Enhanced co-culture and enrichment of human natural killer cells for the selective clearance of senescent cells

Genotoxic stress-induced model of senescence

Human fetal lung fibroblasts (IMR-90) cells were used as a cell culture model of cellular senescence. Robust induction of SA-β-gal activity was observed in 87% of doxorubicin-treated (300 nM) cells, nine days after treatment whereas 1% of non-senescent control cells showed SA-β-gal activity (Figure 1A, 1B). In addition to doxorubicin treatment, additional models of senescence induction such as irradiation (20 Gy) and treatment with the mitochondria damaging agent, etoposide (20 μM, 48 h) were tested. A statistically significant percentage of cells with increased SA-β-gal activity was observed in cells exposed to x-rays or etoposide (87% and 81%, respectively) nine days after treatment compared to non-senescent control cells (1%) (Supplementary Figure 1B, 1C).

The increase in SA-β-gal activity of doxorubicin-treated fibroblasts correlated with persistent DNA damage, as measured by immunofluorescence staining for γ-H2AX (Figure 1C). The percentage of senescent cells with >2 γ-H2AX foci per cell was eight-fold higher compared to that of non-senescent cells (Figure 1D). As expected, mRNA levels of cell cycle checkpoint markers p16^INK4A and p21^CIP1 were significantly elevated in senescent compared to non-senescent cells (Figure 1E). Loss of high mobility group box 1 (HMGB1), a highly conserved nuclear protein from senescent cells, is often used as a biomarker of senescence [29]. The loss of HMGB1 from the nucleus of our senescent fibroblasts was confirmed in doxorubicin-treated cells (Supplementary Figure 1A). Further, loss of Lamin B1 expression, another hallmark of senescence [30], was also observed in senescent compared to non-senescent cells (Figure 1F). Finally, increased mRNA levels of classical SASP factors IL-6, IL-8 and IL-1α in senescent cells was confirmed by qRT-PCR [31, 32] (Figure 1G). Collectively, these findings validated robust induction of a senescence phenotype in our cell culture model.

Isolation and enrichment of NK cells

Freshly drawn whole blood (40 ml) was treated with RosetteSep human NK cell cocktail (Stem Cell Technologies, USA), which removes unwanted cells with Tetrameric Antibody Complexes that binds to white blood cells (except NK cells) and crosslinks them to red blood cells (RBCs). Following density gradient centrifugation, the NK cell population at the interface between the plasma and buoyant density medium was isolated. Freshly isolated NK cells were then incubated in RPMI media supplemented with 20% FBS and 100 IU/ml rIL-2 for 72 hours for enrichment (Figure 2A).

The relative expression of CD56 and absence of CD3 are routinely used to identify human NK cells [33]. Low (CD56^dim) and high (CD56^bright) CD56 expression levels define major subsets of NK cells [33]. CD56^bright CD16^- cells are considered immature NK cells that secrete interferon-γ (IFNγ), whereas CD56^dim CD16⁺ NK cells are responsible for cytotoxicity [34]. Upon physical interaction with target cells, cytotoxic NK cells release perforin, granzymes (serine proteases), and other cytotoxic granules that kill their targets [35]. FACS analysis was performed to characterize NK cells freshly isolated from donors between the ages of 20-42 years. Results revealed a classic distribution of CD56^bright CD16^- and CD56^dim CD16⁺ NK cells, as evidenced by CD16 versus CD56 flow cytometry analysis in the scatter plots of fresh human Peripheral blood mononuclear cells (PBMCs) exemplified by one donor (after gating out CD3+ T cells as shown in Figure 2B-i and Supplementary Figure 2B). Specificity of antibodies was confirmed by Fluorescence minus one (FMO) control experiments with NK cell populations (Supplementary Figure 2C-i, 2C-ii). Interestingly, statistically significant enrichment of both sub-populations of NK cells was observed after culturing cells in media with rIL-2 for three days (Figure 2B-ii). This increase in the numbers of NK cell subtypes was consistent in NK cells isolated and enriched from multiple donors, with a fifteen-fold increase in the numbers of CD56^dim CD16⁺ NK cells and approximately three-fold increase in the numbers of CD56^bright CD16^- NK cells (Figure 2C-i, 2C-ii).

