Hematopoietic Stem/Progenitor Cells Express Myoglobin and Neuroglobin: Adaptation to Hypoxia or Prevention from Oxidative Stress?



Oxidative metabolism and redox signaling prove to play a decisional role in controlling adult hematopoietic stem/progenitor cells (HSPCs) biology. However, HSPCs reside in a hypoxic bone marrow microenvironment raising the question of how oxygen metabolism might be ensued. In this study, we provide for the first time novel functional and molecular evidences that human HSPCs express myoglobin (Mb) at level comparable with that of a muscle-derived cell line. Optical spectroscopy and oxymetry enabled to estimate an O2-sensitive heme-containing protein content of approximately 180 ng globin per 106 HSPC and a P50 of approximately 3 µM O2. Noticeably, expression of Mb mainly occurs through a HIF-1-induced alternative transcript (Mb-V/Mb-N = 35 ± 15, p < .01). A search for other Mb-related globins unveiled significant expression of neuroglobin (Ngb) but not of cytoglobin. Confocal microscopy immune detection of Mb in HSPCs strikingly revealed nuclear localization in cell subsets expressing high level of CD34 (nuclear/cytoplasmic Mb ratios 1.40 ± 0.02 vs. 0.85 ± 0.05, p < .01) whereas Ngb was homogeneously distributed in all the HSPC population. Dual-color fluorescence flow cytometry indicated that while the Mb content was homogeneously distributed in all the HSPC subsets that of Ngb was twofold higher in more immature HSPC. Moreover, we show that HSPCs exhibit a hypoxic nitrite reductase activity releasing NO consistent with described noncanonical functions of globins. Our finding extends the notion that Mb and Ngb can be expressed in nonmuscle and non-neural contexts, respectively, and is suggestive of a differential role of Mb in HSPC in controlling oxidative metabolism at different stages of commitment. Stem Cells 2014;32:1267–1277


Adult hematopoietic stem/progenitor cells (HSPCs) constitute a life-span reservoir for continuous production of all the hematic cell lineages. They reside in the bone marrow (BM) with the more primitive/multipotent among them localized adjacent to the endosteum in a tissular osteoblastic microenvironment known as stem cell niche [1]. The niche is characterized by a very low O2 tension, which is thought to delay aging in HSPC protecting them from cumulative oxidative insults and to impose an anaerobic metabolic adaptation suited to the stem cell quiescent state [2, 3]. However, hypoxia proved to be, paradoxically, a condition that promotes reactive oxygen species (ROS) production by a mitochondria-mediated mechanism [4]. Noticeably, isolated HSPCs contain functioning mitochondria [5]. Following appropriate stimuli, the HSPCs are mobilized from the niche and forced to cope with a more oxygenated perivascular milieu, for example, the vascular niche, where they egress from quiescence [6]. This phase preludes to the stepwise commitment of HSPC, which depends on the orchestrated signaling elicited by specific growth/differentiation factors culminating in genomic reprogramming and activation of specialization genes [7]. In addition, a metabolic adaptation is required to face with the increased energy needs [8-10]. The mitochondrial oxidative phosphorylation is the more efficient cellular ATP source and, consistent with this notion, committed stem cells exhibit a higher content of mitochondria as compared with those in the quiescent state [5, 10, 11]. A shared consensus is emerging that distinct metabolic profiles are not simply markers but determinants of the stem cell fate [12].

The limited solubility and diffusion of O2 in the aqueous cellular milieus is compensated by the activity of reversible O2 binders like globins. These are a related family of proteins, all of which share similar primary and tertiary structure. In addition to myoglobin (Mb) and hemoglobin, prototypes of the family, two new globins have been discovered in mammals, cytoglobin (Cygb) and neuroglobin (Ngb) [13]. The former is ubiquitously expressed in almost all the cell types whereas the latter is mainly located in the brain and both are upregulated under hypoxia [14]. The main function of the globins is to store O2 and to facilitate its intracellular diffusion under condition of O2 shortage although a protective action against oxidative stress has been suggested relying on their scavenging ability toward nitric oxide (NO) or other reactive (oxygen) species [15].

Characterization of globins in HSPCs has never been reported in literature. In this study, following serendipic observations, we present, for the first time, unexpected functional and molecular evidences of the presence in human CD34+Lin HSPCs of members of the globin family.

Materials and Methods


Peripheral blood (PB)-HSPCs were obtained from granulocyte colony-stimulating factor (G-CSF) preconditioned healthy donors for allogeneic transplantation after informed consent; BM-resident HSPCs were from sternal BM-blood aspirate from healthy donors. Eight and three PB- and BM-HSPC samples from as many distinct unrelated subjects were used in this study. Anti-CD34 immunoisolation of both PB- and BM-HSPCs was carried out as previously described [16]. The isolated cells were routinely [mt]98% lineage-committed markers negative. After selection, aliquots of cells were freshly used or cryopreserved in liquid nitrogen. Before use, the frozen cell samples were thawed at room temperature and washed twice in RPMI 1640. Cell viability as determined by trypan blue exclusion was typically 80%–95%. Human rhabdomyosarcoma- and neuroblastoma-derived cell lines (RD and Gi-li-n, respectively) and normal human dermal fibroblasts (NHDF) were grown as monolayer, in complete RPMI 1640 (Gi-li-n) or Dulbecco's modified Eagle's medium (RD and NHDF) +10% fetal bovine serum (FBS) + 1% penicillin/streptomycin in a humidified 5% CO2 incubator at 37°C.

