Retroviral vector production

The process of producing retroviral vectors and transducing T-cells followed previously established methods [32]. In short, HEK293T cells were transiently transfected with two packaging plasmids (PeqPam, RD114) along with the SFG vector containing the CAR construct, using GeneJuice (Merck Millipore, Billerica, MA). Retroviral supernatant was collected after 48 and 72 h and stored at − 80C. 48-h and 72-h supernatants were mixed in a one-to-one ratio prior to transduction.

NK cell isolation, expansion and CAR-NK generation

Peripheral blood mononuclear cells were obtained from peripheral blood by standard Ficoll-paque density gradient centrifugation, and T-cells were depleted via LD MACS columns (Miltenyi, Bergisch Gladbach, Germany) and CD3 microbeads (Miltenyi, Bergisch Gladbach, Germany) in accordance with the manufacturer's instructions. Subsequently, NK cells were activated by four-day co-culture with irradiated feeder cells in a G-Rex 10 vessel (Wilson Wolf Manufacturing, St. Paul, MN) prior to being transduced in non-TC-treated, 24-well plates coated with RetroNectin (Takara Bio, Kusatsu, Japan). The NK cell expansion media consisted of 90% SCGM (CellGenix, Freiburg, Germany) supplemented with 10% Gibco™ Fetal Bovine Serum (ThermoFisher, Waltham, MA), 1% penicillin/streptomycin antibiotic mix (ThermoFisher, Waltham, MA) and 2 mM L-glutamine (Invitrogen, Waltham, MA). To support NK cell expansion, rhIL-15 (Peprotech, Hamburg, Germany) was supplemented at regular intervals to a final concentration of 10 ng/ml. The growing NK cell cultures were split on days 7, 10 and 14 of production. CAR-NK transduction efficiency was measured on day 10 of production via flow cytometry as described previously [32]. The NK cell percentage purity of the cultures was measured at regular intervals via flow cytometry. Functional assays were performed between days 14 and 17 of production.

Cell lines

The K562 CS feeder cell line was a kind gift from Prof. Rooney (Baylor College of Medicine, Houston, TX). The HL-60, Molm-13, OCI-AML2, U937, HEL, IMS-M2, KG1-a, THP1, MV4-11, K562, Molp-8 and RPMI-8226 cell lines were obtained from the DSMZ (German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany). All cell lines were authenticated through the DSMZ prior to their use in experiments. All cell lines were subjected to regular mycoplasma contamination testing using a PCR Mycoplasma Test Kit I/C (PromoCell, Heidelberg, Germany). All cell lines were maintained in RPMI-1640 supplemented with 10% HI-FBS and 2 mM L-glutamine as well as 1% penicillin/streptomycin (ThermoFisher, Waltham, MA) in a humidified atmosphere containing 5% CO2 at 37 °C. The azacitidine/venetoclax resistant cell lines Molm-13, HL-60 and OCI-AML2 were a kind gift from Prof. Müller-Tidow (University Clinic Heidelberg, Heidelberg, Germany). The process of their conditioning is described in detail in [68]. A final concentration of 300 nM venetoclax (Selleck Chemicals, Cologne, Germany) and 1 µM azacitidine (Selleck Chemicals, Cologne, Germany) was added weekly to maintain the resistance of these cell lines. The cells were washed thoroughly of azacitidine and venetoclax immediately prior to commencing experiments.

Primary AML sample culture

Primary AML samples were sourced from the biobank of the University Clinic Heidelberg and maintained in serum-free Stemspan media (StemCell Technologies, Vancouver, Canada) supplemented as described in [69] with 1 µM UM729, 20 ng/ml IL3, 20 ng/ml FLT3-L, 20 ng/l SCF, 20 ng/ml IL-6, 50 ng/ml TPO and 1% penicillin/streptomycin (ThermoFisher, Waltham, MA) for the duration of the experiments.

HPSC culture and CFU assay

HPSCs were isolated via MACS. After thawing, the HPSCs were controlled for CD34 + purity and their expression of CD33 and CD70 was measured via flow cytometry. HPSCs were kept in culture for 24 h after thawing. Colony formation assays were performed with H4434 methylcellulose from StemCell Technologies according to the manufacturer´s instructions. Briefly, the contents of co-cultures containing 2.000 viable HPSC were transferred into 2 ml of methylcellulose media and plated into 12-well tissue-culture treated plates. After 10 days of culture, the colonies formed were counted and visually recorded using a CKX53 microscope (Olympus LS).

