Small Interfering RNA-Mediated Knockdown of Notch Ligands in Primary CD4⁺ T Cells and Dendritic Cells Enhances Cytokine Production

The adaptive immune response is directed by the molecular interactions taking place at the interface between APCs and T cells. The ensuing T cell responses are determined by a combination of three main factors: the strength of the Ag-derived TCR signal, the microenvironment (cytokines, hormones, etc.), and the specific APC-derived costimulatory signals received (1). Integration of these diverse signals defines T cell fate, directing them either to effector phenotypes, such as Th1 or Th2 cells (2), or toward a regulatory (T_reg)² phenotype (3). The repertoire of known ligands and receptors present at the APC:T cell interface has increased rapidly over recent years, such that there are now families of both costimulatory and coinhibitory molecules (4). The importance of costimulatory signals in defining the nature (5) and the longevity (6) of the immune response makes these molecules attractive candidates for immunomodulatory therapeutics.

A recent addition to the list of molecules at work during T cell activation are the Notch receptors and their ligands of the Delta and Jagged families, which are expressed widely by different cells of the central and peripheral immune system (7). The Notch receptors, Notch ligands, and downstream signaling mediators form a highly conserved functional pathway which plays a central role in the cellular differentiation and proliferation events which constitute pattern formation (8). Notch signaling is characterized by receptor-ligand interactions at the cell surface which induce proteolytic cleavage events that release the Notch intracellular domain (N-IC) (reviewed in Ref. 9). In the best-characterized mode of Notch signaling, N-IC translocates to the nucleus and binds a CSL family transcription factor (termed CBF1 in humans, RBP-Jκ in mice), converting it from repressor to transcriptional activator through the recruitment of proteins such as Mastermind and CBP/p300 (10, 11). Significant genetic evidence also exists for CSL-independent Notch signaling, although the molecular components of this pathway and its downstream targets remain largely unknown (12).

In addition to its role in embryonic development, it is now clear that Notch’s ability to regulate cell-fate choices extends into the peripheral immune system, where activation of the Notch signaling pathway can profoundly alter cytokine production in both CD4⁺ and CD8⁺ T cells (13). Although the effect of Notch ligation on CD4⁺ T cell cytokine production varies significantly from one report to another, e.g., cf Refs. 14 and 15 , it is clear that the Notch pathway is an important regulator of T cell activity. Specifically, the available data demonstrate that APCs utilize the Notch pathway to instruct T cell differentiation programs. Although some reports have shown that APC-expressed Notch ligands program the development of Ag-specific regulatory T cells (T_regs) of a CD4⁺ (16) or a CD8⁺ (17) phenotype, Amsen et al. (18) have suggested that Notch ligands expressed on APCs actually coordinate the Th1/Th2 differentiation of naive Th cells. Although a consensus model has not yet emerged, the available data clearly demonstrate that Notch-Notch ligand interaction at the APC:T cell interface defines a further costimulatory pathway modulating TCR activation and subsequent T cell differentiation.

RNA interference (RNAi) was originally identified as an endogenous mechanism for posttranscriptional gene silencing in plants and nematodes and is now an established technique for experimental “knockdown” of gene expression to establish function in mammalian cells (19, 20). RNAi has typically been achieved in cultured mammalian cells by transfection of chemically synthesized gene-specific 21 nt small interfering RNAs (siRNAs). Introduction of these molecules leads to reduced expression of specific mRNAs, thus permitting the analysis of a knockdown phenotype without the need for time-consuming gene-targeting studies. In addition, RNAi can be applied to loss-of-function studies of human genes. Importantly, knockdown phenotypes obtained by RNAi are highly comparable to those observed in corresponding gene-targeted mice (21). Recent reports have shown that siRNAs can also be used to analyze gene function in primary immune cells, including T cells (22) and dendritic cells (DCs) (23, 24, 25).

To analyze the role of Notch signaling during the activation of primary immune cells, we used RNAi to knock down the expression of APC- and T cell-expressed Notch ligands. Using this technique, we sought to investigate the endogenous role of Notch ligands within the immune system without using overexpression studies in which expression levels are not easily controlled. We describe synthetic siRNA sequences which efficiently knockdown expression of the human Notch ligands Delta1, Jagged1, and Jagged2 in both human primary T cells and monocyte-derived DCs. We show that in human DCs, knockdown of either Delta1, Jagged1, or Jagged2 by siRNA leads to enhanced CD4⁺ T cell cytokine production in allogeneic MLR. In contrast, Delta1 siRNA leads to enhanced cytokine production by CD4⁺ T cells in response to polyclonal TCR activation. Reciprocally, strong Notch signals delivered to CD4⁺ T cells by a recombinant form of Delta1 caused a dose-dependent inhibition of IFN-γ production. Strikingly, pharmacological blockade of γ secretase activity abrogated Hes-1 transcription but had no effect on cytokine production, implying that Delta1 regulates T cell cytokine production by a CSL-independent mechanism.

PBMC were isolated from freshly obtained buffy coats or phlebotomy samples by centrifugation over Ficoll-Paque (Amersham Biosciences). PBMC were then subjected to positive selection with anti-CD14 magnetic beads on LS-positive selection columns (Miltenyi Biotec) according to the manufacturer’s instructions. CD14⁺ cells were seeded at 3 × 10⁶ cells/well in 6-well tissue culture plates in RPMI 1640 (Dutch modification without l-glutamine) supplemented with 10% v/v FBS, 50 IU/ml penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, and 5 × 10⁻⁶ M 2-ME (Sigma-Aldrich), and further supplemented with 50 ng/ml recombinant human IL-4 and GM-CSF (PeproTech) in a humidified atmosphere at 37°C in 5% CO₂ for 6 days to promote differentiation to a DC phenotype. DCs were matured by the addition of 20 ng/ml TNF-α (PeproTech) 24 h before use.

