The E3 ubiquitin-protein ligase Triad1 inhibits clonogenic growth of primary myeloid progenitor cells

Granulocytes and monocytes develop from common myeloid progenitor cells through a complex network of cell growth, differentiation, and apoptosis-regulating factors. Alterations in these processes may cause acute myeloid leukemia, which is characterized by uncontrolled proliferation of immature myeloid cells that fail to differentiate toward mature functional cells.^1,2  In leukemia, mutations occur in genes encoding tyrosine kinases such as Fms-like tyrosine kinase 3 (Flt3), c-kit, neuroblastoma ras (N-ras) and transcription factors such as AML1 (acute myeloid leukemia 1), CBFB (core binding factor β), GATA1 (GATA binding protein 1), PU.1, C/EBPα (CCAAT/enhancer binding protein α), and RARα (retinoic acid receptor α).^1,2  Although the transcription factors predominantly play a role in lineage commitment, activation of tyrosine kinases is thought to result in proliferative and/or survival signals.^1,2  Recent studies have shown that several proteins involved in myelopoiesis (including proteins mutated in leukemia) are inactivated through ubiquitin (Ub)-proteasomal degradation pathways.^3-7  This form of targeted protein degradation is accomplished by the covalent conjugation of Ub to substrate proteins, usually in the form of a multi-Ub chain, which marks these proteins for progressive degradation by the 26S proteasome.^8,9  Protein ubiquitination is catalyzed through a cascade of reactions. Ub is first activated by the adenosine triphosphate (ATP)-dependent Ub-activating E1 enzyme and subsequently transferred to one of a set of E2 Ub-conjugating enzymes. The E2 enzymes act in conjunction with accessory E3 Ub protein ligases. In the E2-E3 complex, the E3 component binds to protein substrates, allowing the E2 to form a multi-Ub chain linked to a lysine of the substrate protein.^8,9  Thus, the E3 Ub ligases play a crucial role in this process because these proteins recognize the cellular proteins destined for ubiquitination.¹⁰  To date, several types of E3 Ub ligases have been described. These include HECT (homologous to E6-AP carboxyl terminus), RING (really interesting new gene) finger, U-box, SOCS (suppressor of cytokine signaling)-box, F-box, and cullin ligases.^11-20  Although the first 3 types of proteins are actively involved in substrate ubiquitination, the last 3 do not ubiquitinate substrate proteins themselves but are part of larger protein complexes that exhibit Ub ligase activity.^15-19  Despite several proteins that play an important role in myelopoiesis are known to be targeted by the Ub-proteasome pathway, only a few E3 Ub ligases involved in myelopoiesis have been described. These include the cullin Ub ligase Cul-4A implicated in myelopoiesis.^4,21  Cul-4a heterozygous mice have diminished multipotent progenitors, and enforced Cul-4A expression inhibits granulocytic differentiation of myeloid PLB-985 cells.²¹  An important target of Cul-4A in myelopoiesis may be the transcription factor HoxA9 (homeo box A9) that is targeted by Cul-4A for ubiquitination resulting in its proteasomal degradation.⁴  Another group of Ub ligases involved in myelopoiesis are the SOCS proteins. All of these proteins contain a SOCS box that provides in the scaffold for interaction with Ub ligase complexes.^15,16  One of the main functions of the SOCS box proteins is to recruit substrate proteins to ligase complexes resulting in their ubiquitination. SOCS3 is induced on granulocyte colony-stimulating factor (G-CSF) treatment, binds to the activated G-CSF receptor, and negatively regulates G-CSF receptor-mediated STAT (signal transducer and activator of transcription) activation.²²  In addition, mice with loss of SOCS3 function in the hematopoietic compartment develop neutrophilia.²³  Although it is not exactly clear how SOCS proteins inhibit cytokine signaling, it has been suggested that the ubiquitination and subsequent degradation of signaling molecules may be involved in this process.²⁴  Finally, another important E3 Ub ligase involved in myelopoiesis is the RING finger protein Cbl (CAS-BRM murine ecotropic retroviral transforming sequence homolog). Cbl catalyzes the ubiquitination of several activated receptor tyrosine kinases, including the colony-stimulating factor 1 (CSF-1) receptor.²⁵  The ligand binding stimulated ubiquitination of these receptors results in their internalization and targeting to lysosomes where they are degraded. As this inhibits these receptors to recycle through early endosomes to the cell surface, Cbl negatively regulates receptor signaling.²⁵  Here, we identify a novel E3 Ub ligase, Triad1 (2 RING fingers and DRIL [double RING finger linked] 1), to be strongly induced during granulocytic and monocytic differentiation and show that the Ub ligase activity of Triad1 inhibits the proliferation of committed myeloid progenitor cells.

