Proteasome subunit expression analysis and chemosensitivity in relapsed paediatric acute leukaemia patients receiving bortezomib-containing chemotherapy

The survival of paediatric patients with leukaemia has greatly improved in recent decades. However, 10–30 % of patients still relapse, and outcome after relapse remains poor, with remission rates as low as 10–40 % for poor prognostic subtypes [1, 2]. Hence, there is a need for improvement in therapeutic options for these patients. Based on the proven clinical activity in multiple myeloma (MM) [3] and mantle cell lymphoma [4], as well as encouraging preclinical data from our lab [5] and the Paediatric Preclinical Testing Program [6], the proteasome inhibitor (PI) bortezomib is currently undergoing clinical evaluation in paediatric leukaemia [7, 8].

Proteasomes are large, multi-subunit complexes located in both the cytosol and nucleus and are responsible for the degradation of the majority of intracellular proteins. In addition to constitutive proteasomes (β1 + β2 + β5), the immunoproteasome (β1i + β2i + β5i) represents a proteasome variant that is predominantly expressed in haematopoietic cells [9‐11]. The immunoproteasomes produce peptides similar to constitutive proteasomes. Although the exact role of the immunoproteasome has yet to be defined, the peptide fragments produced by the immunoproteasome are thought to play a role in antigen presentation to MHC class I antigens (reviewed in [12]). Disruption of proteasome activity results in rapid accumulation of toxic regulatory proteins in the cell that results in endoplasmic reticulum (ER) stress and activates the unfolded protein response (UPR), which blocks protein translation and initiates alternative degradation pathways in stress situations [13]. Another mechanism of action of bortezomib is thought to inhibit NF-kB chemotherapy-induced pro-survival pathways in leukaemic blasts by blocking the proteasomal degradation of NF-kB’s natural inhibitor, IkB [14].

Although bortezomib displays only modest single-agent activity in children [15] and adults [16‐18], it has additive or synergistic interactions with cytotoxic agents commonly used to treat acute leukaemia including glucocorticoids, vincristine [5] and anthracyclines [19]. Phase 1/2 studies combining bortezomib with standard induction/consolidation regimens have shown promising clinical activity in both adult AML [20‐22] and paediatric ALL [8, 23]. However, bortezomib did not improve complete remission (CR) rate or overall survival in a paediatric AML cohort [7]. Thus, the clinical impact of PI therapy in combination with cytotoxic chemotherapy has not yet been fully assessed.

Despite the encouraging results of bortezomib in early-phase clinical trials, primary or acquired resistance to bortezomib may limit its efficacy [24, 25] and thus identifying those patients who respond to bortezomib-containing therapy is of great clinical relevance. Recently, we reported that higher ratios of immunoproteasome to constitutive proteasome protein expression in paediatric ALL and AML leukaemia cells at diagnosis predicted ex vivo sensitivity to bortezomib and other PIs [10]. The importance of subunit assembly for PI sensitivity was further investigated in leukaemia cell lines, revealing that interferon-γ-induced upregulation of immunoproteasome subunit expression and concomitant downregulation of constitutive subunit expression, markedly sensitizing bortezomib-resistant cells to PIs, with the β5i being the most important subunit involved in this process [26].

Here, we explored whether ratios of immunoproteasome to constitutive proteasome protein expression correlate with bortezomib response in paediatric leukaemia patients treated with bortezomib-containing re-induction chemotherapy. Specifically, paediatric patients with acute leukaemia in first relapse were enrolled in two Children’s Oncology Group (COG) phase 2 trials of bortezomib combined with re-induction chemotherapy for paediatric ALL (AALL07P1) or paediatric AML (AAML07P1). In addition to (immuno) proteasome profiling, we also examined leukaemia cells for bortezomib-induced alterations in NF-kB activity.

Recently, on the basis of an ex vivo study with acute leukaemia clinical specimen at diagnosis, we provided evidence that ratios of immunoproteasome subunits to constitutive subunits correlate with sensitivity of acute leukaemia cells to PIs [10]. The current research was set up to validate this in a distinct, prospectively collected sample set from patients enrolled on two COG clinical trials, which is the first to study potential biomarkers that may predict clinical sensitivity to bortezomib-containing treatment regimens.

