Translocation of the inhibitor of apoptosis protein c-IAP1 from the nucleus to the Golgi in hematopoietic cells undergoing differentiation: a nuclear export signal-mediated event

The inhibitors of apoptosis proteins (IAPs) have been initially defined as natural cellular inhibitors of cell death. These proteins were identified in baculoviral genome as regulators of host-cell viability during virus infection,¹  and cellular orthologues were subsequently described in yeast, nematodes, drosophila, and mammals. The human genome encodes at least 8 IAPs (X-linked IAP [XIAP], cellular IAP1 [c-IAP1], c-IAP2, melanoma IAP [ML-IAP], neuronal apoptosis inhibitory protein [NAIP], survivin, IAP-like protein 2 [ILP-2], Apollon).²  All of these proteins have in common the presence of 1 to 3 copies of a baculovirus IAP repeat (BIR) domain.¹  These domains are essential for the antiapoptotic properties of the IAPs, which have been attributed to the direct binding and inhibition of caspases. XIAP binds the small subunit of caspase-9 through its BIR3 domain³  and masks the active site of caspase-3 and -7 through a distinct segment, which is immediately amino-terminal to its BIR2 domain.^4,5  c-IAP1 and c-IAP2 bind caspase-3 and -7 but their inhibitory effect on caspases is 2- to 3-log lower than that of XIAP.⁶  Some of the BIR-containing proteins do not have clear links with apoptosis and several members of the family have demonstrated distinct functions including cell cycle regulation,⁷  protein degradation,⁸  and caspase-independent signal transduction.^9-12 

In addition to the BIR domains, several IAPs, including XIAP, c-IAP1, and c-IAP2, contain a highly conserved carboxy-terminal RING finger domain that confers them an enzyme 3 (E3) function in the protein ubiquitylation process. Several proteins specifically targeted for ubiquitylation by IAPs have been identified. At least in vitro, XIAP and c-IAP2 direct the ubiquitylation of caspase-3 and caspase-7,^13,14  whereas c-IAP1 and c-IAP2 mediate ubiquitylation of second mitochondria-derived activator of caspase (Smac)/DIABLO, an antagonist of IAPs.¹⁵  c-IAP1 and c-IAP2 are also components of the type 2 tumor necrosis factor (TNF) receptor complex through interaction with the signaling intermediates TNF receptor-associated factor 1 (TRAF1) and TRAF2.⁹  c-IAP1 could induce the ubiquitylation of TRAF2 and participated in the TNF-α-mediated proteasomal degradation of TRAF2,¹⁶  and c-IAP2 has been involved in the TNF-α signaling leading to nuclear factor-κB (NF-κB) activation.¹⁷ 

The expression and activity of IAPs are regulated at several levels. The transcription factor NF-κB enhances the expression of c-IAP1, c-IAP2, and XIAP, which may contribute to the prosurvival effect exerted in many situations by this transcription factor.^18,19  XIAP translation can be enhanced through the use of an internal ribosomal entry site in the 5′-untranslated region of its messenger RNA.²⁰  IAPs could regulate their own degradation through autoubiquitylation,⁸  whereas the IAP-interacting proteins Smac/DIABLO and Omi/HtrA2 neutralize XIAP and possibly other IAPs when released from the mitochondria under apoptotic stimuli.²¹ 

Another level of regulation of IAP functions is the modulation of their subcellular location. Such a regulation has been described for XIAP whose interaction with the protein XIAP-associated factor 1 (XAF1) induces its sequestration in the nucleus and suppresses its caspase-inhibitory function.²²  The present study demonstrates that c-IAP1 is located in the nucleus of various undifferentiated cells and migrates to the cytoplasm, more specifically to the Golgi apparatus, when these cells undergo differentiation. This redistribution of c-IAP1 involves a nuclear export signal (NES) located in its caspase recruitment domain (CARD). Overexpression of c-IAP1 interferes with 12-O-tetradecanoylphorbol 13-acetate (TPA)-induced differentiation of leukemic cells, a process also inhibited by the nuclear export inhibitor leptomycin B (LMB). Altogether, these observations suggest a role for c-IAP1 in cell differentiation.