NK cell-mediated killing of senescent cells

Senescent cells are known to secrete a variety of factors as part of the SASP that can attract NK cells and either activate or inhibit cytotoxic function [9, 14, 36]. To test whether senescent cells in our senescence model produced increased levels of these factors, we measured mRNA expression of multiple cytokines. A robust and statistically significant increase in the mRNA levels of CCL5, CXCL9 and CXCL11, which are known chemoattractants for NK cells, was observed in senescent cells compared to non-senescent controls (Figure 3A). A modest increase in mRNA expression of CCL2, decreased expression of CXCL12, and non-significant difference in Chemerin compared to non-senescent cells was also observed (Supplementary Figure 2A).

Although the role of NK cells in targeting senescent cells has been previously demonstrated, the protocols used are quite variable and employ conditions that may not be as physiologically representative. Hence the cytotoxic potential of NK cells isolated and enriched with our protocol for the selective killing of senescent cells was determined by lactate dehydrogenase (LDH) release. Even at a T:E ratio of only 1:1, NK cells selectively eliminated senescent fibroblasts at a significantly higher (43%) level compared to non-senescent control cells (15%) after 16 hours of co-culture. In addition, twice (83%) as many senescent fibroblasts were killed by NK cells when the T:E ratio was doubled to 1:2 compared to 41% in NS cells, and further increasing the T:E ratio to 1:3 killed nearly all senescent fibroblasts (93%) compared to 51% in NS cells. However, at higher T:E ratios of 1:2 and 1:3, substantial killing of non-senescent cells was also observed (Figure 3B and Supplementary Figure 3A).

Recent publications have reported substantial variability among senescent cells based on mode of senescence induction and cell type [37]. Thus, whether NK cells enriched by our protocol can eliminate senescent cells induced to senesce by different stressors was investigated. NK cells killed 43-50% of senescent cells whether senescence was induced by doxorubicin (Figure 3C-i), irradiation (Figure 3C-ii) or etoposide (Figure 3C-iii) at statistically higher levels compared to cytotoxicity towards non-senescent cells (15%). In addition, the senescence phenotype may also vary depending on cell type. Thus the efficacy of enriched NK cells in targeting senescent endothelial cells was also investigated [38]. Enriched and activated NK cells were observed to be effective at killing senescent endothelial cells at a three-fold higher rate than NS endothelial cells (Figure 3C-iv) in a statistically significant manner.

Freezing primary NK cells often results in a decline in cytotoxicity [39]. However, our results showed that NK cells isolated and enriched with the protocol described above remained significantly more cytotoxic towards senescent cells compared to non-senescent cells following cryopreservation at a rate comparable to freshly isolated NK cells, as measured by LDH release after 16 hours of co-culture (Supplementary Figure 3B). Interestingly, when co-culture duration was extended to four days, the vast majority (85-90%) of NS cells survived, and only 10% and 30% of senescent cells were viable following co-culture with fresh or revived NK cells, respectively, as determined by Calcein AM (Supplementary Figure 3C).

Results from four days of co-culture were confirmed, demonstrating that only 6% and 17% of senescent cells survived following co-culture with NK cells even when senescence was induced by irradiation (Supplementary Figure 3D-i) or etoposide (Supplementary Figure 3D-ii), respectively, whereas 85-100% of non-senescent cells survived following co-culture.

Finally, to test whether the cytotoxicity of primary NK cells prepared using our protocol was influenced by donor variability, senescent or non-senescent IMR-90 cells were co-cultured with NK cells isolated from multiple donors. A significantly higher cytotoxicity towards senescent compared to non-senescent cells was observed, as measured by LDH assay for each donor (Figure 3D). Furthermore, although the levels of Granzyme B released by activated NK cells correlated with overall cytotoxicity efficiency of individual NK cell donors, NK cells from each donor released similar amounts of Granzyme B upon co-culture with either senescent or non-senescent cells (Supplementary Figure 3E).