Spectrophotometric Analysis

One milliliter of 7 × 107 HSPC lysate in RPMI was supplemented with 200 µg (20 U)/mL glucose oxidase (Sigma) and 4 µg (250 U)/mL catalase (Roche Diagnostics, Penzberg, Germany, http://www.roche-applied-science.com; beef liver) and split in two equal volumes in the spectrophotometric cuvette and in the oxymeter chamber both equipped with a stirring device and thermostated at 37°C. Both suspensions were layered with mineral oil (Sigma) to limit O2 diffusion. The O2 content in both the HSPC lysate suspensions was decreased by adding simultaneously 30 mM glucose. Absolute spectra were recorded at different concentrations of O2 as monitored in parallel by the oxymeter. Appropriate scan and O2 consumption rates were set to minimize the difference in the O2 concentrations between the start and the end of the spectra recordings. The spectrum recorded before addition of glucose was subtracted from those attained at different O2 concentrations to get differential spectra.


The rate of oxygen consumption was measured by high resolution respirometry (Oroboros, oxygraph-2K or Hansatech) with a Clark-type oxygen electrode at 37°C. 6 × 107 viable HSPCs were suspended in RPMI 1640 supplemented with 100 µM diphenylene iodinium (DPI), 0.16 µM purified beef heart cytochrome c oxidase [17], and 16 µM horse cytochrome c (Sigma Aldrich, St. Louis, http://www.sigmaaldrich.com). O2 consumption was initiated by addition of 20 mM ascorbate plus 0.1 mM N,N,N′,N′-tetramethyl-p-phenylenediamine (TMPD). Alternatively to cytochrome c oxidase, 0.5 mg prot/mL of beef heart mitochondria [17] was used as O2 consuming system in the presence of 1 µM antimycin A + 1 µM rotenone.

Quantitative Real Time Reverse Transcriptase Polymerase Chain Reaction

Total RNA was isolated by Trizol reagent (Life Technologies, Paisley, http://www.lifetech.com). One microgram of total RNA was used in a reverse transcription (RT) reaction using the transcriptor first strand cDNA synthesis kit (Roche) according to the manufacturer's instructions. Quantitative real-time polymerase chain reactions (PCR) were performed using Light Cycler 480 II Thermal Cycler (Roche). Quantitect primers for: standard myoglobin (Hs_MB_va.1_SG-QT01004206), alternative myoglobin (HS_MB_vb.1_SG-QT01004213), neuroglobin (HS_NGB_1_SG-QT00044919), and cytoglobin (HS_CYGB_1_SG-QT00016051) were purchased from Qiagen (Basel, Switzerland, http://www1.qiagen.com). The gapdh gene amplification was used as a reference standard to normalize the target signal using the 2ΔΔCt method. The reaction program included preincubation at 95°C for 5 minutes and 45 cycles consisting of denaturation (95°C for 10 seconds), annealing (60°C for 10 seconds), and elongation (72°C for 10 seconds). Melting curves were generated through an additional cycle (95°C for 5 seconds, 65°C for 1 minute, and 97°C continuous).


PBS-washed PB- and BM-HSPCs were suspended in ice-cold lysis buffer (150 mM NaCl, 50 mM Tris-HCl pH 7.4, 0.1% SDS, 1% Triton X-100, 5 mM EDTA, 1% Na-deoxycholate, and protease inhibitor cocktail). Lysates were centrifuged at 12,000 rpm for 15 minutes at 4°C and 40–60 µg of the supernatants were suspended in Laemmli's buffer and run on a 15% SDS-PAGE followed by Western blotting. After the transferring procedure, the membrane was blocked with 10% FBS in TTBS (10 mM Tris-HCl, 150 mM NaCl, 0.1% Tween 20 pH 8) for 1 hour at 37°C and incubated overnight at 4°C with either of anti-Mb (1:100 mouse Mo-Ab from Sigma—N7773 (Sigma Aldrich, St. Louis, http://www.sigmaaldrich.com) or 1:150 rabbit Po-Ab from Santa Cruz Biotechnology, Santa Cruz, CA, http://www.scbt.com—sc-25607), anti-Ngb (1:100 mouse Mo-Ab from BioVendor—RD182043100-C8 (http://www.biovendor.com) or goat Po-Ab from Santa Cruz—sc-22001), anti-GAPDH (1:10,000 mouse mAb from Sigma). The TTBS-washed membrane was incubated for 1 hour at room temperature with 1:8,000 horseradish peroxidase-conjugated secondary antibody (Santa Cruz) and analyzed/visualized by chemiluminescence (SuperSignal West Pico Chemiluminescent Substrate, Thermo Scientific) with the VersaDoc Imaging System (Bio-Rad, Hercules, CA, http://www.bio-rad.com, Quantity One software).

