EGFR-Targeted Granzyme B Expressed in NK Cells Enhances Natural Cytotoxicity and Mediates Specific Killing of Tumor Cells

Natural killer (NK) cells are highly specialized effectors of the innate immune system that hold promise for adoptive cancer immunotherapy. Their cell killing activity is primarily mediated by the pro-apoptotic serine protease granzyme B (GrB), which enters targets cells with the help of the pore-forming protein perforin. We investigated expression of a chimeric GrB fusion protein in NK cells as a means to augment their antitumoral activity. For selective targeting to tumor cells, we fused the epidermal growth factor receptor (EGFR) peptide ligand transforming growth factor α (TGFα) to human pre-pro-GrB. Established human NKL natural killer cells transduced with a lentiviral vector expressed this GrB-TGFα (GrB-T) molecule in amounts comparable to endogenous wildtype GrB. Activation of the genetically modified NK cells by cognate target cells resulted in the release of GrB-T together with endogenous granzymes and perforin, which augmented the effector cells' natural cytotoxicity against NK-sensitive tumor cells. Likewise, GrB-T was released into the extracellular space upon induction of degranulation with PMA and ionomycin. Secreted GrB-T fusion protein displayed specific binding to EGFR-overexpressing tumor cells, enzymatic activity, and selective target cell killing in the presence of an endosomolytic activity. Our data demonstrate that ectopic expression of a targeted GrB fusion protein in NK cells is feasible and can enhance antitumoral activity of the effector cells.


Introduction
Natural killer (NK) cells are highly specialized effectors of the innate immune system. They play an important role in the defense against viral infection and the elimination of neoplastic cells [1]. Natural cytotoxicity of NK cells can be triggered rapidly upon appropriate stimulation, and is regulated by a complex balance of signals from germline-encoded activating and inhibitory cell surface receptors [2]. Following target cell recognition and activation, lytic granules within the effector cells are polarized towards the immunological synapse, where they fuse with the plasma membrane and release their contents into the synaptic cleft between effector and target cell [3,4]. Similar to cytotoxic T cells, cell killing by NK cells is primarily mediated by the granzyme family of serine proteases, and the pore-forming protein perforin [5]. Thereby the pro-apoptotic granzyme B (GrB) plays the most crucial role for cytotoxicity [6]. Initially, GrB is expressed as an inactive precursor protein. This pre-pro-GrB carries an Nterminal signal peptide, directing packaging of the protein into secretory granules, followed by the activation dipeptide Gly-Glu. Removal of this peptide by the cysteine protease cathepsin C generates the enzymatically active form of GrB [7], which is stored together with other granzymes and perforin in the dense core of lytic granules. Upon release from cytotoxic lymphocytes, GrB enters target cells in cooperation with perforin, and rapidly induces apoptosis via caspase-dependent and caspase-independent mechanisms [8].
Owing to the relatively small size of 227 amino acid residues for mature GrB, its broad substrate specificity, and its ability to bypass common apoptosis resistance mechanisms in tumor cells, GrB has been employed as an effector molecule for the generation of recombinant cell death-inducing fusion proteins [9,10]. Since GrB is of human origin, such immunotoxin-like molecules are expected to circumvent immunogenicity and other complications frequently associated with recombinant toxins of plant or bacterial origin [11]. Recombinant GrB and chimeric GrB fusion proteins that harbor peptide ligands or antibody domains for tumor-specific cell recognition have been successfully produced in bacterial, yeast and mammalian expression systems [12,13,14,15,16,17], and have been shown to retain potent cytotoxicity upon targeted delivery into tumor cells [12,13,15,18,19,20].
Here, we investigated feasibility and consequences of expression of a chimeric GrB fusion protein in human NK cells, utilizing established NKL cells as a model. NK cells possess all pathways required for processing, packaging, and triggered release of endogenous wildtype GrB, which may be readily employed by an ectopically expressed retargeted GrB derivative. For selective targeting to tumor cells, we fused the epidermal growth factor receptor (EGFR) peptide ligand transforming growth factor a (TGFa) to human pre-pro-GrB. EGFR overexpression and aberrant activation have been found in many tumors of epithelial origin, and have been shown to contribute to malignant transformation [21]. Due to its accessibility from the extracellular space, EGFR constitutes an attractive target for therapeutic antibodies and cytotoxic antibody or growth factor fusion proteins [15,22,23,24]. NK cells transduced with a lentiviral vector encoding the GrB-TGFa fusion protein expressed the chimeric GrB-T molecule in amounts comparable to endogenous wildtype GrB, which augmented natural cytotoxicity of the genetically modified NK cells against NK-sensitive targets. Furthermore, induction of degranulation resulted in the release of GrB-T from vesicular compartments into the extracellular space. The secreted fusion protein was functionally active, and displayed specific binding to EGFR-overexpressing tumor cells as well as selective target cell killing in the presence of an endosomolytic activity.

