Hyperthermia classic commentary: Activation of natural killer (NK) cells by heat shock protein 70, Gabriele Multhoff, International Journal of Hyperthermia, 2002;18:576–585

Hsp70 activated NK cells in cancer therapy

Dear Sir

Identification of molecular markers that can be utilized as targets for cancer immunotherapy is a critical goal for the development of therapeutics and also for improved diagnostic applications. Ideally, a biomarker should be selectively expressed on tumor but not on normal cells, or at least, should be expressed at a much lower level on normal cells. A large scale screening program on more than 900 tumor biopsies derived from patients with different tumor entities including gastrointestinal, urogenital and bronchial tumors, as well as hematological diseases, and their corresponding normal tissues revealed that a membrane-bound form of Hsp70 could fulfill this criterion. However, an Hsp70 membrane-positive phenotype has also been associated with a higher metastatic potential and an unfavorable prognosis in malignant melanoma 1, lower rectal cancer, squamous cell carcinoma of the lung 2 and in acute myeloid leukemia 3, 4. These latter findings suggest that elevated Hsp70 expression at the plasma membrane may predict a more severe cancer pathology. In line with these findings, the membrane density of Hsp70 has been found to be enhanced in cells within metastases when compared to those in primary tumors using a xenograft tumor mouse model 5.

How might plasma membrane-associated Hsp70 be useful as a therapeutic target? In 1995, we were the first to demonstrate that membrane-bound Hsp70 provides a tumor-specific target structure for lytic natural killer (NK) cells of the innate immune system 6, 7. Moreover, the expression of Hsp70 on the cell surface of tumors plays an immunomodulatory role by increasing tumor cell lysis by NK cells when in the presence of pro-inflammatory cytokines 6, 8, 9. Furthermore, a correlation of the cytolytic activity of NK cells with the Hsp70 density on tumor cells also has been determined 10, 11. Unstimulated NK cells derived from patients did not show any killing activity towards autologous tumors. This might be due to the fact that these NK cells might be in a resting stage or that the tumor has evolved tumor escape mechanisms.

In this commentary we provide an update on the role of Hsp70 for the innate immune system. We will discuss research regarding the anchorage 12 and the export of Hsp70 from tumor cells 13, as well as the mechanism of kill of Hsp70 membrane-positive tumors 14. Furthermore, we will summarize the results of a phase I clinical trial which has been performed in patients with metastasized colorectal and non-small lung cell carcinoma (NSCLC), applying ex vivo Hsp70 peptide-activated, autologous NK cells. Finally, we will report on a targeted NK cell-based adjuvant immunotherapy for the treatment of patients with NSCLC after standard radiochemotherapy which will be initiated this year as a ‘proof of principle’ clinical phase II randomized trial.

Anchorage of Hsp70 in the plasma membrane of tumor cells

As mentioned before, we found Hsp70 to be selectively expressed on the cell surface of human tumors, but not on normal cells. However, it remains unclear how this tumor-specificity is maintained. Hsp70, the major stress-inducible member of the HSP70 family, is a cytosolic protein which lacks a classical transmembrane domain. Therefore, it was assumed that Hsp70 is presented on the plasma membrane via linkage to a cell surface receptor. However, neither high salt conditions nor changes in the pH altered the Hsp70 membrane expression pattern (unpublished observations). Therefore it appears highly unlikely that Hsp70 is bound to a receptor on the plasma membrane. With respect to previously published data by Hightower and Guidon 15 and the group of Antonio DeMaio 16–18 showing a physical interaction of Hsp70 with fatty acid components, in collaboration with Gerd Schmitz, Claudia Steinem and Cliff Lingwood, we performed lipidomic analyses of isolated plasma membranes and cholesterol rich microdomains (CRMs) from tumor cell lines with differential Hsp70 membrane expression. It became apparent that the lipid composition of CRMs of Hsp70 membrane-positive and–negative tumor cells differed in the content of globoyltriaosylceramide (Gb3) suggesting the hypothesis that there is an interaction between Hsp70 and the glycosphingolipid Gb3. This hypothesis could be further supported by vesicle aggregation assays using artificial unilamellar raft-like liposomes which contain Gb3. Since Gb3 is frequently over-expressed in the plasma membrane of human colorectal tumors and metastases derived thereof, this may provide a mechanism by which Hsp70 is expressed on the tumor cell surface 19. The fact that apart from germline B cells, Gb3 is not found in the plasma membrane of normal cells and is over-expressed in CRMs of tumors might explain the tumor-specificity of the Hsp70 cell surface expression. Ongoing research is aiming to further characterize the molecular nature of the Hsp70-lipid interaction.

