Research ArticleT CELLS

T cell receptor–triggered nuclear actin network formation drives CD4+ T cell effector functions

See allHide authors and affiliations

Science Immunology  04 Jan 2019:
Vol. 4, Issue 31, eaav1987
DOI: 10.1126/sciimmunol.aav1987

T cells need nuclear F-actin

T cell activation is regulated by numerous mechanisms upon T cell antigen receptor (TCR) engagement, including induction of specific cytokines by transcription factors like NF-κB and NFAT. Tsopoulidis et al. now show that TCR engagement causes rapid nuclear actin polymerization to create a dynamic actin filament network that is critical to CD4+ T cell effector functions. Nuclear actin filament polymerization involves the nuclear Arp2/3 complex that is induced by nuclear Ca2+ and regulated by N-Wasp and NIK. Specific inhibition of nuclear actin filament formation impairs T cell effector responses, including cytokine expression and CD4+ T cell help for antibody production. Together, these data reveal a role for nuclear actin filaments in driving CD4+ T cell effector functions.

Abstract

T cell antigen receptor (TCR) signaling triggers selective cytokine expression to drive T cell proliferation and differentiation required for immune defense and surveillance. The nuclear signaling events responsible for specificity in cytokine gene expression upon T cell activation are largely unknown. Here, we uncover formation of a dynamic actin filament network in the nucleus that regulates cytokine expression for effector functions of CD4+ T lymphocytes. TCR engagement triggers the rapid and transient formation of a nuclear actin filament network via nuclear Arp2/3 complex, induced by elevated nuclear Ca2+ levels and regulated via N-Wasp and NIK. Specific interference with TCR-induced formation of nuclear actin filaments impairs production of effector cytokines and prevents generation of antigen-specific antibodies but does not interfere with immune synapse formation and cell proliferation. Ca2+-regulated actin polymerization in the nucleus allows CD4+ T cells the rapid conversion of TCR signals into effector functions required for T cell help.

INTRODUCTION

Development, proliferation, and immune functions of T lymphocytes are regulated by their activation state (1, 2). In concert with costimulatory receptors such as CD28, T cell activation is primarily governed by engagement of surface-exposed T cell antigen receptor (TCR/CD3) complexes with major histocompatibility complex II (MHC II)–bound peptides on antigen-presenting cells (APCs). T cell activation not only triggers proliferation of already differentiated effector T cells but also drives polarized differentiation and proliferation of naïve T cells required for their development into effector T cells. T cell activation is thus a prerequisite for the ability of CD4+ T cells to provide help to B lymphocytes in mounting antigen-specific humoral immune responses (3). Physiologically, APC–T cell interactions occur in the context of cell-cell contacts referred to as the immunological synapse (IS). IS formation causes a broad range of downstream signaling events, including the formation of dynamic signaling assemblies at the plasma membrane (PM), sequential tyrosine phosphorylation cascades to activate protein kinase C (PKC), and mitogen-activated protein kinase signaling, as well as rapid elevation of intracellular calcium (Ca2+) levels in T cells (46). TCR engagement also triggers the immediate polymerization of cortical actin that is essential for downstream signaling (7). Polymerization of actin monomers into actin filaments requires the activity of cellular actin nucleators (8). On the basis of their polymerization mechanism, actin nucleators segregate in the Arp2/3 complex and proteins of the formin and spire protein families (8). Cortical actin polymerization of T lymphocytes in response to TCR engagement involves the Arp2/3 complex and several formins to facilitate actin nucleation and proper spatial distribution of signaling-competent protein microclusters and to coordinate TCR and integrin signaling (5, 9, 10). Information generated by these PM-associated and cytoplasmic events is transmitted to the nucleus by activation and importation of transcription factors that launch specific transcriptional profiles characteristic of activated T cells (11, 12). TCR target genes induced in response to TCR activation include those encoding essential cytokines such as interleukin 2 (IL-2), tumor necrosis factor–α (TNF-α), and interferon-γ (IFN-γ), which are critical to driving cell differentiation and proliferation (13).

Whereas activation of cardinal transcription factors such as nuclear factor κB (NF-κB) and nuclear factor of activated T cell (NFAT) is essential to trigger cytokine expression, the identity of these transcription factors alone is insufficient to explain the specificity with which individual genes are induced or remain unaffected by T cell activation, suggesting that additional yet unknown regulatory mechanisms exist. Here, we describe that TCR engagement causes a rapid and transient burst of actin polymerization in the nucleus to generate a dynamic filament network in the nucleus of CD4+ T cells. Nuclear actin polymerization in response to TCR engagement is induced by elevated nuclear Ca2+ levels and requires nuclear Arp2/3 complex. Specific interference with TCR-induced formation of nuclear actin filaments is associated with impaired production of effector cytokines ex vivo and reduces the generation of antigen-specific antibodies in vivo. These results identify nuclear actin dynamics as a previously unrecognized effector function of TCR signaling that is critical for T cell help.

