Human T cells employ conserved AU‐rich elements to fine‐tune IFN‐γ production

Abstract

Long‐lasting CD8⁺ T cell responses are critical in combatting infections and tumors. The pro‐inflammatory cytokine IFN‐γ is a key effector molecule herein. We recently showed that in murine T cells the production of IFN‐γ is tightly regulated through adenylate uridylate–rich elements (AREs) that are located in the 3′ untranslated region (UTR) of the Ifng mRNA molecule. Loss of AREs resulted in prolonged cytokine production in activated T cells and boosted anti‐tumoral T cell responses. Here, we investigated whether these findings can be translated to primary human T cells. Utilizing CRISPR‐Cas9 technology, we deleted the ARE region from the IFNG 3′ UTR in peripheral blood‐derived human T cells. Loss of AREs stabilized the IFNG mRNA in T cells and supported a higher proportion of IFN‐γ protein‐producing T cells. Importantly, combining MART‐1 T cell receptor engineering with ARE‐Del gene editing showed that this was also true for antigen‐specific activation of T cells. MART‐1‐specific ARE‐Del T cells showed higher percentages of IFN‐γ producing T cells in response to MART‐1 expressing tumor cells. Combined, our study reveals that ARE‐mediated posttranscriptional regulation is conserved between murine and human T cells. Furthermore, generating antigen‐specific ARE‐Del T cells is feasible, a feature that could potentially be used for therapeutical purposes.

The protein production of IFN‐γ in T cells is regulated through AU‐rich elements (AREs) that are present in the 3′ untranslated region of the IFNG mRNA. We show here that the role of AREs to regulate IFN‐γ protein production is conserved in primary human T cells. Genetic removal of AREs by CRISPR‐Cas9 technology increased mRNA stability and resulted in higher protein output upon stimulation.

Introduction

CD8⁺ T cells are critical for immunosurveillance and for the protection against invading pathogens. To do so, they produce effector molecules, including granzymes, chemokines, and cytokines. Interferon γ (IFN‐γ) is a key cytokine for CD8⁺ T cells to exert their effector function [1, 2, 3]. IFN‐γ is a pleiotropic cytokine that modulates angiogenesis, hematopoiesis, myelopoiesis, and immune cell functions [4, 5, 6, 7]. For instance, IFN‐γ can suppress the growth of pathogens through upregulation of antiviral factors [8], and attract myeloid cells such as neutrophils to the site of infection [9, 10]. Furthermore, IFN‐γ sensing potentiates the innate immune response of dendritic cells, macrophages, monocytes, and neutrophils against (intra)cellular pathogens [9, 10, 11, 12, 13]. Indeed, point mutations and deletions in humans in the receptors for IFN‐γ, IFNGR1, and IFNGR2, which often lead to premature stop codons, revealed that IFN‐γ sensing is instrumental to protect the host from infections by Mycobacteria species [14, 15, 16].

IFN‐γ also prevents the development of cancers. Mice that lack the Ifng gene, or the signaling protein downstream of IFNGR1/2, STAT1, spontaneously develop tumors [17, 18]. Furthermore, a high IFN‐γ‐mediated gene signature correlates with clinical response rates to immunotherapy in humans [19, 20]. Conversely, copy number alterations of IFN‐γ pathway genes correlate with a poor response to immunotherapy [21].

The regulation of IFN‐γ production is multilayered. The IFNG locus is only demethylated in effector and memory T cells [22], allowing for locus accessibility and transcription upon T cell activation. While the production of T cell effector molecules has been mainly attributed to changes in transcription and the presence of transcription factors [23, 24, 25, 26, 27], recently, the role of posttranscriptional regulation in T cells has also become appreciated [28, 29, 30, 31, 32, 33]. Posttranscriptional regulation is mediated by sequence elements and structures present in both the 5′ and 3′ untranslated regions (UTRs) of mRNA molecules [34, 35, 36, 37] and by nucleoside modifications, such as adenine methylation [38]. By facilitating the binding of RNA‐binding proteins (RBPs), microRNAs, and long noncoding RNAs, these regulators combined determine the actual protein output of a cell [37].

One of these sequence elements are adenylate uridylate‐rich elements (AREs). AREs are AUUUA pentamers present in multimers in the 3′UTR of mRNA molecules [39, 40]. Interestingly, many cytokine transcripts contain AREs [37, 39]. They function as binding hubs for RBPs and microRNAs [39, 40, 41]. Binding to AREs by these factors mediates mRNA stability, localization, and translation, which orchestrates the protein output [30, 41–44]. By deleting the 3′UTR AREs from cytokine mRNA, the protein production is decoupled from ARE‐mediated posttranscriptional regulation [30, 43, 45]. We recently showed that AREs present in the Ifng 3′UTR dampen antitumoral responses in a murine melanoma model [46]. In fact, removal of AREs from the Ifng locus resulted in prolonged IFN‐γ production in a tumor suppressive microenvironment, which translated into substantially delayed tumor outgrowth and prolonged survival [46].

