High Dimensional Functionomic Analysis of Human Hematopoietic Stem and Progenitor Cells at a Single Cell Level

引用 收藏 提问与回复 分享您的反馈 Cited by



Nature Immunology
Jul 2017



The ability to conduct investigation of cellular transcription, signaling, and function at the single-cell level has opened opportunities to examine heterogeneous populations at unprecedented resolutions. Although methods have been developed to evaluate high-dimensional transcriptomic and proteomic data (relating to cellular mRNA and protein), there has not been a method to evaluate corresponding high-dimensional functionomic data (relating to cellular functions) from single cells. Here, we present a protocol to quantitatively measure the differentiation potentials of single human hematopoietic stem and progenitor cells, and then cluster the cells according to these measurements. High dimensional functionomic analysis of cell potential allows cell function to be linked to molecular mechanisms within the same progenitor population.

Keywords: CD34+ hematopoietic stem and progenitor cells (CD34+造血干细胞和祖细胞), Single cell culture (单细胞培养), Differentiation (分化), Quantitative clonal output (定量克隆输出), Barnes-Hut t-SNE (Barnes-Hut t-SNE), Dimension reduction (降维)


The development of techniques for single-cell measurements of cell transcription, signaling, and function at the single-cell level, alongside preexisting technologies such as flow cytometry, has allowed new lenses to examine complex, heterogeneous populations. Such methods generate large amounts of data, which can be interpreted with aid from dimensionality reduction algorithms, as illustrated on single-cell RNA-Seq using Mpath, Monocole, PCA, Wishbone, or diffusion map algorithms (Paul et al., 2016; See et al., 2017), and on CyTOF using tSNE or PhenoGraph (Amir el et al., 2013; Levine et al., 2015).

We developed this protocol to allow functional analysis and subsequent dimension reduction of large-scale culture of hematopoietic progenitors in single-cell environments. In this protocol, we describe a method to culture single cells of human CD34+ hematopoietic stem and progenitor cells (HSPCs) in a stromal cell culture with cytokines, to enumerate the clonal outcome (functionomics) of six different lineages (i.e., granulocyte, monocyte, lymphocyte, CD141+ dendritic cell (conventional type 1 DC (cDC1)), CD1c+ dendritic cell (conventional type 2 DC (cDC2)), and plasmacytoid dendritic cells (pDC)) from each progenitor, and to cluster the progenitors according to function with a dimension reduction method. In our previous paper, we showed that this is feasible for a population of 2,247 progenitors; each progenitor can be individually plotted to form a two-dimensional ‘map’ of 2,247 data points (Lee et al., 2017). Our protocol allows for the functional clustering of single cells. Such high dimensional functionomic analysis aids in linking cell function to molecular mechanism of any given cell population.

Materials and Reagents

  1. 15 cm culture plates (Corning, Falcon®, catalog number: 353025 )
  2. 15 ml Falcon tubes (Corning, Falcon®, catalog number: 352097 )
  3. 96-well V-bottom plates (Thermo Fisher Scientific, Thermo ScientificTM, catalog number: 249570 )
  4. Flat-bottom 96-well plates (Corning, Falcon®, catalog number: 353072 )
  5. 50 ml Falcon tubes (Corning, Falcon®, catalog number: 352098 )
  6. Serological pipettes (Fisher Scientific, catalog number: 13-678-11 )
  7. Opaque Eppendorf tubes (CELLTREAT Scientific Products, catalog number: 229437 )
  8. Eppendorf tubes (Fisher Scientific, catalog number: 05-408-129 )
  9. Pipette tips
  10. Pasteur pipette (Fisher Scientific, catalog number: 13-678-8B )
  11. 100 μm cell strainer (Corning, Falcon®, catalog number: 352360 )
  12. 0.20 μm filter (Corning, catalog number: 431229 )
  13. MS5 (Lee et al., 2015a and 2015b)
  14. OP9 (Lee et al., 2015a and 2017)
  15. dH2O (Thermo Fisher Scientific, InvitrogenTM, catalog number: 10977023 )
  16. Trypsin (Corning, catalog number: 25-052-CI )
  17. Trypan blue (Thermo Fisher Scientific, GibcoTM, catalog number: 15250061 )
  18. 70% ethanol
  19. Ficoll-Paque Plus (GE Healthcare, catalog number: 17144002 )
  20. Anti-human CD34 MACS microbeads (Miltenyi Biotec, catalog number: 130-046-702 )
  21. Recombinant human Flt3L (Celldex Therapeutics, catalog number: CDX-301 )
  22. Recombinant human SCF (PeproTech, catalog number: 300-07 )
  23. Recombinant human GM-CSF (PeproTech, catalog number: 300-03 )
  24. Antibodies (details of clone name, fluorochrome, manufacturer and dilution are listed in Tables 1 and 3)
  25. MEMα with deoxyribonucleosides (Thermo Fisher Scientific, GibcoTM, catalog number: 12571071 )
  26. Fetal bovine serum (FBS) (Thermo Fisher Scientific, GibcoTM, catalog number: 10437028 )
  27. Pen-Strep (Thermo Fisher Scientific, GibcoTM, catalog number: 15140122 )
  28. Mitomycin C (Sigma-Aldrich, catalog number: M4287-5X2MG )
  29. Dulbecco’s phosphate-buffered saline (DPBS) (GE Healthcare, catalog number: SH30378.02 )
  30. Ethylenediaminetetraacetic acid (EDTA) (Corning, catalog number: 46-034-CI )
  31. Bovine serum albumin (BSA) (ThermoFisher, catalog number: BP1600-100 )
  32. FcR blocking buffer (Miltenyi Biotec, catalog number: 130-059-901 )
  33. CountBright Absolute Counting Beads (Thermo Fisher Scientific, InvitrogenTM, catalog number: C36950 )
  34. Complete MEMα medium (see Recipes)
  35. Mitomycin C stock (1 mg/ml) (see Recipes)
  36. FACS buffer (see Recipes)
  37. Anti-CD34 microbeads/FcR blocking mix (see Recipes)


  1. Pipettes
  2. Incubator (Forma Scientific, model: 3354 )
  3. Centrifuge (Eppendorf, model: 5810 R , catalog number: 5811000320)
  4. Vortex with universal holder (VWR, catalog number: 97043-562 )
  5. Microscope
  6. Flow cytometer (BD, models: BD LSRII or LSRFortessaTM )


