never performed using mass cytometry. Here, whole-blood-derived leukocytes from healthy humans and cynomolgus macaques were analyzed with mass cytometry. Two similar panels of around 30 monoclonal antibodies targeting human markers associated with innate myeloid cells to stain fixed human and macaque leukocytes were constructed. To compare the circulating innate cells from the two primate species, an analysis pipeline combining a clustering analysis by the Spanning-tree Progression Analysis of Density-normalized Events (SPADE) algorithm with a two-step hierarchical clustering of cells nodes and markers was used. Identical SPADE settings were applied to both datasets, except for the 20 clustering markers which slightly differed. A correlation analysis designed to compare the phenotypes of human and macaque cell nodes and based on 16 markers, including 15 shared clustering markers and CD19 for humans or CD20 for macaques, revealed similarities and differences between staining patterns. This study unique by the number of individuals (26 humans and 5 macaques) and the use of mass cytometry certainly contributes to better assess the advantages and limits of the use of non-human primates in preclinical research. © 2017 International Society for Advancement of Cytometry.

Research Techniques Made Simple: Mass Cytometry Analysis Tools for Decrypting the Complexity of Biological Systems

Mass cytometry by time-of-flight experiments allow analysis of over 40 functional and phenotypic cellular markers simultaneously at the single-cell level. The data dimensionality escalation accentuates limitations, inherent to manual analysis, as being subjective, labor-intensive, slow, and often incapable of showing the detailed features of each unique cell within populations. The subsequent challenge of examining, visualizing, and presenting mass cytometry data has motivated continuous development of dimensionality reduction methods. As a result, an increasing recognition of the inherent diversity and complexity of cellular networks is emerging, with the discovery of unexpected cell subpopulations, hierarchies, and developmental pathways, such as those existing within the immune system. Here, we briefly review some frequently used and accessible mass cytometry data analysis tools, including principal component analysis (PCA); spanning-tree progression analysis of density-normalized events (SPADE); t-distributed stochastic neighbor embedding (t-SNE)-based visualization (viSNE); automatic classification of cellular expression by nonlinear stochastic embedding (ACCENSE); and cluster identification, characterization, and regression (CITRUS). Mass cytometry, used together with these innovative analytic tools, has the power to lead to key discoveries in investigative dermatology, including but not limited to identifying signaling phenotypes with predictive value for early diagnosis, prognosis, or relapse and a thorough characterization of intratumor heterogeneity and disease-resistant cell populations, that may ultimately unveil novel therapeutic approaches.

Mass Cytometry Reveals PD1 Upregulation Is an Early Step in MDS Disease Progression