We used mouse monoclonal antibodies (mAbs) directed against c-IAP1 (PharMingen, La Jolla, CA); Golgin 97 (clone CDF4; Molecular Probes, Eugene, OR); mitochondrial HSP70 (mHSP70; Affinity BioReagent, Golden, CO); HSC70 (Santa Cruz Biotechnology, Santa Cruz, CA); GM130 (Golgi Matrix protein of 130 kDa; fluorescein isothiocyanate [FITC]-conjugated antibody; Transduction Laboratories, Lexington, KY); rabbit polyclonal Abs targeting c-IAP1 (Santa Cruz Biotechnology and R&D Systems, Abington, United Kingdom); macrophage antigen-1 (Mac-1; phycoerythrin [PE]-conjugated antibody; Pharmingen, Becton Dickinson, Heidelberg, Germany); Bcl-2 (FITC-conjugated antibody; Pharmingen, Becton Dickinson); CD1a (FITC-conjugated antibody; Pharmingen, Becton Dickinson), CD71 (FITC-conjugated antibody; Pharmingen, Becton Dickinson); poly-(adenosine diphosphate-ribose) polymerase (PARP; Boehringer-Mannheim, Mannheim, Germany); XIAP (R&D Systems and Stressgen Biotech, San Diego, CA); protein disulfide isomerase (PDI; Calbiochem, La Jolla, CA); green fluorescent protein (GFP; Invitrogen, Cergy Pontoise, France); and survivin (Novus Biologicals, Littleton, CO). Macrophage colony-stimulating factor (M-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), and interleukin-4 (IL-4) were obtained from R&D Systems; erythropoietin (EPO) was from Amgen (Thousand Oaks, CA); TPA was from Sigma-Aldrich Laboratories (St Quentin Fallavier, France); brefeldin A (BFA) and nocodazole were from Alexis Biochemicals (Lausen, Switzerland); and trypsin-EDTA (ethylenediaminetetraacetic acid) was from Gibco-BRL (Carlsbad, CA). LMB was kindly provided by Dr M. Yoshida (Tokyo, Japan) and thrombopoietin (TPO) was kindly provided by Kirin Brewery (Tokyo, Japan).

Cell lines were obtained from the American Type Culture Collection (ATCC, Rockville, MD) and cultured as described.²³  We also tested the previously described Bcl-2-transfected U937 and HT29 cells and HT29-MTX cells.^23-25  The TPA-resistant variant of U937 cells were kindly provided by Prof P. J. Parker (London, United Kingdom).²⁶  Monocytes from human peripheral blood were obtained with informed consent from healthy donors and 7 patients with chronic myelomonocytic leukemia (CMML) and purified using an isolation kit (Miltenyi Biotec, Paris, France) following the manufacturer's instructions. Cells were differentiated into macrophages or dendritic cells and checked for the expression of differentiation marker CD71 and CD1a as described.²³  Peripheral blood CD34⁺ cells were cultured in liquid conditions in the presence of cytokines to generate megakaryocytes or erythroid cells as described.^27,28  The Bcl-2 transgenic mice were obtained from Irv Weismann (Stanford, CA).²⁹  Bcl-2 overexpression in Mac-1⁺ cells of transgenic mice was verified by flow cytometry using a FACSCalibur cytometer and Cell Quest software (Pharmingen, Becton Dickinson). Femoral bone marrow cells were isolated from 6- to 8-week-old control and transgenic FVB/N female mice and cultured for 4 hours on plastic plates before culturing adherent cells for 6 days in the presence of 10% L929 cell-conditioned medium as source of CSF-1. Macrophage differentiation was assessed by May-Grünwald-Giemsa staining.

