Close this search box.

Resilient anatomy and local plasticity of naive and stress haematopoiesis – Nature

Data reporting

No statistical methods were used to predetermine sample size. This is because image analyses are extremely time consuming, and it is not possible to examine large numbers of samples. We have previously shown that three bones per condition allow identification of sufficient number of cells to detet changes in location and distribution in the bone marrow1. We have thus strived to analyse three bones per condition and included additional bones when possible. All mice were included in the analyses. Mice were randomly allocated to different groups based on the cage, genotype and litter size. For all experiments, we aimed to have the same number of mice in the control and experimental groups. Investigators were not blinded to allocation during experiments and outcome assessment. This is because it was not possible to blind the investigator to the type of bone examined as there are readily identified by shape. Similarly, the insults used generated such evident changes in cellular content (haemorrhage, G-CSF, infection) or shape of the bone (ageing, bones are larger) that it was not possible to blind the investigator to the type of insult examined.


All mouse experiments—except live mouse imaging experiments—were approved by the Institutional Animal Care Committee of Cincinnati Children’s Hospital Medical Center. Live mouse imaging experiments were performed in compliance with institutional guidelines and approved by the Subcommittee on Research Animal Care (SRAC) at Massachusetts General Hospital. The following mouse strains were used: C57BL/6J-Ptprcb (CD45.2), B6.SJL-PtprcaPepcb/BoyJ (CD45.1), B6.Cg-Ndor1Tg(UBC-cre/ERT2)1Ejb/1J (Ubc-creERT2), C57BL/6-Tg(CAG-EGFP)131Osb/LeySopJ (actin-GFP) and B6.129P2-Gt(ROSA)26Sortm1(CAG-Brainbow2.1)Cle (R26R-Confetti). R26R-Confetti mice were crossed with Ubc-creERT2 mice to generate Ubc-creERT2:Confetti mice. All mice were maintained on a C57BL/6 J background. Eight to twelve (2-month-old) and 80 to 100 weeks (20-month-old) male and female mice were used. All mice were bred and aged in our vivarium or purchased from the Jackson Laboratory. Mice were maintained at the vivarium at Cincinnati Children’s Hospital Medical Center under a 14-h light:10-h dark schedule, 30–70% humidity, 22.2 ± 1.1 °C, and specific-pathogen-free conditions.

Tamoxifen treatment

Ubc-creERT2:Confetti mice were treated with two pulses of tamoxifen in the diet (400 mg of tamoxifen citrate per kg of rodent diet, Envigo). Each pulse was two weeks long and pulses were two weeks apart. Since committed haematopoietic progenitors do not persist in vivo for longer than two weeks we chased the mice for eight weeks to ensure that all Confetti-labelled immature and mature haematopoietic cells originated from upstream progenitors.

L. monocytogenes infection

The wild-type virulent L. monocytogenes strain 10403s was back-diluted from overnight culture for 2 h to early log phase growth (OD600 0.1) in BD Difco brain-heart infusion medium (Thermo Fisher Scientific, 237500) at 37 °C, then washed and diluted in 200 μl sterile saline and injected via the lateral tail vein to mice (1 × 104 colony-forming units (CFU) per mouse). Mice were euthanized for bone marrow analyses on day 6 and 20 after infection.

Phlebotomy mice model

To induce erythropoietic stress by blood loss, isoflurane-anaesthetized mice were phlebotomized (15–20 μl blood per gram of body weight from the retro-orbital venous sinus of mice) with a calibrated heparinized capillary tube. Mice were euthanized for bone marrow analyses on day 2, 8, and 20 after phlebotomy.

G-CSF treatment

Mice received subcutaneous injections of G-CSF (R&D) twice a day at a dose of 150 µg kg−1 for four days. Mice were euthanized for bone marrow analyses 2–3 h after the final morning G-CSF dose at day 5, and 30 days after G-CSF treatment. Mice received subcutaneous injections of 0.1% low endotoxin bovine serum albumin (Sigma) were analysed as control.