The 3′UTR of IFN‐γ is highly conserved between mice and men, in particular the region containing the AREs [30]. Therefore, we hypothesized that the regulation of IFN‐γ production is also conserved. To unravel the posttranscriptional regulation of IFN‐γ in primary human T cells, we removed a 160 bp region by CRISPR‐Cas9 technology from the IFNG 3′UTR locus that contained all AREs sequences. Similar to murine Ifng [46], removal of AREs from the human IFNG locus (ARE‐Del) results in increased IFN‐γ production. Combining T cell receptor (TCR) gene transfer with ARE deletion in primary human T cells confirmed increased IFN‐γ production by ARE‐Del T cells in response to antigen expressing tumor cells. The ARE‐mediated regulation of IFN‐γ is thus conserved in human T cells.

Results

Deletion of AREs from the IFNG locus by CRISPR‐Cas9

The human IFNG 3′UTR contains 5 AU‐rich elements (AREs), defined as AUUUA (Fig. 1A, capital underlined sequences). To remove all five ARE sequences within the 3′UTR of the IFNG locus, we designed two CRISPR RNAs (crRNAs) (Fig. 1A, capital bold sequence). As a control, we included non‐targeting crRNAs (control). PBMC‐derived human primary T cells were activated with α‐CD3/α‐CD28 for 3 days prior to nucleofection with 30 μg Cas9 ribonuclear protein (RNP) complexes (Fig. 1B). Using ATTO550‐labeled trans‐activating CRISPR (tracr)RNA allowed us to follow the nucleofection efficiency by flow cytometry. Three days after nucleofection, virtually all CD8⁺ and CD4⁺ T cells were positive for ATTO550 (control 97.8 ± 0.9%, ARE‐Del 98.6 ± 0.4%, Fig. 1C and D). In line with the described cleavage pattern of S. pyogenes Cas9 [47], cleavage occurred between the 17th and 18th base of the designed crRNAs (Supporting Information Fig. 2A). The 160 bp deletion containing the five AREs was efficient, as revealed by both PCR and Sanger sequencing of the genomic IFNG 3′UTR region (Fig. 1E; Supporting Information Fig. 2A). After experiencing cellular stress in the first 24 h, within 72 h, ARE‐Del T and control T cells recovered and were viable (Fig. 1D). Furthermore, ARE‐Del T and control T cells expanded 9.2 ± 1.1 fold and 8.9 ± 1.2 fold expansion within 8 days after removal from α‐CD3/α‐CD28, respectively.

Figure 1.

Generation of primary human ARE‐Del T cells. (A) Sequence of the human IFNG 3′ UTR. Translation stop site is indicated with an asterisk, crRNA sequences are bold capitals, AREs are underlined capitals. (B) Graphical representation of the experimental setup. (C and D) PBMCs were stimulated for 72 h with α‐CD3/α‐CD28 and subsequently nucleofected with Cas9 RNPs with ATTO550‐labeled crRNAs targeting the 3′UTR of IFNG (ARE‐Del), or with non‐targeting control crRNA (Control). (C) Representative flow cytometry histograms of Cas9 RNP‐ATTO550 fluorescence gated on live CD8⁺ T cells (black) and live CD4⁺ T cells (gray). Dotted histograms represent non‐nucleofected T cells. (D) Compiled flow cytometry data on ATTO550 fluorescence (left) and percentage of live cells (right) from n = 5 healthy donors from three independent experiments gated on CD8⁺ and CD4⁺ T cells, respectively. (E) ARE‐deletion was analyzed by genomic PCR and subsequent gel analysis. Representative image of the control PCR performed for each donor in every experiment.

To determine if gene editing affected the IFNG 3′UTR downstream of the ARE region, we extracted RNA from control and ARE‐Del T cells. T cells were first activated for 3 h with α‐CD3/α‐CD28 to increase the expression of IFNG mRNA. After cDNA synthesis, we performed PCR from the end of the IFNG mRNA coding region to the 3′ end of the IFNG 3′UTR, resulting in a 509 bp fragment for the WT and a 349 bp fragment for the ARE‐Del IFNG mRNA (Supporting Information Fig. 2B). The gene editing was thus specific for the ARE region and did not result in a premature transcriptional stop of the IFNG 3′UTR. In conclusion, ARE deletion from the IFNG locus was efficient in primary human T cells.