Cells were fixed in paraformaldehyde (PFA; 2%) for 10 minutes at room temperature, washed twice, saturated in phosphate-buffered saline (PBS) containing 0.1% saponin and 5% nonfat milk, and incubated overnight at room temperature in the presence of primary Ab diluted in PBS containing 0.1% saponin and 0.5% bovine serum albumin (BSA). After washing, cells were incubated for 30 minutes with 488-alexa goat antirabbit or antimouse Ab (Molecular Probes) and washed 3 times with PBS. Nuclei were stained by Hoechst 33342 (Sigma-Aldrich). To demonstrate colocalization of c-IAP1 with Golgin 97 or GM130, cells were first incubated with anti-c-IAP1 Ab overnight at 4°C, then with the secondary biotinylated-immunoglobulin (Ig; 1:100; Amersham Biosciences, Orsay, France) for 1 hour at room temperature, then with a streptavidin-texas red-conjugated Ab (Molecular Probes; 1:2000) for 1 hour. Cells were subsequently incubated for 1 hour at room temperature with anti-GM130-FITC (1:100) or anti-Golgin 97 (1:100), then with FITC-conjugated antimouse Ab. Fluorescence was preserved using the FluorSave mounting medium (Calbiochem). Analysis was performed using either a fluorescence (Nikon Eclipse 80i; Nikon, Champigny, France) or a confocal (Leica TCS SP2; Leica, Bron, France) microscope (objective × 50; original magnification × 500). The images were captured by a 3 CCD (charge-coupled device) color video camera (Sony, Paris, France), digitally saved using Archimed-Pro software (Microvision Instruments, Evry, France), and further processed using Photoshop software (Adobe Systems France, Paris, France).

Whole-cell lysates and nuclear-free extracts were prepared as described.²³  Nuclear and cytoplasmic fractions were obtained by lysing the cells in lysis buffer (10 mM Hepes [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid], 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA [ethyleneglycoltetraacetic acid], 1 mM DTT [dithiothreitol], 0.6% NP-40 [nonidet P-40]) in the presence of the protease inhibitors. Cell lysate was centrifuged at 1200g for 10 minutes. The supernatant was carefully collected (cytoplasmic fraction [C]) and the pellet was washed once then resuspended in lysis buffer (nuclear fraction [N]). Further cell fractionation was performed as described.³⁰  All fractions were stored at -80°C until Western blotting analysis, and protein concentration was measured using the Bio-Rad DC protein assay kit (Hercules, CA). Western blot experiments were performed as previously described.²³ 

Proteins from reticular/microsomal-enriched fraction were digested by 0.05% trypsin in the presence of 0.02% EDTA for 30 minutes at 37°C and analyzed by Western blotting for c-IAP1 content.³¹ 

Plasmid-enhanced GFP (pEGFP)-c-IAP1 was constructed by subcloning full-length c-IAP1 cDNA (kindly provided by J. C. Reed, La Jolla, CA) into the BglII/SalI site of pEGFP-C1 (Clontech, Palo Alto, CA). Sense and antisense oligonucleotides corresponding to leucine-rich motif (LRM) putative NES were as follows: LRM1 sense, 5′-GAT CTT TTT TGG AAA ATT CTC TAG AAA CTC TGA GGA-3′; LRM1 antisense, 5′-GAT CTC CTC AGA GTT TCT AGA GAA TTT TCC AAA AAA-3′; LRM2 sense, 5′-GAT CTC TCT TTC AAC AAT TGA CAT GTG TGC TTC CTA TCC TGG ATA ATC TTT TAA-3′; LRM2 antisense, 5′-GAT CTT AAA AGA TTA TCC AGG ATA GGA AGC ACA CAT GTC AAT TGT TGA AAG AGA-3′; LRM3 sense, 5′-GAT CTC TGT CAC TGG AAG AAC AAT TGA GGA GGT TGC AAA-3′; and LRM3 antisense, 5′-GAT CTT TGC AAC CTC CTC AAT TGT TCT TCC AGT GAC AGA-3′ (Proligo France SAS, Paris, France). Complementary oligonucleotides were annealed and cloned in a sense orientation into the BglII site of pEGFP-C1 (Clontech). All sequences are expressed at the C-terminus of GFP. Full-length c-IAP1 mutants (GFP-c-IAP1-LRM1^*,-LRM2^*, and -LRM3^*) were obtained by mutagenesis of LRM1, 2, and 3, separately or in combination (leucine were replaced by alanine) using the Quick-Change Site-directed Mutagenesis Kit (Stratagene, La Jolla, CA). All constructs were sequenced to ensure the accuracy of the reading frames and the site-directed mutations.