Cell preparation for flow cytometry and cell sorting

Mice were anaesthetized with isoflurane followed by cervical dislocation. For long bones, bone marrow cells were flushed out of the femurs with a 21-gauge needle in 1 ml of ice-cold PEB buffer (2 mM EDTA and 0.5% bovine serum albumin in PBS). For sternum, vertebrae, and skull the bones were chopped into small pieces with scissors in 1 ml of ice-cold PEB buffer. Peripheral blood was collected from the retro-orbital venous sinus of mice, followed by red blood cell lyses with 1 ml of lysis buffer (150 mM NH4Cl, 10 mM NaCO3 and 0.1 mM EDTA). Cells were centrifuged for 5 min at 1,100 rpm under 4 °C, resuspended in ice-cold PEB, and used in subsequent assays. For FACS analyses, cells were stained with a cocktail of biotinylated lineage antibodies for 30 min, washed twice, and stained with streptavidin-conjugated magnetic beads (BD Bioscience, 557812). Magnetic cell depletion was performed according to the manufacturer’s protocol. CountBright Absolute Counting Beads (Thermo Fisher Scientific, C36950) were used to count bone marrow and blood cell numbers in a BD LSRFortessa Flow Cytometer (BD Bioscience).

FACS analyses and LEGENDScreen

Cells were analysed in an LSRFortessa Flow Cytometer (BD Biosciences), LSR II Flow Cytometer (BD Biosciences) or FACS-purified in a FACSAria II Cell Sorter (BD Biosciences) or an SH800S Cell Sorter (Sony Biotechnology). Dead cells and doublets were excluded on the basis of FSC and SSC distribution and DAPI exclusion (Sigma-Aldrich, D9542). Antibodies used were: B220 (clone RA3-6B2), CD2 (clone RM2-5) CD3e (clone 145-2C11), CD4 (clone RM4-5), CD5 (clone 53-7.3), CD8 (clone 53-6.7), CD11b (clone M1/70), CD11c (clone N418), CD16/32 (clone 93), CD24 (clone 30-F1), CD31 (clone A20), CD41 (clone MWReg30), CD42d (clone 1C2), CD43 (clone S11), CD45 (clone 30-F11), CD45.1 (clone A20), CD45.2 (clone 104), CD48 (clone HM48-1), CD71 (clone RI7217), CD105 (clone MJ7/18), CD115 (clone AFS98), CD127 (clone A7R34), CD135 (clone A2F10), CD144 (clone BV13), CD150 (clone TC15-12F12.2), ESAM (clone 1G8), Gr1 (clone RB6-8C5), IgD (clone 11-26 c.2a), IgM (clone RMM-1), Ly6C (clone HK1.4), Ly6G (clone 1A8), Sca-1 (clone D7), Ter119 (clone TER-119), MHCII (clone M5/114.15.2), from BioLegend; CD34 (clone RAM34) and CD117 (clone 2B8), from BioLegend or Thermo Fisher Scientific; CD71 (clone C2) from BD Bioscience. For immunophenotyping experiments, LEGENDScreen Mouse PE Kit (BioLegend, 700005) was used according to the manufacturer’s instructions. In brief, fresh bone marrow cells were stained with a cocktail of biotinylated lineage antibodies for 30 min followed by a stain with streptavidin. Cells were washed twice and resuspended at a concentration of 1 × 107 cells per ml PEB buffer containing antibodies for HSPC identification. Equal amount of cells were transferred into each well of the LEGENDScreen 96-well plates. Plates were incubated for 45 min on ice in the dark. Cells were then washed twice and resuspended in PEB buffer and kept on ice until acquisition on a BD LSRFortessa Flow Cytometer (BD Bioscience). FACS data were analysed with FlowJo software (Tree Star). Dilutions used for each antibody were 1:200, except for CD11b, which was used at 1:500. Gating strategies for all analyses are shown in Supplementary Fig. 10 and Supplementary Fig. 11. Antibodies that did not yield specific signals in confocal imaging are listed in Supplementary Table 5.

