|Home | About | Journals | Submit | Contact Us | Français|
In vivo imaging of transplanted hematopoietic stem and progenitor cells (HSPCs) was developed to investigate the relationship between HSPCs and components of their microenvironment in the bone marrow. In particular, it allows a direct observation of the behavior of hematopoietic cells during the first few days after transplantation, when the critical events in homing and early engraftment are occurring. By directly imaging these events in living animals, this method permits a detailed assessment of functions previously evaluated by crude assessments of cell counts (homing) or after prolonged periods (engraftment). This protocol offers a new means of investigating the role of cell-intrinsic and cell-extrinsic molecular regulators of hematopoiesis during the early stages of transplantation, and it is the first to allow the study of cell-cell interactions within the bone marrow in three dimensions and in real time. In this paper, we describe how to isolate, label and inject HSPCs, as well as how to perform calvarium intravital microscopy and analyze the resulting images. A typical experiment can be performed and analyzed in ~1 week.
Hematopoietic stem cells (HSCs) are responsible for the maintenance of blood and immune cell turnover, both in physiological conditions and in response to injury. They accomplish this because of a tight balance between quiescence, self-renewal and differentiation. The mechanisms regulating HSC fate are a combination of cell-intrinsic and cell-extrinsic molecular signals, most of which are still unknown. In particular, the molecular and cellular components of the HSC microenvironment, or niche, are objects of intense study and the origin of numerous controversies. Knockout and transgenic mouse models with altered osteoblast numbers and function indicate that these bone-making cells are a major component of the HSC niche1–4; however, several other bone marrow stromal cells have been observed in the vicinity of HSCs, and the full nature of the HSC niche remains elusive5–11.
In vivo imaging of mouse calvarium was first performed by von Andrian and co-workers12, using fluorescence microscopy to detect various hematopoietic cell populations rolling within the bone marrow microvasculature. Following the same principles, we used confocal microscopy to observe homing of a leukemia cell line in proximity to calvarium bone marrow vasculature expressing high amounts of stromal cell-derived factor-1 (SDF-1)13. We then combined confocal and two-photon microscopy to simultaneously observe up to five different cellular and extracellular components in the same area (bone collagen was observed through second harmonic generation (SHG)14; osteoblasts through lineage-specific EGFP expression15; hematopoietic stem and progenitor cell (HSPC) populations through ex vivo DiD (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine perchlorate) labeling; autofluorescent cells through confocal microscopy; and vasculature through injection of nontargeted quantum dots)16,17. We have been using calvarial sutures as reliable and consistent spatial reference points, allowing us to scan the entire bone marrow area without bias and, if needed, to revisit specific areas of interest multiple times16.
The methodology described here is widely applicable to studying the effect of genetic or drug-induced alterations in either HSPCs or bone marrow microenvironment components during the early stages of HSPC transplantation. We have shown that Gαs-null bone marrow cells fail to engraft in transplant recipients because they are unable to exit circulation and localize in the bone marrow space18. Colmone et al.19 studied aberrant homing of human HSPCs injected into recipient mice burdened with leukemia. We describe here how to observe the spatial relationship between DiD-labeled HSPC and EGFP-expressing osteoblasts in col2.3-EGFP transplant recipient mice15; however, virtually any bone marrow cellular or extracellular component can be visualized by confocal/multiphoton intravital microscopy, provided that it is selectively labeled with sufficient intensity.
Numerous approaches have been developed to study HSPC localization within the bone marrow, starting from bone sectioning and the use of immunofluorescent techniques4,8,20. Confocal/two-photon intravital microscopy is the only one allowing single-cell resolution imaging in live animals16. A number of methodologies have been developed that allow a direct observation of femur bone marrow. Xie et al.21 used ex vivo femur epiphysis cultures and observed HSPC dynamics in the proximity of the bone resection while the tissue was still viable. Solutions allowing in vivo imaging of femur bone marrow include the drilling of the femoral cortex to a thickness penetrable by light22 or the use of endoscopic probes inserted in the knee area and moved toward the head of the femur23. Calvarium bone marrow imaging is minimally invasive, and even though skull and femur develop from distinct embryonic progenitor cells and through different developmental processes, in both bone types, HSPCs are present at similar frequencies and are functionally identical16.
