Caspase-dependent cleavage of the mono-ADP-ribosyltransferase ARTD10 interferes with its pro-apoptotic function

Authors

  • Nicolas Herzog,

    1. Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Search for more papers by this author
  • Jörg D. H. Hartkamp,

    1. Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Search for more papers by this author
  • Patricia Verheugd,

    1. Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Search for more papers by this author
  • Fabian Treude,

    1. Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Search for more papers by this author
  • Alexandra H. Forst,

    1. Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Search for more papers by this author
  • Karla L. H. Feijs,

    1. Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Search for more papers by this author
  • Barbara E. Lippok,

    1. Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Search for more papers by this author
  • Elisabeth Kremmer,

    1. Helmholtz Zentrum München, Institut für Molekulare Immunologie, München, Germany
    Search for more papers by this author
  • Henning Kleine,

    1. Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Current affiliation:
    1. Abbott GmbH & Co. KG, Wiesbaden, Germany
    Search for more papers by this author
  • Bernhard Lüscher

    Corresponding author
    • Institute of Biochemistry and Molecular Biology, Medical School, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Aachen, Germany
    Search for more papers by this author

Correspondence

B. Lüscher, Institute of Biochemistry and Molecular Biology, Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University, Pauwelsstraße 30, 52074 Aachen, Germany

Fax: +49 241 8082427

Tel: +49 241 8088850

E–mail: Luescher@rwth-aachen.de

Abstract

ADP-ribosylation is a post-translational modification that regulates various physiological processes, including DNA damage repair, gene transcription and signal transduction. Intracellular ADP-ribosyltransferases (ARTDs or PARPs) modify their substrates either by poly- or mono-ADP-ribosylation. Previously we identified ARTD10 (formerly PARP10) as a mono-ADP-ribosyltransferase, and observed that exogenous ARTD10 but not ARTD10-G888W, a catalytically inactive mutant, interferes with cell proliferation. To expand on this observation, we established cell lines with inducible ARTD10 or ARTD10-G888W. Consistent with our previous findings, induction of the wild-type protein but not the mutant inhibited cell proliferation, primarily by inducing apoptosis. During apoptosis, ARTD10 itself was targeted by caspases. We mapped the major cleavage site at EIAMD406↓S, a sequence that was preferentially recognized by caspase–6. Caspase-dependent cleavage inhibited the pro-apoptotic activity of ARTD10, as ARTD10(1–406) and ARTD10(407–1025), either alone or together, were unable to induce apoptosis, despite catalytic activity of the latter. Deletion of the N–terminal RNA recognition motif in ARTD10(257–1025) also resulted in loss of pro-apoptotic activity. Thus our findings indicate that the RNA recognition motif contributes to the pro-apoptotic effect, together with the catalytic domain. We suggest that these two domains must be physically linked to stimulate apoptosis, possibly targeting ARTD10 through the RNA recognition motif to specific substrates that control cell death. Moreover, we established that knockdown of ARTD10 reduced apoptosis in response to DNA-damaging agents. Together, these findings indicate that ARTD10 is involved in the regulation of apoptosis, and that, once apoptosis is activated, ARTD10 is cleaved as part of negative feedback regulation.

Structured digital abstract

Abbreviations
ART

ADP-ribosyltransferase

ARTD

ADP-ribosyltransferase diphtheria toxin-like

FACS

fluorescence-activated cell sorting

HA

hemagglutinin

PI

propidium iodide

RRM

RNA recognition motif

Introduction

ADP-ribosylation is a post-translational modification in which the ADP-ribose moiety of the co-factor β–NAD+ is transferred onto a substrate, thereby releasing nicotinamide. This process is performed by intracellular ADP-ribosyltransferases (ARTDs or PARPs), of which there are 17 members in humans [1, 2]. Some of these enzymes modify their substrates by successively transferring multiple ADP-ribose units to generate poly-ADP-ribose chains. Others are only capable of mono-ADP-ribosylating their substrates. Intracellular poly-ADP-ribosylation is involved in a variety of cellular processes, including DNA repair, gene transcription, chromatin remodeling and signal transduction, with numerous substrates having been described [3-7]. In contrast, relatively little is known about the biological functions and the substrates of mono-ADP-ribosyltransferases.

The founding member of mono-ADP-ribosylating ARTDs is ARTD10 (formerly PARP10, EC 2.4.2.30), which was identified as a MYC-interacting protein that uses substrate-assisted catalysis as its enzymatic mechanism [8, 9]. When the steady-state subcellular distribution is analyzed, ARTD10 is found to be predominantly cytoplasmic. However, live-cell imaging revealed that ARTD10 shuttles between the cytoplasmic and nuclear compartments [10]. It interacts with MYC in the nucleus and co-localizes with the poly-ubiquitin receptor p62/SQSTM1 in the cytosol. p62/SQSTM1 has multiple functions, including a role in regulating the acquisition of cargo by the autophagosomal machinery [11, 12]. Our previous studies suggested that ARTD10 interferes with cell physiology because the wild-type protein inhibits co-transformation of rat embryo fibroblasts by MYC and Ha–RAS [8] and proliferation of established tumor cell lines [9]. However, it remained unclear how ARTD10 affects cell proliferation. To build on these initial observations, we established cell lines that express ARTD10 and the catalytically inactive mutant ARTD10-G888W under the control of a doxycycline-regulated promoter. We found that the wild-type protein but not the catalytically inactive mutant induced apoptosis, thereby explaining the anti-proliferative function of ARTD10.

