Fibroblast nemosis arrests growth and induces differentiation of human leukemia cells



Interactive signaling between cancer cells and stroma plays an important role in determining tumor development. We recently found tumor cell-derived factors to induce fibroblast clustering, and that the high amounts of hepatocyte growth factor/scatter factor (HGF/SF) produced by these cell–cell contact-activated fibroblasts enhanced the invasiveness of c-Met-expressing cancer cells. We now observed that leukemia cells lacking c-Met respond to this novel type of fibroblast activation, nemosis, with growth arrest and differentiation to a dendritic cell-like phenotype. This effect was counteracted by introduction of c-Met expression into these cells. Moreover, those leukemia cell lines harboring properly processed c-Met showed no effect in response to nemosis. We propose this effect to be mediated by nemosis-derived cytokine signals, and present the potential candidates induced in the nemotic fibroblasts: interleukins-1, -6, -8, -11, leukemia inhibitory factor and granulocyte-macrophage-colony-stimulating factor. Our results emphasize the role of activated stromal fibroblasts in controlling progression of certain hematologic malignancies in a c-Met expression-dependent manner. © 2007 Wiley-Liss, Inc.

Control over cell differentiation and proliferation requires the complex interactions of signals to be orchestrated both spatially and temporally. The majority of these developmental cues originate from the mesenchyme surrounding the precursor cells. In mesenchymal tissues, fibroblasts are ubiquitous sentinel cells1, 2 that modulate a series of developmental and pathologic conditions ranging from cell differentiation and organogenesis to inflammation and cancer.2, 3 Being the major stromal cellular constituents, fibroblasts play a dominant role by controlling differentiation and proliferation of hematopoietic precursors,4 and are a rich source of several factors governing hematopoiesis.5 Proper maturation of leukocytes requires strict control over the prevailing proliferative activity of the immature blasts, and is achievable only by a reciprocal complex interaction with their surrounding mesenchyme.4–6 Leukemic cells have, however, lost their ability to translate these regulating signals properly, and prefer an undifferentiated phenotype with sustained intense proliferation.7–9

The hepatocyte growth factor/scatter factor (HGF/SF)-c-Met-pathway is a major regulator of tumor-stromal interactions.10, 11 HGF/SF is a stroma-derived paracrine mediator the effects of which are transmitted via the receptor tyrosine kinase c-Met on target cells.12 Stimulation of this pathway increases cell proliferation and motility, induces morphogenesis, and is thus implicated in organogenesis, regeneration, wound healing, tumor cell invasiveness, metastasis and cancer progression. In addition to its several multifunctional roles as a mitogen, motogen and morphogen,12 HGF/SF is a regulator of hematopoiesis, as well.13, 14 Cancer cells exhibit variable expression of c-Met thus rendering them differently prone for stimulation by stroma-derived HGF/SF.15 A lack of c-Met expression or improper processing of the precursor protein in tumor cells thus makes them more liable to other stimuli, and leads to alternate translation of stromal signals within the tumor microenvironment.

In fibroblasts we recently found a novel biological process that was triggered by cell–cell contacts.16 The contact-activated cells were characterized by massive induction of genes such as cyclooxygenase-2 and HGF/SF.16, 17 On the basis of unique features showing exclusively proinflammatory activity of this distinct type of fibroblast activation by biological means, we designated this process nemosis.17 We subsequently found that exposure to these nemotic fibroblasts dramatically enhanced tumor cell invasiveness. We demonstrated this effect to be exclusively mediated by HGF/SF via a transient phosphorylation of c-Met, detectable only when this receptor underwent proper processing in the tumor cells.17

On the basis of our earlier findings on profuse induction of HGF/SF, we now analyzed the effect of nemosis on hematologic malignancies differently expressing the c-Met receptor. We now report that when cell lines differed in c-Met expression they responded to fibroblast nemosis differently. The c-Met-negative cell lines responded with discernible growth arrest, chemotaxis, and differentiation, whereas the c-Met-positive cells remained unresponsive. We therefore next investigated the extent of secretion of hematopoiesis-associated cytokines from nemotic fibroblasts, and provide here the first insight into the intracellular pathways activated by these fibroblast nemosis-derived signals, in addition to the HGF/SF signal.