The primary limitation of confocal/multiphoton intravital microscopy is the penetration depth, which is currently ~200 µm. Calvarium bone marrow cavities are relatively small; however, we can only visualize approximately the upper half of the bone marrow space. We assume that the deeper half of calvarium bone marrow is structurally and functionally identical to the upper half.
We have proposed that observations made in the calvarium bone marrow might apply to the trabecular area of long bones; however, one has to be aware that the microenvironment of HSPCs located within long bone diaphyses is likely to be very different, for instance, because of the possibility that it may be located at a much greater distance from any endosteal surface. In general, it is impossible to extrapolate observations from calvarium bone marrow structures to other bone marrow areas.
Another limitation of the current imaging methodology is that, despite the calvarial sutures providing spatial reference points, tracking cell positions over long periods of time is inefficient, as the development of scar tissue between imaging sessions impairs further observation of sometimes wide areas. Moreover, little is known about the effects of irradiation, transplantation and even imaging itself on the calvarium bone and stroma, and further studies are necessary to develop highly reliable tracking methodology. Finally, the proliferative characteristics of injected HSPCs dictate the length of time for which they can be observed. Labeled HSPCs injected in non-irradiated recipients do not proliferate and can be detected weeks after transplantation16, whereas the same cells injected in irradiated recipients undergo cell division and dilute the label to undetectable levels within a few days (C.L.C. and C.P.L., unpublished observations).
Before starting an in vivo imaging project, compliance with the relevant local guidelines and regulations for the use of vertebrate species needs to be ensured. Calvarium intravital microscopy experiments are generally classified as generating moderate discomfort and pain to the animals.
Donor and recipient mice need to be syngenic to ensure long-term HSPC engraftment. To observe functional, engrafting HSPCs following transplantation, the recipient mice need to be preconditioned in order to destroy resident hematopoietic cells and make the bone marrow niche permissive to engraftment. In our hands, the most effective and reliable method is a lethal dose of γ-irradiation. We used 9.5 Gy in a single dose or split between two doses of 4.75 Gy 3 h apart for C57/Black6 (C57/B6) mice; however, the most appropriate treatment may vary depending on the mouse strain used and the irradiator available. We therefore recommend titrating the optimal irradiation dose (9–11 Gy) and protocol by monitoring long-term peripheral blood chimerism in a pilot bone marrow transplantation experiment.
If an irradiator is not available, similar engraftment results can be obtained by conditioning the recipient mice with chemotherapeutic agents24.
In principle, any bone marrow HSPC population can be analyzed with in vivo imaging. We describe here the harvesting and preparation of long-term repopulating HSPCs (LT-HSPCs) such as LKS (that is, Lineagelow, c-Kit+ and Sca-1+) CD34−Flk2− or LKS CD48−CD150+; refs. 17,25). In our experience, transplantation of ~10,000 LT-HSPCs leads to observation of ~10 cells homed to the calvarium bone marrow, and this amount of LT-HSPCs can be obtained from four adult C57/B6 donor mice. In our work, lineage depletion with magnetic columns has been highly reliable, although Dynabeads (Invitrogen) provide a reliable alternative.