ARE deletion stabilizes IFNG mRNA and augments protein production in human T cells

We next studied whether ARE deletion from the IFNG locus altered IFNG mRNA levels. In T cells that were cultured for 9 days after nucleofection, IFNG mRNA levels were approximately twofold higher in ARE‐Del T cells compared to control T cells (Fig. 2A). In most donors, IFNG mRNA levels of ARE‐Del T cells were also slightly increased upon 3 h stimulation with α‐CD3/α‐CD28 compared to control T cells (Fig. 2A). TNFA and IL2 mRNA levels remained unaltered, indicating IFNG mRNA‐specific alterations in ARE‐Del T cells (Supporting Information Fig. 3A). To measure the half‐life of IFNG mRNA, we activated control and ARE‐Del T cells for 3 h with α‐CD3/α‐CD28, and then blocked de novo mRNA transcription with actinomycin D [48]. ARE‐deletion resulted in a significant increase of IFNG mRNA stability compared to control T cells (t1/2 > 2 h compared to ∼70 min, Fig. 2B).

Figure 2.

Enhanced IFNG mRNA levels, mRNA stability, and IFN‐γ protein output in ARE‐Del T cells. (A and B) T cells rested 9 days and were left untreated or restimulated for 3 h with α‐CD3/α‐CD28. (A) IFNG mRNA levels as determined by RT‐PCR from n = 6 donors from two independent experiments (paired Student's t‐test; *p < 0.05, ns, not significant). (B) T cells were activated for 3 h with α‐CD3/α‐CD28. Actinomycin D was added to block de novo mRNA transcription, and IFNG mRNA stability was determined following 0, 60, and 120 min of actinomycin D treatment. n = 4 donors from two independent experiments (two‐way ANOVA; *p < 0.05, **p < 0.01). (C and D) Resting T cells were restimulated with α‐CD3/α‐CD28 in the presence of brefeldin A for indicated time points. (C) Representative flow cytometry dot plots of IFN‐γ production gated on live CD8⁺ T cells. (D) Compiled flow cytometry data on IFN‐γ⁺ cells (top row) and IFN‐γ GeoMFI (bottom row) gated on IFN‐γ⁺ CD8⁺ and CD4⁺ T cells from n = 6–13 donors from five to seven independent experiments after 2, 4 and 6 h of stimulation (paired Student's t‐test; *p < 0.05, **p < 0.01, ns, not significant). Brefeldin A was added during the last 2 h of activation. T cells cultured in the absence of stimuli were used as negative control.

To determine whether ARE deletion also modulated the IFN‐γ protein production and/or kinetics, we activated ARE‐Del and control T cells with α‐CD3/α‐CD28. To visualize the IFN‐γ production kinetics, we added brefeldin A for a maximum of the final 2 h of stimulation [49], and measured the production of IFN‐γ protein by intracellular cytokine staining. In line with our previous findings in murine in vitro cultured ARE‐Del T cells [30, 46], we did not detect IFN‐γ protein production in human control or ARE‐Del T cells unless the cells were reactivated with α‐CD3/α‐CD28 (Fig. 2C). Only after 2 h of activation, cytokine production was detectable and peaked at 6 h after stimulation in control and ARE‐Del T cells (Fig. 2C), as previously described [3, 50]. However, already from 2 h of activation onwards, we found increased percentages of IFN‐γ‐producing T cells in ARE‐Del CD8⁺ T cells compared to control T cells (Fig. 2C and D). This increase in cytokine production was specific to IFN‐γ and was not observed for tumor necrosis factor α (TNF‐α) or interleukin 2 (IL‐2, Supporting Information Fig. 3B and C). Even though we did not find overt differences in the IFN‐γ production per cell (Fig. 2D, bottom row), deletion of AREs promoted the IFN‐γ production in both CD8⁺ and in CD4⁺ T cells (Fig. 2D, top row). Thus, ARE deletion in the IFNG 3′UTR increases the stability of IFNG mRNA and augments the production of IFN‐γ protein by human T cells.

Prolonged cytokine production by human ARE‐Del T cells upon removal of stimuli

We next questioned if ARE deletion also modulates IFNG mRNA expression and protein production kinetics upon removal of stimulation. We restimulated ARE‐Del T cells and control T cells with α‐CD3/α‐CD28 for 72 h and transferred the cells to new wells to remove the stimulus. Twenty‐four hours after removing the T cells from the stimulus, control T cells expressed slightly lower IFNG mRNA levels than ARE‐Del T cells, which further dropped after 48 h (Fig. 3A). In contrast, ARE‐Del T cells maintained similarly high IFNG mRNA levels for up to 72 h after removal from stimulation (Fig. 3A).