Material and methods


Antibodies for immunoblotting were rabbit anti-p38 antibody (Ab) (sc-535, Santa Cruz Biotechnology, Santa Cruz, CA), mouse anti-p-p38 Tyr182 monoclonal antibody (MAb) (sc-7973), rabbit anti-JNK Ab (CST-0252, Cell Signaling Technology, Danvers, MA), mouse anti-p-JNK Thr183/Tyr185 MAb (sc-6254), rabbit anti-ERK1/2 Ab (sc-94), mouse anti-p-ERK1/2 Tyr204 MAb (sc-7383), rabbit anti-Akt Ab (CST-9272), rabbit anti-p-Akt Ser473 Ab (CST-9271), rabbit anti-JAK1 Ab (sc-7228), rabbit anti-JAK2 Ab (sc-294), rabbit anti-JAK3 Ab (sc-513), rabbit anti-TYK2 Ab (sc-169), rabbit anti-cleaved caspase-3 Asp175 Ab (CST-9661), mouse anti-full-length caspase-3 Ab (sc-7272), mouse anti-caspase-8 Ab detecting both full length and active fragments (CST-9746), rabbit anti-caspase-9 Ab detecting both full length and 35/37 kDa cleaved fragments (CST-9502), rabbit anti-PARP Ab (CST-9542), rabbit anti-BclxL Ab (sc-7195), goat anti-actin Ab (sc-1615), mouse anti-Bax MAb (sc-7480), mouse anti-Bcl-2 MAb (sc-509), goat anti-COX-1 Ab (sc-1752) and goat anti-COX-2 Ab (sc-1746). Indomethacin (I7378) was from Sigma (St. Louis, MO) and the NS-398 (No. 70590) from Cayman Chemical (Ann Arbor, MI).

Antibodies for flow cytometry (FACS) from the Beckman Coulter Company (Miami, FL) were: anti-CD1a-PE (IM1942), anti-CD3-PE (IM1282), anti-CD10-PE (IM1915), anti-CD11a-FITC (IM0860), anti-CD11b-PE (IM2581), anti-CD11c-PE (IM1760), anti-CD13-PE (IM1427), anti-CD14-FITC (IM0645), anti-CD15-FITC (IM1423), anti-CD16-FITC (IM0814), anti-CD28-FITC (IM1236), anti-CD33-PC5 (IM2647), anti-CD34-PC5 (IM2648), anti-CD38-FITC (IM0775), anti-CD40-PE (IM1936), anti-CD41-FITC (IM0649), anti-CD45-FITC (IM0782), anti-CD45RA-FITC (IM0584), anti-CD45RO-PE (IM1307), anti-CD49d-FITC (IM1404), anti-CD49e-FITC (IM1854), anti-CD51-FITC (IM1855), anti-CD54-PE (IM1239), anti-CD61-FITC (IM1758), anti-CD80-FITC (IM1853), anti-CD83-PC5 (IM3240), anti-CD86-PE (IM2729), anti-CD117-PC5 (IM2657), anti-CD152-PE (IM2282) and anti-HLA-DR-FITC (IM1638). The anti-CD68-FITC (GM-4152) was from Caltag Laboratories (Burlingame, CA).

Cell cultures

Cultures of foreskin-derived human fibroblasts, HFSF-132, were used from passages 7 to 15 as described.16 KG-1, THP-1, U-937, K562, Jurkat and Raji were from the American Type Culture Collection (ATCC, Manassas, VA). All cells were cultured in RPMI 1640 (Life Technologies, Paisley, Scotland) supplemented with 10% fetal bovine serum (Life Technologies), 100 Ag/mL streptomycin and 100 units/mL penicillin.

Spheroid formation was initiated as described by Bizik et al. (2004). Briefly, U-bottom 96-well plates (Costar, Cambridge, MA) were treated with 0.8% LE agarose (BioWhittaker, Rockland, ME) prepared in sterile water to form a thin film of a nonadhesive surface. Fibroblasts were detached from culture dishes by trypsin/ EDTA, and a single cell suspension (4 × 104 cells/mL) was prepared in a complete culture medium. To initiate spheroid formation, 250-mL aliquots were seeded into individual wells and the dishes incubated at +37°C in a 5% CO2 atmosphere.

For the coculture and nemosis stimulation experiments, the leukemia cells were cultured for various time-periods with 24-hr-preformed fibroblast spheroids at a 1:1 leukemia cells:fibroblast ratio. After incubation, the leukemia cells were separated from multicellular spheroids by gravity based on the much higher density of multicellular aggregates as compared to leukemic cells/monocytes. The remaining supernatant contained leukemic cells whereas the spheroids remained in the pellet. For the estimation of growth curves, cell numbers were evaluated by cell-counting in Bürker chambers. For immunoblotting, FACS, and adherence testing, the residual spheroids were removed from cocultures by gravitational differential sedimentation.

Morphology of leukemic cells 96 hr after coculturing was evaluated by phase contrast microscn;well plates as cocultures of 24-hr-preformed fibroblast spheroids with thopy. The leukemic cells' adherence was estimated after 96 hr of coculturing with fibroblast spheroids. Thereafter aliquots of cell lines were seeded onto standard cell-culture dishes for 24 hr. The cultures were washed, and adherent cells were harvested by trypsinization, were counted, and the percentage of these adherent cells was calculated.