BackgroundMDS is characterized by ineffective haematopoiesis and a propensity to leukaemia transformation, with increasing evidence linking immune exhaustion to disease progression. Immune checkpoints are known to be upregulated on T cells in cancer, and inhibitors of CTLA-4 and PD1/PLD1 axis have demonstrated efficacy with likely clinical benefit in MDS. We have previously shown profound changes in both the number and function of components of the adaptive immune system, particularly Tregs, in MDS (Kordasti, Blood 2007). In order to characterise the immune signature in a wider range of T cell subsets simultaneously, with particular emphasis on cells likely to be affected by checkpoint inhibitor therapy (CPI), we analyse CTLA-4 and PD1 expression in MDS by cytometry by time-of-flight (CyTOF). Additionally, we aim to explore whether these differences are accentuated by the absence or presence of somatic mutations, or in morphologically more advanced disease.Materials and MethodsMDS patients (n=56) and age-matched healthy donors (HD, n=6) were stained with two panels of 35 and 34 antibodies for unstimulated and PMA/Ionomycin-stimulated PBMCs, respectively. Samples were run on CyTOF and data analysed using visual stochastic neighbour embedding (viSNE, Cytobank) to generate t-distributed SNE scores by unsupervised multi-dimensional reduction of T cells. Spanning-tree progression analysis of density-normalized events (SPADE), was performed and T cell subsets identified from heat maps based on typical phenotypic markers (Regulatory, Naive, Memory, Effector Memory (EM) Central Memory (CM), Effector, Terminal Effector (TE)). T cells were also clustered based on cytokine secretion (IFN-ϒ, TNF-α, IL-17, IL-2 and IL-10).Somatic mutation analysis was performed on 48 of the MDS patients using our established targeted panel of 24 genes known to be mutated in MDS, for subgroup analysis (Mohamedali, Leukaemia 2015)Results and DiscussionDemographics and subgroups are outlined in figure 1.The number of Tregs was significantly higher in RAEB than non-RAEB MDS (9.6% of total CD4+ cells vs 7.5% p=0.02) and in RAEB versus HD (9.6% vs 5.8% p=0.01). There was a significantly higher proportion of Tregs in MDS patients with somatic mutations compared to those without (8.7% vs 7.06% p<0.05) and HD (8.7% vs 5.9% p<0.05), confirming our previous findings. Amongst two subpopulations of Tregs previously identified (Kordasti, Blood 2016) there was no difference between HD and MDS in terms of their number. However, Tregs A and B have an increased PD1 expression in MDS vs HD (p=0.03 &p=0.003). In addition, CD4+EM and CD4+Memory, CD4+/CD8+ Na•ve T cells and TNF-α secreting cells all expressed more PD1 when compared to HD (p<0.05), see figure 2. However, there were no differences in T cell PD1 expression when comparing low and high risk MDS patients, suggesting immune exhaustion to tumour antigens is a common founder event in MDS.By contrast, CTLA-4 expression was not significantly different in T cell subsets between MDS and HD. Sub-groups based on specific somatic mutations did reveal that CD4+ Memory, CD4+TE and CD8+CM subsets in SF3B1-mutated MDS cases had less CTLA-4 expression compared to MDS cases with other mutations (p<0.05). Furthermore, SF3B1-mutant cases also had less CTLA-4 expression when compared to MDS cases without any detected mutations in CD4+ Memory, CD4+EM and CD8+CM subsets (p<0.05).Taken together this work suggests a possible 3 hit-model for disease progression in MDS. The establishment of MDS is accompanied by increased expression of PD1 most likely in response to increased PDL1 on tumour cells, which argues for early introduction of PD1 inhibitor in the disease course. Increased CTLA-4 expression is likely a later step, dependent on the acquisition of specific mutations by the malignant clone, that further alters the relationship between tumour and immune surveillance. Finally, in disease progression through from non-RAEB to RAEB/AML there is an expansion of suppressive Tregs facilitating tumour outgrowth. Furthermore, pro-inflammatory IL17 secreting T helper cells (Th17) have been shown to be increased in low risk MDS and are known to be associated with autoimmune disease. The trend towards increased PD1 expression in Th17 cells in a group of low risk patients harbouring SF3B1 mutations (p=0.09) is suggestive that acquisition of increased PD1 expression is a key branch point in disease outcome. [Display omitted] [Display omitted] DisclosuresIrish:Incyte: Research Funding; Janssen: Research Funding; Cytobank, Inc.: Equity Ownership, Membership on an entity's Board of Directors or advisory committees.

Revealing the diversity of extracellular vesicles using high-dimensional flow cytometry analyses.

Extracellular vesicles (EV) are small membrane vesicles produced by cells upon activation and apoptosis. EVs are heterogeneous according to their origin, mode of release, membrane composition, organelle and biochemical content, and other factors. Whereas it is apparent that EVs are implicated in intercellular communication, they can also be used as biomarkers. Continuous improvements in pre-analytical parameters and flow cytometry permit more efficient assessment of EVs; however, methods to more objectively distinguish EVs from cells and background, and to interpret multiple single-EV parameters are lacking. We used spanning-tree progression analysis of density-normalized events (SPADE) as a computational approach for the organization of EV subpopulations released by platelets and erythrocytes. SPADE distinguished EVs, and logically organized EVs detected by high-sensitivity flow cytofluorometry based on size estimation, granularity, mitochondrial content, and phosphatidylserine and protein receptor surface expression. Plasma EVs were organized by hierarchy, permitting appreciation of their heterogeneity. Furthermore, SPADE was used to analyze EVs present in the synovial fluid of patients with inflammatory arthritis. Its algorithm efficiently revealed subtypes of arthritic patients based on EV heterogeneity patterns. Our study reveals that computational algorithms are useful for the analysis of high-dimensional single EV data, thereby facilitating comprehension of EV functions and biomarker development.