Export of Hsp70 from tumor cells

It is known that HSPs are found not only in the cytosol, on the plasma membrane but are also found in the extracellular space 20, 21. Members of the Hsp70 family in cooperation with other chaperones are major players in transport mechanisms across lipid membranes 22. Therefore, it is conceivable that Hsp70 might fulfill shuttle functions from the cytosol into the extracellular milieu. Although the mechanism of export remains elusive, release of Hsp70 from viable cells with intact plasma membrane has been documented 23, 24. Blocking experiments indicated that Hsp70 secretion is mediated through a non-classical ER-Golgi pathway since neither brefeldin A nor monensin affect Hsp70 export. Since there is very little spontaneous release of free Hsp70 protein from viable cells, it was speculated that Hsp70 might be released in vesicles. Indeed, following ultracentrifugation of supernatants of tumor cells, detergent-soluble liposomes could be isolated which present Hsp70 on their surface 25. An enrichment of the small GTPase Rab-4 within these vesicles supports the idea of an intracellular transport route from the early endosome to the plasma membrane and finally into the extracellular space. A recent study has documented the presence of Gb3-containing lipid rafts in the vesicular membranes 19. This finding might suggest a mechanism by which Hsp70 can be anchored within the vesicular membranes.

Mechanism of tumor cell kill

Cell surface-bound Hsp70 renders tumor cells more sensitive to the cytolytic attack mediated by NK cells which have been stimulated with a 14-mer amino acid Hsp70 sequence TKD (aa 450–463) plus low dose IL-2. Flow cytometry analysis of these stimulated NK cells compared to unstimulated NK cells revealed an up-regulation of markers including CD56, CD94, NKG2D, natural cytotoxicity receptors (NCRs) and activation markers CD69 and CD25. In contrast, killer cell immunoglobulin-like receptors (KIRs) appeared to be unaffected and the low affinity receptor CD16 was found to be down-regulated by TKD-stimulation 26. Due to their cytosolic motifs KIRs can either inhibit (long ITIM) or activate (short ITAM) the cytolytic activity of NK cells. Overall, the phenotype of the NK cells after TKD/IL-2 stimulation resembled the CD16dim/CD56 bright phenotype, which are considered as cytokine producers such as IFN-gamma upon stimulation with interleukins 27. This NK cell type has been found to possess low amounts of perforin and thus is expected to have a low cytotoxic activity. However, we have shown that our TKD/IL-2 activated NK cells are highly efficient in killing Hsp70 membrane-positive tumor cells. With regard to these findings we studied the mechanism of kill of Hsp70 membrane-positive tumors in more detail. By affinity chromatography, it was shown that the apoptosis-inducing enzyme granzyme B interacts with full length Hsp70 as well as with the Hsp70 peptide TKD. An incubation of tumor cells with differential Hsp70 membrane expression pattern with purified granzyme B in the absence of perforin indicated that only Hsp70 membrane-positive tumors were able to internalize granzyme B into the cytosol. Following uptake and activation of the caspase pathway these tumor cells underwent apoptotic cell death 14. From these results, we propose that Hsp70 on the plasma membrane of tumor cells might serve as an entry port for granzyme B, even in the absence of perforin, and thus might enable apoptotic cell death caused by CD16dim/CD56bright NK cells.