Immunocytochemistry and Laser Scanning Confocal Microscopy Analysis

2 × 105 HSPCs were seeded on poly-lysine-coated glass bottom dishes. Following adhesion, HSPCs were incubated with a 1:100 diluted (mouse) anti-human CD34-fluorescein isothiocyanate (FITC) for 20 minutes in the dark at 4°C; then cells were PBS-washed, fixed in 3.8% paraphormaldeyde (10 minutes), permeabilized with 0.2% Triton X-100 (10 minutes), blocked with 5% FBS in PBS (10 minutes), and incubated with primary anti-Mb (1:100 Mo mouse from Sigma—N7773 or 1:150 Po rabbit from Santa Cruz—sc-25607) or anti-Ngb (1:100 Mo mouse from BioVendor—RD182043100-C8 or Po goat from Santa Cruz—sc-22001) for 1 hour at room temperature followed by 1 hour incubation of 1:100 rhodamine-labeled secondary anti-IgG. Imaging of labeled cells was performed by a Nikon TE 2000 microscope (images collected using a ×60 objective 1.4 NA) coupled to a Radiance 2100 dual laser scanning confocal microscopy system (Bio-Rad). The fluorescent signals emitted by FITC-conjugated Ab (λex, 490 nm; λem, 525 nm) and by Rhodamine-conjugated secondary Abs (λex, 543 nm; λem 572 nm) were from confocal planes of 0.2 µm in thickness examined along the z-axes, going from the bottom to the top of the cells. Acquisition, storage, and analysis of data were made by LaserSharp and LaserPix software from Bio-Rad or by Image J (http://imagej.nih.gov/ij/).

Flow Cytometry and Cell Sorting

For cell analysis 2 × 105 cells were stained with 5 µL of either CD34-phycoerythrin (PE), CD38 PE-Cy-A, CD33 allophycocyanin (APC)-A, CD71 APC-A, CD19 APC-A (Becton Dickinson, Franklin Lakes, NJ, http://www.bd.com) in 100 µL of PBS for 20 minutes at 4°C. Cells were washed in PBS for 3 minutes at 2,000 rpm and then permeabilized and fixed with BD Cytofix-Cytoperm kit (Becton Dickinson) according to the manufacturer's instructions. Primary rabbit antimioglobin antibody (Sigma Aldrich, St. Louis, http://www.sigmaaldrich.com) was incubated 1:50 in Cytoperm solution for 1 hour at 4°C. Cells were then washed and incubated with secondary antibody anti-rabbit Alexa Fluor 488 (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) for 30 minutes at 4°C. Cells were washed and analyzed immediately. Opportune controls were settled in parallel to exclude intrinsic cell autofluorescence or secondary antibody aspecific signals.

Fluorimetric Detection of NO

20 × 106 HSPCs were suspendered in RPMI (without phenol red) supplemented with 10 µM of the NO-probe 4-amino-5-methylamino-2′,7′-difluorofluorescein (DAF-FM, Molecular Probes, Eugene, OR, http://probes.invitrogen.com) diacetate and mitochondria as oxygen consuming system. The suspension was layered with mineral oil and allowed to become anaerobic following the addition of ascorbate plus TMPD (10–15 min incubation). Hence the suspension-containing cuvette was transferred to the spectrofluorimeter (FP-6500, Jasco Analytical Instruments) equipped with a thermostatic control system (T = 37°C) and a stirring device. The instrumental setting was λex = 495 nm, λem = 515 nm, medium gain, in the time-drive mode. After stabilization of a baseline the NO-dependent fluorescence increase was elicited by addition of 100 µM NaNO2.

Statistical Analysis

Two tailed Student's t test was applied to evaluate the significance of differences measured throughout the data sets reported.


HSPCs Contain O2-Releasing/Buffering Systems

Supporting Information Figure S1 shows the absorbance difference spectra (dithionite-reduced minus air-oxidized) of whole HSPC lysate. Spectral features in the visible region indicated the presence of redox-sensitive chromophores partly contributed by mitochondrial cytochromes [5]. However, as dithionite causes also rapid exhaustion of dissolved O2 we performed spectral analysis of HSPCs at different concentrations of O2 (and in the presence of the mitochondrial respiratory chain inhibitor antimycin A) to pinpoint O2-tension-sensitive spectral features. The O2 concentration was gradually reduced by glucose oxidase/glucose and assessed synchronously by oxymetric measurement in twin samples. Figure 1A shows that reducing the O2 concentration in the HSPC-lysate suspension resulted in negative differential peaks at 540 and 578 nm and positive peaks at 557 and 597 nm, which progressively increased with isosbestic points at 523, 552, 567, and 588 nm. These O2-dependent spectral changes were completely reversed following reoxygenation of the cell suspension and entirely abrogated when the HSPC lysate was pretreated with the oxidant ferricyanide (not shown). Plot of the absorbance changes at 578–597 nm as a function of the O2 concentration (Fig. 1B) fitted nicely with an hyperbolic function resulting in an apparent P50 of 3.05 ± 0.29 (n = 5) µM O2 (equivalent to 0.32 kPa, 2.37 mm Hg, 1.5% air saturation).

Figure 1.

HSPCs contain O2-releasing/buffering systems. (A, B): Spectrophotometric analysis of HSPC lysate at different O2 concentrations. (A): Absolute absorbance spectra of lysate from 7 × 107 HSPCs per mL were recorded at different oxygen concentrations (progressively decreased by the glucose/glucose oxidase/catalase system) and corrected for the spectrum initially recorded in fully aerated suspension (see Materials and Methods for details). The dotted line shows the spectrum obtained following reoxygenation of the anaerobic suspension. (B): Plot of the normalized absorbance changes measured as in panel (A) at 578–597 nm versus O2 concentrations; the latter was measured in parallel in a twin sample by respirometry as detailed in Materials and Methods. The values shown are means ± SEM of three different experiments and were fitted with an hyperbolic equation; the indicated C50 is the concentration of O2 resulting in half of the maximal ΔA change. (C, D): Respirometric analysis of the O2 reserve in HSPC. (C): 0.5 µM purified bovine heart cytochrome c oxidase plus 5 µM horse heart cytochrome c were suspended in RPMI without or with 30 × 106 HSPCs (traces (A) and (B), respectively); where indicated 20 mM ascorbate plus 0.2 mM TMPD were added and the O2 consumption recorded as detailed in Materials and Methods section. Following O2 exhaustion, the suspension was reaerated and the O2 consumption followed again upon reclosure of the oxymeter chamber. Trace (C) as trace (B) but with HSPCs pretreated with 50 µM ferricyanide for 10 minutes, spun down, and resuspended in the medium. The histogram below the traces indicate the lag-time for O2 exhaustion to be reached for conditions (A) and (C) normalized to (B); mean values ± SEM of four different experiments with statistical analysis. (D): Dissociation plot of the O2-buffer. The reciprocal value of the normalized O2-release (computed as described in Supporting Information Fig. S2) was plotted as a function of the O2 concentration and fitted with a hyperbolic function. The plot shown is relative to the trace (B) shown in (A). The indicated P50 is the mean value ± SEM of four different assays with as many HSPC samples. Abbreviations: Asc, ascorbate; COX, cytochrome c oxidase; FIC, ferricyanide; HSPCs, hematopoietic stem/progenitor cells; TMPD, N,N,N′,N′-tetramethyl-p-phenylenediamine.

