ORIGINAL ARTICLE Characterization of the Ubiquitin–Proteasome System in Bortezomib-Adapted Cells

Characterization of the ubiquitin–proteasome system in bortezomib-adapted cells

TRu¨ckrich1, M Kraus1, J Gogel1, A Beck2, H Ovaa3, M Verdoes4, HS Overkleeft4, H Kalbacher5 and C Driessen1,6

1Department of Medicine II, University of Tu¨bingen, Tu¨bingen, Germany; 2ZfKM GmbH, Heilbronn, Germany; 3Division of Cell Biology II, The Netherlands Cancer Institute, Amsterdam, The Netherlands; 4Institute of Chemistry, Bio-organic Synthesis, Leiden University, Leiden, The Netherlands; 5Interfaculty Institute of Biochemistry, University of Tu¨bingen, Tu¨bingen, Germany and 6Department of Oncology and Hematology, Cantonal Hospital St Gallen, St Gallen, Switzerland

Resistance towards the proteasome inhibitor bortezomib is Alternative ‘immunoproteasome’ active subunits (b1i, b2i, b5i) may poorly understood. We adapted the HL-60, ARH-77 and AMO-1 be expressed. All active subunits are threonine proteases that can be cell lines (myeloid leukemia, plasmocytoid lymphoma, myelo- 7 ma) to bortezomib exceeding therapeutic plasma levels, and targeted by several inhibitors with differing individual specificities. compared characteristics of the ubiquitin–proteasome system, ATP-dependent substrate recruitment, deubiquitination and unfold- alternative proteases and the unfolded protein response (UPR) ing are executed by the 19S regulatory complex, which together between adapted cells and parental lines. Adapted cells with the 20S core forms the 26S proteasome. Alternative regulatory showed increased transcription rates, activities and polypep- complexes exist, including the interferon-inducible 11S activator, tide levels of the bortezomib-sensitive b5, but also of the b2 also called proteasome activator (PA) 28. proteasome subunit and consistently retained elevated levels of active b1/b5-type proteasome subunits in the presence of Recruitment of putative substrates to the proteasome is therapeutic levels of bortezomib. Bortezomib-adapted HL-60 regulated by the balance between E3 ubiquitin (Ub) ligases cells showed increased expression and proteasome associa- and deubiquitinating enzymes, including the Ub-specific tion of the 11S proteasome activator, and did not accumulate proteases (USP) and Ub C-terminal hydrolases.5,8–10 Substrate poly-ubiquitinated protein, activate the UPR or UPR-mediated recruitment in part occurs through ER-associated degradation, apoptosis in response to bortezomib. The rate of protein where misfolded protein is retro-translocated to the cytosol for biosynthesis was reduced, and the transcription of chaperone 11 genes downmodulated. We did not observe major changes in proteasomal degradation. Accumulation of misfolded proteins the activities of TPPII, cathepsins or deubiquitinating pro- in the ER induces the UPR, resulting in a decreased rate of teases. We conclude that different types of bortezomib-adapted general transcription, whereas protein expression of selected cell lines, including myeloma, show similar patterns of changes chaperones (for example, GRP78/BiP, heat-shock proteins) is in the proteasomal machinery which result in residual protea- increased.12 Prolonged activation of the UPR finally leads to some activity in the presence of bortezomib and a quantitative programmed cell death by several pathways, including a balance between protein biosynthesis and destruction. pathway mediated by caspase 4 and the pro-apoptotic protein Leukemia (2009) 23, 1098–1105; doi:10.1038/leu.2009.8; 12,13 published online 19 February 2009 GADD153/CHOP. Proteasome inhibition by bortezomib Keywords: proteasome inhibitor; bortezomib resistance; active site- leads to the accumulation of misfolded protein especially in the directed probe ER, which triggers the UPR-mediated apoptosis.14 In PI-resistant cells, alternative proteases or proteolytic systems may take over tasks of the proteasome, as suggested for lymphoma cells Introduction adapted to the irreversible proteasome inhibitor 4-hydroxy-5- iodo-3-nitrophenylacetyl-(Leu)3 vinyl sulfone (NLVS), where the Proteasome inhibition has emerged as a new therapeutic option activity of another multi-subunit protease complex, tripeptidyl against cancer.1 The proteasome inhibitor (PI) bortezomib is peptidase II (TPPII), was upregulated and rescued cells from 15,16 approved for the treatment of multiple myeloma and mantle cell PI-challenge. A significant amount of protein breakdown 17 lymphoma, and further clinical applications of bortezomib as also occurs in lysosomes, mostly mediated by cathepsins. an antineoplastic agent are under investigation.2–4 Bortezomib Adaptive changes in the lysosomal proteolytic system may disrupts the equilibrium between protein biosynthesis and similarly contribute to bortezomib resistance. protein degradation, which ultimately results in apoptosis Biology-driven strategies to overcome bortezomib resistance triggered through the unfolded protein response (UPR). How- include the targeting of alternative active sites or proteins ever, not all types of malignancies are bortezomib-sensitive, and associated with the proteasome, combining bortezomib with both primary and secondary bortezomib resistance occur in agents that sensitize for UPR-mediated apoptosis, or combining myeloma.4 Little is known about the molecular changes in the bortezomib with inhibitors of putative proteolytic ‘rescue’- proteolytic network when bortezomib resistance is acquired. pathways, such as TPPII. To further guide such approaches, we The proteasome is responsible for the bulk degradation of (mostly) adapted three different types of cell lines, including a myeloma poly-ubiquitinated cytosolic proteins.5,6 Three types of subunits in cell line, to bortezomib to characterize the fundamental changes the inner b rings of the 20S core particle carry the proteolytic in the proteolytic network of malignant cells with acquired ‘caspase-’ (b1), ‘trypsin-’ (b2) and ‘chymotrypsin-like’ (b5) activities. bortezomib resistance.

