Sterile activation of invariant natural killer T cells by ER-stressed antigen-presenting cells

Significance While there is a clear understanding of how invariant NKT (iNKT) cells are activated in foreign infection, it remains unclear how they are activated during sterile inflammation, including cancer, where they have a well-defined role in tumor immunosurveillance. Here we elucidate a mechanism by which iNKT cells are activated through 1) the presentation of self-lipid antigens by endoplasmic reticulum-stressed antigen-presenting cells and 2) enhanced functional avidity driven by actin cytoskeletal remodeling. We further provide evidence that this mechanism of activation is at play in tumor settings. Here we describe a physiological context, relevant to human health and disease, that drives the presentation of immunogenic self-lipids to activate iNKT cells during sterile inflammation.

(clone EP134Y, ab52632) and the matched isotype control (Mouse IgG1κ, ab170190) and the previously described BiP antibody and the matched isotype control (Rabbit IgG, 3900S), were used for immunohistochemistry. LEAF Purified anti-human CD1d (clone 51.1 from Biolegend) and the corresponding isotype control LEAF Purified Mouse IgG2bκ were used in the CD1d blocking assays.

Isolation of CD14 + monocytes and iNKT cells
PBMCs were isolated from leukocyte cones (purchased from the NHS blood and transport unit) as previously described. The PBMCs were incubated with CD14 + beads (Miltenyi) and the purified using LS MACS columns (Miltenyi) following the manufacturer's instructions. Purified CD14 + monocytes were differentiated into MoDCs, pulsed with αGC and co-cultured with the autologous CD14fraction.
From this, iNKT cells were sorted and expanded as previously described.

Isolation of BMDCs
Wild type and CD1d -/-C57BL/6 mice (described below) were sacrificed following regulations. The long bones (femur and tibia) were collected. The bone marrow, flushed using a 19-gauge needle and syringe with complete medium, was passed through a cell strainer to create a single cell suspension. Cells were plated at 2 million cells/well in a 6-well plate. Medium was replenished with fresh GMCSF (20ng/mL) every 2 days for 5-7 days.

THP1 cells overexpressing modified CD1d
THP1 cells overexpressing wild type and tail -/-CD1d were described previously (2). THP1 cells overexpressing GPI-linked CD1d were made by transducing cells with the CD1d lentiviral vector containing the GPI sequence used in the construct as previously described (3).

Flow Cytometry
Flow cytometry samples were run either on the Dako Cytomation CyAn ADP, BD FACSCanto, Fortessa X50, or the Nxt Attune flow cytometers and analyzed by FlowJo software. The gating was on lymphocytes or APCs (based on forward and side scatter), single cells, and live cells, as illustrated in Supplementary Figure 4B. Cells were counted in specific gates using the Nxt Attune software.

ELISA
The ELISA was performed using Costa half area, high binding polystyrene 96 well plates (Corning), and was performed following the manufacturer's instructions.

Western blotting
Cytosolic protein from cell lysates was quantified using a BCA assay (Thermo Scientific). A constant amount of protein (20µg) was loaded into pre-cast gels (4-12 or 10% polyacrylimide) and run at 120V. The western blotting protocol was performed as per manufacturer's instructions.

XBP1 PCR
RNA was extracted from cell lysates using the RNeasy Mini Kit (Qiagen) following the manufacturer's instructions. cDNA was synthesized using the High-Capacity Reverse Transcription kit (Applied Biosystems) following the manufacturer's instructions. XBP1 mRNA was amplified from the cDNA by PCR using the OneTaq 2X Master Mix with Standard Buffer (New England Biolabs) following the manufacturer's instructions. The forward primer was: 5' -TTACGGGAGAAAACTCACGGC -3'. The reverse primer was: 5'-GGGTCCAACTTGTCCAGAATGC-3'. The primers were purchased from Sigma.
Quantitative RT-PCR Assuming 100% efficiency of the reverse transcriptase PCR, 10ng of cDNA in 5µL nuclease free water was loaded per well of a qPCR plate (MicroAMP® Fast Optical 96-well Reaction Plate with Barcode (0.01mL)). TaqMan probes using the FAM reporter system (Applied Biosystems by Life Technology) were diluted 20X into TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific) and 5 L were loaded into each well. The plates were read using the QuantStudio7 (Life Technology).

iNKT-TCR tetramer assay
Biotinylated NKT-TCR monomer was tetramerized with streptavidin conjugated to PECF 594 (PE-Dazzle, Biolegend), using 12µg of streptavidin for 50µg of monomer. The cells were stained using 0.1µL/well for 40 minutes on ice, based on a previously described protocol (16).