Results
Expression of the granzyme B-TGFa fusion protein GrB-T in NK cells cDNA encoding human pre-pro-GrB was fused via a flexible (Gly 4 Ser) 4 -His 6 linker to a sequence encoding the EGFR-specific peptide ligand TGFa followed by C-terminal Myc and hexahistidine tags in the lentiviral transfer vector pSIEW, that also encodes enhanced green fluorescent protein (EGFP) as a marker (Fig. 1A). After generation of lentiviral particles and transduction of human NKL cells, EGFP-expressing NKL/GrB-T cells were enriched by two rounds of flow cytometric cell sorting. Likewise, NKL/GrB S183A -T control cells were generated by lentiviral transduction with a vector encoding a fusion of TGFa with the enzymatically inactive GrB mutant GrB S183A [14]. In the enriched NKL/GrB-T and NKL/GrB S183A -T cell populations, GrB-T and GrB S183A -T mRNA expression was verified by semi-quantitative RT-PCR analysis (Fig. 1B). GrB-T and GrB S183A -T proteins with an apparent molecular mass of 40 kDa were readily detected in addition to endogenous wildtype GrB by immunoblot analysis of NKL cell lysates with GrB-specific antibody (Fig. 1C), confirming expression and integrity of the fusion proteins. Intracellular staining of NKL/GrB-T and NKL/GrB S183A -T cells with Myctag-specific antibody also verified expression of GrB-T and GrB S183A -T in transduced NKL cells (Fig. 1D).

Ectopic expression of GrB-T enhances natural cytotoxicity of NKL cells
Total levels of GrB in parental and genetically modified NKL cells were analyzed by intracellular staining. We observed a marked increase in total GrB in NKL/GrB-T and NKL/ GrB S183A -T cells in comparison to unmodified NKL, attributed to GrB-T and GrB S183A -T fusion proteins ectopically expressed in addition to endogenous wildtype GrB ( Fig. 2A, left panel). In contrast, expression levels of perforin remained unchanged by transduction with the lentiviral vectors, and were comparable in NKL, NKL/GrB-T and NKL/GrB S183A -T cells ( Fig. 2A, right  panel).
Next, we investigated natural cytotoxicity of NKL/GrB-T and NKL/GrB S183A -T cells in FACS-based assays with NKL-sensitive C1R-neo and Jurkat cells as targets. Parental NKL cells were included for comparison. As expected, co-incubation of Jurkat and C1R-neo cells with unmodified NKL cells for 4 h revealed significant cell killing activity at different E/T ratios, which was very similar for NKL/GrB S183A -T cells that express the enzymatically inactive fusion protein (Fig. 2B, upper panels). In contrast, natural cytotoxicity of NKL/GrB-T cells was markedly increased, resulting in 70% lysis of C1R-neo cells at an E/T ratio of 20:1, while cell killing by NKL/GrB S183A -T and parental NKL was 38 and 30% under these conditions. Also cytotoxicity of NKL/GrB-T cells towards Jurkat cells was enhanced, albeit to a lower extent (45% cell killing at an E/T ratio of 20:1 versus 29 and 33% cell killing for NKL/GrB S183A -T and NKL, respectively).
Release of secretory granular proteins and perforin activity require the presence of Ca 2+ . When cell killing experiments with C1R-neo target cells were performed in the presence of the Ca 2+ chelator EGTA, cytotoxicity of NKL/GrB-T, NKL/GrB S183A -T, and NKL cells was abolished, indicating dependence of cell killing on granule exocytosis and perforin (Fig. 2B, lower panel, left). The similar transfer vector pS-GrB S183A -T-IEW encodes enzymatically inactive mutant GrB S183A fused to TGFa (not shown). After transduction with S-GrB-T-IEW or S-GrB S183A -T-IEW vector particles, EGFP-expressing NKL/ GrB-T and NKL/GrB S183A -T cells were enriched by flow cytometric cell sorting, and analyzed for GrB-T expression. (B) GrB-T mRNA expression in NKL/GrB-T and NKL/GrB S183A -T cells was verified by semi-quantitative RT-PCR. (C) Expression of GrB-T proteins in NKL/GrB-T and NKL/GrB S183A -T cells was investigated by immunoblot analysis of cell lysates with GrBspecific antibody. c-Tubulin was analyzed as a loading control. (D) Expression of GrB-T proteins in NKL/GrB-T and NKL/GrB S183A -T cells was confirmed by intracellular staining with Myc-tag-specific antibody and flow cytometry (open areas). In all experiments parental NKL cells served as controls. doi:10.1371/journal.pone.0061267.g001 C1R-neo and Jurkat cells do not express EGFR (Fig. 2B, lower panel, right). This suggests that in the absence of the TGFa target receptor, GrB-T fusion protein can act in a manner similar to endogenous wildtype GrB and enhance natural cytotoxicity of NKL cells upon encounter of NK-sensitive targets. For this increased target cell killing, enzymatic activity of the GrB-T fusion protein as well as intact granule exocytosis and perforin activity were required.

