Identification of a Distal Locus Enhancer Element That Controls Cell Type–Specific TNF and LTA Gene Expression in Human T Cells

Key Points

• Epigenetic changes predict cell type–specific TNF/LT locus gene expression.

• Distal element hHS-8 regulates TNF and LTA gene expression in activated T cells.

Abstract

The human TNF/LT locus genes TNF, LTA, and LTB are expressed in a cell type–specific manner. In this study, we show that a highly conserved NFAT binding site within the distal noncoding element hHS-8 coordinately controls TNF and LTA gene expression in human T cells. Upon activation of primary human CD4⁺ T cells, hHS-8 and the TNF and LTA promoters display increased H3K27 acetylation and nuclease sensitivity and coordinate induction of TNF, LTA, and hHS-8 enhancer RNA transcription occurs. Functional analyses using CRISPR/dead(d)Cas9 targeting of the hHS-8-NFAT site in the human T cell line CEM demonstrate significant reduction of TNF and LTA mRNA synthesis and of RNA polymerase II recruitment to their promoters. These studies elucidate how a distal element regulates the inducible cell type–specific gene expression program of the human TNF/LT locus and provide an approach for modulation of TNF and LTA transcription in human disease using CRISPR/dCas9.

Introduction

The TNF/LT locus, spanning ∼12 kb on human chromosome 6, has highly conserved architecture containing the three tightly linked genes encoding TNF (TNF), lymphotoxin (LT)-α (LTA), and LT-β (LTB) (1, 2). Although TNF is expressed in multiple cell types, including T cells and monocytic cells/macrophages, the LTA and LTB genes are primarily expressed in lymphocytes and NK cells (2–4). In this study, we examined how the cell type–specific inducible program of the human TNF/LT locus is regulated.

Although TNF was originally defined as a monokine (5–7), it was subsequently shown to be a major product of activated lymphocytes (8–10). In activated T and B cells, TNF gene expression is dependent on calcineurin and the recruitment of the transcription factor NFAT to its promoter region (11–14). Furthermore, regulation of human TNF gene expression is cell type specific (11, 12, 15–21) and requires the assembly of distinct cell type–specific and stimulation-dependent enhanceosome complexes at the TNF promoter region, depending on cell type (20, 21).

Our previous studies identified a distal enhancer element 9 kb upstream of the murine Tnf mRNA cap site (HHS-9), which underwent chromatin remodeling, bound NFATp, and participated in intrachromosomal interactions with the Tnf promoter in murine T cells upon activation (22). The human TNF/LT locus equivalent of this murine element, which is located 8 kb upstream of the TNF promoter (hHS-8), was found to control TNF gene expression in IFN-γ–primed LPS-stimulated monocytic cells via binding of IRF1 to a cognate hHS-8 site in IFN-γ–stimulated monocytes and macrophages (23).

In this study, we show that hHS-8 is coordinately regulated with TNF and LTA gene expression in activated human T cells via a discrete and highly conserved NFAT binding site. In activated primary human CD4⁺ T cells hHS-8 is remodeled in parallel with the TNF and LTA gene promoters, displaying increased H3K27 acetylation and nuclease sensitivity. Upon T cell activation, hHS-8 recruits NFATp and coordinately transcribes hHS-8 enhancer RNA (eRNA) with TNF and LTA mRNA. Specific targeting of the hHS-8-NFAT binding site (NFATbs) by CRISPR/dead(d)Cas9 in the human T cell line CEM resulted in significant inhibition of TNF and LTA gene expression in activated T cells. By contrast, targeting this site had no effect on TNF gene expression in LPS-activated human monocytic cells. Furthermore, CRISPR/dCas9 targeting of the hHS-8-NFATbs in CEM T cells resulted in reduced RNA polymerase II (Pol II) recruitment to the TNF and LTA promoters, indicating that hHS-8 enhances TNF and LTA transcription by increasing Pol II occupancy at the promoters of these two genes.

These studies elucidate how a long-range distal enhancer element upstream of both TNF and LTA drives cell type–specific gene regulation within the highly conserved architecture of the TNF/LT locus. Furthermore, these studies provide a target for potential CRISPR-based approaches to precisely modulate TNF and LTA gene expression in which expression of their protein products in T cells causes pathology.

Materials and Methods

Cells

For isolation of PBMCs, we obtained unidentified, discarded leukocyte packs from the Boston Children’s Hospital Blood Donor Center. Human PBMCs were isolated by Ficoll-Hypaque (Pharmacia) density gradient centrifugation. Primary CD14⁺ monocytes were isolated by positive selection with the EasySep Human CD14 Positive Selection Kit II and CD4⁺ T cells were isolated by negative selection with the EasySep Human CD4⁺ T Cell Isolation Kit (STEMCELL Technologies). Purities of >95% were routinely obtained for both cell types using these methods. Cells were cultured as described previously (23). THP-1 cells (American Type Culture Collection) and CEM cells (a gift from Dr. J. Lieberman) were maintained in RPMI 1640/10% FBS plus gentamicin (Sigma-Aldrich, St. Louis, MO).