HeLa cells were transfected 24 hours after seeding using Superfect transfection reagent (Qiagen, Valencia, CA) following the manufacturer's instructions. Cells were studied 24 hours after transient transfection: nuclei were stained with Hoechst 33342 and cells were fixed with 2% PFA for 5 minutes before studying the subcellular distribution of GFP fusion protein using a fluorescence (Nikon) or a confocal (Leica) microscope. THP1 cells were transiently transfected using the AMAXA nucleofector kit (Amaxa, Köln, Germany) and transfected cells were enriched by a 10-day geneticin selection (0.7 μg/mL) before expansion and treatment.

It has been previously shown that exposure of U937 cells to 20 nM TPA induced their differentiation into macrophage-like cells. Cells become adherent to the culture flask and the expression of CD11b at their plasma membrane increases.²³  We used Western blotting to analyze the expression of XIAP, c-IAP1, c-IAP2, and survivin, 4 proteins that belong to the IAP family, in U937 cells undergoing TPA-induced differentiation (Figure 1A). c-IAP2 could not be detected in undifferentiated U937 cells and remained undetectable at all steps of the differentiation process (not shown). Survivin expression was limited to the nucleus of undifferentiated cells and disappeared upon differentiation. This may be related to the differentiation-associated cell cycle exit since this protein, which has an evolutionarily conserved role as a mitotic spindle checkpoint protein, is expressed mainly in dividing cells.⁷  The expression of XIAP and c-IAP1 was poorly influenced by the differentiation process when studied in whole-cell lysates (Figure 1A left). However, c-IAP1, and to a lesser extent XIAP, progressively accumulated in nuclear-free extracts as the cells underwent differentiation (Figure 1A right). The present study focused on c-IAP1 redistribution.

Differentiation-associated redistribution of c-IAP1 from the nucleus to the cytoplasm was further confirmed by Western blotting analysis of c-IAP1 expression in TPA-treated THP1 cells (Figure 1C) and by fluorescent microscopy analysis of the 2 cell lines (Figure 1B,D). c-IAP1 was located mainly in the nucleus of U937 and THP1 undifferentiated cells and in the cytoplasm of TPA-differentiated cells. A kinetic analysis identified a transient diffuse staining of the cytoplasm in the first hours of TPA treatment. As the cells progressed toward the differentiation process, a more patchy staining close to the nucleus was observed (see THP1 cells in Figure 1D).

To precisely determine the subcellular localization of c-IAP1 in TPA-differentiated cells, we performed Western blot experiments in enriched cellular fractions. Figure 2A shows that c-IAP1 is localized in the nucleus of undifferentiated U937 cells and in the reticular fraction of TPA-differentiated U937 cells. Thus, in accordance with Figure 1A, the protein migrates from the nucleus to the cytoplasm. A similar observation was made by comparing cellular fractions of undifferentiated and differentiated THP1 cells (not shown). Fluorescence microscopy experiments indicated that c-IAP1 colocalized with Golgin 97, a Golgi matrix protein, in TPA-differentiated THP1 (Figure 2B) and U937 (not shown) cells. c-IAP1 also colocalized, although less precisely, with GM130, a protein associated with the cis-Golgi (Figure 2B). Addition of either BFA, a fungal metabolite that causes disintegration of Golgi structure through inhibition of adenosine diphosphate (ADP)-ribosylation factor (ARF) guanosine 5′-triphosphate (GTP)-binding proteins,³²  or nocodazole, a microtubule-depolarizing agent, suppressed the patchy staining of c-IAP1 in TPA-differentiated THP1 (Figure 2C) and U937 (not shown) cells. In the tested conditions, BFA did not modify calnexin C subcellular localization, indicating that the endoplasmic reticulum was not altered (not shown). Altogether, these observations indicated that c-IAP1 was redistributed to the Golgi apparatus in cells undergoing differentiation.