Figure 3.

Prolonged IFN‐γ production in ARE‐Del T cells in the absence of stimulation. Rested T cells (11 days) were restimulated for 72 h with α‐CD3/α‐CD28 and removed from stimulus for indicated time. (A) IFNG mRNA levels as determined by RT‐PCR from n = 4 donors from one experiment (paired Student's t‐test; *p < 0.05, ns, not significant). (B and C) IFN‐γ protein production was assessed by adding brefeldin A 4 h prior to IFN‐γ production assessment. (B) Representative concatenated flow cytometry dot plots of IFN‐γ production of control and ARE‐Del T cells after removal from stimulus for indicated time points. Separately measured samples containing control or ARE‐Del T cells were gated on live CD8⁺ T cells (as depicted in Supporting Information Fig. 1A), exported and subsequently concatenated into one dot plot. IFN‐γ production is shown on the y‐axis; samples are separated on the x‐axis based on SampleID. Control T cells are depicted in gray, and ARE‐Del T cells are depicted in black. (C) Compiled flow cytometry data on IFN‐γ protein production from n = 7 donors from three independent experiments (ratio paired Student's t‐test; *p < 0.05, **p < 0.01, ns, not significant).

We next measured the protein production after removal from stimuli. The production of TNF‐α and IL‐2 was lost in all T cells at 24 h after removal from stimulation (Supporting Information Fig. 3D). Control T cells and ARE‐Del CD8⁺ T cells also produced less IFN‐γ protein at 24 h (Fig. 3B and C). However, in particular ARE‐Del CD8⁺ T cells showed low but sustained protein production of IFN‐γ for up to 72 h (Fig. 3B and C). This also resulted in higher IFN‐γ production per cell at 24 h after antigen removal, as determined by the GeoMFI of the IFN‐γ producing T cells (Fig. 3C). The increased percentage of IFN‐γ producing cells is also observed in CD4⁺ ARE‐Del T cells, albeit only for the first 24 h (Fig. 3C). Together, these data show that, similar to murine T cells [30], deletion of AREs from the IFNG 3′ UTR results in prolonged IFN‐γ production in human T cells.

TCR‐engineered ARE‐Del T cells are superior IFN‐γ producers in response to tumor cells

Finally, we questioned if and how the deletion of AREs in the IFNG gene affects antigen‐specific T cell responses. To this end, we first tested whether CRISPR‐Cas9‐mediated ARE‐Del deletion could be combined with TCR engineering. We first retrovirally transduced human CD8⁺ T cells with a codon‐optimized MART‐1 TCR that recognizes the HLA‐A*0201 restricted epitope of MART‐1 (aa26‐35) [51]. After 5 days, MART‐1 TCR engineered T cells were restimulated with α‐CD3/α‐CD28 for 72 h prior to nucleofection. ARE deletion was also highly efficient in TCR‐engineered T cells (Supporting Information Fig. 3E). Furthermore, the combined TCR engineering with subsequent nucleofection did not affect the viability or the expansion capacity of ARE‐Del versus control MART‐1 TCR‐engineered T cells (92.3 ± 2.3 vs. 91.3 ± 2.6% of live cells at 48 h after nucleofection, and 12.9 ± 5.2 vs. 13.2 ± 4.2 fold expansion within 7 days, respectively).

To study the production of IFN‐γ in response to target cells, we exposed MART‐1 TCR‐engineered ARE‐Del T cells or control T cells to a MART‐1^hi HLA‐A*0201⁺ melanoma cell line (MART‐1⁺), or to a MART‐1^lo HLA‐A*0201⁻ melanoma cell line (MART‐1⁻) [52, 53, 54]. As expected, MART‐1 TCR‐engineered T cells only produce cytokines when co‐cultured with MART‐1⁺ tumor cells (Fig. 4A). Also when T cell activation depended on cognate peptide‐MHC complex, a significantly higher percentage of MART‐1 TCR‐engineered ARE‐Del CD8⁺ T cells produced IFN‐γ after 6 h of co‐culture with MART‐1⁺ cells than control MART‐1 TCR‐engineered T cells (Fig. 4A and B). Again, the higher percentage of IFN‐γ producing ARE‐Del T cells did not translate into significantly increased cytokine production per cell, as determined by the GeoMFI of the IFN‐γ⁺ T cells (Fig. 4B).