Subtractive polymerase chain reaction (PCR) was performed using polyA⁺ mRNA isolated from 0 and 18 hours all-trans retinoic acid (ATRA)-treated NB4 cells (PCR-select cDNA subtraction kit; Clontech, Heidelberg, Germany). Subtracted radiolabeled PCR fragments were isolated from polyacrylamide gel, reamplified, cloned in pCR2.1, and sequenced. Methods for cDNA synthesis and classic and real-time RT-PCR were as described.²⁶  5′-Rapid amplification of cDNA ends (RACE) was performed using the Marathon cDNA amplification kit (Clontech) on 1.0 μg human spleen mRNA. Relevant primers are indicated in Table 1.

Triad1 coding sequences were PCR amplified using cDNA derived from NB4 cells and cloned in pCR2.1. A 10 amino acid (aa) human myc epitope was introduced at the 3′ end of the coding sequence of Triad1 by PCR and cloned into pSG5, and nontagged Triad1 was cloned into pcDNA3.1. Triad1 coding sequences were generated by PCR and cloned in plasmid vector red-shifted green fluorescence protein (pEGFP)-C1 to generate GFP-Triad1, pFastBacHT to generate His-Triad1 in Hi5 cells (Baculovirus expression system; Invitrogen, Karlsruhe, Germany) and in pGEX-2TK to generate glutathione S transferase (GST)-Triad1. For retroviral transduction Triad1 coding sequences were inserted into the retroviral pLRZS vector. For yeast 2-hybrid assays sequences coding for Triad1 aa 1 to 119 and 109 to 493 were cloned in pGAD-C. All primers used for cloning purposes and for generating the Triad1 point mutants H158A and C161A by site-directed mutagenesis are indicated in Table 1. The integrity of all Triad1 sequences was checked by sequence analysis.

In these studies antibodies against CD34, CD3, CD14, CD15, CD45 (Beckman Coulter, Brea, CA), Mac1 (BD Biosciences, Heidelberg, Germany), GFP (7.1/13.1; Roche, Basel, Switzerland), myc (9E10; Santa Cruz Biotechnology, Heidelberg, Germany), UbcH7 and UbcH9 (Transduction Laboratories, Lexington, KY), Ub (6C1; Sigma, Saint Louis, MO), Flag (M2; Sigma), annexinV (AlexaFluor 647; Molecular Probes, Leiden, The Netherlands), and Triad1 were used. Triad1 antibodies were generated by immunizing rabbits with a keyhole limpet hemocyanin (KLH)-coupled peptide corresponding to Triad1 aa 473 to 486, followed by affinity purification using peptides immobilized on Affigel 10 columns (BioRad, Munich, Germany). Human hematopoietic subsets were isolated by fluorescence-activated cell sorting (FACS) with appropriate antibodies. Granulocytes were selected based on CD45 positivity and scatter plot. Cell purity was confirmed to be greater than 95% by flow cytometric analysis. For immunofluorescence experiments cells were stained with indicated primary antibodies followed by staining with a secondary antibody. Samples were mounted in Vectashield medium (H1200; Vector Laboratories, Burlingame, CA) and analyzed with a Leica DMRBE microscope (Leica, Rijswijk, The Netherlands). Pictures were taken at 40 × (oil, 1.00 NA) or 100 × (oil, 1.30 N) magnification with a Leica DC350F Camera using Leica FW400 software and processed with Adobe Photoshop (Adobe Systems, San Jose, CA). For detection of endogenous Triad1 by multicolor flow cytometry (Cytomics FC500; Becton Dickinson, Franklin Lakes, NJ), cells were stained with a Triad1 antibody using fix-and-perm approaches as described²⁷  followed by blocking of cells with rabbit serum and subsequent staining for relevant surface markers. For Triad1 Northern blot hybridization 10 μg total RNA isolated from NB4 cells treated with ATRA (1 × 10^-6 M) for various time points or 1.0 μg mRNA from hematopoietic cells was hybridized using Triad1 clone 14-5 as a probe. For GST-captures, 10 μg immobilized GST or GST-Triad1 was incubated with cell lysates from 1 × 10⁷ NB4 cells. For capture of His-tagged proteins, lysates from transfected 293T cells were incubated with His-select beads (Sigma) under denaturing conditions (6 M guanidiumchloride). For immunoprecipitations, lysates from transfected COS cells were incubated with indicated antibodies. Bound proteins were washed, separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), immunoblotted, and stained with relevant antibodies.