Cytotoxicity of NKL/GrB-T cells towards EGFR-expressing tumor cells
Next, we investigated whether NKL/GrB-T cells also display enhanced cytotoxic activity against EGFR-expressing tumor cells employing established human MDA-MB468 breast carcinoma and A431 squamous cell carcinoma cells as targets. These cells express high levels of EGFR on the cell surface as confirmed by flow cytometry (Fig. 3A). Cell killing activity of NKL/GrB-T and NKL/GrB S183A -T cells against MDA-MB468 and A431 cells was assessed in FACS-based assays upon co-incubation of effector and target cells for 4 h at different ratios. Parental NKL cells were included for comparison (Fig. 3B). In contrast to Jurkat and C1Rneo cells, at the tested E/T ratios EGFR-expressing MDA-MB468 and A431 cells were resistant to lysis by parental NKL cells as well as the genetically modified derivatives NKL/GrB-T and NKL/ GrB S183A -T.
NK cell-mediated cytotoxicity requires efficient activation of the effector cells upon recognition of target cells. To investigate polarization of cytotoxic granules to the site of contact and release of granular contents into the synaptic cleft, we performed confocal laser scanning microscopy analysis with mixed cultures of NKL, and NKL-sensitive or NKL-resistant cancer cells as targets. Parental NKL or NKL/GrB-T cells were added to C1R-neo or MDA-MB468 cells, and stained with perforin-specific antibody to visualize cytotoxic granules and distinguish effector and target cells. NK cells in the absence of targets were included for comparison. In NK cells incubated for 1 h with C1R-neo, perforin-containing granules were concentrated at the interface with the target cells (Fig. 4A, middle panel), indicative of effector cell activation. In contrast, upon incubation with MDA-MB468,  the majority of NKL and NKL/GrB-T cells formed conjugates with the targets, but showed little or no polarization of cytotoxic granules towards the site of contact (Fig. 4A, right panel). This resembled the distribution pattern observed in unstimulated NKL cultured without target cells (Fig. 4A, left panel). To assess differences in effector cell activation by the cancer cells, we subsequently analyzed activation-induced degranulation of NKL and NKL/GrB-T cells upon contact with C1R-neo and MDA-MB468. Cells were co-incubated at an E/T ratio of 1:1 for 5 h, before surface expression of the lysosomal associated membrane protein LAMP-1 (CD107a) was measured by flow cytometry. This functional marker for CTL and NK cell activity resides in the membrane of lytic granules, and is mobilized to the cell surface following activation-induced granule exocytosis. While CD107a levels were markedly increased on the surface of NKL and NKL/ GrB-T cells upon encounter of C1R-neo cells (Fig. 4B, upper panel), only marginal changes in CD107a expression were observed upon contact of the effector cells with MDA-MB468 target cells (Fig. 4B, lower panel).
Hence, while NKL and NKL/GrB-T cells formed conjugates with the EGFR-expressing targets, this interaction did not induce effector cell activation, degranulation and target cell killing.

