TNF-α Has Tropic Rather than Apoptotic Activity in Human Hematopoietic Progenitors: Involvement of TNF Receptor-1 and Caspase-8§


  • Keren Mizrahi,

    1. Frankel Laboratory, Center for Stem Cell ResearchDepartment of Pediatric Hematology-Oncology, Schneider Children's Medical Center of Israel, Petach Tikva, Israel
    2. Department of Surgery, Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • Jerry Stein,

    1. Bone marrow Transplant Unit, Department of Pediatric Hematology-Oncology, Schneider Children's Medical Center of Israel, Petach Tikva, Israel
    Search for more papers by this author
  • Isaac Yaniv,

    1. Bone marrow Transplant Unit, Department of Pediatric Hematology-Oncology, Schneider Children's Medical Center of Israel, Petach Tikva, Israel
    Search for more papers by this author
  • Offer Kaplan,

    1. Department of Surgery, Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • Nadir Askenasy

    Corresponding author
    1. Frankel Laboratory, Center for Stem Cell ResearchDepartment of Pediatric Hematology-Oncology, Schneider Children's Medical Center of Israel, Petach Tikva, Israel
    • Frankel Laboratory, Center for Stem Cell Research, Schneider Children's Medical Center of Israel, 14 Kaplan Street, Petach Tikva 49202, Israel

    Search for more papers by this author
    • Telephone: 972-3921-3954; Fax: 972-3921-4156

  • Author contributions: K.M.: performance of experiments and collection of data; J.S. and I.Y.: collection of data and data analysis and interpretation; O.K.: collection of data and manuscript writing; N.A.: performance of experiments, conception and design, and manuscript writing.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS October 18, 2012.


Tumor necrosis factor-α (TNF-α) has been suggested to exert detrimental effects on hematopoietic progenitor function that might limit the success of transplants. In this study, we assessed the influences of TNF-α and its two cognate receptors on the function of fresh umbilical cord blood (UCB) and cryopreserved mobilized peripheral blood (mPB). CD34+ progenitors from both sources are less susceptible to spontaneous apoptosis than lineage-committed cells and are not induced into apoptosis by TNF-α. Consequently, the activity of UCB-derived severe combined immune deficiency (SCID) reconstituting cells and long-term culture-initiating cells is unaffected by this cytokine. On the contrary, transient exposure of cells from both sources to TNF-α stimulates the activity of myeloid progenitors, which persists in vivo in UCB cell transplants. Progenitor stimulation is selectively mediated by TNF-R1 and involves activation of caspase-8, without redundant activity of TNF-R2. Despite significant differences between fresh UCB cells and cryopreserved mPB cells in susceptibility to apoptosis and time to activation, TNF-α is primarily involved in tropic signaling in hematopoietic progenitors from both sources. Cytokine-mediated tropism cautions against TNF-α neutralization under conditions of stress hematopoiesis and may be particularly beneficial in overcoming the limitations of UCB cell transplants. Stem Cells2013;31:156–166


Members of the tumor necrosis factor (TNF) superfamily have been long recognized to mediate negative regulation of the immune system through activation-induced cell death [1, 2]. Similar to this pivotal mechanism of immune homeostasis, the TNF family receptors have been demonstrated to suppress proliferation and arrest growth of clones in distal stages of hematopoietic cell differentiation [3–5]. The involvement of TNF receptor/ligand interactions in the proximal stages of hematopoietic cell development and function is less characterized and can be not extrapolated from the negative regulatory attributes in terminally differentiated hematopoietic and immune cells [6–9]. For example, early evidence showed differential effects of TNF-α on myeloid cells in various stages of differentiation [10–12]. Insensitivity of bone marrow-derived progenitors to apoptosis under steady-state conditions has been attributed to low-level expression of the TNF family receptors [13–15]. In addition, several antiapoptotic factors overexpressed in hematopoietic progenitors have been associated with protection from apoptosis including nuclear factor-κB (NFκB) [16–18], Bcl-2 [19, 20], c-FLICE inhibitory protein (c-FLIP) [21, 22], and inhibitor of apoptosis proteins (IAP) [23, 24], which converge to suppress caspase-8-mediated activation of downstream effector caspases.