Correspondence: Professor Dr C Driessen, Onkologie/Ha¨matologie, Kantonsspital St Gallen, Rorschacherstrasse, St Gallen CH-9001, Materials and methods Switzerland. E-mail: [email protected] Received 14 July 2008; revised 23 November 2008; accepted 6 For a more detailed description, please refer to the Materials and January 2009; published online 19 February 2009 methods section in Supplementary Information. Bortezomib-adapted cells TRu¨ckrich et al 1099 Cell culture, lysis and immunoblot contained 20 nM bortezomib (Ortho Biotech, Neuss, Germany). Cell lines were cultured in FCS-supplemented RPMI-1640 with All antibodies except those against different USP were obtained penicillin/streptomycin. Medium for adapted cells additionally commercially.

Metabolic labeling For quantiﬁcation of protein biosynthesis, cells were labeled with 35S-methionine/cysteine and the radioactivity incorporated into the protein fraction of cell lysates was measured using a scintillation counter.

Activity assays Active site-directed probes were employed for (semi-) quantitative assessment of protease activities: labeling of active proteasome subunits was performed in live cells with the proteasome-specific affinity probes MV151 or dansyl-sulfona- midohexanoyl-(6-aminohexanoyl)2-(Leu)3 vinyl sulfone, as described.18,19 Cell lysates were separated by SDS-PAGE. The fluorescent probe MV151 allows for visualization of proteasome activities by scanning the gel for fluorescence of its bodipy moiety. Quantification of fluorescent bands was performed using the program AIDA.

Immunoprecipitation For immunoprecipitation of proteasome, cell lysates were incubated with anti-a2-agarose beads (Biomol, Hamburg, Germany) in IP buffer (25 mM Tris-HCl pH 7.3, 10 mM KCl, 0.5 mM EDTA, 2 mM MgCl2,1mM DTT, 1 mM ATP, 5% glycerol). Beads were washed with 0.1 M NaCl in IP buffer and eluted with 2 M NaCl in IP buffer (without DTT, ATP, glycerol). Eluates were concentrated by ultraﬁltration and analyzed by SDS-PAGE and immunoblot.

Cytotoxicity assay Cytotoxicity of drugs was determined by an MTS assay (Promega, Mannheim, Germany) and absorbance measured at 492 nm. Results represent mean values from triplicates.

Results

Myeloma cell lines can be adapted to bortezomib In analogy to the bortezomib-adapted HL-60a leukemia cell subline reported earlier,20 we adapted the myeloma cell line AMO-1 and the plasmocytoid lymphoma-type subline ARH-77 to bortezomib by continuous incubation with increasing concentrations of the drug. The subclones HL-60a, AMO-1a, ARH-77a showed an IC 50 of 4600, 4160 and 500 nM, respectively, compared to 30, 8 and 20 nM for the parental cell lines (Figures 1a–c), corresponding to a consistent at least 20-fold increase in the tolerated bortezomib dose. When

Figure 1 Resistance to the proteasome inhibitor bortezomib in bortezomib-adapted cells. (a) Bortezomib-adapted HL-60a cells were compared to non-adapted HL-60 cells and to adapted cells cultured without bortezomib for 14 days (HL-60a 14d) in an MTS cytotoxicity assay. After incubation with different concentrations of bortezomib for 48 h, cellular dehydrogenase activity was measured by photometric detection of MTS turnover. (b) Bortezomib-adapted ARH-77a and AMO-1a cells were examined in an MTS assay in comparison to their non-adapted counterparts (ARH-77, AMO-1), as in (a). (c) Estimated IC50 values (nM) for non-adapted and adapted cells treated with bortezomib.