Unless indicated otherwise, CD4⁺ T cells were prepared directly from CD14-depleted PBMC by positive selection by incubation with anti-CD4 magnetic beads and selection on LS-positive selection columns according to the manufacturer’s instructions (Miltenyi Biotec).

Total RNA was extracted from between 5 × 10⁶ and 1 × 10⁷ cells using the RNeasy system (Qiagen) according to the manufacturer’s instructions. cDNA was synthesized from 1 μg of total RNA using Superscript II reverse transcriptase (Invitrogen Life Technologies) primed with oligo(dT) and Retroscript random decamers (Ambion). Q-PCR was performed using Quanitect SYBR green detection chemistry (Qiagen) with the Roche LightCycler system. Relative expression levels were determined by dividing the calculated expression level for the target gene by the calculated expression level of the 18S rRNA gene. Gene-specific quantification was determined by comparing each cDNA to a known serially diluted standard plasmid containing the appropriate cloned amplicon. Primer sequences used were, 5′–3′: 18S rRNA, forward GTAACCCGTTGAACCCCATT, reverse CCATCCAATCGGTAGTAGCG; hDelta1, forward GGGAATGCAAGTGCAGAGTGG, reverse AGGTGGCTCCATTCTTGCAGG; hJagged1, forward GCTGCCTTTCAGTTTCGCCTGG, reverse GCAGAACTTATTGCAGCCAAAGCC; hJagged2, forward CGAGCGAGTGTCGCATGCCGG, reverse TGTTGCCGTACTGGTCGCAGG; and hHes-1, forward CTCTCTTCCCTCCGGACTCT, reverse AGGCGCAATCCAATATGAAC.

cDNAs encoding all Notch ligands used in this study were derived from human fetal poly(A)⁺ RNA (Stratagene) which was reverse transcribed using the Marathon system (Clontech) to generate double-stranded cDNAs with a high proportion of the 5′ ends of long transcripts represented. Full-length human Delta1 cDNA, minus the termination codon, was PCR amplified using Advantage-GC2 polymerase (Clontech) using primers forward 5′-CAT GGG CAG TCG GTG CGC GCT GG-3′ and reverse 5′-GAC TTC AGT TGC TAT GAC GCA CTC ATC C-3′. Resulting PCR products were TA cloned into pCR2.1 (Invitrogen Life Technologies) and sequenced. Mutations introduced during the PCR process were repaired using the QuickChange site-directed mutagenesis system (Stratagene) and verified by sequencing. To generate a carboxyl-terminal-V5-tagged version of full-length Delta1, the cDNA insert (which did not contain a termination codon) was excised from the pCR2.1 holding vector using the HindIII and EcoRV sites flanking the insert and subcloned into the pCDNA3.1-V5HisB vector using the same enzymes.

Full-length Jagged1 cDNA, minus the termination codon, was PCR amplified using Advantage-2 Polymerase (Clontech) and the primers forward 5′-GCG ATG CGT TCC CCA CGG ACG-3′ and reverse 5′-TAC GAT GTA CTC CAT TCG GTT TAA GC-3′. Resulting PCR products were cloned, sequenced, and mutagenized as described for Delta1. Carboxyl-terminal-V5-tagged Jagged1 cDNA was generated by HindIII/EcoRV ligation into pcDNA3.1-V5HisB as for Delta1.

Jagged2 was PCR amplified from placental cDNA using the Advantage GC-2 polymerase system (Clontech) primers forward 5′-GGG GCA ATG CGG GCG CAG GGC-3′ and reverse 5′-CCC CTA CTC CTT GCC GGC GTA GCG-3′. Amplification products were cloned into the TA cloning vector pCR2.1 (Invitrogen Life Technologies), sequenced, and mutations corrected as described above. The 3′ end of the Jagged2 cDNA was amplified from this construct using PfuUltra polymerase (Stratagene) and the primers forward 5′-AAC GAC TGC CTT CCC GAT CCC TGC-3′ and reverse 5′-GAT CCT CGA GGT ACT CCT TGC CGG CGT AGC GGG C-3′, which removes the endogenous stop codon. The product was digested with NotI, which cuts at an endogenous site within the Jagged2 cDNA positioned 3′ of the forward primer annealing site, and XhoI, which cuts at the end of the reverse primer (shown in bold), and the digestion product was ligated into the NotI and XhoI sites of pcDNA3.1-V5HisA (Invitrogen Life Technologies). The 5′ end of Jagged2 was transplanted from the original Jagged2 clone in pCR2.1 as a HindIII/NotI fragment and reconnected to the 3′ end in the pcDNA3.1-V5HisA vector to generate carboxyl-terminal-V5-tagged Jagged2.

To generate Notch ligand/luciferase fusion constructs, the firefly luciferase open reading frame was PCR amplified from pGL3Basic (Promega) and cloned in frame 3′ of the Notch ligand open reading frame using the XhoI and and XbaI sites within the pcDNA3.1-V5HisA or B vectors. For generating the Delta1 fusion, the forward primer for cloning the luciferase gene was 5′-ACG TCT CGA GTA TGG AAG ACG CCA AAA ACA TAA AGA AAG GCC-3′. For the Jagged1 and Jagged2 fusion, the forward primer was 5′-ACG TCT CGA GAT GGA AGA CGC CAA AAA CAT AAA GAA AGG CC-3′. In all three cases, the reverse primer for the firefly luciferase gene was 5′-ACG TTC TAG AAT TAC ACG GCG ATC TTT CC-3′. Expression of these constructs was confirmed by luciferase activity in transient transfection experiments.

pCMV-Luc was constructed by removing the firefly luciferase gene from pGL3Basic as a HindIII/XbaI fragment and subcloning it into the corresponding restriction sites in the expression vector pcDNA3.1 (Invitrogen).