Work performed in models of stress hematopoiesis and transplantation using murine and human cells has yielded controversial evidence and interpretation of the role of TNF family receptors as dual mediators of apoptosis and stimulation [25–27]. On one hand, in vitro experiments have demonstrated detrimental effects of TNF-α on hematopoietic cell function [28–30] through induction of apoptosis [31–33] mediated both by direct interaction with its cognate receptors [11, 30, 34] and by indirect upregulation of the Fas receptor [13–15]. On the other hand, it has been emphasized that TNF-mediated apoptosis is restricted to cytokine-stimulated human bone marrow and umbilical cord blood (UCB) progenitors in vitro [22, 35-37] and is accentuated by cytokine withdrawal [38]. We have recently demonstrated tropic signaling of TNF-α in murine hematopoietic progenitors [39] and of TNF-related apoptosis-inducing ligand (TRAIL) in UCB progenitors [40]. Similar divisive evidence has been presented by in vivo studies. The traditional concept of a negative regulatory role of the TNF receptors in hematopoietic progenitors is supported by impaired engraftment [30, 41, 42], which has been substantiated in murine transplants using TNF receptor-deficient donor cells [33, 43]. Contrasting evidence has documented reduced hematopoietic progenitor activity and ineffective durable hematopoietic reconstitution of mice grafted with bone marrow cells deficient in TNF-α and its cognate receptors [39, 44, 45]. Furthermore, the most primitive progenitors become responsive to TNF family ligands through ubiquitous upregulation of the TNF [39], TRAIL [40], and Fas receptors [46] soon after transplantation without sensitization to apoptosis. Thus, one set of experiments would suggest that neutralization of TNF-α, a major inflammatory factor released as a consequence of conditioning-induced injury, would be beneficial to hematopoietic progenitor engraftment. Another set of experiments implies that the most primitive hematopoietic progenitors can operate as repair and reconstitution units under injury environments such as stress hematopoiesis following chemoradiotherapy, because injury factors such as TNF-α augment their activity by transduction of stimulatory signals.

In this study, we assessed the impact of TNF-α and its receptors on fresh UCB cells using surrogate in vitro assays and xenogeneic transplants. The activities of severe combined immune deficiency (SCID)-repopulating cells (SRC), long-term culture-initiating cells (LTC-IC), and colony-forming units (CFU) in semisolid cultures reflect the activity of progenitors positioned in progressive stages of commitment and differentiation. Complementary assays of apoptosis, proliferation, and cell cycle phase document tropic effects of TNF-α in human UCB progenitors transduced by TNF-R1 and involving caspase-8 activation, with concurrent negative regulation of the differentiated progeny. These findings in fresh UCB cells have been validated in cryopreserved mobilized peripheral blood (mPB) samples to demonstrate that responsiveness to TNF-α signaling is a general feature of human hematopoietic progenitors.


Cell Isolation, Characterization, and Staining

UCB was obtained from healthy donors pending informed consent according to the Institutional guidelines of the Beilinson Campus (IRB file 3656: 920050264). UCB samples harvested before placental delivery with citrate-phosphate-dextrose adenine-1 were diluted twofold in phosphate buffered saline (PBS) containing 0.5% human serum albumin and 2 mM EDTA and collected by centrifugation over Lymphocyte Separation Medium (1.077–1.08 g/ml, MP Biomedical, Illkirch, France; [40]. Mononuclear cells (MNC) were washed twice, and the lineage−/low (lin) subset was immunomagnetically isolated using the Human Lineage Cell Depletion Kit (Miltenyi Biotec, Bergisch Gladbach, Germany; The efficiency of the lin cell separation was reassessed as detailed above using a cocktail of Phycoerythrin (PE)-labeled antibodies. Samples of cryopreserved mPB were thawed at 37°C and processed by gradient centrifugation for collection of viable MNCs.

Flow Cytometry

Measurements were performed with a Vantage SE flow cytometer (Becton Dickinson, Franklin Lakes, NJ; [39, 40]. CD34+ cells were identified using Allophycocyanin-labeled antibodies (clone 8G12, Pharmingen, San Diego, CA; and lineages were quantified using PE-labeled monoclonal antibodies (mAb): anti-human CD3 (clone OKT3, BD Pharmingen, PE), anti-human CD19 (clone HD37, IQProducts, Groningen, Netherlands;, and anti-human CD33 (clone WM53, IQProducts). Expression of receptors in UCB cells was determined using mAb against Fas (clone DX2, Miltenyi), TNF-R1 (clone 16803, R&D Systems, Abingdon, U.K.;, and TNF-R2 (clone 22235; R&D Systems) [39]. Cell death was determined in cells incubated with 5 μg/ml 7-aminoactinomycin-D (Sigma, St. Louis, MO; and apoptosis using Annexin-V (IQ products) and caspase-3 activity (Apoptosis Kit, Pharmingen). Engraftment of UCB cells in NOD.SCID xenochimeras was determined in the bone marrow after 12 weeks using antibodies against murine CD45 (clone 30-F11, eBioscience, San Diego, CA; and human CD45 (clone ML2, IQ Products) [40]. Proliferation was determined from dilution of the intracellular dye 5-(and-6-)-carboxyfluorescein diacetate succinimidyl ester (Molecular Probes, Eugene, OR; [47]. Cell cycle distribution was analyzed with propidium iodide and Ki-67-FITC staining in cells fixed with 100% cold ethanol for 20 minutes on ice and permeabilized with 0.2% Tween 20 (Sigma) for 15 minutes at 37°C. Proliferation and fractional distribution of cells in the G0/G1, S, and G2/M phases were calculated using the ModFit software (Verity Software House;