Moreover, we observed that ARTD10 is substrate of caspases during apoptosis, resulting in two ARTD10 fragments, the first comprising amino acids 1–406 and the second comprising amino acids 407–1025. This cleavage separates the C–terminal catalytic domain from an N–terminal region that contains an RNA recognition motif (RRM) and a glycine-rich region. RRMs belong to a group of RNA-binding domains that are found in proteins involved in diverse aspects of RNA metabolism, ranging from splicing and editing, to export, translational regulation and degradation [13-15]. RRMs are domains of approximately 80–90 amino acids that are very abundant in eukaryotic proteins [14-16]. Combinations of various RNA-binding domains are common, and glycine-rich domains accompany RRMs, especially in plants [17, 18]. These two domains are also found in some animal proteins, for example nucleolin possesses an RRM and a glycine-rich domain [19]. In addition to interacting with RNA, many recent studies have documented that certain RRMs also interact with DNA or protein [20]. Several proteins involved in RNA metabolism and translation are targeted by caspases, including members of the heterogeneous nuclear ribonucleoproteins family [21-23]. This is part of a broad response in apoptotic cells to halt many housekeeping processes in order to allow efficient apoptosis [24]. Although the precise molecular functions of ARTD10 require further investigation, our findings suggest that the mono-ADP-ribosyltransferase activity is necessary but not sufficient to induce apoptosis, and that additional N–terminal regions contribute to the pro-apoptotic phenotype. Moreover, we find that reducing the expression of ARTD10 is protective to DNA damage-inducing, apoptosis-stimulating conditions. This suggests that ARTD10 activity must be tightly controlled in cells to balance apoptotic signaling.

Results

ARTD10 over-expression induces cell death

Previous findings indicated that ARTD10 inhibits cell proliferation, dependent on its catalytic activity [9]. To further evaluate this observation, we generated HeLa Flp–In T–REx cell clones with a stably integrated single copy of a transgene encoding either wild-type ARTD10 or ARTD10-G888W, a catalytically inactive mutant. Expression of the transgenes is induced by the addition of doxycycline (dox) (Fig. 1A). First, the growth characteristics of stable cell clones were assessed. Colony formation assays revealed that, upon induction of ARTD10 but not ARTD10-G888W, the cells did not form stainable colonies (Fig. 1B). This was observed with at least five individual clones, and results are shown here for two clones that express wild-type ARTD10 (wt3 and wt5) and two ARTD10-G888W clones (G888W1 and G888W4). Because the colonies developed from single cells and could only be assessed after 10–12 days, we next determined whether the repression of proliferation was an early or a late effect of ARTD10 expression. Therefore, we measured cell proliferation by counting living cells. Induction of ARTD10 by doxycycline considerably reduced the number of cells over 5 days, with clear effects being visible from day 3 onwards (Fig. 1C). In contrast, control cells or ARTD10-G888W-expressing cells were unaffected by the addition of doxycycline (Fig. 1C). To address the fate of the cells, we performed fluorescence-activated cell sorting (FACS) analysis. Expression of ARTD10, but not the catalytically inactive mutant, resulted in a substantial increase in propidium iodide (PI)-positive cells (Fig. 1D), indicative of cell death. Together, these findings suggest that over-expression of ARTD10 is toxic to these cells. Because ARTD10-G888W had no effect, it is likely that the effect of ARTD10 is due to inappropriate mono-ADP-ribosylation of its substrates.

Figure 1.

ARTD10 inhibits cell proliferation. (A) HeLa Flp–In ARTD10 (wt) and ARTD10-G888W (GW) cells were treated with 1 μg·mL−1 doxycycline for the indicated times (left panel, pool of wild-type cells) or for 16 h (right panel, individual clones of stably transfected cells). Protein expression was analyzed by immunoblotting with the indicated antibodies; tubulin was used as a loading control. (B) For the colony formation assay, 100 cells were seeded and grown for 11 days with or without doxycycline and subsequently stained with methylene blue. Two different clones each for ARTD10 and ARTD10-G888W were analyzed. (C) HeLa Flp–In ARTD10, ARTD10-G888W or vector control cells were grown for 5 days with or without the addition of doxycycline and counted daily. (D) Cells of the clones wt3 and G888W4 were grown with or without doxycycline for 3 days, and then stained with propidium iodide (PI) for 1 h prior to FACS analysis.

Induction of apoptosis by ARTD10

The induction of ARTD10 interferes with cell proliferation, and the appearance of PI-positive cells suggested that the cells die (Fig. 1). To determine whether ARTD10 stimulates apoptosis, the cells were analyzed by staining for annexin V at the cell surface. Annexin V staining is an indicator of the early stages of apoptosis, during which the membrane lipid phosphatidylserine, which is recognized by annexin V, accumulates in the outer leaflet of the plasma membrane [25]. ARTD10 or ARTD10-G888W expression was induced, the cells were stained with annexin V, and the PI-positive cells were excluded. Cells expressing ARTD10, but not the catalytically inactive mutant, stained positive for annexin V, an indication of apoptosis induction (Fig. 2A). Control cells were treated with the topoisomerase II inhibitor etoposide, a well-described inducer of apoptosis [26]. To validate these findings, cleavage of ARTD1 (formerly PARP1, EC 2.4.2.30) was monitored by immunoblotting. ARTD1, the best-characterized member of the ARTD family, is efficiently processed by caspases and serves as a marker of apoptosis [27, 28]. Cleaved ARTD1 became apparent in response to ARTD10 expression (Fig. 2B), supporting the concept that ARTD10 induces apoptosis. As expected, ARTD10-G888W was unable to stimulate ARTD1 processing.