More recent data have shown that Hsp70 membrane-negative tumors which do express NKG2D ligands such as MICA/B on their cell surface can also be killed by TKD/IL-2 activated NK cells 28. The best lysis was observed in tumors that express both Hsp70 and NKG2D ligands on their cell surface 28. It can be speculated that the synergistic effects of two immunological danger signals might result in an enhanced cytotoxic activity of NK cells against tumors. The role of standard radiochemotherapy 29 and hyperthermia as potential tools to enhance the expression of danger signals on the cell surface of tumor cells is presently under investigation 30. Indeed, recent studies show the potential for hyperthermia to increase the expression of NKG2D ligands on the surface of tumor cells 31.

Clinical application of Hsp70 peptide activated NK cells

An Hsp70 membrane-positive phenotype was associated with poor clinical outcome in a large number of tumor entities 2. We have shown in tumor mouse models that TKD/IL-2 activated NK cells have the capacity to kill Hsp70 membrane-positive tumors and to enhance overall survival of tumor-bearing animals. As mentioned earlier, unstimulated NK cells of patients are unable to kill Hsp70 membrane-positive tumor cells. These promising pre-clinical results encouraged us to initiate a clinical phase I trial in order to study feasibility, safety, and tolerability of ex vivo Hsp70-peptide TKD plus low dose IL-2 activated NK cells 32. Patients with multiple metastasized colorectal and non-small lung cell carcinoma (NSCLC) that were refractory to standard radiochemotherapy were enrolled into the trial. Up to six repeated reinfusions of complete autologous leukapheresis products that had been stimulated with TKD/IL-2 was found to be feasible, safe and well tolerated by all patients. Although the leukapheresis and reinfusions were performed every two weeks, the total number of lymphocytes and the composition of lymphocyte subpopulations in the leukapheresis product remained stable. A key finding of the trial was that the cytolytic of NK cells derived from pre-treated patients following radiochemotherapy could be stimulated by in vitro incubation with TKD/IL-2. Patients who received 4 reinfusions of TKD/IL-2 activated effector cells showed an enhanced in vivo cytolytic activity corresponding to an enhanced cell surface density of activating NK cell surface markers. Moreover, 2 of 5 patients receiving more than 4 treatment cycles showed clinical responses: a stable disease and a mixed response 32.

Based on these promising findings a 'proof of principle’ randomized phase II clinical study is underway in NSCLC patients. Lung cancer is the second most frequent tumor type in western societies and the leading cause of cancer death. Previously, we have shown that radiochemotherapy efficiently enhances Hsp70 membrane expression on tumor cells (unpublished observation). Therefore, we now want to adoptively transfer TKD/IL-2-activated, autologous NK cells into NSCLC patients stage IIIA and IIIB following successful radiochemotherapy. An Hsp70 membrane positive tumor phenotype is a key inclusion criteria for enrolling patients into the study. Recent data have shown that the DNA damage response induced by ionizing radiation enhances the expression of NKG2D ligands 33, 34 and thus this could be beneficial for clinical outcome. Another approach to increase radiosensitization is the use of Hsp90 inhibitors which are known to enhance the Hsp70 synthesis 34. A major goal of the study will be is to prolong survival of patients with a better quality of life. An additional goal in this study will be to measure the phenotype and the cytolytic capacity of immune effector cells in the patients before, during and after immunotherapy. The follow up of the patients will be 18 months from start of the NK cell infusions.

In summary, our paper republished here in the International Journal of Hyperthermia was the start of a new direction in our research which has taken us from basic laboratory observations to the implementation of new clinical trials. Overall, we expect that membrane-bound Hsp70 might provide a tumor-specific target structure for future cell-based and/or antibody-based therapeutic approaches.

Acknowledgements

The author would like to thank Armin Bareiss for technical assistance.

Grant support: Deutsche Forschungsgemeinschaft (DFG, MU 1238/7-2), European Union EU-STEMDIAGNOSTICS (LSHB CT 2007 037703), EU-CARDIORISK (FP7 2007–2001, #211403), BMBF–MOBITUM (Watching Tumor Biology by Molecular Imaging–MoBiMed), BMBF–Innovative Therapies (funding in progress) and multimmune GmbH.

Declaration of interest: The author reports no conflicts of interest. The author alone is responsible for the content and writing of the paper.