Apoptotic Challenge

UCB cells were incubated (5 × 106 cells per milliliter) for variable times in α-MEM culture medium (Beit Haemek, Israel; supplemented with StemPro Nutrient Supplement (Stem Cell Technologies, Vancouver, BC;, 2 mM L-glutamine, and 50 μM 2β-mercaptoethanol purchased from PeproTech (Rocky Hill, NJ; [40]. Apoptosis was induced by supplementation of 20 ng/ml recombinant TNF-α (PeproTech) and was evaluated in human Jurkat cells (E6-1TIB-152, ATCC, Rockville, MD; used as a positive control. Caspase-8 and caspase-3 were inhibited by the addition of Z-IETD-fmk and Z-DEVD-fmk (R&D Systems), respectively [39].

CFU-GM Assays

CFU assays were performed in 1.2% methylcellulose containing Iscove-modified Dulbecco medium (Beit Haemek) by plating 2.5 × 103 MNC per well or 103 CD34+ and lin cells per well [40]. Cultures were supplemented with 2 mM L-glutamine, 30% fetal calf serum, 50 ng/ml stem cell factor (SCF), and 10 ng/ml interleukin-3 (IL-3), purchased from PeproTech. Cultures were stimulated with 10 ng/ml recombinant mouse (rm) granulocyte-macrophage colony-stimulating factor (rmGM-CSF) for determination of myeloid colonies (CFU-GM) and additional 10 U/ml recombinant human erythropoietin (EPO) was used for mixed colonies. Small (30–50 cells) and large (>50 cells) colonies were counted after 14 days using an Inverted Microscope (Olympus Optical, Tokyo, Japan; Cultures were supplemented with 10 ng/ml SuperFasL (AXORA, San Diego, CA, and 20 ng/ml TNF-α (PeproTech), and blocking antibodies against Fas (clone ZB4, Millipore, MA;, TNF-R1 (clone 16803, R&D Systems), and TNF-R2 (clone 22221, R&D Systems).


UCB-derived mononuclear and lineage-negative cells were seeded on feeder layers of irradiated (3,000 rad, RadSource 2000; human mesenchymal stromal cells derived from UCB blood (PromoCell, Heidelberg Germany; The cultures were refreshed every week by replacing half the culture medium (Myelocult; Stem Cell Technologies) and supplementation of 20 ng/ml TNF-α along control cultures from the same cord unit without supplementation of ligand. After 5 weeks, the cells were harvested, washed, and plated in methylcellulose.

Animal Preparation and Transplantation

Mice used in this study were NOD.CB17-Prkdcscid/J (NOD.SCID) purchased from Jackson Laboratories (Bar Harbor, ME; www.jax,org) and housed in a barrier facility. All procedures were approved by the Institutional Animal Care Committee. Recipients were routinely conditioned with two daily doses of 25 μg/g busulfan (Sigma), administered 2 days before cell transplantation. Cells suspended in 0.2 ml PBS (Biological Industries) were infused into the lateral tail vein.

Statistical Analysis

Data are presented as means ± SD for each experimental protocol. Results in each experimental group were evaluated for reproducibility by linear regression of duplicate measurements. Differences between the experimental protocols were estimated with a post hoc Scheffe's t test and significance was considered at p < .05.


Apoptotic Activity of TNF Receptors

Assessment of 35 UCB units revealed variable expression of the TNF receptors, which are expressed by small fractions of MNCs (∼20%) and CD34+ progenitors (∼10%), with differential lower expression of TNF-R1 only in isolated lineage-negative (lin) progenitors (Fig. 1A). Both gated CD34+ and isolated lin progenitors were less sensitive to spontaneous apoptosis as compared to the bulk UCB population and were not affected by exposure to toxic doses of TNF-α (Fig. 1B). Cells expressing either one of the TNF receptors were particularly susceptible to spontaneous apoptosis; however, the cognate ligand failed to induce apoptosis (Fig. 1C) or modulate the proliferation rates (Fig. 1D). These data indicate that UCB cells expressing both TNF receptors are largely more susceptible to spontaneous apoptosis in liquid culture; however, they are unaffected by the cognate ligand. Notably, UCB cells were incubated without supporting growth factors that often modulate cell phenotype, proliferation, and activity in vitro and consequently modulate the effects of TNF-α [22, 30, 35-37].

Figure 1.

TNF receptor expression and apoptosis in fresh umbilical cord blood (UCB) cells. (A): The expression of the TNF receptors was determined in MNC and gated CD34+ and lineage-negative (lin) progenitors from 35 UCB units. (B): Apoptosis of MNC (n = 15–33), gated CD34+ (n = 12–27), and isolated lin progenitors (n = 6–14) after 1–3 days of incubation in medium and with 20 ng/ml TNF-α, as determined from Annexin-V uptake. (C): Apoptosis of UCB cells in reference to TNF receptor expression following incubation in medium and with TNF-α for 48 hours (n = 12). Right panels are demonstrative measurements of apoptosis on the gated population expressing TNF-R1. (D): Proliferation rates in reference to TNF receptor expression (n = 7), as determined from CFSE dilution and quantified with the ModFit software. Right panels are demonstrative plots of CFSE dilution gated on cells expressing TNF-R1. Abbreviations: 7-AAD, 7-aminoactinomycin-D; CFSE, carboxyfluorescein diacetate succinimidyl ester; MNC, mononuclear cells; TNF, tumor necrosis factor.