Figure 1C shows the outcome of a respirometric assay designed to unveil the presence of an O2 reservoir in the HSPC sample. Briefly, a catalytic amount of purified cytochrome c oxidase was supplemented to the medium without and with HSPCs (treated with antymicin A), and the O2 consumption was monitored to complete exhaustion upon addition of the respiratory substrates ascorbate plus TMPD. In the absence of cells, the rate of O2 consumption was constant up to 15–20 µM of O2 slowing down to lower concentrations. Since the KM to O2 of the cytochrome c oxidase is in the submicromolar range [18], the sluggish phase likely reflected a limited sensitivity of the oxygen electrode at the lowest concentrations of the O2. In the presence of HSPCs, a diversion from linearity in the O2 consumption begun at 100 µM of O2 and the sluggish phase was much more evident. Reaeration of the oxymetric chamber, following O2 exhaustion, resulted in a successive cycle of O2-consumption comparable with the previous one. Importantly the slow O2-consumption phase was completely abrogated by pretreating the HSPC sample with the oxidant ferricyanide. The outcome of the assay was compatible with a kinetic model assuming the presence in the HSPCs of an O2 buffering system, which rapidly released O2 thus causing an apparent decrease in the oxidase-catalyzed O2-consumption rate (Supporting Information Fig. S2A). Thus, the oxymetric traces provide quantitative information about the O2 binding/releasing activity of the O2-buffer computable as shown in Supporting Information Figure S2B, S2C and described in the legend thereof. The percentage of saturation of the putative buffer at different O2 concentrations fitted a hyperbolic function displaying a P50 = 3.53 ± 0.14 µM O2 (n = 5) (Fig. 1D) closely resembling that attained from the O2-dependent spectrophotometric changes.

All together the results presented in Figure 1 provide evidences for the occurrence in HSPCs of detectable amount of a redox-sensitive O2-liganding/exchanging system with a relatively high affinity and spectral features compatible with heme-containing protein(s).

HSPCs Express Detectable Amounts of Mb and Ngb but Not of Cygb

The capability to reversibly bind O2 in the low micromolar P50 range (and without cooperativity) is a hallmark of the human globin family comprising Mb and the more recently discovered Cygb and Ngb [13]. Figure 2A illustrates the result of a quantitative reverse transcriptase PCR (RT-PCR) assaying the transcript levels of Mb, Ngb, and Cygb in HSPCs. For comparative purpose, the expression level of each globin was normalized to that of representative cytotypes: (a) human rabdomyosarcoma-derived cell line (Rd) for Mb [19]; (b) neuroblastoma-derived cell line (Gi-li-n) for Ngb [20]; (c) primary NHDF for Cygb [21]. It is shown that HSPCs express levels of Mb and Ngb transcripts comparable to that of the respective representative cell types whereas Cygb was scarcely expressed (Fig. 2A).

Figure 2.

Expression of globins in HSPCs. (A): Results of quantitative real-time reverse transcriptase polymerase chain reaction for Mb, Ngb, and Cygb transcripts; the globin expression levels in HSPCs were normalized to those of RD, G-LI-n, and NHDF cells for Mb, Ngb, and Cygb, respectively. The expression of the Mb transcript variant (Mb-V) normalized to that of the standard normal transcript (Mb-N) is also shown for HSPC and RD cells. The values are means ± SEM of six to eight different determinations from four different HSPC samples; the p values of statistically significant differences is also shown. For details see Materials and Methods section. (B, C): Western blotting of Mb and Ngb, respectively, on total protein extracts from PB and BM HSPCs and RD (B) and Glin (C). The monomeric (mon) and dimeric (dim) forms of the globins are indicated on the basis of the apparent molecular weights. Both (B) and (C) are representative of three different assays each yielding similar results. The histograms below each WB report the corresponding densitometric analysis; the bars are the mean values ± SEM (n = 3) of the ratios of the immune-detected bands with respect to the GAPDH and normalized to the values of the reference cell type. For details see Materials and Methods section. Abbreviations: BM, bone marrow; Cygb, cytoglobin; Dim, dimeric; HSPC, hematopoietic stem/progenitor cells; Mon, monomeric; Mb, myoglobin; Ngb, neuroglobin; NHDF, normal human dermal fibroblast; PB, peripheral blood.