Microparticle and mitochondrial release during extended storage of different types of platelet concentrates

On activation, platelets release vesicles called microparticles (MPs). MPs are heterogeneous with regard to the presence or absence of mitochondria. We quantified MPs in platelet concentrates (PCs) taking their mitochondrial content into account. Platelet-rich plasma (PRP), buffy coat (BC) and apheresis (AP) PCs were tested through 7 days of storage. A combination of flow cytometry and spanning-tree progression analysis of density-normalized events (SPADE) was used to determine MP and mitochondrial release during storage. All the PC biochemical parameters complied with transfusion standards at all times. Platelet activation markers increased during storage and were higher for PRP than other types of PCs. Concentrations of MPs and extracellular mitochondria interpreted by SPADE algorithm were significantly higher in PRP than other in PCs and were stable throughout storage. The mode of preparation, rather than storage duration, impacts the release of MPs and mitochondria in PCs.

Visualization and cellular hierarchy inference of single-cell data using SPADE.

High-throughput single-cell technologies provide an unprecedented view into cellular heterogeneity, yet they pose new challenges in data analysis and interpretation. In this protocol, we describe the use of Spanning-tree Progression Analysis of Density-normalized Events (SPADE), a density-based algorithm for visualizing single-cell data and enabling cellular hierarchy inference among subpopulations of similar cells. It was initially developed for flow and mass cytometry single-cell data. We describe SPADE's implementation and application using an open-source R package that runs on Mac OS X, Linux and Windows systems. A typical SPADE analysis on a 2.27-GHz processor laptop takes ∼5 min. We demonstrate the applicability of SPADE to single-cell RNA-seq data. We compare SPADE with recently developed single-cell visualization approaches based on the t-distribution stochastic neighborhood embedding (t-SNE) algorithm. We contrast the implementation and outputs of these methods for normal and malignant hematopoietic cells analyzed by mass cytometry and provide recommendations for appropriate use. Finally, we provide an integrative strategy that combines the strengths of t-SNE and SPADE to infer cellular hierarchy from high-dimensional single-cell data.

Identification of Vaccine-Altered Circulating B Cell Phenotypes Using Mass Cytometry and a Two-Step Clustering Analysis.

Broadening our understanding of the abundance and phenotype of B cell subsets that are induced or perturbed by exogenous Ags will improve the vaccine evaluation process. Mass cytometry (CyTOF) is being used to increase the number of markers that can be investigated in single cells, and therefore characterize cell phenotype at an unprecedented level. We designed a panel of CyTOF Abs to compare the B cell response in cynomolgus macaques at baseline, and 8 and 28 d after the second homologous immunization with modified vaccinia virus Ankara. The spanning-tree progression analysis of density-normalized events (SPADE) algorithm was used to identify clusters of CD20(+) B cells. Our data revealed the phenotypic complexity and diversity of circulating B cells at steady-state and significant vaccine-induced changes in the proportions of some B cell clusters. All SPADE clusters, including those altered quantitatively by vaccination, were characterized phenotypically and compared using double hierarchical clustering. Vaccine-altered clusters composed of previously described subsets including CD27(hi)CD21(lo) activated memory and CD27(+)CD21(+) resting memory B cells, and subphenotypes with novel patterns of marker coexpression. The expansion, followed by the contraction, of a single memory B cell SPADE cluster was positively correlated with serum anti-vaccine Ab titers. Similar results were generated by a different algorithm, automatic classification of cellular expression by nonlinear stochastic embedding. In conclusion, we present an in-depth characterization of B cell subphenotypes and proportions, before and after vaccination, using a two-step clustering analysis of CyTOF data, which is suitable for longitudinal studies and B cell subsets and biomarkers discovery.

Multiparametric Analysis of Intra-Tumoral T-Cells in Hodgkin's Lymphoma Using Mass Cytometry (CyTOF)