DNase I hypersensitivity analysis

DNase I hypersensitivity analyses were performed as described previously (22, 23). DNA was digested with the restriction enzymes shown in Fig. 1A–C and analyzed by Southern blotting, using radiolabeled ³²P probes described in the figure legend to Fig. 1.

FIGURE 1.

Transcriptional and epigenetic profiling of the TNF/LT locus in primary human CD4⁺ T cells and monocytes. (A) Primary human CD4⁺ T cells were stimulated with P + I, and primary human monocytes were stimulated with LPS for the times indicated (three independent donors). Mean transcript copy numbers per ng of total RNA are shown with pre-mRNA copy numbers indicated by the left y-axis and mRNA copy numbers indicated by the right y-axis. (B) Nuclei from freshly isolated primary human monocytes or primary human CD4⁺ T cells from at least four independent donors were digested with increasing amounts of DNase I and purified DNA was subsequently digested with ScaI (left panel) or BamHI (right panel). A representative donor is shown. (C) After mock stimulation or stimulation with P + I for 1 h, freshly isolated CD4⁺ T cells were analyzed as described in (B). (D) After mock stimulation or stimulation with 1 μg/ml LPS for 1 h, freshly isolated monocytes were analyzed as described in (B). Bands correspond to the restriction enzyme fragments shown in the schematic figure of the TNF/LT locus shown below (D) and are designated by arrows and named based on distance from the TNF TSS: purple (occurring in monocytes and CD4⁺ T cells) and blue (CD4⁺ T cell–specific sites). In (D), the nuclease-sensitive sites found in CD4⁺ T cells are presented in pale blue to denote the sites’ putative location on the blot, even though they are absent in monocytes. (E) Histone H3 occupancy. Primary human CD4⁺ T cells and monocytes were mock-stimulated or stimulated with P + I or LPS as described above, nuclei were isolated and sonicated, and chromatin was precipitated with a rabbit mAb to histone H3. Precipitated DNA was quantitated by TaqMan qPCR assay, using probes and primers shown in Supplemental Table I. H3 occupancy was calculated by dividing H3 signal by chromatin input. Mean values from four donors are shown. TSH2B indicates the somatically silenced TSH2B gene, which served as a control genomic site. Error bars represent SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

Precursor mRNA, mRNA, and eRNA analysis

To evaluate the kinetics of TNF, LTA, and LTB gene expression, primary human CD4⁺ T cells and monocytes were seeded at 1 × 10⁶ cells/ml and incubated overnight in complete medium. CD4⁺ T cells were stimulated with 20 ng/ml PMA and 1 μM ionomycin (PMA + I; both from Calbiochem) or with Dynabeads Human T-Activator CD3/CD28 for T Cell Expansion and Activation (catalog no. 11161D; Thermo Fisher Scientific). Cells were stimulated with the beads at a bead/cell ratio of 1:1.

Monocytes were stimulated with 1 μg/ml LPS (Sigma-Aldrich) for 0, 0.25, 0.5, 1, 2, 4, 8, or 16 h. Prior to cell harvest, RNA synthesis was stopped using 5 μg/ml actinomycin D (Sigma-Aldrich). Cells were immediately diluted 2:1 in ice-cold PBS, pelleted at 300 × g, and total RNA was prepared as previously described (23). For quantitative transcript measurement, precursor mRNA (pre-mRNA) and mRNA TNF, LTA, and LTB amplicons were cloned into the pTOPO vector (Life Technologies), and quantitative PCR (qPCR) was performed using serial dilutions of known plasmid amounts. Slopes for all qPCR primer sets were between −3.21 and −3.53.

To measure transcription at hHS-8, CD4⁺ T cells were prepared and seeded as described above. Prior to P + I stimulation, cells were mock treated or treated with 1 μM cyclosporin A (CsA; Sigma-Aldrich) for 10 min and processed as above. To measure hHS-8 eRNA, cDNA synthesis was performed on 400 ng of total RNA with random hexamers (Integrated DNA Technologies) as described above, and hHS-8 eRNA transcripts were quantitated by qPCR (see Supplemental Table I for primer sequences). To measure TNF, LTA, and LTB transcript levels, cDNA synthesis was performed with oligo deoxythymidine primer, and these mRNAs were normalized against cyclophilin B mRNA levels after qPCR.

Chromatin immunoprecipitation

For analysis of H3K4me1 and H3K27ac enrichment, primary monocytes were left unstimulated or treated with 1 μg/ml LPS for 1 h, primary CD4 T cells were left unstimulated or treated with PMA + I for 1 h, and chromatin immunoprecipitation (ChIP) was performed as described previously (23), using one of the following Abs: rabbit IgG (Diagenode), rabbit anti-histone H3 (Abcam), rabbit anti-H3K4me1 (Abcam), or rabbit anti-H3K27ac (Active Motif). Primers and probes are shown in Supplemental Table I. Analysis of data from each donor was performed as follows: first, percentage of input was calculated for each pulldown. Next, the percentage input value for each pulldown was divided by the percentage input value for total H3 to adjust for H3 occupancy levels.