CFU assay

FACS-purified cells were suspended in IMDM + 2% FBS, added into the methylcellulose culture medium (Stem Cell Technologies, MethoCult M3334, M3434, M3436 and M3534), mixed thoroughly, plated in duplicate 35 mm culture dishes (Greiner Bio-One, 627160), and incubated at 37 °C with 5% CO2 in air and ≥ 95% humidity, for 7–10 days. Colonies were identified and counted based on cluster size and cell morphology using a Nikon Eclipse Ti inverted microscope (Nikon Instruments) equipped with 4×, 10× and 40× objectives.

Extreme limiting dilution assays

Adult CD45.1+ recipient mice were lethally irradiated (700 rad plus 475 rad, 3 h apart). Then 15, 7, 3 or 1 FACS-purified CD45.2+ LT-HSCs or ST-HSCs were mixed with 2 × 105 CD45.1+ competitor mouse bone marrow cells and transplanted by retro-orbital venous sinus into lethally irradiated CD45.1+ recipients within 6 h after the second irradiation. Peripheral blood chimerism was determined by FACS analyses at week 16 post-transplant. HSC frequencies were calculated by using extreme limiting dilution analysis36.

Transplant of ESAM+ and ESAM progenitor subsets in sublethally irradiated recipients

Adult CD45.1+ recipient mice were sublethally conditioned with a single dose of 700 rad. The indicated number of FACS-purified ESAM+ or ESAM HSPCs was transplanted via retro-orbital venous sinus injection within 6 h after irradiation. Peripheral blood chimerism was determined by FACS analyses on day 10, 20, 30 and 40 post-transplant.

For transplants of pre Meg-E subsets, we transferred 2,000 ESAM+ or ESAM pre Meg-E purified from Ubc-GFP mice into CD45.1+ recipient mice sublethally conditioned with a single dose of 700 rads. Peripheral blood chimerism (including platelets and red blood cells) was determined by FACS analyses on day 6, 12 and 18 post-transplant.

Whole-mount immunostaining

In experiments requiring visualization of blood vessels in the absence of ESAM, mice were intravenously injected with 10 μg of Alexa Fluor 647 anti-mouse CD144 antibody (BV13, BioLegend) and euthanized 10 min after injection. In experiments requiring visualization of CLP, mice were intravenously injected with 2 μg of Alexa Fluor 647 anti-mouse CD127 antibody (A7R34, BioLegend) and euthanized 5 min after injection. Whole-mount sternum immunostaining has been described37. In brief, the sterna were dissected and cleaned of soft and connective tissue, followed by sectioning along the sagittal or coronal plane to expose the bone marrow under a dissecting microscope (Nikon SMZ1500 Stereomicroscope). Each half of the sternum was fixed in 4% PFA (Electron Microscopy Sciences, 15710) in DPBS (Thermo Fisher Scientific, 14190144) for 3 h on ice. Each fragment was further washed with DPBS after fixation and blocked with 10% goat serum (Sigma-Aldrich, G9023) for 1 h, followed by staining with 100 µl staining buffer (2% goat serum in DPBS and the indicated antibodies) on ice. For whole-mount analyses of tibia and humerus the bones were cleaned and soft and connective tissue and bisected along the sagittal plane to expose the bone marrow and then processed as the sternum segments above. For whole-mount analyses of the L5 vertebrae we cleaned the soft and connective tissue and removed the spinal cord. With a surgical blade we removed the body of the vertebrae and bisected it to expose the marrow. For the whole-mount analyses of the lambdoid sutures we dissected the top of the skull from the frontal to occipital bones. Then we used a surgical blade to bisect the lambdoid sutures along the transversal plane. The exposed suture was further bisected by cutting along the horizontal plane to expose the bone marrow inside. All bones were then stained as indicated above for the sternum.