Figure 4.

Enhanced IFN‐γ production by ARE‐Del T cells upon target cell recognition. MART‐1 TCR‐engineered T cells were co‐cultured with MART‐1⁺ and MART‐1⁻ cell lines expressing MART‐1 peptide for indicated time points. Brefeldin A was added 2 h prior to IFN‐γ production assessment. To define IFN‐γ producing T cells, MART‐1 TCR⁺ T cells co‐cultured with MART‐1⁻ cells were used. Control T cells are depicted in gray, and ARE‐Del T cells are depicted in black. (A) Representative concatenated flow cytometry dot plots of IFN‐γ production of control and ARE‐Del T cells after 6, 24, and 48 h of co‐culture with MART‐1⁺ and MART‐1⁻ tumor cells. Separately measured samples containing control or ARE‐Del T cells were gated on live CD8⁺ T MART‐1 TCR⁺ T cells (as depicted in Supporting Information Fig. 1B), exported and subsequently concatenated into one dot plot. IFN‐γ production is shown on the y‐axis; samples are separated on the x‐axis based on SampleID. Control T cells are depicted in gray, and ARE‐Del T cells are depicted in black. (B) Compiled flow cytometry data from n = 4 donors from three independent experiments (paired Student's t‐test; *p < 0.05, ns, not significant).

The percentage of IFN‐γ producing cells was substantially reduced after 24 h of co‐culture in both control and ARE‐Del T cells, yet the percentage of IFN‐γ−producing ARE‐Del MART‐1 TCR‐engineered T cells was higher compared to control T cells (Fig. 4B). We also transferred the T cells to freshly seeded tumor cells for another 24 h. At this time point, that is, 48 h of co‐culture, differences in cytokine production were lost (Fig. 4B). Overall, these finding show that the generation of antigen‐specific ARE‐Del T cells is feasible and that the ARE‐deletion promotes the IFN‐γ production in response to target cells.

Discussion

Here, we show that the deletion of AREs from the IFNG 3′UTR in primary human T cells results in higher IFNG mRNA levels, enhanced numbers of IFN‐γ‐producing T cells, and prolonged IFN‐γ production. These findings are in line with previous findings with murine ARE‐Del T cells [43, 46]. Since exogenously expressed GFP reporter genes containing a full‐length IFNG 3′UTR variant with point mutations in all five AREs showed similar results in human T cells as ARE‐Del T cells we generated here [30], we postulate that the observed enhanced IFN‐γ production is at least in large part driven by the ARE deletion, and not merely a result of shortening the IFNG 3′UTR. Thus, the highly conserved ARE region in the IFNG 3′UTR locus also provides conserved posttranscriptional regulation in T cells.

Intriguingly, even though nonactivated murine memory T cells show leaky cytokine production directly ex vivo [30], this feature was not conserved in in vitro cultured T cells [30]. The lack of leaky cytokine production in the absence of stimulation is also observed in in vitro cultured human ARE‐Del cells. This could either stem from leaky IFN‐γ production levels that are below the detection limit, or from lack of additional signals that are required to off‐set the protein production, and that are only present in vivo.

We also observed differences between murine and human ARE‐Del T cells, which are primarily of quantitative nature. Murine ARE‐Del T cells produce more IFN‐γ per cell than WT T cells [46]. We did not observe this difference in human T cells. The overall mRNA stability in activated human control T cells is also lower than that of murine T cells. While WT murine T cells show a t1/2 > 2 h upon antigen stimulation [30, 33], human control T cells activated with α‐CD3/α‐CD28 only reach a t1/2 ∼ 70 min. This also translates into lower overall IFN‐γ protein output in human T cells, both in terms of responding T cells (60–80% in murine WT T cells versus 30–50% in human T cells upon PMA‐Ionomycin stimulation [49, 50]), and in terms of production kinetics. In fact, even though human and murine T cells initiate the IFN‐γ production at similar time points, human T cells cease to produce IFN‐γ faster. To date, it remains unresolved whether these differences stem from different activation methods in murine and human T cells (MEF cells expressing antigen and B7.1 in murine [46] vs. α‐CD3/α‐CD28 in human T cells), from different T cell responsiveness due to different requirements on nutrients and stimuli during culture conditions, from higher diversity in human CD8⁺ T cell populations [55], or from intrinsic differences between murine and human T cells. Irrespective of the observed quantitative differences, we show here that ARE sequences are essential for the regulation of IFNG mRNA stability and IFN‐γ protein output in both murine [46] and human T cells.