A transcript variant of Mb (Mb-V) has been recently reported to be expressed in breast cancer cells. This variant harbors an extended 5′-untranslated region under control of a hypoxia-sensitive promoter [22]. Figure 2A shows that the level of the Mb-mRNA variant in HSPC resulted on an average basis about 15 ± 8-fold that expressed in Rd cells (p < .01, n = 4). Most notably, the Mb-V/Mb expression ratio was about 35 ± 15 (p < .01, n = 4) in HSPCs whereas it was about 1 in Rd cells. Consistent with this observation, HSPCs displayed immunodetectable amount of the subunit isoforms of the hypoxia inducible factor (HIF), HIF-1α and HIF-2α (Supporting Information Fig. S3). To notice, the stabilized amounts of the HIF subunits were not further enhanced by treatment with the hypoxia-mimetic desferoxamine. These observations confirmed and extended previous report showing HIF activation under normoxic condition in HSPCs [23].

To confirm the Mb and Ngb expression at the protein level, immunoblotting with commercially available antibodies was carried out on total protein extract of HSPC lysate and compared with that of reference cell types. Figure 2B, 2C show that both PB- and BM-derived HSPCs displayed detectable immune-reacting bands for Mb and Ngb. To note immune-blotting for Mb resulted in two distinct bands with apparent molecular weights consistent with the monomeric and dimeric form of human Mb as described in [24]. No change in the Mb protein is expected irrespective of the Mb transcript (i.e., Mb-N vs. Mb-V) [22]. Blotting of HSPC lysates was carried out with two different commercial Mb-Abs resulting in similar outcome (not shown). As to Ngb, a single immune-detected band was observed with a molecular weight compatible with the dimeric form. Challenging Ngb-Abs against purified horse Mb did not result in crossreactivity whereas Mb-Abs recognized it (not shown). Densitometric analysis of the immune-reacting bands both in PB- and BM-HSPCs resulted in values comparable with those of the Mb and Ngb present in the reference cytotype. The presence of Mb and Ngb in BM-HSPCs (i.e., obtained without G-CSF-induced HSPCs mobilization) ruled out any effect of cytokine treatment.

The Intracellular Compartmentalization of Mb Changes with the Expression of the CD34 Marker

To investigate the cellular localization of Mb and Ngb, the HSPC samples were subjected to immunocytochemical imaging by confocal microscopy. Figure 3A, 3B show that treatment of the HSPC with either of the Mb-Ab and Ngb-Ab resulted in an intracellular fluorescence signal (red fluorescence). Treatment of the same sample with a CD34-Ab resulted in a fluorescence signal located, as expected, at the surface of the HSPCs (green fluorescence). The relative specificity of the Abs used in the immune-cytochemistry assay was assessed as described in Supporting Information Figure S4 and in its relative legend.

Figure 3.

Laser scanning confocal microscopy (LSCM) immunocytochemical analysis of Mb and Ngb in hematopoietic stem/progenitor cells (HSPCs). Two samples of permeabilized peripheral blood-HSPCs were treated in parallel with the indicated primary Mb- and Ngb-Ab (panels (A) and (B), respectively) along with a FITC-conjugated Ab for CD34 (green fluorescence); the secondary anti-IgG Abs against the two anti-globins Abs were rhodamine-conjugated (red fluorescence). The LSCM analysis and processing of the images were performed as described under Materials and Methods section. The single color channels and the merged image are shown. The results shown are representative of five different preparations from as many different HSPC samples yielding similar outcomes. Abbreviations: FITC, fluorescein isothiocyanate; Mb, myoglobin; Ngb, neuroglobin; Rhod, rhodamine.

A closer analysis of the laser scanning confocal microscopy (LSCM) images shown in Figure 3 unveiled a nonhomogeneous pattern of the CD34-Ab-related fluorescence signal. Roughly 25% of the HSPCs displayed a brighter fluorescence (CD34high) (Supporting Information Fig. S5). This subset of the cell population represents more primitive HSPCs endowed with higher clonogenic and long-term hemopoietic recovery capacity [25, 26]. Most notably, although the whole cell content of Mb did not change in CD34-related HSPCs subsets, nevertheless, a different intracellular distribution of the Mb-related fluorescence was evident thereof. Figure 4A shows that in CD34high-HSPCs Mb was more concentrated in the central nuclear part of the cell whereas in CD34low-HSPCs the Mb-related fluorescence signal was mainly localized in the cytoplasmic rift surrounding the nucleus. The nuclear/cytoplamic Mb ratios were 1.40 ± 0.02 and 0.85 ± 0.05 (p < .0005) in CD34high- and CD34low-HSPCs, respectively. At difference of Mb, Ngb displayed a cytoplasmic localization irrespective of the CD34 expression level in HSPCs (Fig. 4B).

Figure 4.

Intracellular localization of Mb and Ngb in HSPCs. LSCM analysis of the intracellular distribution of the rhodamine-related fluorescence signals in HSPCs treated with anti-Mb and anti-Ngb (panels (A) and (B)) was as described in Figure 3. The single color channels and merged images of enlarged selected pairs of HSPCs representative of the CD34high and CD34low subsets, along with the false colors imaging of the rhodamine-related fluorescence are shown. Bar histograms on the right: statistical analysis of the intracellular distribution of the anti-Mb and anti-Ngb rhodamine-related fluorescence signals. The nuclear area for each cell was selected as a circle centered within the cell area with a diameter of two third of that of the cell; the pixel intensity of the region of interest (ROI) was estimated using dedicated tools of the image analyzer software Image J (http://imagej.nih.gov/ij/); the fluorescence of the cytoplasmic area was obtained subtracting that of the ROI from that of the whole cell. The horizontal bars are means ± SEM of at least 150–300 randomly selected single cells from three optical fields under each condition and from two to three different HSPC preparations; when statistical significant p value is also shown. C, cytoplasm; Mb, myoglobin; N, nucleus; N/C nucleus/cytoplasm ratio; T, total cell.