Confocal imaging

Confocal imaging was performed in a Nikon A1R GaAsP Inverted Confocal Microscope, Nikon A1R LUN-V Inverted Confocal Microscope, or Nikon AXR Inverted Confocal Microscope. Specifications for the Nikon A1R GaAsP Inverted Confocal Microscope: high-power 405 nm, 442 nm, 488 nm, 561 nm, 640 nm and 730 nm solid-state diode lasers. Specifications for the A1R LUN-V Inverted Confocal Microscope: high-power 405 nm, 445 nm, 488 nm, 514 nm, 561 nm and 647 nm solid-state diode lasers. Specifications for the AXR Inverted Confocal Microscope: high-power 405 nm, 445 nm, 488 nm, 514 nm, 561 nm, 594 nm, 640 nm and 730 nm solid-state diode lasers. All microscopes were equipped with a fully encoded scanning xy motorized stage, piezo-z nosepiece for high-speed z-stack acquisition, resonant and galvanometric scanners, 1 high-quantum efficiency, low-noise Hamamatsu photomultiplier tube, and three high-quantum efficiency gallium arsenide phosphide photomultiplier tubes (GaAsP-PMTs) for overall 400–820 nm detection. An LWD Lambda S 20XC water-immersion objective (Nikon, MRD77200) was used and images were taken using the resonant scanner with 8× line averaging, 1,024 × 1,024 pixels resolution, and 2-μm z-step. For high-power images we used a LWD Lambda S 40XC water-immersion objective (Nikon, MRD77410) with a resonant scanner and 8× line averaging, 1,024 × 1,024 pixels resolution, 0.5-μm z-step.

Image and distance analyses

Original images (.ND2 format file) were denoised by a built-in artificial intelligence algorithm (Denoise.AI) and stitched together using the NIS Elements software (Nikon, version 5.20.02 and 5.30.03). The denoised and stitched ND2 files were converted to Imaris (.IMS) files using Imaris software (Bitplane, version 9.5 to 9.9). Because not all antibodies penetrate to the same depth within the tissue, we only examine the first 35 µm of the sternum image, which we have previously shown are uniformly stained through the tissue1. Cells of interest were labelled with dots with the Imaris Spots function in manual mode and the x, y and z coordinates of dots were automatically computed. Sinusoids, arterioles, and megakaryocytes were segmented based on channels of CD144, CD41, ESAM and Ly6C using the Imaris Surface function. The diameters of each type of cell were measured manually in 3D view in Imaris software and were as follows: CD41 LT-HSC, 8.67 ± 1.23 μm; CD41+ LT-HSC, 8.94 ± 0.91 μm; ST-HSC, 8.68 ± 1.10 μm; MPP2, 7.98 ± 1.05 μm; MPP3, 8.48 ± 1.32 μm; MkP, 14.45 ± 3.88 μm; pre Meg-E, 9.49 ± 1.34 μm; pre CFU-E, 13.92 ± 1.70 μm; CFU-E, 12.67 ± 1.88 μm; early erythroblast, 8.86 ± 1.61 μm; late erythroblasts, 7.92 ± 1.36 μm; reticulocytes, 5.17 ± 0.76 μm; RBC, 4.38 ± 0.60 μm; CLPs, 7.40 ± 0.97 μm; pre-pro B, 8.9 ± 0.61 μm; pro B, 7.71 ± 1.23 μm; pre B, 6.10 ± 0.61 μm; MDP, 12.13 ± 1.19 μm; GP, 11.70 ± 0.99 μm; PN, 10.21 ± 1.08 μm; Ly6Clow Mo, 9.30 ± 1.17 μm; cDC, 12.33 ± 2.69 μm. The distance from each cell to the closest vascular structures and megakaryocytes was obtained with the Imaris Distance Transform Matlab Xtension and then subtracted the mean radius for each cell type. The distance between cells was calculated using Matlab software (MathWorks, version 2018a) with the coordinates exported from Imaris and then subtracted the mean radius for each cell. All software were installed in HP Z4 windows 10 x64 workstations equipped with Dual Intel Xeon processor W-2145, 192GB ECC-RAM, and an Nvidia Quadro RTX 5000 16GB GDDR6 graphics card.