  1. Culture of stromal cells (day -4 – day 0)
    1. Passaging stromal cells (maximum of 13 times)
      1. Grow MS5 and OP9 cells in complete MEMα medium. For 15 cm plates, seed 1.5 million cells in 20 ml medium.
      2. For normal passage, microscopically verify that the cells are 80-90% confluent.
      3. Remove medium from the culture plate by aspiration and wash cells with 10 ml sterile PBS to remove medium and FBS.
      4. Remove the PBS, and then add 3 ml trypsin pre-warmed to 37 °C.
      5. Incubate for 2-3 min at 37 °C, and microscopically verify that cells have started to detach.
      6. Add 10 ml complete MEMα medium pre-warmed to 37 °C to stop trypsin activity.
      7. Resuspend cells by pipetting up and down, and then transfer cells to 15 ml collecting tubes.
      8. Centrifuge at 450 x g (1,500 rpm) for 5 min at 4 °C.
      9. Remove supernatant, and then resuspend the cells in 1 ml complete MEMα medium (Cell Concentrate).
      10. In a new 15 cm plate, add 20 ml complete MEMα medium. Transfer 100 μl cell concentrate to the new plate. This new plate will become confluent after 3-4 days.
      11. Incubate at 37 °C, 5% CO2 for 3-4 days, or until cells are 80-90% confluent.
      12. Passage the cells again, to a maximum of 13 passages.
      13. For MS5 and OP9 cells, cell yield from a 15 cm plate with 80-90% confluency is 15-20 million cells.
    2. Treating stromal cells with Mitomycin-C
      1. Verify that MS5/OP9 cells are 80-90% confluent.
      2. Label collecting tubes (‘MS5 + mitC’, ‘OP9 + mitC’).
      3. Add Mitomycin C stock (1 mg/ml, 100x) directly into the medium to achieve a final concentration of 10 μg/ml. For example, aspirate medium and add back exactly 10 ml, and add 100 μl Mitomycin C stock.
      4. Mix by swirling gently for 3-5 sec.
      5. Incubate at 37 °C for 3 h.
    3. Plating stromal cells
      1. Remove medium, and then add 10 ml PBS. Gently swirl the plate for 3-5 sec.
      2. Remove the PBS, and then add 3 ml trypsin.
      3. Incubate for 2-3 min at 37 °C, and use a microscope to verify that cells have started to detach.
      4. Add 10 ml complete MEMα medium to stop trypsin.
      5. Resuspend cells by pipetting up and down, and then transfer cells to labeled harvesting tubes.
      6. Centrifuge at 450 x g for 5 min at 4 °C.
      7. Remove medium, and then resuspend in 1 ml complete MEMα medium.
      8. Count cells, and calculate the number and volume of cells needed for the final plating. Each well in a flat-bottom 96-well plate, should contain 6,250 OP9 cells and 37,500 MS5 cells in 50 μl complete MEMα medium. To prepare for one 96-well plate, resuspend 0.625 x 106 OP9 cells and 3.75 x 106 MS5 cells in 5 ml complete MEMα medium, mix well, and then distribute 50 μl into each well.
      9. Incubate cells at 37 °C overnight.
      10. Verify that cells are securely attached to plate by shaking gently while viewing under a light microscope. With secure attachment, the cells should not drift around with the fluid movement.
      11. Within 1-3 days, sort and plate progenitors on these cells.

  2. Purification of CD34+ hematopoietic stem and progenitor cells from human cord blood (day 0)
    1. Isolation of cord blood mononuclear cells using Ficoll-Paque
      Note: This step should be handled at room temperature.
      1. Add 25 ml PBS into multiple 50 ml tubes.
      2. Sterilize the outside of the cord blood sample container with ethanol. Samples usually contain about 100 ml of blood.
      3. In 15 ml portions, evenly distribute the cord blood sample into the prepared 50 ml Falcon tubes containing PBS.
      4. Rinse the blood sample container with some additional room temperature PBS, and use the rinse to ‘top off’ the last Falcon tube (such that each tube has a fluid volume of 40 ml). On average, use 7 Falcon tubes per bag of cord blood.
      5. Tightly seal each of the tubes, and gently invert them 3-5 times to resuspend any sedimented cells.
      6. Underlay the cell suspension with 10.5 ml of Ficoll-Paque with a 10 ml Serological pipette, in the following manner:
        1. Load 10.5 ml into the pipette.
        2. Without releasing any fluid, gently bring the pipette to the bottom of the tube.
        3. For the first 2 ml, very slowly release Ficoll-Paque into the bottom of the tube. The goal is to have a pure layer of Ficoll-Paque at the bottom of the tube.
        4. For the next 8 ml, the Ficoll-Paque release can be slightly sped up, but use caution to avoid turbulence that may disrupt the Ficoll-Paque-blood boundary.
        5. For the last 0.5 ml, slow down the Ficoll-Paque release, and observe the meniscus in the pipette. Use caution to prevent introducing air bubbles into Ficoll-Paque.
      7. Centrifuge at 450 x g (1,500 rpm) for 25 min at room temperature. Use lower acceleration (level 3 of 9) and no brakes for deceleration. Orient the opaque sections of the test tube face towards the central motor of the centrifuge, so that cell deposits will collect on a transparent section of the test tube, away from the motor, and so can be easily visualized.
      8. After centrifugation, there should be four layers:
        1. The topmost layer (yellow/clear, ~35 ml) is a PBS/serum mix.
        2. The second layer from top (cloudy, ~1 ml) is a leukocyte-containing buffy coat.
        3. The third layer from top (clear, ~10 ml) contains Ficoll-Paque.
        4. The bottommost layer (red, ~5 ml) contains RBCs.
        5. Additionally, at the buffy coat layer, a smeared pellet of cells (cloudy, ~1 mm) should be deposited along the test tube wall.
      9. Transfer the collected cells into buffy collection tube:
        1. Prepare one 15 ml buffy collection tube for each 50 ml tube from the previous steps. Add 2 ml PBS in each tube.
        2. Remove the PBS/serum layer until only 10 ml of this layer remains, using caution to avoid turbulence by aspirating from the top of the layer. (i.e., if the buffy coat line is at the 30 ml line, then retain 40 ml total fluid remains.)
        3. With a 10 ml Serological pipette, transfer the buffy coat layer to a buffy collection tube, and use caution to minimize fluid transfer. To do so, hold the pipette tip directly above the buffy coat, and slowly aspirate, then eject fluid to the buffy collection tube. Leave ~5 ml of PBS/serum for the next step.
        4. Resuspend the smeared pellet of cells in the Ficoll-Paque by gently rocking the tube, using caution to avoid mixing the fluid layers. The rocking motion allows the liquid-liquid interface between the Ficoll-Paque and the PBS/serum mix to rub the pellet off the wall.
        5. Transfer as much Ficoll-Paque layer to the buffy collection tube as possible, while completely avoiding the RBC layer. Avoiding the PBS/serum layer is ideal, but not critical.
      10. Centrifuge at 450 x g (1,500 rpm) for 10 min at 4 °C.
      11. Remove supernatant, and resuspend cells in 1 ml FACS buffer.
      12. Collect cell suspensions from all collection tubes into a single 50 ml Falcon tube.
      13. Wash the cells by adding FACS buffer until the total fluid volume is 50 ml.
      14. Count the cells. Each bag of cord blood yields ~300-1,000 million cells.
      15. Centrifuge at 450 x g for 5 min at 4 °C.
      16. Remove supernatant, deriving a mononuclear cell pellet.
    2. Enrichment of CD34+ cells 
      1. Prepare anti-CD34 microbeads/FcR blocking mix.
      2. In a 15 ml conical tube, dilute the cells in 2 μl anti-CD34 microbeads/FcR blocking mix per 1 million nucleated cells.
      3. Incubate at 4 °C for 30-40 min.
      4. Add FACS buffer up to the 15 ml line.
      5. Centrifuge at 450 x g (1,500 rpm) for 5 min at 4 °C.
      6. Remove supernatant.
      7. Resuspend cells in FACS buffer (1 ml per 100 million cord blood mononuclear cells).
      8. Prepare an elution tube: 15 ml Falcon tube + 2 ml FACS buffer.
      9. Perform positive selection for CD34+ cells with an LS MACS column:
        1. Place a 100 μm cell strainer on top of the column.
        2. Equilibrate the column by filtering 5 ml FACS buffer through the strainer; discard the flow-through.
        3. Gently vortex the cells to ensure suspension.
        4. Filter the cells through the LS MACS column; retain the flow-through.
        5. Optional: wash the column once by passing 4 ml FACS buffer through it; discard this wash. Then, re-load the flow-through from the previous step on the column to increase cell yield.
        6. Wash the column by passing 4 ml FACS buffer through it; discard the wash. Repeat the wash a second time, and discard the wash again.
        7. Remove the MACS column from its magnetic holder, and place it on top of the elution tube.
        8. Add 4 ml of FACS buffer to the column, and allow gravity elution to occur.
        9. Add 4 ml of FACS buffer, taking caution to prevent the column from drying out, and use a plunger to push the FACS buffer through.
      10. Count the cells. Each bag of cord blood yields 300-1,000 x 106 CD34+ cells.
      11. Centrifuge at 450 x g for 5 min at 4 °C.
      12. Remove supernatant.
      13. (Optional) If necessary, cells can be stored overnight on ice in 4 °C, or frozen in 10% DMSO + FBS media; however, proceeding with fresh cells without an intermediate freezing step is recommended to avoid affecting clonal output and cell viability.