Because of the high basal levels of total histone H3 in the nucleus, H3 occupancy was measured using a modified ChIP protocol, in which Ab/chromatin ratio was increased to improve the signal. To measure 3xHA-dCas9 recruitment to hHS-8, the modified ChIP protocol was used with sonicated nuclei prepared from mock-stimulated or stimulated CEM/CRISPR-Ctrl, CEM/CRISPR-hHS-8-NFATbs THP-1/CRISPR-Ctrl, and THP-1/CRISPR-hHS-8-NFATbs cells and rabbit anti–HA tag Ab (Cell Signaling Technology) and measured with primers flanking the NFATbs in hHS-8 (Supplemental Table I). The modified ChIP protocol was also followed to measure Pol II loading at the TNF and LTA promoters by qPCR, using the primer sets shown in Supplemental Table I.

Sequence alignment

Nucleotides −6898 to −6668 (relative to the TNF transcription start site [TSS]) from the human TNF/LT locus were aligned with sequences from chimpanzee (Pan troglodytes verus), gorilla (Gorilla gorilla), Bornean orangutan (Pongo pygmaeus), Sumatran orangutan (Pongo abelii), lar gibbon (Hylobates lar), green monkey (Chlorocebus sabaeus), rhesus macaque (Macaca mulatta), and common marmoset (Callithrix jacchus), which were obtained by PCR amplification with Phusion High-Fidelity DNA Polymerase (New England Biolabs) of the hHS-8 region in samples from primate species that had previously been evaluated in the Goldfeld Lab (24, 25); they were compared with the house mouse (Mus musculus) from the National Center for Biotechnology Information database. For Homo sapiens, G. gorilla, P. pygmaeus, and P. abelii, a consensus sequence from three individuals was compiled and used in the alignment, whereas for the remaining species the hHS-8 sequence from one representative animal was used in the alignment.

EMSA

EMSA was performed as described previously (22, 23). A total of 5 × 10⁶ primary human CD4⁺ T cells (or CEM cells) were mock treated or treated with 1 μg/ml CsA for 30 min and then left unstimulated or stimulated with 1 μM ionomycin for 30 min. Nuclear extracts were prepared as described but with 0.025% NP-40 in the membrane lysis buffer, and radiolabeled probe (sequence in Fig. 3B) was mixed with extracts in the absence or presence of anti-NFATp Ab (Thermo Fisher Scientific).

FIGURE 3.

Analysis of hHS-8 in primary human CD4⁺ T cells. (A) Alignment of nonhuman primate and murine sequences with nt −6898 to −6725 upstream of the human TNF TSS. The NFAT and IRF1 binding motifs are noted. (B) Nuclear extracts were prepared from primary human CD4⁺ T cells that were mock stimulated or stimulated with 1 μM ionomycin for 30 min or were pretreated with 1 μM CsA for 15 min as indicated in the presence or absence of an NFATp Ab, as indicated by the arrow. The sequence used for the probe is boxed in the human sequence in (A). (C) Primary human CD4⁺ T cells from three independent donors were mock treated or pretreated with 1 μM CsA for 10 min and then mock stimulated or stimulated with P + I. Cells were harvested at 0, 5, 15, 30, and 60 min poststimulation and hHS-8 eRNA, TNF, LTA, and LTB mRNA levels were measured after normalization to cyclophilin B transcript levels (primers shown in Supplemental Table I). Mean and SEM from three independent donors are presented.

Construction of CRISPR/dCas9 CEM and THP-1 cell lines

A lentiviral vector described previously in detail (23) was modified to encode a specific guide RNA (gRNA), which is under the control of the human U6 Pol III promoter, and a 3xHA-dCas9-2A-E2Crimson-puroR cassette driven by the human EFS promoter. gRNAs were designed using the design tool at http://crispr.mit.edu and were predicted to have very low off-target propensity. Primer sequences used for the generation of each gRNA are shown in Supplemental Table I.

Lentiviruses were prepared using standard second-generation techniques and after transduction by spinoculation were successfully introduced into CEM and THP-1 cells, which were selected with increasing concentrations of puromycin (1–5 μg/ml) for 3 wk. At the time of experimental analysis, cells were seeded at 5 × 10⁵ cells/ml in the absence of puromycin.

Statistics

The unpaired Student t test was used to compare ChIP data between CD4⁺ T cells and monocytes and differences in gene expression in the CRISPR THP-1 and CEM cell lines. The paired Student t test was used to compare changes in mRNA and eRNA transcription in response to stimulation within the same donor monocytes or CD4⁺ T cells. Statistical significance was defined as p < 0.05.