Several combinations of cell surface markers can be used to identify and isolate HSPC sub-populations. In our study, reliable in vivo data for LT-HSPCs were obtained when sorting LKS CD34−Flk2− or LKS CD48−CD150+ populations17,25. In principle, any other hematopoietic population can be used for in vivo imaging experiments and most fluorescence-activated cell sorting (FACS) instruments allow collection of multiple cell populations at the same time (multiway sorting). Similar to the imaging of multiple parameters, the cocktails of antibodies used for sorting need to be conjugated to fluorophores with appropriate excitation and emission spectra, in order to be easily distinguished. Examples of such antibodies and fluorophores that can be used to identify LT-HSPCs are provided in Table 1. Briefly, whole bone marrow cells are first labeled with a cocktail of biotin-conjugated antibodies against differentiation markers (see Table 1 ‘Lineage cocktail’); this is followed by incubation with streptavidin-coated magnetic beads. When flowing through columns placed inside a magnet, differentiated cells (labeled by the antibody cocktail) are retained in the column, whereas undifferentiated or little-differentiated cells flow through. We refer to the eluted cells as lineage-depleted bone marrow. In a second step, lineage-depleted cells are labeled with antibodies, allowing the identification of HSCs (see Table 1 ‘FACS’); together with fluorophore-conjugated streptavidin, this allows the elimination of the remaining Lineage-positive (i.e., differentiated) cells. When labeled cells flow through the FACS instrument, initial gates are drawn on the basis of the size and granularity of the cells to eliminate debris and cell doublets. A subsequent gate containing c-Kit–bright, Lineage-dim cells separates the Lineagelow cell population. Within this population, c-Kit+ Sca1+ cells are the so-called LKS population. Long-term repopulating HSCs are subsequently identified as CD34−Flk2− or CD150+CD48− cells. Further subgates can be used to obtain purer populations; however, in our study, they are unlikely to provide sufficient numbers of cells for in vivo imaging purposes.
For each sorting, it is essential to prepare not only the cell suspension stained with the cocktail of chosen antibodies but also a series of controls (compensation controls or single-color controls) necessary to arrange the correct compensation settings. To prepare compensation controls, some total bone marrow cells are aliquotted and stained individually with each fluorophore-conjugated antibody used in the FACS cocktail, and some Lineage-depleted cells are stained with fluorophore-conjugated streptavidin. Lineage-depleted cells are used as a single-color Lineage control because whole bone marrow cells would produce a brighter signal than the sorted sample, leading to unnecessary compensation. One further aliquot is left unstained. The unstained sample is used to set up the sorter voltage settings and each single-color control is run to set up the appropriate compensation settings (mathematical algorithms that eliminate ‘bleed through’ signal of one channel into another, so that a bright signal in one channel is not erroneously collected as a signal in a different channel)26. Further information on setting up the sorter voltage settings, compensations settings and carrying out the cell sorting can be found in refs25–27.
It is good practice to check the purity of the obtained cell population by occasionally analyzing some of the obtained cells (at least once per each sorter used). This should be done by running some of the obtained cells through the cell sorter to check what percentage of them falls again within the gates used for sorting. If they all do, then sorting purity is 100%. Because of the small numbers of LT-HSCs typically obtained, it may not possible to perform such analysis after every sort.
Particular care must be taken when choosing the label for imaging cells. Fluorescent signals need to be very intense in order to be detected efficiently through bone. DiD and other similar lipophilic dyes (e.g., Invitrogen Vybrant labeling solutions) are nontoxic to mouse HSPCs and provide uniform bright labeling16. They were initially developed to visualize membrane dynamics; depending on the cell type and the staining protocol used, it has been reported that the membrane distribution of the dyes can vary28. The labeling needs to be optimized for each cell type of interest in order to select the most efficient and least toxic reagent and protocol. Carboxyfluorescein succinimidyl ester (CFSE), PKH and quantum dot-based dyes are valid alternatives to the lipophilic dyes29.
Transgenic reporter strains are extremely valuable to highlight specific bone marrow compartments or hematopoietic lineages30,31. Strong promoters should be chosen to ensure efficient imaging of fluorescent reporters expressed at the highest levels. Lower fluorescence might still permit some detection, however, albeit with a high risk of missing part of the cells or structures of interest (C.L.C., unpublished observation).
We have visualized bone marrow vasculature by injecting Qtracker 800 (nontargeted quantum dots 800, Invitrogen). Depending on each specific experimental setup, other circulating fluorophores (e.g., Qtracker 655, Angiosense (VisEn) or fluorescently labeled dextran)12,16,32 can be valid alternatives.