Figure 2.

ARTD10 induces apoptosis. (A) ARTD10-expressing wt3, ARTD10-G888W-expressing G888W4 and control cells were treated with or without doxycycline for 3 days. As a positive control, cells were treated with etoposide for 24 h. The cells were subsequently stained for annexin V, with PI-positive cells excluded. (B) wt3 and G888W4 cells were treated with doxycycline for the indicated times, and cleavage of ARTD1 was analyzed by immunoblotting. Tubulin was analyzed as a control. (C) The cells were treated as in (B), and the cell-cycle profile was determined by FACS analysis. (D) ARTD10 expression was monitored in wt3 and G888W4 cells in response to doxycycline.

In addition to annexin V staining and ARTD1 cleavage, we monitored the cell cycle using the live cell stain Vybrant DyeCycle Violet. Induction of ARTD10 did not result in any obvious alteration in cell-cycle distribution, but stimulated the appearance of a sub-G1 peak, supporting the notion that ARTD10 induces apoptosis (Fig. 2C). Again the effect was only observed as a consequence of ARTD10 induction, and ARTD10-G888W did not cause an accumulation of cells in the sub-G1 fraction. Together, these data strongly suggest that induction of apoptosis is not merely a protein over-expression artifact but is dependent on catalytic activity. While monitoring the effects of ARTD10 in the HeLa Flp–In ARTD10 cells, we noticed that the induction of apoptosis was accompanied by the appearance of a roughly 50 kDa polypeptide (Fig. 2D, labeled N). Because this polypeptide was detected using an antibody that recognizes an epitope in the N–terminal half of ARTD10 (see Fig. 3A for a summary of the regions recognized by the various antibodies used) and because it was not seen in cells expressing ARTD10-G888W (Fig. 2D), we consider this fragment to represent a processed form of ARTD10 that appears upon activation of apoptosis.

Figure 3.

ARTD10 is a substrate of caspases during apoptosis. (A) Schematic representation of ARTD10. The following domains are indicated (from left to right): RRM, RNA recognition motif (amino acids 11–85); G–rich, glycine-rich region (amino acids 281–399); nuclear targeting, conserved region that mediates nuclear uptake (amino acids 435–528); E–rich, glutamate-rich region (amino acids 588–697); NES, nuclear export sequence (amino acids 598–607); UIMs, ubiquitin interaction motifs (amino acids 650–667 and amino acids 673–690); catalytic domain (amino acids 818–1013). The caspase cleavage site at EIAMD406↓S as well as the binding regions of the antibodies used in this study are indicated. (B) wt3 cells with induced ARTD10 expression (16 h) were treated with UV (20 mJ·cm−2, 254 nm) for the indicated times, and ARTD10 was analyzed using the monoclonal anitbody 5H11. (C) ARTD10 was induced for 16 h in wt3 cells, and these were then treated with doxorubicin (5 μm) for the indicated times. ARTD10 was analyzed using the monoclonal antibody 5H11 by Western blotting. (D) The experiment was performed as in (C), but using etoposide or staurosporine instead of doxorubicin. (E) THP–1 cells were treated with staurosporine (1 μm), and endogenous ARTD10 was analyzed by immunoblotting using the indicated antibodies. (F) The indicated cells were treated with UV in the presence or absence of Z–VAD–FMK. ARTD10 was analyzed using the indicated antibodies [see (A) for the epitopes recognized]. ARTD1 and tubulin were analyzed as controls in the various panels.

ARTD10 is a substrate of caspases

The findings described above not only indicate that ARTD10 induces apoptosis but also that this protein may be cleaved during apoptosis. Knowing that the best-studied family member ARTD1 is hydrolyzed at a specific site during apoptosis [27], we considered that ARTD10 may also be a caspase substrate [29]. To investigate this hypothesis, we used various antibodies to visualize the appearance of processed forms of ARTD10 (Fig. 3A). We treated ARTD10-expressing wt3 or THP–1 cells with UV, staurosporine, doxorubicin or etoposide, which efficiently trigger apoptosis. Analysis of ARTD10 processing in response to these treatments revealed two major fragments (Fig. 3B–E). Because the monoclonal antibody 5H11 recognizes an epitope within ARTD10(250–400) (Fig. 3A), the findings suggest that the apoptosis-specific 50 kDa fragment was derived from an N–terminal region of ARTD10 (Fig. 3). This fragment was not only detected for exogenous ARTD10, but also when endogenous ARTD10 was analyzed in THP–1 cells treated with staurosporine or etoposide (Fig. 3D). Using a polyclonal antiserum (891–6, see Fig. 3A), an additional weak band of roughly 95 kDa became apparent (Fig. 3E, labeled C). To more specifically address the identity of the C–terminal polypeptide, which contains the catalytic domain, we generated monoclonal antibodies against a C–terminal peptide (Fig. 3A). We selected two antibodies, 3H5 and 9E12, that are specific for ARTD10 and recognized the C–terminal ARTD10 fragment in THP–1 cells treated with etoposide or staurosporine (Fig. 3D). The combined apparent molecular masses of the two fragments added up to approximately the molecular mass of the full-length protein. Together with the specificity of the applied antibodies, the findings suggest that ARTD10 is cleaved once during apoptosis, producing an N- and a C–terminal fragment.