Leukemia Bortezomib-adapted cells TRu¨ckrich et al 1100

Figure 2 Increased proteasome activity in bortezomib-adapted cells. (a) HL-60 cells and bortezomib-adapted HL-60a cells were incubated with the fluorophor-coupled activity-based probe MV151. The probe covalently binds to the b1, b2, b5 proteasome subunits in an activity-dependent fashion. After cell lysis and separation by gel electrophoresis, MV151-labeled active proteasome subunits (b1, b2, b5) were detected using a fluorescence scanner. Cells incubated with the proteasome inhibitor ZL3VS were used as a negative control. Protein loading was assayed in a western blot for GAPDH. (b) ARH-77, AMO-1 and bortezomib-adapted ARH-77a and AMO-1a cells were treated with 20 nM bortezomib or 50 mM NLVS (negative control) for 16 h, before incubation with the MV151 probe for proteasomal subunit activities. Active subunits of the proteasome and immunoproteasome (i) were detected as described above. Protein loading was assayed by western blot for b-actin. (c) Quantification of active proteasomal subunits from bortezomib-adapted and non-adapted cells. Signal intensities of the fluorescent activity-based probe were directly measured after separation by SDS-PAGE using a fluorescence laser scanner. Intensities of bands corresponding to the b2 (black bars) and b5/ b1 (striped bars) proteasomal subunits in untreated cells are shown normalized to band densities of their respective protein loading controls.

HL-60a cells were grown without bortezomib for 14 days also showed a sizable increase in active b1/b5 subunits, (HL-60a 14d) and then re-challenged by addition of the drug, ARH-77a cells showed no such change. HL-60a, ARH-77a cell viability was identical to HL-60a cells (Figure 1a), indicating and AMO-1a cells all retained higher levels of residual active that bortezomib-adapted cells undergo a persistent adaptation b1/b5 proteasome subunits in the presence of bortezomib than process. control cells treated with the drug. This bortezomib-induced change in the proteasomal activity pattern persisted when HL-60a cells were incubated in bortezomib-free medium (wash Bortezomib-adapted cells show similar patterns of out) for up to 7 days, either with or without re-challenge with changes in active proteasome subunits pulsed bortezomib (Supplementary Figure S1). To characterize the activity profile of proteasome subunits, we used the cell-permeable, activity-specific proteasome affinity label MV151. In agreement with our previous observations in Bortezomib-adapted cells are uniformly resistant to HL-60a cells (20, Supplementary Figure S1), labeling of b1/ b5-targeted proteasome inhibitors but show proteasome active subunits in intact HL-60a cells confirmed variability in their sensitivity towards pan-proteasome the upregulation of both the b1/b5-type and the b2-type of inhibitors proteasome activity (Figure 2a). Similarly, an increase in active To establish the selectivity of bortezomib resistance b2-type of subunits was also observed with ARH-77a cells and in bortezomib-adapted cells, the different cell lines were AMO-1a cells (Figure 2b). However, although AMO-1a cells challenged with different types of proteasome inhibitors as well

Leukemia Bortezomib-adapted cells TRu¨ckrich et al 1101 subunits of both the constitutive and the immunoproteasome in an irreversible manner. As expected, the viability of all bortezomib-adapted cell lines was unaffected by 40 nM bortezomib, in contrast to the parental control lines (Figure 3). All three types of bortezomib-adapted cells showed a similar selectivity in tolerance for bortezomib, NLVS and ZL3VS, whereas lactacystin and epoxomicin were less selectively tolerated by the bortezomib-adapted cells. By contrast, HL-60 and ARH-77 cells showed only minor variability in the dose– response relationship to daunorubicin between adapted cells and control cells, whereas AMO-1a cells were less sensitive towards daunorubicin than their parental line.

Upregulation and variation in the composition of proteasomes in bortezomib-adapted cells To analyze the composition and expression of the proteasome in bortezomib-adapted cells in further detail, we ﬁrst focused on HL-60a cells (Figure 4a). The expression of immunoproteasome subunits showed no gross difference between bortezomib- adapted and parental cells, and little effect was also observed for the non-active a4 subunit. Interestingly, the 11S activator a-subunit (PA28a) was upregulated in HL-60a cells, in contrast to the 19S subunit Rpn1. We did not detect changes on the polypeptide level that were directly related to either the addition of pulsed bortezomib or to the duration of the bortezomib washout phase in HL-60a cells, indicating that the changes observed on the polypeptide level were adaptive and not reactive in nature. The increased expression of the 11S polypeptide, in contrast to 19S, in HL-60a cells suggested a higher proportion of 11S–20S proteasome complexes in comparison to non-adapted cells. Indeed, using a 20S-directed antiserum, a signiﬁcant amount of 11S polypeptide (PA28a) was co-immunoprecipitated with 20S proteasomes only from HL-60a cells, but not from the HL-60 parental line (Figure 4b). Thus, bortezomib-adapted HL-60a cells contained higher levels of proteasome proteins and were also characterized by a larger fraction of proteasomes associated with the 11S activator protein. A similar change in the expression level of proteasome polypeptides was also observed in ARH-77a and AMO-1a cells: in particular the b2 and b5-types of polypeptides increased in the adapted cells, whereas an increase in b1 expression was only observed in AMO-1a cells (Figure 4c).