RESULTS

TCR signaling induces nuclear actin polymerization

Whereas the roles of the actin cytoskeleton in the cytoplasm of mammalian cells are well studied, relatively little is known about the functions of nuclear actin dynamics. Only recently, nuclear targeting actin filament probes such as lifeact or utrophin and optimized protocols for staining with phalloidin allowed visualizing of nuclear actin filaments in mammalian cells (1417). Dynamic nuclear actin filaments have thus far only been visualized in immortalized cell lines, but their existence and relevance have not been analyzed in immune cells, and the approach to visualize nuclear actin filaments has to be optimized for each cell type. Because studies in serum-stimulated fibroblasts suggested that nuclear actin dynamics may exert important function in signal transduction, we hypothesized that such processes may be involved in TCR signaling in CD4+ T cells. To test this hypothesis, we generated Jurkat and A3.01 CD4+ T cell lines that stably express the nuclear F-actin probe lifeact.GFP (Jurkat NLA and A3.01 NLA) and a Jurkat NLA cell line with additional expression of cytoplasmic lifeact.mCherry (Jurkat NCLA), which were analyzed by live-cell spinning disc microscopy (Fig. 1A). TCR stimulation of Jurkat NCLA cells with surface-bound anti-CD3/CD28 antibodies resulted in characteristic cell spreading accompanied by formation of a dense actin ring structure at the cell periphery (fig. S1, A and B) (18). Cells further displayed a transient network of nuclear actin filaments (Fig. 1A, top; movie S1; and fig. S1B). Nuclear actin filaments formed within seconds of cell contact with the TCR-stimulatory surface and were detectable for short periods of time ranging from 60 s to 8 min (Fig. 1A, bottom). Formation of nuclear actin filaments preceded formation of F-actin rings at the cell periphery (Fig. 1A, bottom). Transient formation of a nuclear F-actin network before polymerization of cortical actin was also observed in primary human CD4+ T cells upon IS formation with Staphylococcus aureus, enterotoxin type B (SEB)–loaded Raji B cells (Fig. 1B and movie S2). The time spans during which actin filaments were detectable in the nucleus and at the IS were not correlated (fig. S1C), suggesting that actin polymerization in the nucleus and cytoplasm is largely uncoupled. Contact with nonstimulatory surfaces such as poly-lysine (polyK) only rarely induced the formation of nuclear F-actin (Fig. 1, C (top) and D, and movie S3). Stimulation with anti-CD3 alone was sufficient to trigger nuclear actin polymerization, but stimulation with anti-CD28 alone or engagement of integrins with fibronectin (FN) had no effect [Fig. 1, C (top) and D, and movie S3]. Nuclear actin polymerization was also induced by TCR stimulation via soluble anti-CD3/28 or anti-CD3 alone [Fig. 1, C (bottom) and E]. Stimulating T cells downstream of TCR engagement at the PM by the PKC activator phorbol 12-myristate 13-acetate (PMA) and the ionophore ionomycin (PMA/Iono) (19) or Iono alone also triggered the formation of nuclear actin filaments [Fig. 1, C (bottom) and E, and movie S3]. The formation of a dynamic network of nuclear actin filaments after PMA/Iono treatment was visualized by nuc.lifeact.GFP in several T cell lines including Jurkat and A3.01 cells (fig. S1, D and E) and in primary human CD4+ T cells (Fig. 1F). In addition, we developed a protocol that allowed the visualization of endogenous nuclear actin filaments in the absence of a nuclear actin filament reporter in fixed cells (Fig. 1G). TCR signaling thus induces the rapid and transient formation of a dynamic nuclear actin filament network in CD4+ T lymphocytes.

Fig. 1 TCR signaling induces nuclear actin polymerization.

(A) Top: Jurkat NCLA cells were placed on TCR-stimulatory coverslips and subjected to live-cell microscopy. Shown are still images at the indicated time after contact with the stimulatory surface (see also movie S1). Bottom: Single-cell analysis of actin dynamics over 10 min after activation. Individual cells were imaged as in (A), and the periods in which polymerized F-actin was detected in the nucleus (green) or at the PM (red) are indicated. p.a., post activation. (B) Primary human CD4+ T cell expressing nuc.lifeact interacting with SEB-pulsed Raji B cells. Top: Representative still images of live-cell movies. Inset: Boxed contact region (cont.) in high contrast (see also movie S2). Bottom: Single-cell analysis of actin dynamics over 50 min after contact with the Raji B cell. (C to E) Nuclear actin polymerization in Jurkat NLA cells in response to surface-bound (C, top, and D) or soluble (C, bottom, and E) stimuli. Shown are representative images (see also movie S3) and relative occurrence of nuclear F-actin filaments (mean and SD from three experiments in which at least 30 cells were analyzed per condition each). Statistical significance relative to the αCD3/28 (D) or PMA/Iono (E) control was assessed by one-sample t test. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. DMSO, dimethyl sulfoxide. Visualization of nuclear F-actin with nuclear lifeact.GFP after (+) or without (−) stimulation with PMA/Iono in primary human CD4+ T cells (F). Visualization of endogenous nuclear actin filaments with Alexa Fluor 488 Phalloidin in A3.01 T cells without stimulation (−) or after stimulation (+) with PMA/Iono (G). Scale bars, 5 μm. Arrowheads indicate examples of nuclear F-actin filaments.

Superresolution microscopy of nuclear F-actin reveals highly connective networks

Staining of endogenous nuclear actin filaments after T cell activation enabled us to visualize this network at high resolution with stimulated emission depletion (STED) microscopy (Fig. 2A). Whereas nuclear actin filaments were undetectable in control A3.01 cells (left), treatment with PMA/Iono (right) induced a highly connective network of nuclear actin filaments. To obtain more insight into the organization of nuclear actin filaments, we applied computational segmentation through machine learning to quantitatively analyze the STED-resolved endogenous nuclear actin network (Fig. 2B). By this analysis, the nuclear actin filament network in A3.01 cells treated with PMA/Iono was indistinguishable from that induced in NIH3T3 fibroblasts by serum stimulation (Fig. 2B) but was clearly distinct from cytoplasmic actin stress fibers (Fig. 2C). Cells displayed nuclear actin filaments of an average length of 494.8 ± 7.856 nm (A3.01) or 523.5 ± 7.013 nm (NIH3T3), respectively (Fig. 2D). Because this indicated high connectivity of the network, individual nuclear actin filaments typically encountered about two apparent junctions with other filaments in both cell types (Fig. 2E). Presumably because of the smaller size of nuclei in A3.01 versus NIH3T3 cells, they contained about 75 ± 9 (A3.01) to 150 ± 17 (NIH3T3) filaments per cell (Fig. 2F), resulting in a network of 37.34 ± 3.90 μm (A3.01) to 78.76 ± 8.35 μm (NIH3T3) in size (Fig. 2G). However, when normalized to the nuclear area, the density of nuclear actin filaments [0.999 ± 0.096 filaments/μm2 (A3.01) versus 0.933 ± 0.076 filaments/μm2 (NIH3T3)] or the size of the filament network [0.922 ± 0.098 μm/μm2 (A3.01) versus 0.792 ± 0.070 μm/μm2 (NIH3T3)] was comparable in both cell types (Fig. 2, H and I). Anisotropy analysis revealed that, in contrast to actin stress fibers in the cytoplasm, nuclear actin filaments did not favor a specific orientation (Fig. 2, J and K). The nuclear actin filament network induced by T cell activation therefore is highly connective and spans the entire nucleus without apparent polarity.