The hydrolysis of ARTD10 was rapid in response to the various apoptosis-inducing treatments. Within 2 h of UV treatment, the 50 kDa ARTD10 fragment became visible in wt3 cells, a time point at which no signs of apoptosis were detected in response to ARTD10 expression induced with doxycycline overnight. The timing of the appearance of this fragment was comparable to that of cleaved ARTD1 (Fig. 3B), indicating that ARTD10 is an early substrate during apoptosis. Similar observations were made when endogenous ARTD10 was analyzed and compared to ARTD1 in THP–1 cells (Fig. 3E).

To further evaluate a role for caspases in the cleavage of ARTD10, HeLa cells transiently transfected with an expression vector for hemagglutinin (HA)-tagged ARTD10 or wt3 cells with induced ARTD10 were treated with UV in the presence or absence of Z–VAD–FMK, a broad specificity caspase inhibitor [30]. Z–VAD–FMK prevented ARTD10 cleavage, and neither the N- nor the C–terminal fragment was detectable (Fig. 3F). Thus, these findings provide evidence for cleavage of ARTD10 by caspases during apoptosis, most likely at a single site.

Identification of the caspase cleavage site in ARTD10

In order to identify caspases that are capable of hydrolyzing ARTD10, tandem affinity-purified ARTD10-G888W was subjected to a cleavage assay with recombinant caspases. Caspases-1, -6, -7 and -8 (EC 3.4.22.52, EC 3.4.22.59, EC 3.4.22.60, and EC 3.4.22.61, respectively) were able to hydrolyze ARTD10 to various degrees in vitro, with caspase–6 (EC 3.4.22.59) being the most efficient (Fig. 4A). In addition to the N–terminal fragment, several other cleavage products were apparent, but these are not further analyzed here because we did not obtain evidence that these are also generated in cells. The size of the fragments generated in cells and in vitro suggested a cleavage site at approximately amino acid 400. Therefore, we subjected GST–ARTD10(206–459), a fragment of ARTD10 that was expected to contain the target site, to caspase treatment. This analysis revealed that this fragment was indeed cut by caspase–6 and caspase-8, comparable to the full-length protein, further supporting the notion that ARTD10 is a caspase substrate and that the cleavage site is within the predicted region of the protein (Fig. 4B).

Figure 4.

ARTD10 is cleaved at position D406 by caspase–6. (A) In vitro caspase assay using TAP-tagged ARTD10-G888W and recombinant caspases. The assay was stopped by adding sample buffer, and the proteins were subjected to gel electrophoresis and immunoblotting using the indicated antibodies. (B) In vitro caspase assay using a GST-tagged recombinant fragment of ARTD10 containing amino acids 206–459 and recombinant caspases. Proteins were separated by SDS/PAGE and stained with Coomassie Brilliant Blue. Cleaved proteins are indicated by arrows. (C) Control cells or HeLa cells expressing either HA-tagged ARTD10 or HA-tagged ARDT10-D406G were left untreated or were treated with staurosporine or UV (20 mJ·cm−2, 254 nm). In addition,cells were incubated with or without Z–VAD. ARTD10 expression was analyzed using the indicated antibodies. (D) HA-tagged ARTD10 and HA-tagged ARTD10-D406G were immunoprecipitated from lysates of transiently transfected HEK293 cells. The immunoprecipitated proteins were subjected to an in vitro caspase assay and the proteins were analyzed by Western blotting.

Comparing the amino acid sequence of ARTD10 with known target sequences of caspases, and taking the size of the N–terminal fragment into account, we identified a possible caspase target sequence at EIAMD406↓S. To evaluate this site, we mutated the aspartate at position 406 to a glycine (D406G) to abolish recognition and subsequent cleavage. HA-tagged ARTD10 and ARTD10-D406G were transiently expressed in HeLa cells, and their hydrolysis was analyzed in response to apoptosis induced by UV or staurosporine treatment. While the wild-type protein was efficiently fragmented, ARTD10-D406G was resistant, demonstrating that the hypothesized sequence is a major caspase cleavage site (Fig. 4C). In control experiments, cleavage of ARTD10 and ARTD1 was inhibited by Z–VAD–FMK. In addition to these cell-based assays, we also performed in vitro cleavage assays with immunoprecipitated ARTD10 and ARTD10-D406G. As shown before, caspase–6 was able to cut the wild-type but not the mutant protein (Fig. 4D). Caspase–1 (EC 3.4.22.36) was still able to cleave the mutant protein, generating an N–terminal fragment of approximately 70 kDa, a fragment that was already evident in the assay described above (Fig. 4A). Thus, these findings indicate that apoptotic caspases, exemplified by the effector caspase–6, hydrolyze ARTD10 at aspartate 406. We note that additional caspase–1 and caspase–7 cleavage sites exist in vitro that appear to be located further C–terminal than D406. We have not considered the potential caspase–7 site further because we did not detect a corresponding ARTD10 fragment in apoptotic cells.