  3. Culture of purified cells in single-cell environment (day 0-7)
    1. Single-cell sorting and plating
      1. Stain CD34+ HSPCs with antibody mix. Use 10 μl antibody mix for 1 x 106 cells according to the following dilution (Table 1):

        Table 1. Antibody preparation for purification of progenitors from human cord blood

      2. Incubate cells on ice for 40 min, and using flow cytometry, sort cells according to following gating (Table 2, Figure 1): 

        Table 2. Characterization of progenitor populations in human cord blood

        Figure 1. Visual gating strategy of progenitor populations in human cord blood. Flow cytometry plots show separation of CD34+ cord blood cells into ten populations, HSC, MPP, LMPP, MLP, BNKP, CMP, MEP, GMDP, MDP and CDP.

      3. Sort single cells into each well of stromal-cell-containing 96-well plates with 50 μl complete MEMα medium per well.
    2. Cell culture over time
      1. Prepare a 2x mixture of cytokines in complete medium:
        1. 100 ng/ml Flt3L (final 1x concentration 50 ng/ml).
        2. 20 ng/ml SCF (final 1x concentration 10 ng/ml).
        3. 10 ng/ml GM-CSF (final 1x concentration 5 ng/ml) (not required to culture cells; stimulates and inhibits proliferation of various cell subtypes).
      2. Add the 2x mixture of cytokines to seeded stromal cell wells. (Add 50 μl for 96-well plate)
      3. Replenish media with cytokine every 7 days.
        1. On day 7, for each well, prepare 50 μl medium containing cytokines at 3x concentration. Gently add 50 μl fresh media with 3x cytokines into the well by pipetting onto the wall. 
          Note: A higher concentration of cytokines is prepared because the final media volume will be increased, while the volume of media-cytokines mixture remains the same. Media is not removed from the wells yet, due to the possibility of disturbing cell attachment.
        2. On day 14, prepare 50 μl medium containing cytokine at 3x concentration as in Step C2c i. Gently remove 50 μl of the old medium from each well, and then add 50 μl fresh media into the well by pipetting onto the wall.

  4. Harvest and analyze cells (day 7~21)
    1. Read cells via flow cytometry
      1. Prepare antibody mixture in FACS buffer using the following dilution (Table 3).

        Table 3. Antibody preparation for analysis of progeny from human cord blood

      2. Add PBS such that each well is filled, using a multichannel pipet. (250 μl on a 96-well plate)
      3. Pipet up and down 10-20 times to resuspend cells, and transfer 150 μl to 96-well V-bottom plates for staining.
      4. Centrifuge at 450 x g for 5 min at 4 °C.
      5. Discard the supernatant by microtip aspiration, using caution to keep the fluid level above the slanted portion of the well.
      6. Transfer the remaining 100 μl of cells to the 96-well V-bottom plates.
      7. Centrifuge at 450 x g for 4 min at 4 °C.
      8. Discard the supernatant, as before.
      9. Gently vortex the plate to resuspend the cells, using caution to avoid any fluid spillage. To do so, reduce the vortex speed to minimum, grasp the plate firmly, and firmly place it on the vortex mixer. Slowly increase the vortex speed, and gently reduce pressure on the vortex mixer so that the vortex mixer can shake. Then, decrease the vortex speed and remove the plate. As needed, this process may be repeated in each of the plate’s four quadrants. It is advised to practice this technique with a test plate and tap water to ensure no sample loss. (An alternative resuspension method is to gently pipet 10-20 times.)
      10. Add 4 μl of antibodies.
      11. Gently vortex the plate to mix cells with the antibody, as before.
      12. Centrifuge with a quick spin (50 x g, 5 sec) to bring down the cells and staining mix to the bottom of the well.
      13. Incubate in the dark for 40 min at 4 °C.
      14. Add 200 μl PBS to wash.
      15. Centrifuge at 450 x g for 5 min at 4 °C. Discard the supernatant.
      16. Resuspend in 50 μl FACS buffer.
      17. Vortex the counting beads stock solution for 30 sec, and make a 1:10 dilution of counting beads in FACS buffer.
      18. Add 10 μl of dilute counting beads to each well (~1,000 beads per well).
      19. Acquire cells on a flow cytometer (BD-LSRII or LSR Fortessa), and gate cells according to the following gating (Table 4) (Figure 2 and Lee et al., 2017).

        Table 4. Characterization of progeny populations in human cord blood

    2. Proceed to data analysis

      Figure 2. Visual gating strategy for analysis of progeny populations in human cord blood. Flow cytometry plots show phenotype of six cell types differentiated from CD34+ cells cultured in MP+FSG, granulocytes, monocytes, cDC1, cDC2, pDC and B/NK (lymphoid) cell.

Data analysis

  1. Data preparation
    1. Export flow cytometry data and save in a .csv file, such that:
      1. Each data column contains one ‘dimension’ of data.
      2. Each row contains one sample’s data.
      3. One column ‘Batch’ exists (case sensitive); all samples from one patient should have identical batch IDs.
      4. One column ‘Progenitor’ exists (case sensitive).
      5. All data columns are consecutive, each containing the number of relevant reads in each sample-progeny data point. (Figure 3)