Results

Cell type–specific TNF, LTA, and LTB transcription in activated primary human CD4⁺ T cells and monocytes

To determine the pattern of TNF, LTA, and LTB gene expression in primary human T cells and monocytes, we separated cells from three independent healthy donors and stimulated CD4⁺ T cells with P + I and stimulated monocytes with LPS. As shown in Fig. 1A, prior to stimulation TNF and LTA gene expression were nearly undetectable in resting primary human CD4⁺ T cells but were strongly induced by P + I (Fig. 1A, first and third panel down). By contrast, LTB transcripts were detectable in CD4⁺ T cells from all donors prior to stimulation and declined in response to P + I stimulation (Fig. 1A, bottom panel). We also stimulated CD4⁺ T cells from three independent donors with the TCR ligands anti-CD3 and anti-CD28 and found the same pattern of transcriptional induction of the TNF, LTA, and LTB genes as after P + I stimulation (Supplemental Fig. 1). This was consistent with our previous results demonstrating P + I to be a reliable surrogate for TCR activation in analyses of inducible TNF gene expression in T cells (11, 12, 15–20).

TNF was also rapidly induced in LPS-stimulated monocytes (Fig. 1A, second panel). By contrast, LTA and LTB transcript levels were negligible both prior to and in response to LPS treatment (see inset in Fig. 1A, second panel down from top). Notably, maximal TNF mRNA levels in P + I–stimulated CD4⁺ T cells were ∼10-fold higher compared with maximal TNF mRNA levels in LPS-stimulated monocytes, consistent with previous studies showing that TNF is highly inducible and transcribed as an early-immediate gene in activated T cells (11). We also note that patterns of pre-mRNA synthesis for each of the three genes mirrored what was seen for its corresponding mRNA in all cases, indicating that the differences in mRNA levels seen for TNF, LTA, LTB were due to differences in transcriptional induction.

Distinct cell type–specific chromatin signatures at the TNF/LT locus in primary human CD4⁺ T cells and monocytes

To discover genomic elements controlling T cell–specific expression of TNF/LT locus genes, we mapped chromatin-accessibility signatures across the TNF/LT locus in resting primary human CD4⁺ T cells and resting primary monocytes from four independent donors. We found that resting CD4⁺ T cells possess four TNF/LT locus DNase I–hypersensitive (DH) sites (left panel of Fig. 1B, 1C, marked in blue) that are unique to T cells and absent in monocytes (see schematic figure below blots for expected cleavage patterns for each enzyme used in DH analysis). Three of the four CD4⁺ T cell–specific sites correspond to the LTA promoter region (hHS-3.6, hHS-3.3, and hHS-2.8 kb, left panel of Fig. 1B, 1C), and a fourth site corresponds to the LTB promoter region (hHS+7, right panel of Fig. 1B, 1C). Three other DH sites (marked in purple) were present in both primary CD4⁺ T cells and monocytes. One of these sites maps to the TNF promoter (hHS-0.8) (right panel of Fig. 1B, 1C), one maps to hHS-8 (left panel of Fig. 1B, 1C), and another maps to a site 3′ to TNF that is within the LTB coding region (hHS+5; right panel of Fig. 1B, 1C).

Upon stimulation of CD4⁺ T cells, we observed an increase in nuclease sensitivity at three positions: the TNF promoter region, hHS-8, and at the LTA promoter region. By contrast, we detected decreased nuclease accessibility after stimulation at the LTB promoter region (hHS+7) (Fig. 1C). In LPS-stimulated monocytes, we observed that the TNF promoter and hHS-8 were constitutively open and did not display any change in nuclease sensitivity with LPS stimulation (Fig. 1D). This is consistent with our previous study, which showed constitutive accessibility at the TNF promoter and hHS-8 in THP-1 monocytic cells and primary human macrophages, with an increase in nuclease accessibility at hHS-8 only occurring when IFN-γ treatment preceded LPS stimulation (23). Although the LTB coding region hHS+5 site was constitutively open in monocytes as well as CD4⁺ T cells, the LTA and LTB promoter region hypersensitive sites that were readily detectable in T cells were not present in monocytes, even in response to LPS activation (Fig. 1D).

Using a different method to assess nuclease accessibility, we examined nucleosome eviction by measuring histone H3 occupancy (26) at the TNF/LT locus in primary human CD4⁺ T cells and monocytes from four donors (Fig. 1E). As a control, we evaluated histone H3 occupancy at the TSH2B locus, which is not expressed in somatic tissues (27). Histone H3 levels were particularly low at hHS-8 in both cell types (Fig. 1E), consistent with a potential regulatory role for hHS-8 in both T cells and monocytes. In T cells after stimulation, histone H3 levels declined further at hHS-8 (p < 0.001), and at the promoter regions of both TNF (p = 0.002) and LTA (p < 0.001) and at the LTB coding region hHS+5 (p = 0.009) (Fig. 1E), mirroring the cell type–specific patterns we observed in the DH analyses shown in Fig. 1B–D. No changes in histone H3 occupancy were observed at any site in the TNF/LT locus in primary monocytes following LPS activation, also consistent with what we observed using the nuclease accessibility assays.