Notch ligand siRNAs were 21nt long duplexes designed, synthesized, and, unless otherwise stated, were all supplied by Ambion. The following human GenBank accession numbers were used to derive these sequences: Delta1, GI 6180181; Jagged1, GI 4557678; and Jagged2, GI 21704276. Three candidate sequences were tested for each gene in preliminary experiments. The optimal sequences for each gene, selected using various criteria (described in Results) are as follows: for Delta1, sense 5′-GGAGUUCGUCAACAAGAAGtt-3′, antisense 3′-CUUCUUGUUGACGAACUCCtg-5′. The selected Jagged1 siRNA consisted of the sense strand 5′-GGUCCUAUACGUUGCUUGUtt-3′ and the antisense strand 3′-ACAAGCAACGUAUAGGACCtc-5′. The selected Jagged2 siRNA consisted of the sense strand 5′-GGAGUGCAAGGAAGCUGUGtt-3′ and the antisense strand 5′-CACAGCUUCCUUGCACUCCtt-3′ The firefly luciferase siRNA consisted of the sense strand 5′- GAUUAUGUCCGGUUAUGUAtt-3′ and the antisense strand 5′-UACAUAACCGGACAUAAUCtt-3′. Scrambled siRNAs contained the same nucleotide content as the selected siRNAs but in a random sequence, and with no calculated target gene specificity as assessed by BLASTing against all human sequence databases. A preoptimized siRNA targeting the firefly luciferase transcript was supplied by Dharmacon (Lafayette).

CHO cells were transfected with both plasmids and siRNAs using Lipofectamine 2000 (Invitrogen Life Technologies). Cells were seeded in DMEM supplemented with 10% v/v FBS, 50 IU/ml penicillin, 50 μg/ml streptomycin, and 2 mM l-glutamine 24 h before transfection (6 × 10⁵ cells/well for 6-well plates, 5 × 10⁴ cells/well for 24-well plates). Complexes of 100 nM siRNA with Lipofectamine 2000, and plasmid DNA with Lipofectamine 2000 were prepared in optimem, incubated for 20 min at room temperature, and then added to cells. After a 4-h incubation, transfection complexes were removed, replaced with fresh medium, and the cells incubated for another 24 h.

CD4⁺ T cells and CD14⁺DCs were transfected using the Amaxa Nucleofector System. Briefly, 5 × 10⁶ (CD4⁺ T cells) and 2 × 10⁶ cells (CD14⁺DCs), respectively, were resuspended in 100 μl of the appropriate Amaxa solution and transfected with 100 nM (CD4⁺) or 500 nM final concentration of siRNA (CD14⁺DCs) using the manufacturer’s protocols U-14 or U-02, respectively. Cells were immediately transferred to 1 ml of RPMI 1640 (Dutch modification without l-glutamine) supplemented with 10% v/v FBS, 50 IU/ml penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, and 5 × 10⁻⁶ M 2-ME and seeded as required.

An equal volume of Titer-Glo reagent (Promega) was added to cells in culture 24 h posttransfection, lysed for 5 min, and 175-μl aliquots transferred to white 96-well plates and luminescence measured using Top-Count plate reader (Packard Bell). The number of viable cells was determined using a standard curve of luminescence measured for a known number of untransfected cells.

A CHO cell line (CHO-N2) was established which coexpressed a full-length human Notch2 cDNA and a Notch-responsive reporter plasmid consisting of 10 CBF1 binding sites cloned upstream of firefly luciferase in pGL3Basic (Promega) (17). For experiments assaying Notch ligand knockdown by siRNA, 2 × 10⁴ CHO-N2 reporter cells per well were seeded in 96-well plates in DMEM supplemented with 10% v/v FBS, 50 IU/ml penicillin, 50 μg/ml streptomycin, and 2 mM l-glutamine. Briefly, 5 × 10⁴ CHO cells, which had been transiently cotransfected with plasmids expressing full length, V5-tagged Notch ligand expression construct, pRL-TK (Promega), and Notch ligand-specific siRNAs were added to the reporter cells and incubated overnight. Firefly and Renilla luciferase activities were quantified the following day using Dual-Glo assay reagent (Promega) in a Top-Count plate reader. An additional CHO cell line stably transfected with a full-length human Delta1 cDNA was used as a positive assay control by coculturing these cells with CHO-N2 and assessing luciferase activity as described. In experiments assaying the activity of plate-bound recombinant Notch ligand, Delta1-Fc was coated directly onto 96-well plates for 2 h at 37°C, washed with PBS, and 2 × 10⁴ CHO-N2 cells plated. After overnight incubation, luciferase activity was determined as described above.

CHO cells were transfected with full-length V5-tagged Notch ligands, with or without siRNA, lysed, denatured, and protein quantified using the Bradford protein assay. Equal quantities of protein were loaded and electrophoresed on 10% (w/v) SDS-PAGE. Proteins were semidry blotted onto nitrocellulose ECL plus membrane, blocked in 5% nonfat milk in TBS/Tween 20 for 90 min at room temperature, and probed with anti-V5 HRP (Invitrogen Life Technologies) diluted 1/1000 in 5% nonfat milk/TBS/Tween 20. The membrane was incubated with ECL reagents (Amersham Biosciences) and exposed to film.

