BACKGROUND:Protein metabolism is an innovative potential therapeutic target for AML. Proteotoxic stress (PS) sensitizes malignant cells for proteasome inhibitor treatment. Some HIV protease inhibitors (HIV-PI) induce PS and may therefore be combined with proteasome inhibitors to achieve PS-targeted therapy of AML.METHODS:We investigated the effects of all nine approved HIV-PI alone and in combination with proteasome inhibitors on AML cell lines and primary cells in vitro.RESULTS:Ritonavir induced cytotoxicity and PS at clinically achievable concentrations, and induced synergistic PS-triggered apoptosis with bortezomib. Saquinavir, nelfinavir and lopinavir were likewise cytotoxic against primary AML cells, triggered PS-induced apoptosis, inhibited AKT-phosphorylation and showed synergistic cytotoxicity with bortezomib and carfilzomib at low micromolar concentrations. Exclusively nelfinavir inhibited intracellular proteasome activity, including the β2 proteasome activity that is not targeted by bortezomib/carfilzomib.CONCLUSIONS:Of the nine currently approved HIV-PI, ritonavir, saquinavir, nelfinavir and lopinavir can sensitize AML primary cells for proteasome inhibitor treatment at low micromolar concentrations and may therefore be tested clinically toward a proteotoxic stress targeted therapy of AML.

The interaction of developing thymocytes with peptide-MHC complexes on thymic antigen presenting cells (APC) is crucial for T cell development, both for positive selection of "useful" thymocytes as well as negative selection of autoreactive thymocytes to prevent autoimmunity. The peptides presented on MHC II molecules are generated by lysosomal proteases such as the cathepsins. At the same time, lysosomal proteases will also destroy other potential T cell epitopes from self-antigens. This will lead to a lack of presentation on negatively selecting thymic antigen presenting cells and consequently, escape of autoreactive T cells recognizing these epitopes. In order to understand the processes that govern generation or destruction of self-epitopes in thymic APC, we studied the antigen processing machinery and epitope processing in the human thymus. We find that each type of thymic APC expresses a different signature of lysosomal proteases, providing indirect evidence that positive and negative selection of CD4+ T cells might occur on different sets of peptides, in analogy to what has been proposed for CD8+ T cells. We also find that myeloid dendritic cells (DC) are more efficient in processing autoantigen than plasmacytoid DC. In addition, we observed that cathepsin S plays a central role in processing of the autoantigens myelin basic protein and proinsulin in thymic dendritic cells. Cathepsin S destroyed a number of known T cell epitopes, which would be expected to result in lack of presentation and consequently, escape of autoreactive T cells. Cathepsin S therefore appears to be an important factor that influences selection of autoreactive T cells.

Imatinib (IM) has been described to modulate the function of dendritic cells and T lymphocytes and to affect the expression of antigen in CML cells. In our study, we investigated the effect of the tyrosine kinase inhibitors IM and nilotinib (NI) on antigen presentation and processing by analyzing the proteasomal activity in CML cell lines and patient samples. We used a biotinylated active site-directed probe, which covalently binds to the proteasomally active beta-subunits in an activity-dependent fashion. Additionally, we analyzed the cleavage and processing of HLA-A3/11- and HLA-B8-binding peptides derived from BCR-ABL by IM- or NI-treated isolated 20S immunoproteasomes using mass spectrometry. We found that IM treatment leads to a reduction in MHC-class I expression which is in line with the inhibition of proteasomal activity. This process is independent of BCR-ABL or apoptosis induction. In vitro digestion experiments using purified proteasomes showed that generation of epitope-precursor peptides was significantly altered in the presence of NI and IM. Treatment of the immunoproteasome with these compounds resulted in an almost complete reduction in the generation of long precursor peptides for the HLA-A3/A11 and −B8 epitopes while processing of the short peptide sequences increased. Treatment of isolated 20S proteasomes with serine-/threonine- and tyrosine-specific phosphatases induced a significant downregulation of the proteasomal activity further indicating that phosphorylation of the proteasome regulates its function and antigen processing. Our results demonstrate that IM and NI can affect the immunogenicity of malignant cells by modulating proteasomal degradation and the repertoire of processed T cell epitopes.