To determine the topologic orientation of c-IAP1 in the Golgi compartment of differentiated cells, we isolated the microsomal fraction from TPA-treated U937 cells and submitted this fraction to tryptic-limited digestion before Western blotting analysis (Figure 2D). Addition of trypsin resulted in complete digestion of Golgin 97, which is located on the cytoplasmic side of the Golgi apparatus, whereas the endoplasmic reticulum-lumenal PDI was resistant to trypsin digestion. In these conditions, trypsin completely digested c-IAP1, indicating that the protein may be located to the external cytoplasmic side of the Golgi apparatus.

To characterize the mechanisms that are responsible for the nuclear export of c-IAP1, we first used LMB, a specific inhibitor of exportin 1 (also known as CRM1 [chromosomal region maintenance-1]), which is the receptor for leucine-rich NES.³³  Addition of 100 nM LMB for 24 hours to TPA-treated THP1 (Figure 3A) and U937 (not shown) cells prevented the redistribution of c-IAP1. To confirm the ability of LMB to inhibit the nuclear export of c-IAP1, we used a construct encoding full-length c-IAP1 associated, through its N-terminus, to GFP. This construct was transiently expressed in HeLa and 293T cell lines, in which the transfection rate was much higher than in leukemic cell lines. Twenty-four hours after transfection of GFP-c-IAP1 construct in these cells, the fluorescence was detected in both the nucleus and the cytoplasm. In the presence of LMB, the protein accumulated in the nucleus (see HeLa cells on Figure 3B). These results suggested that the nuclear export of c-IAP1 involved an NES and CRM1.

A software-based search in the protein sequence of c-IAP1 identified 3 hydrophobic LRMs that were consensus sequences for potential NESs. The first one was located in the BIR2 domain (LRM1), the second one in the CARD (LRM2), and the last one between the CARD and the RING domain (LRM3; Figure 4A). To determine whether one or several of these motifs played a role in c-IAP1 nuclear export, we cloned the sequences encoding these motifs in a GFP-encoding vector and expressed them by transient transfection in HeLa and 293T cells. The subcellular location of GFP fusion proteins was examined 24 hours after transfection by conventional (Figure 4B) and confocal laser (not shown) microscopy. While GFP-associated LRM1 and LRM3 were expressed in both the nucleus and the cytoplasm, GFP-associated LRM2 was almost exclusively expressed in the cytoplasm (Figure 4B-C). In addition, exposure to leptomycin B induced accumulation of the GFP-LRM2 protein in the nucleus (Figure 4C). These results indicated that LRM2 was the only sequence to behave as a functional NES.

To determine whether this potential NES was functional in the whole protein, a series of mutants were prepared in which leucine amino acids in the LRMs were replaced by alanine residues. The mutated constructs fused to GFP in a plasmid vector were transiently transfected in HeLa (Figure 5) and THP1 (not shown) cells, and their subcellular location was analyzed by fluorescence microscopy and Western blotting 24 hours later. As shown previously (Figure 3B), overexpressed wild-type c-IAP1 in HeLa cells demonstrated a cytoplasmic and nuclear pattern of expression. Mutations in either LRM1 or LRM3 or both did not affect the cellular distribution of the protein, whereas all the LRM2 mutants accumulated in the nucleus (Figure 5A). For unknown reasons, LRM2 mutant was less expressed than wild-type protein and other mutants. These observations were confirmed by immunoblotting the nuclear and cytoplasmic fractions of transiently transfected HeLa (Figure 5B) and THP1 (not shown) cells with an anti-GFP Ab. These experiments indicated that overexpressed wild-type c-IAP1 was detected mainly in the cytoplasmic fraction. When leucine residues in LRM2 were mutated, GFP-associated protein was located in the nucleus. When leucine residues in LRM1 or LRM3 were mutated, GFP-associated c-IAP1 demonstrated a cytoplasmic and nuclear expression similar to that of the wild-type protein. Altogether, these results indicated that LRM2 was a functional NES in c-IAP1.

