Integrate Imaging Flow Cytometry and Transcriptomic Profiling to Evaluate Altered Endocytic CD1d Trafficking

Summary

Imaging flow cytometry provides an ideal approach to detect the morphological and functional alteration of cells at individual and populational levels. Disrupted endocytic function for lipid antigen presentation in pollutant-exposed human dendritic cells was demonstrated with a combined transcriptomic profiling of gene expression and morphological demonstration of protein trafficking.

Abstract

Populational analyses of the morphological and functional alteration of endocytic proteins are challenging due to the demand of image capture at a single cell level and statistical image analysis at a populational level. To overcome this difficulty, we used imaging flow cytometry and transcriptomic profiling (RNA-seq) to determine altered subcellular localization of the cluster of differentiation 1d protein (CD1d) associated with impaired endocytic gene expression in human dendritic cells (DCs), which were exposed to the common lipophilic air pollutant benzo[a]pyrene. The colocalization of CD1d and endocytic marker Lamp1 proteins from thousands of cell images captured with imaging flow cytometry was analyzed using IDEAS and ImageJ-Fiji programs. Numerous cellular images with co-stained CD1d and Lamp1 proteins were visualized after gating on CD1d⁺Lamp1⁺ DCs using IDEAS. The enhanced CD1d and Lamp1 colocalization upon BaP exposure was further demonstrated using thresholded scatterplots, tested with Mander's coefficients for co-localized intensity, and plotted based on the percentage of co-localized areas using ImageJ-Fiji. Our data provide an advantageous instrumental and bioinformatic approach to measure protein colocalization at both single and populational cellular levels, supporting an impaired functional outcome of transcriptomic alteration in pollutant-exposed human DCs.

Introduction

Antigen presentation typically involves intracellular protein trafficking, which has been often investigated using morphological characterization and phenotypic profiling of antigen presenting cells^1,2,3. To integrate the advantages of imaging and phenotyping methods, we describe an imaging analysis platform at both single cell and population levels to demonstrate an altered protein colocalization in human dendritic cells (DCs). In peptide antigen presentation, major histocompatibility complex (MHC) class I molecules bind a short peptide (8-10 residues) in the endoplasmic reticulum to activate conventional CD8⁺ T cells, while MHC class II molecules bind a relatively longer peptide (~20 residues) in endocytic compartments to activate conventional CD4⁺ T cells^1,4. In contrast, lipid-specific T cells are activated by CD1 proteins with lipid antigens loaded mainly in endocytic compartments^5,6. Lipid antigen presentation requires the supply of lipid metabolites produced in lipid metabolism^7,8,9,10 and the loading of functional lipid metabolites to CD1 proteins in endocytic compartments^5,6. In this context, various cellular factors modulating lipid antigen presentation, especially in environmental exposure of lipophilic pollutants and immune disorders, are critical to be defined. In this study, we used transcriptomic analysis, image profiling, and cellular and populational imaging analysis of human monocyte-derived DCs to determine the endocytic protein trafficking contributing to lipid antigen presentation in pollutant exposure. Certainly, this combined platform can be applied to studying subcellular protein trafficking and colocalization in different biological processes.

Technically, subcellular protein localization was usually demonstrated using confocal microscopy and statistically analyzed in a limited number of detected cells^1,2,3,11. Moreover, flow cytometry has been broadly applied to gate cell populations with co-stained signals of multiple proteins at a cellular level¹²; however, this lacks a detailed visualization of subcellular protein colocalization. To achieve comprehensive and statistical analyses of percentage protein colocalization at both cellular and population levels, we incorporate imaging profiling and analysis approaches to determine the features of protein colocalization with biological relevance. Specifically, we use imaging flow cytometry to detect the colocalization of CD1d and lysosomal-associated membrane protein 1 (Lamp1) proteins in this study. Quantitative analysis of colocalized molecules was previously difficult to be performed at a populational scale. In this study, we adapted the ImageJ-Fiji program to examine the percentage of protein colocalization with a large number of co-stained cells at both populational and individual cellular levels. Specifically, we measured co-localized areas, intensity, and populational size to support the conclusion that the CD1d protein was largely retained in late endocytic compartments of human DCs in exposure to a lipophilic pollutant benzo[a]pyrene (BaP)¹³. This combined cellular and population imaging analysis provided highly reproducible, comprehensive, and statistically significant results of CD1d-Lamp1 colocalization relevant to inhibited lipid antigen presentation.

