The iFAST3D imaging protocol allows high-resolution imaging of HSCs and their niche components within intact mouse tissues while preserving their spatial organization (Saçma et al., 2022). Briefly, the protocol starts with sample preparation, where femurs, tibiae or humeri were harvested from mice and fixed in paraformaldehyde to maintain tissue integrity.

The bones were shaved by a cryotome until the BM was fully exposed, ensuring optimal antibody penetration. Immunofluorescence staining followed, involving permeabilization for antibody access, blocking to prevent nonspecific binding and incubation with primary fluorophore-conjugated antibodies targeting HSC markers such as CD150 and CD48, as well as niche components like sinusoids and arterioles. If applicable, secondary antibodies were used for signal amplification. Imaging was performed using confocal laser scanning microscopy for capturing z-stack images to assess the information of the morphology and 3D organization of HSCs and their niches. Data analysis involved image processing to reconstruct 3D visualizations, followed by quantification of HSC size, shape and spatial positioning relative to niche structures, enabling statistical assessment of niche-dependent HSC characteristics. Cell fixation that is part of iFAST3D can affect the cell size, morphology and position, which might affect the correct quantiative determination of parameters. We used an even further optimized iFAST3D protocol from Saçma et al., 2022 that is designed to even better preserve cell morphology and tissue architecture, by minimizing the use of chemical solvents, incubation time and mechanical stress.

To assess the size and polarity of FACS isolated single HSCs, specifically focusing on Tubulin and Cdc42, we employed a protocol adapted from the recent published methodology CellDetail (Schuster et al., 2024). Briefly, this approach involves immunofluorescence staining of FACS-isolated and fixed HSCs, followed by epifluorescence or confocal microscopy to capture subcellular high-resolution images. Next, the spatial distribution of Tubulin and Cdc42 within the cells was analyzed using the specialized image analysis software CellDetail, which calculates polarity values based on fluorescence intensity distributions. Further, this method allowed for precise measurement of cell size and the valuation of polarity status in both young and aged isolated HSCs.

The absolute size of individual HSPC populations from young and aged mice was determined by flow cytometry using forward scatter (FSC) measurements calibrated with reference size beads. FSC-W histograms were created using 7 µm, 10 µm, and 16 µm reference beads from the SPHEROTM Flow Cytometry Particle Size Standard Kit.

A calibration plot was constructed to establish a standard curve with FSC measurements proportional to the bead diameter and a significant R² value, using the formula for cell size calculation. Bead-referenced average cell sizes were determined for LT-HSCs, ST-HSCs, and LMPPs isolated from both endosteal and central BM of young and aged mice.

Using 3D confocal microscopy, we first confirmed that HSCs (LSK, CD34^-, Flk2^-) from BM of aged mice (>80 weeks old) were not only on average larger (increase in volume) than HSCs from young mice (8–20 weeks old), but aged HSCs also showed on average a larger (increase in volume) nucleus (Supplementary Figure S1A,B), as previously published (Grigoryan et al., 2018). To determine the size of HSCs in distinct niches in BM of long bones of C57BL/6 mice, we next used the iFAST3D method for IF staining of HSCs within BM¹¹ (Supplementary Figure S1C,D). In young as well as in aged mice, HSCs (CD150⁺CD41^-CD48⁻LIN^-) that were located in the central BM were on average smaller compared to HSCs located in endosteal niches (Figures 1A,B). The smaller HSCs in the central BM were found preferentially adjacent to central sinusoids and megakaryocytes, whereas larger HSCs at the endosteum were in most cases positioned adjacent endosteal arterioles (Figures 1C,D; Supplementary Figures S1C,S1D). HSCs size is therefore distinct among HSCs from distinct regions and niches in BM. In aged mice, the relative difference in size of HSCs between the two niches was maintained, while old HSCs were on average larger than young HSCs (Figures 1A–D). Interestingly, young periarteriolar HSCs in endosteal BM were though on average larger than aged periarteriolar HSCs in central BM (Figure 1D). Notably, the proportions of HSCs adjacent to sinusoids and megakaryocytes in the central BM (cBM) were significantly higher than those in the endosteal BM (eBM) in both young and aged mice (HSCs adjacent to sinusoids: young cBM 85%, young eBM 15%, old cBM 86%, old eBM 14%; HSCs adjacent to megakaryocytes: young cBM 76%, young eBM 24%, old cBM 71%, old eBM 29%). In contrast, the proportion of HSCs adjacent to arterioles was higher in eBM than in cBM only in young mice (HSCs adjacent to arterioles: young cBM 37%, young eBM 63%, old cBM 67%, old eBM 33%, Supplementary Figure S1E). These results suggest a shift, in which fewer HSCs remain in the eBM in aged mice, likely due to age-related remodeling of BM niches, affecting either accessibility or function. (Saçma et al., 2019; Maryanovich et al., 2018; Ho and Toro, 2019). The absolute size of distinct HSPCs populations from distinct BM niches from young and aged mice was determined by flow cytometry (forward scatter, FSC) calibrated with reference size beads (Supplementary Figures S2A,S2B). HSCs from central bone marrow (sinusoidal or megakaryocytic niches) were also with respect to their absolute size smaller than HSCs located at endosteal niches, while in general the absolute size of HSCs from aged mice was increased in both types of niches when compared to young HSCs with the respective niche localization (Supplementary Figure S2C). Also in absolute size, young HSCs from endostal niches were larger than aged HSCs from central niches. Interestingly, short-term repopulating HSCs (ST-HSCs), and especially lympho-myeloid primed progenitor cells (LMPPs), which are only slightly more differentiated compared to HSCs, show already a strong increase in overall cell size (young and aged) compared to their respective HSCs (Supplementary Figure S2C).