Release of GrB-T protein from NKL/GrB-T cells upon activation-induced degranulation
To facilitate release of GrB-T fusion protein from NKL cells for functional analysis, NKL/GrB-T and NKL/GrB S183A -T cells were stimulated with PMA and ionomycin, and induction of degranulation was confirmed by demonstrating enhanced expression of CD107a on the cell surface (Fig. 5A). Enzymatic activity of released proteins was analyzed using the GrB-specific peptide substrate Ac-IETD-pNA (Fig. 5B). Thereby concentration-dependent GrB activity was found in supernatant of activated parental NKL cells. As expected, this was very similar for NKL/GrB S183A -T cells that express enzymatically inactive fusion protein but also harbor endogenous wildtype GrB. In contrast, markedly enhanced GrB activity was detected in supernatant from activated NKL/ GrB-T, indicating release of enzymatically active GrB-T fusion protein from these cells in addition to endogenous wildtype GrB.
Functionality of the TGFa domain of GrB-T was investigated in cell binding experiments with EGFR-overexpressing MDA-MB468 cells. As a control, human MDA-MB453 breast carcinoma cells were included, which are negative for EGFR [20]. Tumor cells were incubated with culture supernatants of activated NKL/ GrB-T and NKL/GrB S183A -T cells, and surface-bound proteins were detected by flow cytometry with fluorochrome-labeled GrBspecific antibody. For proteins from NKL/GrB-T and NKL/ GrB S183A -T cells, strong binding to the surface of MDA-MB468 cells, but not EGFR-negative MDA-MB453 cells was found (Fig. 5C). Endogenous GrB released from parental NKL cells displayed only marginal binding to MDA-MB468 and MDA-MB453 cells.
These data show that GrB-T and GrB S183A -T fusion proteins were indeed released from NKL cells upon activation-induced degranulation. While GrB-T was bifunctional and displayed enzymatic activity as well as EGFR-specific binding, the GrB S183A -T control protein retained cell binding, but as expected did not cleave GrB substrate.

Cytotoxic activity of GrB-T protein from activated NKL cells
To analyze cytotoxic activity of GrB-T protein from NKL/ GrB-T cells, the cells were treated with PMA and ionomycin as described above, cell-free supernatant was collected, and increasing concentrations from 50 to 200 mg/mL of total granular proteins were added to MDA-MB468 cells. Equal concentrations of proteins released by NKL/GrB S183A -T and parental NKL cells were included as controls. After 24 h, the relative number of viable cells in comparison to target cells cultured in PMA-and ionomycin-containing medium in the absence of NK cell proteins was determined in WST-1 metabolization assays (Fig. 6A). A slight reduction in cell viability was observed at high protein concentrations. However, this was independent from the source of the granular proteins, and most likely due to the activity of endogenous granzymes and perforin released by the different NKL cell derivatives. To test whether insufficient intracellular uptake of GrB-T protein was responsible for the lack of specific cell killing, MDA-MB468 cells were treated with supernatant from activated NKL/GrB-T cells, and uptake of GrB-T after 1.5 h at 37uC was analyzed by confocal laser scanning microscopy using fluorochrome-conjugated GrB antibody. In the majority of target cells, punctuate staining within the cytoplasm was found (Fig. 6B, upper panel), suggesting receptor-mediated endocytosis of GrB-T upon binding to EGFR and routing to endosome-like vesicular structures. MDA-MB468 cells incubated with the same amount of protein from activated parental NKL cells containing endogenous wildtype GrB did not display intracellular staining with GrBspecific antibody (Fig. 6B, lower panel).
To test whether release from endosomal vesicles in target cells could facilitate access of GrB-T to cytosolic GrB substrates and enable cytotoxic activity, next we treated target cells with increasing concentrations of total granular proteins from activated NKL/GrB-T cells in the presence of chloroquine. Indeed, addition of the endosomolytic reagent strongly enhanced cytotoxicity of GrB-T protein, resulting in effective and concentrationdependent killing of target cells (Fig. 6C). Interestingly, chloroquine also increased cytotoxic activity of endogenous granular proteins from NKL and NKL/GrB S183A -T cells. However, in these cases much higher protein concentrations were required to achieve significant cell killing.

Specificity of GrB-T-mediated cell killing
Apoptotic cell death is a hallmark of GrB-induced cytotoxicity.
To examine whether increased cell death observed upon treatment with GrB-T was due to GrB-induced apoptosis, MDA-MB468 cells were incubated with supernatant from activated NKL/GrB-T cells containing 100 mg/mL of total granular proteins in the presence of chloroquine. Induction of apoptosis was measured by determining the percentage of Annexin V and propidium iodide double-positive cells. After 24 h, 34% of GrB-T-treated cells were apoptotic, while treatment with granular proteins from NKL/ GrB S183A -T and parental NKL cells both resulted in 20% target cell apoptosis (Fig. 7A).
Next, we investigated whether enhanced cytotoxicity of GrB-T protein was dependent on EGFR expression on target cells. EGFR-negative MDA-MB453 and EGFR-positive MDA-MB468 cells were each treated for 24 h with supernatant from activated NKL/GrB-T cells containing 100 mg/mL of total granular proteins in the presence of chloroquine, and effects on cell viability were determined. As observed before, GrB-T-containing supernatant was highly toxic for MDA-MB468 cells (Fig 7B, left  panel), while no significant reduction in viability was observed upon treatment of MDA-MB453 cells (Fig 7B, right panel). To confirm specificity of GrB-T-mediated cell killing, a similar experiment was performed with MDA-MB468 cells pre-treated with EGFR-specific antibody 425 [25]. While in the absence of competitor GrB-T treatment resulted in 54% cell killing, blockade of the ligand binding site of EGFR by the antagonistic antibody abrogated GrB-T binding and reduced cytotoxicity to 30%, which was similar to the levels observed for granular proteins from NKL and NKL/GrB S183A -T cells (Fig. 7C).
The inability of mutant GrB S183A -T to enhance cell death indicates that the enzymatic activity of the GrB domain is required for specific cell killing by GrB-T. This was confirmed by pretreatment of GrB-T from activated NKL/GrB-T cells with the GrB-specific peptide aldehyde inhibitor Ac-IETD-CHO, which markedly reduced specific cytotoxicity (Fig. 7D). Specific binding of GrB-T from NKL/GrB-T cells and high and selective cytotoxicity in the presence of chloroquine was also observed for other EGFR-overexpressing tumor cells such as A431, confirming that these effects are not restricted to MDA-MB468 target cells (Fig. 8).
Taken together, these data demonstrate that GrB-T released by NK cells is bifunctional and selectively cytotoxic for EGFRoverexpressing target cells. Cell killing by GrB-T required TGFamediated binding to EGFR on the cell surface, release from vesicular compartments in target cells, and subsequent cleavage of intracellular GrB substrates.