For yeast 2-hybrid analyses UbcH bait²⁸  and Triad1 prey constructs were transfected to the yeast strain AH109 (Clontech). This strain contains 2 selectable markers (histidine and adenine) and a lacZ reporter. A positive interaction in yeast was defined by growth of blue colonies 4 days after plating cells on leucine, tryptophane, adenine, and histidine-deficient synthetic defined (SD) medium containing X-α-gal.

Proteasome activity was determined in primary murine bone marrow cells after treatment with 1 × 10^-8 M MG132. For measurements, 67.5 μg protein lysates (concentration determined using Bradford assay) were incubated with the proteasome substrate Suc-Leu-Leu-Val-Tyr-AMC (Boston Biochem, Cambridge, MA) for 2 hours at 37°C as described.²⁹  7-Amino-4-methylcoumarin (AMC) liberated from the substrate by proteasome activity was determined by measuring fluorescence (Fluorstar, BMG; Isogen Life Sciences, Maarssen, The Netherlands).²⁹  For in vitro ubiquitination assays His-Triad1 was isolated from Hi5 cells (Baculovirus Bac-to-Bac system; Invitrogen) using His-select spin columns followed by elution using 250 nM imidazole. Imidazole was removed from His-Triad1 protein eluates using a Microcon YM-10 filter (Amicon, Bedford, MA), and protein concentrations were determined with the Bradford assay. Ubiquitination assays were performed using 1.6 μg reticulocyte fraction IIA (Boston Biochem), bovine Ub (Sigma), or recombinant UbK48-only (Boston Biochem) with freshly prepared His-Triad1 for 60 minutes at 37°C followed by staining of ubiquitinated proteins in Western blot analysis using a Ub-specific antibody (C61; Sigma) as described.³⁰ 

Retroviral vectors (pLZRS) were transfected to retroviral producer phoenix cells (kindly provided by Dr G. Nolan). Primary murine bone marrow cells isolated from femurs from 6-week-old female CBA/CA mice were prestimulated for 2 days with stem cell factor (SCF), interleukin 12 (IL-12), IL-3, Flt-3, thrombopoietin (TPO), and α transforming growth factor-β (α-TGFβ) and subsequently infected with retroviruses on retronectin-coated dishes. After 24 hours GFP-positive cells were sorted by FACS and seeded in semisolid medium (5000 cells/well) containing pokeweed mitogen-stimulated mouse spleen conditioned medium, horse serum, and SCF. Colonies were counted 5 days later yielding between 83 and 125 colonies after granulocyte-macrophage colony-forming unit (CFU-GM) of empty vector transduced cells and morphologically analyzed after May-Grünwald-Giemsa staining. To determine proliferation in liquid assays, 5000 cells were seeded in 96-well plates using the above-mentioned medium without agar. Relative cell numbers were determined by FACS using fluorescent beads. Apoptosis and DNA histograms were determined by flow cytometry using AnnexinV and propidium iodide staining, respectively. Percentages of cells in various phases of cell cycle were determined using ModFit software (Verity Software House, Topsham, ME). These studies were approved by the DEC Committee, RUNMC, Nijmegen, The Netherlands.