Fig. 2 Comparative analysis of the nuclear F-actin networks in A3.01 T cells and NIH3T3 fibroblasts.

(A) STED microscopy of endogenous nuclear actin filaments with Alexa Fluor 488 and Atto 647N Phalloidin in A3.01 T cells without stimulation (−) or after stimulation (+) with PMA/Iono. (B) Workflow for the quantification of nuclear actin filaments. STED images of endogenous nuclear actin filaments and segmentation based on supervised machine learning for comparative quantification between A3.01 T cells stimulated with PMA/Iono (left) and NIH3T3 (3 T3) cells stimulated by serum (right). (C) Example of NIH3T3 cytoplasmic actin stress fibers. Stimulated A3.01 T cells and NIH3T3 show nuclear filaments that are similar in size (D) and do not differ in the number of junctions per filament (E). ns, not significant. A3.01 show fewer nuclear actin filaments per cell (F) compared with stimulated NIH3T3 cells. The overall size of the nuclear actin filament network (sum of all nuclear actin filaments) is higher in NIH3T3 cells than in A3.01 cells (G), but the number of filaments in the nucleus is comparable (H). When normalized to the nuclear area, the network size between A3.01 and NIH3T3 cells appears to be similar (I). (J and K) Anisotropy of nuclear actin filaments: No orientation is favored by nuclear actin filaments, and all angles in the orientation range are occupied to a similar extent by nuclear actin filaments in stimulated A3.01 (J) and NIH3T3 (K) cells. For comparison, cytoplasmic actin stress fibers (stress) are plotted in (J) and show a characteristic isotropic peak in the range of orientation. Scale bars, 5 μm. Arrowheads indicate examples of nuclear F-actin filaments (B) or stress fibers (C). Statistical significance was determined by unpaired t test (n = 19 A3.01 T and n = 18 NIH3T3 cells quantified from two independent experiments). Only filaments longer than 200 nm are considered in the quantitative analysis. **P ≤ 0.01, ***P ≤ 0.001.

Nuclear calcium signaling drives nuclear actin polymerization in T cells

The rapid onset of nuclear actin filament formation subsequent to TCR engagement and the ability of Iono to induce these filaments led us to investigate the role of Ca2+ signaling in this process. As expected (5, 20), intracellular Ca2+ dyes allowed to visualize a rapid and transient burst of Ca2+ release in the cytoplasm and nucleus upon CD4+ T cells making contact with the TCR-stimulatory surface (Fig. 3, A and B). Ca2+ release coincided or shortly preceded the formation of nuclear F-actin filaments, which was followed by the formation of the circumferential actin ring with a distinct delay (Fig. 3, A and C; movie S4; and fig. S2A). Consistent with a critical role of Ca2+ release in nuclear actin filament formation, blocking Ca2+ release by the addition of the Ca2+ chelator 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetoxymethyl ester (BAPTA-AM) abrogated nuclear actin polymerization but had no effect on actin polymerization into an actin-rich circumferential ring at the PM (Fig. 3, D to G, and fig. S2, B and C). In turn, PMA/Iono alone induced the formation of nuclear actin filaments without triggering major actin polymerization at the PM (Fig. 1G). Pharmacological inhibition of individual steps of Ca2+ signaling downstream of the TCR (21) revealed a critical role for the Ca2+-binding protein calmodulin (CaM) and partial involvement of CaM kinase II (CaMKII). Inhibitors of calcineurin or CaMK kinases (CaMKK) had no effect (Fig. 3H). To distinguish whether cytoplasmic or nuclear Ca2+ served as the trigger of TCR-induced actin polymerization in the nucleus, we expressed CaMBP4, a strictly nuclear inhibitor of CaM that selectively blocks nuclear Ca2+ signaling in T cells upon TCR engagement (20). CaMBP4 efficiently suppressed the formation of nuclear actin filaments but did not affect actin polymerization at the PM (Fig. 3, I and J; see movie S5 for localization of CaMBP4). Formation of actin filaments downstream of TCR engagement is thus triggered by the increase in the nuclear Ca2+ pool.

Fig. 3 Nuclear calcium signaling drives nuclear actin polymerization in T cells.

(A) Live-cell imaging of Ca2+ release (top, X-Rhod-1 AM) and nuclear actin polymerization (bottom) in Jurkat NLA cells upon contact with a TCR-stimulatory surface. Shown are still images at the indicated time after exposure to the stimulatory surface (see also movie S4). (B) Quantification of Ca2+ release triggered upon contact with a TCR-stimulatory surface in cells analyzed as in (A). Depicted are means ± SD of total and nuclear Ca2+ levels from eight cells analyzed. FC, fold change. (C) Single-cell analysis over 20 min after activation. Occurrence of actin polymerization at the PM (red) or the nucleus (green) as well as Ca2+ release (yellow) is shown. (D to G) Visualization of Ca2+ release and nuclear actin polymerization upon PMA/Iono stimulation in A3.01 NLA cells. (D and E) Stills of live-cell imaging movies of Ca2+ release of the same cell before (t = 0 s) and after (t = 44 s) stimulation with PMA/Iono. Dashed circles indicate the position of the nucleus. (F and G) Quantification of Ca2+ release (means ± SD from at least six cells each; intensities normalized to the frame before stimulation). Cells were treated with DMSO (D and F) or the Ca2+ chelator BAPTA-AM (E and G). (H) Relative occurrence of nuclear actin filaments in A3.01 NLA cells after treatment with the indicated stimuli and small-molecule inhibitors of Ca2+ signaling (means ± SD from three experiments with at least 30 cells evaluated each per condition). (I and J) Inhibition of nuclear CaM impairs the formation of nuclear F-actin filaments. Representative micrographs (I) and quantification of occurrence of nuclear or PM F-actin (means ± SD of three experiments with 30 cells evaluated each per condition) (J). A3.01 NLA (nuclear F-actin) or Jurkat NLA (F-actin ring) cells transiently expressing mCherry or the nuclear CaM inhibitor CaMBP4.mCherry were analyzed 58 s or 5 min after activation of a TCR-stimulatory surface, respectively. Asterisks indicate mCherry-positive cells. Statistical significance relative to the PMA/Iono (H) or mCherry (J) control was assessed by one-way analysis of variance (ANOVA) or one-sample t test, respectively. Scale bars, 5 μm. Arrowheads indicate examples of nuclear F-actin filaments.