To determine whether the observed features are specific to UCB cells or to progenitors in general, we used a different prevalent source of cells for transplantation: cryopreserved mPB. The profile of TNF receptor expression showed predominance of TNF-R1 over TNF-R2 in thawed mPB cells, including CD34+ progenitors (Fig. 2A). Remarkably, only TNF-R2 was acutely upregulated in all mPB cells during short incubation, irrespective of the presence of TNF-α in the culture medium. Both initial and postincubation levels of the receptors had no significant impact on the sensitivity of mPB cells to apoptosis induced by the cognate ligand (Fig. 2B). Although spontaneous thawing-related apoptosis was higher in the CD34+ subset than in the whole mPB population, progenitors sustained superior survival than lineage-positive subsets during incubation for 4–16 hours. Taken together, these data demonstrate that CD34+ progenitors from both sources are less susceptible to spontaneous apoptosis than mature lineages, with marked variations in susceptibility to spontaneous apoptosis in reference to the source of cells and their preservation.

Figure 2.

TNF receptor expression and apoptosis in cryopreserved mobilized peripheral blood (mPB) cells. (A): The expression of the TNF receptors was determined in MNC and gated CD34+ progenitors from six thawed mPB samples and after incubation for 16 hours. Right panels are demonstrative plots of receptor expression in thawed mPB cells (black line) and after 16 hours of incubation (gray line) as compared to isotype control (shaded). (B): Apoptosis of MNC (n = 8), gated CD34+ progenitors (n = 5) immediately after thawing and after 4 and 16 hours of incubation in medium and with 20 ng/ml TNF-α. Right panels are representative measurements of dead mPB cells upon thawing (upper) and apoptosis during 4 hours of incubation with TNF-α. Abbreviations: 7-AAD, 7-aminoactinomycin-D; MNC, mononuclear cells; TNF, tumor necrosis factor.

Ex Vivo Exposure to TNF-α Does Not Impair Quantitative SCID Repopulating Cell Activity

TNF receptor activation might not induce apoptosis but rather affect the function of UCB-derived hematopoietic progenitors. To evaluate the implications of TNF signaling in hematopoietic cell activity, we used several surrogate functional assays for progenitors in different stages of commitment. Engraftment of UCB cells in NOD.SCID mice represents a relatively primitive subset of SRC, which is somewhat related to the human reconstituting cell [48, 49]. In these experiments, we used equal initial numbers of cells to determine the effects of pre-exposure to TNF-α on SRC function (Fig. 3A). Incubations were performed for 25–48 hours considering that beyond this period the culture impairs significantly human cell engraftment [22, 30, 37, 40, 50]. Despite wide variability in the levels of xenochimerism, exposure of UCB cells to TNF-α for 24–48 hours had no apparent impact on SCR activity (Fig. 3B), suggesting insensitivity of this subset to apoptotic signaling through the TNF receptors. Prior studies have also documented that deficient engraftment induced by extended ex vivo culture is associated with cell proliferation and egress from the G0/G1 cell cycle phase [50–54]. The proliferation rates (Fig. 3C) and the cell cycle phase of UCB cells, gated CD34+, and isolated lin progenitors were largely unaffected by TNF-α under our experimental conditions (Fig. 3D), consistent with preserved engraftment of SRC restricted to the G0/G1 cycle phase.

Figure 3.

SCID repopulating assays. (A): Umbilical cord blood (UCB) cells were incubated in medium and with 20 ng/ml TNF-α for variable periods of time and equal numbers of initial cells were grafted into NOD.SCID mice conditioned with two doses of 25 μg/g busulfan. (B): Quantitative analysis of human donor chimerism was determined in the bone marrow at 12 weeks following transplantation of various numbers of viable UCB cells from the same unit preincubated in medium and with 20 ng/ml TNF-α. Data represent levels of chimerism in four to five individual mice grafted after incubation for 24 hours (5 UCB units) and 48 hours (6 UCB units) in medium and with TNF-α. The bottom panels represent flow cytometric measurements of human xenochimerism following transplant of fresh UCB cells and incubation for 24 hours in medium and with TNF-α. (C, D): MNC, gated CD34+ and isolated lineage-negative (lin) progenitors were assessed after 48 hours of culture with and without 20 ng/ml TNF-α for: (C) proliferation rates as determined from carboxyfluorescein diacetate succinimidyl ester dilution (n = 7) and (D) fractions of cells positioned in G0/G1 cycle phase as quantified by nucleic incorporation of propidium iodide (PI) (n = 6–9). Demonstrative plots of PI uptake following incubation in medium (open line) and with TNF (shaded). Abbreviations: MNC, mononuclear cells; TNF, tumor necrosis factor.