DCs were cultured for 5 days as described in IL-4 and GM-CSF, and matured by overnight incubation with 20 ng/ml TNF-α. DCs were then transfected as described and rested overnight. They were then washed twice in PBS and resuspended in RPMI 1640 (Dutch modification without l-glutamine) supplemented with 5% (v/v) human serum (Sigma-Aldrich), 50 IU/ml penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, and 5 × 10⁻⁶ M 2-ME. CD4⁺ T cells were prepared as described and resuspended in RPMI 1640 (Dutch modification without l-glutamine) supplemented with 5% (v/v) human serum, 50 IU/ml penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, and 5 × 10⁻⁶ M 2-ME. CD4⁺ T cells were seeded at 4 × 10⁵ cells/well in round-bottom 96-well plates, and transfected DCs were added at varying ratios of CD4⁺:DC ratios as described. Cells were incubated for 5 days, at which point supernatants were harvested and analyzed by ELISA.

The response of CD4⁺ T cells transfected with siRNAs was assessed as follows. Purified CD4⁺ cells (5 × 10⁶) were transfected with 100 nM Notch ligand or scrambled control siRNA by Amaxa nucleofection as described above. Cells were seeded at 1 × 10⁶ cells/well in 1 ml of medium containing 2 μg/ml soluble anti-CD28 (clone CD28.2; BD Biosciences) onto 24-well tissue culture plates that had been precoated overnight at 4°C with 10 μg/ml anti-CD3 (clone UCHT-1; BD Biosciences). Supernatants were harvested 72 h posttransfection and analyzed by ELISA as described below.

To examine the effect of recombinant human Delta1-IgG4Fc (hereafter referred to as Delta1-Fc) on CD4⁺ T cells, 96-well plates were first coated overnight at 4°C with a mixture of capture Abs consisting of anti-mouse IgG2a (1 μg/ml; BD Biosciences) and anti-human IgG4 (hIgG4, 1 μg/ml; BD Biosciences). Plates were washed with PBS and incubated for 2 h at 37°C with a mixture of anti-CD3 (3 μg/ml, clone Hit3a; BD Biosciences) and Delta1-Fc (3, 10, or 30 μg/ml) or hIgG4 control (Sigma-Aldrich). Plates were again washed with PBS and purified CD4⁺ T cells were seeded at 2 × 10⁵/well in the presence of soluble anti-CD28 (2 μg/ml, clone CD28.2; BD Biosciences). The human Delta1-hIgG4Fc fusion protein used in this study was expressed and purified as previously described (26).

IL-2, IFN-γ, IL-5, and IL-10 were measured from culture supernatants using the appropriate OptEIA ELISA kit (BD Biosciences), which contains optimized Ab pairs and standards. The manufacturer’s instructions were followed throughout and end point absorbance was measured using a transmission plate reader.

DCs were transfected as described and stained with PE- or FITC-conjugated mouse anti-human Abs recognizing CD40, CD86 (BD Biosciences), and CD83, HLA-DR (Immunotech) for 30 min at 4°C, in the dark. Cells were washed and resuspended in PBS supplemented with 1% FCS and 0.025% sodium azide and analyzed using a Beckman Coulter XL Flow Cytometer with EXPO32 software.

Purified CD4⁺ cells were resuspended to 1 × 10⁷ cells/ml in PBS supplemented with 5% FCS. CFSE (Molecular Probes) was diluted in PBS to 50 μM solution and 110 μl added to 1-ml cell aliquots, mixed, and incubated for 5 min at 22°C. Cells were washed three times with 10 volumes of PBS supplemented with 5% FCS and centrifuged at 300 × g for 5 min. CFSE-labeled cells were then transfected as described and incubated for 4 days at 37°C in 5% CO₂. Cells were then washed and resuspended in PBS supplemented with 1% FCS and 0.025% sodium azide and analyzed using a Coulter FACS analyzer with EXPO32 software.

To identify which Notch ligands to target with siRNA in the peripheral immune system, we determined the endogenous expression patterns of Notch ligands on primary human CD4⁺ T cells and CD14⁺-derived, TNF-α-matured DCs. Q-PCR analysis showed that CD4⁺ T cells expressed Delta1, Jagged1, and Jagged2 (Fig. 1,A), while the other known Notch ligands Delta3 and Delta4 were not detectable (data not shown), confirming expression profiles that have been reported previously (27). Strikingly, while CD4⁺ T cells expressed approximately similar levels of all three ligands, DCs expressed high levels of Jagged1 and Jagged2 mRNA, but a comparatively small amount of Delta1 message (Fig. 1 B). Again, Delta3 and Delta4 were undetectable. Based on these analyses, we chose to develop siRNAs against Delta1, Jagged1, and Jagged2 for subsequent functional studies in primary immune cells.

To identify and characterize siRNA sequences capable of selectively knocking down Notch ligand expression in primary immune cells, we used three distinct assays. First, we constructed plasmids designed to express a fusion product encoding Delta1, Jagged1, or Jagged2 fused to the firefly luciferase gene. These constructs were transfected into CHO cells along with a characterized luciferase-specific siRNA. Luciferase siRNA significantly reduced activity of the three different fusion constructs as assessed by luminescence readout (data not shown). Having confirmed the integrity of the assay, we then analyzed the effects of three different siRNAs for each Notch ligand. CHO cells were cotransfected with a Notch ligand-firefly luciferase fusion construct (NL-Luc), three individual Notch ligand-specific siRNAs, a scrambled siRNA-negative control, or firefly luciferase siRNA as a positive control for knockdown. For each NL-Luc fusion construct, significant knockdown was achieved with at least one of the siRNAs designed. For each Notch ligand, the siRNA generating the greatest knockdown in this assay was selected and used throughout the rest of this study. Data representing the selected siRNA generating the greatest knockdown of each NL-Luc fusion protein is shown in Fig. 2 A.