Chemotaxis of leukemic cells was performed in agarose-treated 6-well plates as cocultures of 24-hr-preformed fibroblast spheroids with the naïve leukemia cell lines. We calculated with an ocular grid the number of leukemic cells located at a distance from the spheroid double its own diameter, and measured these cells around 15 spheroids per well.

Lentiviral vector transduction

293FT cells were transfected together with pRRLsinPPT. CMV.MCS.METwt.Wpreplasmid18 and second-generation packaging vectors pHCMV-G and pCMVΔ8.91 using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). After 72 h, the supernatant was collected, the viral particles were concentrated using ultracentrifugation (26,000g for 1.5 h), and were resuspended in PBS. GFP-expression lentivirus (pLV-PGK/GFP) used as control was a kind gift from professor Seppo Ylä-Herttuala (AIV-Institute, Kuopio, Finland). Virus stocks were stored in −70°C until transduction.

THP-1 cells were seeded (5 × 105 cells/well) into a 6-well plate and transduced with 1:20 dilution of virus concentrate and titer of 1.4 × 107 in the presence of polybrene (8 μg/mL). After 16 h, virus-containing medium was removed, cell were washed and resuspended in normal growth medium for experimentation.

Cell cycle analysis

For DNA histograms and cell cycle analyses, leukemic cells were cocultured for 96 hr with 24-hr-preformed fibroblast spheroids, and separated from fibroblast clusters by sedimentation. These cells were then washed with PBS and fixed in 1% paraformaldehyde, were treated with RNAse (100 μg/mL), their DNA was stained with propidium iodine (50 μg/mL), and they were analyzed by FACS.


Cell samples were lysed directly in SDS-PAGE sample-loading buffer: 62.5 mmol/L Tris-HCl (pH 6.8), 2% SDS, 20% glycerol, 5% β-mercaptoethanol and 0.005% bromophenol blue, supplemented with Complete Mini-protease inhibitor mixture tablets (Roche, Mannheim, Germany) and boiled for 5 min. Lysates were centrifuged at 14,000 rpm for 15 min to sediment particulate-insoluble material. These samples were separated in SDS-PAGE (gradient of polyacrylamide 5–15%, 3.5% stacking gel). The proteins were transferred electrophoretically from the gel to a nitrocellulose membrane (Schleicher & Schuell, Dassel, Germany), with transfer efficiency verified by Ponceau-S staining. After blocking of the membrane with 2.5% low-fat dry milk in TBS, 20 mmol/L Tris-HCl, 150 mmol/L NaCl and 0.1% Tween 20 at pH 7.5, it was incubated with specific primary antibodies, followed by an alkaline phosphatase-conjugated secondary antibody (Promega, Madison, WI). Protein bands were visualized according to manufacturer's recommendations.

Flow cytometry

For flow cytometric analysis, the leukemia cells cocultured for indicated time points and after differential sedimentation to remove spheroids were incubated on ice with antigen-specific antibodies or with isotype-matched antibodies as controls, and fixed in 1% paraformaldehyde. FACS analysis was done by an EPICS ALTRA flow cytometer with the EXPO32 analysis program (both from Beckman Coulter, Fullerton, CA).

Measurements of cytokine concentrations by enzyme-linked immunoassays

Fibroblast spheroid-conditioned medium was collected at 96 hr after initiation of spheroid formation from the 96-well plates. Concentrations of IL-1β, IL-6, IL-8, IL-11, GM-CSF, LIF, oncostatin M and TNF-α were quantified by commercial ELISA kits and reagents according to manufacturers' instructions. The human IL-1β, human IL-11, human LIF, human TNF-α and human oncostatin M ELISAs were from R&D Systems (Minneapolis, MN), the human IL-6 and human IL-8 ELISAs were from the Central Laboratory of the Netherlands Red Cross (CLB, Amsterdam). Cytokine quantification in the nemotic fibroblast-conditioned medium for human IL-2, IL-4, IL-5, IL-10, IL-12, IL-13, GM-CSF, interferon-γ (IFN-γ) and TNF-α was carried out with the Bio-Plex Human Cytokine Th1/Th2 Panel (Bio-Rad Laboratories, Hercules, CA, catalogue number 171-A11081), by the Luminex 100 System (Luminex, Austin, TX).


c-Met expression in leukemia cells, their growth characteristics and cell cycle analysis

In analysis of leukemia cell lines for their expression of the HGF/SF receptor c-Met, expression of the properly processed form appeared on U-937, Jurkat, Raji and K562 cell lines but not on THP-1 or KG-1 cells (Figure 1a). All these cells were cocultured with nemotic fibroblasts for their growth characteristics. The cell lines THP-1 and KG-1 lacking c-Met responded with discernible growth arrest, whereas the cell lines expressing c-Met showed no significant alterations in their proliferation rates. For subsequent experiments on stimulation with nemotic fibroblasts, we chose the c-Met-negative THP-1 and KG-1 cell lines, and the c-Met-positive U-937 cell line as control.