Cleavage of ARTD10 inactivates its pro-apoptotic function

Exogenous expression of ARTD10 induces apoptosis, dependent on its catalytic activity (Fig. 2). Therefore, we addressed whether cleavage of ARTD10 was important to induce apoptosis. We expressed the non-cleavable ARTD10-D406G mutant in HeLa Flp–In T–Rex cells (Fig. 5A), and determined whether its induction interfered with cell proliferation (Fig. 5E). Similar to the wild-type protein, over-expression of ARTD10-D406G efficiently inhibited cell proliferation (measured in the two clones D406G1 and 2). Microscopic inspection of the cells suggested that this was indeed due to apoptosis. Approximately half of the cells became PI-positive and accumulated in the sub-G1 fraction 48 h after induction of ARTD10-D406G (Fig. 5G,H). All the ARTD10-D406G clones grew less well than the wild-type cells (Fig. 5E). This may be due to low-level expression of the transgene in the absence of doxycycline and a more efficient activation of apoptosis by ARTD10-D406G compared to the wild-type protein. This interpretation is consistent with the observation that slightly more cells were PI-positive and present in the sub-G1 fraction in response to ARTD10-D406G induction (Fig. 5G,H). This was true in all individual experiments performed. Thus, cleavage of ARTD10 is unlikely to induce pro-apoptotic activity of ARTD10.

Figure 5.

Cleavage of ARTD10 interferes with its pro-apoptotic function. (A) The expression of ARTD10, ARTD10-G888W and ARTD10-D406G was analyzed in lysates of stably transfected HeLa cells with or without doxycycline treatment. Glyceraldehyde-3–phosphate dehydrogenase (GAPDH) was used as a control. (B) Experiment performed as in (A) using stable cell lines expressing ARTD10 fragments. Tubulin was analyzed as a loading control. (C) Stable cell lines expressing ARTD10 and ARTD10(1–406) were treated with increasing amounts of doxycycline as indicated. The expression of the ARTD10 proteins was analyzed and compared to GAPDH as a control. (D) Radioactive ADP-ribosylation assay using immunoprecipitated GFP-tagged ARTD10 constructs expressed in U2OS cells. The GST-tagged catalytic domain [GST–ARTD10(818–1025)] was used as positive control, and the inactive version [GST–ARTD10(818–1025/G888W)] served as substrate. The Coomassie Brilliant Blue-stained gel is shown on the left, and autoradiographs are shown on the right. The right upper panel shows auto-modification of the GFP-tagged proteins; the right lower panel shows modification of the GST proteins. Short and long exposure refer to overnight at room temperature or at −80 °C, respectively. (E) The proliferation of HeLa cells expressing either ARTD10 or ARTD10-D406G was analyzed. The cells were treated with or without doxycycline for 5 days. The cells were counted daily. (F) Experimental performed as in (E) using HeLa cells expressing ARTD10(257–1025) or ARTD10(407–1025). (G) The number of propidium iodide-stained cells in response to doxycycline-induced protein expression was analyzed by FACS at the indicated time points. From left to right, the cells used express ARTD10, ARTD10-G888W, ARTD10-D406G, ARTD10(1–406), ARTD10(407–1025) and ARTD10(257–1025). (H) Experiment performed as in (G). The percentage of sub-G1 cells is shown.

To address the role of the fragments in apoptosis, we determined whether the C–terminal catalytic fragment is sufficient to induce apoptosis. We generated HeLa Flp–In T–Rex cells that express the C–terminal cleavage fragment ARTD10(407–1025) or the N–terminal fragment ARTD10(1–406). These fragments were efficiently expressed (Fig. 5B,C). Moreover, the C–terminal fragment, expressed and immunoprecipitated as a GFP-tagged protein, possessed catalytic activity (Fig. 5D). ARTD10(407–1025) was able to modify itself and a catalytically inactive ARTD10(818–1025/G888W) fragment to a similar extent as the wild-type protein. Thus, the C–terminal ARTD10 fragment generated in response to caspase activation is catalytically active. Neither of the fragments influenced cell proliferation (Fig. 5F) nor induced apoptosis (Fig. 5G,H). Thus, although catalytic activity of ARTD10 is necessary, it is not sufficient to trigger cell death.

To more specifically address the role of the RRM in the N–terminal region, we generated ARTD10(257–1025) (Fig. 5B). This longer mutant of ARTD10, similarly to ARTD10(407–1025), neither inhibited proliferation nor induced apoptosis (Fig. 5E–G). Thus, these findings suggest that the integrity of at least two domains of ARTD10, i.e. the catalytic domain and the RRM, are necessary for inhibition of proliferation and induction of apoptosis. Co-expressing ARTD10(1–406) and ARTD10(407–1025) was also not sufficient to induce apoptosis (data not shown), further indicating that the two domains must be physically linked.

DNA damage-induced apoptosis is ameliorated by knockdown of ARTD10

The findings above suggested that ARTD10 induces apoptosis when over-expressed. To evaluate the role of the endogenous protein, we knocked down ARTD10 and treated U2OS cells with DNA-damaging agents. We observed that the cells with reduced ARTD10 expression were less sensitive to UV and doxorubicin treatment. This was apparent when the morphology of cells was analyzed (data not shown) and when the cleavage of ARTD1 was monitored (Fig. 6). These findings suggest that ARTD10 sensitizes cells to apoptosis when stimulated with DNA-damaging agents.