LMB was observed to prevent TPA-induced differentiation of THP1, as demonstrated by studying the CD11b marker (Figure 6A), which suggested that a nucleocytoplasmic redistribution of proteins was required for the differentiation process. In an attempt to determine whether c-IAP1 was one of the proteins whose nuclear export was a key event in this process, we transiently overexpressed the GFP-tagged LRM2 mutant of c-IAP1 in THP1 cells. The protein was located mainly in the nucleus of transfected cells (Figure 6B) and TPA exposure failed to increase CD11b expression in GFP-tagged cells (Figure 6C-D). In addition, adhesion of GFP-positive cells to the culture flasks was delayed (not shown). However, similar results were obtained when wild-type c-IAP1 was transiently overexpressed in THP1 cells (Figure 6B-C). These results indicated that c-IAP1 overexpression could interfere with cell differentiation.

c-IAP1 was observed to be present mainly in the nucleus of the CD34⁺ progenitor, in both the nucleus and the cytoplasm of peripheral blood monocytes, and exclusively in the cytoplasm of macrophages and dendritic cells obtained by ex vivo differentiation of monocytes (Figure 7A) as well as erythroblasts and megakaryocytes obtained by ex vivo differentiation of CD34⁺ cells (Figure 7B).^27,28  As previously observed in TPA-differentiated cells (Figure 1B), c-IAP1 demonstrated a punctuated expression in the peri-nuclear zone and colocalized with Golgin 97 in macrophages (Figure 7C) and dendritic cells (not shown) obtained by differentiation of normal peripheral blood monocytes. A differentiation-associated redistribution of c-IAP1 from the nucleus to the cytoplasm was also observed in nonhematopoietic cells (ie, in HT29 human colon carcinoma cells undergoing partial differentiation when grown at confluence).³⁴  c-IAP1 also demonstrated a cytoplasmic expression in the well-differentiated, mucus-secreting HT29/MTX clone (Figure 7D).²⁵ 

The redistribution of c-IAP1 observed in parental U937 cells when undergoing TPA-induced differentiation was not identified in a TPA-resistant U937 cell clone treated under the same conditions (Figure 7E). We have previously shown that Bcl-2 overexpression prevented TPA-induced differentiation in U937 human leukemia cells.²³  The nucleocytoplasmic redistribution of c-IAP1 observed in TPA-treated parental cells transfected with an empty vector was not identified in Bcl-2-overexpressing U937 cells treated under similar conditions (Figure 7E). To confirm this latter observation, we tested the differentiation of bone marrow monocytes obtained from control and transgenic FVB/N mice overexpressing Bcl-2 in Mac-1⁺ cells. Whereas c-IAP1 translocation to the cytoplasm was observed in control cells induced to differentiate into macrophages by ex vivo culture in the presence of CSF-1-containing medium, no redistribution of the protein could be detected in cells from transgenic FVB/N mice cultured under similar conditions (Figure 7F). We also cultured peripheral blood mononuclear cells from 7 patients with chronic myelomonocytic leukemia (CMML) in the presence of M-CSF or the GM-CSF/IL-4 combination for 6 days. Cell differentiation was assessed morphologically and confirmed by studying cell phenotype using CD71 and CD1a to identify macrophages and dendritic cells, respectively. Mononuclear cells from these patients failed to differentiate, and this blockade in cell differentiation was associated with a lack of c-IAP1 nucleus export (see Figure 7G for example).