Nuclear Arp2/3 mediates formation of TCR-induced nuclear actin filaments

We next sought to define the actin polymerization machinery responsible for the formation of TCR-induced nuclear actin filaments. The identification of the relevant actin nucleator in our system was complicated by the fact that many actin nucleators play critical roles for actin rearrangements at the IS (10, 22), and interference with their activity may thus have indirect effects on nuclear actin dynamics when triggered via the TCR. We therefore capitalized on the above finding that PMA/Iono was sufficient to induce nuclear actin polymerization without actin polymerization at the PM. Nuclear actin filament formation in response to PMA/Iono was disrupted by treatment with the actin polymerization inhibitor latrunculin B (fig. S3, A and B) or with inhibitors of the Arp2/3 complex (CK-869 and CK-666) or all formins (SMIFH2) (fig. S3, C and D), suggesting that actin filaments detected in the nucleus did not result from bundling of preexisting filaments but were formed de novo. To identify the specific actin nucleators responsible for the formation of nuclear actin filaments induced by PMA/Iono, we conducted a short hairpin RNA (shRNA)–based screen targeting human actin nucleators (fig. S3, E to G). Silencing expression of Arp3, an essential subunit of Arp2/3 complex (23, 24), with several independent shRNAs resulted in the strongest reduction of nuclear F-actin formation after T cell activation (Fig. 4, A and B, and movie S6). Suppression of nuclear actin filament formation with CK-869 was dose dependent (Fig. 4, C and D, and movie S7) but did not interfere with Ca2+ release in response to PMA/Iono (fig. S4, A to D). TCR-induced actin polymerization in the nucleus also required the activity of the formin FMN2, which likely cooperated with Arp2/3 to form the nuclear actin filament network, whereas the formin DIAPH1 required for formation of serum-induced nuclear actin filament networks (15) was dispensable (fig. S3, E to G).

Fig. 4 Nuclear Arp3 mediates formation of TCR-induced nuclear actin filaments.

(A and B) Silencing of Arp3 expression impairs formation of TCR-induced nuclear actin filaments. (A) Representative images of Jurkat NLA cells expressing the indicated shRNAs after PMA/Iono stimulation. (B) Relative occurrence of cells with nuclear F-actin (top) and Western blot expression analysis (bottom). GAPDH, glyceraldehyde phosphate dehydrogenase. (C and D) Pharmacological inhibition of Arp2/3 complex impairs formation of TCR-induced nuclear actin filaments. Jurkat NLA cells were treated with DMSO or increasing amounts of the Arp2/3 inhibitor CK-869, stimulated with PMA/Iono, and analyzed for formation of nuclear actin filaments. (C) Representative images. (D) Relative occurrence of cells with nuclear F-actin. (E to G) Inhibition of nuclear Arp2/3 impairs formation of TCR-induced nuclear actin filaments at the IS. Primary human CD4+ T cells expressing mCherry or nucleus-targeted dn Arp2.mCherry (nuc.dnArp2.mCherry) together with nuc.lifeact.GFP were incubated with SEB-loaded Raji B cells (red) analyzed for the formation of F-actin in the nucleus and at the cell-cell contact. (E) Representative micrographs. Scale bar, 5 μm. Yellow arrowheads indicate F-actin at the IS, and white arrowheads indicate F-actin in the nucleus (see also movie S9). (F) Quantification of cells with nuclear F-actin. (G) Quantification of cells with F-actin at the IS. D1, donor 1. Silencing of NIK (H and I) or NWASP (J and K) expression impairs formation of TCR-induced nuclear actin filaments. (H and J) Representative images of Jurkat NLA cells expressing the indicated shRNAs after PMA/Iono stimulation. Quantifications (B, D, I, and K) depict means ± SD from three experiments with at least 30 cells evaluated each per condition. Statistical significance relative to shC (B, I, and K) and DMSO-treated (D) controls was assessed by one-sample t test.

To assess the relevance of nuclear Arp2/3 upon T cell activation by TCR engagement, we tested the effect of CK-869 in the context of the IS. However, as expected by the essential role of Arp2/3 for formation and function of the IS (9), treatment with the inhibitor interfered with the polymerization of cortical actin upon contact with TCR-stimulatory surfaces or SEB-loaded Raji cells (fig. S4, E to I). To assess the specific role of nuclear actin dynamics after engagement of the TCR, we therefore developed a tool allowing us to selectively interfere with Arp2/3 activity in the nucleus while leaving cytoplasmic Arp2/3 unaffected. Because we found that Arp3 resides in the nucleus in nonstimulated T cells without increased nuclear import upon PMA/Iono treatment (fig. S5), interference could not occur at the level of nuclear import but had to target nuclear Arp2/3 directly. To this end, we engineered a strictly nuclear version of a recently reported dominant negative (dn) of the Arp2/3 subunit Arp2. dnArp2 lacks several sites of phosphorylation critical for its function and blocks cytoplasmic Arp2/3 activity when expressed in the cytoplasm (25). Fusing dnArp2 to a nuclear localization signal and mCherry (nuc.dnArp2.mCherry) and expression in primary human CD4+ T cells after lentiviral transduction resulted in exclusively nuclear and robust expression of nuc.dnArp2.mCherry but only in a small number of cells (~1 to 5%). These low rates of expression made functional or biochemical characterization of bulk cultures of these cells challenging, in particular when considering the asynchronous, rapid, and transient nature of nuclear actin filament formation in response to T cell activation. We therefore visualized individual cells by high-speed multipositioning spinning disc microscopy to quantitatively analyze transduced cells. This analysis revealed that nuc.dnArp2.mCherry substantially and selectively reduced the formation of actin filaments in the nucleus upon IS formation (Fig. 4, E to G). Inhibition of nuclear actin polymerization upon TCR engagement did not affect cytoplasmic actin polymerization or Ca2+ release upon IS formation (Fig. 4, E to G; fig. S6, A and B; and movie S9) (see movie S8 for localization of nuc.dnArp2.mCherry) or contact with a TCR-stimulatory surface (fig. S6, C and D). Consistent with the interference with nuclear actin filament formation by nuc.dnArp2.mCherry, silencing the expression of Nck-interacting kinase (NIK), the kinase that phosphorylates the residues of Arp2 that are mutated in dnArp2 (25), also resulted in a reduction of nuclear actin filament formation (Fig. 4, H and I). This suggested that the NIK may control the activity of nuclear Arp2/3 in this process and that nuc.dnArp2.mCherry acts by assembling inactive Arp2/3 complexes. Arp2/3 by its own is a weak actin nucleator and requires activation by a nucleation-promoting factor (NPF) for efficient actin polymerization (8). The NPF N-Wasp is known to interact with CaM (26) and may thus be regulated by Ca2+. Consistently, silencing the expression of N-Wasp revealed an essential role of the NPF for the formation of nuclear actin filaments after T cell activation by PMA/Iono (Fig. 4, J and K). Arp2/3-mediated formation of F-actin foci at the IS requires activation by Wasp but not by N-Wasp (27), suggesting that the identity of the NPF involved provides selectivity for Arp2/3 in different cellular compartments. The formation of nuclear actin filaments triggered by TCR signaling is thus driven by nuclear pools of Arp2/3 that are subject to complex regulation, which likely include Ca2+-mediated activation of the NPF N-Wasp.