Figure 6.

Reduced apoptosis after ARTD10 knockdown. (A) U2OS cells were transiently transfected using Oligofectamine™ (Invitrogen) and an siOligo pool targeting ARTD10 mRNA. After 48 h, the cells were treated with UV (20 mJ·cm−2, 254 nm) and incubated further for the indicated times. Cells were lysed and knockdown efficiency and ARTD1 cleavage were analyzed by Western blotting. Tubulin was used as a loading control. (B) Experiment performed as in (A), except that the cells were treated with doxorubicin (5 μm) for the indicated times. The proteins were analyzed as in (A).

Discussion

Mono-ADP-ribosylation by resident intracellular ADP-ribosyltransferases has only recently been suggested as a post-translational modification, with ARTD10 being the founding member of this sub-class of ARTD enzymes and the model enzyme for their catalytic mechanism [9]. The functional consequences of ARTD10, and mono-ADP-ribosylation more generally, are just starting to be unraveled. In addition to the catalytic domain, ARTD10 harbors various distinct protein elements, including ubiquitin interaction motifs, a nuclear export sequence and a nuclear-targeting sequence, as well as an RNA recognition motif (RRM) and a glycine-rich domain (see Fig. 3A for the localization of the domains), which suggest that ARTD10 is involved in various processes throughout the cell. Recent findings demonstrate that the nuclear export sequence and the nuclear-targeting sequence are functional [8, 10], promoting shuttling of ARTD10 between the nuclear and cytoplasmic compartments. In the present study, we expand on our previous observation that ARTD10 interferes with cell proliferation [9]. We provide evidence that ARTD10 induces apoptosis, explaining how this protein inhibits proliferation. Moreover, we find that repression of ARTD10 interferes with induction of apoptosis by DNA-damaging agents.

The comparison between ARTD10 and ARTD10-G888W indicated that the ability to mono-ADP-ribosylate is critical for inducing apoptosis. Therefore, it was surprising that the C–terminal fragment ARTD10(407–1025), which contains the catalytic domain, did not trigger this self-destructive pathway. As expected from various recombinant ARTD10 proteins analyzed previously [9], ARTD10(407–1025) possesses ADP-ribosylating activity. Moreover, ARTD10(257–1025) was also unable to stimulate cell death, despite a shorter N–terminal deletion compared to ARTD10(407–1025). This indicates that, although catalytic activity is important for induction of apoptosis, it is not sufficient. Thus, the question arises why these N–terminally truncated mutants of ARTD10 are incapable of inducing apoptosis. We note that both mutants lack the RRM, suggesting that this domain is functionally relevant. Our findings indicate that the RRM, by binding to RNA but potentially also to other macromolecules [20], may affect targeting of ARTD10 to distinct substrates and thus their mono-ADP-ribosylation, thereby affecting specific cellular functions. The ability of ARTD10 to shuttle between the nuclear and cytoplasmic compartments suggests a possible link of ARTD10 to RNA transport [10]. Alternatively, it is also possible that the interaction with nucleic acids through the RRM is involved in regulating catalytic activity. We consider the latter possibility less likely because we have previously tested whether RNA or DNA affect catalytic activity [9]. These experiments did not reveal any striking regulation. A role for the RRM and possibly the glycine-rich domain in targeting ARTD10 to specific molecules or structures and their subsequent regulation by mono-ADP-ribosylation is an attractive model that needs to be addressed in future studies.

In vitro, caspase–6 was the enzyme that cleaved ARTD10 at D406 most efficiently. Although this caspase has been categorized as an executioner caspase because of its role in cleaving nuclear lamins [31], relatively few apoptotic substrates are known. More recent studies have demonstrated that caspase–6 hydrolyzes caspase-2, caspase–3 and caspase–8. Moreover, it has a number of neuronal substrates that appear to be cleaved at least in part independently of apoptosis [32]. As ARTD10 is expressed in neuronal cells (B.E. Lippok and B. Lüscher, unpublished results), it will be of interest to determine whether caspase–6-dependent processing of ARTD10 is relevant to neurons, possibly independently of apoptosis.

The cleavage site EIAMD406↓SP in ARTD10 is unusual, as the P2–P4 positions do not contain acidic residues [33, 34]. Instead, an acidic residue is found at the P5 position. There is little information about the relevance of P5. Other determinants within ARTD10 may contribute to the observed selectivity. Moreover, an SP motif is present C–terminal to the cleavage site. This offers the possibility that the hydrolysis of ARTD10 by caspases is regulated by phosphorylation of the SP motif in ARTD10 by proline-directed kinases. Indeed, many examples exist that demonstrate regulation of cleavage by P1 phosphorylation, and this is particularly well documented for Casein kinase 2-dependent phosphorylation [35-37]. Thus, it will be important to address whether S407 is phosphorylated, and, if so, how this affects ARTD10 processing by caspases.