Discussion
Significant progress has been made over the last decade towards realizing the potential of natural killer cells for cancer immunotherapy [1,26]. NK cells can respond rapidly to transformed and stressed cells, and have the intrinsic potential to extravasate and reach their targets in almost all body tissues. In experimental models, tumor-specificity of NK cells has been enhanced by genetic modification for expression of chimeric antigen receptors (CAR), that trigger the effector cells' endogenous cytotoxicity upon antibody-mediated recognition of a defined tumor cell surface antigen [27,28,29,30,31]. Alternatively, antitumoral activity may be augmented by providing NK cells with a targeted effector molecule that is released upon NK cell activation in soluble form, and can act in concert with endogenous natural cytotoxicity mechanisms. As a first step in this direction, here we expressed a chimeric granzyme B fusion protein in human NK cells that carries the peptide ligand TGFa for targeting to EGFR, and evaluated its effect on natural cytotoxicity and its cell killing activity towards EGFR-expressing tumor cells.
We used established NKL cells as a model system [32]. These cells display natural cytotoxicity and can mediate antibodydependent cell-mediated cytotoxicity (ADCC), but their overall cytotoxic potential is moderate [33]. Transduction with a GrB-Tencoding lentiviral vector resulted in the integration of the provirus into the NKL cells' genome (data not shown), and stable long-term expression of the chimeric fusion protein at levels comparable to endogenous wildtype GrB in these cells. To direct ectopically expressed GrB-T to the lytic granules of the NK cells and facilitate activation-triggered release, we employed complete pre-pro-GrB as an effector domain. Consequently, the chimeric fusion protein like wildtype GrB depends on removal of the signal peptide and subsequent processing by cathepsin C to generate the free N-terminus of mature GrB, that is required for enzymatic activity [7,14]. Expression and granular storage of GrB-T resulted in a marked increase in total intracellular GrB levels, while not affecting the levels of other granular proteins like perforin.
NKL/GrB-T cells displayed significantly enhanced natural cytotoxicity towards NK-sensitive target cells. This could be attributed to the activity of GrB-T, since NKL/GrB S183A -T cells that produce similar amounts of the enzymatically inactive fusion protein only showed cell killing comparable to unmodified parental NKL. Cytotoxicity of NKL/GrB-T cells was abrogated by addition of EGTA, which chelates Ca 2+ essential for the release of granular proteins and for perforin activity [34,35,36]. This suggests that ectopically expressed GrB-T is released via the same mechanism as endogenous wildtype GrB, and likely employs perforin to gain access to GrB substrates in target cells. In contrast to EGFR-negative leukemia and lymphoma cells, EGFR-overexpressing cancer cells of solid tumor origin were resistant to the cytolytic activity of NKL cells. Unexpectedly, this was also the case for NKL/GrB-T cells that express the EGFR-specific fusion protein. This could be explained by insufficient activation of the NK cells upon contact with the NKL-resistant targets. While conjugates were formed between effector and MDA-MB468 target cells, lytic granules were not re-oriented towards the site of contact, and degranulation of NKL cells and release of granular proteins into the immunological synapse was not induced. Hence, GrB-T did not gain access to the target cells.
To facilitate release of GrB-T along with endogenous granular proteins from vesicular compartments, we stimulated genetically modified NKL cells with the phorbol ester PMA and the calcium ionophore ionomycin [37,38]. Culture supernatant was collected and used for subsequent analysis of the fusion protein. The presence of GrB-T in supernatant and its functionality were confirmed in cell binding experiments and enzymatic assays with a synthetic GrB substrate. We observed strong and selective binding of the fusion protein to EGFR-expressing tumor cells. Despite efficient cell binding, however, GrB-T released from activated NKL/GrB-T cells did not facilitate killing of EGFR-overexpressing tumor cells in the absence of an exogenous endosomolytic activity. Instead, we found the protein trapped in target cells within intracellular vesicles, where it did not have access to its cytosolic substrates.
Under physiological conditions, delivery of wildtype GrB into the cytosol of target cells is aided by the pore-forming protein perforin. This is accomplished either via formation of active perforin pores directly in the cell membrane, or disruption of vesicular membranes after co-endocytosis of GrB and perforin [39,40,41]. Enhanced natural cytotoxicity of NKL/GrB-T cells towards EGFR-negative targets suggests that GrB-T like wildtype GrB can cooperate with perforin to enter cells and induce cell death. The TGFa domain contributes approximately 6 kDa to the molecular mass of GrB-T, which is well within the size limit previously suggested for cooperation of GrB fusion proteins or conjugates with perforin [33]. Within the synaptic cleft between fully activated effector cells and targets, high local perforin concentrations can be expected. In contrast, the amount of perforin in culture supernatants of PMA and ionomycin stimulated NKL cells was below the detection limit of immunoblots, and may not have supported cytosolic delivery of GrB-T.
However, when the endosomolytic reagent chloroquine was present, GrB-T from NKL/GrB-T supernatant efficiently killed EGFR-overexpressing tumor cells. Chloroquine is a weak base that accumulates in acidic compartments such as late endosomes and lysosomes, where it interferes with the pH equilibrium, leading to osmotic rupture of the vesicles [42]. It has previously been used to facilitate cytosolic delivery of recombinant GrB fusion proteins and other targeted cell-death inducing proteins in the absence of perforin [15,20,43]. Importantly, cell killing activity of GrB-T in the presence of chloroquine was strictly dependent on specific TGFa -EGFR interaction as well as enzymatic activity of the GrB domain. Thereby viability of EGFR-overexpressing breast carcinoma and squamous cell carcinoma cells was reduced to about 50% at a concentration of 100 mg/mL of total granular proteins from activated NKL/GrB-T cells, which corresponds to approximately 5 to 10 ng/mL (125 to 250 pM) of GrB-T as estimated in cytotoxicity assays employing recombinant GrB-T expressed in yeast as a standard [15] (data not shown).
Our data demonstrate that ectopic expression of a targeted GrB fusion protein can enhance antitumoral activity of NK cells. Upon proper activation by target cells, the chimeric GrB-T molecule was released, exhibiting functionality similar to wildtype GrB and markedly enhancing natural cytotoxicity. In addition, the fusion protein displayed EGFR-specific binding and the ability to kill EGFR-expressing tumor cells specifically. Hence, NK cells modified to express a targeted GrB derivative may become therapeutically useful, if NK cell activation at the tumor site and triggered release of the effector molecule can be ensured. This may for example be achieved by co-expressing a targeted GrB fusion protein together with a tumor-specific chimeric antigen receptor of the same or different specificity, that can bypass natural mechanisms preventing activation [31]. Experiments in this direction have been initiated.