Figure 1.

c-Met expression in leukemia cell lines, and growth characteristics of selected cell lines subjected to nemosis. (a) Expression of the c-Met receptor in leukemia cell lines KG-1, THP-1, U-937, Jurkat, Raji and K562. Arrow indicates position of the properly processed form of c-Met (145 kDa). (b) Proliferation kinetics of leukemia cell lines KG-1, THP-1 and U-937 with and without stimulation by nemotic fibroblast spheroids in coculture. *p < 0.05, **p < 0.01, ***p < 0.001 between treatments at indicated time-points. (c) DNA histogram data and percentage of cells as divided into cell cycle G0G1, G2 and S phases of leukemia cell lines KG-1, THP-1 and U-937 with (closed bars) and without (open bars) stimulation by nemotic fibroblast spheroids in coculture for 96 hr.

Cell cycle analysis of KG-1, THP-1 and U-937 cells

Nemotic fibroblasts induced a dramatic growth inhibition of the cells lacking c-Met, whereas growth of the c-Met-positive cell lines showed only marginal, if any, inhibition (Figure 1b). Attenuation of growth by nemosis-derived signals was evident at 72 hr for the KG-1 cells, but was already evident at 48 hr in the THP-1 cell line, suggesting enhanced sensitivity to nemosis of the latter cells. In the control U-937 cells only a modest and delayed effect on proliferation was apparent after 96 hr of incubation. The nemosis-arrested proliferation of the responder cell lines persisted throughout the study. With the control cells reaching their growth plateau at 168 hr, nemosis inhibited the proliferation of the cell lines by 67% for KG-1, 83% for THP-1 and 6% for U-937 cells.

Figure 1c shows the leukemic cell lines' cell-cycle phase distribution as evaluated by DNA histograms. When treated with nemosis, 30.2% of the KG-1 and 31.3% of the THP-1 cell populations accumulated in the G0G1 phase. This shift away from the M and S phases indicates cell-cycle arrest associated with differentiation. The U-937 cells retained their cell-cycle characteristics, with no population shifts in response to fibroblast nemosis. Moreover, treatment of clustering fibroblasts with the nonsteroidal anti-inflammatory cylooxygenase-inhibitors, NS-398 and indomethacin, to prevent prostaglandin production16 had no effect on inhibition of proliferation by nemosis (data not shown).

c-Met expression and dependence of nemosis response

Stimulation of the monocytoid leukemic cells by nemosis did not alter their expression levels of c-Met (Fig. 2a). To show dependency of the nemosis effect on the HGF/SF-c-Met-pathway, we introduced wild type human c-Met expression into THP-1 cells using a lentiviral vector, and nemosis response in terms of growth arrest to GFP-transduced control cells. Receptor expression was evident only in the c-Met-transduced cells. Increased expression of c-Met persisted throughout the 5 days of experimentation (Fig. 2b) for growth curve analysis of cells subjected to fibroblast nemosis. Although the intensity of c-Met expression declined during the experiment, the c-Met-transduced cells resisted growth arrest whereas the GFP-transduced cells entered growth arrest (Fig. 2c). The nemosis-uninfluenced GFP and c-Met expressing cells exhibited similar growth characteristics (Fig. 2c).

Figure 2.

Lentiviral vector transduction of c-Met into THP-1 cells. (a) Expression of c-Met in the leukemia cell lines KG-1, THP-1 and U-937 stimulated for 96 hr by nemotic fibroblast spheroids (+) and without stimulation (−). (b) c-Met expression kinetics in GFP-control and c-Met-transduced THP-1 cells at indicated time points. (c) Proliferation kinetics of THP-1 cells transduced with GFP-control or c-Met lentiviral vector with or without stimulation by nemotic fibroblast spheroids in coculture. *p < 0.05, **p < 0.01, ***p < 0.001 between treatments at indicated time-points.

Morphological and functional characteristics of leukemic cells in response to nemosis

Cell cycle arrest at the G0G1 phase is associated with induction of differentiation.19 In KG-1 and THP-1 cells, nemosis led to an increased proportion of adherent cells by 19.8 and 31.6% (Fig. 3a). These cells showed morphological features of a dendritic-cell-like phenotype with cell elongation and formation of stellate pseudopodia (Fig. 3b), but in response to nemosis, the U-937 cells changed neither their morphology nor their pattern of adherence (Fig. 3b). Induction of an adherent phenotype in the nemosis-responsive cell lines was associated with increased expression of intercellular adhesion molecule-1 (ICAM-1) (Fig. 3c).