Cell culture

Human primary NK cells were maintained at 37° C in humidified air containing 5% CO₂, and complete media containing RPMI-1640 medium (ATCC, USA; Cat#30-2001) supplemented with 20% Fetal Bovine Serum (FBS) (Millipore Sigma, USA; Cat# F4135), 1X Penicillin–Streptomycin (Corning; Cat# 30-001-CI), and 100 IU/ml human rIL-2 (recombinant Interleukin-2) (TECIN teceleukin; Bulk Ro 23-6019). IMR-90 primary lung fibroblasts (ATCC, USA: Cat# CCL-186) were used at population doubling level (PDL) 30-47 and were cultured in an atmosphere with 5% CO₂ and 3% O₂ with complete media containing Dulbecco’s Modified Eagle’s Medium (DMEM) (Corning; Cat# 10-013-CV) supplemented with 10% Fetal Bovine Serum (FBS) (Millipore Sigma, US; Cat# F4135) and 1X Penicillin–Streptomycin (Corning; Cat# 30-001-CI). Cumulative PDL was calculated using the following equation:

$$\text{PDL}=\frac{\text{log}\text{H}-\text{log}\text{S}}{\text{log}\text{2}}$$

where H is the number of cells at harvest and S is the number of cells seeded. Primary human arterial endothelial cells purchased from Coriell Institute for medical research (AG10770) were maintained in promo cell basal medium MV2 (PromoCell; Cat# C-22221) supplemented with Growth Medium MV 2 Supplement Pack (PromoCell; Cat# C-39221) and assayed within 10 or less passages. Endothelial cells were maintained at 37° C in humidified air containing 5% CO₂.

Senescence induction

Senescence was induced in IMR-90 fibroblasts as described before with some modifications [63]. Cells were treated with 300 nM doxorubicin hydrochloride (Millipore Sigma, USA; Cat# 504042) in DMEM complete media for 24 h and maintained in culture as described for 10 days (d). Human endothelial cells were treated with 250 nM doxorubicin in promo cell basal medium MV2 supplemented with Growth Medium MV 2 Supplement Pack for 24 h and maintained in culture as described [63]. For cells induced to senesce by irradiation, IMR-90 cells were treated with ionizing radiation (20 Gy X-ray). For cells induced to senesce by etoposide, fibroblasts were treated with 20 μM of etoposide (Millipore Sigma, USA; Cat# E1383) for 48 h and maintained in culture as described before.

Senescence-associated ß-galactosidase staining

Senescence in IMR-90 cells was determined by senescence-associated ß-galactosidase (SA-ß-gal) activity, as reported [32], using the Senescence Detection Kit (BioVision; Cat# K320), following the manufacturer’s instructions. Fibroblasts were plated (5 x 10⁴ per well) 1 day before treatment in a 6-well cell culture plate (Greiner Bio-One; Cat# 657160) with 2.5 ml of DMEM complete media per well. Non-senescent cells were plated (1 x 10⁵) also in 6-well cell culture plates 1 d before staining. Staining was performed 9 d after doxorubicin treatment. During staining, cells were incubated for 48 h at 37° C in the absence of CO₂, then visualized by bright-field microscopy and imaged. Percentage of SA-ß-gal-positive cells was counted manually.

Real-time quantitative PCR

Non-senescent IMR-90 fibroblasts were plated (1 x 10⁶) in a T-75 cell culture flask (Cellstar; Cat# 658170) with 10 ml of DMEM complete media 2 d before cell pellet harvest. For senescent cells, fibroblasts were plated (1 x 10⁶) 1 d before doxorubicin treatment also in a T-75 cell culture flask, with 10 ml of DMEM complete media. Senescent cells were pelleted 9 d after doxorubicin treatment. All cell pellets were stored at -80° C before RNA isolation. Total RNA was isolated from cell pellets using Quick-RNA MiniPrep (Zymo Research; Cat# R1055), following the manufacturer’s protocol. 1 μg of total RNA per sample was reverse transcribed using PrimeScript RT Master Mix (Takara; Cat# RR036B), and cDNA was analyzed by real-time qPCR using TaqMan Fast Advanced Master Mix (Applied Biosystems; Cat# 4444557) (StepOnePlus™ Real-Time PCR System). Gene expression analyses were performed with Applied Biosystems TaqMan Gene Expression single-tube assays. All reactions were performed in triplicate, and relative expression levels of each gene were normalized to actin. The relative expression of mRNA was determined using the comparative threshold (Ct) method by normalizing target cDNA Ct values to that of actin.