Two main functions have been assigned to c-IAP1. The protein is involved in the signaling induced by engagement of several members of the TNF receptor family including TNF-R2,⁹  CD40,³⁵  and the lymphotoxin-β receptor.³⁶  c-IAP1 was also described as an endogenous inhibitor of apoptosis through direct binding to the active sites of caspase-3 and -7.³⁷  The recently described role of the protein in ubiquitylation may contribute to both functions (eg, by regulating the cellular level of the adaptor molecule TRAF2¹⁶  or the apoptosis inducer Smac/DIABLO).¹⁵  These functions imply a cytoplasmic localization of the protein. We show here that c-IAP1 is present almost exclusively in the nucleus of the studied undifferentiated cells, translocates to the cytoplasm when these cells undergo differentiation, and localizes mainly to the Golgi apparatus in differentiated cells.

Several arguments suggest a translocation of the protein rather than a degradation followed by a synthesis in a distinct cellular compartment. First, c-IAP1 mRNA (not shown) and protein levels remain stable along the differentiation. Secondly, by immunofluorescence analysis, the protein was identified in the nucleus of undifferentiated cells, in the cytosol of cells undergoing differentiation, and in the Golgi of differentiated cells, suggesting a migration. Third, we have identified a functional NES in the protein. The nuclear transport of proteins through nuclear pores requires the presence of specific signals such as nuclear localization signals (NLSs) and NESs. The size of c-IAP1 prevents it from passively diffusing through nuclear pores³⁸  and we did not identify any classical NLSs in the protein, which suggests that the protein is either carried by a cofactor or enters the nucleus using a nonconventional mechanism of active import. On the other hand, the protein export may involve an NES-mediated, LMB-sensitive mechanism. The retention of c-IAP1 in the nucleus of undifferentiated cells suggests inhibition of this NES-mediated export. Activation of such an active export mechanism has been shown in other proteins to require an event that promotes their interaction with CRM1 such as phosphorylation,³⁹  monoubiquitylation,⁴⁰  conformational changes,⁴¹  or proteolytic cleavage.⁴²  So far, we did not detect any posttranslational modification of the protein in cells undergoing differentiation. The location of the functional NES of c-IAP1 in its CARD, a motif involved in protein-protein interactions, could also indicate a role for an unidentified protein partner in the modulation of c-IAP1 translocation. Such a protein partner has been identified for XIAP, which can be retained in the nucleus through interaction with XAF1.²² 

We have observed that c-IAP1 was located in the Golgi apparatus of differentiated cells. Another IAP, referred to as BRUCE in mice and Apollon in humans, also localizes to the Golgi compartment and the vesicular system.⁴³  In addition to being related to IAPs through a BIR motif, this giant protein is an ubiquitin-conjugating enzyme (E2).⁴⁴  However, the functions of both Apollon and c-IAP1 in the Golgi apparatus remain to be elucidated. Proteins that accumulate in the Golgi structure can be secreted.⁴⁵  We have observed that c-IAP1 did not accumulate in cholesterol- and sphingomyelin-enriched fractions of differentiated cells (data not shown), suggesting that the protein did not associate with microdomains described in the secretory pathway.⁴⁶  c-IAP1 was shown to alter the cellular distribution of coexpressed reaper and grim drosophila proteins in mammalian cells, suggesting that c-IAP1 could modulate signaling pathways by sequestration of proteins in cell compartments.⁴⁷ 

c-IAP1 is not the only IAP to translocate from the nucleus to the cytoplasm. The main isoform of survivin, a single-BIR-containing IAP survivin involved in the control of the mitotic spindle checkpoint,⁷  can be exported from the nucleus through an LMB-sensitive mechanism, whereas its alternative isoform survivin-ΔEx3 is driven to the nucleus by a C-terminal NLS.⁴⁸  In the present study, we have observed that XIAP was also exported from the nucleus in U937 cells undergoing differentiation, together with c-IAP1. Of importance, survivin is ubiquitylated and degraded when cells exit the cell cycle, whereas c-IAP1 and XIAP protein levels remain stable along the differentiation process, which suggest that cellular redistribution of these latter proteins may be essential in their regulation.

Two IAPs have previously been shown to interfere with cell differentiation (ie, overexpressed human NAIP prevents neuronal differentiation of PC12 cells,⁴⁹  whereas transgenic mice overexpressing XIAP in their lymphocytes demonstrate altered T-cell maturation).⁵⁰  Here, we show that overexpression of either c-IAP1 or its NES-mutated protein interferes with TPA-induced differentiation of human leukemic cells. How these IAPs interfere with cell differentiation remains unidentified. The first function assigned to c-IAP1 has been its ability to interact with and to inhibit caspases.³⁸  These enzymes have been involved in cell differentiation processes, both in humans^23,27  and drosophila.⁵¹  Obviously, caspase activity must be carefully controlled when associated with cell differentiation to prevent cell death by apoptosis and it is attractive to speculate that IAPs play a role in this control. In accordance with this hypothesis, the BIR-containing protein dBRUCE was proposed to bind to and degrade caspases involved in drosophila spermatogenesis.⁵¹  We have recently shown that a limited activation of several caspases was required for the differentiation of human peripheral blood monocytes into macrophages, whereas their differentiation into dendritic cells did not depend on caspase activation.²³  Since the nucleocytoplasmic translocation of c-IAP1 was observed in both caspase-dependent (macrophages) and -independent (dendritic cells) pathways of differentiation, the negative control of caspase activity may not be the function of c-IAP1 shuttling. In addition, caspases have been observed moving from the cytoplasm to other cell compartments in leukemic cells undergoing differentiation but were not identified to associate with the Golgi apparatus.⁵²  Finally, we have shown previously that the postmitochondrial pathway to apoptosis remained functional in TPA-differentiated U937 cells, indicating that redistributed c-IAP1 did not prevent caspase activation by cytochrome-c released from the mitochondria.

Another function assigned to c-IAP1 and other IAPs containing a RING finger motif is ubiquitylation of proteins. c-IAP1 catalyzes its own ubiquitylation in vitro,⁸  promotes ubiquitylation of the apoptosis inducer Smac/DIABLO,¹⁵  and modulates cell response to TNF-α through the regulation of the intracellular level of TRAF2¹⁶  and NF-κB essential modulator (NEMO).⁵³  Preliminary studies suggest that c-IAP1 down-regulation decreases the proliferation rate of THP1 cells (data not shown). Together with the role of the ubiquitin/proteasome pathway in the regulation of cell cycle, this observation could indicate a connection between c-IAP1 redistribution and growth inhibition in cells undergoing differentiation.

The nucleocytoplasmic traffic of proteins modulates cellular functions.⁵⁴  Accordingly, LMB prevents TPA-induced differentiation of U937 and THP1 cells. However, the role of the inhibition of c-IAP1 translocation in LMB-induced inhibition of cell differentiation remains to be determined. On the other hand, inhibition of the differentiation process correlates with a lack of nuclear export of c-IAP1, as observed in Bcl-2-overexpressing cells, in TPA-resistant U937 cells, and in monocytes from patients with CMML that do not respond to M-CSF-induced differentiation. Bcl-2 and related proteins were involved in the regulation of differentiation (eg, in restricting lineage determination during hematopoietic differentiation).⁵⁵  The ability of overexpressed Bcl-2 to prevent the translocation of c-IAP1 could be a nonspecific consequence of its influence on the differentiation process. However, we cannot rule out that Bcl-2 directly interferes with the translocation of c-IAP1, since the protein is expressed at multiple cellular sites including the nuclear outer membrane⁵⁶  and the nucleus.⁵⁷ 

Altogether, the present study suggests that c-IAP1 functions may be regulated in part by the subcellular location of the protein that is present mainly in the nucleus of various types of undifferentiated cells and translocates to the cytoplasm when these cells undergo differentiation. Ongoing studies may indicate whether this redistribution of c-IAP1 plays an active role in the differentiation process or confers specific functions to differentiated cells.

The author thanks M. Yoshida for kindly providing LMB; J. C. Reed for c-IAP1 cDNA; P. J. Parker, T. Lesuffleur, and J. Breard for cell lines; Irving Weismann and Eric Lagasse for the MRP8BCL-2 transgenic mice; and B. Goud, S. Khochbin, O. Hermine, and M. Fontenay for fruitful discussions.