Acute promyelocytic leukemia cells can be forced to differentiate along the granulocytic lineage by the vitamin A derivative ATRA. To identify genes involved in granulocytic differentiation we performed a subtractive PCR screen using RNA isolated from promyelocytic leukemia (NB4) cells before and 18 hours after ATRA treatment. Northern blotting using one of the subtracted clones (clone 14-5, GenBank accession number AF099153) showed the induction of a 4-kilobase (Kb) mRNA in NB4 cells 24 hours after ATRA treatment to reach maximum levels at 48 hours (Figure 1A). The subtracted clone appeared identical to human randomly cloned expressed cDNA sequences identified in database searches (not shown). By using these sequences in database walks a 3.9-Kb cDNA contig was constructed. This contig was confirmed by amplification of overlapping RT-PCR fragments and 5′- and 3′-RACE (Figure 1B-C). Sequence analysis of the 3.9-Kb transcript identified a poly-adenylation signal (ATAAA) 24 base pair (bp) upstream of the 3′ end of the transcript and a translational start codon at cDNA position 145. This start codon is accompanied by a consensus Kozak sequence and is followed by a 1479-bp ORF (Figure S1). The coding sequences were confirmed by sequence analysis of 12 independent clones (GenBank accession number AF099149). In line with multiple bands observed in RT-PCR (Figure 1B), 3 Triad1 clones contained different insertions of extra sequences present at cDNA position 399 possibly representing splice variants (GenBank accession numbers AF099150, AF099151, and AF099152).

The 1479-bp ORF in the 3.9-Kb transcript predicts the formation of a 493-aa protein (Figure 1D). We have called this protein Triad1 referring to its cysteine-histidine rich TRIAD motif.³¹  The TRIAD motif consists of 2 RING fingers that flank the DRIL domain. The DRIL (also known as IBR [in between RING fingers]³² ) domain resembles other cysteine-rich motifs as the LAP (leukemia-associated protein)/PHD, LIM, and RING finger domains.³¹  Recent data indicate that the C-terminal RING finger (RING2) of the TRIAD motif has a distinct topology from the classic RING finger structure and that it binds only 1 zinc atom instead of 2.³³  Further Triad1 protein analysis identified an acidic N-terminal region, 2 C-terminal coiled coil regions, and 2 putative nuclear localization signals. To determine whether Triad1 is conserved during evolution we performed database searches and identified full-length murine Drosophila melanogaster and Caenorhabditis elegans orthologs (Figure S1). The conservation in the last 231 aa C-terminal part of Triad1 between D melanogaster and C elegans compared with human Triad1 is very high, being 80% and 64% identical, respectively. Comparison of complete murine and human Triad1 shows that the proteins are 98% identical. The strong evolutionary conservation of Triad1 suggests that it exerts important biologic functions.

TRIAD1 gene expression is induced during forced granulocytic differentiation of promyelocytic leukemia (NB4) cells. To determine the relevance of this observation we analyzed TRIAD1 expression in normal human hematopoiesis. TRIAD1 Northern blot hybridization using RNA isolated from total bone marrow, mature granulocytes, and T cells revealed strongest TRIAD1 expression in granulocytes (Figure 2A). In agreement, TRIAD1 quantitative RT-PCR detected high expression in mature granulocytes and low expression in T cells and immature CD34⁺ cells, indicating that TRIAD1 is induced during maturation of immature CD34⁺ cells toward mature granulocytes (Figure 2B). Further expression studies showed detectable TRIAD1 expression by real-time PCR in a large set of human tissues tested, indicating that the function of Triad1 is probably not limited to hematopoiesis (Figure 2C).

To determine the subcellular localization of Triad1 a C-terminally myc epitope-tagged Triad1 (Triad1-myc) expression vector was transfected to L88/5 stromal cells. Indirect immunofluorescence (IF) using the myc epitope recognizing antibody 9E10 revealed a clear nuclear diffuse staining with exclusion of the nucleoli (Figure 3A). In cells with highest IF levels, some cytoplasmic staining was also observed. To determine the endogenous Triad1 expression in primary hematopoietic cells an affinity-purified Triad1 antibody was used in indirect IF. Analysis of immature CD34⁺ and T cells revealed a weak nuclear staining (Figure 3B). In contrast, a strong nuclear localization was observed in both granulocytes and monocytes. Although the localization in monocytes was nuclear diffuse, a ring-shaped nuclear organization was observed in granulocytes (Figure 3B). To determine at what stage Triad1 was induced during myelopoiesis we stained fresh bone marrow and leukapheresis samples with a Triad1 antibody and analyzed the cells by flow cytometry. Costaining with CD34 (immature progenitors), CD3 (T cells), CD14 (monocytes), or CD15 (granulocytes) showed that Triad1 is lowly expressed in immature CD34⁺ and T cells and that the expression gradually increases on maturation toward committed myelomonocytic progenitors to reach highest levels in mature monocytes and granulocytes (Figure 3C; Figure S2). Together, these data indicate that Triad1 is induced at the mRNA and protein level during differentiation of immature blood cells toward both monocytes and granulocytes and that it localizes predominantly to the nucleus in primary human hematopoietic cells.