Arp2/3-mediated formation of nuclear actin network is essential for CD4+ T cell effector functions

We next sought to assess the functional role of nuclear actin filaments in T cell activation by using primary CD4+ T lymphocytes isolated from peripheral blood of healthy donors. The low transduction efficiency precluded generating sufficient material for a meaningful analysis of cells in which nuclear actin polymerization was impaired by expression of nuc.dnArp2.mCherry. However, treatment with Arp2/3 inhibitor CK-869 combined to bypassing of PM actin dynamics via activation by PMA/Iono enabled us to selectively interfere with nuclear Arp2/3 activity. Stimulation for 15 min resulted in the frequent and immediate formation of nuclear actin filaments but no detectable induction of actin rearrangements in the cytoplasm or at the PM (Fig. 5A) in control cells. The presence of CK-869 during activation reduced nuclear actin filament formation to background levels (Fig. 5, A and B, and movie S10).

Fig. 5 Arp2/3-mediated formation of nuclear actin networks is essential for CD4+ T cell effector functions.

(A and B) Presence of the Arp2/3 inhibitor CK-869 during 10 min of PMA/Iono stimulation inhibits nuclear actin network formation in primary human CD4+ T cells. (A) Representative images of primary human CD4+ T cells expressing nuc.lifeact.GFP. (B) Occurrence of nuclear F-actin relative to the PMA/Iono control (mean ± SD of cells from four different donors with 30 cells evaluated each per condition). (C) Primary human CD4+ T cells isolated from three healthy donors as depicted in fig. S7A were either left unstimulated or stimulated by PMA/Iono in the presence or absence of CK-869 for 24 hours. Shown are cytokine concentrations detected in the supernatants of the respective treatments that were potently blocked by the presence of CK-869 during T cell stimulation. n.d., not detected. (D) Schematic overview and experimental workflow of the in vivo analysis of T cell help. Transferred cells expressed transgenic BCR and TCR recognizing HEL and OVA antigens, respectively. (E) Cytokine production of murine CD4+ T cells expressing nuc.dnArp2.mCherry relative to mCherry control shown as log2 fold changes for cells from three animals. (F and G) Expression of nuc.dnArp2.mCherry in CD4+ T lymphocytes impairs production of antigen-specific antibodies. CD4+ T cells were transduced to express mCherry or nuc.dnArp2.mCherry, and mCherry-positive cells were adoptively transferred together with B cells into recipient mice, which were later immunized with HEL-OVA antigen (see legend to fig. S10 for details). At the indicated time points after HEL-OVA immunization, IgG and IgM antibody production was quantified (F and G). At day 12 or 14 after immunization, cells were isolated from the lymph node of recipient animals to determine the frequency of mCherry-positive cells among the lymphocyte population (H). Ex vivo proliferation assay of murine OT-II cells expressing mCherry or nuc.dnArp2.mCherry activated by coculture with SW-HEL B cells with different concentrations of HEL-OVA (I). Statistical significance relative to DMSO-treated (B) or mCherry (F) controls was assessed by one-sample t test (B) or Mann-Whitney test (F; mean ± SD from six mice per group and two independent experiments). Scale bar, 5 μm. Arrowheads indicate examples of nuclear F-actin filaments.

We assessed whether the formation of nuclear actin filaments is linked to a cytokine expression profile (fig. S7A). When analyzed 24 hours after stimulation by PMA/Iono, expression and secretion of a broad panel of 20 cytokines were induced relative to untreated control cells. Suppression of nuclear actin polymerization by CK-869 at the time point of T cell activation did not significantly affect production of a number of cytokines, including IL-5, IL-8, IL-13, and IL-17A, demonstrating that cytokine production was not generally impaired (fig. S7B). In contrast, suppression of nuclear actin polymerization was associated with a significant reduction in levels of key effector cytokines such as IL-2, IL-6, IL-9, IL-10, IL-21, IFN-γ, and TNF-α in the cell culture supernatant (Fig. 5C). This dysregulation was apparent on the mRNA and protein level as early as 4 hours after activation (fig. S7, C to E). Because this suggested the involvement of gene expression regulation, we determined the mRNA profile of primary human CD4+ T lymphocytes from several independent donors in response to activation in the absence or presence of CK-869 at several time points after activation. Only 86 of the 31,100 genes analyzed (0.3%) were dysregulated upon suppression of nuclear actin filament formation by CK-869 (table S1), excluding general effects of CK-869 on transcription. Dysregulation included up- and down-regulation of a specific set of genes early after activation (fig. S8A, 45 min) with suppression of an increasing number of genes at later time points (fig. S8A, 90 and 240 min after stimulation), some of which were recently reported to be controlled by nuclear Ca2+ (fig. S8B) (20). Analysis of the top 40 dysregulated genes 45 min after activation from both donors analyzed revealed that the pattern of dysregulation was remarkably conserved, contained mainly cytokine genes, and occurred already early after activation (fig. S9A). This dysregulation preceded alteration of cell cycle–related genes (fig. S9B) and was associated with mild to strong reduction in proliferation for cells from individual donors (fig. S9C). These results suggest dysregulation of cytokine gene expression as an important mechanism by which nuclear actin dynamics affect cytokine production of CD4+ T cells in response to activation.