CellTiter Glo viability assay and IC50 calculation

Cells were dispensed in 96-well flat-bottom, white, chimney-well plates in 100 µl of IMDM (Invitrogen, Waltham, MA) supplemented with 10% HI-FBS (ThermoFisher, Waltham, MA), then exposed to a serial dilution of Bortezomib or Carfilzomib in DMSO, or a DMSO-only control. After 24 h, total cell viability was measured using the CellTiter Glo luminescent viability assay (Promega, Fitchburg, USA) on a Tecan Spark microplate reader (Tecan, Männedorf, Switzerland), following the manufacturer's instructions. The absolute luminescence values were normalized to those of the DMSO controls. The IC50 values for each cell line or primary sample were calculated in GraphPad Prism7 using the nonlinear fit of [Inhibitor] versus response (four parameters). All measurements were performed in technical triplicate unless specifically stated otherwise.

Flow cytometry

The fluorochrome-conjugated anti-human antibodies were purchased from BD Biosciences (San Jose, CA) or Biolegend (San Diego, CA). Biotinylated protein L and fluorochrome-conjugated streptavidin for CD33 CAR detection were purchased from Biolegend (San Diego, CA). A complete list of all antibodies used in this work, including catalog numbers, is provided in the supplementary methods section (Figure S4). Flow cytometric data were acquired on the BD-LSR II (BD Bioscience, San Jose, CA) and BD-Celesta (BD Bioscience, San Jose, CA) flow cytometers using the high-throughput sampling module. Flow cytometric data was analyzed with FlowJo, version 10 (Tree Star, Ashland, OR).

In vitro NK cell functional analysis

Following 14 days of culture, NK cells were washed and resuspended in RPMI-1640 supplemented with 10% FBS. To easily distinguish target from effector cells, AML cell lines were tagged with CellTracker Green CMFDA (ThermoFisher, Waltham, MA) at a concentration of 1 µM following the manufacturer's instructions. After the fluorescent tagging step, the AML cell lines were incubated with Bortezomib or Carfilzomib at the determined IC50 dose, or a DMSO-control for 24 h as indicated. Following this pre-treatment, the target cells were thoroughly washed, and viable cell counts were determined using CountBright absolute counting beads (ThermoFisher, Waltham MA, USA) on a BD-Celesta flow cytometer before seeding in plates in the presence of NK cells.

Co-culture assay

AML cell lines were pre-treated for 24 h with DMSO or IC50 Bortezomib or Carfilzomib, respectively. The cells were then washed thoroughly and tagged with a 1 μM concentration of CellTracker Green, following the manufacturer´s instructions. After determining the counts and viability of the AML cells in each group via flow cytometry, 25,000 viable tumor cells in 100 μl of complete media were dispensed in each well of a 96-well flat bottom TC-plate. Following a 24-h incubation the contents of each well were harvested, mixed with counting beads and 7-AAD, then measured using the high-throughput sampling module attached to a BD Celesta flow cytometer. For co-culture assays performed in a 96-well plate format, 2.5E4 viable AML cells were dispensed per well. NK cells were added to the AML cells at a ratio of 2-to-1, 1-to-1 or 1-to-2 as stated in the corresponding figure legend. Twenty-four hours after the start of co-culture, cells were washed, CountBright beads (ThermoFisher, Waltham MA, USA) were added, and the co-culture was stained with 7-AAD (Biolegend, San Diego, CA). The co-cultures were acquired on a BD-Celesta flow cytometer (BD Bioscience, San Jose, CA) using a high-throughput sampling unit. The absolute cell counts per well were determined by normalizing 7AADneg viable cell counts to absolute counting beads in accordance with the manufacturer´s instructions.