Immunofluorescence (IF)

IMR-90 fibroblasts were plated for senescence induction (2 x 10⁴/well) 1 d before doxorubicin treatment in a black 96-well plate with square wells and a flat, clear bottom (Ibidi; Cat# 89626), with 250 μl/well of DMEM complete media. Non-senescent cells were plated (2 x 10^5/well) 2 d before staining. All immunostaining was performed 8 d after doxorubicin treatment. Cells were fixed with 200 μl/well of 4% paraformaldehyde in 1X PBS (Thermo Scientific; Cat# AAJ19943K2) for 15 min at room temperature, carefully rinsed with 1X PBS (Corning; Cat# 21-031-CV), then permeabilized with 300 μl per well of 0.5% Triton X-100 for 10 min at room temperature. Cells were rinsed once with 1X PBS and incubated with 250 μl per well of anti-γH2AX [p Ser139] (Novus Biologicals; Cat# NB100-74435) and anti-HMGB1 (abcam; Cat# ab18256) antibodies diluted in 5% BSA (Research Products International; Cat# A30075) in 1X PBS overnight at 4° C. Subsequently, cells were washed 5 times with 1X PBS and incubated with 250 μl/well of both Alexa Fluor 488 goat anti-mouse antibody (Invitrogen; Cat# A11029) and Alexa Fluor 546 goat anti-rabbit antibody (Invitrogen; Cat# A11010) along with Hoechst 33342, trihydrochloride, trihydrate (Invitrogen; Cat# H3570) in 5% BSA for 20 min at room temperature in the dark. Cells were washed 5 times with 1X PBS, then 200 μl of 1X PBS was added to each well before images were acquired with Molecular Devices Image Express Micro (Molecular Devices, San Jose, CA, USA). Seven images were acquired per well, with five wells captured per condition (NS or S). Cells with >2 γ-H2AX foci per nucleus were defined as senescent (S).

NK cell isolation and enrichment

Blood samples were obtained from healthy donors (n = 5, age range 20-42) in heparin coated vacutainers. All subjects provided informed written consent. Inclusion criteria for healthy individuals included those who did not take medication that could impact immunity (e.g., corticosteroids) and had no clinical indication of immunodeficiency. NK cells were isolated from freshly drawn human peripheral blood with RosetteSep Human NK Cell Enrichment Cocktail (Stemcell Technology, USA; Cat# 15065) 3 d before co-culture with target cells. Rosette antibody cocktail was added to each blood sample (30 μl/ml) to negatively select unwanted cells with Tetrameric Antibody Complexes that crosslink non-NK cells in human whole blood to red blood cells (RBCs). Blood samples were then incubated at room temperature for 40 min, diluted with 1X PBS (Corning; Cat# 21-031-CV), and combined with Lymphocyte Separation Medium (Corning; Cat# 25-072-CI) followed by density gradient centrifugation, according to the manufacturer’s instructions. Once isolated, NK cells were enriched and activated for 3 days in RPMI complete media with 100 IU/ml human rIL-2 (recombinant Interleukin-2) (R and D Systems; Cat# 8879-IL-010) before co-culturing with senescent or non-senescent IMR-90 fibroblasts.