Arp2/3-mediated formation of nuclear actin network is associated with efficient T cell help in mice

To address whether nuclear actin dynamics contributes to T cell helper function in vivo (13), we made use of an adoptive transfer mouse model. This approach was chosen because only relatively low cell numbers are required, and a retroviral transduction/selection procedure (28) allows us to isolate sufficient numbers of cells in which nuclear actin polymerization is selectively impaired by expression of nuc.dnArp2.mCherry. To assess the role of nuclear Arp2/3 for the production of antigen-specific antibodies in mice, CD4+ T cells from mice transgenic for ovalbumin (OVA) peptide (OVA 323 to 339)–specific TCR were adoptively transferred into recipient SMARTA mice with a transgenic TCR specific for an unrelated peptide combined with B cells of mice transgenic for hen egg lysozyme (HEL)–specific B cell antigen receptor (BCR). Subsequently, these mice were immunized with conjugated HEL-OVA antigen (Fig. 5D and fig. S10, A and B). Analysis of culture supernatants of mCherry- or nuc.dnArp2.mCherry-expressing mouse CD4+ T cells at the time of adoptive transfer revealed that, similar to CK-869 in human T cells in response to PMA/Iono, nuc.dnArp2.mCherry resulted in a complex dysregulation of cytokine production in response to antigen-specific T cell activation, including reduced production of IFN-γ, colony stimulating factor 2 (CSF2), IL-7, and IL-13 in cells from all and IL-2 and IL-21 in cells from most animals analyzed, respectively (Fig. 5E). Transfer of mCherry-expressing control T cells in recipient mice resulted in efficient immunoglobulin G (IgG) and IgM responses against HEL (Fig. 5, F and G), demonstrating their ability to provide adequate T cell help. Although survival and proliferation rates were comparable to control cells (Fig. 5, H and I), cells expressing nuc.dnArp2.mCherry were markedly less efficient in supporting the production of HEL IgG (Fig. 5F). Because this may indicate impairment in germinal center reaction (29), nuc.dnArp2.mCherry also showed slightly reduced production of αHEL IgM (Fig. 5G). We conclude that the role of nuclear actin filaments in T cell activation is critical to CD4+ T cell helper function in vivo.

DISCUSSION

Antigen-specific activation of CD4+ T lymphocytes triggers the expression and secretion of specific cytokines required for cell proliferation and differentiation and thus T cell help. In addition to induction of the activity of cardinal transcription factors such as NF-κB and NFAT and epigenetic regulation of TCR target genes (30), we identify herein the formation of nuclear actin filaments as an unexpected and previously unknown effector mechanism of TCR signaling that regulates cytokine production. Specific interference with the nuclear machinery mediating the formation of nuclear actin filaments impairs mounting of efficient antigen-specific humoral immunity in mice, suggesting that nuclear actin dynamics in response to TCR engagement is essential for T cell help. The formation of nuclear actin networks has recently been described in fibroblasts for a growing number of biological process such as the response to stimulation with serum (15), integrin engagement (31), or induction of DNA damage (16) and repair (32, 33) and may represent a cell-intrinsic process that is analogous to baculovirus-induced nuclear actin assembly (34). The findings of this study expand the involvement of nuclear actin dynamics to immune cell signaling such as activation of CD4+ T cells, where nuclear actin filament formation is associated with elevated production of specific cytokines. This effect is rapidly exerted after TCR engagement and manifest at the protein and the transcriptional level. Transcriptional dysregulation upon interference with nuclear actin dynamics affected a small set of target genes among which cytokines genes were strongly enriched. This suggests that the formation of nuclear actin filaments upon T cell activation promotes cytokine production by selective induction of their genes; however, indirect effects of the nuclear actin network on cytokine secretion may also contribute. Cells in which cytokine production was impaired because of interference with nuclear actin dynamics expanded normally but failed to mount an efficient humoral immune response after T cell activation in vivo. However, the precise step in immune responses such as T cell differentiation or the germinal center reaction affected by interference with nuclear actin dynamics could not be defined in the context of this study. In addition, the known effects by cytokines such as IL-10 (35) are likely to involve effects of noncytokine-encoding genes such as g. CD40L, which are dysregulated upon interference with nuclear Arp2/3. Through this combination of effects, the formation of TCR-induced nuclear actin filaments provides CD4+ T cells with a mechanism to rapidly convert from a resting state into a cytokine-secreting T effector cell in vivo.

On the mechanistic level, our study revealed a signaling cascade coupling TCR engagement to nuclear actin polymerization. On the basis of the time of onset, their duration, and the possibility for selective induction or inhibition, the formation of nuclear actin filaments is uncoupled from actin polymerization events at the PM. The central player is a nuclear pool of Arp2/3, which, probably in cooperation with other actin nucleators such as FMN2, mediates nuclear actin polymerization after T cell activation. Considering that nuclear actin polymerization precedes that at the PM, a key question was how Arp2/3 activity can be so rapidly induced in the nucleus. Our results define that Arp2/3 is already abundant in T cell nuclei before activation and hence only requires activation but not nuclear import. Together with the kinase NIK and nuclear Ca2+ waves, we identified two regulators of nuclear Arp2/3 after T cell activation, illustrating a direct link between Ca2+ signaling and actin polymerization. Of particular interest, the release of nuclear Ca2+ upon TCR engagement was recently shown to regulate expression of TCR target genes (20) including many cytokine genes (fig. S8), suggesting nuclear Arp2/3 as an effector of nuclear Ca2+. The molecular mechanism that couples Ca2+ to nuclear actin polymerization remains unclear but may involve activation of CaM binding of the essential upstream NPF N-Wasp (26) to selectively activate nuclear Arp2/3 (27). Activation of N-Wasp-Arp2/3 by nuclear Ca2+ and NIK may thus act as a module downstream of TCR engagement that provides a molecular switch that converts nuclear Ca2+ waves into immediate cytokine secretion programs.