Classic Hodgkin lymphoma (cHL) is characterized by a rich non-malignant immune infiltrate. T-cells are key components of the antitumoral immune response and studies characterizing subsets in cHL have yielded conflicting results. Initial studies suggested a predominance of TH2-polarized CD4+ T-cells, thought to allow tumor progression due to exhaustion and hypofunctionality. More recent data contest these findings, supporting the theory of tumor progression through evasion from a TH1-rich infiltrate that is potentially functional.The role of tumor evasion in cHL has been highlighted by compelling early clinical data with the use of PD-1 blockade in patients with advanced disease. A similar trial in patients with non-Hodgkin lymphoma (NHL) yielded far more modest results. Intrinsic differences in T-cell subpopulations in the tumor microenvironment may correlate to response to immune checkpoint inhibitor therapy.CyTOF or mass cytometry is a platform able to evaluate more than 45 simultaneous parameters on a single-cell level using nonradioactive nonbiological isotopes tagged to monoclonal antibodies. Measurements are made based on mass spectrometry, avoiding the hurdles of interference and spectral overlap experienced with fluorochromes. This constitutes an ideal tool for the study of the tumor microenvironment given its ability to assess a large number of parameters and resolve differences in a heterogeneous population.We hypothesize that the phenotype of intratumoral lymphocytes in cHL identifies T-cells that can effectively eradicate malignant cells. To test this hypothesis, we compared the phenotype of intratumoral T-cells in cHL to that of NHL and nodular lymphocyte-predominant Hodgkin Lymphoma (nlpHL). Tonsil and hyperplastic lymph node (LN) tissues were used as normal controls.Single-cell suspensions created from tumor specimens were stained with a metal-tagged antibody panel containing 31 surface markers and acquired on CyTOF. Multiparametric data analysis was performed on Cytobank using spanning-tree progression analysis of density-normalized events (SPADE) and t-Distributed Stochastic Neighbor Embedding (viSNE) algorithms. Inferential statistical analyses were performed with JMP®, Version 10.0.0 (SAS Institute Inc., Cary, NC, 1989-2007) using two-tailed tests and a 95% confidence interval. Cell subsets are expressed as percentages of parent population (CD45+CD3+CD19-).A total of 10 samples were studied (4 cHL, 1 nlpHL, 3 NHL, 1 tonsil, 1 LN). The total T-cell population ranged from 30.52 to 67.05% in cHL and 15.36 to 47% in NHL compared to 4.02% and 24.58% in tonsil and LN respectively. The CD4+ T-cell subset ranged from 58.05 to 35.3% in cHL, 50.03 to 82.61% in NHL and corresponded to 82.74% and 87.07% in tonsil and LN respectively. SPADE analysis identified two areas of asymmetric frequency of events amongst samples (figure 1 and 2). The CD4+ Tnaive subset (CD4+CD45RA+CCR7+) ranged from 7.8 to 31.2% of total T-cells in cHL compared to 10.7% in nlpHL, 0.17 to 3.02% in NHL and 6.2 to 6.7% in controls. The pooled mean frequency of CD4+ Tnaive subset was significantly higher in HL (cHL + nlpHL) compared to NHL (14.3% vs. 1.55%; p<0.05; figure 3A). The regulatory T-cell subset (Treg; CD25+CCR4+) ranged from 0.49 to 1.84% of total T-cells in HL compared to 9.3 to 21.04% in NHL, and 4.45 to 8.28% in controls. The pooled mean frequency of the Treg subset was significantly smaller in HL compared to NHL (1.28% vs. 16.23%; p<0.05; figure 3B).Our data supports the use of mass cytometry as a platform to study the tumor microenvironment in B-cell lymphomas. Multiparametric data analysis revealed significant differences in the intratumoral T-cell population between HL and NHL samples, namely in the CD4+ Tnaive and Treg subsets. Further validation in a larger sample is underway and will include panels to evaluate intracellular cytokine production and cell signaling pathways. Correlation between specific intratumoral T-cell phenotypic signatures and clinical outcomes may identify prognostic and predictive characteristics and provide insight to mechanisms of resistance to immunotherapy. [Display omitted] [Display omitted] [Display omitted] DisclosuresNo relevant conflicts of interest to declare.