ARE‐mediated regulation of mRNA transcripts depends at least in part on mRNA binding of RBPs [37]. Which ARE‐binding proteins regulate the production of IFN‐γ in activated human T cells remains to be determined. Our laboratory previously reported that the ARE‐binding protein Zfp36l2 blocks the translation of preformed Ifng mRNA in murine memory T cells [30]. Zfp36l2, however, releases Ifng mRNA upon reactivation, allowing T cells to instantaneously produce IFN‐γ from this preformed mRNA [30, 49]. Because Ifng mRNA is released from Zfp36l2 upon murine T cell activation, we consider it unlikely that this RBP drives ARE‐mediated posttranscriptional events in activated human T cells, at least in the early phase of T cell activation. Other possible candidates to modulate the fate of IFNG mRNA are the family members of ZFP36L2, that is, ZFP36 and ZFP36L1 [56]. Their potential contribution in regulating the production of IFN‐γ in human T cells is, however, yet to be experimentally addressed.

Excessive production of IFN‐γ can induce immunopathology [57, 58, 59, 60, 61, 62, 63, 64, 65]. However, the AREs solely fine‐tune posttranscriptional regulation of IFN‐γ production, while leaving the requirement of antigen recognition and transcriptional regulation in murine and human T cells intact ([46], this study). Indeed, in studies with long‐term memory ARE‐Del T cells in a Listeria monocytogenes infection and in a B16 melanoma model, we find the increased and prolonged IFN‐γ production only in the presence of tumor cells or after antigenic challenge [46]. We therefore anticipate only limited side effects of the ARE deletion in T cells.

We here combined retroviral TCR engineering with efficient CRISPR‐Cas9‐mediated genome editing. This experimental setup is a powerful tool to study gene modification in human T cells in an antigen‐specific setting. Furthermore, TCR engineering with CRISPR‐Cas9 technology could also be useful for therapeutic purposes. IFN‐γ‐mediated signaling is key for effective antitumoral T cell responses and immunotherapy efficiency [19, 20, 21, 66]. Currently, several different types of cancer are treated with genetically engineered T cells that express a CAR against leukemic antigens like CD19 [67], or a TCR that is specific for a tumor (neo)antigen [68, 69, 70]. It is therefore tempting to speculate that modifying genetically engineered T cell products with an additional IFNG ARE deletion could help maintaining their potential to specifically produce more IFN‐γ in the tumor environment and thus help increase their antitumoral potential. It is noteworthy that upon CRISPR‐Cas9‐mediated genome editing, TCR‐engineered T cells maintained their capacity to expand, a feature that is critical for generating the large numbers of T cells required for cellular products.

In conclusion, the generation of primary human ARE‐Del T cells revealed that fundamental posttranscriptional mechanisms of IFNG regulation are conserved throughout evolution.

Materials and methods

Human PBMCs and cell culture

Studies with human T cells from anonymized healthy donors were performed in accordance with the Declaration of Helsinki (Seventh Revision, 2013) after written informed consent (Sanquin). Peripheral blood mononuclear cells (PBMCs) were isolated through Lymphoprep density gradient separation (Stemcell Technologies). Cells were used after cryopreservation.

Human T cells were cultured in T cell mixed media (Miltenyi) supplemented with 5% heat‐inactivated human AB0 serum (Sanquin) and fetal bovine serum (FBS; Bodinco), 2 mM l‐glutamine, 20 IU/mL penicillin G sodium salts, 20 μg/mL streptomycin sulfate (all Sigma Aldrich), 100 IU/mL recombinant human (rh) IL‐2 (Proleukin, Novartis), and 10 ng/mL rhIL‐15 (Peprotech), and were cultured in a humidified incubator at 37°C + 5% CO₂. Cells were cultured at a density between 0.5 and 1 × 10⁶ cells/mL. Medium was refreshed every 3 days.

T cell activation

T cells were activated as previously described [50]. Briefly, 24‐well plates were precoated overnight with 0.5 μg/mL rat α‐mouse IgG2a (clone MW1483; Sanquin) at 4°C. Plates were washed once with PBS, coated with 1 μg/mL α−CD3 (clone Hit3a, eBioscience) for a minimum of 2 h at 37°C, and washed once with PBS. 1 × 10⁶ PBMCs/well were added in medium supplemented with 1 μg/mL α‐CD28 (clone CD28.2; eBioscience). Cells were activated for 72 h in a humidified incubator at 37°C and 5% CO₂.

crRNA and sequence primer design

crRNAs and sequence primers were designed using the CRISPR and Primer design tools in Benchling (https://benchling.com, Table 1). Sequences were verified to be specific for the target of interest via BlastN and PrimerBlast (both NCBI).