Taken together, the strong stimulation-dependent induction of TNF and LTA gene expression and the poststimulation diminution of LTB expression in CD4⁺ T cells corresponded to nuclease accessibility and histone H3 eviction patterns. In monocytes, LPS-induced TNF gene expression was associated with constitutive nuclease sensitivity and low histone H3 levels at both the TNF promoter and hHS-8. Consistent with the lack of transcriptional expression of LTA and LTB in monocytic cells, remodeling of chromatin in the regions of the LTA and LTB genes was not observed in LPS-stimulated monocytes.

H3K4me1 and H3K27ac modifications and coordinate TNF, LTA, and hHS-8 eRNA transcription in activated primary T cells

Strong H3K4me1 enrichment is associated with enhancer regions in both poised and active states (28–30) and is also found at poised/active promoters, particularly in zones upstream of transcription start sites (31). In primary CD4⁺ T cells, we observed markedly higher levels of constitutive H3K4me1 at the TNF, LTA, and LTB promoters and at hHS-8 as compared with what we observed at these TNF/LT locus sites in monocytes (Fig. 2A), consistent with the transcriptional activity of the three genes in primary CD4⁺ T cells versus monocytes. In response to activation, little change occurred in H3K4me1 enrichment at any site in either cell type (Fig. 2A). This suggests that H3K4me1 is a relatively static mark, serving as a signpost of active gene expression or of the potential for transcriptional engagement at the TNF/LT locus once the appropriate cellular stimulus is received (32).

FIGURE 2.

H3K4me1 and H3K27ac enrichment at DH sites across the TNF/LT locus in primary CD4⁺ T cells and hHS-8 eRNA transcription. Primary human CD4⁺ T cells and monocytes were mock stimulated or stimulated with P + I or LPS as described above. After sonication, chromatin was precipitated with rabbit mAbs against (A) H3K4me1 or (B) H3K27ac. Precipitated DNA was analyzed as above. Enrichments of H3K4me1 and H3K27ac were calculated by dividing specific Ab signal by H3 signal in the same sample. (C) hHS-8 eRNA was measured after 0 and 90 min of P + I or anti-CD3/anti-CD28 stimulation of primary human CD4⁺ T cells. Error bars represent SEM. *p < 0.05, **p < 0.01.

H3K27ac marks active promoters and, in combination with H3K4me1, marks active enhancers (28, 29, 33). In response to CD4⁺ T cell activation, H3K27ac increased at hHS-8 (p = 0.027), the LTA promoter (p = 0.023), and at the TNF promoter (Fig. 2B). We note that the dramatic reduction in H3 occupancy at the TNF promoter following T cell activation (see Fig. 1E) resulted in an extremely low qPCR signal for immunoprecipitated H3K27ac, leading to large donor-to-donor variability in calculated H3K27ac/H3 enrichment at the TNF promoter; thus, although an increase in H3K27ac enrichment at this site was evident following stimulation, it did not reach statistical significance (Fig. 2B).

By contrast, H3K27ac enrichment at the LTB promoter region (hHS+7) declined in response to P + I stimulation of CD4⁺ T cells (p = 0.048), mirroring the decrease in nuclease sensitivity and the decline in LTB mRNA levels following P + I activation of CD4⁺ T cells. In monocytes, H3K27ac enrichment was present at both hHS-8 and the TNF promoter prior to LPS activation, especially in comparison with the low levels of basal H3K27ac enrichment at the LTA and LTB promoter regions, which was barely detectable.

Given that activation of human primary CD4⁺ T cells resulted in inducible gene expression of TNF and LTA and that we observed a concordant pattern of increased DNase I accessibility, histone H3 eviction, and H3K27ac enrichment at the TNF and LTA promoter regions and at hHS-8, we next investigated whether hHS-8 eRNA was induced in CD4⁺ T cells from healthy donors stimulated with P + I (n = 3) or anti-CD3/anti-CD28 (n = 6) for 90 min. We observed similar and significant induction of hHS-8 eRNA after both stimuli (Fig. 2C), consistent with our data showing that P + I and anti-CD3/anti-CD28 stimulation equivalently regulate TNF, LTA, and LTB transcription in primary CD4⁺T cells (Supplemental Fig. 1).

The hHS-8-NFAT binding element is highly conserved, and hHS-8 eRNA transcription in T cells is blocked by CsA

TNF gene expression is dependent on the recruitment of the calcium-inducible transcription factor NFAT to its promoter region in T cells and is blocked by the calcineurin inhibitor CsA (11, 12, 17), which blocks NFAT dephosphorylation and its nuclear translocation (34). NFAT has also been implicated in LTA gene expression (35). We previously showed that HSS-9, the murine equivalent of the hHS-8 element, contains an NFATbs that matches a consensus NFATp dimer binding site (5′-GGAAAGTCC-3′) (36). This site binds rNFATp in a quantitative DNase I–footprinting analysis and binds NFATp in vivo in primary murine T cells stimulated with anti-CD3/CD28 in a ChIP assay (22). Furthermore, the 1.2-kb murine HSS-9 element can act as a CsA-sensitive enhancer element in T cells when fused to a minimal Tnf promoter luciferase reporter gene (22). However, in these pre-CRISPR era analyses, we were unable to show the element’s function within its chromosomal context or whether it was required for enhancing TNF expression in T cells.