The transcriptome of BaP-exposed human DCs strongly supported the hypothesis that endocytic lipid metabolism and CD1d endocytic trafficking were impaired in BaP exposure. To test this hypothesis, we applied imaging flow cytometry to profile the images of DC population that were co-stained with multiple proteins including CD1d, endocytic markers, and DC markers. Finally, co-stained cells were statistically analyzed to demonstrate the percentage of intensity and areas of CD1d and Lamp1 colocalization.

Protocol

Human protocols in this study were approved by the Institutional Review Board of the University of Cincinnati and all methods were performed in accordance with the relevant guidelines and regulations. Blood samples from healthy donors were obtained from the Hoxworth Blood Center at the University of Cincinnati Medical Center.

1. Transcriptomic Profiling of Pollutant-exposed Human Monocyte-derived DCs

1. Total RNA extraction of BaP-exposed DCs
   1. Differentiate human DCs using cytokines GM-CSF and IL-4, and expose DCs to BaP for 4 days¹³.
   2. Sort BaP-exposed DCs using flow cytometry based on the surface expressed markers (Lineage^-HLA-DR⁺), the majority of which consists of conventional DCs (CD11c⁺CD1c⁺)¹³. Typically, sort 10,000 DCs into culture medium containing 10% FBS, centrifuge DCs at 600 x g and 4 °C for 6 min, and immediately remove the supernatant by aspiration using a 1 mL pipette tip.
   3. Once the supernatant is removed by aspiration, lyse the cell pellet by adding 0.4 mL of Lysis/Binding Buffer from the RNA isolation kit for storage in -80 °C before RNA extraction.
   4. Use the RNA isolation kit with the total RNA extraction protocol to extract the total RNA¹⁴. Measure the RNA integrity using a Bioanalyzer¹⁵.
2. Transcriptomic analysis of sorted DCs (Figure 1A)
   1. Prepare the library for RNA-seq using a RNA Library preparation kit. In short, fragment the isolated poly-A RNA (~200 bp), reverse transcribe the fragments to the 1st strand of the cDNA, and follow by the second strand cDNA synthesis labeled with dUTP.
   2. Ligate double strand cDNA to the adapter with a stem-loop structure after bead purification, end-repairing and dA-tailing. Excise the uracil residues in the 2nd strand of the cDNA and adapter loop with USER (Uracil-Specific Excision Reagent) enzyme to maintain strand specificity and open the adapter loop for PCR.
      1. Perform 13 cycles of PCR using universal and index-specific primers to enrich the indexed library. Clean up the library and run on a Bioanalyzer DNA High Sensitivity chip to check the library quality and size distribution.
   3. Use a library quantification kit and a real-time PCR system combined with library size information to calculate library concentration via a standard curve method. 
   4. Proportionally pool individually indexed compatible libraries and adjust the final total concentration to 15 pM. Perform library cluster generation and sequence the library at a setting of single read 1x 50 bp to generate ~25 million reads per sample.
3. Sequencing
   1. Load the sequencing and indexing reagents to the SBS and PE reagent racks, respectively. Place the reagents in a laboratory-grade water bath for 1 h until all the ice has melted and the reagents in each bottle/tube are mixed properly.
   2. Prepare the ICB mix by adding the thawed dye and -20 °C enzyme to the bottle and mix. Prepare a NaOH solution according to the sequencing instructions. Place all reagents at 4 °C until ready to use.
   3. During the 1 h waiting period, power on the sequencer. Wait for the DONOTEJECT drive to appear and connect the computer to a network drive.
   4. Launch the sequencer control software.
   5. Prepare 2 L of Maintenance Wash solution that contains 0.5% Tween 20 and 0.03% ProClin 300 in laboratory-grade water.
   6. In the SBS reagent rack, add ~100 mL of Maintenance Wash solution to each of the 8 bottles, and screw funnel caps to the bottles. In the PE reagent rack, add ~12 mL of Maintenance Wash solution to each of the ten 15 mL conical tubes, and discard the caps.
   7. Load the two racks with the solution filled bottles/tubes to the sequencer.
   8. From the sequencer control software, choose the Maintenance Wash; follow the instructions on screen to clean the sequencer fluid system until the process is completed.
   9. Start a New Run in the "SEQUENCE" tab from the software; direct the output data to a network drive. Choose parameters for single read 1x 50 bp with single index multiplexed libraries.
   10. Optionally, log into the BaseSpace Sequence Hub so that the sequencing status can be remotely monitored via a computer or smart phone.
   11. Upload a Sample Sheet for demultiplexing and provide reagent information according to the software requirement.
   12. Load SBS and PE reagents to the sequencer. Prime the system with a used flow cell (~15 min).
   13. Once the cluster generation is completed (~4.5 h), take the flow cell out, lightly spray the flow cell with water, and wipe it dry using lens paper. Lightly spray the flow cell with 95% ethanol and wipe it dry. Check against a light to make sure that the surface is clean without debris or salt residue.
   14. After the Prime step is completed, load the clustered flow cell and start the sequencing. The Sequence Analysis Viewer software will automatically be started.
   15. Monitor the sequencing data quality via SAV including cluster density, reads pass filter, cluster pass filter %, % ≥Q30, Legacy phase/prophase %, indexing QC, etc. This helps to understand the data quality and troubleshoot.
   16. Change the flow cell gasket and perform a Maintenance Wash after the sequencing is completed. The sequencer is ready for the next run.
4. Bioinformatic analysis
   1. Perform bioinformatics RNA-seq data analysis¹³.