The specificity of siRNA for the target Notch ligand was confirmed by cotransfecting CHO cells with a NL-Luc construct, the optimal siRNA for each ligand, corresponding scrambled siRNA-negative control or firefly luciferase siRNA as positive control. When targeting luciferase expression from Delta1-Luc, only Delta1- or luciferase-specific siRNA reduced expression (Fig. 2 B). Similar specificity data were also obtained for the Jagged1 and Jagged2 siRNAs (data not shown).

Second, we confirmed that the preselected siRNAs for Delta1, Jagged1, and Jagged2 reduced Notch ligand protein expression. Constructs expressing V5-tagged full-length Notch ligands were transiently transfected into CHO cells with the optimal siRNA sequence for each ligand and a scrambled siRNA control and was analyzed by Western blot using an anti-V5 Ab (Fig. 2 C). Significant knockdown was observed for each siRNA compared with scrambled control, confirming the results generated with the NL-Luc fusion constructs.

Third and final, we modified a Notch-dependent reporter assay which has been previously described (17) to show that siRNA-mediated knockdown of Notch ligand not only affected protein expression but also led to a reduction in Notch signaling activity. In this assay, a CHO cell line (CHO-N2) constitutively expressing a full-length Notch2 cDNA and a multimerized CBF1-luciferase reporter was used to measure Notch signaling from CHO cells transiently transfected with full-length Delta1, Jagged1, or Jagged2 constructs. This assay confirmed that the optimal siRNA sequences were efficacious in a functional assay, as assessed by a decrease in Notch signaling to the CHO-N2 cells (Fig. 2 D).

Having selected and characterized Notch ligand siRNAs for further study, we next developed a protocol for introducing them into primary immune cells. We chose to use the Amaxa Nucleofection System for transfection (28). Initially, transfection and siRNA-mediated knockdown conditions for CD4⁺ T cells and DCs were optimized by transfecting with a construct expressing firefly luciferase (pCMV-luc) along with varying amounts of either luciferase-specific or scrambled siRNAs. The optimal siRNA concentration for both cell types reduced luciferase expression by ∼70% (Fig. 3, A and C, respectively). All concentrations of siRNA tested led to a significant reduction in luciferase activity. The effect of transfection on cell viability was also determined with no significant difference being observed for cells transfected with pCMV-Luc alone or when in the presence of scrambled or luciferase-specific siRNA (Fig. 3, B and D). The optimal concentration of siRNA to achieve reduction in expression while retaining viability was determined to be 100 nM for CD4⁺ T cells and 500 nM for DCs.

Gene-specific knockdown of Notch ligands in DCs was demonstrated by transfecting matured DCs with Notch ligand-specific or scrambled siRNA. Q-PCR showed that endogenously expressed Notch ligand was reduced by a minimum 50%, as shown in Fig. 4,A. We sought to determine whether reduced Notch ligand expression levels affected the cell surface phenotype and therefore the ability of the DC to signal to and present Ag to CD4⁺ T cells. Matured DCs were transfected with Notch ligand-specific or scrambled siRNA and analyzed by flow cytometry for any phenotypic changes. No difference in phenotype was observed between Notch ligand or scrambled siRNAs for any of a panel of activation markers we examined (CD40, CD83, CD86, and HLA-DR; Fig. 4 B).

Having established the systems and reagents required for Notch ligand siRNA in primary human immune cells, we sought to determine whether a reduction in expression of Notch ligands in DCs affected their T cell activation capabilities, and, if so, whether this effect was ligand specific. Matured DCs were transfected with Notch ligand-specific or scrambled siRNA, MLRs were performed, and cytokine production by allogeneic T cells was measured. Reduction in expression of all three Notch ligands resulted in an increased production of IFN-γ (Fig. 4 C). IL-4 was undetectable in this system, with or without any of the Notch ligand siRNAs (data not shown).

Since Notch signaling has been shown to impact upon cytokine production from peripheral T cells, we next asked whether our selected Notch ligand siRNAs could influence cytokine production from purified polyclonally activated CD4⁺ T cells. Quantitative RT-PCR was used to confirm that expression of endogenous ligand was reduced by siRNA by up to 80% (Fig. 5,A). This effect was specific since knockdown of a particular Notch ligand with its optimal siRNA did not affect the expression of others when analyzed by quantitative RT-PCR (Fig. 5,B). CD4⁺ T cells transfected with Notch ligand-specific or scrambled control siRNA were then activated with anti-CD3/anti-CD28 Abs and analyzed for cytokine expression (Fig. 5,C). Delta1 siRNA selectively increased the production of IFN-γ, IL-2, and IL-5; in contrast, Jagged1 and Jagged2 siRNAs had no significant effects on these cytokine levels. To eliminate the possibility that a difference in the proliferation rate was affecting the cytokine response, CFSE-labeled CD4⁺ cells were transfected with Notch ligand-specific or scrambled siRNAs, incubated for 4 days with anti-CD3/anti-CD28 stimulation and analyzed for proliferation by flow cytometry. Transfected, stimulated cell populations typically progressed through a minimum of three rounds of proliferation by 96 h, and this was not affected by any of the Notch ligand-specific siRNAs compared with their scrambled counterparts (Fig. 5,D). In direct contrast to the Delta1 siRNA, treating CD4⁺ T cells with the γ secretase inhibitor N-[N-(3,5-difluorophenacetyl)-l-alanyl]S-phenylglycine t-butyl ester (DAPT) to block Notch ligand-mediated Notch cleavage had no effect on cytokine production, despite its reduction of the Notch target gene Hes-1 to basal levels as assessed by Q-PCR (Fig. 5 E). Collectively, these data suggest a selective cytokine-suppressing role for Delta1-mediated Notch signaling during T-T interactions and that this suppression may involve a cleavage-independent Notch signal.