Table 1.

Primers used in this study

Name	Sequence (5′→3′)
ARE‐Del gRNA 1	GTGAGACAGTCACAGGATAT
ARE‐Del gRNA 2	ATTTAAAATTCAAATATTGC
ARE‐Del PCR F	TCGAGGTCGAAGAGCATCCC
ARE‐Del PCR R	CCCATGGGATCTTGCTTAGGT
IFNG_3′PCR_F	TCGAGGTCGAAGAGCATCCC
IFNG_3′PCR_R	GGAAGCACCAGGCATGAAATC

Genetic modification of human T cells with Cas9‐RNPs

Cas9 RNP production and T cell nucleofection was performed as previously described [71]. Briefly, Alt‐R crRNA and ATTO550‐labeled tracr‐RNA were reconstituted to 100 μM in Nuclease Free Duplex buffer (all Integrated DNA Technologies). As a negative control, non‐targeting negative control crRNA #1 was used (Integrated DNA Technologies). Oligos were mixed at equimolar ratios (i.e., 4.5 μL total crRNA + 4.5 μL transactivating CRISPR RNA) in nuclease‐free PCR tubes and denatured by heating at 95°C for 5 min in a thermocycler. Nucleic acids were cooled down to room temperature prior to mixing them via pipetting with 30 μg TrueCut Cas9 V2 (Invitrogen) to produce Cas9 RNPs. Mixture was incubated at room temperature for at least 10 min prior to nucleofection.

For nucleofection, 1 × 10⁶ T cells/condition were harvested 72 h after α‐CD3/α‐CD28 activation and transferred to DNA Lo‐binding Eppendorf tubes (Eppendorf). Cells were washed once with PBS and supernatant was completely removed. Cells were resuspended in 20 μL P2 Buffer (Lonza), Cas9 RNPs were added, and incubated for 2 min. Cells were then electroporated in 16‐well strips in a 4D Nucleofector X unit (Lonza) with program EH‐100. 100 μL of prewarmed complete medium was added after nucleofection and cells were allowed to recover for 5 min in a humidified incubator at 37°C and 5% CO₂. Cells were transferred to 48‐well plates containing 500 μL prewarmed complete medium and cultured in a humidified incubator at 37°C + 5% CO₂.

CRISPR‐mediated gene editing was tested on genomic DNA. Snapfrozen cell pellets (1 × 10⁶ cells) were incubated overnight at 56°C, while rotating at 850 RPM in 50 μL lysis buffer (50 mM Tris‐HCl, 2.5 mM EDTA, 50 mM KCl, and 0.45% Tergitol at pH 8.0) that was freshly supplemented with 1 μg/mL proteinase K (Roche). After deactivating proteinase K by incubating for 15 min at 95°C, lysed cells were centrifuged for 10 min at 13 000 RPM (20 000g). Supernatant was transferred to a new tube. The IFNG genomic locus was amplified by PCR (30 s at 98°C, (10 s at 98°C, 30 s at 65°C, 30 s at 72°C) × 40, 7 min at 72°C, 2 min at 15°C) with DreamTaq HotStart Green Polymerase (ThermoScientific) and primerpair ARE‐Del_PCR_F and ARE‐Del_PCR_R (Table 1) and subsequently analyzed on 0.8% agarose gel. For Sanger sequencing, PCR product was purified with NucleoSpin Gel and PCR Clean‐up kit (Macherey Nagel) according to manufacturer's protocol (Baseclear).

ARE deletion from IFNG mRNA was determined by total RNA extraction from snapfrozen pellets of 0.2 × 10⁶ cells activated for 3 h with α‐CD3/α‐CD28‐activated T cells using Trizol (Invitrogen). DNA contamination was removed with RNA Clean & Concentrator 5 kit (Zymo Research) according to manufacturer's protocol. cDNA synthesis was performed in the presence or absence of SuperScript III Reverse Transcriptase according to manufacturer's protocol (Invitrogen). The resulting cDNA was amplified by PCR (30 s at 98°C, (10 s at 98°C, 30 s at 62°C, 30 s at 72°C) × 40, 7 min at 72°C, 2 min at 15°C) with DreamTaq HotStart Green Polymerase and primerpair IFNG_3′PCR_F and IFNG_3′PCR_R (Table 1) and subsequently analyzed on a 2% agarose gel.