Ngb but Not Mb Correlates with CD34 Expression

The correlation between globins content and markers of hematopoietic stem and precursors cells was further investigated by “dual-colors” flow-cytometric analyses. Figure 5A shows that Mb content in HSPCs displayed no significant correlation with either CD34 and the lineage-uncommitted precursor marker CD38 as well as with markers of erythroid (CD71), lymphoid (CD19), and myeloid (CD33) progenitors. When the same analysis was carried out for Ngb a significant positive correlation was observed between the globin content and the CD34 expression such that the Ngb expression in CD34high/bright-HSPCs was twofold that observed in the CD34low cell subset (Fig. 5B). As for Mb no correlation was found between Ngb content and expression of lineage-committed markers. These results showed that while the Mb protein is homogeneously expressed in all the HSPC compartment irrespective of the degree of lineage-commitment, the Ngb protein content appeared to prevail in the uncommitted HSPC.

Figure 5.

Flow cytometry analysis of the distribution of MB and Ngb in hematopoietic stem/progenitor cell (HSPC) subsets. HSPC samples were co-immunostained for Mb (clouds diagrams in (A)) or Ngb (cloud diagrams in (B)) with either of the indicated CD markers (i.e., CD34, CD38, CD33, CD71, CD19). In the case of costaining with CD34 the distribution of Mb- and Ngb-related fluorescence, along with resulting parameters, is also shown for CD34low (blue) and CD34high(bright) (red) gated populations. The results shown are from a single HSPC sample and are representative of three different preparations yielding similar outcomes. See Materials and Methods section for experimental details. Abbreviations: FITC, fluorescein isothiocyanate; Mb, myoglobin; Ngb, neuroglobin; PE, phycoerythrin.

In order to complement the above reported observations at the gene expression level, the relative amount of the globin transcripts was assessed by quantitative RT-PCR on subsets of HSPC sorted on the basis of their CD34 content (Supporting Information Fig. S6). The results obtained show that while the normal MB transcript did not change with the level of CD34 its variant was significantly higher in the HSPC-CD34low subset as compared with the HSPC-CD34high and HSPC-CD34mid. As for the Ngb transcript its level increased progressively from HSPC-CD34high to HSPC-CD34low.

Although counterintuitive the different profile of globin expression at the mRNA and protein level is not surprising given the growing evidence that the transcription of the majority of the genes does not correlate with the content of the corresponding proteins [27]. This occurs as a consequence of different processes—that is, post-transcriptional, translational, and protein degradation regulation—in controlling steady-state protein abundances.

HSPCs Possess a Nitrite Reductase Activity Generating NO

In addition to the canonical functions of O2 binding/transport, a recently unveiled function of globins is their capability, under hypoxic/anoxic conditions, to reduce nitrite to NO [28, 29]. To test the nitrite reductase activity-dependent NO-generating capacity of globins-containing HSPCs, we developed a variant of a described experimental protocol [28], which exploits the ability of NO to inhibit mitochondrial respiration. HSPCs supplemented with isolated mitochondria were allowed to become anoxic following exhaustion of dissolved O2 by the presence of excess of respiratory substrate; after then the chamber lid of the oxymetric chamber was removed and the air oxygen was allowed to flux therein. The detection of O2 in the opened chamber is expected, under this condition, to result from the balance between the opposing rates of the gas influx and of the mitochondrial respiration as well as of the globin-mediated O2 binding. Figure 6A shows, indeed, that following air admission in the anaerobic suspension, a steady-state equilibrium between O2 influx and consumption was achieved. When nitrite was added to the anoxic suspension, the steady-state level of detectable O2 was significantly larger suggesting inhibition of the mitochondrial respiration. In the absence of HSPCs, the suspension of sole mitochondria resulted to behave qualitatively as that supplemented with HSPCs. This is consistent with the described nitrite reductase activity of hypoxic cytochrome c oxidase [30]. However, the inhibition of the mitochondrial respiration was significantly larger in the presence of HSPCs clearly indicating in these the presence of a nitrite reductase activity generating NO. Purified horse heart Mb supplemented to the mitochondrial suspension mimicked the results shown in Figure 6 under identical experimental protocol (data not shown).

Figure 6.

Measurements of the nitrite reductase-dependent production of NO in HSPCs. (A): Oxymetric detection of nitrite-mediated inhibition of mitochondrial respiration. Suspensions of 0.5 mg prot/mL isolated BHM ± 20 × 106 per mL HSPCs were allowed to became anoxic following the addition of 20 mM ascorbate plus 0.1 mM TMPD in sealed twin oxymetric chambers as described in the legend to Figure 2A. Then the stirring was switched off, the chambers unplugged (Air ON) and the stirrer switched on at a rate of 200 rpm. After achievement of a steady O2 detection the chambers were plugged (Air OFF) and the O2 allowed to exhaust. Hence, 100 µM of NaNO2 was added and the reoxygenation cycle repeated as indicated. The histogram on the right indicates the stationary level of O2 attained under the indicated different conditions; values are means ±SEM of three different experiments with as many HSPC samples. (B): Fluorimetric detection of nitrite-induced NO generation. Suspensions of BHM ± HSPCs ((A) and (B)) were prepared as in (A) in the presence of 10 µM 4-amino-5-methylamino-2′,7′-difluorofluorescein, placed in a 0.5 mL of a fluorimetric cuvette provided with a magnetic bar, and overlaid with 0.5 mL of mineral oil. After addition of ascorbate and TMPD by a Hamilton microsyringe, 100 µM NaNO2 was added after 15 minutes and fluorescence was recorded as described in Materials and Methods section. In (C) 2 mM KCN was added before NaNO2. The histogram on the right shows the initial rates of the fluorescence increase attained following the addition of NaNO2; values are means ±SEM of three different experiments under each condition with as many HSPC samples. Abbreviations: BHM, bovine heart mitochondria; HSPCs, hematopoietic stem/progenitor cells; KCN, potassium cyanide.