Comparative gene expression analysis reveals upregulation of proteasomal subunits and Figure 3 Bortezomib-adapted cells are less sensitive towards downregulation of chaperones in bortezomib-adapted different proteasome inhibitors. HL-60, ARH-77 and AMO-1 cells cells (black bars) and their bortezomib-adapted counterparts (striped bars) To examine the transcription levels of proteasome genes, and to were incubated for 48 h with proteasome inhibitors bortezomib (bort.), identify additional genes in the ubiquitin–proteasome system epoxomicin (epoxo.), lactacystin (lacta.), NLVS and ZL3VS, or with the cytotoxic daunorubicin (dauno.) at the concentrations indicated. that may mediate resistance to proteasome inhibitors, we Sensitivity to the drugs was assessed by measuring metabolic activity employed a comparative gene expression analysis of bortezo- of the cells as turnover of the MTS dehydrogenase substrate. Results mib-adapted and control cells. We used a commercial were normalized to untreated control cells. microarray, which contains 1270 genes associated with the ubiquitin–proteasome system. To focus on persistent (adaptive) cellular changes associated with resistance to PI and to exclude short-lived (reactive) effects of proteasome inhibition or residual as with daunorubicin as a conventional cytotoxic drug control. bortezomib activity, HL-60a cells were kept in bortezomib-free The vinylsulfone-type inhibitors carboxybenzyl-(Leu)3 vinyl culture for 3 days (HL60a 3d) before comparison to HL-60 cells. sulfone (ZL3VS) and NLVS preferentially inhibit proteasomal Supplementary Tables 1 and 2 include mean results of two b1/b5-type active subunits in an irreversible fashion, whereas independent experiments; only genes that consistently showed the borate-type inhibitor bortezomib binds these targets in a 41.7-fold homonymous changes were included. Based on this reversible manner. Epoxomicin and lactacystin target all active analysis, 23 genes of the ubiquitin–proteasome system were

Leukemia Bortezomib-adapted cells TRu¨ckrich et al 1102 detected as upregulated in HL-60a cells (Supplementary Table 1), and 34 as downregulated (Supplementary Table 2), compared to HL-60 control cells. The top five genes with the highest increase in mRNA expression included the PA28 proteasomal activator b subunit. In general, we observed consistent upregulation of genes encoding proteasomal subunits (b1, b2 and a7 of the 20S proteasome), as well as of the a- and b-subunits of the PA28 (11S) proteasomal activator. Among the 34 genes that were consistently downregulated in HL-60a cells, the top five genes included four genes that are linked to protein folding and reactions to misfolded protein accumulation: HSP40 and HSP70 heat-shock proteins (mean reduction 89 and 78%, respectively), the ER-resident homo- cysteine-responsive protein HERPUD1 (reduction 74%) as well as pro-apoptotic GADD153/CHOP (reduction 75%), which is induced during ER stress-mediated apoptosis. The quantitatively most significant results from this gene expression analysis fit into the following overall pattern: Genes indicating ER stress and activation of UPR-mediated apoptosis were consistently downregulated (HSPs and other chaperones, CHOP), whereas genes facilitating proteasomal degradation (expression of proteasome activators, subunits, ubiquitin and ubiquitin-like proteins) were upregulated.

HL-60a cells are characterized by reduced protein biosynthesis and failure to accumulate poly- ubiquitinated protein and to trigger ER stress-mediated apoptosis in response to bortezomib Compared to untreated parental cells, adapted cells showed similar (ARH-77a, AMO-1a) or slightly reduced (HL-60a) growth kinetics in the absence of bortezomib. The presence of the therapeutic dose of 20 nM bortezomib did not substantially reduce the growth rate of the adapted cells (Figure 5a). We compared the rate of protein biosynthesis in adapted vs non- adapted cells by metabolic labeling of newly synthesized proteins. Bortezomib-adapted HL-60a cells showed a 75% reduction of the rate of protein biosynthesis compared to parental cells. Activation of the UPR is known to reduce protein biosynthesis. Accordingly, the addition of 20 nM bortezomib led to a 40% decrease in the rate of protein biosynthesis in HL-60 cells, but not in HL-60a cells, indicating that bortezomib did not activate the UPR in HL-60a cells (Figure 5b). Consistent with this interpretation, bortezomib treatment resulted in a dose-dependent increase in poly-Ub proteins in HL-60 cells, but not in HL-60a cells (Figure 5c). Similarly, expression of the UPR-associated chaperone BiP, as well as of markers of