The RING finger structure provides an interface for interaction with E2 Ubcs, and all RING finger proteins that bind Ubcs exhibit Ub ligase activity.²⁰  Because Triad1 harbors 2 RING fingers we tested whether it binds to Ubcs using yeast 2-hybrid assays. For these experiments 2 Triad1 prey-fragments were used, 1 coding for amino acids 1 to 119, Triad1^[1-119], lacking the TRIAD structure and 1 for amino acids 109 to 493, Triad1^[109-493], including the TRIAD structure. No interaction was observed between Triad1^[1-119] and a panel of Ubcs²⁸  tested, including CDC34, UbcH5, UbcH6, UbcH7, UbcH8, UbcH9, UbcH10, and UbcH13 (not shown). In contrast, a clear interaction between Triad1^[109-493] and UbcH7 was observed as indicated by growth of yeast cells on medium lacking both histidine and adenine (Figure 4A). When less stringent yeast 2-hybrid conditions were used (histidine selection only) an interaction between Triad1^[109-493] and UbcH6 and UbcH13 (also known as Bendless) was also found (data not shown). Because endogenous UbcH7 is readily detectable in NB4 cells, we incubated cell lysates from these cells with GST-Triad1 to confirm the UbcH7 interaction as observed in yeast 2-hybrid assays. Bound proteins were stained with an UbcH7 antibody in Western blot analysis, showing that Triad1 efficiently binds endogenous UbcH7 from NB4 cells (Figure 4B). Staining for UbcH9 in these experiments was negative, whereas UbcH9 is also easily detectable in NB4 cells by Western blotting (not shown). To test whether the N-terminal RING finger was involved in UbcH7 binding we constructed 2 independent Triad1 RING finger point mutants in which 1 of the predicted zinc-coordinating amino acids were changed into alanine (H158A, C161A). Mutations were introduced in the Triad1 prey-construct (Triad1^[109-493]) and used in yeast 2-hybrid assays. Although normal Triad1^[109-493] interacts with UbcH7 in yeast 2-hybrid assays, both point mutants did not interact with UbcH7, indicating that the N-terminal RING finger of Triad1 is essential for UbcH7 binding (Figure 4A). To further test whether Triad1 might function as Ub ligase, we determined whether Triad1 binds ubiquitinated proteins. To this end, Triad1-myc and Flag-tagged Ub (Flag-Ub) were coexpressed in the presence or absence of proteasome inhibitors followed by myc IP. Triad1 staining of myc-immunoprecipitated proteins in Western blot analysis revealed that proteasome inhibition resulted in stabilization of Triad1. Flag staining of proteins bound to Triad1 detected a smear of ubiquitinated proteins that appeared more intense when transfected cells were treated with a proteasome inhibitor, suggesting that Triad1 binds ubiquitinated proteins (Figure 4C). Because the observed smear might represent exclusively ubiquitinated Triad1 species instead of ubiquitinated proteins bound to Triad1, we repeated the experiment using GFP-Triad1 and Flag-Ub. Subsequent GFP IP followed by Triad1 staining in Western blot analysis revealed the presence of unmodified Triad1, whereas Flag staining revealed a smear of ubiquitinated proteins, including proteins with sizes smaller compared with Triad1. Because the smaller proteins do not represent Triad1, we concluded that Triad1 binds ubiquitinated proteins (Figure 4D). To further test whether Triad1 functions as E3 Ub ligase, we performed an in vitro ubiquitination assay using reticulocyte fraction IIA (enriched in E1 and E2 activity and devoid of Ub and proteasome activity), with or without Ub and His-Triad1. Staining of reaction mixtures with a Ub-specific antibody in Western blot analysis revealed the presence of Triad1-dependent ubiquitinated proteins (Figure 4E). Triad1 binds to UbcH7. In concert with E3 Ub ligases, UbcH7 supports the formation of Ub chains linked through lysine 48 (K48) that are recognized by the proteasome. To test whether Triad1 can catalyze the formation of Ub chains linked through K48, we also used recombinant Ub carrying only 1 lysine residue at position 48 (UbK48-only), showing that Triad1 catalyzes the formation of Ub chains linked through K48. The interaction between Triad1 and UbcH7 as well as ubiquitinated proteins and the in vitro ubiquitination assay strongly suggest that Triad1 functions as E3 Ub ligase that may target proteins for proteasomal degradation.