Examination and comparison of “hHS-8” sequences from humans, great apes, Old and New World monkeys, and mice reveals that the hHS-8-NFATbs is completely conserved (Fig. 3A). Furthermore, using nuclear extracts from human CD4⁺ primary T cells stimulated with ionomycin, we observed binding of NFATp to this site in an EMSA, which was inhibited by an NFATp Ab and blocked by CsA (Fig. 3B) as expected and consistent with our previous data (22).

Within minutes after stimulation of primary human CD4⁺ T cells, we observed hHS-8 eRNA synthesis that was abrogated by CsA pretreatment of the cells (Fig. 3C). Furthermore, blocking hHS-8 eRNA transcription with CsA paralleled CsA’s inhibitory effect on TNF and LTA gene transcription (Fig. 3C). By contrast, LTB transcription, which was diminished in activated T cells, was not impacted by CsA treatment (Fig. 3C), consistent with a previous report (35). Previous studies have shown the link between eRNA synthesis and commissioning of enhancers (37, 38). Our data are consistent with hHS-8 being commissioned by NFATp binding, resulting in eRNA transcription and enhancement of TNF and LTA gene expression in activated human T cells.

Functional interrogation of hHS-8 in vivo with CRISPR/dCas9

To determine the function of hHS-8 in the coordinate induction of TNF and LTA gene transcription in T cells, we used a dCas9 CRISPR approach to specifically disrupt NFAT recruitment to hHS-8 in the CEM human T cell line. We chose CEM cells because they display the same pattern of CsA-sensitive TNF and LTA gene expression that we observed in primary human CD4⁺ T cells stimulated with P + I (Supplemental Fig. 2A), thus providing an appropriate T cell model system to study NFATp-dependent hHS-8 transcriptional regulation of TNF and LTA.

CEM T cells were then transduced with lentiviral vectors encoding a gRNA that overlaps the hHS-8-NFAT site to bring a dCas9 tagged with HA to this site in hHS-8 (Supplemental Fig. 2B; designated CRISPR-hHS-8-NFATbs). As a control, we transduced CEM T cells with a lentiviral vector encoding a gRNA that does not match any human sequence (designated CRISPR-Ctrl) (see Supplemental Table I for sequences). We also transduced monocytic THP-1 cells as a negative control with lentiviral vectors encoding HA-tagged dCas9 and either the gRNA that overlaps the hHS-8-NFAT site or the gRNA that does not match any human sequence, as described above. We chose to express dCas9 without any linked transcriptional repression domain, such as the KRAB domain, to ensure that any effects seen on TNF and LTA transcription in response to dCas9 targeting to hHS-8 were specific and not due to nonspecific KRAB-dependent heterochromatin spreading (39).

We note that expressing dCas9 with an HA tag allowed us to specifically confirm that the NFAT site in hHS-8 was efficiently occupied and blocked by dCas9 in both CEM T and THP-1 monocytic cells by performing a ChIP assay with an Ab to the HA tag. As shown in Fig. 4A, the NFAT site in hHS-8 was efficiently occupied and blocked by dCas9 in both CEM T cells and in THP-1 monocytic cells by ChIP assay.

FIGURE 4.

Targeting the hHS-8 site with CRISPR/dCas9. (A) ChIP was performed with rabbit preimmune IgG or rabbit anti–HA tag Ab on sonicated chromatin from the CRISPR-Ctrl and CRISPR-hHS-8-NFATbs CEM (left panel) and THP-1 (right panel) cell lines to measure specific recruitment of 3xHA-dCas9 to the NFATbs in hHS-8. Mean and SD of three independent experiments is presented. (B) Impact of 3xHA-dCas9 targeting to the hHS-8-NFATbs on the induction of TNF (far left panel), LTA (middle left panel), and IFN-γ (middle right panel) mRNA in CEM cells and TNF (far right panel) mRNA in THP-1 cells. Mean and SD of three independent experiments is shown. **p < 0.01, ***p < 0.001. ns, not significant. (C) ChIP was performed with rabbit anti–Pol II Ab on sonicated chromatin from the CRISPR-Ctrl and CRISPR-hHS-8-NFATbs CEM cell lines that had been mock stimulated or stimulated with P + I for 1 h to measure Pol II loading at the TNF (left panel) and LTA (right panel) promoter regions. Mean and SD of three independent experiments is shown. *p < 0.05, **p < 0.01.