2. Pathway Analysis of Transcriptomic Profiles (Figure 1B)

1. Use an edgeR Bioconductor to compare resultant gene expression intensity counts between BaP-exposed and non-exposed DCs from three donors. Then, identify differentially expressed genes between BaP-exposed and non-exposed DCs based on the absolute fold change (>2 folds) and the false discovery rate (FDR)-adjusted p-values (<0.05).
2. To predict the functional clusters of differentially expressed genes, use the ToppCluster software package to search the altered genes against several databases including KEGG and REACTOME and generate clustering data. ToppCluster uses the hypergeometric test to obtain functional enrichment via the gene list enrichment analysis¹⁶.
3. Further input the results from these function clusters to Cytoscape^17,18 Version 3.3.0, a broadly used open source software platform for visualizing complex networks. Thus, the genes involved in different clusters or pathways, including endocytic clusters and lipid metabolism, are shown in a network with color annotation of the averaged fold change of gene expression (Figure 1B)¹³.

3. Imaging Flow Cytometry of CD1d and Lamp1 Colocalization

1. Antibody labeling of BaP-exposed DCs
   1. Expose 0.5 x 10⁶ human DCs to 5.94 μM BaP for 4 days at 37 °C in 2 mL of complete Dulbecco’s Modified Eagle’s Medium. Harvest DCs by centrifugation at 400 x g for 10 min. Block the differentiated DCs by incubating cells with human serum blocker and anti-human Fc receptor antibodies for 10 min in ice, including anti-human CD16, CD32, and CD64 antibodies.
   2. Incubate Brilliant Violet 421-anti-CD1c (L161), phycoerythrin/cyanine dye 7 (PE/Cy7)-anti-HLA-DR, and PE/Cy5-anti-CD11c with DCs for 20 min in ice. Table 1 shows this list of antibodies.
   3. Fix DCs with 4% paraformaldehyde in PBS and permeabilize them with Permeabilization Wash Buffer. Perform the intracellular staining with a mixture of PE-labeled purified anti-human CD1d (51.1) and Alexa Fluor 647-labeled anti-Lamp1 (CD107a) (H4A3).
2. Imaging flow cytometry measurement (Figure 2)
   1. Analyze the stained samples using an imaging flow cytometer at the flow cytometry core of Cincinnati Children’s Hospital using a 40X objective to yield 300 pixels for a cell with around 10 μm in diameter. One-pixel size is approximately 0.5 μm by 0.5 μm of the object.
   2. Acquire 10,000 cellular images in an unbiased manner according to the manufacturer’s instruction.