In addition to their substrates, many Ub ligases are ubiquitinated themselves.¹³  Ubiquitination of Triad1 was studied by coexpressing Triad1 and His-tagged Ub (His-Ub) followed by selection of His-tagged proteins under denaturing conditions. Staining with a Triad1-specific antibody detected discrete bands with higher molecular weight compared with unmodified Triad1, indicating that Triad1 is ubiquitinated (Figure 5A). Proteasome inhibition resulted in an increase in ubiquitinated Triad1 forms, indicating that ubiquitinated Triad1 is subject to proteasomal degradation. Similar results were obtained when Triad1 and Flag-Ub were coexpressed followed by Flag IP under nondenaturing conditions (Figure 5B). The discrete Triad1 Ub-modified forms observed in these IPs combined with the smear of ubiquitinated proteins detected on Triad1 IP when cotransfected with Ub as shown in Figure 4C-D further indicate that, apart from being ubiquitinated itself, Triad1 also binds ubiquitinated proteins.

To determine the biologic relevance of the induction of Triad1 during myeloid differentiation, we tested whether Triad1 expression affected the clonogenic growth of primary immature murine bone marrow cells in CFU-GM assays. Primary murine bone marrow cells were transduced with pLZRS-derived viruses containing a GFP-IRES (internal ribosome entry site)-Triad1 sequence allowing for separate GFP and Triad1 expression. Fractions with low, intermediate and high GFP expression were sorted. GFP expression correlated with Triad1 expression as shown by Western blot analysis (Figure 6A; Figure S3A). Compared with empty vector controls (GFP alone), Triad1 expression resulted in a concentration-dependent inhibition of clonogenic growth. Although low Triad1 expression did not inhibit clonogenic growth, intermediate Triad1 expression resulted in an inhibition of greater than 60%, whereas highest Triad1 expression resulted in a strong inhibition of clonogenic growth of greater than 80% (Figure 6A). In addition, colonies derived from high and intermediate Triad1-expressing cells were significantly smaller compared with empty vector controls (Figure S3B). To determine whether Triad1 affects myeloid differentiation, colonies were recovered from semisolid medium and morphologically analyzed. This revealed a modest Triad1-dependent increase in the percentage of granulocytes and decrease in erythroblasts compared with empty vector transduced cells (Figure 6B).

The observed inhibition of clonogenic growth suggests a role for Triad1 in the development of immature-committed murine bone marrow cells toward mature myeloid cells. In many Ub ligases the RING finger is essential for Ubc binding as well as Ub ligase activity.²⁰  Mutations in RING fingers abrogating the interaction with Ubcs result in impaired ubiquitination. To test whether the Ub ligase activity of Triad1 was involved in the clonogenic inhibition of myeloid progenitor cells, we introduced the H158A and C161A RING finger mutations that abrogate UbcH7 binding in full-length Triad1-coding sequences in the retroviral vectors and used these constructs in CFU-GM colony assays. Strikingly, in contrast to normal Triad1, expression of the Triad1 point mutants did not affect the clonogenic growth of primary immature cells (Figure 6C). Because the mutant proteins were expressed at levels comparable to wild-type Triad1 as shown by Western blot analysis, this indicates that the N-terminal RING finger of Triad1 plays an important role in the regulation of proliferation and/or differentiation of myeloid progenitor cells. In E3 Ub ligases, the RING finger structure has been shown to be essential for Ub ligase activity.²⁰  Because UbcH7 can catalyze the formation of Ub chains resulting in protein degradation and the Triad1 RING finger mutations abrogate the inhibitory effect on clonogenic growth, our data suggest that the ubiquitination and subsequent proteasomal degradation of Triad1 target proteins results in clonogenic inhibition. To test this, we determined whether the inhibition of clonogenic growth exerted by Triad1 could be relieved by proteasome inhibition. For these experiments, 1 × 10^-8 M of the proteasome inhibitor MG132 was used because at this concentration limited toxicity on empty vector-transduced cells was observed in liquid assays (Figure 6D). At this concentration, the bone marrow cells showed an increase in ubiquitinated species and decreased proteasomal activity (Figures 6E-F, respectively). Interestingly, proteasome inhibition partially reversed the Triad1-induced suppression of colony formation (Figure 6G). When corrected for the suppressive effects of MG132 on empty vector-transduced cells, a greater than 2-fold increase in Triad1-positive colonies was observed when the proteasome was inhibited (Figure 6H). Moreover, the colonies derived from MG132-treated Triad1-transduced cells were significantly larger as compared with nontreated cells (Figure S3B). In addition to the proteasome, MG132 inhibits other proteases. To rule out the possibility that the increase in colony numbers of Triad1-transduced cells after treatment with MG132 depended on other mechanisms rather then inhibition of the 26S proteasome, we repeated the experiments using other more specific proteasome inhibitors, including PS-341, epoxomycin, and lactacystin. This yielded similar results as observed for MG132 (Figure 6H). Thus, the growth inhibitory effect exerted by Triad1 relies on the N-terminal RING finger and can be counteracted by proteasome inhibition, indicating that degradation of Triad1 targets are implicated in myelopoiesis.