To further prove more directly the nitrite reductase activity of globin-containing HSPCs, NO production was assessed under anoxic conditions by the membrane permeant fluorescent probe DAF-FM diacetate. Figure 6B shows that addition of 100 µM nitrite to an anaerobic suspension of HSPCs and isolated mitochondria resulted in a DAF-mediated fluorescence increase. The rate of the fluorescence increase was significantly higher than that elicited by mitochondria alone with the latter being completely abrogated by the cytochrome c oxidase inhibitor KCN (potassium cyanide).


The main and novel result presented in this study is that human HSPCs express detectable amount of functional Mb and Ngb. Importantly, globins' expression was confirmed by five independent experimental approaches including differential optical spectral analysis, respirometry, quantitative-PCR, Western blotting, and immunocytochemistry. The dogma confining Mb exclusively in cardiac and skeletal muscle has been recently challenged by the discovery of Mb in nonmuscle contests, for example, in hypoxia tolerant fish [31] and in nonmyogenic human tumors and in several cancer-derived cell lines [22, 32-34]. Likewise, Ngb, initially found exclusively expressed in nervous system and retina [13], has been reported in non-small-cell lung cancer [33]. However, to the best of our knowledge this is the first report showing the presence of Mb and Ngb in “normal” nonmuscular and non-neural mammalian cells, respectively.

From the optical spectra a content of approximately 0.01 nmol heme per 106 HSPCs was estimated corresponding to approximately 180 ng globin per 106 HSPCs and to an intracellular concentration of approximately 25 µM (the HSPC volume being approximately 400 µm3 on an average basis). These values are significantly higher than the upper range reported in tumor samples [32] but about one order of magnitude lower than in muscular cells. In keeping that the O2 concentration in the BM is likely to be lower than 13 µM (corresponding to 10 mm Hg) [2, 3] our estimates strongly indicate a physiological relevance of globins in the HSPC biology.

A further intriguing finding about the Mb expression in HSPCs is that it occurred largely through a HIF-1/2α-dependent alternative transcript recently described in breast cancer cells [22]. Accordingly, we found normoxic stabilization of HIF-1/2α in HSPCs confirming and extending previous observations [23]. In the case of Ngb, apparently lacking conventional hypoxia responsive elements, a number of studies have, nevertheless, clearly established that hypoxia and HIF-1 are necessary for the upregulation of Ngb expression in the nervous system [35].

The flow-cytometry analysis carried out in this study failed to indicate significant correlations of Mb with markers of stem/progenitors cell subsets whereas Ngb resulted to be more expressed in immature precursors. These observations would suggest that the functional role of the globins might be pursued in the earlier step(s) of the stem cell commitment.

The last feature emerged from this study was the heterogeneous intracellular compartmentalization of Mb in distinct subsets of HSPCs with a significantly higher nuclear/cytoplasmic ratio in CD34high-HSPCs. Interestingly, nuclear localization of cytoglobin has been reported [36] suggesting the possibility that globin-folded proteins might function as transcriptional regulators. Unlike Mb, the intracellular distribution of Ngb was mainly cytoplasmic irrespective of the level of CD34 expression. A recent comparative study reports that the protein interactomes of Ngb and Mb overlap substantially and are modified by hypoxia. The globins-interacting proteins include partners consistent with both antioxidative and antiapoptotic functions as well as, in the case of Mb, nuclear located proteins [37].

Our observations do not allow drawing conclusions about the specific functions of Mb and Ngb in HSPCs. However, taking cue from studies carried out in cancer cells where the Mb gene was overexpressed [38] or silenced [22] it is likely that its role in HSPCs is unrelated or not solely related to the O2 storage/delivery.

We would like to propose a hypothesis (see Fig. 7) whereby the functional role of cytoplasmic Ngb/Mb is linked, under hypoxic condition, to generation of NO through the described nitrite reductase activity of their deoxygenated forms [28, 29]. This activity has also been proved in HSPCs in this study. NO has been reported to inhibit the HIF-hydroxylating prolylhydroxylases (PHDs) thereby contributing to stabilize HIF-1/2α [39]. HIF is more and more emerging as an essential transcription factor controlling oxygen homeostasis and energy metabolism for maintenance of HSC function and long-term self-renewal [9, 40]. NO is also a powerful competitive inhibitor of cytochrome c oxidase [28, 41]. Thus the combined action of NO would inhibit the mitochondrial oxidative phosphorylation and foster primitive HSPCs to rely mainly on glycolysis-based bioenergetics [8]. However, as side effect, inhibition of mitochondrial respiration would lead to production of ROS as repeatedly observed under hypoxia [4]. Mitochondrial ROS might also contribute to HIF-1/2α stabilization by oxidative inhibition of PHDs [42]. To prevent redox-signaling-mediated activation of proliferation/differentiation, the nuclear localization of Mb might function as a localized antioxidant buffer preserving a quiescent state in HSPCs [43]. Consistent with this hypothesis is the notion that more primitive HSPCs expressing relatively high levels of CD34 have a more pronounced long-term in vivo repopulating capacity [26, 44], a lower ROS production [45] and mitochondria content [5, 46] and a higher level of hypoxicity markers [47]. Conversely, the CD34low-HSPCs exhibit opposite phenotypic features and are considered as precommitted cells [47]. These subset of HSPCs, likely located in a less hypoxic BM environment, for example, the vascular niche, are more prone to undergo oxidative metabolism, in virtue of a higher mitochondrial content, and as such more responsive to proliferative/differentiative stimuli. Relocation of Mb from nuclei to cytoplasm might favor redox signaling-mediated expression of specialization genes [48].