Our typical setup allows the observation of (i) bone through the SHG signal of collagen14, one of the main components of calcified bone; (ii) osteoblasts by means of high EGFP expression characterizing the col2.3-EGFP reporter strain15; (iii) transplanted cells by brightly labeling them with the lipophilic dye DiD; and (iv) vasculature by injecting nontargeted quantum dots 800 (ref. 16). As all fluorophores are distributed along the electromagnetic spectrum, both for their absorption and emission, a window is left to analyze autofluorescent signal using the 532-nm laser and a 560- to 640-nm emission filter. Autofluorescent cells are present in high numbers in the bone marrow; however, their identity is unknown. They produce signals of identical shape and similar intensity in both the DiD and the ‘autofluorescence’ channel, and sometimes also in the EGFP channel. Because they appear in all channels, whereas labeling fluorophores have a distinctive excitation-emission spectrum, autofluorescent cells are easily identified and excluded from further analysis. We work with ×25 or ×30 water immersion objectives, using the combination of lasers and filters described in Table 2.
In vivo imaging experiments usually require some troubleshooting, and therefore, we recommend imaging no more than one or two mice in each experiment, at least initially. Mouse and microscope setup may require careful and detailed optimization, often for each imaging session, and results might not be comparable if the last mouse is imaged many hours (e.g., >8 h) after the first one. It is possible to delay the injection of cells; however, keeping primary hematopoietic cells ex vivo can be detrimental to their function and viability. We have observed highly reproducible results when injecting the same cell population in experimental replicates on separate days; therefore, we recommend obtaining statistically significant data through smaller repeated experiments rather than larger, less consistent ones.
Several software packages can be used for the final image analysis, from the open source ImageJ, to Adobe Photoshop, to more complex 3D-oriented packages. To date, we have been unable to identify a software package that allows full automation of the process while generating reliable positional data. We recommend working with ImageJ, as it is a fully open-source software platform, therefore ensuring complete control over image processing and avoiding any risk of artifact generation.
Generally, high-quality images are the easiest to analyze; however, it is not always possible to acquire such images as a result of fluctuations in microscope performance and because of the nature of the tissue surrounding the cells of interest. Moreover, better-quality images inevitably require longer exposure of the tissue to the laser, thus increasing the likelihood of damaging the tissue. It is therefore important to establish an efficient imaging routine, compromising between speed of acquisition and image quality, so that the necessary information can be obtained while causing as little damage as possible to the tissue.
The intensity of the signals collected through intravital microscopy depends not only on the brightness of the label or fluorophore used but also on the composition of the surrounding tissue. For this reason, we have not taken into account discrepancies in signal levels and we have limited our analysis to the identification of cells and structures with signal above noise levels.
Different fluorophores are detected with varying efficiency. If you are injecting multiple cell populations, we recommend performing dye-swap control experiments to ensure that similar numbers of cells are observed with each dye; in particular, ensure that different labeling protocols do not affect cell migration and position. For example, if cell population A is labeled with fluorophore 1 and cell population B is labeled with fluorophore 2, the experiment should be repeated by labeling cell population A with fluorophore 2 and cell population B with fluorophore 1.
It is also ideal to image test and control cells or recipient mice within the same experiment, for example, by purifying test and control cells and injecting them in littermate recipients or by splitting the same cell population between test and control recipient mouse. Even if a large amount of cells can be obtained, we recommend imaging one test and one control condition each day and repeating the experimental setup rather than imaging three test recipient mice on one day and three control mice on a separate day. We have not observed differences in HSPC localization when imaging mice within few hours from transplantation; however, we recommend swapping the order of recipient mice during repeats (e.g., during the first experiment, the test mouse is imaged first and the control mouse second, but during the second experiment, the control mouse is imaged first and the test mouse second).
Finally, for each cell imaged, it is important to validate the signal of the label against autofluorescence in order to avoid further analysis of false-positive events. Imaging one or two mice that have not been injected with any cell or label can be helpful to become familiar with bone and autofluorescent signal, especially if low fluorescence is expected.
All reagents need to be sterile.
Harvest medium is composed of PBS with 2% (vol/vol) FCS. Prepare at least 500 ml. It can be prepared in advance and stored at 4 °C for a few days.
Sorting collection medium is made up of PBS with 10% (vol/vol) FCS. Prepare 1–2 ml. It can be prepared in advance and stored at 4 °C for a few days.
The cocktail is composed of biotin-conjugated Ter119, Gr1, B220, Mac1, CD3, CD4 and CD8 antibodies (see Table 1 for details). Mix these in a 1:1:1:1:1:1:1 ratio. It can be prepared in advance and stored in aliquots of 0.5–1 ml at 4 °C until needed.