Figure 3.

Adherence, morphology and chemotactic response of the leukemic cells to nemosis. (a) Percentage of cells from the total cell population adhering to culture dish after coculture stimulation by nemotic fibroblasts as compared to unstimulated cells of leukemia cell lines KG-1, THP-1 and U-937. ***p < 0.001 compared to respective control cells. (b) Morphology of adherent cells with or without stimulation by nemotic fibroblast spheroids. Cell elongation and presence of pseudopodia evident in stimulated KG-1 and THP-1 cells whereas, after stimulation, U-937 cells retain their phenotype. (c) Expression of intercellular adhesion molecule-1 (ICAM-1) in KG-1, THP-1 and U-937 cells with (+) or without (−) stimulation by nemotic fibroblast spheroids in coculture. Increased ICAM-1 expression of the nemosis-responsive cell lines KG-1 and THP-1 associated with increased adherence to the cell culture dish. (d) Quantification and morphology of leukemia-cell (KG-1, THP-1, U-937) chemotactic movement towards fibroblast clusters in coculture. Increased chemotactic accumulation of KG-1 and THP-1 cells visible around a nemotic spheroid, whereas U-937 cells are unresponsive. ***p < 0.001 compared to U-397 cells.

Increased adherence of KG-1 and THP-1 cells by nemosis suggests that the clustered fibroblasts can also produce factors affecting cell motility and chemotaxis. We therefore evaluated the chemotactic responses of KG-1, THP-1 and U-937 cells in coculture with the nemotic fibroblasts. Comparison of the responses of the analyzed cell lines showed that both KG-1 and THP-1 were chemotactically drawn towards the fibroblast clusters undergoing nemosis, whereas the U-937 cells were unresponsive (Fig. 3d). Compared to the U-937 cells, nemosis attracted the KG-1 and THP-1 cells to accumulate in the vicinity of the fibroblast clusters at an 11- and a 22-fold enhanced density.

Since monocyte maturation and differentiation are associated with decreased antigen uptake through macropinocytosis,20 we studied, as a parameter of differentiation on a functional scale, the effect of nemosis on pinocytotic activity. The ability of the leukemia cell lines to repel FITC-labeled dextran was assessed at 24 and 96 hr of coincubation of leukemia cells with nemotic spheroids. After stimulation of the cell lines with nemosis followed by incubation with FITC-dextran, flow cytometry revealed a distinct inhibition of FITC-dextran uptake in nemosis-treated KG-1 (mean intensity change −11.92) and THP-1 cells (mean intensity change −3.86), with the U397 cells showing an increase of 0.91 in intensity (data not shown). These data are in accordance with morphological characterization showing induction of adherence and the presence of dendritic cell-like pseudopodia on the nemosis-responsive KG-1 and THP-1 cells.

Changes in surface antigen expression of nemosis-stimulated cell lines

The phenotypic characterization of nemosis-treated cells was carried out by FACS analysis of the cell-surface antigens, as shown in the Supplementary Table, comparing control antigen intensity to that of nemosis-treated cells. From this analysis, a clear induction pattern of 5 cell-surface markers emerged. Interestingly, and in accordance with the other cell responses, these antigens were induced only in KG-1 and THP-1 cells. No induction of these surface antigens was evident in U-937 cells, which responded with an overall expressional down-regulation.

In the nemosis-responsive cell lines we identified induction of the following: dendritic cell marker CD11c, the leukocyte common antigen CD45RA, the adhesion molecule CD54, the dendritic cell-associated T-cell costimulatory molecule CD86, and the membrane peptidase CD13. The time-dependence of CD86 induction (Fig. 4) showed response kinetics similar to the growth arrest in KG-1 and THP-1 cells, with the latter cell line reacting more promptly also by this parameter. No effect on the induction of CD86 occurred when, prior to coculture with the leukemia cells, fibroblast clusters were formed in the presence of NS-398 or indomethacin (data not shown).

Figure 4.

Time-dependence of CD86 surface antigen expression. CD86 expression evaluated by FACS at indicated time points for leukemia cell lines KG-1, THP-1 and U-937 in coculture with nemotic fibroblast spheroids.

Further population analysis was carried out based on differential expression of CD45 in various lineages and differentiation stages.21 By gating on CD45 new subpopulations emerged in nemosis-treated KG-1 and THP-1 cells (Fig. 5), but no changes occurred in response to nemosis in the subpopulation characteristics of U-937 cells. The emerging populations were positive for all the nemosis-induced markers, especially for CD11c and CD13, in contrast to a nonresponsive similar CD45-positive population of U-937 cells. Nemosis increased CD11c and CD86 positivity in a CD45-positive population with low SSC values of KG-1 and THP-1 cells. Considered together, these results suggest that in responsive cell lines nemosis induces expression of antigens involved in antigen presentation and T-cell stimulation, along with dendritic cell characteristics.