Confetti imaging

For our imaging experiments we used 6 fluorescent channels (405 nm, 445 nm, 488 nm, 514 nm, 561 nm and 647 nm). In the Confetti model, Cre recombination leads to expression of GFP (488 nm), YFP (514 nm), RFP (561 nm) and CFP (445 nm), thus occupying 4 out of 6 channels used for imaging. To overcome this limitation and analyse spatial relationships between Confetti-labelled cells we routinely used a dump channel with Alexa 488 or FITC-labelled antibodies (same fluorescence as GFP). We discarded cells showing green fluorescent from analyses and compared YFP, RFP and CFP labelled cells of interest. To analyse the clonal relationships between CFU-E and erythroblasts, we used Ly6C-Alexa 488, and discarded Ly6C+GFP+ cells from analyses. To analyse the clonal relationships between CLP and B precursors, we used Lin Alexa 488 (the Lin panel contains CD2, CD3e, CD5, CD8, CD11b, Ter119, Ly6G, IgM and IgD), and discarded Lin+GFP+ cells from analyses.

Random simulations

Sternal fragments were stained with anti-CD45 and anti-Ter119 antibodies to detect all haematopoietic cells, with anti-CD144, anti-ESAM, anti-CD41 and anti-Ly6C to detect sinusoids, arterioles, and megakaryocytes. 3D binary segmentation tools in NIS Elements software were used to automatically annotate CD45+ or Ter119+ cells. In brief, high-resolution images (0.31 μm per pixel xy, 0.6 μm per pixel z) acquired with a 40× water-immersion objective (NA 1.15) were deconvolved, and CD45 and Ter119 fluorescent membrane channels were added into a single channel with the floating-point math, converted into 12-bit data, and pre-processed to normalize intensities in-depth and min/max intensities. The ‘3D darkspot detection’ algorithm enables the detection of cells of different sizes. This segmentation algorithm considers the distribution of intensities in x, y and z 3D region watershed dark centroid to bright membrane. This will account for non-spherical cells and include all dark space inside the cell membrane stain. The generated ‘inside cell’ binary data was exported to the Imaris software and used to place dots representing each haematopoietic cell (48,964 to 81,248 cells) in each 35-µm optical slice z-stack of each sternum fragment. We then used Research Randomizer38 to randomly select dots representing each type of haematopoietic cell at the same frequencies found in vivo through the bone marrow cavity and measured the distances between these random cells or with vessels as above. Each random simulation was repeated 100–200 times.

To generate random distributions of cells in experiments using Confetti mice, we first obtained the coordinates and Confetti colour for each type of cell in each section analysed. Then we used Research Randomizer to randomize the Confetti label while maintaining the spatial coordinates of each cell. We then measured the distances between these cells with randomized colours. Each random simulation was repeated 100–200 times.

Production site identification

Production sites for each lineage were identified by comparing the observed distributions of distances with that of random cells as described in each figure.