Having demonstrated an enhanced cytokine production by CD4⁺ T cells transfected with Delta1 siRNA, we finally assessed the effect of ectopic Delta1-mediated Notch signals on CD4⁺ T cells using recombinant Delta1 fusion protein. The ability of this protein to drive CSL-dependent Notch signals was confirmed using the heterologous CHO-N2 reporter assay we have described previously (17). In this assay, recombinant Delta1-Fc fusion protein was plate bound and the CHO-N2 reporter cell line plated upon the immobilized ligand. Measuring CSL-dependent luciferase activity confirmed a dose response to recombinant Delta1 (Fig. 6,A). This activity was confirmed by Q-PCR quantitation of the Hes-1 message in human CD4⁺ T cells stimulated with anti-CD3 and anti-CD28 and either IgG4 or recombinant Delta1 (Fig. 6,B), confirming the delivery of a functional CSL-dependent Notch signal to CD4⁺ T cells by this form of Notch ligand. Importantly, the ability of Delta1 to activate Hes-1 transcription was completely blocked by the γ secretase inhibitor DAPT. Primary CD4⁺ T cells were then stimulated with anti-CD3/anti-CD28 for 72 h with either IgG4 or recombinant Delta1, and cytokine production was assessed. As shown in Fig. 6 C, recombinant Delta1 signaled a dose-dependent reduction in both IFN-γ and IL-5 production. Thus, although reducing Delta1 expression using siRNA enhances cytokine production by human CD4⁺ T cells, ectopic signaling via recombinant protein leads to an inhibition of cytokine production.

Understanding the signals which regulate the peripheral immune system is an ongoing challenge for immunologists. Defining the costimulatory molecules, the phenotypic outcome of their ligation and the downstream signaling pathways by which they operate promises to provide an enhanced mechanistic description of immune responses and additional therapeutic targets for treating disease states characterized by dysregulated immune function. In this study, we have, for the first time, used RNAi to target the Notch signaling pathway in both primary human CD4⁺ T cells and DCs. Knockdown of Notch ligand gene expression is associated with an enhanced cytokine response, in both polyclonal activation of CD4⁺ T cells and in allogeneic MLR, demonstrating that Notch signaling plays a key regulatory role during T cell activation. Importantly, the enhanced cytokine production observed with Delta1 siRNA is consistent with the reduced cytokine production associated with strong Delta1 signals delivered by a recombinant ligand.

RNAi has previously been used to modulate the immune response by targeting the expression of the cytokines IL-10 (23) or IL-12 (24) in DCs. These studies clearly demonstrated the utility of siRNAs in primary immune cells: knockdown of cytokine function in these studies confirms results obtained either with neutralizing Abs or with cytokine-deficient mice. Our initial justification for targeting the Notch pathway using this technique was to assess the role of endogenous Notch signaling without resorting to either overexpression experiments or pharmacological blockade. A number of studies have assessed Notch signaling in immune cells by blocking Notch cleavage with γ secretase inhibitors (15, 29). By blocking the proteolytic release of N-IC, these compounds block the classical CSL-dependent Notch signaling pathway. Importantly, however, mutant forms of Notch which are resistant to furin cleavage in the secretory pathway, and therefore are resistant to subsequent γ secretase cleavage, are still able to inhibit myogenesis in a ligand-dependent but CSL-independent manner (30). This demonstrates the existence of signaling events that are cleavage independent, and thus it is likely that pharmacological blockade of N-IC production with γ secretase inhibitors will inhibit only a subset of Notch’s functions. With the exception of constitutively activating mutations of Notch receptors which occur in some forms of leukemia (31), all known Notch signaling is dependent on ligand interaction. Therefore, γ secretase inhibitors are also unable to discriminate between the effects of specific Notch ligands. By developing siRNAs specific for Notch ligand knockdown, we have avoided the use of these compounds and can truly assess the effect of a reduction in Notch signaling and specifically attribute this effect to a particular Notch ligand.

Using a broad range of assays, both analytical and functional, we identified siRNA sequences that cause the knockdown of three of the five mammalian Notch ligands, namely, Delta1, Jagged1, and Jagged2. As shown in Fig. 5, A and B, these siRNAs are specific for the Notch ligand against which they were designed and exert no detectable effects upon the expression of the other Notch ligands. By using scrambled controls comprising the same nucleotide composition as the Notch ligand-specific siRNAs throughout the numerous assays in this study, we believe that our results represent a true characterization of Notch ligand knockdown. We have not attempted to rescue the siRNA-mediated knockdown phenotype by overexpression here. Overexpressing Notch ligands in immune cells can lead to dose-dependent effects (G.J.M., unpublished data; Refs. 14 and 32), and therefore we chose not to perform such experiments due to potential problems with interpreting cytokine phenotypes.