Generation of MART‐1 TCR expressing T cells

PBMCs were activated for 72 h with α‐CD3/α‐CD28, harvested, and transduced with MART‐1 TCR retrovirus as previously described [51]. Briefly, nontissue culture treated 24‐well plates were coated with 50 μg/mL Retronectin (Takara) overnight at 4°C and washed once with PBS. Subsequently, 300 μL viral supernatant was added per well and was centrifuged for 1 h at 4°C at 4000 rpm (2820g). Note that 1 × 10⁶ freshly activated T cells were added per well, spun for 10 min at 1000 rpm (180g), and cultured at 37°C and 5% CO₂ in a humidified incubator. After 24 h, cells were harvested and cultured in T25 flasks at a concentration of 0.8 × 10⁶ cells/mL for 3 days in complete medium.

Functional assays

T cells were stimulated with 1 μg/mL precoated α‐CD3 and 1 μg/mL soluble α‐CD28 for indicated time points. MART‐1 TCR‐transduced T cells were co‐cultured with HLA‐A*0201⁺ MART1^hi Mel 526 (MART1⁺), or HLA‐A*0201⁻ MART1^lo Mel 888 (MART1⁻) tumor cell lines [52, 53, 54], in a 1:1 effector to target (E:T) ratio for indicated time points. 1 μg/mL brefeldin A (BD Biosciences) was added as indicated. Nonactivated T cells were used as a negative control. All stimulations were performed in T cell mixed medium supplemented with 10% FBS, 2 mM l‐glutamine, 20 U/mL penicillin G sodium salts, and 20 μg/mL streptomycin sulfate.

Flow cytometry and intracellular cytokine staining

For flow cytometric analysis, cells were washed with FACS buffer (PBS, containing 1% FBS and 2 mM EDTA) and labeled with monoclonal antibodies α‐CD4 (clone SK3), α‐CD8 (clone SK1), α‐murine TCR beta (clone H57‐597; all BD Biosciences), α‐IFN‐γ (clone 4S.B3), α‐IL‐2 (clone MQ1‐17H12), and α‐TNF‐α (clone MAb11) (all Biolegend). Near‐IR (Life Technologies) was used to exclude dead cells from analysis. For intracellular cytokine staining, cells were cultured in the presence of 1 μg/mL brefeldin A for indicated time points and were fixed and permeabilized with Cytofix/Cytoperm kit (BD Biosciences) according to manufacturer's protocol. Expression levels were acquired using FACSymphony (BD Biosciences) and data were analyzed using FlowJo (FlowJo LLC, version 10). For gating strategy, see Supporting Information Fig. 1A and B. The guidelines for the use of flow cytometry and cell sorting in immunological studies [72] were followed as noted in sections III.1–5 of ref. [72].

Real‐time PCR analysis

Total RNA was extracted using Trizol (Invitrogen). cDNA synthesis was performed with SuperScript III Reverse Transcriptase (Invitrogen), and real‐time PCR was performed with SYBR Green and a StepOne Plus RT‐PCR system (both Applied Biosystems). Reactions were performed in duplicate or triplicate, and cycle threshold values were normalized to 18S levels. Primer sequences for IFNG, TNF, and IL2 mRNA were previously described [50].

To determine the t_1/2 of IFNG mRNA, T cells were activated for 3 h with indicated stimuli, and then treated with 5 μg/mL actinomycin D (ActD) (Sigma‐Aldrich). Data were analyzed using StepOne Plus software (Applied Biosystems).

Statistical analysis

Results are expressed as mean ± SD when indicated. Statistical analysis between groups was performed with Prism (GraphPad Software, version 8), using paired or ratio‐paired Student's t‐test when comparing two groups as indicated, or two‐way ANOVA with Bonferroni's correction for multiple comparisons when comparing more than two groups. p values < 0.05 were considered to be statistically significant.

Author Contributions

J.J.F.H., B.P., and M.C.W. designed experiments; J.J.F.H., B.P., and A.G. performed experiments; J.J.F.H., B.P., and M.C.W. analyzed data; and J.J.F.H. and M.C.W. wrote the manuscript. M.C.W. supervised the project.

The authors would like to thank Dr. Ton Schumacher (Netherlands Cancer Institute) for providing MART‐1 TCR viral supernatants, Nordin Zandhuis for technical assistance, and the Wolkers’ lab for scientific input and critical reading of the manuscript.

Conflict of interest

The authors declare no financial or commercial conflict of interest.

Abbreviations

ARE

adenylate uridylate‐rich element

crRNA

CRISPR RNA

FBS

fetal bovine serum

RBP

RNA‐binding protein

RNP

ribonuclear protein

UTR

untranslated region

Supporting information

Supporting Information