Ex Vivo Exposure to TNF-α Improves Engraftment and Myeloid Differentiation

Analysis of the contents of the bone marrow of NOD.SCID mice grafted with UCB cells showed quantitative and qualitative variations following exposure to TNF-α. The absolute contents of human CD34+ cells in murine bone marrow at 12 weeks post-transplantation were increased by pre-exposure to TNF-α for 24 and 48 hours by 34% ± 30% (nonsignificant) and 21% ± 24% (nonsignificant) as compared to control medium, respectively. Qualitatively, grafts pre-exposed to TNF-α ex vivo yielded superior myelomonocytic progeny in vivo (p < .005 for 24 and 48 hours vs. control), with a reciprocal fractional decrease in lineage-negative cells (Fig. 4A). These findings were further assessed in functional assays. In first stage, we sought to determine the impact of TNF-α on progenitors responsible for durable reconstitution by transplantation of the marrow contents of primary recipients into secondary recipients. Half of the femoral contents of the primary recipients contained 3.2 ± 1.4 × 105 and 5.3 ± 3.1 × 105 human CD34+ cells (not significant) following incubation for 24 hours in control medium and with TNF-α, respectively. Transplantation of half of femoral contents into secondary NOD.SCID recipients displayed increased levels of human chimerism following pre-exposure to TNF-α (Fig. 4B), validating the absence of detrimental effects of the cytokine on human progenitors. In second stage, the apparent increase in myeloid phenotypes was further assessed in methylcellulose cultures stimulated with human cytokines, showing increased myeloid clonogenic activity following exposure to TNF-α (Fig. 4C). Increased progenitor numbers and enhanced myeloid repopulation point to effective self renewal of UCB progenitors that is augmented by ex vivo exposure to TNF-α, an inductive event that persisted in vivo.

Figure 4.

Characteristics of SCID-repopulating cells. (A): Qualitative analysis of the bone marrow at 12 weeks post-transplantation of umbilical cord blood (UCB) cells from same unit incubated in medium for 24 hours (5 UCB units in ten mice) and with 20 ng/ml TNF (4 UCB units in eight mice) and for 48 hours in medium (4 UCB units in six mice) and with TNF (4 UCB units in seven mice) revealed qualitative variations: human lineage-negative and CD34 progenitors, CD3 T-cells, CD19 B lymphocytes, CD33 myelocytes, and CD14 monocytes. Right panels are representative for determination of multilineage engraftment using human-specific antibodies of grafts preincubated for 48 hours with TNF. (B): At 12 weeks post-transplantation, half of the femoral cellular contents primary recipients of UCB cells preincubated for 24 hours in medium and with TNF were grafted into secondary busulfan-conditioned NOD.SCID mice. Human xenochimerism was measured in the bone marrow after 12 weeks (3 UCB units in six mice in each group). (C): The bone marrow contents of human progenitors were assessed at the experimental end point in semisolid methylcellulose assays stimulated with human stem cell factor, IL-3, and granulocyte-macrophage colony-stimulating factor. Colonies of recipients of UCB cells preincubated for 24 hours (n = 6 from 3 UCB units) and 48 hours (n = 7 from 3 UCB units) were normalized to the clonogenic activity measured in recipients of cells preincubated in medium (normalized). Abbreviation: TNF, tumor necrosis factor.

Ex Vivo Exposure to TNF-α Stimulates Myeloid Progenitor Activity In Vivo

To validate the inductive effect of TNF-α on myeloid reconstitution, fresh UCB and cryopreserved mPB cells were pulsed with TNF-α for variable periods of time before plating in semisolid methylcellulose cultures. Exposure of UCB cells for 48 hours (Fig. 5A) and of mPB cells for 16 hours to TNF-α (Fig. 5C) resulted in significant increase in myeloid clonogenic activity as compared to cells preincubated in medium, consistent with induction of the myeloid lineage in vivo (Fig. 4A). Preincubated UCB cells also showed increased activity in mixed colony assays in response to GM-CSF and EPO (Fig. 5B). Notably, in these assays, equal initial cell numbers from the same UCB unit were compared to estimate the relative activities of myeloid progenitors at each time point.

Figure 5.