To further investigate how Triad1 affected CFU-GM colony formation, we cultured Triad1 retrovirally transduced bone marrow cells in liquid medium and measured proliferation, cell cycle progression (DNA histograms), differentiation, and apoptosis. Compared with empty vector control, Triad1 expression resulted in a dramatic inhibition of proliferation (Figure 7A). This was accompanied by a marked increase in the percentage of cells in G₀/G₁ phase, and an increase in annexin V positivity (Figures 7B-C). In these experiments no significant changes in myeloid differentiation as indicated by Mac1 positivity was observed (Figure 7D). Thus, the inhibition of clonogenic growth by Triad1 is probably caused by a cell-cycle arrest associated with increased apoptosis. From these findings we conclude that the ubiquitination and degradation of Triad1 target proteins contributes to the development of early committed myeloid progenitors toward mature monocytes and granulocytes.

The modification of proteins with ubiquitin plays an important role in a variety of basic cellular processes, including cell division, differentiation, and apoptosis. Protein poly-ubiquitination linked through lysine 48 of Ub results in degradation of tagged proteins by the 26S proteasome. Proteins are ubiquitinated by E2-E3 Ub ligase complexes. The E3 Ub ligases play a key role in this process because these enzymes determine which cellular proteins are ubiquitinated. We report here that the E3 Ub ligase Triad1 is induced during both granulocytic and monocytic differentiation (Figures 1, 2, 3). Triad1 contains 2 RING finger structures. By providing an interface for interaction with E2 Ub-conjugating enzymes, the RING finger constitutes an essential domain in many E3 Ub ligases. In line with this, we found that Triad1 physically interacts with the E2 Ub-conjugating enzyme UbcH7 through its N-terminal RING finger and to ubiquitinated proteins (Figure 4A-D). Moreover, in vitro ubiquitination experiments showed that Triad1 supports ubiquitination of both wild-type and UbK48-only (Figure 4E). Ub chains linked through lysine 48 are efficiently recognized by the proteasome, resulting in substrate degradation. This indicates that Triad1 may mark proteins for proteasomal degradation.

Triad1 exhibits a predominantly nuclear staining in all cell types tested (Figure 3). In stromal cells (L88/5) and in primary monocytes the nuclear staining is diffuse, whereas a ring-shaped localization is observed within the nuclear lobes of primary granulocytes (Figure 3). Although these data may suggest that Triad1 substrates are primarily nuclear proteins, a protein-interaction map of the Drosophila proteome³⁴  revealed that Drosophila Triad1 (CG5709) binds in addition to Ubc7 (CG12799) to both cytoplasmic and nuclear proteins. These include cytoplasmic orthologs of signaling molecule Traf3 (TNF [tumor necrosis factor] receptor-associated factor-3) involved in TNF-receptor-mediated signal transduction and E3 ub ligase Hakai involved in ubiquitination of E-cadherin (respectively, CG4394 and CG10263), but also nuclear proteins, including the ortholog of the transcription factor HOX11L2 (CG7937) implicated in acute lymphoblastic leukemia development and the ortholog of the helix-loop-helix transcription factors NHLH1 and NHLH2 (CG3052) implicated in neurogenesis.^35-38 

