SWIFT clustering analysis of intracellular cytokine staining flow cytometry data of the HVTN 105 vaccine trial reveals high frequencies of HIV-specific CD4+ T cell responses and associations with humoral responses.

Introduction: The HVTN 105 vaccine clinical trial tested four combinations of two immunogens - the DNA vaccine DNA-HIV-PT123, and the protein vaccine AIDSVAX B/E. All combinations induced substantial antibody and CD4+ T cell responses in many participants. We have now re-examined the intracellular cytokine staining flow cytometry data using the high-resolution SWIFT clustering algorithm, which is very effective for enumerating rare populations such as antigen-responsive T cells, and also determined correlations between the antibody and T cell responses. Methods: Flow cytometry samples across all the analysis batches were registered using the swiftReg registration tool, which reduces batch variation without compromising biological variation. Registered data were clustered using the SWIFT algorithm, and cluster template competition was used to identify clusters of antigen-responsive T cells and to separate these from constitutive cytokine producing cell clusters. Results: Registration strongly reduced batch variation among batches analyzed across several months. This in-depth clustering analysis identified a greater proportion of responders than the original analysis. A subset of antigen-responsive clusters producing IL-21 was identified. The cytokine patterns in each vaccine group were related to the type of vaccine - protein antigens tended to induce more cells producing IL-2 but not IFN-γ, whereas DNA vaccines tended to induce more IL-2+ IFN-γ+ CD4 T cells. Several significant correlations were identified between specific antibody responses and antigen-responsive T cell clusters. The best correlations were not necessarily observed with the strongest antibody or T cell responses. Conclusion: In the complex HVTN105 dataset, alternative analysis methods increased sensitivity of the detection of antigen-specific T cells; increased the number of identified vaccine responders; identified a small IL-21-producing T cell population; and demonstrated significant correlations between specific T cell populations and serum antibody responses. Multiple analysis strategies may be valuable for extracting the most information from large, complex studies.

Development of a functional salivary gland tissue chip with potential for high-content drug screening.

Radiation therapy for head and neck cancers causes salivary gland dysfunction leading to permanent xerostomia. Limited progress in the discovery of new therapeutic strategies is attributed to the lack of in vitro models that mimic salivary gland function and allow high-throughput drug screening. We address this limitation by combining engineered extracellular matrices with microbubble (MB) array technology to develop functional tissue mimetics for mouse and human salivary glands. We demonstrate that mouse and human salivary tissues encapsulated within matrix metalloproteinase-degradable poly(ethylene glycol) hydrogels formed in MB arrays are viable, express key salivary gland markers, and exhibit polarized localization of functional proteins. The salivary gland mimetics (SGm) respond to calcium signaling agonists and secrete salivary proteins. SGm were then used to evaluate radiosensitivity and mitigation of radiation damage using a radioprotective compound. Altogether, SGm exhibit phenotypic and functional parameters of salivary glands, and provide an enabling technology for high-content/throughput drug testing.

SwiftReg cluster registration automatically reduces flow cytometry data variability including batch effects.

Biological differences of interest in large, high-dimensional flow cytometry datasets are often obscured by undesired variations caused by differences in cytometers, reagents, or operators. Each variation type requires a different correction strategy, and their unknown contributions to overall variability hinder automated correction. We now describe swiftReg, an automated method that reduces undesired sources of variability between samples and particularly between batches. A high-resolution cluster map representing the multidimensional data is generated using the SWIFT algorithm, and shifts in cluster positions between samples are measured. Subpopulations are aligned between samples by displacing cell parameter values according to registration vectors derived from independent or locally-averaged cluster shifts. Batch variation is addressed by registering batch control or consensus samples, and applying the resulting shifts to individual samples. swiftReg selectively reduces batch variation, enhancing detection of biological differences. swiftReg outputs registered datasets as standard .FCS files to facilitate further analysis by other tools.

Automated Analysis of Flow Cytometry Data to Reduce Inter-Lab Variation in the Detection of Major Histocompatibility Complex Multimer-Binding T Cells.