Lentiviral GrB-T expression constructs and transduction of NKL cells
cDNA encoding a fusion of human pre-pro-GrB, linked via a flexible peptide linker to human TGFa and C-terminal Myc and His 6 tags was assembled by stepwise PCR utilizing full-length GrB cDNA [14] and plasmid pPIC9-GrB-T [15] as templates. The GrB-TGFa (GrB-T) sequence was inserted into plasmid pHR'SIN-cPPT-SIEW (pSIEW) [44] upstream of IRES and EGFP sequences of the vector, resulting in the lentiviral transfer plasmid pS-GrB-T-IEW which allows co-expression of GrB-T and EGFP. Following a similar strategy, plasmid pS-GrB S183A -T-IEW was generated that encodes the enzymatically inactive control protein GrB S183A -T [15]. VSV-G pseudotyped lentiviral vector particles were produced by co-transfecting 293T cells with the respective lentiviral transfer plasmid together with packaging and envelope plasmids pCMVDR8.91 and pMD2.G [45] using standard calcium phosphate transfection. Culture supernatants containing pseudotyped lentiviral vector particles were collected 48 h later, and sterile filtered to remove cell debris. For transduction, lentiviral particle-containing supernatants were added to NKL cells in the presence of 8 mg/mL polybrene, and the mixtures were centrifuged for 90 min at 32uC and 1,8006g. Then, the cells were cultured overnight at 37uC before replacing the medium. After 48 h, cells were analyzed for EGFP expression by direct flow cytometry using FACSCalibur or FACSCanto II flow cytometers (BD Biosciences, Heidelberg, Germany) and CELLQuest Pro or FACSDiva software (BD Biosciences). Homogeneous pools of EGFP-expressing NKL cells were obtained by two rounds of sorting using a FACSAria fluorescence-activated cell sorter (BD Biosciences).
For intracellular staining, up to 1610 6 NKL/GrB-T, NKL/ GrB S183A -T, or parental NKL cells were cultured in the presence of GolgiStop TM for 5 h at 37uC, collected by centrifugation, washed once with DPBS, and fixed and permeabilized by adding 200 mL of BD Cytofix/Cytoperm TM solution (BD Biosciences) for 30 min at room temperature. Then, cells were washed twice with BD Perm/Wash TM buffer (BD Biosciences), and incubated with either Alexa Fluor 647-conjugated Myc-tag-specific antibody 9E10 (Santa Cruz Biotechnology) for detection of GrB-T and GrB S183A -T fusion proteins, Alexa Fluor 647-conjugated human GrBspecific antibody GB11 (BD Biosciences) for analysis of total GrB levels, or phycoerythrin-conjugated perforin-specific antibody dG9 (BD Biosciences) for analysis of total perforin levels. Subsequently, cells were analyzed using a FACSCalibur flow cytometer and CELLQuest Pro software.