Transfection of these siRNAs into DCs led to enhanced IFN-γ production in MLR experiments irrespective of the siRNA used, without exerting any measurable effect on the activation status of the DCs as assessed with a panel of surface activation markers. These data suggest that DCs physiologically use all three of Delta1, Jagged1, and Jagged2 to deliver immunomodulatory signals to engaged T cells and that their presence serves as an endogenous regulator of TCR-mediated cytokine production. Interestingly, in our system, knocking down these three different ligands had identical effects on the production of IFN-γ, the characteristic Th1 cytokine, suggesting that, in this particular activation context, Notch ligands do not differentially promote functional skewing of T cell differentiation. These data contrast directly with a murine study which concludes that Delta family ligands promote Th1 differentiation, while Jagged-derived signals orchestrate STAT6-independent Th2 differentiation by directly regulating transcription of the il4 gene (18). In our system, knocking down expression of endogenous Delta and Jagged ligands by RNAi has a similar effect. The reasons for this discrepancy are unclear, but it is possible that the L cells used as artificial APCs by Amsen et al. (18) are functionally distinct from the naturally occurring DCs used in our study. It is also possible that there are differences in Notch ligand utilization in human and mouse systems. It seems certain that discrepancies of this sort are dependent on the experimental system used and will only be resolved by the generation of the relevant conditionally gene-targeted mice or perhaps further development of in vivo siRNA technologies.

In direct contrast to the data obtained in MLR, transfection of Notch ligand siRNA sequences into primary human CD4⁺ T cells demonstrated that although all three were functional as assessed by mRNA knockdown, only Delta1 knockdown led to significant alterations in the cytokine profile driven by polyclonal activation in an APC-depleted system. This reduction in Delta1 expression resulted in an increase in the production of IL-2, IL-5, and IFN-γ, whereas knockdown of Jagged1 or Jagged2 had no effect. These data were not attributable to reduced proliferation or cell viability, and therefore must reflect the loss of a naturally occurring signal among populations of CD4⁺ T cells as they respond to ligation of the Ag receptor. A number of previous studies have demonstrated that Notch signaling impacts upon TCR signaling, although a consistent model has not yet been agreed upon. Although some studies have shown that Notch activation potentiates TCR signaling by enhancing IL-2Rα (CD25) expression (29), others have suggested that coincident Notch ligation blocks TCR-mediated activation of the central signaling molecule Akt (15). Since Akt is the key mediator of IL-2 and IFN-γ production downstream of CD28 (33), our observation that Delta1 siRNA enhances production of these cytokines might be explained if Delta1/Notch signals serve to naturally limit Akt activity. A lack of effect for Jagged1 and Jagged2, in contrast, suggests that these ligands do not play a similar role in homeostatic interactions which appear to take place between T cells. We have not attempted any experiments to explain why this should be the case, although it is clear that glycosylation of Notch receptors by the three mammalian Fringe genes alters their ability to bind distinct Delta or Jagged ligands (34). Furthermore, mice transgenic for Lunatic Fringe display altered patterns of lymphocyte development (35), suggesting that specific Notch-Notch ligand interactions in the immune system are also under the control of Fringe proteins. It is therefore possible that Fringe activity regulates distinct receptor-ligand interactions in peripheral T cells and that this activity prevents Jagged-dependent signaling between T cells. This remains to be investigated.

Although Delta1 siRNA shows a clear enhancement of IFN-γ production, the γ secretase inhibitor DAPT has no functional effect upon polyclonal CD4⁺ T cell activation either here (Fig. 5,E) or in numerous other assays we have performed with primary immune cells, including MLR (data not shown). Q-PCR analysis of DAPT-treated CD4⁺ T cells, however, clearly shows that γ secretase inhibition reduces expression of the Notch target gene Hes-1 to basal levels (Fig. 5 E), demonstrating the pharmacological efficacy of this compound in blocking endogenous CBF1-dependent Notch signals. In addition, we have tested DAPT in the CHO-N2 reporter system used in this study where it completely inhibits Delta1-mediated activation of a CBF1-dependent reporter construct at concentrations as low as 1 μM (data not shown). In CD4⁺ T cell populations, therefore, our data strongly suggest that only Delta1-dependent Notch signaling is physiologically relevant. Furthermore, the DAPT data imply that these signals must take place through a mechanism that is independent of Notch receptor cleavage, or else DAPT would have shown its effect upon cytokine production. Since Notch cleavage is absolutely required to generate the N-IC fragment to activate CBF1, the implication is that a cleavage-independent pathway must also be a CBF1-independent pathway. The existence of such an alternative pathway originates from an analysis of Drosophila mutants, where the Notch phenotype is stronger than the phenotype for the Drosophila CSL transcription factor Su(H) (12). In conjunction with more recent data demonstrating Su(H)-independent Notch mutant phenotypes (36), there is now clear evidence for the existence of CSL-independent Notch signals. In our hands, Delta1 siRNA treatment leads to enhanced IFN-γ production while DAPT has no effect. These data can only be reconciled if Delta1 signals regulate IFN-γ production by human CD4⁺ T cells in a CBF1-independent manner. In support of the existence of such a pathway in the immune system, we have identified several genes which are regulated by Delta1 in a CSL-independent manner in CD4⁺ T cells (G.A.W., Y.S., and G.J.M, unpublished data).