Single Cell Mass Cytometry for Phenotypic Analysis of Diffuse Large B-Cell Lymphoma

Diffuse Large B-cell Lymphoma (DLBCL) is the most common histologic subtype of non-Hodgkin Lymphoma (NHL). Despite its improved outcome with R-CHOP chemotherapy, 40% of patients still suffer with relapsed or refractory disease. Further investigation is needed to understand the complexity of DLBCL. Time-of-flight mass cytometry (CyTOF) is a recently developed technology that combines traditional flow cytometry with mass spectrometry and supports analysis of up to 30-40 parameters simultaneously at the single cell level with minimal spectral overlap (Bendall et al, Science 2011). In this study, we sought to develop a CyTOF method for phenotypic analysis of DLBCL to obtain high resolution profiles of the malignant clone(s) represented in individual tumor samples and resolve any underlying population substructure that might be informative in understanding the clinical and biologic heterogeneity of this disease.We designed a two-tube assay for this study. Tube #1 contained 35 cell surface markers including CD45, CD19, CD20, CD22, CD79B, IgM, IgD, Ig Kappa, Ig Lambda, CD5, CD10, CD23, CD43, CD38, CD138, CD44, CD21, CD24, CD40, CD72, CD80, CD45RA, CD49D, CD49F, CD62L, CD25, CD27, CD30, CD127, CD184, CD194, CD200, CD34, HLA-DR, and CD3. Tube #2 contained 37 markers in total including 17 cell surface markers overlapping with Tube #1 plus an additional 20 intracellular markers (BCL2, BCL6, IRF4/MUM1, LMO2, MYC, MCL1, MEF2B, KAT3B/P300, CBP, FOXP1, RUNX1, Ikaros, EZH2, BMI1, NOTCH1, CARD11, IkBa, phospho-Rb, Ki67, and CyclinD2). Metal-conjugated antibodies were purchased from DVS Sciences or unconjugated antibodies labeled in-house using DVS MaxPar metal labeling kits. Data were acquired using a DVS CyTOF2 instrument. We examined both fresh and viably frozen single cell suspensions from diagnostic lymph node biopsy samples received for flow cytometric analysis at the BC Cancer Agency. We used SPADE (spanning-tree progression analysis of density-normalized events) for initial data analysis.We first performed preliminary validation studies including cross-comparison of mass cytometry (CyTOF2) vs. flow cytometry (BD Canto2) datasets and fresh vs. previously frozen cell suspension material. Although individual marker intensities using matched antibody clones were in general lower by CyTOF, eight-parameter surface staining results were qualitatively comparable between the two platforms. Also there were essentially no differences observed in CyTOF staining profiles between fresh and previously frozen samples. Preliminary clustering analysis of cell populations using SPADE revealed clear separation between normal and malignant B cell populations as well as apparent substructure to the malignant population in a subset of DLBCL samples. These findings suggest intratumoral heterogeneity can be resolved by high dimensional CyTOF analysis. Ongoing efforts will focus on determining if phenotypically defined subsets show enrichment for subclonal mutations. DisclosuresNo relevant conflicts of interest to declare.

Anatomical localization of progenitor cells in human breast tissue reveals enrichment of uncommitted cells within immature lobules.

IntroductionLineage tracing studies in mice have revealed the localization and existence of lineage-restricted mammary epithelial progenitor cells that functionally contribute to expansive growth during puberty and differentiation during pregnancy. However, extensive anatomical differences between mouse and human mammary tissues preclude the direct translation of rodent findings to the human breast. Therefore, here we characterize the mammary progenitor cell hierarchy and identify the anatomic location of progenitor cells within human breast tissues.MethodsMammary epithelial cells (MECs) were isolated from disease-free reduction mammoplasty tissues and assayed for stem/progenitor activity in vitro and in vivo. MECs were sorted and evaluated for growth on collagen and expression of lineages markers. Breast lobules were microdissected and individually characterized based on lineage markers and steroid receptor expression to identify the anatomic location of progenitor cells. Spanning-tree progression analysis of density-normalized events (SPADE) was used to identify the cellular hierarchy of MECs within lobules from high-dimensional cytometry data.ResultsIntegrating multiple assays for progenitor activity, we identified the presence of luminal alveolar and basal ductal progenitors. Further, we show that Type I lobules of the human breast were the least mature, demonstrating an unrestricted pattern of expression of luminal and basal lineage markers. Consistent with this, SPADE analysis revealed that immature lobules were enriched for basal progenitor cells, while mature lobules consisted of increased hierarchal complexity of cells within the luminal lineages.ConclusionsThese results reveal underlying differences in the human breast epithelial hierarchy and suggest that with increasing glandular maturity, the epithelial hierarchy also becomes more complex.Electronic supplementary materialThe online version of this article (doi:10.1186/s13058-014-0453-3) contains supplementary material, which is available to authorized users.