        Figure 3. Example of first few rows of a valid .csv file viewed in Excel

  2. Data normalization in R
    1. Open cPlot.R in RStudio, then Source the file. (Code > Source) (The file is available online at https://github.com/kangliulab/cplot/blob/master/cPlot_v0.7.R)
    2. Change working directory to the location of data files. (Session > Set Working Directory > Choose Directory)
      Note: Do not use a working directory in a synced folder (OneDrive, Dropbox, etc.), as this may cause some problems with file outputs.
    3. Normalize data using one of the following functions. To use, enter the function into the console, replacing inFile with the file name, firstData with the first data column (leftmost column is #1), and dataCols with the number of data columns. Keyword may be omitted if desired; if used, it adds the keyword to the start of each row to help differentiate between files. When replacing words, keep any quotation marks that are denoted below.
      1. addCB (“inFile”, firstData, dataCols, “keyword”) is used for samples that must be normalized both by subject (i.e., batch) and by progeny (i.e., lineage).
      2. addPatient (“inFile”, firstData, dataCols, “keyword”) is used for samples that must be normalized only by subject (i.e., batch).
      3. addPreNorm (“inFile”, firstData, dataCols, “keyword”) is used for samples that have already been normalized.
      4. Examples:
        1. addCB (“data-file-v123_for-test-use.csv”, 5, 6).
        2. addCB (“data-file-v123_for-test-use.csv”, 5, 6, “v123”).
    4. Generate t-SNE maps using the below function. To use, enter it into the console. You may choose to omit all numbers, which causes the script to use the default numbers (denoted below). To replace numbers, indicate both the variable to be changed and its new value.
      1. generateManyMaps(dataCols = 6, perVector = c(50), thetaVector = c(0.10), iterVector = c(1000), etaVector = c(200), n = 1, seedSet = FALSE, randSeed = 6) allows multiple parameters to be simultaneously tested. It can also be used to generate multiple maps with a single set of parameters. Note that it will try every possible combination of parameters, and will try each combination n times. dataCols should be equal to your number of data columns, perVector selects perplexity values, thetaVector selects theta values, iterVector selects iterations to run, etaVector selects eta values, n selects number of repetitions per combination of parameters, seedSet forces each plot to have the same random seed, and randSeed selects the random seed of the first plot (or, the random seed of all plots if seedSet is TRUE).
      2. Examples:
        1. generateManyMaps().
        2. generateManyMaps(n = 10, seedSet = TRUE).
        3. generateManyMaps(perVector = c(40, 50, 60), etaVector = c(150, 200, 250), n = 3).
    5. Assign cell lineages using each of the below functions in sequential order.
      1. pickMap(“inFile”) allows a particular t-SNE map to be imported for further analysis. inFile should represent a data file that was created by generateManyMaps().
      2. assignLineages() automatically calculates an appropriate lineage track for each data point. It can only be used after pickMap() has been used. It can take a long time (3-8 h) to run, especially for larger data files!
      3. compileAllData(“inFile”) generates a .csv with raw data, normalized data, tSNE coordinates, and assigned lineages. It assumes that data has already been imported into RStudio. If RStudio was closed sometime between importing the data (in step 3) and this step, then you should re-import the data before running compileAllData(). inFile should represent a data file that was created by assignLineages().
      4. Examples:
        1. pickMap(“data1_p50_t0.1_s6_i1000.csv”).
        2. assignLineages().
        3. compileAllData(“2assignedLineages_FC0.7.csv”).

Above, we describe an outlined approach to our data. As described in our previous paper, each progenitor’s output is imagined as a 6D vector. A preliminary data cleanup step involves neglecting low cell counts: below 2 cells for CDPs and below 6 cells for all other cell types; the neglect occurs by changing the reading to 0 cells. Progenitors with a total of 0 cell output after data cleanup should then be purged. Data normalization of cell type yield follows the DESeq procedure, assuming the geometric mean of total clonal output for a single progenitor phenotype across different donors should be similar. A second normalization of cell type yield was also done via DESeq, to ensure similar geometric mean of each progeny type yield across all progenitors.
Barnes-Hut t-SNE (Van Der Maaten, 2014) was performed to visualize the inferred pair-wise similarity data in 2 dimensions. Potency similarity between any pair of cells is inferred by their proximity in this space using Gaussian diffusion kernel.
To assign lineages to each progenitor, we begin by defining six ‘backbones’; one for each type of progeny output. The backbone of each progeny type consists of any progenitors with > 70% commitment degree to that progeny; commitment degree is defined as (yield of one progeny type)/(total yield of all progeny types). Then, distance of each progenitor to any given track is defined as the 6D Euclidean distance between a progenitor of interest and all backbone progenitors of that track. The progenitor is assigned to the lineage corresponding with the track that it has the lowest distance to.


Although this culture protocol only reports differentiation potential of granulocyte, monocyte, lymphocyte, CD141+ dendritic cell (cDC1), CD1c+ dendritic cell (cDC2), and plasmacytoid dendritic cells (pDC) of hematopoietic lineage, the analysis method can be applied to any culture system that nurtures different lineages as far as that the cellular output can be quantified.


  1. Complete MEMα medium
    10% FBS
    1x Pen/Strep
  2. Mitomycin C stock (100x, 1 mg/ml)
    1. Suspend 2 mg Mitomycin C powder in 2 ml dH2O
    2. Vortex solution
    3. Filter solution with a 0.20 μm filter
    4. Keep stock for up to 3 months at -20 °C
  3. FACS buffer
    2 mM EDTA
    0.5% BSA
  4. Anti-CD34 microbeads/FcR blocking mix
    Prepare 2 μl of the mix per 1 million total nucleated cells
    15% FcR blocking buffer
    25% anti-CD34 MACS beads
    60% FACS buffer


This work was supported by the Empire State Stem Cell Fund through the New York State Department of Health (C029562 to K.L.) and The US National Institute of Health (AI101251 and OD023291 to K.L.). Research reported in this manuscript was performed partly by the Columbia Center for Translational Immunology Flow Cytometry Core, supported in part by the Office of the Director of the US National Institutes of Health (S10RR027050 and S10OD020056). This protocol was adapted from previous work Lee et al., 2017, Lee et al., 2015a. Authors declare no conflicts of interest or competing interests. The recombinant human Flt3L was a gift from Celldex Therapeutics.


  1. Amir el, A. D., Davis, K. L., Tadmor, M. D., Simonds, E. F., Levine, J. H., Bendall, S. C., Shenfeld, D. K., Krishnaswamy, S., Nolan, G. P. and Pe'er, D. (2013). viSNE enables visualization of high dimensional single-cell data and reveals phenotypic heterogeneity of leukemia. Nat Biotechnol 31(6): 545-552.
  2. Levine, J. H., Simonds, E. F., Bendall, S. C., Davis, K. L., Amir el, A. D., Tadmor, M. D., Litvin, O., Fienberg, H. G., Jager, A., Zunder, E. R., Finck, R., Gedman, A. L., Radtke, I., Downing, J. R., Pe'er, D. and Nolan, G. P. (2015). Data-driven phenotypic dissection of AML reveals progenitor-like cells that correlate with prognosis. Cell 162(1): 184-197.
  3. Lee, J., Zhou, Y. J., Ma, W., Zhang, W., Aljoufi, A., Luh, T., Lucero, K., Liang, D., Thomsen, M., Bhagat, G., Shen, Y. and Liu, K. (2017). Lineage specification of human dendritic cells is marked by IRF8 expression in hematopoietic stem cells and multipotent progenitors. Nat Immunol 18(8): 877-888.
  4. Lee, J., Breton, G., Aljoufi, A., Zhou, Y. J., Puhr, S., Nussenzweig, M. C. and Liu, K. (2015a). Clonal analysis of human dendritic cell progenitor using a stromal cell culture. J Immunol Methods 425: 21-26.
  5. Lee, J., Breton, G., Oliveira, T. Y., Zhou, Y. J., Aljoufi, A., Puhr, S., Cameron, M. J., Sekaly, R. P., Nussenzweig, M. C. and Liu, K. (2015b). Restricted dendritic cell and monocyte progenitors in human cord blood and bone marrow. J Exp Med 212(3): 385-399.
  6. Paul, F., Arkin, Y., Giladi, A., Jaitin, D. A., Kenigsberg, E., Keren-Shaul, H., Winter, D., Lara-Astiaso, D., Gury, M., Weiner, A., David, E., Cohen, N., Lauridsen, F. K. B., Haas, S., Schlitzer, A., Mildner, A., Ginhoux, F., Jung, S., Trumpp, A., Porse, B. T., Tanay, A. and Amit, I. (2016). Transcriptional heterogeneity and lineage commitment in myeloid progenitors. Cell 164(1-2): 325.
  7. See, P., Dutertre, C. A., Chen, J., Gunther, P., McGovern, N., Irac, S. E., Gunawan, M., Beyer, M., Handler, K., Duan, K., Sumatoh, H. R. B., Ruffin, N., Jouve, M., Gea-Mallorqui, E., Hennekam, R. C. M., Lim, T., Yip, C. C., Wen, M., Malleret, B., Low, I., Shadan, N. B., Fen, C. F. S., Tay, A., Lum, J., Zolezzi, F., Larbi, A., Poidinger, M., Chan, J. K. Y., Chen, Q., Renia, L., Haniffa, M., Benaroch, P., Schlitzer, A., Schultze, J. L., Newell, E. W. and Ginhoux, F. (2017). Mapping the human DC lineage through the integration of high-dimensional techniques. Science 356(6342).
  8. Van Der Maaten, L. (2014). Accelerating t-sne using tree-based algorithms. J Mach Learn Res 15: 3221-3245.