In summary, our findings indicate that ARTD10 mono-ADP-ribosylates key substrate(s), which subsequently induce apoptosis and activation of caspases. This is compatible with the finding that knockdown of ARTD10 reduces UV- and doxorubicin-induced apoptosis, indicating that ARTD10 sensitizes cells to DNA damage. However, analysis of the N–terminal deletion mutants suggest that the relevant substrates are only accessible when the N–terminal region is present, implying that targeting and substrate recognition are most likely essential for this effect. The RRM, possibly together with the glycine-rich region, may mediate this targeting. It is of note that the pro-apoptotic function of ARTD10 is inhibited once apoptosis has been initiated and caspases have been activated. A possible explanation for this effect is that ARTD10 fulfils a housekeeping function that must be inhibited during the ordered breakdown of the cell. Indeed, many proteins associated with such basal functions, for example proteins involved in gene transcription and protein translation, are targeted by caspases during apoptosis [22, 24]. We postulate two such housekeeping functions for ARTD10 that may be of sufficient importance to justify cleavage. ARTD10 shuttles between the cytoplasmic and nuclear compartments [10]. Moreover, we have recently observed that ARTD10 interacts with RNA (B.E. Lippok and B. Lüscher, unpublished results). Thus, ARTD10 may be involved in the transport of RNA from the nucleus to the cytosol. Such a hypothesis is also supported by our identification of the small GTPase Ran, which controls nuclear–cytosolic transport, as an ARTD10 substrate [38]. RNA transport is a housekeeping function that may well be repressed during apoptosis, e.g. to conserve energy. A second function may relate to the use of NAD+. This co-factor is important for glycolysis, and loss of NAD+ induces necrosis because of the resulting inhibition of ATP production and because of the energy consumption during re-synthesis of NAD+. This is best exemplified by ARTD1, which is inactivated by caspase–3-dependent cleavage [7]. Although cleavage of ARTD10 by a caspase during apoptosis does not interfere with catalytic activity per se, it may, as discussed above, alter targeting of the enzyme and thereby reduce NAD+ consumption. These hypotheses need to be addressed in the future to understand more completely the role that ARTD10 plays in controlling apoptosis.

Experimental procedures

Cell culture and transfections

All cell lines were cultured at 37 °C with 5% CO2. RPMI-1640 (Life Technologies, Carlsbad, CA, USA) was used for THP–1 cells. Dulbecco's modified Eagle's medium/Glutamax I (Life Technologies) medium was used for all other cell lines. All media were supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. In addition, 5 μg·mL−1 blasticidin and 100 μg·mL−1 hygromycin were added as selection markers to the medium used for the stably transfected cell lines. Transfections of plasmids were performed using the calcium phosphate method [39]. UV (20 mJ·cm−2 at 254 nm) was applied to cells washed once in NaCl/Pi. Knockdown of ARTD10 was mediated by transfection of an siRNA oligonucleotide pool (Dharmacon, Lafayette, CO, USA) targeting ARTD10 mRNA using Oligofectamine™ (Invitrogen Life Technologies, Carlsbad, CA, USA) according to the manufacturer's instructions.

Stable cell lines

HeLa Flp–In T–Rex cells (a kind gift from Steven Taylor, The University of Manchester, Manchester, UK) were transfected with pcDNA5/FRT/TO-ARTD10wt or plasmids containing the respective mutants and pOG44 (Invitrogen). The transfected cells were selected using 5 μg·mL−1 blasticidin and 200 μg·mL−1 hygromycin. After initial selection, monoclonal cell lines were established.

Plasmids and mutagenesis

pEVRF0-HA-ARTD10 has been described previously [9]. pEVRF0-HA-ARTD10-D406G was constructed by site-directed mutagenesis. pcDNA5/FRT/TO-ARTD10wt, G888W or D406G were constructed using pcDN5/FRT/TO (Invitrogen) and inserts from the respective pEVRF0-HA vectors. pcDNA5/FRT/TO-ARTD10(1–406), pcDNA5/FRT/TO-ARTD10(407–1025) and pcDNA5/FRT/TO-ARTD10(257–1025) were created using the Gateway system (Invitrogen).

Antibodies

ARTD10 was detected by various antibodies. The regions recognized by the 5H11 (monoclonal, rat), 3H5 (monoclonal, rat), 9E12 (monoclonal, rat), 891–6 (polyclonal, rabbit) and E09 (polyclonal, rabbit) antibodies are indicated in Fig. 3A. Monoclonal antibodies recognizing the C–terminus of ARTD10, i.e. 3H5 and 9E12, were generated against the C–terminal peptide VPRASPDDPSGLPGRSPDT(1007–1025). Additional antibodies used were anti-α–tubulin (Sigma Aldrich, Saint Louis, MO, USA, T–5168), anti-ARTD1/PARP1 (Roche, Penzberg, Germany, 118352380001) and anti-HA (Roche, 3F10). Secondary antibodies used were horseradish peroxidase-labeled goat anti-mouse (115–035–146), goat anti-rabbit (111–035–144) and goat anti-rat (112–035–008), purchased from Jackson Immunoresearch (West Grove, PA, USA).