NK cell characterization

Peripheral blood mononuclear cells (PBMCs) or enriched NK cells were resuspended to a concentration of 1 x 10⁶ cells/ml in FACS buffer (1% BSA in PBS + 0.01% sodium azide) and aliquoted. Cells were washed 2 times with 1X PBS and resuspended in FACS buffer. Cells were then incubated with APC-conjugated anti-human CD3 antibody (Miltenyi Biotec; Cat# 130-113-135), PE-conjugated anti-human CD16 antibody (Miltenyi Biotec; Cat# 130-113-393) and FITC-conjugated anti-human CD56 antibody (Miltenyi Biotec; Cat# 130-114-549) for 30 min at room temperature in the dark. Then, cells were washed 3 times with FACS buffer and resuspended in 200 μl of FACS buffer. Finally, 500,000 events were collected by flow cytometer (DB Accuri C6). Cell viability was determined by PI staining and live cells were gated for downstream analysis. Lymphocytes were gated based on forward and side scatter followed by gating to CD3- cells as NK cells. Finally, percentages of CD56dim CD16+ and CD56bright CD16- NK cells were determined across multiple donors. The data were acquired for at least 100,000 cells/sample. Data were analyzed using Flowlogic software (Miltenyi Biotech, Germany).

Method for co-culture of NK effector cells with IMR-90 target cells

IMR-90 fibroblasts were seeded (2 x 10⁴/well) onto 24-well cell culture plates (Greiner Bio-One; Cat# 662160) with a 1 ml volume of DMEM complete media 10 d before co-culture. Senescence was induced 9 d before co-culture by treatment with 300 nM of doxorubicin for 24 h. Media was replaced with fresh DMEM complete media every 2-3 d. 3 d before co-culture with NK cells, non-senescent IMR-90 fibroblasts were seeded (1 x 10^4 per well) onto 24-well cell culture plates (Greiner Bio-One; Cat# 662160) with 1 ml DMEM complete media. 24 h before co-culture, each well was replaced with 250 μl/well of DMEM complete media to allow for SASP accumulation from senescent cells. Additional wells containing no target cells were included for media only and NK cell spontaneous lactate dehydrogenase (LDH) release controls. At the start of co-culture, NK cells were added in 250μl/well of RPMI complete media at a target to effector (T:E) ratio of 1:1 for a total of volume of 500 μl per well. Co-cultures were incubated at 37° C in humidified air containing 5% CO₂ for 16 h. Bright-field images were taken after 16 h of co-culture for qualitative analysis of NK cell cytotoxicity.

Cytotoxicity quantification

Cytotoxicity was assessed after 16 h of co-culture incubation by quantifying LDH release levels in target cells using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega, Madison, WI, USA), which quantitatively measures LDH released upon cell lysis. Baseline expression of LDH in culture medium alone, medium with target cells only, and medium with NK cells only were used as negative controls. Fibroblasts treated with 0.2% Triton X-100 served as positive controls. Absorbance was recorded at 490 nm using SpectraMax i3 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA, USA). Average absorbance values for the culture medium background were subtracted from absorbance values for experimental and target cell spontaneous LDH release. The data are presented as an index calculated as:

$$\% Cytotoxicity=\frac{\begin{array}{c}Experimental-Effector \\ Spontaneous\\ -Target Spontaneous\end{array}}{\begin{array}{c}Positive control\\ -Target Spontaneous\end{array}} \times 100$$

Granzyme B release assay

Secretion of granzyme B was determined in the supernatant of NK/IMR90-cell co-cultures after 16 h. Co-culture plates were subject to centrifugation at 400xg for 5 min. Supernatant was collected and used to measure human granzyme B by enzyme-linked immunosorbent assay (ELISA) kit (R and D Systems; DY2906-05), according to the manufacturer’s instructions.

Cell viability assay with Calcein AM

Fibroblasts were carefully rinsed 4 times with 1X PBS (1 ml/well) to wash away effector (NK cells) and dead detached cells. Target cells were stained with 1 μM of Calcein AM (diluted from a 4 mM stock solution in dimethyl sulfoxide (DMSO), Invitrogen; Cat# C3099) in no-serum DMEM for 30 min in a 37° C incubator at a volume of 250 μl/well. Following incubation, fluorescence values were recorded at 530 nm using SpectraMax i3 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA, USA).