在单细胞水平进行细胞转录,信号传导和功能调查的能力为以前所未有的决议研究异质人群开启了机会。 尽管已经开发了评估高维转录组学和蛋白质组学数据(与细胞mRNA和蛋白质有关)的方法,但尚未有方法从单个细胞评估相应的高维功能组学数据(与细胞功能有关)。 在这里,我们提出了一种方案来定量测量单个人造血干细胞和祖细胞的分化潜能,然后根据这些测量结果聚集细胞。 细胞电位的高维功能组分析允许细胞功能与相同祖细胞群体内的分子机制相关联。

【背景】单细胞水平的细胞转录,信号传导和功能单细胞测量技术的发展,以及流式细胞仪等先前存在的技术的发展,使得新镜头能够检测复杂的异质群体。这些方法产生大量数据,这可以借助于降维算法来解释,如使用Mpath,Monocole,PCA,Wishbone或扩散图算法在单细胞RNA-Seq上所示的(Paul等, 2016年;参见 et al。,2017),以及使用tSNE或PhenoGraph的CyTOF(Amir et al。,2013; Levine et al 。,2015)。

我们开发了这个协议,以允许在单细胞环境中对造血祖细胞的大规模培养物进行功能分析和随后的降维。在这个协议中,我们描述了在细胞因子的基质细胞培养物中培养人CD34 +造血干细胞和祖细胞(HSPC)的单个细胞以列举六种不同的克隆结果(功能组学)谱系(em),粒细胞,单核细胞,淋巴细胞,CD141 +树突细胞(常规1型DC(cDC1)),CD1c +树突细胞(常规2型DC(cDC2))和浆细胞样树突状细胞(pDC)),并根据功能用降维法聚集祖细胞。在我们以前的文章中,我们表明这对于2,247个祖先的人群是可行的;每个祖细胞可以分别绘制成2,247个数据点的二维“图谱”(Lee等人,2017)。我们的协议允许单细胞的功能聚类。这种高维功能组学分析有助于将细胞功能与任何给定细胞群的分子机制联系起来。

关键字:CD34+造血干细胞和祖细胞, 单细胞培养, 分化, 定量克隆输出, Barnes-Hut t-SNE, 降维


  1. 15厘米培养板(Corning,Falcon ,目录号:353025)
  2. 15毫升Falcon管(Corning,Falcon ,目录号:352097)
  3. 96孔V底板(Thermo Fisher Scientific,Thermo Scientific TM,目录号:249570)
  4. 平底96孔板(Corning,Falcon ,产品目录号:353072)
  5. 50ml Falcon管(Corning,Falcon ,目录号:352098)
  6. 血清移液器(Fisher Scientific,目录号:13-678-11)
  7. 不透明Eppendorf管(CELLTREAT科学产品,目录号:229437)
  8. Eppendorf管(Fisher Scientific,目录号:05-408-129)
  9. 移液器吸头
  10. 巴斯德吸管(Fisher Scientific,目录号:13-678-8B)
  11. 100μm细胞过滤器(Corning,Falcon ,目录号:352360)
  12. 0.20μm过滤器(Corning,目录号:431229)
  13. MS5(Lee等人,2015a和2015b)
  14. OP9(Lee等人,2015a和2017)
  15. (Thermo Fisher Scientific,Invitrogen TM,目录号:10977023)。
  16. 胰蛋白酶(康宁,目录号:25-052-CI)
  17. 台盼蓝(Thermo Fisher Scientific,Gibco TM,目录号:15250061)
  18. 70%乙醇
  19. Ficoll-Paque Plus(GE Healthcare,目录号:17144002)
  20. 抗人CD34 MACS微珠(Miltenyi Biotec,目录号:130-046-702)
  21. 重组人Flt3L(Celldex Therapeutics,目录号:CDX-301)
  22. 重组人SCF(PeproTech,目录号:300-07)
  23. 重组人GM-CSF(PeproTech,目录号:300-03)

  24. 抗体(克隆名称,荧光染料,生产商和稀释度的详细信息列在表1和表3中)
  25. 含有脱氧核糖核苷的MEM(Thermo Fisher Scientific,Gibco TM,目录号:12571071)
  26. 胎牛血清(FBS)(Thermo Fisher Scientific,Gibco TM,目录号:10437028)
  27. Pen-Strep(Thermo Fisher Scientific,Gibco TM,目录号:15140122)
  28. 丝裂霉素C(Sigma-Aldrich,目录号:M4287-5X2MG)
  29. Dulbecco磷酸盐缓冲盐水(DPBS)(GE Healthcare,目录号:SH30378.02)
  30. 乙二胺四乙酸(EDTA)(Corning,目录号:46-034-CI)
  31. 牛血清白蛋白(BSA)(ThermoFisher,目录号:BP1600-100)
  32. FcR封闭缓冲液(Miltenyi Biotec,目录号:130-059-901)
  33. CountBright Absolute Counting Beads(Thermo Fisher Scientific,Invitrogen TM,目录号:C36950)
  34. 完整的MEMα培养基(见食谱)
  35. 丝裂霉素C储备(1毫克/毫升)(见食谱)
  36. FACS缓冲液(见食谱)
  37. 抗CD34微珠/ FcR阻断混合物(见食谱)


  1. 移液器
  2. 孵化器(Forma Scientific,型号:3354)
  3. 离心机(Eppendorf,型号:5810 R,目录号:5811000320)
  4. 用通用支架涡旋(VWR,目录号:97043-562)
  5. 显微镜
  6. 流式细胞仪(BD,型号:BD LSRII或LSRFortessa TM)


  1. 基质细胞培养(第-4天 - 第0天)
    1. 传代基质细胞(最多13次)
      1. 在完全MEMα培养基中培养MS5和OP9细胞。对于15厘米平板,在20毫升培养基中种植150万个细胞。
      2. 对于正常通道,显微镜检查细胞是80-90%汇合。
      3. 通过抽吸从培养板中除去培养基,用10ml无菌PBS清洗细胞以除去培养基和FBS。
      4. 取出PBS,然后加入3 ml预热至37°C的胰蛋白酶。

      5. 在37°C孵育2-3分钟,显微镜检查细胞已经开始分离。

      6. 加入10 ml完全MEMα培养基预热至37°C停止胰蛋白酶活性
      7. 通过上下移液重悬细胞,然后将细胞转移到15 ml收集管中。
      8. 在4℃下以450×g(1,500rpm)离心5分钟。
      9. 除去上清液,然后将细胞重悬于1ml完全MEMα培养基中。 (细胞浓缩)
      10. 在新的15厘米板中,加入20毫升完整的MEMα培养基。将100μl细胞浓缩液转移到新的平板上。
      11. 在37℃,5%CO 2下孵育3-4天,或直到细胞达到80-90%汇合。