Manual analysis of flow cytometry data and subjective gate-border decisions taken by individuals continue to be a source of variation in the assessment of antigen-specific T cells when comparing data across laboratories, and also over time in individual labs. Therefore, strategies to provide automated analysis of major histocompatibility complex (MHC) multimer-binding T cells represent an attractive solution to decrease subjectivity and technical variation. The challenge of using an automated analysis approach is that MHC multimer-binding T cell populations are often rare and therefore difficult to detect. We used a highly heterogeneous dataset from a recent MHC multimer proficiency panel to assess if MHC multimer-binding CD8+ T cells could be analyzed with computational solutions currently available, and if such analyses would reduce the technical variation across different laboratories. We used three different methods, FLOw Clustering without K (FLOCK), Scalable Weighted Iterative Flow-clustering Technique (SWIFT), and ReFlow to analyze flow cytometry data files from 28 laboratories. Each laboratory screened for antigen-responsive T cell populations with frequency ranging from 0.01 to 1.5% of lymphocytes within samples from two donors. Experience from this analysis shows that all three programs can be used for the identification of high to intermediate frequency of MHC multimer-binding T cell populations, with results very similar to that of manual gating. For the less frequent populations (<0.1% of live, single lymphocytes), SWIFT outperformed the other tools. As used in this study, none of the algorithms offered a completely automated pipeline for identification of MHC multimer populations, as varying degrees of human interventions were needed to complete the analysis. In this study, we demonstrate the feasibility of using automated analysis pipelines for assessing and identifying even rare populations of antigen-responsive T cells and discuss the main properties, differences, and advantages of the different methods tested.

Competitive SWIFT cluster templates enhance detection of aging changes.

Clustering-based algorithms for automated analysis of flow cytometry datasets have achieved more efficient and objective analysis than manual processing. Clustering organizes flow cytometry data into subpopulations with substantially homogenous characteristics but does not directly address the important problem of identifying the salient differences in subpopulations between subjects and groups. Here, we address this problem by augmenting SWIFT--a mixture model based clustering algorithm reported previously. First, we show that SWIFT clustering using a "template" mixture model, in which all subpopulations are represented, identifies small differences in cell numbers per subpopulation between samples. Second, we demonstrate that resolution of inter-sample differences is increased by "competition" wherein a joint model is formed by combining the mixture model templates obtained from different groups. In the joint model, clusters from individual groups compete for the assignment of cells, sharpening differences between samples, particularly differences representing subpopulation shifts that are masked under clustering with a single template model. The benefit of competition was demonstrated first with a semisynthetic dataset obtained by deliberately shifting a known subpopulation within an actual flow cytometry sample. Single templates correctly identified changes in the number of cells in the subpopulation, but only the competition method detected small changes in median fluorescence. In further validation studies, competition identified a larger number of significantly altered subpopulations between young and elderly subjects. This enrichment was specific, because competition between templates from consensus male and female samples did not improve the detection of age-related differences. Several changes between the young and elderly identified by SWIFT template competition were consistent with known alterations in the elderly, and additional altered subpopulations were also identified. Alternative algorithms detected far fewer significantly altered clusters. Thus SWIFT template competition is a powerful approach to sharpen comparisons between selected groups in flow cytometry datasets.

Microbubble array diffusion assay for the detection of cell secreted factors.

The therapeutic potential of monoclonal antibodies (mAbs) makes them an ideal tool in both clinical and research applications due to their ability to recognize and bind specific epitopes with high affinity and selectivity. While mAbs offer significant therapeutic potential, their utility is overshadowed by the cost associated with their production, which often relies on the ability to identify minor antigen-specific cells out of a heterogeneous population. To address concerns with suboptimal methods for screening cells, we have developed a cell-sorting array composed of nanoliter spherical cell culture compartments termed microbubble (MB) wells. We demonstrate a proof-of-concept system for the detection of cell secreted factors from both immortalized cell lines and primary B cell samples. Exploiting the unique ability of the MB well architecture to accumulate cell secreted factors as well as affinity capture coatings, we demonstrate on-chip detection and recovery of antibody-secreting cells for sequencing of immunoglobin genes. Furthermore, rapid image capture and analysis capabilities were developed for the processing of large MB arrays, thus facilitating the ability to conduct high-throughput screening of heterogeneous cell samples faster and more efficiently than ever before. The proof-of-concept assays presented herein lay the groundwork for the progression of MB well arrays as an advanced on-chip cell sorting technology.

An animated landscape representation of CD4+ T-cell differentiation, variability, and plasticity: insights into the behavior of populations versus cells.