Single-Cell Mass Cytometry Reveals Phenotypic and Functional Heterogeneity In Acute Myeloid Leukemia At Diagnosis and In Remission

Acute myeloid leukemia (AML) is organized in a hierarchy with a rare population known as leukemia stem cells (LSC) capable of self-renewal and propagation of the disease. Characterization of the unique phenotypes and complex signaling pathways in LSCs that survive induction chemotherapy is essential for understanding of the mechanisms of chemoresistance and designing the strategies to eliminate residual leukemia clones.In this study, we compared signaling profiles of distinct phenotypic AML subsets in paired bone marrow (BM) samples collected at diagnosis and after achieving the complete remission (CR). Cell surface characteristics and signaling pathways activated within sub-populations of AML samples were defined using the novel technology of time-of-flight mass cytometry (CyTOF) that has the ability to perform up to 100 mutiparameter assays in single cells (Bendall et al, Science 2011). First, we validated CyTOF measurements by performing cross-comparisons of surface markers and intracellular proteins measured in AML cells with traditional multi-parametric flow cytometry (FCM). Frequencies of CD123+CD99+ population within CD34+CD38- cells were 73.7%±1.8% and 78.5%±3.7% by CyTOF and FCM. Patterns of specific activation of the intracellular proteins pSTAT5, pERK1/2 and pAKT by GM-CSF, PMA and SCF, and inhibition by selective kinase inhibitors showed excellent cross-platform consistency between CyTOF, FCM and immunoblotting. Next, mononuclear cells of 5 paired AML (at diagnosis and in CR) and of 3 normal BM (NBM) were stained with 11 cell surface markers (CD34, CD38, CD123, CD99, CD45, CD33, CD117, CD7, CD4, CD90 and CD133) and 8 intracellular markers (p-4EBP1, p-NF¦ÊB, p-STAT3, p-AKT, p-mTOR, p-ERK, p-S6 and p-STAT5). A SPADE (spanning-tree progression analysis of density-normalized events) (Qiu et al, Nat Biotechnol. 2011) tree plot was generated, representing clustered expression of the cell-surface antigens. Boundaries and annotations of the AML cells were manually defined to represent distinct cell subsets (Figure 1). We used the pooled data from NBM samples, which showed identical patterns, as a reference. SPADE analysis revealed several subsets unique to the diagnostic AML samples, which were eliminated by chemotherapy; and phenotypically distinct subsets in diagnostic samples that persisted in CR. Notably, a subset defined by the “traditional LSC” markers (CD45dimCD34+CD38lowCD90-CD33-CD117+; annotation #2) was readily identified in diagnostic samples and was significantly reduced by induction chemotherapy in 2 of the 5 AML samples. In one of these samples we identified a distinct subset co-expressing LSC markers CD45dimCD34+CD38lowCD33-CD117-CD99lowCD133low (annotation #3) that was present in both diagnostic (1.1%) and CR (1.7%) BM; this subset may have contributed to the MRD detected by standard leukemia-associated immunophenotypes. [Display omitted] We next investigated intracellular signaling pathways in antigen-defined AML subpopulations using CyTOF. Activation of p-AKT and pS6 showed similar pattern in subsets defined by annotations 1, 9 and 10 at diagnosis (Figure 2A), and was largely reduced in the CR BM. In turn, activation of p-4EBP1 and p-mTOR were observed in multiple subsets (#1-5 and 9-11) in all diagnostic AML samples, especially in a subset 1 characterized by the “Progenitor” phenotype, and remained heightened in the CR samples (Figure 2B). [Display omitted] In summary, using CyTOF and SPADE, we characterized phenotype-specific intracellular signaling pathways in AML samples at diagnosis and in CR. Persistent activation of p-mTOR and p-4EBP1 are identified in the subpopulations of AML progenitors in CR, and may present the potentially targetable pathways in AML. The study is ongoing with prospective CyTOF analysis of a larger set of paired AML samples at diagnosis, CR and relapse coupled with the molecular analysis of the distinct subpopulations. Disclosures:No relevant conflicts of interest to declare.