      12. 再次通过细胞,最多可以传代13次。
      13. 对于MS5和OP9细胞,具有80-90%汇合度的15cm平板的细胞产量是15-20万个细胞。
    2. 用丝裂霉素C处理基质细胞
      1. 确认MS5 / OP9细胞是80-90%汇合。
      2. 标签收集管('MS5 + mitC','OP9 + mitC')。
      3. 直接将丝裂霉素C贮存液(1mg / ml,100x)加入培养基中以达到10μg/ ml的终浓度。例如,吸出培养基并加回10毫升,然后加入100微升丝裂霉素C原液。

      4. 轻轻旋转混合3-5秒。

      5. 在37°C孵育3小时。
    3. 电镀基质细胞
      1. 取出培养基,然后加入10 ml PBS。轻轻旋转盘子3-5秒。
      2. 取出PBS,然后加入3毫升胰蛋白酶。

      3. 在37°C孵育2-3分钟,并使用显微镜确认细胞已经开始分离。

      4. 加入10ml完整的MEMα培养基以终止胰蛋白酶。
      5. 通过上下移液重悬细胞,然后将细胞转移到标记的收获管中。

      6. 在450℃下离心5分钟4分钟。
      7. 移除培养基,然后重悬于1ml完全MEMα培养基中。
      8. 计数细胞,并计算最终电镀所需细胞的数量和体积。平底96孔板中的每个孔应在50μl完全MEMα培养基中含有6,250个OP9细胞和37,500个MS5细胞。为了准备一块96孔板,在5ml完全MEMα培养基中重悬0.625×10 6个OP9细胞和3.75×10 6个MS5细胞,充分混合,然后分配每孔加入50μl。

      9. 在37°C培养细胞过夜。
      10. 通过在光学显微镜下观察时轻轻摇动,确认细胞牢固地附着在平板上。通过安全连接,细胞不应该随着流体运动而漂移。
      11. 在1-3天内,对这些细胞进行分选和平板祖细胞。

  2. 从人脐带血中纯化CD34 +造血干细胞和祖细胞(第0天)
    1. 1.使用Ficoll-Paque分离脐血单核细胞

      1. 添加25毫升PBS到多个50毫升管中。
      2. 用乙醇消毒脐血样本容器的外部。样品通常含有约100毫升的血液。
      3. 以15毫升的份量,将脐血样本均匀分布到准备好的含有PBS的50毫升Falcon管中。
      4. 用一些额外的室温PBS冲洗血液样本容器,并使用冲洗来“倒掉”最后的Falcon管(使得每个管具有40ml的流体体积)。平均每袋脐带血使用7只Falcon管。

      5. 紧密密封每个试管,并轻轻颠倒3-5次以重新悬浮任何沉淀细胞。
      6. 用10ml血清移液管以10.5ml Ficoll-Paque以下述方式置于细胞悬液的下面:
        1. 加入10.5毫升的移液器。
        2. 在不释放任何液体的情况下,轻轻地将移液管放在管的底部。
        3. 对于第一个2毫升,非常缓慢地释放Ficoll-Paque到管底部。我们的目标是在管子的底部涂上一层Ficoll-Paque。
        4. 对于接下来的8毫升,Ficoll-Paque释放可以稍微加快,但要小心避免可能扰乱Ficoll-Paque血液界限的湍流。
        5. 最后0.5 ml,减慢Ficoll-Paque释放,并观察吸管中的弯月面。请小心,以防止将气泡引入Ficoll-Paque。
      7. 在室温下450×g(1,500rpm)离心25分钟。使用较低的加速度(9级中的3级)并且没有制动器用于减速。将试管表面的不透明部分朝向离心机的中心马达定位,以便细胞沉积物将聚集在试管的透明部分上,远离马达,因此可以很容易地观察到。
      8. 离心后,应该有四层:
        1. 最上层(黄色/透明,约35毫升)是PBS /血清混合物。

        2. 上面的第二层(混浊,约1毫升)是含白细胞的血沉棕黄层
        3. 从顶部开始的第三层(透明,约10毫升)含有Ficoll-Paque。
        4. 最底层(红色,约5毫升)含有红细胞。
        5. 另外,在血沉棕黄层上,应该在试管壁上沉积一层细胞(多云,〜1mm)的污迹颗粒。
      9. 将收集的细胞转移到buffy收集管中:
        1. 从前面的步骤中为每个50 ml试管准备一个15 ml的buffy收集管。每管加入2 ml PBS。
        2. 除去PBS /血清层直到剩下10毫升的这一层,小心避免从层的顶部吸入湍流。 (即,如果血沉棕黄层线在30毫升线上,则保留40毫升总体流体)。
        3. 使用10 ml血清移液管,将血沉棕黄层转移至血沉棕黄色采集管,并小心使流体转移最小化。为此,请将移液器尖端直接放在血沉棕黄层上方,然后缓慢抽吸,然后将液体喷射到血沉棕黄色采集管。
          留下约5毫升PBS /血清用于下一步
        4. 轻轻地摇动管子,重新悬浮Ficoll-Paque中的细胞涂抹的小球,小心避免混合流体层。摇摆运动允许Ficoll-Paque和PBS /血清混合物之间的液 - 液界面将颗粒从墙上刮下。
        5. 尽可能多地将Ficoll-Paque层转移到buffy采集管,同时完全避免RBC层。避免PBS /血清层是理想的,但不是关键。
      10. 在450℃下离心450分钟(1,500rpm)10分钟。
      11. 取出上清液,并将细胞重悬于1ml FACS缓冲液中。
      12. 从所有收集管收集细胞悬液到一个50毫升Falcon管中。
      13. 通过加入FACS缓冲液清洗细胞,直到总流体体积为50毫升。
      14. 计数细胞。每袋脐血产生约300-1,000万个细胞。

      15. 在450℃下离心5分钟4分钟。
      16. 去除上清液,得到单核细胞颗粒。
    2. 富集CD34 +细胞 
      1. 准备抗CD34微珠/ FcR阻断混合物。
      2. 在15ml锥形管中,将细胞稀释成每1百万有核细胞2μl抗-CD34微珠/ FcR阻断混合物。

      3. 4°C孵育30-40分钟

      4. 添加FACS缓冲液至15 ml生产线
      5. 在4℃下以450×g(1,500rpm)离心5分钟。
      6. 去除上清液。
      7. 在FACS缓冲液中重悬细胞(每1亿脐血单核细胞1ml)。
      8. 准备洗脱管:15毫升Falcon管+ 2毫升FACS缓冲液。
      9. 用LS MACS柱对CD34 +细胞进行正向选择:

        1. 在柱子顶部放置一个100μm的细胞过滤器。
        2. 通过过滤器过滤5ml FACS缓冲液以平衡柱子;丢弃流通。
        3. 轻轻旋转细胞以确保悬浮。
        4. 通过LS MACS列过滤单元格;保留流通。
        5. 可选:将4 ml FACS缓冲液通过柱子,洗涤柱子一次;放弃这种洗涤。然后,重新加载柱上前一步的流通量以增加电池产量。
        6. 通过4毫升FACS缓冲液洗涤柱子;丢弃洗涤物。再次重复洗涤,再次丢弃洗涤液。