Figure 5.

Cell population analysis of leukemia cell lines KG-1, THP-1 and U-937 based on expression of CD45. Subpopulation analysis based on CD45 expression and divided into dual-labeling of cells for expression of indicated surface antigens (CD11c, CD86, CD54, CD13). Top figures show expression of CD45 at baseline and with cells stimulated with nemotic fibroblasts in coculture. On the basis of CD45 expression, untreated U-937 cells show 2 populations with high and low granularity and show no change after treatment, whereas in KG-1 and THP-1 cells a new emerging population with high granularity is evident after stimulation with nemotic fibroblast spheroids. In these nemosis-responsive cell lines, expression of CD11c and CD86 is enhanced also in the cell population with low granularity.

Cytokine production in nemotic fibroblasts

We previously reported that nemotic fibroblasts are an ample source of the c-Met ligand HGF/SF.17 As present data indicates growth arrest and differentiation of leukemia cells in response to fibroblast nemosis to be overcome by HGF/SF-c-Met-signaling, we evaluated a pattern of cytokines known to be associated with modulation of chemotaxis and leukemia cell proliferation. We found induced release of interleukin(IL)-1β, IL-6, IL-8, IL-11, granulocyte-macrophage colony-stimulating factor (GM-CSF), and leukemia inhibitory factor (LIF) from nemotic spheroids compared to the corresponding monolayer cultures at 96 hr from culture initiation (Fig. 6). The levels of IL-2, IL-4, IL-5, IL-10, IL-12, IL-13, interferon-γ, oncostatin M and TNF-α remained low or undetectable, and were unaffected by cell culture arrangement. The cytokines most abundantly produced by the nemotic fibroblasts were IL-6 and IL-8, with fold-inductions (mean production in spheroids) of 3.7 (25.4 ng/mL) and of 8.0 (134.2 ng/mL) as compared to the corresponding monolayer cultures. Production of IL-1β and LIF was induced 18.3 and 3.8 fold in the nemotic fibroblasts, which also produced GM-CSF undetectable from monolayer cultures (Fig. 6). Our results thus suggest that nemosis is a fundamental source of an array of proinflammatory cytokines and growth factors regulating monocyte functions.

Figure 6.

Hematopoiesis-associated cytokine release from fibroblast spheroid clusters and monolayers. Analysis of cytokine (interleukin IL-1, IL-6, IL-8, IL-11, granulocyte-macrophage colony-stimulating factor, GM-CSF and leukemia inhibitory factor, LIF) production from fibroblasts cultured as spheroids or as monolayers at the corresponding time (96 hr) after spheroid formation. ***p < 0.001 between groups.

Apoptosis-related intracellular changes in the leukemia cell lines by nemosis

Inhibition of tumor cell growth is usually accompanied by induction of apoptosis. We therefore evaluated the known apoptotic pathways in the growth-arrest and differentiation responses of THP-1 and KG-1 cells to nemosis. In Figure 7a, expression of several apoptosis-associated proteins revealed that the cleaved, active form of the universal apoptosis executor, caspase-3, occurs in response to nemosis only in the nemosis-responsive cell lines. That the unresponsive U-937 showed no effect suggests activation of apoptosis in the nemosis-responsive cells. No changes in the expression of the active cleaved forms of the initiator caspases -8 and -9 were detectable in any of the cell lines, but expression of the full-length caspase-8 was induced by nemosis in the THP-1 and KG-1 cells. In this case, as well, the U-937 cells showed no effect. The increases in caspase-3 cleavage and caspase-8 expression were not, however, reflected in the expression of the apoptosis-regulating Bcl-xL, Bcl-2, or Bax proteins, suggesting their apoptosis-unrelated mechanism of action.

Figure 7.

Immunoblot analysis of apoptosis-related and intracellular signaling proteins in leukemia cells. (a) Expression of apoptosis-related molecules in leukemia cell lines KG-1, THP-1 and U-937 with (+) or without (−) stimulation by nemotic fibroblast spheroids in coculture for 96 hr. In nemosis-responsive cell lines KG-1 and THP-1, activation-associated cleavage of caspase-3 and -8 is evident. No cleavage products of caspase-9 and reduced expression of the cleaved form of poly(ADP-ribose)polymerase (PARP) are visible. Phenotype differences between nemosis-responders and the nemosis-unresponsive cell line U-937 evident in expression levels of the pro-apoptotic Bax protein. (b) Expression of phosphorylated and total levels of mitogen-activated protein kinases (MAPK) p38, JNK, ERK1/2 and the Akt kinase in comparison to expression of the differentiation-associated Janus-kinase family JAK1, JAK2, JAK3 and TYK2 in leukemia cell lines KG-1, THP-1 and U-937, stimulated (+) or unstimulated (−) with nemotic fibroblast spheroids. Increased dephosphorylation of p38 and ERK1/2 is evident in the nemosis-responsive cell lines KG-1 and THP-1 together with increased expression of JAK1 and JAK3. The nemosis-unresponsive cell line U-937 showed no expressional differences for these proteins.