The mechanism by which nuclear actin filaments that are induced by TCR signaling can regulate cytokine expression remains a key open question. Nuclear actin filaments have recently been shown to facilitate nuclear expansion at mitotic exit and localization of double-strand breaks to the nuclear periphery (32, 33, 36), and it is currently unclear how such biophysical properties might relate to the control of cytokine expression and release in CD4+ T cells. On the basis of their large size and short duration, TCR-induced nuclear actin filaments are also unlikely to exert direct physical effects on PolII transcription, suggesting that the role of nuclear Arp2/3 complex described here is distinct from previously described in vitro effects on monomeric actin in the context of transcription (37, 38). The function of nuclear Arp2/3 as a TCR effector instead resembles that of formin nucleators responsible for the formation of a nuclear actin filament network required for fibroblast responses to stimulation with serum (15) or triggering of integrin signaling (31). The coupling of nuclear actin dynamics to gene expression may emerge as a general and physiological mechanism of many signal transduction pathways. The involvement of specific actin nucleation machinery may allow adaptation of this principle to signaling pathway- and cell type–specific regulation, whereas the regulation by nuclear Ca2+ identified herein may represent a generalized mechanism.

In summary, our study describes a previously unidentified role of nuclear actin dynamics in cytokine expression control of CD4+ T lymphocytes in response to antigen-specific stimulation. Cytokine storm and cytokine release syndrome (CRS) are frequent side effects of cancer immunotherapy based on chimeric antigen receptor T cells. CRS is characterized by an early and rapid increase of cytokine levels in the blood, including IFN-γ, IL-2, IL-6, and IL-10 (39), whose expression is also controlled by nuclear actin polymerization early on in our system. Intensive efforts are aiming at neutralizing the effect of cytokines in CRS (40). The discovery of nuclear actin dynamics as a regulator of cytokine expression in T cells may open new avenues toward therapeutic control of advert cytokine responses in the clinic.

MATERIALS AND METHODS

Live-cell imaging of actin dynamics

Live microscopy of actin dynamics was performed with a Nikon Ti PerkinElmer UltraVIEW VoX spinning disc confocal microscope equipped with a perfect focus system (PFS), a 60× oil objective (numerical aperture, 1.49), Hamamatsu ORCA-flash 4.0 scientific complementary metal-oxide semiconductor camera, and an environmental control chamber (37°C, 5% CO2). Acquisition settings are as follows: exposure time, 160 ms; frame rate, 6 to 10 frames/s, number of Z planes, 10; Z-stack spacing, 0.5 μm; 488 nm, laser power 5.5%; and total acquisition time, 3 to 10 min. T cells stably expressing lifeact versions were washed with phosphate-buffered saline (PBS) and split 24 hours before the experiment to a density of 3 × 105/ml. The next day, 3 × 105 cells were harvested, washed with PBS, and resuspended in 100-μl live-cell imaging media. To adjust the PFS system, first, a low amount of highly diluted cells was placed on a 0.01% polyK-coated or FN-coated (30 μg/ml) (both from Sigma) glass bottom dish, a single cell was centered to the field of view, and the PFS was adjusted to automatically focus on the glass-cell contact site. Subsequently, the stage was moved to a cell-free area, and 100 μl of the cell suspension (3 × 105/100 μl) was added to the glass bottom dish with simultaneously recording cells while making contact with the glass surface. For filming of actin dynamics in cells resting on the glass surface, 100 μl of cells (3 × 105/ml) was plated on polyK-coated glass-bottom dishes, allowed to adhere for 5 min, and then stimulated with PMA/Iono in live-cell imaging media. For DNA staining, cells were incubated in 1:100,000 Hoechst 33342 (Thermo Fisher Scientific) solution for 15 min, followed by a washing step with PBS.

Primary CD4+ T cells (1 × 106) expressing nuc.lifeact.GFP were plated on 0.1% polyK-coated glass bottom dishes, and movies were acquired as single-plane images over multiple positions. For quantification by live microscopy, at least 20 primary T cells or 30 cells of T cell lines per condition were analyzed. Because the overall frequency of cells in which nuclear actin filaments were visualized after activation varied between cell lines and primary cells from different donors, frequencies of cells nuclear actin filaments were expressed relative to the nonstimulated control that was arbitrarily set to 1. Nuclear actin filaments were observed with nuc.lifeact.GFP in the following approximate fraction of all stimulated cells: Jurkat NLA, PMA/Iono, 29% and stimulatory surface, 22%; A3.01 NLA, 54%; primary CD4 T cells, 56%.

Phalloidin staining for nuclear actin filaments

T cells were washed and adjusted to cell density 1 day before the experiment, as described above. The next day, 5 × 105 cells were collected and resuspended in 100-μl live-cell imaging media. Cells were allowed to adhere for 5 min on polyK-coated glass bottom dishes. Stimulation was performed by adding 100 μl of PMA/Iono solution dropwise to the cell suspension. Cells were activated for 30 s to 3 min and then permeabilized and stained with a 100-μl mixture containing 0.3% Triton X-100 + Alexa Fluor Phalloidin 488 (1:2000) in cytoskeleton buffer [10 mM MES, 138 mM KCl, 3 mM MgCl, 2 mM EGTA, and 0.32 M sucrose (pH 7.2)] for 1 min. Cells were fixed with 2 ml of 4% methanol-free formaldehyde (Pierce) in cytoskeleton buffer and incubated for 25 min. Subsequently, the fixed cells were washed twice with cytoskeleton buffer, blocked with 5% bovine serum albumin (BSA) in cytoskeleton buffer, and stained with 1:500 Alexa Fluor Phalloidin 488 or 647 in cytoskeleton buffer for 1 hour at room temperature (RT) or overnight (ON) at 4°C. Endogenous nuclear actin filaments in the absence of nuc.lifeact.GFP were observed in the following approximate fraction of all stimulated cells: Jurkat (7%) and A3.01 (19%). Serum-stimulated NIH3T3 cells were prepared for microscopy as previously described (15).