The biosynthesis of immunoglobulin leads to constitutive endoplasmic reticulum (ER) stress in myeloma cells, which activates the unfolded protein response (UPR). The UPR promotes protein folding by chaperones and increases proteasomal degradation of misfolded protein. Excessive ER stress induces apoptosis and represents a molecular basis for the bortezomib sensitivity of myeloma. Most solid malignancies such as sarcoma, by contrast, are poorly bortezomib sensitive and display low levels of ER stress. We hypothesized that pharmacologic induction of ER stress might sensitize malignancies to bortezomib treatment. We show that the HIV protease inhibitor ritonavir induces ER stress in bortezomib-resistant sarcoma cells. Ritonavir triggered the UPR, decreased the degradation of newly synthesized protein, but did not directly inhibit proteasomal active sites in the therapeutic dose range in contrast to bortezomib. Whereas neither bortezomib nor ritonavir monotherapy translated into significant apoptosis at therapeutic drug levels, the combination strongly increased the level of ER stress and activated PERK, IRE1, and ATF6, synergistically induced CHOP, JNK, caspase-4, and caspase-9, and resulted in >90% apoptosis. In summary, ritonavir increases the level of ER stress induced by bortezomib, which sensitizes bortezomib-resistant cells to bortezomib-induced apoptosis. Ritonavir may therefore be tested clinically to improve the sensitivity of solid malignancies toward bortezomib treatment.

Imatinib mesylate (Gleevec) is a specific tyrosine kinase inhibitor, which inhibits phosphorylation of downstream proteins involved in BCR-ABL signal transduction. In the treatment of CML it has become indispensable and shows few side effects. Nevertheless, imatinib has been described to influence the function and differentiation of APCs, inhibit the effector function of T lymphocytes and decrease the immunogenicity of CML cells. In the present study, we analyzed the possible effects of imatinib on proteasomal activity in CML cells. Proteasomal activity was determined using a biotinylated active site-directed probe which, depending on their activity, covalently binds and labels proteasomal subunits beta-1, beta-2 and beta-5. We show that Imatinib treatment leads to a decreased activity of the proteasome in BCR-ABL positive cells. Imatinib sensitive K-562 cells as well as HLA-A2 transfected K-562 cells displayed a profound downregulation of proteasome activity at sub-therapeutic concentrations of imatinib. This effect was not due to downregulation of the protein expression of proteasome subunits or the induction of apoptosis. In imatinib resistant K-562R cells a higher concentration of imatinib was required for a detectable inhibition of proteasomal activity corresponding to a higher expression of BCR-ABL in these cells. Interestingly, the use of specific siRNA against the BCR-ABL fusion site b3a2 to downregulate the expression of the protein in K-562 and K-562R cells did not influence the activity of the proteasome indicating that this effect is independent of the expression and activity of BCR-ABL. In line with these results transfection of BaF/3 cells with BCR-ABL had no effect on proteasomal activity. To further analyze the mechanisms involved in proteasome inhibition by imatinib we performed activity assays with isolated 20S and 26S proteasomes. Incubation of both proteasome subtypes with imatinib resulted in a reduced acitivity in a concentration dependent manner as shown by reduced proteolytic cleavage of fluorogenic substrates. As the reduction in activity was seen in the ATP-independent 20S proteasome as well as the ATP-dependent 26S proteasome, a competition of imatinib with ATP similar to the mechanism of BCR-ABL inhibition is not responsible for this effect. In order to expand these studies, we additionally examined HLA class I expression on the cell surface to determine the influence of imatinib and subsequent proteasome inhibition on antigen presentation. Correlating with the inhibition of proteasome activity, HLA class I expression was strongly reduced on HLA-A2 transfected K-562 cells after imatinib treatment. Taken together these findings show that imatinib directly affects proteasomal activity in vitro. This effect is more pronounced in BCR-ABL positive cells, which might be due to BCR-ABL mediated deregulation of cellular pathways, but seems to be, however, independent of BCR-ABL.