Because of the evident cleavage and increased expression of the active form of caspase-3 caused by nemosis in THP-1 and KG-1 cells, we evaluated the extent of poly-ADP-ribose-polymerase (PARP) cleavage associated with DNA damage, apoptosis and caspase-3 activity.22 Figure 7a shows the expression pattern of full-length PARP (p116) and its cleaved inactive form p89 in leukemia cell lines subjected to nemosis.

Effects of nemosis on leukemia cells' intracellular signaling cascades

We evaluated the involvement of MAPKs c-jun N-terminal kinase (JNK), extracellular signal-regulated kinases p44/p42 (ERK1/2) and p38 in leukemia cell responsiveness to nemosis (Fig. 7b). In all the naïve leukemia cell lines, we found constitutive phosphorylation of these kinases, reflecting their mitotically active phenotype. In the KG-1 and THP-1 cell lines, which responded to nemosis with growth arrest and differentiation, phosphorylation of p38 MAPK was dramatically quenched, whereas no changes in p38 MAPK phosphorylation in response to nemosis occurred in the U-937 cell line. A similar but less pronounced inhibition of phosphorylation was evident in ERK1/2, with the U-937 cells again left unresponsive to nemosis. The level of phosphorylation and expression of JNK in all cell lines remained relatively unchanged. In the KG-1 and THP-1 cells induced to undergo differentiation by fibroblast nemosis, expression of JAK1 and JAK3 was increased. In the nonresponsive U-937 cell line, expression of JAK1 was downregulated, with no visible expression of JAK3 (Fig. 7b). Phenotypic differences between U-937 and the nemosis-responsive KG-1 and THP-1 cells were reflected in TYK2 as well as in JAK3 expression. JAK2 expression correlated neither with growth arrest nor with differentiation (Fig. 7b).


Nemotic fibroblasts, activated by cell–cell contacts, induce a growth inhibitory and differentiating response in leukemia cells specifically lacking c-Met. This effect is not seen in cells expressing c-Met, and can be counteracted by introducing c-Met to the responsive cells suggesting that HGF/SF-c-Met signaling can be utilized by leukemic cells to evade stromal growth control signals. The nonreactivity in terms of cell function, morphology and antigen expression of c-Met expressing leukemia cells suggests also that alternate mechanisms for motility, chemotaxis and cell differentiation are used by monocytes in contrast to epithelial cells, and demonstrates an alternate strategy for utilization of c-Met in tumor cells derived thereof. Moreover, we show that in nemosis, a novel type of stromal cell biological reactivity, production of inflammation-, cell growth- and differentiation-associated cytokines IL-1β, IL-6, IL-8, IL-11, LIF and GM-CSF is induced. It must be stressed that neither in this study nor in our previous or ongoing work have we found any induction of anti-inflammatory cytokines in nemosis. This suggests that fibroblast cell-cell contacts activate a biologically diverse, yet directed, paracrine signaling cascade. Furthermore, our results suggest that translation of stroma-derived nemosis signaling with massive induction of signaling molecules within a given cell microenvironment is largely determined by the phenotype and receptor-expression profile of the target cells.

Tumor cells, showing an imbalance between cell survival and death, prefer a growth promoting undifferentiated phenotype. We show that leukemic cells responding with growth arrest also underwent differentiation when subjected to fibroblast nemosis. Using FACS-gating on CD45, the leukocyte common antigen, we found in the nemosis-responder cell lines emerging new populations with enhanced expression of CD11c and CD86. CD54 and CD13 cell-surface antigens were also increased in the emerging high-granular population suggesting that the leukemic cells influenced with nemosis gained a phenotype reminiscent of antigen-presenting cells.

Expression of the CD11c, a marker for myeloid dendritic cells23 associated with differentiation and maturation,24 was coinduced with CD86 in the nemosis-responders. Similar to CD11c, expression of CD86 on leukemia cells is associated with a dendritic cell-like phenotype.25 In addition to CD11c, increased adherence of the growth-arrested cells is explained by increased expression of CD54 (ICAM-1), which mediates adhesive interactions, for example by binding to the β2 subfamily integrins CD11a/CD18 (LFA-1) and CD11b/CD18 (Mac-1).26 Similar to CD11c, also ICAM-1 is required for leukocyte migration and, like CD86, ICAM-1 binding functions as a costimulatory signal for the activation of T cells in antigen presentation.27