Figure 4 Bortezomib-adapted cells show higher expression of constitutive proteasome subunits, and of the 11S activator. (a) Western blot for proteasome subunits in HL-60 and HL-60a cells: Equal amounts of protein from cells incubated with or without bortezomib (B) were probed with antibodies against subunits of the constitutive 20S proteasome (a4, b1, b2, b5), immunoproteasome (b1i, b2i, b5i), the 19S regulator (Rpn1) and 11S activator (PA28a). Adapted HL-60a were additionally tested after bortezomib washout phases of 3 and 7 days; b-actin served as a loading control. (b) 20S proteasomes were immunoprecipitated from HL-60 and HL-60a cell lysates by an anti- a2-Ab. Lysates (lys) and precipitates (IP) were analyzed by western blot probed with Ab against 20S (a4) and 11S (PA28a) subunits. On the blot, ‘IP’ lanes contain 1/20 of the protein amount used in the ‘lys’ lanes. (c) Western blot for proteasome subunits (b1, b2, b5) and b-actin in ARH-77, AMO-1 and bortezomib-adapted ARH-77a and AMO-1a cells incubated with or without 20 nM bortezomib (B) for 16 h.

Leukemia Bortezomib-adapted cells TRu¨ckrich et al 1103

Figure 5 Bortezomib-adapted HL-60a cells are characterized by a reduced metabolism and lack of metabolic stress upon treatment with bortezomib. (a) HL-60 (upper panel), ARH-77 (middle panel) and AMO-1 (lower panel) cells in their non-adapted (circles) or bortezomib-adapted (triangles) forms were incubated with 20 nM bortezomib (white symbols) for 48 h or left untreated (black symbols). Their metabolic activity was measured in an MTS assay by photometric detection of the product of dehydrogenase-catalyzed MTS turnover during 3 h. (b) Incorporation of 35 S-labeled methionine/cysteine into protein is shown for HL-60 and HL-60a cells left untreated (black bars) or treated with 20 nM bortezomib (gray bars). Bars represent mean results from ﬁve experiments (t-test: *Po0.07, **Po0.05). (c) HL-60 and HL-60a were treated with bortezomib at given concentrations for 24 h and cell lysates were probed in a western blot with antibodies against poly-ubiquitin chains (poly-Ub), ER chaperone BiP, UPR-associated caspase 4 (Casp4, pro-enzyme and activated form), processed PARP-1 and GAPDH (loading control).

UPR-mediated apoptosis (active caspase 4 and cleaved PARP-1), obtained from studies analyzing individual non-myeloma type was not triggered by increasing concentrations of bortezomib in of cell lines adapted to standard laboratory-use proteasome HL-60a cells. Together, these results illustrate that bortezomib inhibitors in vitro. The current study is the ﬁrst to demonstrate treatment does not induce an imbalance between protein that (i) adaptation to bortezomib can indeed be achieved in a biosynthesis and protein destruction in HL-60a cells. This is not myeloma cell line, that (ii) the adaptation to bortezomib induces due to an increased expression and/or activity in compensating similar patterns of changes in the proteasome system in different proteolytic systems like lysosomal proteases, TPPII, or ubiquitin- types of cell lines and that (iii) bortezomib-adapted cells reduce speciﬁc proteases, because we did not observe consistent changes their rate of protein biosynthesis and do not activate the UPR in any of these systems in the bortezomib-adapted cells, and the upon exposure to bortezomib. inhibition of cathepsins or TPPII did not induce the accumulation Adaptation of a myeloma cell line or a plasmocytoid of poly-ubiquitinated protein in HL-60a cells (Supplementary lymphoma cell line towards bortezomib or other proteasome Figures S2 and S3). Hence, changes in the expression, activity and inhibitors has not been reported to date. It has been speculated composition of proteasomes together with a decreased rate of that due to the special biology of the protein biosynthesis and protein biosynthesis are the major features characterizing the quality control machinery in myeloma cells such an adaptation homeostasis of proteins in bortezomib-adapted cells. might not be possible.21 The emergence of bortezomib-resistant myeloma in the clinical setting challenges this assumption. Our data demonstrate that myeloma-type AMO-1 cells and plasmo- Discussion cytoid-type lymphoma cells ARH-77 can be adapted towards high bortezomib concentrations, resulting in a more than We have little insight into the cell biology of resistance against 20-fold increase in the respective IC50 of the drug. These proteasome inhibitors. Most of the information available is bortezomib-adapted myeloma/plasmocytoid lymphoma cell