Phalloidin staining of circumferential F-actin at the PM

Visualization of the circumferential F-actin ring upon TCR engagement was performed as described before (41). JTAg cells (3 × 105) were resuspended in 50 μl of cell culture medium and placed on a stimulatory coverslip, followed by 5-min incubation at 37°C. Cells were fixed with 3% paraformaldehyde for 20 min and washed thrice with PBS before permeabilization for 5 min with 0.5% Triton X-100 in PBS. Triton X-100 was washed off, and the coverslips were blocked in 3% BSA for 30 min at RT. F-actin was stained with 1:1000 Alexa Fluor Phalloidin 488 for 1 hour at RT. Samples were mounted on glass slides with Mowiol and analyzed by epifluorescence (Olympus IX81 S1F-3, cellM software). For quantification of phenotype frequencies, at least 100 cells were analyzed.

Antibody production in mice after adoptive transfer and recovery of transferred cells

For collaborative response of OT-II T cells and SWHEL B cells, 2 × 104 FACS (fluorescence-activated cell sorting)–sorted mCherry-positive OT-II T cells and 6 × 104 SWHEL B cells were adoptively transferred into SMARTA mice. After 24 hours, mice were immunized subcutaneously with 10 to 20 μg of chemical conjugates of OVA323–339 peptide and HEL emulsified in Freund’s complete adjuvant (Santa Cruz Biotechnology) (42). Blood was collected from mice before and after immunization via facial vein puncture and production of IgM and IgG antibodies was tested by enzyme-linked immunosorbent assay. Briefly, 96-well plates were coated with HEL (2 μg/ml) (Sigma) ON. The next day, the plates were washed with PBS (with 0.1% Tween 20). Sera extracted from the blood were then added, and plates were incubated for 2 hours. Then, the plates were washed and incubated with biotinylated anti-IgM and anti-IgG antibodies (Bio-Rad), followed by horseradish peroxidase–conjugated streptavidin (Thermo Fisher Scientific). Ortho-phenylenediamine substrate (Thermo Fisher Scientific) was used to develop the reaction that was then stopped using 3 M H2SO4.

For comparing the cell survival and expansion potential of control and dnArp2-expressing T cells, draining lymph nodes were harvested from the immunized mice after 2 weeks, and percentage of mCherry-positive cells in the cell suspension was determined by flow cytometry.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/4/31/eaav1987/DC1

Materials and Methods

Fig. S1. Visualization of nuclear and cytoplasmic actin dynamics upon TCR stimulation.

Fig. S2. Calcium and actin dynamics during T cell stimulation.

Fig. S3. Screen for actin nucleators involved in nuclear actin polymerization upon T cell stimulation.

Fig. S4. Effect of CK-869 on Ca2+ and actin dynamics.

Fig. S5. Effect of T cell activation on subcellular distribution of Arp3.

Fig. S6. Influence of nuc.dnArp2.mCherry on actin and Ca2+ dynamics upon T cell stimulation.

Fig. S7. Cytokine release upon T cell stimulation in the presence of CK-869.

Fig. S8. Differentially expressed genes upon T cell stimulation and inhibition of Arp2/3 or nuclear calcium.

Fig. S9. Gene expression of cytokines is affected early after activation in the presence of CK-869.

Fig. S10. In vivo model system used for analysis of T cell help.

Table S1. Time-dependent differential gene expression of primary human CD4+ T cells upon PMA/Iono activation and CK-869 treatment.

Table S2. Raw data of all figure graphs.

Movie S1. Nuclear actin polymerization precedes formation of circumferential F-actin rings at the PM triggered by TCR engagement.

Movie S2. Nuclear actin polymerization upon immune synapse formation.

Movie S3. TCR signaling induces nuclear actin polymerization.

Movie S4. Calcium release and nuclear actin polymerization coincide.

Movie S5. Inhibition of nuclear calcium signaling inhibits nuclear actin polymerization.

Movie S6. Three different shRNAs targeting Arp3 expressed in Jurkat NLA impair formation of PMA/Iono-induced nuclear actin filaments.

Movie S7. Pharmacological inhibition of Arp2/3 complex impairs formation of TCR-induced nuclear actin filaments.

Movie S8. Inhibition of nuclear Arp2/3 impairs formation of TCR-induced nuclear actin filaments.

Movie S9. Specific inhibition of nuclear Arp2/3 and the effect on nuclear actin polymerization on immune synapse formation in T cells.

Movie S10. Pharmacological inhibition of Arp2/3 inhibits nuclear actin polymerization in primary T cells.

References (4352)

REFERENCES AND NOTES

Acknowledgments: We are grateful to I. Ambiel and N. Tibroni for expert technical help, L. LeClaire and H. Bading for providing reagents, and M. Lusic, H. Bading, F. Frischknecht, and A. Ruggieri for critical discussion and comments. We thank the microarray unit of the DKFZ Genomics and Proteomics Core Facility for providing the Illumina Whole-Genome Expression BeadChips, the FACS facility unit of the ZMBH for service, and M. Langlotz for help. Funding: O.T.F. is a member of the CellNetworks Cluster of Excellence (ECX81). This research was supported by funding by the Deutsche Forschungsgemeinschaft (grants FA378-1/15-1 and SFB1129 to O.T.F. and fellowship to N.T. from the HBIGS graduate school) and HFSP (grant RGP0021/2016 to RG). Author contributions: N.T. carried out all experiments including statistical analysis with the exceptions mentioned below. O.T.F. conceived the study and O.T.F. and N.T. wrote the manuscript. O.T.F., R.G., and N.T. interpreted the data and designed experiments. R.G. and C.B. provided essential concepts and reagents. V.L. provided technical support in microscopy and generated and analyzed STED images. S. Kaw designed, performed, and interpreted the in vivo experiments. S. Kutscheidt contributed nuclear-cytoplasmic fractionations. B.S. designed the in vivo experimental system and coordinated animal work with the Heidelberg University animal facility. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All necessary data related to this research are included in this publication within the text or the Supplementary Materials. The microarray data are MIAME compliant and have been submitted to NCBI’s Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE122531. Questions and reagent request can be made to O.T.F.
View Abstract

Navigate This Article