Although approved and efficacious for treatment of MM, bortezomib treatment does not lead to response in all MM patients due to emergence of drug resistance [24]. Conceivably, resistance phenomena could also limit the clinical utility of this agent in relapsed paediatric leukaemia. Hence, parameters that could indicate the likelihood of clinical response to PI therapy would provide a mechanism to determine which patients are likely to benefit from the addition of PI therapy to their re-induction regimen. Several mechanisms have been reported to underlie bortezomib resistance [29, 30]. Bortezomib-resistant cell lines are frequently characterized by mutations in the PSMB5 gene encoding the β5 subunit [27, 31]. To date, however, no PSMB5 mutations have been found in patients clinically resistant to bortezomib [32‐34]. Also in the current study, no PSMB5-associated mutations in exon 2 of the gene were identified in either end of induction (n = 15) or relapse (n = 3) samples (data not shown). This implies that, in acute leukaemia, β5 mutations make only a minor contribution to bortezomib resistance and that other common resistance mechanisms, such as overexpression of β5, play a more important role [27, 29, 31, 35]. Confirming our earlier ex vivo work in newly diagnosed patient samples [10], the current study shows that constitutive (β5 and β1) proteasome subunit expression was significantly lower in patients with relapsed ALL vs. AML, whereas β1i and β5i immunoproteasome subunit expression was higher (though it did not reach significance). These results were recently confirmed by de Bruin et al. [36], who showed by labelling analysis of primary patient cells that B cell ALL patients have higher immune/constitutive subunit expression ratios than AML cells and T-ALL cells. Given the significance of correlation between the subunit expression and catalytic activity (Fig. 3a), the subunit activity-based probes may prove useful for examining clinical specimen. Others have also reported that bortezomib sensitivity relates to proteasome expression. In particular, increased PSMB5 mRNA expression was found in a myeloma patient who subsequently developed bortezomib resistance [33]. Moreover, bortezomib-sensitive hematologic cell lines harboured higher immunoproteasome expression levels compared to relatively bortezomib-resistant solid-tumour cell lines [37]. Lastly, a higher β2/β1 + β5 activity ratio correlated with higher bortezomib sensitivity in a cell line panel of hematologic malignancies [38]. In a recent study, we showed that interferon-γ-induced upregulation of immunoproteasome expression and concurrent constitutive proteasome subunit downregulation in bortezomib-resistant hematologic tumour cell lines resulted in the sensitization for PI treatment [26]. Consistently, knockdown of constitutive β5 in the AML cell line THP1 resulted in increased sensitivity to bortezomib [31].

NF-kB activity was evaluated by ELISA in 26 pre-B ALL, 10 T-ALL and 12 AML samples after the first dose of bortezomib. NF-kB is constitutively active in the majority of ALL patients [23, 39] and AML patients [40]. Though a significant decline in NF-kB activity was observed in pre-B ALL patients 24 h after the first bortezomib dose, this decline in NF-kB activity did not correlate with bortezomib response. Of note, Magrangeas et al. [41] found that low pre-treatment levels of NF-kB were associated with a higher response rate to bortezomib-based induction in newly diagnosed MM, but we were unable to find a similar correlation between pre-treatment NF-kB activity and bortezomib. Overall, in our current study, changes in NF-kB activity during treatment with bortezomib-containing chemotherapy were not associated with achievement of post-induction CR, and neither pre-treatment NF-kB activity nor decreases in NF-kB activity after 24 h correlated with clinical response.

This current cohort provided the first set of samples collected from patients who received bortezomib treatment and could serve as a potential validation cohort for our former test cohort [10]. Since the design of this add-on study was not yet optimized for the proteasome-based assays, there are some limitations that have to be considered. First, although only samples that contained >20 % of blasts were used, there was a large variation in blast percentage between samples (mean 72 % ± 23 %). To address the impact of normal cells, we measured proteasome subunit expression levels in normal PBMCs and found that these had a relative minor contribution on proteasome subunit expression in leukaemic cells and thus did not largely influence our analyses. Notwithstanding this fact, purification of lymphoblast/myeloblast purification is recommended in future follow-up studies. Another important issue is the fact that patients in this study were treated with combination therapy. Hence, any outcome variable is a cumulative effect of all drugs, which can confound the relation between our measurements and bortezomib sensitivity.

In summary, cells of relapsed ALL patients had significantly higher β5i/β5 and β1i/β1 proteasome subunit protein expression ratios, and lower β5 and β1 constitutive subunit expression compared to cells of relapsed AML patients. Interestingly, AML patients who achieved CR after bortezomib-containing therapy had higher pre-treatment immunoproteasome/constitutive proteasome expression ratios compared to patients who did not achieve CR. Further studies using purified blast populations are being conducted to confirm that higher ratios of immuno-/constitutive proteasome in pre-treatment ALL and AML cells are associated with an initial clinical response to bortezomib-containing re-induction treatment.