Cytotoxicity assay
Cytotoxic activity of NKL effector cells towards target cells was analyzed in FACS-based assays as described [31]. Briefly, target cells were labeled with calcein violet AM (Molecular Probes, Invitrogen, Karlsruhe, Germany), washed, and then co-cultured with effector cells at various effector to target (E/T) ratios for 4-6 h at 37uC. After co-culture, cells were centrifuged, and 200 mL of a 1 mg/mL propidium iodide (PI) solution were added to each sample shortly before flow cytometric analysis with a FACSCanto II flow cytometer. Specific cytotoxicity was calculated using FACSDiva software. Dead target cells were determined as calcein violet AM and PI double positive. Spontaneous target cell lysis was analyzed in samples containing only labeled target cells, and subtracted to obtain specific cell killing.

Degranulation of NKL cells
Activation of NKL cells and induction of degranulation was investigated by analyzing surface expression of lysosomal-associated membrane protein LAMP-1 (CD107a) using BD FastImmune TM CD107a APC detection antibody (BD Biosciences) according to the manufacturer's instructions. Briefly, NKL cells (1610 6 cells/mL) were either mixed with target cells at an E/T ratio of 1:1, or stimulated with 1 mg/mL phorbol 12-myristate 13-acetate (PMA) and 1 mg/mL ionomycin (both Sigma-Aldrich) for 60 min at 37uC in the presence of 5 mL of anti-CD107a-APC antibody. Then, 1 mL of GolgiStop TM was added to each sample, followed by incubation for another 4 h at 37uC. Cells were washed, and CD107a expression was analyzed using a FACSCalibur cytometer and CellQuestPro software. To collect proteins released from cytotoxic granules, NKL cells (up to 5610 6 cells/mL) were stimulated with 1 mg/mL each of PMA and ionomycin for 6 h at 37uC in serum-free RPMI 1640 medium supplemented with 2 mM L-glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin, and 200 IU/mL IL-2. Culture supernatants were collected and sterile filtered to remove cell debris before subsequent analysis. Yields of total granular proteins from 5610 6 cells were typically in the range of 200 to 750 mg.

GrB activity assay
Enzymatic activity of GrB proteins in supernatants from activated NKL cells was analyzed using the synthetic GrB substrate Ac-IETD-pNA (Alexis, Grünberg, Germany) as described [20]. Cleavage reactions were prepared in duplicates with reaction buffer (10 mM Hepes pH 7.4, 140 mM NaCl, 2.5 mM CaCl 2 ) and 200 mM Ac-IETD-pNA in a total volume of 150 mL per sample in 96-well plates. Samples were incubated for 20 h at 37uC, and substrate cleavage was quantified by measuring the absorbance (A) at 405 nm (corrected for background by subtracting A490) with a microplate reader (Molecular Devices, Ismaning, Germany). Peptide substrate incubated in reaction buffer without granular proteins served as blank.