Given this evidence of specific targeting of the hHS-8-NFAT site, the CEM-CRISPR-hHS-8-NFATbs and control cells were mock stimulated or stimulated with P + I, and TNF and LTA gene expression were evaluated. As shown in Fig. 4B, TNF (p = 0.002) and LTA (p < 0.001) gene expression were significantly reduced in P + I–stimulated CRISPR-hHS-8-NFATbs CEM cells as compared with CRISPR-Ctrl CEM cells. Furthermore, the decrease in TNF and LTA gene expression in the CRISPR-hHS-8-NFATbs cells was not secondary to nonspecific effects, as expression of another NFAT-dependent gene, IFN-γ (40), was not affected in the CRISPR-hHS-8-NFATbs CEM cells (middle right panel in Fig. 4B). As a control for the T cell specificity of the hHS-8-NFATbs we also tested the effect of blocking the hHS-8-NFATbs in THP-1 cells; as expected, there was no impairment in LPS-induced TNF gene expression in hHS-8-NFATbs-deficient THP-1 cells (far right panel in Fig. 4B). Thus, the hHS-8-NFAT site plays a critical and specific role in inducible expression of TNF and LTA gene expression in activated T cells.

Distal enhancers can augment transcriptional activation by promoting Pol II recruitment to a specific promoter (41). To determine whether enhancement of TNF and LTA gene expression by hHS-8 in T cells involved changes in Pol II occupancy at the promoters of these genes, we measured Pol II loading at the TNF and LTA promoters in the CRISPR-hHS-8-NFATbs and CRISPR-Ctrl CEM cells by ChIP. Pol II loading was significantly reduced at both the TNF (p = 0.002) and LTA (p = 0.04) promoters in activated CRISPR-hHS-8-NFATbs CEM T relative to control cells (Fig. 4C). Thus, the TNF/LT locus element hHS-8 acts as a cell type–specific NFAT-dependent long-range enhancer of TNF and LT gene expression in human CEM T cells, and our data suggest that its mechanism of action involves enhancing Pol II recruitment to the TNF and LTA promoters.

Discussion

Our findings in this study shed light on two fundamental questions: 1) what mechanisms regulate differential patterns of expression of the three TNF/LT locus genes in human T cells, and 2) what mechanisms control differential gene expression of the human TNF/LT locus genes in T cells and monocytes?

With respect to the first question, we have discovered that control of TNF, LTA, and LTB transcription in primary human CD4⁺ T cells is linked to specific epigenetic signatures. Following stimulation of primary CD4⁺ T cells, we observed coordinate increases in nuclease accessibility, histone H3 eviction, and H3K27 acetylation at the TNF and LTA promoters and at the distal noncoding element hHS-8. This was accompanied by sharp increases in CsA-sensitive hHS-8 eRNA and TNF and LTA mRNA transcriptional induction. By contrast, we find >50% decline in H3K27ac enrichment at the LTB promoter accompanied by diminished nuclease sensitivity and a decrease in LTB transcription upon activation of human primary CD4⁺ T cells, which is not affected by CsA. Thus, specific patterns of epigenetic remodeling are tightly associated with regulation of mRNA expression of the three TNF/LT locus genes and of hHS-8 eRNA transcription in primary T cells.

When CRISPR/dCas9 was targeted to the hHS-8-NFAT site in the CEM T cell line, induction of TNF and LTA gene expression in activated cells was significantly and specifically decreased. Furthermore, Pol II recruitment to the TNF and LTA promoters was inhibited when NFAT binding to the hHS-8-NFATbs was blocked by dCas9 in CEM T cells. Thus, the inducible expression of TNF and LTA mRNA in activated CEM T cells is enhanced by hHS-8, consistent with the patterns of epigenetic remodeling of hHS-8 in primary CD4⁺ T cells.

With respect to the second question, what mechanisms control cell type–specific gene expression of the TNF/LT locus genes in T cells and myeloid cells, our data show that distinct epigenetic signatures are associated with and predict TNF/LT locus gene regulation in T cells and monocytes. Although all three genes are regulated in activated T cells, only the TNF gene is induced by LPS in primary human monocytes. Consistent with these findings, the TNF promoter in monocytes is constitutively nuclease accessible and exhibits appreciable H3K27ac enrichment prior to monocyte cellular activation, whereas nuclease accessibility and H3K27ac enrichment are extremely low at the promoters of LTA and LTB, and little change occurs following LPS stimulation.

Intriguingly, the location of the distal noncoding enhancer element hHS-8, which is 8 and 4 kb upstream of the TNF and LTA genes, respectively, is consistent with the remarkably conserved architectural order, or microsynteny, of the TNF/LT locus. This gene order of TNF and LTA in tandem and in the same orientation with LTB downstream of TNF and in inverse orientation predates the split from teleost fishes (42–44). The position of hHS-8 and its role in the regulation of both genes suggests that it has been a cis-regulatory constraint (45) factor in the highly conserved architecture of the TNF/LT locus. Furthermore, hHS-8’s enhancer activity upon the TNF and LTA genes and its dependence upon NFAT in T cells is consistent with the speculation that the TNF and LTA genes arose from an ancient gene duplication event (46).