Exposure to TNF-α fosters myeloid progenitor activity in vitro. Cells were incubated for variable periods of time with and without 20 ng/ml TNF-α and were plated in methylcellulose cultures stimulated with GM-CSF for determination of myeloid colonies (CFU-GM) and with addition of erythropoietin for determination of mixed colonies (CFU). Data are presented as CFU per 1,000 total initial cells. (A): CFU-GM in cultures of fresh UCB cells (control, n = 35) and following incubation for 24–72 hours with (n = 32–34) and without TNF-α (n = 32–34). (B): Mixed colonies in cultures of fresh UCB cells (control, n = 28) and following incubation for 24–48 hours with (n = 19–28) and without TNF-α (n = 19–28). (C): CFU-GM in cultures of thawed mPB cells (control, n = 8) and after incubation for 4 (n = 7) and 16 hours with (n = 4) and without TNF-α (n = 4). Abbreviations: CFU, colony-forming unit; mPB, mobilized peripheral blood; TNF, tumor necrosis factor; UCB, umbilical cord blood.

TNF-α Stimulates Myeloid Progenitors In Vitro

Surrogate in vitro assays for hematopoietic progenitor activity include long-term cultures over mesenchymal stromal cells (LTC-IC), which represent a more primitive subset than colonies formed by committed progenitors in semisolid methylcellulose cultures. Exposure of UCB cells to 20 ng/ml TNF-α, during the entire 5 weeks period of the long-term cultures, did not impact LTC-IC activity (Fig. 6A), consistent with overall resistance of the CD34+ and lin progenitors to TNF-mediated apoptotic signaling. Similar results were obtained when cells were pre-exposed to TNF-α before plating in LTC-IC cultures without the continuous presence of the cytokine in the cultures. In next stage, TNF-α was continuously applied to methylcellulose cultures stimulated by SCF, IL-3, and GM-CSF. The presence of TNF-α increased CFU activity within the lineage-negative progenitor subset and decreased their activity in bulk UCB cultures (Fig. 6B). Indirect inhibitory effects of death ligands on myeloid progenitor activity in whole cell preparations have been shown to be caused by apoptosis of nonprogenitor cells [55].

Figure 6.

The influence of TNF-α on clonogenic activity of UCB cells. (A): LTC-IC activity of lineage-negative (lin) progenitors seeded for 5 weeks on UCB-derived MSC feeder layers with (n = 18) and without (medium, n = 35) supplementation of 20 ng/ml TNF-α and subsequently quantified in semisolid cultures for 2 weeks without TNF-α (in triplicates). Data are presented as CFU frequency in 1,000 cells. (B): The clonogenic activity of mononuclear UCB cells (n = 32) and lin progenitors (n = 21) was assessed in methylcellulose cultures stimulated with stem cell factor, IL-3, and GM-CSF for 2 weeks with and without the continuous presence of 20 ng/ml TNF-α. (C): The stimulatory effect of 20 ng/ml TNF-α on the clonogenic activity of lin progenitors (n = 21) is reversed by inhibition of both TNF receptors and TNF-R1 but not by inhibition of TNF-R2 (n = 12 in each group) and Fas (n = 11). (D): The influence of TNF-α on the ratio between large (>50 cells) and small (<30 cells) colonies (n = 21), and with selective inhibition of the TNF and Fas receptors (n = 11–12). (E): The impact of caspase-8 (33 mM Z-IETD-fmk) and caspase-3 (33 mM Z-DEVD-fmk) inhibition on the clonogenic activity of MNC (n = 9) and lin progenitors (n = 7–9). (F): Inhibition of the activity of caspase-8 (IETD) and capspase-3 (DEVD) increases colony size (n = 7–9). Abbreviations: CFU, colony-forming unit; LTC-IC, long-term culture-initiating cell; MNC, mononuclear cells; mPB, mobilized peripheral blood; TNF, tumor necrosis factor; UCB, umbilical cord blood.

The Mechanism of Myeloid Progenitor Stimulation by TNF-α

Stimulation of UCB progenitors by TNF-α was further characterized in a series of inhibition experiments. Neutralization of the Fas receptor had no significant impact on clonogenic activity (Fig. 6C), pointing to the TNF receptors as direct mediators of TNF-α signaling. Consistently, TNF-mediated stimulation was effectively reversed by neutralization of TNF-R1, but not of TNF-R2, evidence of nonredundant activation of these receptors by the common cognate ligand. In addition to recruitment of progenitors to form colonies, the size of myeloid colonies reflects TNF-mediated signaling in the differentiated progeny. The continuous presence of the ligand in cultures decreased colony size, an effect also mediated by TNF-R1 (Fig. 6D). However, the decrease in colony size was also reversed by neutralization of Fas, corroborating the contention that negative regulation of differentiated hematopoietic cells exerted by TNF-α is partially mediated by the Fas receptor [13–15]. These data imply a dynamic change in responsiveness of developing myeloid cells to signal transduction through TNF-R1: tropic signaling in progenitors and restriction of the differentiated progeny.