Figure 7.

Schematic view of the suggested functional role of globins under hypoxic and normoxic conditions in quiescent and early committed HSPCs. It is proposed that under hypoxic conditions (A) the deoxygenated forms of Mb and Ngb release NO by means of their nitrite reductase activity. NO inhibits the mitochondrial RC activity and oxidative phosphorylation and leads to enhanced ROS production. These in turn stabilize the HIF-1/2α by inhibiting the PHDs thereby setting conditions for cell survival and metabolic quiescence in HSC or early progenitors; the nuclear localization of Mb, observed in the more primitive CD34high-HSPCs might prevent oxidant-mediated activation of specialization genes. Under normoxia (B) the oxygenated forms of Mb and Ngb are unable to generate NO but rather function as NO-oxygenases. This unblocks the RC fostering oxidative phosphorylation and a more active metabolic state in CD34low-HSPC. Redistribution of Mb from the nucleus to the cytoplasm might serve on one hand to favor proliferation/differentiation by NOX-generated ROS-mediated signaling and on the other hand to support intracellular O2- or FA-diffusion. See Discussion section for further explanations. Abbreviations: FA, fatty acid; HSPC, hematopoietic stem/progenitor cells; HIF-1/2α, hypoxia inducible factors 1/2α; Mb, myoglobin; Ngb, neuroglobin; NOX, NADPH oxidase; PHDs, prolyl-hydroxylases; RC, respiratory chain; ROS, reactive oxygen species; RIRS, ROS-induced ROS release.

Notably, 50% of the O2 consumed by HSPC under normoxia is due to constitutively active isoforms of the NADPH oxidase (NOX) [5, 49], which deliberately generates ROS thus contributing to redox signaling [43, 49-53]. An intriguing aspect reported in various cell systems is a cross-talk between NOX and mitochondrial ROS [54]. Mitochondrial H2O2 regulates NOX activity and, conversely, NOX activation induces mitochondrial H2O2 formation thus supporting a positive feed-forward mechanism that promotes sustained H2O2 production and activation of redox signaling. This may be part of a more general phenomenon described as “ROS-induced ROS release.” Whether this regulatory mechanism is involved in stem and progenitor cell function is the subject of future investigation [55].

Under normoxic conditions and/or in CD34low-HSPCs the higher cytoplasmic/nuclear ratio of Mb/Ngb would facilitate a continuous flux of O2 from the extracellular to the perimitochondrial space. Notably, the oxygenated forms of globins do not exhibit nitrite reductase but rather NO-dioxygenase activity [41] thereby NO is converted to nitrate and this would unleash the NO-dependent inhibition of mitochondrial respiration. These combined effects would ensure a more efficient ATP production by oxidative phosphorylation to cope with the high-energy demand required to attain the proliferative/differentiative HSPCs reprogramming [11, 12]. Another described property of Mb is its fatty-acid binding/release activity [56] that might be functional, under condition requiring high metabolic activity, to promote mitochondrial fatty acids oxidation (FAO). Consistent with this notion is a recent report showing that the maintenance of HSC is regulated by the PML-PPARγ-FAO pathway controlling the asymmetric division of HSC. This finding further highlights the leading role of a metabolic switch in tuning the stem cell fate [57].


In this study, it is shown for the first time that human HSPCs express significant amount of two members of the O2-binding globin family, Mb and Ngb. The control exerted by HIF-1/2α on the expression of Mb suggests that the globin function(s) in this peculiar cell contest is likely related to adaptation to hypoxia. The hypoxic nitrite-related NO production of HSPCs would imply a previously unrecognized mechanism to keep in check the mitochondrial oxidative metabolism. Moreover, the finding of different intracellular compartmentalization of Mb, as a function of the stemness hierarchy, would indicate an additional role in the maintenance of HSC likely related to control of redox-signaling pathways. Genetic manipulation of the expression of both/either Mb and Ngb in HSPCs are expected to provide clues about their physiological role and possible impact in malignant transformation.


This work is dedicated to the memory of the never forgotten Antonio Tabilio who enthusiastically inspired the Foggia group to this line of research. We thank Dr. Giuseppe Capitanio from the University of Bari for kindly providing purified bovine heart cytochrome c oxidase. This work was supported by grants from the Italian Ministry of University and Research (PRIN-2008FJJHKM_001) to N.C. and from the Italian Association against Leukemia-Lymphoma and Myeloma (AIL-Foggia) to C.P.

Author Contributions

N.C.: conception and design and manuscript writing; C.P.: conception and design and manuscript revising; A.D., R.S., G.Q., M.G., L.D.V., and T.T.: collection, analysis, and interpretation of data; F.F.: provision of study materials; M.D.I.: provision of study materials.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.