4. Imaging Analyses of Flow Cytometry Images

1. Colocalization analysis using IDEAS software (Figure 2)
   1. Perform compensation with single-stained versus non-stained samples by setting the fluorescence intensity of non-stained samples below the threshold, including non-stained BaP-exposed DCs with BaP autofluorescence, as in routine flow cytometry data analysis.
   2. Gate stained cells to obtain the subsets of HLA-DR⁺CD11c⁺CD1d⁺Lamp1⁺ cells (Figure 2A) and show cellular images in the gated subsets (Figure 2B).
   3. Save cell images in a png format based on two technical inclusion criteria, the visual presence of strong co-stained signals and subcellular localization of CD1d and Lamp1 proteins, for the colocalization analyses using ImageJ-Fiji (Figure 2B).
2. Use software ImageJ-Fiji to perform colocalization analysis (Figure 3).
   1. Prepare the input image files by merging the 100 saved cell images for non-exposed and BaP-exposed human DCs, respectively (Figure 3A).
   2. Analyze CD1d (Red) and Lamp1 (Green) colocalization using a scatterplot.
      1. Run ImageJ-Fiji program. Open the .png file with 100 cell images (Figure 3B and Figure 4A): "File | Open".
      2. Split the image with merged red and green channels into two images with either a red or a green channel: "Image | Color | Split channels".
      3. Draw a scatterplot using the commands "Analyze | Colocalization | Colocalization Threshold". Save the scatterplot using the "PrintScreen" key.
   3. Calculate Mander's colocalization coefficients for each single cellular image (n=100) (Figure 4B)
      1. Select a single cell image on the image file with split channels using the "Oval" selection tool.
      2. Use the commands "Analyze | Colocalization | Colocalization Threshold". Select "Channel 1" from the dialogue box of region of interest (ROI). Keep all calculation options including "Mander’s using thresholds" for each cell image.
      3. Repeat this calculation for all 100 cell images.
      4. Save and open the results using a spreadsheet.
      5. Plot the average and standard errors for “thresholded Mander’s coefficients” (n=100, 0 means no colocalization and 1 means perfect colocalization). Use Student’s t-test to calculate the p value for the comparison between BaP-exposed and non-exposed groups (Figure 4B).
   4. Calculate the percent of thresholded pixel intensity co-localized between CD1d and Lamp1 for multiple cell images (n=100) (Figure 4C).
      1. Use the same analysis protocol in 4.2.3 and additionally select the result option "% intensity above threshold colocalized".
      2. Perform this analysis together with Mander’s colocalization coefficients.
      3. Similarly plot the average and standard errors for % intensity colocalized between CD1d and Lamp1. Use Student’s t-test to calculate the p value for comparison between BaP-exposed and non-exposed groups (Figure 4C).

Results

The lipophilic pollutant BaP alters endocytic gene clusters in human DCs. Human monocyte-derived DCs from each donor (n=3) were incubated with BaP and sorted for conventional DCs, which were further used for RNA extraction and transcriptomic analysis as described. Upon the normalization of gene expression, altered genes between BaP-exposed and non-exposed groups were clustered according to the functional correlation of differentially expressed genes. We input the altered gene list to the ...

Discussion

Functional confirmation of an altered gene pathway is challenging, because of widely impacted gene expression involving multiple pathways and the difficulty to integrate individual and populational cellular activities. We employed imaging flow cytometry to specifically test CD1d trafficking in endocytic compartments. Imaging flow cytometry integrates the populational measurement of cells and the individual demonstration of subcellular colocalization of multiple proteins. Confocal microscopy has previously provided h...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Transcriptomics	Illumina	HiSeq 2500 v4	Illumina HiSeq system
ImagingStream X	Millipore	100220	Imaging flow cytometry
mirVana miRNA Isolation Kit	Thermo Fisher		Total RNA extraction
Agilent RNA 6000 Pico Kit	Agilent		Total RNA QC analysis
Veriti 96-Well Fast Thermal Cycler	Thermo Fisher		PCR, enzyme reaction
NEBNext Poly(A) mRNA Magnetic Isolation Module	New England Biolab		PolyA RNA purification
Automated SMARTer Apollo system	Takara		PolyA RNA purification
NEBNext Ultra RNA Library Prep Kit for Illumina	New England Biolab		Library preparation
Agencourt AMPure XP magnetic beads	Beckman Coulter		DNA purification
2100 Bioanalyzer	Agilent		Library QC, size distribution analysis
Agilent High Sensitivity DNA Kit	Agilent		Library QC, size distribution analysis
QuantStudio 5 Real-Time PCR System (Thermo Fisher)	Thermo Fisher		Library quantification
NEBNext Library Quant Kit	New England Biolab		Library quantification
cBot	Illumina		Library cluster generation
TruSeq SR Cluster Kit v3 - cBot – HS	Illumina		Library cluster generation
HiSeq 1000	Illumina		Sequencing
TruSeq SBS Kit v3 - HS (50-cycles)	Illumina		Sequencing
Phycoerythrin/Cyanine 7 (PE/Cy7)	Bio Legend	L243
Phycoerythrin/Cyanine 5 (PE/Cy5)	Bio Legend	3.9
Brilliant violet 421	Bio Legend	L161
PE-anti-mouse IgG2b	Bio Legend	51.1
Alexa Fluor 647 (AF647)	Bio Legend	CD107a(H4A3)