The signal transduction pathway of TNF-R1 was further assessed for involvement of caspases in induction of clonogenic activity. Inhibition of caspase-3 more than caspase-8 caused a small increase in CFU activity in bulk UCB cultures (Fig. 6E), consistent with indirect inhibition of clonogenic activity by dead cells in the cultures [55]. In variance, only inhibition of caspase-8 reduced significantly the development of human myeloid colonies induced by TNF-R1, indicating selective involvement of this caspase in transduction of tropic signaling. This behavior in human cells is different from murine bone marrow cells where inhibition of both caspases reverted the inductive effect of TNF-α [39]. As expected from the restrictive effect of TNF-α on the myeloid progeny, inhibition of either one of the caspases increased the size of myeloid colonies (Fig. 6F). Notably, these influences were observed under continuous presence of TNF-α in the cultures, therefore they do not represent influences of the cytokine upon growth factor withdrawal [38].


TNF-α is involved in a series of activities in human UCB and mPB cells, including induction of progenitor activity and negative regulation of expanding clones. In addition to insensitivity to receptor-mediated apoptosis, several functional assays indicate that TNF-α mediates tropic signals in hematopoietic progenitors. Signaling through TNF-R1 fosters myeloid activity and synergizes with growth factors in recruitment of cytokine-activated CFU, involving the activity of caspase-8. There was no detectable function of TNF-R2, although it was acutely upregulated in cultured mPB cells, and deficiency in this receptor has been shown to repress durable hematopoietic reconstitution in mice [39].

Prior studies have attributed the impaired hematopoietic reconstitution capacity of cytokine-activated cycling progenitors to detrimental effects of TNF-α as an apoptotic factor through direct interaction with its cognate receptors and indirectly through Fas [13-15, 30-33, 36, 41, 42, 56]. We found that resistance of unstimulated UCB progenitors to TNF-induced apoptosis results in preserved activity of SRC and LTC-IC. Ex vivo incubation of UCB cells with growth factors is generally associated with extensive changes in phenotype [36, 37] that decrease homing to and engraftment in the bone marrow of SCID mice [30-33, 40, 44], to the extent of loss of the reconstitution potential after prolonged (3–5 days) culture [22, 42, 50]. It is unclear to what extent UCB-derived SRC correspond to short-term or long-term human repopulating cells [48, 49]; however, these cells represent a more primitive subset than LTC-IC, which are equally unaffected by exposure to TNF-α throughout the entire culture duration of several weeks. One could argue that the presence of mesenchymal cells abolished the potentially detrimental influences of TNF-α, yet in view of stimulation of committed myeloid progenitors in semisolid cultures it is evident that TNF-α has no negative impact on hematopoietic progenitors in UCB.

Our findings in fresh UCB cells concur with the relatively low levels of TNF receptor expression in human bone marrow-derived hematopoietic precursors under steady-state conditions [13–15], however are inconsistent with the concept that low frequency of the receptors is the major mechanism of progenitor protection from detrimental influences of TNF-α [27–33]. Similar to murine bone marrow [39, 46], the expression of TNF family receptors is dissociated from and does not predict the sensitivity of hematopoietic progenitors to apoptosis. The dissociation is evident from failure of TNF-α to induce receptor-mediated apoptosis, although the susceptibility to spontaneous death in culture was markedly increased in cells expressing both TNF receptors. Likewise, the fraction of thawed mPB induced into apoptosis during ex vivo incubation (>50%) exceeded the relatively high levels of TNF receptor expression (∼ 30%) following mobilization with G-CSF.

The physiological significance of differential sensitivities to apoptosis in the transplant setting is the priority of more primitive apoptosis-insensitive precursors for engraftment in the hostile bone marrow environment by preventing early colonization of the marrow by apoptosis-sensitive, fast-expanding committed progenitors. This competitive process of progenitor prioritization for engraftment by differential sensitivity to apoptosis has been also demonstrated in murine bone marrow cells [46]. Fas, TNF-R1, and TRAIL-R2 (the only murine TRAIL receptor) are ubiquitously upregulated early after transplantation without sensitizing the most primitive precursor subsets to apoptosis [39, 40, 55]. Similar resistance to apoptotic signaling is demonstrated for TNF-R1, as previously demonstrated for TRAIL-R2 in UCB progenitors [40]. Hematopoietic cells are submitted to autocrine influences of TNF-α secreted by the cells themselves [57] and paracrine influences of this injury and inflammatory cytokine secreted by bone marrow stroma and immune cells [39, 50]. Reciprocally, TNF-α mediates interactions between hematopoietic progenitors and other cell types, including induction of the Jagged ligand in bone marrow-derived endothelial cells [58], support of plasmacytoid dendritic cell expansion [45], and sensitization of NK cells to apoptosis [59].