Resistance towards proteasome inhibition by Bortezomib (Velcade®) represents a challenge for myeloma therapy. Its biology has not yet been characterized in detail. We have demonstrated that Bortezomib-sensitive malignant haematopoetic cells can acquire secondary resistance to Bortezomib in vitro. We here present the first analysis of proteasome biology and activity, alternative proteolytic pathways, ubiquitin-specific proteases (USP) and the ER stress response (unfolded protein response, UPR) upstream of the proteasome, as well as in vitro cytotoxicity of conventional cytotoxic drugs, alternative proteasome inhibitors and agents that target the UPR in Bortezomib-resistant (BR) cells, compared to wild type (WT) controls. BR cells had higher activities of all subunits of the constitutive and the immunoproteasome, as deferred from turnover of fluorogenic substrates as well as affinity-labelling of active proteasome subunits in intact cells. This was mirrored by increased levels of proteasomal β1 and β2, but especially β5 polypeptides, implicating a homeostatic system that senses and corrects low proteasome activity in cells chronically exposed to Bortezomib. While the vinylsulfone-type proteasome inhibitor NLVS abrogated detectable proteasome activity in both BR and WT cells, Bortezomib at therapeutic concentrations eliminated proteasomal β1 and β5-type activity only in WT cells, while BR cells retained residual activity. These changes in proteasome biology appear to be the molecular hallmark of required Bortezomib resistance, since no changes were observed between WT and BR cells in alternative cytosolic or lysosomal proteolytic pathways, UPR activity as well as the gross activity pattern of USP. As expected, this translated into sensitivity against cytotoxic drugs in vitro: BR cells were less sensitive towards alternative proteasome inhibitors. However, while the IC50 for pan-proteasome inhibitors was only roughly doubled in BR cells, it was nearly tenfold elevated for the β5-preferring vinylsulfone inhibitor NLVS. By contrast, sensitivity towards anthracyclines or cytotoxicity induced by ER stressors as well as the synergy between proteasome inhibitors and UPR-activators remained unaffected in BR cells. Based on our data, proteasome inhibitors with activity profiles different from that of Bortezomib, alone or in combination with induction of the UPR, may represent an appropriate concept to overcome secondary Bortezomib resistance.

Cathepsin S (CatS) is a lysosomal cysteine protease of the papain family, the members of which possess relatively broad substrate specificities. It has distinct roles in major histocompatibility complex (MHC) class II-associated peptide loading and in antigen processing in both the MHC class I and class II pathways. It may therefore represent a target for interference with antigen presentation, which could be of value in the therapy of (auto) immune diseases. To obtain more detailed information on the specificity of CatS, we mapped its cleavage site preferences at subsites S3-S1' by in vitro processing of a peptide library. Only five amino acid residues at the substrate's P2 position allowed for cleavage by CatS under time-limited conditions. Preferences for groups of amino acid residues were also observed at positions P3, P1 and P1'. Based on these results, we developed highly CatS-sensitive peptides. After processing of MHC class II-associated invariant chain (Ii), a natural protein substrate of CatS, we identified CatS cleavage sites in Ii of which a majority matched the amino acid residue preference data obtained with peptides. These observed cleavage sites in Ii might be of relevance for its in vivo processing by CatS.

Upregulation of alternative proteolytic pathways characterizes malignant cells that overcome proteasome inhibition in vitro. The proteasome inhibitor Bortezomib (Velcade®), which selectively targets only one of the three active subunits of the proteasome, has shown limited activity in AML. Ritonavir (Norvir®) is an aspartate protease inhibitor used in intensive HIV-therapy, where therapeutic levels of 5–20μM are reached with an oral dose of 1200 mg/d. A cytotoxic effect of Ritonavir against malignant cells due to proteasome inhibition has been suggested (Gaedicke et al., Cancer Research 62, December 1, 2002). We have here tested the effect of Ritonavir on AML cells, both as single agent and in combination with Bortezomib. Ritonavir induced cytotoxic death in AML cell lines and primary AML blasts with an IC50 of 30–40 μM in vitro. The combination of Ritonavir and Bortezomib was synergistic in vitro, i.e. subtoxic concentrations of Ritonavir at 10 μM combined with subtoxic Bortezomib 5–10 nM induced robust cytotoxicity in AML cell lines and freshly isolated primary AML blasts. Using a novel chemical probe that for the first time allows to visualize the individual activity of proteasomal subunits in intact AML blasts, we show that Velcade selectively abrogates β5 proteasomal activity at 20 nM in AML cells, as expected. Ritonavir, by contrast, had no effect on active proteasomal subunits up to 50 μM. Thus, the synergistic effect of Ritonavir with Bortezomib on AML cells is not due to inhibition of the same proteasomal target by both drugs, but more likely mediated by blocking alternative proteolytic pathways. One individual patient aged 72 years with an early relaps of AML was treated with Ritonavir 400–600 mg/d p.o.. During treatment, the absolute leukocyte count dropped from 24000/μl to 8000/μl while the ANC raised from 185/μl to 1530/μl. Ritonavir was withdrawn due to diarrhoea and abdominal cramps, leading to a sharp increase in peripheral blood blasts and leukocytes. Retreatment with Ritonavir at a reduced dose of 200 mg/d combined with Velcade 1mg/sqm was tolerated and stabilized leukocyte counts for a short period of time. We conclude that Ritonavir has activity against chemotherapy-refractory AML in vitro and in vivo. The combination of Velcade and Ritonavir might allow to synergistically target the proteolytic machinery of AML blasts with tolerable toxicity.