Primary AML sample co-cultures

Prior to starting the co-culture, six primary AML samples were fluorescently tagged using CellTracker Green CMFDA (ThermoFisher, Waltham MA, USA) as described above, then pre-treated with a fixed concentration of 1 nM Bortezomib (Selleck Chemicals, Cologne, Germany) or 2 nM Carfilzomib (Selleck Chemicals, Cologne, Germany). To maintain AML cell viability and prevent in vitro differentiation during co-culture, both the AML cells and the NK cell effectors were washed and resuspended in serum-free primary Stemspan AML media supplemented as described above. The primary AML sample co-cultures were performed in a 384-well format. The viable cell count dispensed per well was adjusted to 1,25E4 to better fit the smaller wells. Staining and analysis were performed in the same fashion as the AML cell line co-cultures.

Stable conjugate formation assay

Stable conjugate formation was measured as described previously [70] and shown schematically (Figure S8B). Briefly, AML cell lines were fluorescently tagged and pre-treated with proteasome inhibitors at IC50 concentrations or a DMSO control for 24 h. Following this, AML cells were washed, counted, and mixed with NK cells at a 1-to-1 ratio in 5 ml round-bottom polypropylene FACS tubes. Five minutes later, the co-cultures were fixed with ice-cold 70% ethanol while vortexing vigorously, before being put on ice and immediately proceeding to flow cytometric analysis.

Interferon-γ ELISA

Following 24-h co-culture, 100 µl of supernatant were collected from each well and stored at -80 C. After thawing the supernatants, the concentration of Interferon-γ was measured using a standard Human Duo-Set ELISA kit and Ancillary Reagent kit II (R&D Systems, Minneapolis, USA) following the manufacturer's instructions.

Concurrent combinatorial treatment assay

ZsGreen-expressing U-937, Molm-13 and HL-60 AML cells were seeded in technical triplicate in flat-bottom 96-well TC plates. The culture media contained the respective IC50 concentration of Bortezomib or Carfilzomib, or a DMSO control. Concurrently, NK cells were seeded at a ratio of 1-to-1. After 48 h of co-culture, the first row of technical replicates was harvested, stained and the absolute counts of NK- and tumor cells were measured via flow cytometry. In all remaining wells, 100 µl of supernatant was removed before being exchanged with 100 µl of cell media containing proteasome inhibitors and a further 2,5E4 viable AML cells. The cells were then left in co-culture for a further 48 h before a second measurement was performed in analog to the first.

Western blot

Cell pellets were lysed in ice-cold RIPA-buffer containing 1X HALT protease inhibitor cocktail (ThermoFisher, Waltham, MA, USA). The protein concentration of the lysates was determined with a Pierce BCA Protein Assay Kit (ThermoFisher, Waltham, MA, USA) and measured on a TecanSpark microplate reader (Tecan, Männedorf, Switzerland). The protein samples were separated via SDS-PAGE gel electrophoresis and blotted on a nitrocellulose membrane as described previously [71]. 10 µg of total protein were pipetted in each pocket of the gel comb. The membranes were blocked with 1% BSA (ThermoFisher, Waltham, MA, USA) in TBS-T and stained overnight with the primary antibodies listed in supplementary Figure S4 followed by a stain with a mouse anti-rabbit IgG HRP-conjugate and chemiluminescent image capture.

Dynamic NFkB activity monitoring using the Jurkat TPR

The Jurkat triple parameter reporter cell line (JurkatTPR) [72] was a kind gift from the Steinberger lab (University Clinic Vienna, Austria). The cells were maintained in RPMI-1640 supplemented with 10% HI-FBS and 2 mM l-glutamine as well as 1% penicillin/streptomycin (ThermoFisher, Waltham, MA) in a humidified atmosphere containing 5% CO2 at 37 °C. JurkatTPR cells were activated using the Cell Activation Cocktail at a 1:500 dilution (without Brefeldin A) (Biolegend, San Diego, CA) for 12 h before Bortezomib, Carfilzomib or a DMSO control were added. Twelve hours later, the cells were washed and stained on ice prior to acquisition on a BD-LSR II flow cytometer.

Two-step RT-qPCR

Wild-type Molm-13 cells were treated with a serial dilution of Bortezomib for 24 h before lysis and RNA extraction using a commercial silica-column based kit (RNeasy Mini) (Qiagen). After extraction, the concentration and purity of the extracts were measured using a NanoDrop device. Following that, cDNA was synthesized using a PrimeScript RT-PCR kit (Takara Bio, Kusatsu, Japan) and random hexamer primers. qPCR was then performed in technical triplicate using a SYBR Green PCR Master Mix (ThermoFisher). All values were normalized to 18S rRNA as a housekeeping gene, then the experimental samples under proteasome inhibition were normalized to the DMSO controls to show transcript fold-change under proteasomal inhibition.