        7. 从磁力架上取下MACS柱,并将其放在洗脱管的顶部。
        8. 将4 ml FACS缓冲液加入色谱柱,并进行重力洗脱。
        9. 加入4 ml FACS缓冲液,小心避免柱子变干,并使用活塞推动FACS缓冲液。
      10. 计数细胞。每袋脐带血产生300-1,000×10 6个CD34 +细胞。

      11. 在450℃下离心5分钟4分钟。
      12. 去除上清液。
      13. (可选)如有必要,细胞可以在4°C的冰上放置过夜,或在10%DMSO + FBS培养基中冷冻;然而,建议在不进行中间冷冻步骤的情况下进行新鲜细胞以避免影响克隆输出和细胞活力。

  3. 纯细胞在单细胞环境中培养(第0-7天)
    1. 单细胞分选和电镀
      1. 用抗体混合物染色CD34 + HSPC。根据以下稀释度,使用10μl抗体混合物用于1×10 6个细胞(表1):


      2. 在冰上孵育细胞40分钟,并使用流式细胞术,根据以下门控对细胞进行分选(表2,图1): 


        图1.人脐血中祖细胞群的视觉门控策略流式细胞术图显示CD34 + /脐带血细胞分离为10个群体,HSC,MPP,LMPP,MLP ,BNKP,CMP,MEP,GMDP,MDP和CDP。

      3. 将单细胞分入含有基质细胞的96孔板的每个孔中,每孔含50μl完全MEMα培养基。
    2. 随时间推移的细胞培养
      1. 在完全培养基中制备2倍细胞因子混合物:
        1. 100 ng / ml Flt3L(最终1x浓度50 ng / ml)。
        2. 20 ng / ml SCF(最终1x浓度10 ng / ml)。
        3. 10ng / ml GM-CSF(最终1x浓度5ng / ml)(不需要培养细胞;刺激并抑制各种细胞亚型的增殖)。
      2. 添加2倍的细胞因子混合物到种子基质细胞孔。 (为96孔板加50μl)
      3. 每7天用细胞因子补充培养基。
        1. 在第7天,为每个孔准备50μl含有3倍浓度细胞因子的培养基。
          注意:制备较高浓度的细胞因子是因为最终的培养基体积会增加,而培养基 - 细胞因子混合物的体积保持不变。由于可能会干扰细胞附着,因此培养基尚未从孔中移除。
        2. 在第14天,如步骤C2c i所示,制备含有3倍浓度的细胞因子的50μl培养基。轻轻地从每个孔中取出50μl旧培养基,然后通过移液器将50μl新鲜培养基加入孔中。

  4. 收获和分析细胞(第7〜21天)
    1. 通过流式细胞仪读取细胞

      1. 使用以下稀释液在FACS缓冲液中制备抗体混合物(表3)。


      2. 使用多道移液器添加PBS,使得每个孔都被填充。 (在96孔板上250μl)
      3. 吸取上下10-20次以重悬细胞,并将150μl转移至96孔V底板进行染色。

      4. 在450℃下离心5分钟4分钟。
      5. 通过微尖抽吸弃去上清液,小心保持液位高于斜井部分。
      6. 将剩余的100μl细胞转移到96孔V底板。

      7. 在450℃下离心4分钟4分钟。
      8. 像以前一样丢弃上清。
      9. 轻轻地旋转平板以重悬细胞,小心避免任何流体溢出。为此,将涡旋速度降至最低,牢牢抓住平板,并将其牢牢放在涡旋混合器上。慢慢地增加涡流速度,并且轻轻地减小涡流混合器上的压力,以便涡流混合器可以振动。然后,降低涡旋速度并取下板。根据需要,这个过程可能会在每个板块的四个象限中重复出现。建议用测试板和自来水练习此技术以确保没有样品损失。 (另一种重悬方法是轻轻吸取10-20次。)
      10. 加入4μl抗体。
      11. 像以前一样轻轻地旋转平板以混合细胞与抗体。
      12. 快速旋转离心(50×g,5秒),使细胞和染色混合物下沉至孔底。

      13. 在黑暗中孵育40分钟
      14. 加入200μlPBS清洗。
      15. 在450℃下离心5分钟4分钟。丢弃上清液。
      16. 重悬于50μlFACS缓冲液中。
      17. 漩涡计数珠储备液30秒,并在FACS缓冲液中1:10稀释计数珠。

      18. 每孔加入10μl稀释的计数珠(每孔约1000个珠)。
      19. 在流式细胞仪(BD-LSRII或LSR Fortessa)上获取细胞,并根据以下门控(表4)(图2和Lee等人,2017)获得细胞。


    2. 继续进行数据分析

      图2.用于分析人脐带血中后代群体的视觉门控策略流式细胞术图显示了在MP + FSG中培养的与CD34 +细胞分化的6种细胞类型的表型,粒细胞,单核细胞,cDC1,cDC2,pDC和B / NK(淋巴细胞)。


  1. 数据准备
    1. 导出流式细胞术数据并保存在.csv文件中,以便:
      1. 每个数据列都包含一个“维度”数据。
      2. 每行包含一个样本的数据。
      3. 一列'批'存在(区分大小写);
      4. 一列'祖'存在(区分大小写)。
      5. 所有数据列都是连续的,每个列都包含每个样本子代数据点中相关读数的数量。 (图3)


  2. R中的数据规范化
    1. 在RStudio中打开cPlot.R,然后输入文件。 (Code> Source)(该文件可在 https:/ /github.com/kangliulab/cplot/blob/master/cPlot_v0.7.R。
    2. 将工作目录更改为数据文件的位置。 (会话>设置工作目录>选择目录)
    3. 使用以下功能之一对数据进行标准化。要使用,请在控制台中输入函数,将inFile替换为文件名firstData与第一个数据列(最左边的列为#1),dataCols替换为数据列数。如果需要,关键字可以省略;如果使用,它会将关键字添加到每行的起始处以帮助区分文件。替换单词时,请保留下面标注的任何引号。
      1. addCB(“inFile”,firstData,dataCols,“keyword”)用于必须由主题( ie ,批处理)和后代( >即,血统)。
      2. addPatient(“inFile”,firstData,dataCols,“keyword”)用于只能由主题(即,批次)进行归一化的样本。
      3. addPreNorm(“inFile”,firstData,dataCols,“keyword”)用于已经规范化的样本。
      4. 例子:
        1. addCB(“data-file-v123_for-test-use.csv”,5,6)。
        2. addCB(“data-file-v123_for-test-use.csv”,5,6,“v123”)。
    4. 使用下面的函数生成t-SNE地图。要使用,请将其输入到控制台中。您可以选择省略所有数字,这会导致脚本使用默认数字(如下所示)。要替换数字,请指明要更改的变量及其新值。
      1. generateManyMaps(dataCols = 6,perVector = c(50),thetaVector = c(0.10),iterVector = c(1000),etaVector = c(200),n = 1,seedSet = FALSE,randSeed = 6)允许同时测试多个参数。它也可以用于通过一组参数生成多个地图。请注意,它会尝试每种可能的参数组合,并会尝试每次组合n次。 dataCols应该等于您的数据列数,perVector选择困惑值,thetaVector选择theta值,iterVector选择迭代运行,etaVector选择eta值,n选择每个参数组合的重复次数,seedSet强制每个plot具有相同的值随机种子和randSeed选择第一个图的随机种子(或者,如果seedSet为TRUE,则所有图的随机种子)。
      2. 例子:
        1. generateManyMaps()。
        2. generateManyMaps(n = 10,seedSet = TRUE)。
        3. generateManyMaps(perVector = c(40,50,60),etaVector = c(150,200,250),n = 3)。
    5. 按顺序使用以下每个函数分配细胞谱系。
      1. pickMap(“inFile”)允许导入一个特定的t-SNE地图用于进一步分析。 inFile应该代表由 generateManyMaps()创建的数据文件。
      2. assignLineages()自动为每个数据点计算适当的沿袭轨迹。它只能在pickMap()被使用后才能使用。可能需要很长时间(3-8小时)才能运行,尤其是对于较大的数据文件!
      3. compileAllData(“inFile”)使用原始数据,标准化数据,tSNE坐标和指定的谱系生成.csv。它假定数据已经被导入到RStudio中。如果在导入数据(步骤3)和这一步之间有一段时间关闭了RStudio,那么您应该在运行compileAllData()之前重新导入数据。 inFile应该代表由assignLineages()创建的数据文件。
      4. 例子:
        1. pickMap(“data1_p50_t0.1_s6_i1000.csv”)。
        2. assignLineages()。
        3. compileAllData(“2assignedLineages_FC0.7.csv”)。