Inferring Phenotypic Properties from Single-Cell Characteristics

Flow cytometry provides multi-dimensional data at the single-cell level. Such data contain information about the cellular heterogeneity of bulk samples, making it possible to correlate single-cell features with phenotypic properties of bulk tissues. Predicting phenotypes from single-cell measurements is a difficult challenge that has not been extensively studied. The 6th Dialogue for Reverse Engineering Assessments and Methods (DREAM6) invited the research community to develop solutions to a computational challenge: classifying acute myeloid leukemia (AML) positive patients and healthy donors using flow cytometry data. DREAM6 provided flow cytometry data for 359 normal and AML samples, and the class labels for half of the samples. Researchers were asked to predict the class labels of the remaining half. This paper describes one solution that was constructed by combining three algorithms: spanning-tree progression analysis of density-normalized events (SPADE), earth mover’s distance, and a nearest-neighbor classifier called Relief. This solution was among the top-performing methods that achieved 100% prediction accuracy.

Mass Cytometry Organizes the Heterogeneity of Pediatric B Cell Acute Lymphoblastic Leukemia

Abstract 753Pediatric B Cell Acute Lymphoblastic Leukemia: Common and Unique Differentiation States Defined by Signaling Response States Background:When mutations or regulatory dysfunction drive inappropriate cell division or survival this sets the stage for cancer initiation or progression. At what point do cells in an early cancer veer from their normal cellular routines to no longer participate in the development of a normal cellular tissue architecture & differentiation heirarchy? Do they still attempt to play out their programming to some degree, or are they “free actors”? The fact that cancers of any given type appear to remain tied to phenotypic classification schemes is illustrated in traditional clinical stratification systems. Paradoxically, cancers are considered (by some) as “heterogeneous”, whereas as a class they appear to recapitulate patterns of clinical responses, gene expression, signaling states. To what extent are these repeated molecular patterns mirrored at the level of differentiation? We mapped, at the single cell level, patterns of expression of markers and phenotypic traits that might be similar or unique across patient subgroups of pediatric B-cell ALL samples using 35 parameter proteomic mass cytometry. Using the features assignable to each single cell, with a statistical reconstruction of most likely similarity of features in 35 dimensional space, we mapped a “common” differentiation tree previously unrecognized by conventional analyses, and demonstrate here how differences in certain markers distinguish, or create, common phenotypic co-expression features across these ALL patients, or suggest patient-specific derailing of differentiation associated with changes in signaling module expression states. Methods:Cryopreserved cells were obtained from 8 pediatric B cell ALL patients under informed consent. 19 surface markers were used to cluster data into metacells of localized similarity displayed in a tree of local relationships via hierarchical cell lineage clustering and spanning-tree progression analysis of density-normalized events (SPADE) (Figure 1) (Bendall et al, Science 6 May2011; 332: 687–696). Results:The blast cell subpopulations comprised the areas of greatest density on the spanning trees–consistent with the fact that these cells are the most prevalent. Blast populations displayed variable expression of B cell precursor surface markers such as CD10, CD19, CD34 and CD38—even though clinical phenotyping placed all patients in a similar class. SPADE analysis detailed unexpected sub-branches prominently observed in certain patients, while absent or weakly represented in others. This confirms common signaling and differentiation states can be observed across patients, but individual patients can manifest unique and prevalent outgrowths of these common malignant differentiation states.Three patients' disease appears to gain a transformative event during a common point at a “pre-B cell development state, leading to local expansion at this halted population of characteristic immunophenotype. Sample ALL04 appears to have an outgrowth of cells with the earliest pre-B cell progenitors, whereas ALL01 is consistent with a maturing pre B cell, gaining expression of CD45. Notably, ALL03, characterized by an MLL rearrangement, clusters with myeloid cells and is CD10 negative, suggesting transformation prior to lymphocyte commitment. The cellular responses to perturbation provides added structure to the cellular subsets; responses to stimulation of the preB cell receptor are present within the majority of blast populations for ALL01 but absent in most for ALL03 and ALL04. We will present this and other findings related to these patients, as well as application of this approach to other tumor types.Conclusion:This high-dimensional immunophenotypic analysis of single cells from primary patient samples reveals an unseen developmental structure within pediatric B precursor acute lymphoblastic leukemia. The developmental stage at which the transformative event occurs informs the characteristic response to perturbation and critically, to drug treatment. *KD and ES contributed equally to this work. [Display omitted] Disclosures:Fantl:Nodality, Inc.: Equity Ownership.