Leukemia Bortezomib-adapted cells TRu¨ckrich et al 1104 lines proliferated at a similar rate compared to their parental In addition to increased proteasome active subunit expres- lines, independently of the absence or presence of bortezomib. sion, HL-60a showed increased expression of the 11S activator, Affinity labeling of active proteasome subunits using a cell- both on the mRNA and protein levels, whereas no changes in permeable activity-based probe allows the assessment of the the immunoproteasome levels were observed. Although higher active proteasomal subunits in live cells. The use of the levels of the IFN-g-inducible 11S activator without increased fluorescent probe MV-151 enables a quantitative detection of numbers of immunoproteasomes may be counter-intuitive, our the fluorescence signal, which selectively decorates active results show that a higher portion of 11S is also bound to the 20S proteasome subunits, directly from the SDS-PAGE. This method core particle in HL-60a cells. In contrast, Burkitt lymphoma cells eliminates limitations of peptide-based fluorogenic substrates adapted to bortezomib expressed higher levels of constitutive for assessing proteasome activity, because the latter are not 20S and 19S subunits, whereas immunoproteasome and 11S enzyme-specific and require disruption of the cells prior to subunits were downregulated.24 However, Burkitt lymphoma analysis. We therefore here present the first quantitative relies less on proteasome activity due to its exceptionally high measurement of changes in active proteasome subunits in intact activity of the cytosolic protease TPPII, so that different cell bortezomib-adapted cells. Our data illustrate that bortezomib- types may well differ in this respect. adapted cells undergo qualitatively quite similar changes in their Proteasome inhibition by bortezomib triggers the UPR due to active proteasome subunits, albeit with significant quantitative the accumulation of unprocessed, misfolded protein, which differences: they increase b2 activity from 50% to nearly ultimately results in UPR-mediated apoptosis, when sufficient 10-fold, and similarly, activity signals for the b1/b5-type of ER stress is applied. HL-60a cells show a number of features activities were differentially increased in two adapted cell lines. suggesting that bortezomib fails to induce ER stress or that the These increases at least in part result from the increased level of ER stress is at least minimized: transcription rates of proteasome genes and expression of the In a striking difference to the hyperproliferation of bortezo- respective polypeptides, as observed by western blot and gene mib-adapted Burkitt lymphoma cells,24 HL-60a cells markedly expression analysis. The failure to detect a significant increase in reduced their rate of protein biosynthesis, although they the amount of active b1/b5-type of subunits in ARH-77a cells proliferated at only slightly decreased speed. First and foremost, may in part reflect the fact that these two subunits can barely be the combination of reduced protein synthesis and increased distinguished by SDS-PAGE, whereas the respective comparison baseline proteasomal activity is likely to facilitate cellular of proteasome polypeptides by western blot analysis shows an homeostasis in the presence of bortezomib (through a decrease increase in b5 expression, but not in b1 expression. Importantly, in production of defective ribosomal products and residual all adapted cell types retained detectable activity of the targeted proteolytic activity). Also, the reduction of gene expression of b1/b5 active sites in the presence of 20 nM bortezomib, in E3 Ub ligase components observed in HL-60a cells can be contrast to control cells, underscoring the apparent need for integrated into this model: a downmodulation of the Ub residual proteasome activity. conjugation machinery can be predicted to prepare a less To date, different individual cell lines have been adapted substrate for degradation by the proteasome, thereby contribut- to increased bortezomib levels and different changes in ing to the balance between proteasome substrates and protea- proteasome activity have been reported.21–24 Because of some activity in the absence of the inhibitor. We observed less the different types of analysis, assay conditions, as well as mRNA expression of the major HSP in HL-60a cells during differences in generating the individual bortezomib-adapted bortezomib-free culture, indicating the basal levels of cellular cell clones, it is difficult to assess to what extent bortezomib stress and hence the need for chaperones is considerably lower resistance follows a more general and qualitatively reprodu- in HL-60a than in control cells. Therefore, even when cible reaction pattern in different cells. Mutations in the active exogenous ER stress is applied in HL-60a cells, these cells center of the b5 polypeptide have been repeatedly reported, may more easily upregulate chaperones to compensate the ER which in part did, and in part did not lead to increased stress and to avoid UPR-mediated apoptosis. chymotryptic activity with and without overexpression of When the proteasome is inhibited, alternative proteases individual proteasome subunits. The currently most compre- (such as autophagy-mediating lysosomal proteases, or cytosolic hensive report21 analyzes different clones of one myeloid TPPII) can compensate for the lack of proteasome leukemia cell line, which show overexpression of b1, 2 and 5 activity.15,16,25 We did not observe such an increased activity polypeptides, and an increase in active b2, but a decrease in of TPPII using an assay based on a fluorogenic substrate. Also, active b1 and b5 by intracellular affinity labeling. This we found no evidence for an increased activity of key lysosomal discrepancy is explained by a point mutation in the active proteases such as cathepsins Z, B or S that may take over a center of b5, which likely interferes with binding of the probe. part of the proteasome’s function in HL-60a cells. Although A different point mutation in the active center of the b5 we cannot formally exclude that protease activities not polypeptide has been observed in T-cell lymphoma-type Jurkat examined by us could be upregulated, the increase in cells, together with an increase in transcription of the respective proteasome activity together with the lack of effect of a TPPII gene.22,23 Here, we directly compare the adaptive changes in inhibitor or leupeptin, which broadly inhibits cysteine and three different cell lines generated under identical conditions, serine proteases, suggest to us that alternative proteolytic and observe an increased expression and also increased pathways are not key features that mediate the bortezomib- active species of the targeted as well as of the non-targeted adapted phenotype. active proteasome subunits. This argues that the resistance Bortezomib-adapted cells showed resistance towards different towards proteasome inhibitors is characterized by a mostly types of PI. This argues against the use of alternative proteasome uniform reaction pattern consisting of the upregulation of inhibitors to target bortezomib-adapted cells. In contrast to their proteasome transcription, expression and activity, which is PI resistance, they remained sensitive to daunorubicin (as well triggered by proteasome inhibition. The increase in labeled as to further standard cytotoxic drugs, data not shown). Because b1/b5 activity observed in our study argues against mutations daunorubicin is a substrate of multi-drug resistance transporters, in the active center of the b5 polypeptide at least in AMO-1a resistance to bortezomib is unlikely to be mediated by increased and HL-60a cells. drug efflux through these transport systems.