Forced Triad1 expression in primary murine bone marrow cells inhibited colony formation in CFU-GM assays. Two N-terminal RING finger point mutants that are unable to bind the E2 UbcH7 did not inhibit clonogenic growth, indicating that the N-terminal RING finger of Triad1 is involved in the inhibition of clonogenic growth (Figure 6A,C). As the RING finger is involved in the catalytic activity (ie, Ub conjugation)^13,20  in virtually all E3 RING Ub ligases, this suggested that the ubiquitination of Triad1 target proteins caused the inhibition of clonogenic growth. In conjunction with E3 Ub ligases, UbcH7 is known to catalyze the formation of Ub chains that result in proteasomal degradation of tagged proteins. To further study the clonogenic inhibition of Triad1 in CFU-GM assays we repeated the experiments in the presence of proteasome inhibitors and found that this partially reversed the effects of Triad1 (Figure 6G-H). We conclude that proteasomal degradation of Triad1 targets plays an important role in myelopoiesis. In liquid cultures of transduced bone marrow cells, Triad1 expression resulted in a strong inhibition of proliferation whereas differentiation remained unaltered. The inhibition of proliferation was associated with a G₀/G₁ accumulation and increased apoptosis (Figure 7).

Triad1 is a member of the class of TRIAD (or RING-IBR-RING) proteins.^31,32  Since the description of this class of proteins, several TRIAD proteins have been reported to function as E3 Ub ligase, including Triad3, Parkin, Parc (pulmonary and activation-regulated chemokine), and p53RFP (p53-associated Parkin-like cytoplasmic protein).^39-42  These Ub ligases are involved in various biologic processes, including immunity, neuronal development, and p53 functioning. Triad3 targets Toll-like receptors 4 and 9 for ubiquitination.⁴⁰  By sensing antigens, Toll-like receptors play a role in both innate and adapted immunity. Ligand binding of these receptors result in nuclear factor[kappa]B (NF-κB) activation and Triad3 regulates the ubiquitination and concomitant turnover of these receptors, thereby controlling the intensity and duration of Toll-like receptor function.⁴⁰  Another well-studied TRIAD protein is Parkin, which is, like Triad1, highly expressed in fetal brain and involved in neuronal development.⁴³  Congenital mutations in Parkin cause juvenile parkinsonism, and acquired loss-of-heterozygosity of Parkin has been described in various human malignancies.^44-48  The TRIAD proteins Parc and p53RFP are involved in p53 function.^41,42  Parc functions as E3 Ub ligase that binds to p53. Although Parc does not ubiquitinate p53, it sequesters this protein in the cytoplasm. The other TRIAD protein, p53RFP, is a p53 target gene. p53RFP binds to the cyclin-dependent kinase inhibitor p21^waf1. Because p53RFP overexpression results in reduced p21^waf1 levels, p53RFP might function as Ub ligase for p21waf1.⁴¹  The majority of the reported TRIAD proteins have been shown to interact with UbcH7, which is in line with the observation that Triad1 also interacts with UbcH7. By using less stringent conditions in yeast 2-hybrid assays, we also observed an interaction between Triad1 and UbcH13 (not shown). Interestingly, UbcH13 catalyzes the formation of Ub chains (linked through lysine 63) that are not recognized by the proteasome but that alter the activity of proteins.^49-51  Two recent studies have shown that the TRIAD protein Parkin also binds to UbcH13 in addition to UbcH7 and that it can catalyze the formation of both lysine 48- and 63-linked Ub chains.^30,52  As Triad1 may bind UbcH13 it will be important to test whether it also exhibits activity that supports ubiquitination linked through lysine 63 of Ub. Nevertheless, the inhibition of clonogenic growth exerted by Triad1 described here depends at least for a part on targeting proteins for proteasomal degradation, because proteasome inhibition partially reversed the effects of Triad1. To further understand how the Ub-proteasome pathway contributes to myelopoiesis, it will be of great interest to identify which cellular proteins are targeted for ubiquitination by Triad1.

We thank Dr Martin Scheffner for valuable discussions and help in setting up ubiquitination assays and for providing us with pcDNA3.1-His-Ub, Herman Swarts for generating Triad1 Baculovirus, Dr Philip James for 2 hybrid vectors, Dr Yukio Okano for Ubc yeast 2-hybrid bait constructs, Dr Toshiaki Suzuki for pcDNA3.1-Flag-Ub, Dr G. Nolan for pLZRS and the phoenix cell line ϕ-NX-A.