The protein products of LTA and LTB combine to form the heterotrimeric LT-α1/LT-β2 ligand, which binds to and signals through the LT-β receptor (LT-βR) (47), which is highly expressed on monocytes, macrophages, and dendritic cells (48, 49). LT-βR signaling is critical for lymphoid system development and maintenance (4) and has also been associated with type I IFN production (50, 51) and is proposed to play a role in autoimmunity (50, 51). Thus, the tight control and lack of LTA and LTB expression in the monocytes is consistent with limiting potential cell-autonomous signaling in monocytes via the LT-βR, resulting in excessive inflammatory responses.

In the case of activated T cells, our demonstration of the decline in LTB mRNA levels as LTA mRNA levels rise would therefore limit at the transcriptional level the formation of the LT-α1/LT-β2 heterotrimer, and it would favor LT-α3 homotrimer formation resulting from excess LT-α protein not incorporated into LT-α1/LT-β2 heterotrimers. We note that LT-α3 homotrimers signal through the TNFR, similar to TNF homotrimers (52, 53). Thus, following T cell activation, the amplification of both TNF and LTA gene expression by hHS-8 could be expected to enhance TNFR signaling via the protein products of both genes, whereas the concomitant decrease of LTB mRNA levels limits the formation of the LT-α1/LT-β2 heterotrimers and their subsequent signaling via the LT-βR.

In activated human T cells, NFATp binds to six sites within the 5′ region of TNF, where it anchors the stimulation-induced TNF enhanceosome, which is critical for induction of TNF gene expression in T cells (11, 12, 20, 22). In this study, we show that blocking access to the highly conserved NFATbs in hHS-8 using a CRISPR/dCas9–based approach in the human CEM T cell line reduced both TNF and LTA gene expression and Pol II recruitment to their promoters upon T cell activation. Taken together, our data indicate that NFAT binding at hHS-8 plays a fundamental role in the formation of a transcriptional hub involving Pol II, TNF, and LTA. Current experiments investigating whether hHS-8 forms T cell–specific intrachromosomal interactions with the TNF and LTA regulatory regions and other potential regulatory elements in the TNF/LT locus are underway.

CRISPR/dCas9, either in the absence or presence of chromatin-modifying domains linked to dCas9, has been used to confirm and/or identify short regions within distal noncoding elements that promote the expression of genes within complex genomic environments (54–58). To our knowledge, our laboratory was one of the first to employ this technology to interrogate the role of a distal enhancer element within its native chromatin context, showing that targeting dCas9-KRAB to an IRF1 binding site within hHS-8 in THP-1 monocytic cells significantly impaired IFN-γ priming of LPS-induced TNF gene expression and eRNA production (23). In this study, we have demonstrated that dCas9, in the absence of any linked repressive or other enzymatic domain is a useful tool to investigate the functional importance of epigenetic modifications that accompany changes in local gene expression. Furthermore, this approach has allowed us to show that the distal enhancer hHS-8 controls cell type–specific TNF/LT locus gene expression via discrete activators, NFAT or IRF-1, which are recruited to closely positioned unique cognate hHS-8 binding sites. These binding events commission hHS-8, resulting in eRNA production and its function as a cell type–specific enhancer element depending on cell type.

Currently, several Food and Drug Administration–approved biological inhibitors that broadly inhibit TNF activity across all cell types are in clinical use for the treatment of diverse autoimmune disorders (59). Although of great clinical value, these agents are also complicated by serious adverse events in some patients because of their broad mode of action (60, 61). The ability to selectively inhibit TNFR signaling by inhibiting TNF in a cell- and/or inducer-specific manner, and to inhibit LT-α expression in T cells that are relevant to specific disease pathologies would be a major advance. We note that a cell type–specific targeting approach at the TNF promoter is not feasible, as DNA sequences in the promoter bind distinct factors, depending on cell type, stimulus type, and the ambient nuclear concentration of each transcription factor. For example, the same 5′ sites that bind NFAT in T cells, bind Ets, Sp1, and IRF proteins in myeloid cells after LPS and IFN-γ stimulation (20, 21, 23). By contrast, the targeting of hHS-8, for example, via its discrete nonoverlapping cognate binding sites for IRF-1 or NFAT, would inhibit expression of hHS-8 eRNA and TNF and LTA in a cell type- and inducer-specific–specific manner.

In conclusion, these studies demonstrate how a long-range enhancer element upstream of both the TNF and LTA genes controls cell type–specific gene expression within the gene-dense TNF/LT locus. They also provide insight into the high degree of microsynteny of the locus throughout evolutionary history. Furthermore, they identify a genomic target for potential CRISPR-based approaches to precisely modulate TNF and LTA gene expression in a T cell type–specific manner.

Disclosures

The authors have no financial conflicts of interest.