Prepare ~40 ml of PBS in a 50-ml Falcon tube, connect it to a Steriflip filter and keep it under vacuum for a few minutes without filtering it.
This reagent is best prepared fresh just before use. Eliminating gas from all solutions applied to the columns increases the efficiency of the purification step.
To a bottle containing 10 ml ketamine HCl (50 mg ml−1; 500 mg), add 750 µl xylazine (100 mg ml−1; 75 mg). Shake well. Store at room temperature (20–25 °C) in the dark for up to 3 months.
Several cell sorters are available, and it is important to optimize the antibody staining and fluorophore combination according to the advice of the available FACS instrument. For each sort, unstained and single-color controls must be prepared for compensation purposes, using the same cell type (e.g., bone marrow cells) but only a single antibody (at the same concentration used in the cocktail staining) in each tube (see Experimental design).
A schematic representation of our custom-made mouse imaging setup is presented in Figure 1. The in vivo microscope needs to be equipped with a heated mouse holder, either custom made or commercially available (e.g., Stoelting homeothermic blanket system, cat. no. 50300V). A 3D electronically controlled stage is helpful for tracking x-y positions and for z-stack acquisition. Immersion objectives (for example ×25 or ×30) with 0.9 numerical aperture are ideal for imaging through physiological solution or aqueous medium. We use water immersion objectives and image through a cover slip separating the glycerol solution from the water. Alternatively, water dipping objectives with similar magnification and numerical aperture can be used without cover slips and directly positioned on top of the saline-covered calvarium. Non-descanned detectors maximize the efficiency of two-photon microscopy signal acquisition and improve collection of low signals.
Table 2 summarizes our preferred combination of lasers and filters. Lasers and filter sets can differ from the ones listed and should be chosen on the basis of the fluorophores used for each experiment. It is important to minimize overlap between the collected signals. A pulsed femtosecond laser is crucial to detect collagen by SHG signal.
When repeatedly injecting similar numbers of HSPCs, similar numbers of cells should be detected in the calvarium bone marrow. We typically observe 10–15 events when injecting ~10,000 LT-HSPCs (LKS CD34−Flk2− or LKS CD48−CD150+), with a detection limit set at ~5,000 LT-HSPCs injected. These numbers vary for different cell populations, possibly because of their different ability to localize in the bone marrow space.
Although some variability is observed between independent recipient mice, we observed consistent trends throughout our experiments, and usually no statistically significant differences are observed when the same HSPC populations are injected in equivalent recipient mice. As differences in bone marrow localization of distinct HSPC populations can be subtle, data from cells observed in multiple recipients can be pooled in order to statistically compare different cell types. In any case, the average distance from any HSPC population to osteoblasts or bone has a very wide standard deviation. Moreover, each measurement is subject to errors. Accordingly, we recommend not to focus on the exact distance (in µm) between specific cell types; rather, we suggest careful analysis of the shapes and shifts of the distributions. Even though specific localization of LT-HSPCs relative to osteoblasts may vary between experiments, the facts remain that most LT-HSPCs are most likely to be located in the proximity of osteoblasts, that the likelihood of identifying LT-HSPCs rapidly decreases when moving further from osteoblasts and that only a few LT-HSPCs are directly adjacent to osteoblasts.
We thank F. Ferraro, S. Lane, S. Lymperi, E. Ozcivici, A. Sanchez-Aguilera and M. Spitaler for critical input on the manuscript. C.L.C. was funded by the European Molecular Biology Organization and Human Frontier Science Program and is currently funded by Imperial College London, Kay Kendall Leukaemia Foundation and Cancer Research UK. C.P.L. and D.T.S. are funded by multiple NIH grants.
AUTHOR CONTRIBUTIONS C.P.L. developed the confocal/multiphoton imaging system for calvarium bone marrow intravital microscopy. C.L.C. developed, performed and undertook troubleshooting of the described protocol. D.T.S. provided guidance and critical input throughout the development of the methodology.
COMPETING FINANCIAL INTERESTS The authors declare competing financial interests (see the HTML version of this article for details).