Leukemia Bortezomib-adapted cells TRu¨ckrich et al 1105 In summary, bortezomib-adapted cells can likewise emerge 8 Ardley HC, Robinson PA. E3 ubiquitin ligases. Essays Biochem from myeloma, plasmocytoid lymphoma or myeloid leukemia 2005; 41: 15–30. cells and show similar patterns of changes in their proteasome 9 Amerik AY, Hochstrasser M. Mechanism and function of deubiqui- biology. They combine a number of different features that are tinating enzymes. Biochim Biophys Acta 2004; 1695: 189–207. 10 Ovaa H, Kessler BM, Rolen U, Galardy PJ, Ploegh HL, Masucci likely to make their UPR/proteasome system less vulnerable to MG. Activity-based ubiquitin-specific protease (USP) profiling of bortezomib. Some of these features, such as upregulating or virus-infected and malignant human cells. Proc Natl Acad Sci USA mutating the b5 subunit, are specific to the target of bortezomib, 2004; 101: 2253–2258. and may therefore be overcome by alternative proteasome 11 Meusser B, Hirsch C, Jarosch E, Sommer T. ERAD: the long road to inhibitors. However, other features like a general downregula- destruction. Nat Cell Biol 2005; 7: 766–772. tion of protein biosynthesis and UPR chaperones, as well as the 12 Zhang K, Kaufman RJ. The unfolded protein response: a stress signaling pathway critical for health and disease. Neurology 2006; upregulation of the remaining active proteasome subunits, affect 66: S102–S109. the entire UPR/proteasome system and are not specific for the 13 Hitomi J, Katayama T, Eguchi Y, Kudo T, Taniguchi M, Koyama Y inhibition of b5 by bortezomib. Therefore, in bortezomib- et al. Involvement of caspase-4 in endoplasmic reticulum stress- adapted cells, targeting the proteasome/UPR machinery by induced apoptosis and Ab-induced cell death. J Cell Biol 2004; alternative proteasome inhibitors and/or combination therapies 165: 347–356. of bortezomib with drugs that increase ER stress (such as HSP 14 Obeng EA, Carlson LM, Gutman DM, Harrington Jr WJ, Lee KP, 26 Boise LH. Proteasome inhibitors induce a terminal unfolded inhibitors, HIV protease inhibitors ) or interfering with alter- protein response in multiple myeloma cells. Blood 2006; 107: native proteolytic pathways (cathepsin inhibitors, autophagy 4907–4916. inhibitors, TPPII-inhibitors) appear to be less promising strate- 15 Glas R, Bogyo M, McMaster JS, Gaczynska M, Ploegh HL. gies than the targeting of pathways entirely unrelated to the A proteolytic system that compensates for loss of proteasome proteasome. function. Nature 1998; 392: 618–622. 16 Princiotta MF, Schubert U, Chen W, Bennink JR, Myung J, Crews CM et al. Cells adapted to the proteasome inhibitor 4-hydroxy-5- Acknowledgements iodo-3-nitrophenylacetyl-Leu-Leu-leucinal-vinyl sulfone require enzymatically active proteasomes for continued survival. Proc Natl Acad Sci USA 2001; 98: 513–518. We thank the Proteom-Centrum at the Interfaculty Institute for Cell 17 Turk V, Turk B, Turk D. Lysosomal cysteine proteases: facts and Biology, University of Tu¨bingen (Professor A Nordheim) for use of opportunities. EMBO J 2001; 20: 4629–4633. the fluorescence scanner and Carsten Henneges for help with data 18 Berkers CR, Verdoes M, Lichtman E, Fiebiger E, Kessler BM, analysis. Research was supported by grants from Deutsche Anderson KC et al. Activity probe for in vivo profiling of the Forschungsgemeinschaft (SFB685), the Deutsche Krebshilfe and specificity of proteasome inhibitor bortezomib. Nat Methods 2005; 2: 357–362. the Swiss National Science Foundation (to CD) as well as the 19 Verdoes M, Florea BI, Menendez-Benito V, Maynard CJ, Witte Clinical Trials Unit at the Cantonal Hospital St Gallen (CTU- MD, van der Linden WA et al. A fluorescent broad-spectrum KSSG). proteasome inhibitor for labeling proteasomes in vitro and in vivo. Chem Biol 2006; 13: 1217–1226. 