Recent advances in understanding CD4(+) T-cell differentiation suggest that previous models of a few distinct, stable effector phenotypes were too simplistic. Although several well-characterized phenotypes are still recognized, some states display plasticity, and intermediate phenotypes exist. As a framework for reexamining these concepts, we use Waddington's landscape paradigm, augmented with explicit consideration of stochastic variations. Our animation program "LAVA" visualizes T-cell differentiation as cells moving across a landscape of hills and valleys, leading to attractor basins representing stable or semistable differentiation states. The model illustrates several principles, including: (i) cell populations may behave more predictably than individual cells; (ii) analogous to reticulate evolution, differentiation may proceed through a network of interconnected states, rather than a single well-defined pathway; (iii) relatively minor changes in the barriers between attractor basins can change the stability or plasticity of a population; (iv) intrapopulation variability of gene expression may be an important regulator of differentiation, rather than inconsequential noise; (v) the behavior of some populations may be defined mainly by the behavior of outlier cells. While not a quantitative representation of actual differentiation, our model is intended to provoke discussion of T-cell differentiation pathways, particularly highlighting a probabilistic view of transitions between states.

SWIFT-scalable clustering for automated identification of rare cell populations in large, high-dimensional flow cytometry datasets, part 1: algorithm design.

We present a model-based clustering method, SWIFT (Scalable Weighted Iterative Flow-clustering Technique), for digesting high-dimensional large-sized datasets obtained via modern flow cytometry into more compact representations that are well-suited for further automated or manual analysis. Key attributes of the method include the following: (a) the analysis is conducted in the multidimensional space retaining the semantics of the data, (b) an iterative weighted sampling procedure is utilized to maintain modest computational complexity and to retain discrimination of extremely small subpopulations (hundreds of cells from datasets containing tens of millions), and (c) a splitting and merging procedure is incorporated in the algorithm to preserve distinguishability between biologically distinct populations, while still providing a significant compaction relative to the original data. This article presents a detailed algorithmic description of SWIFT, outlining the application-driven motivations for the different design choices, a discussion of computational complexity of the different steps, and results obtained with SWIFT for synthetic data and relatively simple experimental data that allow validation of the desirable attributes. A companion paper (Part 2) highlights the use of SWIFT, in combination with additional computational tools, for more challenging biological problems.

SWIFT-scalable clustering for automated identification of rare cell populations in large, high-dimensional flow cytometry datasets, part 2: biological evaluation.

A multistage clustering and data processing method, SWIFT (detailed in a companion manuscript), has been developed to detect rare subpopulations in large, high-dimensional flow cytometry datasets. An iterative sampling procedure initially fits the data to multidimensional Gaussian distributions, then splitting and merging stages use a criterion of unimodality to optimize the detection of rare subpopulations, to converge on a consistent cluster number, and to describe non-Gaussian distributions. Probabilistic assignment of cells to clusters, visualization, and manipulation of clusters by their cluster medians, facilitate application of expert knowledge using standard flow cytometry programs. The dual problems of rigorously comparing similar complex samples, and enumerating absent or very rare cell subpopulations in negative controls, were solved by assigning cells in multiple samples to a cluster template derived from a single or combined sample. Comparison of antigen-stimulated and control human peripheral blood cell samples demonstrated that SWIFT could identify biologically significant subpopulations, such as rare cytokine-producing influenza-specific T cells. A sensitivity of better than one part per million was attained in very large samples. Results were highly consistent on biological replicates, yet the analysis was sensitive enough to show that multiple samples from the same subject were more similar than samples from different subjects. A companion manuscript (Part 1) details the algorithmic development of SWIFT.

Modulation of single-cell IgG secretion frequency and rates in human memory B cells by CpG DNA, CD40L, IL-21, and cell division.

During the recall response by CD27(+) IgG class-switched human memory B cells, total IgG secreted is a function of the following: 1) the number of IgG-secreting cells (IgG-SC), and 2) the secretion rate of each cell. In this study, we report the quantitative ELISPOT method for simultaneous estimation of single-cell IgG secretion rates and secreting cell frequencies in human B cell populations. We found that CD27(+) IgM(-) memory B cells activated with CpG and cytokines had considerable heterogeneity in the IgG secretion rates, with two major secretion rate subpopulations. BCR cross-linking reduced the frequency of cells with high per-cell IgG secretion rates, with a parallel decrease in CD27(high) B cell blasts. Increased cell death may account for the BCR-stimulated reduction in high-rate IgG-SC CD27(high) B cell blasts. In contrast, the addition of IL-21 to CD40L plus IL-4-activated human memory B cells induced a high-rate IgG-SC population in B cells with otherwise low per-cell IgG secretion rates. The profiles of human B cell IgG secretion rates followed the same biphasic distribution and range irrespective of division class. This, along with the presence of non-IgG-producing, dividing B cells in CpG plus cytokine-activated B memory B cell populations, is suggestive of an on/off switch regulating IgG secretion. Finally, these data support a mixture model of IgG secretion in which IgG secreted over time is modulated by the frequency of IgG-SC and the distribution of their IgG secretion rates.

Automated analysis of two- and three-color fluorescent Elispot (Fluorospot) assays for cytokine secretion.

The Elispot effectively measures the frequencies of cells secreting particular molecules, especially low-frequency cells such as antigen-specific T cells. The Fluorospot assay adapted this analysis to two products per cell, and this has now been extended to three-color measurement of both mouse and human cytokine-secreting cells. Due to the increased data complexity, and particularly the need to define single-, double- and triple-producing cells, it is critical to objectively quantify spot number, size, intensity, and coincidence with other spots. An automated counting program, Exploraspot, was therefore developed to detect and quantify Fluorospots in automated fluorescence microscope images. Morphological parameters, including size, intensity, location, circularity and others are calculated for each spot, exported in FCS format, and further analyzed by gating and graphical display in popular flow cytometry analysis programs. The utility of Exploraspot is demonstrated by identification of single-, double- and triple-secreting T cells; tolerance of variable background fluorescence; and estimation of the numbers of genuine versus random multiple events.

Collagen distribution and expression of collagen-binding alpha1beta1 (VLA-1) and alpha2beta1 (VLA-2) integrins on CD4 and CD8 T cells during influenza infection.

Most viral infections occur in extralymphoid tissues, yet the mechanisms that regulate lymphocytes in these environments are poorly understood. One feature common to many extralymphoid environments is an abundance of extracellular matrix. We have studied the expression of two members of the beta(1) integrin family of collagen-binding receptors, alpha(1)beta(1) and alpha(2)beta(1) (CD49a, VLA-1 and CD49b, VLA-2, respectively), on CD4 and CD8 T cells during the response to influenza infection in the lung. Flow cytometry showed that whereas T cells infiltrating the lung and airways can express both CD49a and CD49b, CD49a expression was most strongly associated with the CD8+ subset. Conversely, though fewer CD4+ T cells expressed CD49a, most CD4+ cells in the lung tissue or airways expressed CD49b. This reciprocal pattern suggested that CD4 and CD8 T cells might localize differently within the lung tissue and this was supported by immunofluorescent analysis. CD8+ cells tended to localize in close proximity to the collagen IV-rich basement membranes of either the airways or blood vessels, whereas CD4+ cells tended to localize in the collagen I-rich interstitial spaces, with few in the airways. These observations suggest that CD4 T cell interaction with the tissue microenvironment is distinct from CD8 T cells and support the concept that CD4+ T cells in peripheral tissues are regulated differently than the CD8 subset.

T regulatory and primed uncommitted CD4 T cells express CD73, which suppresses effector CD4 T cells by converting 5'-adenosine monophosphate to adenosine.

CD73 (5'-ectonucleotidase) is expressed by two distinct mouse CD4 T cell populations: CD25+ (FoxP3+) T regulatory (Treg) cells that suppress T cell proliferation but do not secrete IL-2, and CD25- uncommitted primed precursor Th (Thpp) cells that secrete IL-2 but do not suppress in standard Treg suppressor assays. CD73 on both Treg and Thpp cells converted extracellular 5'-AMP to adenosine. Adenosine suppressed proliferation and cytokine secretion of Th1 and Th2 effector cells, even when target cells were activated by anti-CD3 and anti-CD28. This represents an additional suppressive mechanism of Treg cells and a previously unrecognized suppressive activity of Thpp cells. Infiltration of either Treg or Thpp cells at inflammatory sites could potentially convert 5'-AMP generated by neutrophils or dying cells into the anti-inflammatory mediator adenosine, thus dampening excessive immune reactions.