Colony forming unit assay

The methylcellulose colony forming unit assay was performed using MethoCult H4434 (StemCell Technologies) according to the manufacturer’s instructions. Briefly, CD34pos human stem-and-progenitor cells (HPSC) were isolated via MACS. A total of 2,000 viable HPSC were dispensed per well of a 96-well U-bottom tissue culture treated plate and co-cultured with 20,000 viable (CAR-) NK cells.

Xenograft models of AML, bioluminescent imaging

Six-to-twelve-week-old NOD-Prkdcscid-IL2rgTm1/Rj (NSG) (NSG) mice were transferred from the DKFZ animal facility or sourced from internal breeding at the Interfacultary Biomedical Research Facility at the University of Heidelberg (IBF), as available. The animals were maintained under pathogen free conditions in ventilated cages at the IBF animal facility. The animal experiments described were approved by the Animal Welfare Committee of the Karlsruhe regional council and performed under approval number 35-9185.81/G291/21. The NSG mice received intravenous injections of tumor cells and NK cells via the tail vein after infrared light illumination. Proteasome inhibitors were injected in the tail vein at the doses stated in the relevant results section. Prior to injection, the proteasome inhibitors Bortezomib and Carfilzomib were mixed with PEG300 (SelleckChem, Cologne, Germany) and Tween80 (SelleckChem, Cologne, Germany) at the ratio recommended for in vivo use by the company. All AML cell lines used in the animal experiments were stably transduced with a construct encoding zsGreen and Click Beetle Green to allow for their flow-cytometric and bioluminescent tracking, respectively. The AML-cell burden was monitored by bioluminescent imaging (BLI; photons per second per cm2 spectral radiance), using the in-house IBF Xenogen In Vivo Imaging System (IVIS; Caliper Life Sciences, Hopkinton, MA) ten minutes after the intraperitoneal injection of D-Luciferin substrate (PerkinElmer, Waltham, MA).

Calculating synergy

Synergy was calculated using Jin’s modified Bürgi formula as described previously [27, 28].

$$Combinatorial\, Index\, \left(q\right)\,=\,\frac$$

Synergy is defined as a combinatorial index (q) greater than 1. A combinatorial index of 1 indicates additivity and a combinatorial index of less than 1 indicates anti-synergy.

To calculate the combinatorial index, we first defined the effect being measured. Our parameter of interest is the antileukemic potency of the proposed treatment and the surrogate marker is the viability loss of the AML cells.

Viability loss was defined as follows:

$$Viability\; loss = \frac \right)}} \right)}} 1$$

The predicted viability loss was defined as the theoretical loss of viability caused by the antileukemic efficacy of NK cells and proteasome inhibitors if these were purely additive.

Thus, the final equation for synergy calculations was:

$$\tiny Combinatorial\, Index \,\left(q\right)\,=\, \frac$$

$$Combinatorial\, Index\, (q)\,=\,1\, Additive\, effect \,only$$

$$Combinatorial \,Index\, (q)\,<1\, Antagonism$$

$$Combinatorial \,Index\, (q)\,>1\, Synergy$$

Statistical testing was done by calculating the Combinatorial Index as described above and performing a one-sample t-test against the hypothetical value of 1.0.

Statistical analysis and data visualization

Statistical analysis was performed using Prism 7 (GraphPad Software, San Diego, CA). The 2-tailed Student t test was used for comparisons between two groups. When comparing the means of more than two groups, a post-hoc multiple comparison analysis was performed. Dunnett’s multiple comparison correction was used when comparing multiple groups to a control group. Tukey’s multiple comparison correction was used when comparing the means of each group with the means of every other group. Survival was compared using log-rank (Mantel-Cox) tests after constructing Kaplan–Meier curves. Data was visualized using Prism 7 (GraphPad Software, San Diego, CA). Unless explicitly stated otherwise, all error bars indicate the means and the standard error of the mean (S.E.M).