Similarly, when differentiating between small, medium and large cells based on forward scatter (Supplementary Figure S2D), large LMPP and large LSK cells were both significantly larger compared to HSCs, irrespective of age or niche position (Supplementary Figure S2E). Progenitors cells show on average a high turnover and are thus frequently actively cycling (Rossi et al., 2007), and active cells may need to maintain a large cell size to allow for cell cycle progression and subsequent division (Conlon and Raff, 1999), which might explain the relatively large size of, for example, LMPPs. Finally, HSCs outside their native microenvironment like in our flow experiments have a smaller cell size compared to HSCs in their native microenvironment in BM in situ (Figure 1B; Supplementary Figure S2C).

To compare the stem cell and multilineage reconstitution potential of HSCs that are located in the 2 distinct BM regions and which are therefore distinct in their average size, we competitively transplanted 200 LT-HSCs from either central (smaller HSCs) or endosteal (larger HSCs) BM alongside 3 × 10⁵ BoyJ competitor BM cells into sublethally irradiated BoyJ mice (Figure 2A). Endosteal HSCs from young mice (larger) showed a bias towards B-lymphoid differentiation, whereas HSCs from central BM (smaller) of young mice were more myeloid-biased (Figure 2B). The on average larger size of HSCs located at the endosteum (Figure 1D) is thus linked to a preferential lymphoid reconstitution potential, and the on average smaller sized central BM HSCs are linked to preferential myeloid-biased reconstitution. These data support previous suggestions that lymphoid-biased HSCs are preferentially located in periarteriolar niches at the endosteum and myeloid-biased HSCs close to perisinusoidal MKs in the central BM (Pinho et al., 2018). In contrast, HSCs from old mice from both the endosteal (larger size) and the central part (smaller size) of the BM show a similar level of engraftment and multilineage reconstitution (Figure 2C). As aged HSCs located at the endosteum are on average larger than HSCs located in the central BM, differences in cell size among aged HSCs in the endosteal and central regions do not correlate with differences in potential. Interestingly, label-retaining HSCs (Figure 2D), which represent the most quiescent HSC subpopulation with the highest regenerative potential in old mice and are thus functionally more young-like (Saçma et al., 2019), exhibit a cell size similar to that of all old HSCs (Figure 2E). These findings suggest that size alone is insufficient to predict HSC function in aged mice.

In general, among aged HSCs there is a significantly reduced frequency of HSCs polar for distinct types of polarity markers. This loss of cell polarity is tighlty linked to the aging associated dysfunction of aged HSCs (Florian et al., 2012). We thus investigated the extent to which the frequency of aged polar HSCs is distinct among central or endosteal HSCs. Single-cell IF analyses of young and old central and endosteal LT-HSCs (Figures 3A,B) revealed that young central HSCs are more polar (cells with a higher dipole moment) for the established polarity markers tubulin and Cdc42 compared to endosteal young HSCs, and especially to aged HSCs (Figures 3C,D). Among aged endosteal or central HSCs, there was no difference in the frequency of polar HSCs. The level of polarity does correlate better than the cell size to the potential of HSCs determind in the transplantation experiments (Figures 2A–C).