We extended our evaluation of nemosis-induced leukemia cell growth arrest and differentiation to the molecular intracellular level by evaluating the involvement of the major pathways associated with growth arrest and differentiation. We found intensive cleavage of the executor caspase-3, in contrast to unchanged expression of Bcl-2, Bcl-2XL and Bax, accompanied by no changes in cleavage of the initiator caspases-8 and -9. In contrast to the executor caspases requiring proteolytic cleavage for activation, the initiators are also activated without cleavage by oligomerization or dimerization.28 Interestingly, caspases 3 and 8 also regulate differentiation.29 Furthermore, the fact that cell numbers after nemosis treatment remained unchanged implies that the nemosis-responsive phenotype favors differentiation over apoptotic death. Surprisingly, no increased proteolytic processing of PARP, a caspase-3 substrate,22 occurred; this suggests for caspases-3 and -8 an apoptosis-unrelated differentiating function. PARP is associated with DNA repair, with cell proliferation and differentiation, and with transcriptional regulation.22 We never found increases in protein levels of Bax, Bcl-2 and Bcl-2XL to be associated with reduced PARP expression, suggesting an apoptosis-unrelated and differentiation-associated mechanism for PARP downregulation.30 PARP expression and activity may also reflect the changes in mitotic rate of leukemic cells; after cells undergo growth arrest, for example in response to nemosis, they have less need for PARP expression and activity in the nucleus to protect the fragile opened DNA of mitosis.31

The MAPK cascades involving JNK, ERKs and p38s are tightly associated with cell proliferation, differentiation and death.32 Despite no changes in JNK expression or phosphorylation, the phosphorylation of p38 MAPK–and to a minor extent also that of ERK1/2–was significantly suppressed in cells responding to nemosis with differentiation and growth arrest. Active p38 prevents Jurkat T-cell apoptosis,33 and inhibits differentiation of promyelocytic cells.34 This is in agreement with our data showing that inhibition of p38 MAPK phosphorylation occurs only in those cells responding to nemosis with differentation and growth arrest. It thus suggests that the highly proliferating phenotype is associated with active p38 MAPK and may indicate an altered or constitutively active p38 pathway in these cells.

Signal transduction by the Janus protein tyrosine kinase family (JAK) members (JAK1, JAK2, JAK3 and TYK2) is intimately associated with monocyte and leukemia cell differentiation.35 We found induction of the JAK kinases JAK1 and JAK3 in the nemosis-responsive cell lines. This is concordant with the prevailing view of expressional control of JAK136 and JAK337 in hematopoietic differentiation. Induction of the JAKs in the nemosis-responsive cell lines suggests involvement of IL-6 and GM-CSF in mediating the effect of differentiation. The detailed intracellular mechanisms involved require further investigations currently ongoing in our laboratory.

Fibroblast nemosis is a unique novel type of inherent stromal activation that leads to production of a distinct set of paracrine mediators guiding differentiation and growth in a target cell phenotype-dependent manner. Through direct effects on differentiation of hematological tumor cells, in a way dependent on HGF/SF-c-Met pathway, nemosis may influence responses of the immune system to malignancy (Fig. 8). Nemosis, activated by the biological means of fibroblast cell–cell contacts, represents a novel type of stromal reactivity with signals showing proinflammatory, growth and differentiation functions. Differentiation of leukemic cells into the dendritic cell lineage can stimulate anti-leukemic actions of T-cells38, 39; such differentiation can be suggested as immunotherapy.40 Our results present the first in vitro evidence that homotypic stromal cell–cell interactions leading to nemosis can provide sufficient signaling to modulate and restrain neoplastic growth.

Figure 8.

Schematic model summarizing effects of nemosis-derived signals on solid and hematopoietic tumor cells. Schematic presentation of fibroblast nemosis effects and mediators inducing invasiveness of epithelial tumor cells together with arrest of growth and induction of differentiation of leukemic cells in a target-cell phenotype-dependent manner. Dashed lines link possible biological and pathological processes with this cascade. [Color figure can be viewed in the online issue, which is available at]


The authors thank Mona Schoultz for her diligent expert assistance on flow cytometry, Lahja Eurajoki and Alena Kadnarova for their skillful technical assistance, and Libusa Stevulova for preparing the figures on flow cytometry. We thank Jana Jakubikova for performing the Luminex analysis and Carol Norris for author editing the language.

Note Added in Proof

Jozef Bizik, Esko Kankuri, Pertteli Salmenperä, and Antti Vaheri contributed to the discovery of specific gene expression and data on cytokine mRNA. Whereas this data is not presented in the manuscript, it is fully acknowledged as an important source of information to this paper. This data on gene expression and cytokine mRNA and protein profiling have now been submitted for publication (Pertteli Salmenperä et al., submitted; Anna Enzerink et al., submitted).