CD1d plate bound assay
The CD1d plate bound assay was performed as previously described (77). Lipid fractions were added, in duplicate, in 50uL 50mM citrate-phosphate buffer pH5-6 and incubated overnight.
Generation of IRE1, PERK, ATF6α, ATF4 shRNA knock down THP-1 cells 20,000 wild type THP-1 cells in 100µL were plated in a 48-well plate. 50,000 shRNA-loaded pKLO.1 lentiviral particles (Mission shRNA, Sigma) were added to the cells in 100µL of medium (MOI 2.5). Cells with expanded and maintained in R-10 with 2µg/mL puromycin. Knock down was confirmed by western blot.

Lipid Isolation and Fractionation
Approximately 25X10 6 THP1 WT or PERK KD cells were treated for six hours with thapsigargin 0.03µM or left untreated. The cells were washed and left to incubate overnight, after which they were pelleted and snap frozen in dry ice.
Pellets were taken up in 0.9 mL 1:10 PBS: MilliQ water and subjected to 3 freeze-thaw cycles. Chloroform (1.5 mL) and methanol (1.5 mL) was added and centrifuged 3,000 rpm for 10 minutes to remove insoluble/precipitated protein. on repetition rate as previously described and analyzed by a lab-developed script (6, 7). Lipids were identified by LipidXplorer software (8).

Cell labeling for live cell sFCS diffusion measurements
Antibodies against CD1d were produced and affinity-purified from the hybridoma 51.1.3 that has been previously described (9)

Analysis of sFCS data
Recently, we published a paper on software dedicated for the analysis of sFCS data called FoCuS-scan (10). This software was used to perform autocorrelation analysis. For our experiments, we restricted our autocorrelation analysis for 10-20s. The initial few seconds of the intensity carpets that included substantial photobleaching were cropped-out. The obtained autocorrelation curve was generated based on a previous publication (11). Therefore, all 'τ D ' values were converted to their ln values and the frequencies were normalized to 1 with the maximum counts being 100%. The two ln distributions were then compared using two-tailed unpaired t-test with Welch's correction with p-value being reported with a statistical significance of <0.0001. The ln (transit time, ms) was also fitted with the non-linear regression Gaussian fitting algorithm of the GraphPad Prism 7.
The mean +/-SD and confidence interval (CI) values obtained from the fits were reported thereafter (Table 1).

THP1 CD1d (G-actin citrine) cells under different treatment conditions were used
for FRAP experiments. The cells were prepared by washing them twice in L15 buffer then releasing them slowly in a L15 media containing ethanol-cleaned coverslips. Cells were allowed to attach to the coverslips for ~20 min at 37 °C before proceeding for FRAP data acquisitions. The instrumental settings for FRAP data acquisition were based on a previous publication (12). The data were exported to OriginPro 9.1 (Origin Labs, USA) for the analysis. The fluorescence recovery data was normalized to 1 with the maximum intensity -mean intensity of the first two recorded images before photobleaching -as 100%. The normalized fluorescence recovery curves were then plotted against time using two-component exponential fitting strategy described (12). The half-life of the fluorescence recovery was calculated as previously described (13). Half-life and fraction of populations of species (mobile or immobile) obtained from each curves were then compared using a two-tailed unpaired t-test and using Mann-Whitney test ( *, **, ***, **** being p-values less than 0.05, 0.01, 0.001 and 0.0001 respectively).

Image Analysis : Quantification of Surface Area for Actin distribution
THP1 CD1d (G-actin citrine) cells under different treatment conditions were prepared as above. Thereafter, microscopic images of fluorescent actin were taken. Quantification was performed using Fiji. Quantified area for each cells in different samples were then compared using GraphPad Prism 7 using two-tailed unpaired t-test with Welch's correction (**** is p-value <0.0001).

Spatial Autocorrelation analysis
For spatial autocorrelation analysis, fixed THP1 CD1d cells were prepared as

In vivo
Animal studies were performed with appropriate UK Home Office licenses, with ethical approval from the University of Oxford. C57BL/6 wild-type and CD1d -/-(C57BL/6-Cd1d1 tm1.2Aben /J, JAX stock # 017294) were cared for at the Biomedical Service Unit (John Radcliffe Hospital, Oxford, UK). Intravenous tailvein injections were performed using 500,000 or 1X10 6 ex vivo differentiated bone marrow-derived dendritic cells suspended in 100µL or 150 L PBS, respectively. Spleens and lungs were harvested, processed into a single cell suspension (lungs using 0.75mg/mL DNAse and 0.5mg/mL collagenase at 37°C for 1 hour), and treated with red blood cell lysis buffer before being stained for flow cytometry analysis. Blood was obtained from recipient mice via tail-bleeding let to clot at room temperature for 30 minutes, and spun at 4000g for 15min to separate the serum used in the ELISA.

Statistical Analysis
Statistical analysis was performed where biological replicates N were equal to or greater than three. IFN-γ secretion was assumed to follow a Gaussian distribution. Points represent the mean of technical duplicates for each biological experiment, and the error bars represent standard error around the mean.
Statistical analysis was performed in Graphpad Prism Version 5.0a.