Microscopy-guided HSC transplantation in the bone marrow of live animals

Microscopy-guided HSC transplants into the skull of living mice have been reported in detail before29. In brief, Tie2+CD150+CD48low/−CD135LinSca1+Kit+ LT-HSCs were purified from actin-GFP mice or stained with Dil. The skull was then exposed, and the vasculature visualized by rhodamine-B,70 kMW dextran injection. Second harmonic generation was used to localize bone marrow cavities. Then laser ablation was used to etch a microwell in the bone, with a small opening (about one cell diameter) at the bottom of the microwell that connects to the bone marrow cavity. The opening of the bone marrow cavity was confirmed by lack of second harmonic generation signal and bone marrow leakage. HSCs were loaded in a straight glass micropipette (28–32 µm diameter, Origio) attached to a pump (SAS11/2-E, Research Instruments). Single (1) or multiple (5) HSCs were slowly released into the optical tweezer one at a time and the trapped cells were guided to the bottom of the microwell under image guidance. For the transplant of 17, 19, and 22 cells, multiple cells were first released into the microwell from the micropipette, and the laser tweezer was used to move the cells down to the bottom of the microwell. After the delivery, imaging was performed every 5 min for up to 15 min to ensure that the cell remained at the delivery site. Subsequent imaging was performed as described29.

Live-imaging analyses of haematopoietic behaviour after cell division

HSC (LinCD117+ScaI+CD48), MPPs (LinCD117+ScaI+CD48+), MDPs, granulocyte progenitors and CFU-Es were purified by FACS and plated in 18-well microplates with liquid medium. Live-cell images were taken using a CIC widefield Nikon Ti2 inverted SpectraX system. Cells were cultured in a Tokai Hit incubation system for 12 h to make sure cells were fully decanted. Live-cell images were taken every 15 min for 36 h. HSC and MPP were cultured in F12 medium supplemented with 10 mM HEPES, 1× penicillin–streptomycin–glutamine (P/S/G), 1× insulin–transferrin–selenium–ethanolamine (ITSX), 1 mg ml−1 polyvinyl alcohol (PVA), 100 ng ml−1 thrombopoietin (TPO), and 10 ng ml−1 stem cell factor (SCF). MDP and granulocyte progenitors were cultured in Iscove’s Modified Dulbecco’s Medium with 25 mM HEPES and l-glutamine containing 10% (vol/vol) FBS, 1 mM sodium pyruvate, penicillin (100 U ml−1) and streptomycin (100 μg ml−1) with a combination of cytokines (50 ng ml−1 SCF, 20 ng ml−1 LIF, 10 ng ml−1 IL-3, 20 ng ml−1 IL-6). CFU-E were cultured in Iscove’s Modified Dulbecco’s Medium with 25 mM HEPES and l-glutamine containing 10% (vol/vol) FBS, 1 mM sodium pyruvate, penicillin (100 U ml−1) and streptomycin (100 μg ml−1) with a combination of cytokines (3.0 U ml−1 recombinant human EPO, 10 ng ml−1 recombinant mouse IL-3, 10 ng ml−1 recombinant mouse IL-6, 25 ng ml−1 recombinant mouse SCF and 50 ng ml−1 recombinant mouse TPO). All cytokines were purchased from Stem Cell Technologies.

Quantifications of vessel length, diameter and branching

Bone marrow vessels were detected based on ESAM and Ly6C expression (sinusoids ESAM+Ly6C, arterioles ESAM+Ly6C+). We defined a branch as the point where two or more lumens connect. A vessel is a vascular structure—with a continuous lumen—between two branching points. Vessel length and diameter were measured manually using the measurement tool in Imaris. Diameter reported was the largest value for the whole vessel.


All statistical analyses were performed using Prism 9 (GraphPad Software). For graphs quantifying cells in different mice, we indicate the mean, and each dot corresponds to one mouse. For graphs showing distances between cells or structures, or quantifying cells in production sites, we indicate the median or mean respectively, and each dot corresponds to one cell or production site as indicated. Statistical analyses between two samples were performed by using Student’s t-test if the data were normally distributed and Mann–Whitney test if the data were not normally distributed. For statistical analysis between multiple samples analyses were performed using two-way ANOVA followed by Sidak’s multiple comparisons test if the data were normally distributed or Kruskal–Wallis test if they were not normally distributed. No statistical methods were used to predetermine sample size.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.