Binding analysis
To examine EGFR surface expression, target cells (5610 5 ) were incubated with anti-human EGFR antibody R1 (Santa Cruz Biotechnology) for 60 min at 4uC, followed by APC-conjugated goat anti-mouse secondary antibody (Jackson ImmunoResearch, West Grove, PA, USA), and analysis with a FACSCalibur flow cytometer and CELLQuest Pro software. To determine cell binding of GrB-T and GrB S183A -T expressed in NK cells, target cells (5610 5 ) were incubated with up to 50 mg of total granular proteins from activated NKL cells for 60 min at 4uC. Unbound proteins were removed, cells were washed, and bound GrB fusion proteins were detected with Alexa Fluor 647-conjugated antihuman GrB antibody GB11 by flow cytometry as described above.

Cell viability and apoptosis assays
Cytotoxic activity of GrB fusion proteins from NKL cell supernatants was analyzed in WST-1 cell viability assays (Roche Diagnostics) following the manufacturer's recommendations. Target cells were seeded in 96-well plates at a density of 122610 4 cells/well in triplicates, and incubated for 24 h at 37uC with 50 to 200 mg/mL of total granular proteins from supernatants of activated NKL cells in the presence or absence of 25 mM chloroquine. After addition of WST-1 reagent for 4 h, the relative number of viable cells in comparison to medium-treated control cells was determined by measuring the absorbance (A) at 450 nm (corrected for background by subtracting A690). To compete binding of GrB-T and GrB S183A -T to EGFR, target cells were pre-incubated with 50 mg/mL of anti-EGFR antibody 425 [25] or an isotype-matched control antibody for 30 min at 4uC prior to the addition of NKL cell supernatant and analysis of cytotoxicity. Dependence of cytotoxic activity on the enzymatic activity of GrB was confirmed by pre-incubation of total granular proteins from supernatants of activated NKL cells with 400 mM of GrB-specific peptide aldehyde inhibitor Ac-IETD-CHO (Alexis) for 30 min at room temperature before addition to target cells and analysis of cytotoxicity. To determine induction of apoptosis, target cells were seeded in 24-well plates at a density of 2610 5 cells/well, and incubated for 24 h at 37uC with 100 mg/mL of total granular proteins from supernatants of activated NKL cells in the presence of 25 mM chloroquine. Control cells were treated with chloroquine-containing medium. For each sample, adherent cells were detached with trypsin-EDTA and combined with cells floating in the culture medium. Harvested cells were washed with PBS, and labeled with Annexin V-APC and PI using the Annexin V Apoptosis Detection Kit (eBiosciences, Frankfurt, Germany) according to the manufacturer's instructions. Fluorescence of cells was measured with a FACSCanto II flow cytometer, and the percentage of apoptotic and dead cells was calculated using FACSDiva software.

Confocal laser scanning microscopy
To investigate conjugate formation and activation of NKL cells upon target cell recognition, NKL cells were mixed with target cells (E/T ratio of 1:1), and adhered to glass slides coated with poly-L-lysine (Sigma-Aldrich). After co-incubation for 60 min at 37uC, cells were washed, fixed with 4% paraformaldehyde in PBS for 20 min at room temperature, and permeabilized and blocked for 60 min in buffer containing 0.1% Triton X-100, 3% BSA in PBS. Then cells were incubated with anti-human perforin antibody dG9, followed by Alexa Fluor 546-conjugated antimouse IgG (Molecular Probes). Cells were washed twice with PBS, and analyzed with a Leica CTR 6500 laser scanning microscope (Leica Microsystems, Bensheim, Germany). For analysis of uptake of GrB-T fusion protein, MDA-MB468 target cells were grown overnight on coverslips, and then treated for 60 min at 4uC with supernatant of activated NKL/GrB-T or parental NKL cells containing 200 mg/mL of total granular proteins. Then cells were washed, and incubated for another 90 min at 37uC in fresh culture medium to initiate uptake of GrB-T. Cells were fixed with 4% paraformaldehyde in PBS for 20 min at room temperature, permeabilized, and blocked for 60 min in buffer containing 0.1% Triton X-100, 3% BSA in PBS. This was followed by incubation of the cells with 2 mg/mL Alexa Fluor 647-conjugated anti-human GrB antibody GB11 in 3% BSA in PBS for detection of GrB-T. Nucleic acids were counterstained with DAPI. Cells were washed twice with PBS, and analyzed with a Leica CTR 6500 laser scanning microscope.

Statistical analysis
Differences between values were evaluated using the two-tailed unpaired Student's t test. P values ,0.05 were considered significant. Statistical calculations were done using Prism 5 software (GraphPad Software, La Jolla, CA).