Protein purification

GST fusion proteins were expressed in Escherichia coli BL21 (DE3)pLysS (Stratagene, La Jolla, CA, USA). Bacteria were grown in Luria–Bertani medium at 37 °C to an attenuance at 600 nm of 0.6, and expression was induced by addition of isopropyl thio-β–d–galactoside (1 mm), before further growth for 4 h. After lysis of the bacterial pellet in lysis buffer [20 mm Tris/HCl, pH 8.0, 150 mm NaCl, 1 mm EDTA, 5 mm dithiothreitol, 1 mm Pefa-Bloc (Roche), 1% v/v aprotinin (Sigma-Aldrich)] complemented with lysozyme (100 mg·mL−1), the GST fusion proteins were bound to glutathione Sepharose (Amersham Biosciences, GE Healthcare, Little Chalfond, UK). The beads were washed three times in 100 mm Tris/HCl, pH 8.0, 120 mm NaCl, at 4°C and eluted with elution buffer (100 mm Tris/HCl, pH 8.0, 120 mm NaCl, 20 mm glutathione). Tandem affinity purification-tagged ARTD10 was purified as described previously [9].

Colony formation assay

HeLa cells expressing the various ARTD10 proteins were seeded at 100 cells/6 cm well. Then 1 μg·mL−1 of doxycycline was added to induce expression of the transgene. Doxycycline was replenished every 2 days. Medium was changed after 4 days. On day 11, the cells were washed once in phosphate-buffered saline at room temperature and subsequently stained with 0.2% methylene blue in methanol for 30 min. The plates were washed, dried and scanned for documentation.

Cell proliferation assays

HeLa cells were seeded in triplicate at a density of 3 × 104 cells per well. The cells were then induced for expression of the protein by addition of doxycycline (1 μg·mL−1). Counting was performed each day using the CASY® Technology cell counter (BD, Franklin Lakes, NJ, USA) with three repetitions per sample. For analysis of the cell cycle, the respective clones were seeded in 6 cm dishes at a density of 1 × 104 cells per dish. Expression of the proteins was induced by doxycycline (0.2 μg·mL−1) for 0–72 h. The cells were harvested from the plates and from the supernatant and stained with Vybrant Dyecycle Violet (1 μL/106 cells in 1 mL Dulbecco's modified Eagle's medium) and PI (50 μg·mL−1) for 30 min, and subsequently 2 × 105 cells were measured using a FACSCanto II (BD, Franklin Lakes, NJ, USA). Early apoptotic cells were detected using an annexin V binding assay. The cells were seeded on 6 cm dishes at a density of 8 × 104 cells per dish, incubated for 24 h and then stimulated with 1 μg·mL−1 doxycycline for 3 days. Control cells were treated with 20 μm etoposide. At the indicated time points, the cells were treated using an FITC Annexin V apoptosis detection kit I according to the manufacturer's instructions (BD Pharmingen, Franklin Lakes, NJ, USA), and analyzed by FACS.

ADP-ribosylation assay

GFP-tagged ARTD10 constructs were expressed in U2OS cells and immunoprecipitated using a GFP antibody. The activity of ARTD10 was measured by incorporation of radioactively labeled 32P–NAD+. The immunoprecipitated proteins and substrates were incubated in 30 μL assay buffer (50 mm Tris pH 8.0, 0.2 mm dithiothreitol, 4 mm MgCl2) at 30 °C. The reaction was performed in assay buffer additionally containing 50 μm β–NAD+ and 1 μCi 32P–NAD+. After 30 min, the reaction was stopped by the addition of 4× sample buffer, and the samples were subjected to SDS/PAGE. After Coomassie Brilliant Blue staining and drying of the gel, medical X–ray films (Fujifilm, Tokyo, Japan, 100NIF) were used to visualize modified proteins.

Caspase assay

Recombinant human caspases (0.1 unit of each enzyme; PromoKine, Heidelberg, Germany) were incubated with the protein of interest in 30 μL caspase reaction buffer (50 mm HEPES pH 7.2, 50 mm NaCl, 0.1% CHAPS, 10 mm EDTA, 2.5% glycerol, 5 mm dithiothreitol). After incubation for 1–2 h at 37 °C, the reaction was stopped by addition of 4× sample buffer, and the fragments were analyzed by SDS/PAGE followed by Western blotting and immunodetection or Coomassie Brilliant Blue staining.

Immunoblotting and immunoprecipitation

Cells that were subjected to immunoblot analysis were lysed in RIPA buffer (10 mm Tris/HCl, pH 7.4, 150 mm NaCl, 1% Nonidet P–40, 1% deoxycholic acid, 0.1% SDS, 0.5% Trasylol) containing a protease inhibitor cocktail (Proteobloc, Fermentas; Waltham, MA, USA), on ice. The lysates were sonicated for 15 min using a BioRaptor (Diagenode, Liege, Belgium) to destroy the genomic DNA, and cleared by centrifugation at 16 100 g. Proteins were separated by 10–12% SDS/PAGE, and then blotted on nitrocellulose membranes for further detection by specific antibodies. Lysates used for immunoprecipitation of proteins subjected to in vitro caspase treatment were also generated in RIPA buffer. The lysates were incubated with IgG Sepharose and a specific antibody at 4 °C for 24 h. The beads were washed twice with RIPA and twice with high-salt RIPA buffer (containing 500 mm NaCl), and then used immediately in a caspase assay.

Acknowledgements

We thank Steven Taylor (The University of Manchester, Manchester, UK) for the HeLa Flp–In T–Rex cells. This work was supported by a grant from the Deutsche Forschungsgemeinschaft (LU 466/15–1) and by the START program of the Medical School of the Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University to B. Lüscher.

Ancillary