Resistance to apoptosis makes way to transduction of tropic signals and supports progenitor engraftment. Activation of myeloid progenitors by transient exposure to TNF-α ex vivo persisted to increase the absolute numbers of CD34+ progenitors and to enhance myelomonocyte differentiation of human UCB progenitors in vivo. Consistently, continuous presence of TNF-α fostered the activation of myeloid progenitors by growth factors in semisolid cultures. TNF-α has been shown to synergize with IL-3 and SCF in enhancing human progenitor activity under GM-CSF stimulation [10-12, 35, 60], as also observed in murine bone marrow cells [55]. Gradual sensitization of the progeny to TNF-mediated apoptosis along the process of differentiation is emphasized by the decrease in colony size in cultures continuously exposed to this cytokine. Both Fas and TNF-R1 receptors, but not TRAIL [40], are negative regulators of expanding clones in the human [4, 5, 10, 14, 28, 29, 32, 36] and murine hematopoietic systems [6, 7, 33, 39, 46]. Consistent with the marked variations in susceptibility to spontaneous apoptosis between UCB and mPB cells, stimulation of myeloid activity in the latter (16 hours) was shorter than the period required to activate UCB cells (24–48 hours). It is likely that the shorter time to apoptosis and stimulation of mPB cells is caused by an intrinsic state of activation following stimulation of the bone marrow with G-CSF.

TNF-R1 is the dominant receptor involved in transduction of both survival signals in progenitors and apoptotic signals in the differentiated progeny. Whereas TNF-R1 has been considered to suppress the activity of UCB progenitors [8, 14, 15, 30, 33], this receptor also mediates synergism with GM-CSF in induction of dendritic cell differentiation [61] and is shown here to stimulate the clonogenic activity of lineage-negative UCB cells. There was no detectable role of TNF-R2 in our studies, although this receptor participates in transduction of apoptotic signals under growth factor stimulation [6] as well as tropic signaling in erythroleukemic cell lines [8]. The function of TNF-R2 remains to be determined in view of the severely impaired durable engraftment of donor cells deficient in either one of the TNF receptors [39]. This receptor was acutely upregulated during short incubation of mPB cells, possibly similar to release of cytosolic vesicles under stress as described for TNF-R1 in granulocytes [62]. The TNF receptors are not redundant, as neutralization of TNF-R1 was not functionally compensated by TNF-R2, consistent with prior observations in murine bone marrow cells [39]. In variance from dual activity of TNF-R1, complementary activity has been observed for the TRAIL receptors, where TRAIL-R1 (DR4) is endowed with dominant tropic activity and TRAIL-R2 (DR5) has primarily suppressive function in UCB cells [40]. Differential transduction of apoptotic signals by the TNF receptors in studies that use exogenous supplementation of the ligand may be partially related to differential affinities: soluble TNF-α activates preferentially TNF-R1 [63], whereas TNF-R2 is activated predominantly by membrane-bound TNF-α [64].

TNF-R1 is primarily associated with an intracellular death domain that recruits the apoptotic adaptor molecule Fas-associated death domains (FADD), whereas TNF-R2 is primarily associated with the TNFR-associated factor-2 [65–67]. Survival of human UCB hematopoietic progenitors under TNF-R1 signaling is attributed primarily to activation of the NFκB pathway and modulation of caspase-8 activity [16-18, 20-22, 68-70]. The NFκB signaling pathway and caspase-8 are inherently upregulated at the transcriptional level in UCB progenitors insensitive to Fas-mediated apoptosis (Mizrahi et al. unpublished data), where NFκB is considered to be an intrinsic mechanism of protection from apoptosis, in addition to activation of this pathway by TNF-α [65–70]. Although prior studies have shown that complex signaling through TNF-R1 [66] results in downregulation of caspase-8 activity through the inhibitory effects of c-FLIP [21, 22], cIAP1/2 [23], and the NFκB pathway [67, 70], we found involvement of caspase-8 in the stimulatory activity of that the activity of TNF-α. Tropic signaling mediated by TNF-R1 involves activation of caspase-8, similar to progenitor stimulation by TRAIL [71], in hematopoietic progenitors that are inherently resistant to apoptotic signaling [40]. One of the species-specific characteristics of tropic signaling mediated by TNF-R1 through caspase-8 in human UCB cells is the additional involvement of caspase-3 in murine bone marrow cell activation [55].


In summary, TNF-α is a modulator of hematopoietic activity in fresh UCB and cryopreserved mPB samples, with significant tropic activity mediated by TNF-R1 and transduced through caspase-8. Since TNF-α is involved in the pathophysiology of graft versus host disease, neutralization of the cytokine has been considered as a therapeutic option. Our data suggest that neutralization of TNF-α in the early stages of engraftment should be carefully considered in view of the small yet significant supportive effects of this cytokine on hematopoietic progenitor activity. Hematopoietic cell transplants rather benefit enhanced myeloid progenitor function induced by exposure to TNF-α ex vivo, which can alleviate the limitations of UCB as an excellent source of donor cells: small numbers of progenitors and delayed engraftment. As a major injury and inflammatory cytokine, TNF-α is demonstrated to play a significant role in hematopoietic reconstitution under stress conditions.


This work was funded by grants from the Frankel Trust for Experimental Bone Marrow Transplantation. We thank Ela Zuzovsky and Ana Zemliansky for their outstanding technical support.


The authors indicate no potential conflicts of interest.