Barnes-Hut t-SNE(Van Der Maaten,2014)被执行以在二维中可视化推断的成对相似性数据。
任何一对细胞之间的效能相似性都是通过它们在这个空间中的接近性使用高斯扩散核来推断的 为了给每个祖先分配谱系,我们首先定义六个'骨干';一种用于每种子代产出。每个子代类型的主干由任何具有>对该后代70%的承诺程度;承诺程度定义为(一个后代类型的产量)/(所有后代类型的总产量)。然后,将每个祖先与任何给定轨迹的距离定义为感兴趣的祖先与该轨迹的所有主干祖先之间的6D欧几里德距离。祖先被分配到与距离最近的轨迹相对应的谱系。


虽然这种培养方案仅报告了粒细胞,单核细胞,淋巴细胞,CD141 +树突细胞(cDC1),CD1c +树突细胞(cDC2)和浆细胞样树突状细胞( pDC)造血谱系,该分析方法可以应用于培养不同谱系的任何培养系统,只要细胞输出可以被量化即可。


  1. 完整的MEMα培养基
    1x Pen / Strep
  2. 丝裂霉素C储备液(100x,1mg / ml)
    1. 将2毫克丝裂霉素C粉末悬浮于2毫升dH 2 O
    2. 涡流解决方案
    3. 使用0.20μm过滤器过滤溶液

    4. 在-20°C储存长达3个月
  3. FACS缓冲液
    2 mM EDTA
  4. 抗CD34微珠/ FcR阻断混合物
    25%抗CD34 MACS珠


这项工作得到了帝国州干细胞基金通过纽约州卫生部(C029562至K.L.)和美国国家卫生研究院(AI101251和OD023291至K.L.)的支持。在这份手稿中报道的研究部分由哥伦比亚转化免疫学流式细胞术核心中心进行,部分由美国国立卫生研究院院长办公室(S10RR027050和S10OD020056)支持。该协议是从以前的工作Lee等人,2017年,Lee等人,2015a改编的。作者声明不存在利益冲突或利益冲突。重组人Flt3L是来自Celldex Therapeutics的礼物。


  1. Amir el,A. D.,Davis,K. L.,Tadmor,M. D.,Simonds,E. F.,Levine,J. H.,Bendall,S. C.,Shenfeld,D. K.,Krishnaswamy,S.,Nolan,G. P.和Pe'er,D.(2013)。 viSNE可以显示高维度单细胞数据并揭示白血病的表型异质性。 Nat Biotechnol 31(6):545-552。
  2. Levine,JH,Simonds,EF,Bendall,SC,Davis,KL,Amir el,AD,Tadmor,MD,Litvin,O.,Fienberg,HG,Jager,A.,Zunder,ER,Finck,R.,Gedman, AL,Radtke,I.,Downing,JR,Pe'er,D。和Nolan,GP(2015)。 数据驱动的AML表型解剖显示与预后相关的祖细胞样细胞。 Cell 162(1):184-197。
  3. Lee,J.,Zhou,YJ,Ma,W.,Zhang,W.,Aljoufi,A.,Luh,T.,Lucero,K.,Liang,D.,Thomsen,M.,Bhagat,G.,Shen ,Y.和Liu,K。(2017)。 人类树突状细胞的谱系特征以造血干细胞和多潜能祖细胞中的IRF8表达为标志。 a> Nat Immunol 18(8):877-888。
  4. Lee,J.,Breton,G.,Aljoufi,A.,Zhou,Y.J.,Puhr,S.,Nussenzweig,M.C。和Liu,K.(2015a)。 使用基质细胞培养的人树突细胞祖细胞的克隆分析 J免疫学方法 425:21-26。
  5. Lee,J.,Breton,G.,Oliveira,T.Y.,Zhou,Y.J.,Aljoufi,A.,Puhr,S.,Cameron,M.J。,Sekaly,R.P.,Nussenzweig,M.C.and Liu,K.(2015b)。 限制人脐带血和骨髓中的树突状细胞和单核祖细胞 J Exp Med 212(3):385-399。
  6. Paul,F.,Arkin,Y.,Giladi,A.,Jaitin,DA,Kenigsberg,E.,Keren-Shaul,H.,Winter,D.,Lara-Astiaso,D.,Gury,M.,Weiner, A.,David,E.,Cohen,N.,Lauridsen,FKB,Haas,S.,Schlitzer,A.,Mildner,A.,Ginhoux,F.,Jung,S.,Trumpp,A.,Porse,BT ,Tanay,A。和Amit,I.(2016)。 骨髓祖细胞中的转录异质性和谱系承诺 Cell 164(1-2):325。
  7. 参见P.,Dutertre,CA,Chen,J.,Gunther,P.,McGovern,N.,Irac,SE,Gunawan,M.,Beyer,M.,Handler,K.,Duan,K.,Sumatoh, HRB,Ruffin,N.,Jouve,M.,Gea-Mallorqui,E.,Hennekam,RCM,Lim,T.,Yip,CC,Wen,M.,Malleret,B.,Low,I.,Shadan,NB ,Fen,CFS,Tay,A.,Lum,J.,Zolezzi,F.,Larbi,A.,Poidinger,M.,Chan,JKY,Chen,Q.,Renia,L.,Haniffa,M.,Benaroch ,P.,Schlitzer,A.,Schultze,JL,Newell,EW和Ginhoux,F。(2017)。 通过整合高维技术来映射人类DC系。 科学 356(6342)。
  8. Van Der Maaten,L。(2014)。 使用基于树的算法加速t-sne。 J Mach Learn Res 15:3221-3245。
  • English
  • 中文翻译
免责声明 × 为了向广大用户提供经翻译的内容,www.bio-protocol.org 采用人工翻译与计算机翻译结合的技术翻译了本文章。基于计算机的翻译质量再高,也不及 100% 的人工翻译的质量。为此,我们始终建议用户参考原始英文版本。 Bio-protocol., LLC对翻译版本的准确性不承担任何责任。
Copyright: © 2018 The Authors; exclusive licensee Bio-protocol LLC.
引用:Luh, T., Lucero, K., Ma, W., Lee, J., Zhou, Y. J., Shen, Y. and Liu, K. (2018). High Dimensional Functionomic Analysis of Human Hematopoietic Stem and Progenitor Cells at a Single Cell Level. Bio-protocol 8(10): e2851. DOI: 10.21769/BioProtoc.2851.