The suggestion that Notch might influence cytokine production without being cleaved requires some speculation as to a potential mechanism by which this may take place. Although the molecular components of a CSL-independent signaling pathway remain unknown, genetic evidence has provided a compelling case for its existence (reviewed in Ref. 12). Recent data have demonstrated that Notch associates with CD4 and the TCR complex in CD4⁺ T cells (27), a result which lead the authors to hypothesize that Notch/Notch ligand ligation plays a role at the MHC:TCR interface. It is therefore tempting to speculate that Notch receptors and ligands are either naturally associated with, or are recruited to, the MHC-TCR complex in response to Ag. Reducing the levels of endogenous Delta1 in CD4⁺ T cells using siRNA would therefore lead to a reduction in the level of Delta1 available for inclusion in these complexes. Eager et al. (15) have reported that activation of Notch1 with an Ab blocks TCR-mediated phosphorylation of the central signaling molecule Akt/PKB, a key regulator of the signaling pathways regulating the production of cytokines such as IFN-γ (33). This effect is observed in a matter of minutes and therefore is unlikely to be the result of new transcription mediated by N-IC activation of CSL, which, by elimination, implies that this effect is a consequence of a CSL-independent Notch function. It is possible that, by way of its RAM domain and ankyrin repeats, ligand-dependent but cleavage-independent Notch activation directs recruitment of signalasome components following TCR activation, thereby controlling the initiation of downstream signaling cascades. In line with this speculation, it has been reported that Notch-IC can associate with p56^lck (37) and that Deltex, a cytoplasmic binding partner for Notch-IC, binds the multifunctional adaptor molecule Grb2 (38). It is therefore possible that, besides its effects as a “membrane-bound transcription factor” (39), the Notch receptor also serves as an orchestrator of the immune signalasome by either promoting or preventing the recruitment of specific signaling modules.

Our data complement some of the results obtained using CD4⁺ T cells from conditional RBP-Jκ-deficient mice (40). In this study, the authors show that RBP-Jκ-deficient CD4⁺ T cells fail to proliferate in response to APC stimulation, but respond normally to polyclonal anti-CD3/anti-CD28 stimulation. This observation suggests that the Notch signaling pathway activated as a result of APC-T cell interactions is CSL-dependent and is mechanistically distinct from the signal arising from T-T interactions. Although there is currently no data available detailing the response of these mice to either model Ags or pathogens, it appears that CD4⁺ T cells from these mice tend to differentiate toward a Th1 phenotype in the absence of Notch/RBP-Jk signals. This observation, which implies that RBP-Jκ-dependent Notch signals between APCs and T cells exert a homeostatic influence which controls the extent of Th1 differentiation, is consistent with our RNAi data demonstrating that reduction of Notch ligand expression in DCs leads to enhanced T cell IFN-γ production. In agreement with this, activation of Notch1 in naive CD4⁺ T cells with an agonist mAb can block IFN-γ production (15). Mechanistic insight into this process may also be inferred from a recent study using mice which express a dominant-negative form of the Notch transcriptional coactivator Mastermind in T cells, and thus are incapable of CSL-dependent Notch signaling (41). Despite this blockade, these mice mount wild-type responses to the model Th1-provoking parasite Leishmania major, but exhibit defective Th2 responses to intestinal nematodes. Assuming that CSL-independent Notch signals are still functional in these mice, and since we and others have clearly shown that Notch can regulate IFN-γ, it seems likely that the Th1-instructing signal downstream of Notch is not dependent on the classical Notch signaling pathway. Our data using siRNA suggest a model where Notch signaling naturally regulates IFN-γ production by CD4⁺ T cells. Inducible deletion of the murine Delta1 gene in mice has been reported (42): these mice display impaired marginal B cell development, but normal T cell development. Although no data have yet been reported for peripheral immune responses in these mice, our RNAi data suggest that these mice may demonstrate enhanced cytokine production in response to T cell-dependent Ags.

The results presented here provide additional support for the therapeutic potential of the Notch signaling pathway. Previous studies have shown that activation of the Notch pathway can suppress immune responses to either model (43) or transplantation (17) Ags, results which suggest that activating the Notch pathway may lead to Ag-specific tolerance. Our data demonstrates that the Notch pathway is equally amenable to antagonism and that Notch ligand knockdown leads to enhanced cytokine responses. It is tempting to speculate that RNAi-mediated knockdown of Delta1 expression in our experiments leads to a loss of T_reg activity and a subsequent loss of control over the development of effector CD4⁺ T cell responses. If this is the case, then Notch antagonism may be of benefit in cancer, where there is accumulating evidence that human tumors evade immune detection by promoting the development of tumor-specific T_reg activity (44). Blockade of either T_reg development or function by antagonizing Notch might therefore neutralize existing T_reg populations in cancer, permitting normal immune surveillance to recommence. This could be achieved, using sequences described here, by the use of siRNA. Significant recent advances in in vivo delivery technologies may make RNAi a therapeutic reality in the near future (45). Alternatively, Notch blockade has been described using a recombinant protein consisting of the DSL-binding domain of human Notch1 (epidermal growth factor-like repeats 11 and 12) as an antagonist, which presumably functions by sequestering Notch ligands (46).

A complete understanding of the Notch pathway in immune cell function still remains some way off. The data presented in this study show that knocking down Notch ligand expression leads to enhanced cytokine production by CD4⁺ T cells. By using siRNA to reduce the levels of Notch ligands, we conclude that the endogenous role of Notch ligands is to restrain T cell activation. Furthermore, our observations have also led us to the intriguing hypothesis that Delta1/Notch signals occurring between CD4⁺ T cells do not involve the canonical cleavage-dependent CSL-dependent Notch signaling pathway. Understanding the mechanisms by which these processes take place remains an ongoing challenge.

We thank Prof. Maggie Dallman for insightful discussions throughout the course of this study.

B.R.C. declares a commercial interest in Lorantis Ltd., Cambridge, U.K.