20 Kraus M, Ru¨ckrich T, Reich M, Gogel J, Beck A, Kammer W et al. References Activity patterns of proteasome subunits reflect bortezomib sensitivity of hematologic malignancies and are variable in 1 Richardson PG, Mitsiades C, Hideshima T, Anderson KC. primary human leukemia cells. Leukemia 2007; 21: 84–92. Proteasome inhibition in the treatment of cancer. Cell Cycle 21 Oerlemans R, Franke NE, Assaraf YG, Cloos J, van Zantwijk I, 2005; 4: 290–296. Berkers CR et al. Molecular basis of bortezomib resistance: 2 Stapnes C, Doskeland AP, Hatfield K, Ersvaer E, Ryningen A, proteasome subunit beta5 (PSMB5) gene mutation and over- Lorens JB et al. The proteasome inhibitors bortezomib and PR-171 expression of PSMB5 protein. Blood 2008; 112: 2489–2499. have antiproliferative and proapoptotic effects on primary 22 Lu¨ S, Chen Z, Yang J, Chen L, Gong S, Zhou H et al. human acute myeloid leukaemia cells. Br J Haematol 2007; 136: Overexpression of the PSMB5 gene contributes to bortezomib 814–828. resistance in T-lymphoblastic lymphoma/leukemia cells derived 3 Strauss SJ, Higginbottom K, Juliger S, Maharaj L, Allen P, from Jurkat line. Exp Hematol 2008; 36: 1278–1284. Schenkein D et al. The proteasome inhibitor bortezomib acts 23 Lu¨ S, Yang J, Song X, Gong S, Zhou H, Guo L et al. Point mutation independently of p53 and induces cell death via apoptosis and of the proteasome beta5 subunit gene is an important mechanism mitotic catastrophe in B-cell lymphoma cell lines. Cancer Res of bortezomib resistance in bortezomib-selected variants of Jurkat 2007; 67: 2783–2790. T cell lymphoblastic lymphoma/leukemia line. J Pharmacol Exp 4 Orlowski RZ, Kuhn DJ. Proteasome inhibitors in cancer Ther 2008; 326: 423–431. therapy: lessons from the first decade. Clin Cancer Res 2008; 14: 24 Fuchs D, Berges C, Opelz G, Daniel V, Naujokat C. Increased 1649–1657. expression and altered subunit composition of proteasomes 5 Glickman MH, Ciechanover A. The ubiquitin-proteasome proteo- induced by continuous proteasome inhibition establish apoptosis lytic pathway: destruction for the sake of construction. Physiol Rev resistance and hyperproliferation of Burkitt lymphoma cells. J Cell 2002; 82: 373–428. Biochem 2008; 103: 270–283. 6 Heinemeyer W, Ramos PC, Dohmen RJ. The ultimate nanoscale 25 Wang EW, Kessler BM, Borodovsky A, Cravatt BF, Bogyo M, mincer: assembly, structure and active sites of the 20S proteasome Ploegh HL et al. Integration of the ubiquitin-proteasome pathway core. Cell Mol Life Sci 2004; 61: 1562–1578. with a cytosolic oligopeptidase activity. Proc Natl Acad Sci USA 7 Bogyo M, McMaster JS, Gaczynska M, Tortorella D, Goldberg AL, 2000; 97: 9990–9995. Ploegh H. Covalent modification of the active site threonine of 26 Kraus M, Malenke E, Gogel J, Mu¨ller H, Ru¨ckrich T, Overkleeft H proteasomal beta subunits and the Escherichia coli homolog HslV et al. Ritonavir induces ER-stress and sensitizes sarcoma cells by a new class of inhibitors. Proc Natl Acad Sci USA 1997; 94: towards bortezomib-induced apoptosis. Mol Cancer Ther 2008; 7: 6629–6634. 1940–1948.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia