The DISC1–Girdin complex – a missing link in signaling to the T cell cytoskeleton

When T cells engage antigen-presenting cells they form a specialized contact site known as the immunological synapse. Classically, the synapse can be described in terms of concentric zones of receptors, adhesion proteins and cytoskeletal elements (Monks et al., 1998; Bunnell et al., 2001; Kupfer et al., 1987; Grakoui et al., 1999). The central region, known as the central supramolecular activation cluster (cSMAC), is characterized by the accumulation of T cell receptors (TCR) and protein kinase C-theta (PKC-θ) (Monks et al., 1998). Surrounding the cSMAC is an outer zone known as the peripheral supramolecular activation cluster (pSMAC), which is characterized by clusters of the lymphocyte function-associated protein 1 (LFA-1, also known as integrin alpha-L). Finally, there is the outermost zone known as the distal supramolecular activation cluster (dSMAC), which is enriched in lamellipodial actin (Freiberg et al., 2002).

One of the prominent features of the immunological synapse is the accumulation of actin, which is seen as a ring-like structure that forms at the edges of the cell lamellae as they spread over the target cell. The most obvious actin assembly at the synapse is thought to be triggered by formation of phosphatidylinositol (3,4,5)-trisphosphate (PIP3), which in turn leads to activation of the WASP-family verprolin homologous protein WAVE2 and the actin-related protein (ARP) 2/3 structure (Le Floc'h et al., 2013; Chen et al., 2017; Basu et al., 2016). However, recent studies have revealed a remarkable complexity to the regulation of actin, involving multiple actin regulators and nucleators with a variety of effects on T cell functions (Le Floc'h et al., 2013; Kumari et al., 2015; Jankowska et al., 2018; Janssen et al., 2016; Comrie et al., 2015). Defects in proper actin assembly have been linked generally to immunodeficiency and autoimmune dysfunctions (Wickramarachchi et al., 2010). More specifically, actin dynamics have been linked to sustained T cell signaling, cell spreading, calcium entry, formation of stable adhesions and target cell stimulated secretion in cytotoxic T lymphocytes (CTLs) (Babich et al., 2012; Carisey et al., 2018; Nolz et al., 2006).

In this study, we introduce new players in the T cell actin schema, Girdin (girders of actin filaments) and DISC1 (disrupted in schizophrenia 1). DISC1 is a scaffolding protein that, in addition to dynein, interacts with over 100 different proteins including those associated with centromeres, the cytoskeleton, cell signaling and neuronal synapses (Camargo et al., 2007; Chubb et al., 2008). We previously showed that DISC1 forms a complex with dynein, NDE1 (neurodevelopment protein 1, also known as nuclear distribution protein nudE homolog 1) and LIS1 (lissencephaly 1) in Jurkat cells (Nath et al., 2016) but its function was not explored. Girdin (also known as GIV, G α-interacting vesicle-associated protein) exhibits a variety of signaling functions and is involved in cytoskeletal reorganization and integrin signaling (Leyme et al., 2016, 2015; Aznar et al., 2016; Weng et al., 2010).

We initially showed there are two DISC1 isoforms expressed in Jurkat cells, the full-length L isoform (DISC1L) and the Lv splice variant (DISC1Lv) (Nakata et al., 2009). We show that the L isoform accumulates at the immunological synapse, whereas DISC1Lv is associated with mitochondria. When DISC1 was deleted using CRISPR/Cas9, actin accumulation at the immunological synapse was greatly reduced and members of the dynein complex (dynein, NDE1 and LIS1), which colocalizes with ADAP (FYN-binding protein 1) at the pSMAC, remain clustered together near the center of the synapse (Combs et al., 2006; Nath et al., 2016). We also found that DISC1 forms a complex with Girdin and that deletion of Girdin gave a phenotype similar to that of DISC1 knockout (DISC1-KO) cells. When DISC1L or Girdin were reintroduced as cDNAs into their respective deletion mutant cell lines, the DISC1–dynein complex was again localized at the pSMAC and actin accumulation at the synapse was restored to levels seen in wild-type (WT) Jurkat cells.

The loss of actin accumulation and failure of the dynein complex to locate at the pSMAC suggested that there was a connection between actin assembly and localization of the NDE1–dynein complex. To explore this further, we treated Jurkat cells with cytochalasin B (CytB) or latrunculin B (LatB) to disrupt actin polymerization prior to formation of conjugates with staphylococcal enterotoxin E (SEE)-coated Raji cells. These treatments resulted in loss of the dynein complex from the synapse and a failure of microtubule organizing center (MTOC) translocation to the immunological synapse. Moreover, we showed that after Jurkat cells are activated by anti-TCR Ig, DISC1 co-immunoprecipitates with talin. Because talin binds to LFA-1, this finding may explain how the dynein complex becomes associated with the pSMAC (Klapholz and Brown, 2017).

Initial immunostaining studies showed that DISC was concentrated around the MTOC in unstimulated Jurkat cells but accumulated at the synapse after conjugation with SEE-coated Raji cells (Fig. 1A). These results were not unique to Jurkat cells as DISC1 also accumulates at the immunological synapse in NK-92-Daudi cell pairs and in mouse OT-1 CTLs engaged with peptide-pulsed EL4 cells (Fig. S1).

We then proceeded to clone DISC1 from a Jurkat cDNA library. Sequencing of the clones revealed two previously identified isoforms, L and Lv (Nakata et al., 2009). We used these two clones to produce DISC1-eGFP chimeras, which were expressed in Jurkat cells. We found that DISC1 isoform Lv localized to organelles that accumulate near the synapse after stimulation with SEE-coated Raji cells (Fig. 1B). These organelles were subsequently identified as mitochondria, which are known to accumulate at the immunological synapse in a microtubule-dependent manner (He et al., 2019; Maccari et al., 2016; Quintana et al., 2007). Isoform L concentrated around the MTOC in unstimulated Jurkat cells but also became localized to the synapse after stimulation with SEE-coated Raji cells (Fig. 1C).

To explore the function of DISC1, we initially introduced siRNA to reduce DISC1 expression. The results showed that DISC1 expression was only partially reduced and that cells quickly recovered. As an alternative, a CRISPR/Cas9 construct targeted to DISC1 was used to disrupt the gene. Chemical selection and cell sorting were then used to obtain a pure DISC1 deletion cell line (DISC1-KO). Absence of DISC1 expression in this cell line was verified by immunoblotting and an absence of DISC1 immunofluorescence at the synapse (Fig. S2).

One of the most obvious effects of DISC1 deletion was the loss of actin accumulation at the immunological synapse, as detected by staining with phalloidin-TRITC (Fig. 2A,B). Actin accumulation was analyzed by plotting average phalloidin-TRITC fluorescence for pixels within segments spanning the width of the synapse; 30 wild-type (WT) and 30 DISC1-KO cells were analyzed (Fig. 2C). For the five segments closest to the immunological synapse there was significant difference in phalloidin-TRITC fluorescence (P<0.001). Additionally, to determine whether disruption of DISC1 had an effect on LFA-1 recruitment to the synapse, we immunostained for the LFA-1 adapter talin in WT and DISC1-KO cells (Fig. 2A,B). The results show that talin was recruited to the immunological synapse in the absence of DISC1.

To verify that the loss of actin at the immunological synapse was specific to deletion of DISC1, we introduced DISC1-eGFP constructs for L and Lv isoforms into DISC1-KO cells. However, neither of the DISC1-eGFP isoforms restored actin accumulation at the synapse (Fig. S3). We then repeated the experiment using DISC1 constructs without the fused eGFP. In this case, only expression of the L isoform was able to restore visible actin staining at the immunological synapse (Fig. 2D,E). Finally, to determine whether detectable actin remained at the immunological synapse in the DISC1-KO cell line, we used confocal microscopy together with phalloidin-TRITC staining to compare WT Jurkat cells and DISC1-KO cells that were either untreated or treated with LatB to disrupt actin (Fig. 2F–H). The results show that treatment of DISC1-KO cells with LatB visibly reduced actin staining below the levels seen in untreated DISC1-KO cells. We found that the fluorescent intensity around the synapse was reduced by 42% in DISC1-KO cells compared with the WT, subtracting for background and using LatB-treated cells as a baseline for 0% (n=10; P<0.05). This suggests that some polymerized actin remains in the DISC1-KO cells.

In searching for a functional connection between DISC1 and actin at the synapse, immunoprecipitation experiments revealed that Girdin was part of the DISC1 complex (Fig. 3A). Like DISC1, Girdin also accumulated at the synapse of Jurkat cells activated by SEE-coated Raji cells (Fig. S4A,B). To determine whether Girdin was needed for actin accumulation at the synapse, a CRISPR/Cas9 construct targeted to Girdin was used to disrupt the gene. Chemical selection and cell sorting were then used to obtain a pure Girdin knockout cell line (Girdin-KO). Absence of Girdin expression was verified by immunoblotting and by absence of any accumulated immunofluorescence at the synapse (Fig. S4C–F).

When Girdin-KO cells were paired with SEE-coated Raji cells, fixed and stained with phalloidin-TRITC, the normally obvious actin ring at the immunological synapse was not detected, similar to observations for DISC1-KO cells (Fig. 3B,C). When Girdin was reintroduced into Girdin-KO cells as an eGFP-fusion protein, a repeat of the phalloidin-TRITC staining experiment showed that actin was clearly present at the synapse (Fig. 3D). Plots of average phalloidin-TRITC fluorescence for segments across the width of the synapse were generated from 30 WT cells, Girdin-KO cells or Girdin-KO cells expressing Girdin-eGFP (Fig. 3E). Comparisons of the first five segments showed a significant difference between WT and Girdin-KO cells (P<0.001) and no significant difference between WT cells and Girdin-KO cells expressing Girdin-eGFP (P>0.05).

Previous studies have shown that dynein complexes immunoprecipitated from Jurkat cells contained NDE1 and LIS1 and that all three proteins form a ring-like pattern corresponding to the pSMAC (Nath et al., 2016). To determine whether DISC1 was needed to recruit the NDE1–LIS1 complex to the immunological synapse, WT and DISC1-KO Jurkat cells were immunostained for NDE1 and LIS1 (Fig. S5A,B). Images showed that both NDE1 and LIS1 accumulated at the synapse in the absence of DISC1. When the average NDE1 fluorescence was plotted for segments across the cell, the average NDE1 fluorescence at the Jurkat–Raji junction was slightly but significantly different across the first five segments from the synapse (Fig. S5C; P<0.05).

To investigate further how these dynein-associated proteins were distributed at the synapse, Jurkat cells were settled on anti-TCR-coated coverslips, fixed and immunostained for the dynein intermediate chain (DIC). In WT Jurkat cells, dynein showed a ring-like pattern (Fig. 4A). In the DISC1-KO cells, dynein was confined to the central region of the synapse (Fig. 4B). When DISC1L-eGFP was expressed in DISC1-KO cells, the distribution of dynein was essentially identical to that seen for WT Jurkat cells (Fig. 4C). When DISC1Lv was expressed in DISC1-KO cells, the distribution of dynein and its associated proteins remained in the center (Fig. 4D). In Girdin-KO cells, dynein was confined to a central position, resembling the distribution in DISC1-KO cells (Fig. 4E).

We repeated the experiment, but co-immunostained for NDE1 and LIS1 (Fig. 4F–J). We found that NDE1–LIS1 similarly formed a peripheral ring at the synapse in WT cells, which was not seen in DISC1-KO and Girdin-KO cell lines. This ring formation was similarly recovered by reintroducing the DISC1L-eGFP construct but not the DISC1Lv-eGFP construct. Taking 30 cells of each type, we grouped cells based on whether they formed a ring, a central spot or a miscellaneous pattern (Fig. 4K) and compared the results with a chi-squared test. This showed that WT cells were significantly different in actin ring formation compared with DISC1-KO cells, χ² (1, N=30)=28.1, P<0.001; DISC1-KO cells expressing DISC1Lv-eGFP, χ² (1, N=30)=28.8, P<0.001; and Girdin-KO cells, χ² (1, N=30)=11.6, P<0.01. Interestingly, in Girdin-KO cells, loss of the ring-like distribution of NDE1 and LIS1 was not as severe as that seen for DISC1-KO cells. WT cells showed no significant difference in ring distribution compared with DISC1-KO cells expressing DISC1L-eGFP, χ² (1, N=30)=0.33, P>0.05.

We next looked at the distribution of the dynein–NDE1 complex using talin as a marker for the pSMAC. We also used interference reflection microscopy (IRM) to image the total adhesion area between the cell and the coverslip. In WT cells, we found that NDE1 and talin colocalize at the pSMAC (Fig. 4L). In DISC1-KO cells, an apparent pSMAC still formed, as evidenced by the ring of talin, but NDE1 remained at the center of the synapse (Fig. 4M).

We next sought to track the temporal movements of the DISC1 and the dynein complexes. For these studies, DISC1L-eGFP was expressed in WT and DISC1-KO cell lines (Fig. S6). Because NDE1 is complexed to and colocalizes with dynein, we used the NDE1-mCherry chimera to follow the dynein complex in WT, DISC1-KO, or DISC1-KO cells expressing the eGFP fusions of either DISC1L or Lv. In WT Jurkat cells settled on anti-TCR coated coverslips, we found that DISC1L-eGFP and NDE1-mCherry initially accumulated at the center of the synapse before spreading out into a ring corresponding to the pSMAC. In the absence of DISC1, NDE1-mCherry failed to spread into a ring and remained at the center of the immunological synapse.

The data in Fig. 4 show that the dynein complex ultimately formed a ring at the pSMAC when Jurkat cells were bound to TCR-coated coverslips. However, the dynein complex failed to locate at the pSMAC in the absence of DISC1. Given that the pSMAC is often defined in terms of LFA-1 clustering, we looked for a direct link between DISC1 and LFA-1. Unfortunately, data from cell sorting, western blots and attempts to visualize ICAM (intercellular cell adhesion molecule) clustering on supported lipid bilayers all indicated that LFA-1 was expressed at very low levels on Jurkat cells. As such, we were not successful at immunostaining for LFA-1 or detecting LFA-1 in DISC1 immunoprecipitates. However, interference reflection microscopy showed that the contact site formed on supported lipid bilayers was substantially smaller in DISC1-KO cells than in WT Jurkat cells (Fig. S7). Furthermore, to the extent that it could be discerned, ICAM clustering in DISC1-KO cells was confined to a smaller ring than seen for WT Jurkat cells or DISC1-KO cells expressing isoform L.

Although we could not show that DISC1 binds to LFA-1, we were able to show an association between DISC1 and talin, which is known to bind LFA-1 (Tadokoro et al., 2003; Simonson et al., 2006). We immunostained Jurkat–Raji pairs for both DISC1 and talin and found that they colocalized at the synapse (Fig. 5A). We also found that DISC1 and talin co-immunoprecipitated when Jurkat cells were first treated with anti-TCR Ig (Fig. 5B). To determine which DISC1 isoform bound to talin, we used anti-GFP antibody to pull down each DISC1 isoform and then probed for talin on blots (Fig. 5C). The results showed that talin is primarily associated with isoform L.

The loss of actin and failure of the dynein complex to localize at the pSMAC suggested that DISC1 might be linked to actin, perhaps through Girdin. To test this idea, we treated Jurkat cells with either CytB or LatB to disrupt actin assembly and then immunostained Jurkat–Raji pairs for members of the dynein complex. The results show that treatment with either drug results in the loss of all members of the dynein complex (dynein, NDE1, LIS1) from the synapse (Fig. 6A,B). These results were confirmed by comparing average NDE1 fluorescence plotted from 30 treated and 30 untreated cells (Fig. 6C). From the segment-by-segment analysis of the plots, we found that fluorescence was significantly diminished in the first five segments of both CytB- and LatB-treated cells compared with WT cells (P<0.001).

To test how disrupting actin polymerization affects MTOC translocation, we immunostained Jurkat–Raji conjugates for tubulin to compare MTOC position in the presence or absence of actin-disrupting drugs (Fig. 7A–D). The distance of the MTOC was recorded for 33 cells of WT and DISC1-KO cell types, as well as for CytB- and LatB-treated cells (Fig. 7E). In WT cells, the distance of the MTOC was an average of 1.03±0.47 µm away from the synapse. In the absence of DISC1, the MTOC was an average of 2.05±1.22 µm away from the synapse, a small but significant difference compared with the distance in WT cells (P<0.001). However, for cells treated with either CytB or LatB, the MTOC remained much further away from the synapse than in WT (3.62±1.86 µm for CytB and 4.34±1.80 µm for LatB) indicating that little or no MTOC translocation had taken place (P<0.001). Finally, we saw no significant difference in MTOC position between CytB and LatB treatments (P>0.05).

Previous studies have shown that polarization of the microtubule cytoskeleton begins with a group of microtubules projecting to the central region of the interface between a T cell and a target (Kuhn and Poenie, 2002). These microtubules then fan out to form a hollow cone that projects from the MTOC to the pSMAC. It was noted that microtubules exhibit sharp bends in the zone where LFA-1 is clustered, and we speculated that these bends might be due to dynein. Tracking of the MTOC showed that, initially, it moves linearly towards the center of the contact site but begins oscillating laterally as it nears the immunological synapse. Subsequently, dynein forms a ring that colocalizes with ADAP, a protein that together with SKAP55 plays an important role in formation of the pSMAC (Combs et al., 2006). This ADAP complex colocalizes with the pSMAC by activating integrin clustering and binding to integrin complexes (Kliche et al., 2006; Wang et al., 2009; Burbach et al., 2011). Finally, our previous work showed that dynein at the immunological synapse is found in a complex with NDE1, LIS1 and DISC1 (Nath et al., 2016). We also showed that incompletely functional NDE1 could localize to the synapse in the absence of dynein, but that fully functional NDE1 is required to bind and recruit dynein.

The present study explores the function of DISC1 in more detail. We began by showing that there are two isoforms of DISC1 expressed in Jurkat cells (L and Lv). We focused primarily on the L isoform that accumulates at the immunological synapse upon activation of Jurkat cells. The Lv isoform accumulates around mitochondria, where it is known to bind the mitochondrial proteins Trak and Miro (James et al., 2004; Norkett et al., 2016). As we will detail in a separate study, DISC1Lv is required for accumulation of mitochondria near the immunological synapse.

Our results show that when Jurkat cells are activated by SEE-coated Raji cells, DISC1L, Girdin and members of the dynein complex first accumulate at or near the central region of the nascent immunological synapse and then become associated with the pSMAC. CRISPR-mediated deletion of either DISC1 or Girdin results in a loss of bulk actin accumulation at the immunological synapse and failure of the dynein complex to form a peripheral ring corresponding to the pSMAC.

To examine the specificity of the CRISPR-mediated deletion of Girdin, a Girdin-eGFP construct was introduced into the Girdin-KO line and this restored the accumulation of actin at the synapse. Adding back DISC1Lv-eGFP did not restore actin at the synapse nor association of the DISC1–dynein complex with the pSMAC. Expression of DISC1L-eGFP showed that this isoform could bind to talin and restore accumulation of the NDE1–dynein complex at the pSMAC, but does not support the polymerization of actin. Expression of the DISC1L isoform without the fused eGFP was able to restore actin accumulation at the immunological synapse. We suspect that eGFP placed at the N terminus of DISC1, near where Girdin binds, interferes with the binding or function of Girdin (Enomoto et al., 2009). It should also be noted that we tested a construct in which eGFP was placed at the C terminus of DISC1. This construct failed to accumulate at the synapse. It is known that the C terminus of DISC1 contains binding regions for NDE1 and LIS1 (Burdick et al., 2008; Sanchez-Pulido and Ponting, 2011), so in this case the eGFP tag may interfere with DISC1 binding to the dynein complex.

In both the DISC1 and Girdin KO cell lines, the NDE1–dynein complex remains in an approximately central location at the synapse. This dynein complex is sufficient to draw the MTOC close but not all the way to the immunological synapse. Indeed, the difference in MTOC position seen in WT Jurkat and DISC1-KO cells is subtle and was initially missed. However, the MTOC–cell edge measurements on at least 30 cells of each category show a significant difference. When Jurkat cells were treated with either CytB or LatB and then paired with SEE-coated Raji cells, the dynein complex was absent from the synapse and the MTOC failed to translocate. These results support the idea that the NDE1–dynein complex might be directly or indirectly linked to actin, and explains why dynein-dependent translocation of the MTOC to the synapse is blocked by treatment with these drugs (Filbert et al., 2012; Orange et al., 2003; Wulfing et al., 2003). However, although a small amount of actin in the center of the immunological synapse has been reported, we were not able to detect it by phalloidin staining (Sanchez et al., 2019; Murugesan et al., 2016).

The dramatic decrease in actin in the absence of DISC1L or Girdin, as well as the failure of the dynein complex to relocate to the pSMAC initially, suggests that these two events are linked. This idea was supported by data showing that CytB or LatB can reduce synaptic phalloidin staining below that seen for the DISC1-KO or Girdin-KO lines and that these treatments appear to eliminate the dynein complex from the synapse. One potentially attractive model is that the early DISC1–dynein complex is linked to a small patch of actin at the center of the immunological synapse and progresses to the periphery on a wave of actin polymerization. This appears unlikely given what is known about actin polymerization at the synapse, where formation of the actin network begins at the periphery instead of spreading outward. Although there is some evidence suggesting that actin is needed to recruit dynein to the synapse, at present we do not know how the initial dynein complex is anchored to the central region of the synapse.

The results obtained when DISC1L-eGFP was expressed in the DISC1-KO cells show that dynein relocates to the pSMAC in the absence of the actin ring. Thus DISC1, if linked to actin, is not linked to the same set of actin filaments that Girdin generates. In principle, DISC1 could bind to actin through LIS1. LIS1 interacts with actin filaments through the actin-binding protein IQGAP and also interacts with the actin guanine-nucleotide exchange factors (GEFs) Cdc42, Rac and RhoA. Additionally, formins are required for MTOC translocation, so formin-generated actin filaments might link actin to the DISC1 complex (Gomez et al., 2007). Formins are needed to generate the concentric arcs of actin seen in the pSMAC. Because these actin arcs are linked to LFA1, and DISC1 is bound to talin, this seems to be a plausible connection (Murugesan et al., 2016). However, we note that formins are also involved in microtubule acetylation, which in turn can affect MTOC translocation, so this issue needs to be resolved (Thurston et al., 2012).

From our studies, the single biggest defect we see in this loss of actin is a failure of the cell to spread, as seen from interference reflection microscopy on supported lipid bilayers. The use of ICAM as a marker on these lipid bilayers shows that the diameter of the LFA-1 ring, insofar as one can see it, is also reduced. This is consistent with earlier studies showing that polymerization of actin drives spreading of the contact site (Bunnell et al., 2001; Barda-Saad et al., 2005). More recently, actin polymerization at the synapse has been shown to depend on WAVE2 and associated proteins (Nolz et al., 2006; Zipfel et al., 2006). Blocking the ability of the WAVE2 complex to activate ARP2/3 also blocks spreading.

This spreading defect could be one possible explanation for the failure of the NDE1–dynein complex to form a peripheral ring in the absence of DISC1. In this case, the failure to spread could be a physical barrier that prevents outward relocation of the dynein complex to the pSMAC. To examine this question, we used coverslips coated with anti-TCR Ig and poly-L-lysine where Jurkat cells bind and form a larger contact area (Bunnell et al., 2001). We also immunostained these preparations for talin to identify the putative pSMAC. The results show that in the absence of DISC1 the NDE1–dynein complex remains confined to the central region of the synapse, even though a peripheral ring of talin is evident. Thus, spreading as a physical barrier might be necessary for the peripheral spread of the dynein complex, but that alone is not sufficient.

At present, it is also not clear how the DISC1–Girdin complex regulates actin polymerization. Girdin is essential for cytoskeletal reorganization in a number of systems (Enomoto et al., 2005; Wang et al., 2018; Gu et al., 2014). Furthermore, there are several reports showing that reduced Girdin expression leads to a reduction in polymerized actin (Gu et al., 2014; Wu et al., 2016; Enomoto et al., 2005). There are also studies showing that depletion of DISC1 is associated with a reduction of actin, although this might also be a result of the link between DISC1 and Girdin (Steinecke et al., 2014).

There are several possible ways that Girdin could be involved in pathways that stimulate actin polymerization. Girdin is a non-receptor GEF for the Gαi group of trimeric G proteins and there is evidence that it links activation of receptor tyrosine kinases to signaling through these G proteins (Lin et al., 2014). T cells are known to express Gαi, which is important for T cell development, so this signaling pathway could be in play (Hwang et al., 2017; Leyme et al., 2016). Another possible signaling pathway is through the binding of DISC1 to the Ras effector RASSF7, which could play a role in activating the Ras signaling pathways (Wang et al., 2016). It should be noted that Girdin was first identified as an enhancer of AKT signaling (Enomoto et al., 2005; Anai et al., 2005) and might be part of a FAK-dependent positive feedback loop whereby integrin stimulation leads to activation of phosphoinositide 3 (PI-3) kinase (Leyme et al., 2016). Regardless of how it takes place, it is clear that Girdin serves as a hub for amplifying signaling through the PI-3 kinase/AKT pathway (Lin et al., 2014, 2011; Enomoto et al., 2005; Ni et al., 2015; Wu et al., 2016).

Of the various signaling pathways in which Girdin participates, the ones that appears most closely related to actin signaling are signaling through the Ras pathway and the PI-3 kinase pathway (Le Floc'h et al., 2013; Hammer et al., 2019). Interestingly, Le Floc'h and colleagues suggest that the Ras and PI-3 kinase pathways are early signaling steps that lead to actin polymerization (Le Floc'h et al., 2013). Clearly, more work needs to be done to understand the role of the DISC1–Girdin complex in regulating actin.

The results obtained here complement those of Kuhn and Poenie (2002). In that study, microtubules were seen to initially concentrate at the center of the synapse and then fan out to form a hollow cone that projects to the pSMAC. In tracking MTOC translocation to the immunological synapse, they found that the MTOC initially moved linearly toward the synapse. Then, as the MTOC came near the immunological synapse, it began to oscillate laterally. Here we expand on those results and propose a revised model for translocation of the MTOC to the immunological synapse (Fig. 8). We have shown that the NDE1–dynein complex first accumulates at the center of the immunological synapse. Dynein at this central location would tend to pull the MTOC straight towards the immunological synapse. Then, the dynein complex moves peripherally to associate with the pSMAC. At this point, the MTOC would be at the center of the dynein ring and opposing dynein forces would act to pull the MTOC laterally in an oscillating manner. However, in the absence of DISC1, the dynein complex is unable to move peripherally and the MTOC is not brought directly adjacent to the center of the synapse by opposing dynein forces.

A final point that should be mentioned concerns the role of DISC1 in schizophrenia. Up to this point, almost all studies of DISC1 have focused on its role in the brain and in neurons. However, the connection between schizophrenia and the immune system has recently come to the fore. As Debnath stated in a review, “immunopathogenesis has emerged as one of the most compelling etiological models of schizophrenia” (Debnath, 2015). Although models relating immune effects to schizophrenia are largely correlative, this study provides a concrete link between a known genetic risk factor for schizophrenia and T cells.

The goal of this study was to characterize the role of DISC1 in the formation of the immunological synapse. Images in each experiment were taken under the same conditions and brightness/contrast was adjusted to the same degree and in the same manner for all representative images of the same experiment. Cell selections were imaged randomly in all experiments. Measurements of cell pairs were conducted blind of the experimental groups being measured. For all experiments, a minimum of 30 cells or cell pairs were used per experimental group. The minimum number of cell pairs needed to avoid a type II (nonrejection of a false hypothesis) error was established as 30 through power analysis calculations made using the R programming environment. For each experiment, a value for power of at least 80% could be expected.

The Jurkat (E6.1), Raji and Daudi cell lines were obtained from the American Type Central Collection. The EL-4 cell line was obtained from Dr Anne-Marie Schmitt-Verhulst (Centre d'Immunologie de Marseille-Luminy, Marseille, France). OT-1 splenocytes were obtained from Dr Lauren Ehrlich (Molecular Biosciences, University of Texas at Austin). These C57BL/6-Tg(TCRaTCRb)1100Mjb/J (OT-I) mice were sourced from Jackson Laboratories and bred in house. All strains were bred and maintained under specific pathogen-free conditions in the University of Texas at Austin animal facility. Experiments were performed using mice 1-3 months of age of mixed sex. Mouse maintenance and experimental procedures were carried out with approval from the Institutional Animal Care and Use at the University of Texas at Austin.

Opti-MEM cell media was obtained from Gibco Thermo-Fisher (Cat # 31985062). Heat-inactivated fetal bovine serum (FBS) was obtained from Atlas Biologicals (Cat # F-0500-D). The 4 mm gap transfection cuvettes were obtained from Fisher Scientific (Cat # FB104). Goat serum (Cat # G2093), poly-L-lysine (Cat # P2636) and cytochalasin B (Cat # C6762) were obtained from Sigma-Aldrich. SuperSignal West Pico chemiluminescent substrate solution (Cat # 34580) and X-ray film (Cat # 34090) were obtained from Thermo Scientific. G418 Sulfate was purchased from Gold Biotechnology (Cat # G-418-5). All restriction enzymes were obtained from New England Biolabs. Mini Plasmid and Midi Fast Ion Plasmid Kits were obtained from IBI Scientific (Cat # IB47111 and IB47111). Xfect transfection reagent was obtained from Clonetech (Cat # 631318). ProLong Gold Anti-Fade Mounting Reagent was obtained from Life Technologies (Cat # P36930). The Cas9, DISC1 and Girdin sgRNA plasmids were obtained from Genecopoeia (Cat # CP-LvC9NU-02-B, HCP268459-LvSG03-1-B and HCP259879-LvSG03-1-B). Partially purified staphylococcal enterotoxin E (SEE) was obtained from Toxin Technologies (Cat # ET404).

DISC1 rabbit polyclonal antibody (Cat # PA2023) and Talin mouse monoclonal antibody (Cat # MA1092) were obtained from Boster Biological. CCDC88A (Girdin) rabbit polyclonal antibody (Cat # A16132) was obtained from Abclonal. LIS1 mouse monoclonal antibody (Cat # L7391), eGFP polyclonal rabbit antibody (Cat # G1544), Dynein (intermediate chain) mouse antibody (Cat # D5167) and β-tubulin mouse monoclonal antibody (Cat # T8328) were obtained from Sigma-Aldrich. The TCR Vβ8 mouse monoclonal antibody was obtained from BD Biosciences (Cat # 555604). Rabbit anti-NDE1 antibody was obtained from Proteintech Group (Cat # 10233-1-AP). Goat anti-rabbit AlexaFluor 594-conjugated antibody (Cat # A11037) and goat anti-mouse FIT-conjugated antibody (Cat # F2012) were obtained from Invitrogen. Goat anti-mouse IgM FITC-conjugated antibody (Cat # F2959), goat anti-mouse horse radish peroxidase (HRP)-conjugated antibody (Cat # A9917) and goat anti-rabbit HRP-conjugated antibody (Cat # A0545) were obtained from Sigma-Aldrich. Cell Tracker Blue (Cat # C2110) was obtained from Invitrogen. TRITC-conjugated phalloidin (Cat # P-1951) was obtained from Sigma-Aldrich.

Jurkat cells and Raji cells were grown in RPMI 1640 supplemented with 24 mM sodium bicarbonate, 1 mM sodium pyruvate, 2 mM L-glutamine, 50 µM β-mercaptoethanol, 10,000 U/ml penicillin, 10 mg/ml streptomycin and 10% (v/v) FBS (ACC growth media). Gryphon cells were grown in DMEM supplemented with 44 mM sodium bicarbonate, 1 mM sodium pyruvate, 2 mM L-glutamine, 50 µM β-mercaptoethanol, 10,000 U/ml penicillin, 10 mg/ml streptomycin, and 10% (v/v) FBS. All cells were cultured at 37°C in 5% CO₂.

For expansion and stimulation of OT-1 cells, EL-4 cells were treated with 50 µg/ml mitomycin C for 2 h, washed thoroughly and then treated with 1 µM Ova peptide. These EL-4 cells were then mixed with OT-1 splenocytes or activated OT-1 CTLs, or used as targets in immunostaining experiments. OT-1 CTLs were maintained in ACC growth media supplemented with 20 U/ml IL-2.

We began identifying possible DISC1 isoforms expressed in Jurkat cells by adapting an RT-PCR method used by Nakata et al. (2009). First, total Jurkat mRNA was isolated using the RNeasy Midi Kit. This mRNA was converted into a cDNA library using a MMLV reverse transcriptase kit. Using eight different sets of primers, DISC1 exon fragments were identified from this cDNA library through PCR. Through this method, DISC1L and Lv isoforms were identified and verified through Sanger sequencing on an Applied Biosystems 3730 DNA Analyzer (UT ICMB Core Facilities).

Full-sized DNA fragments for DISC1 isoforms L and Lv were synthesized through PCR of Jurkat cDNA. A tagless DISC1 construct was made by making DNA fragments of DISC1 isoforms L and Lv containing the XhoI and XmaI restriction sites. These fragments were inserted into a peGFP-C1 vector to generate DISC1-eGFP constructs, or into a peGFP-N1 vector that contained a premature stop codon at the start of the eGFP sequence, created through site-directed mutagenesis.

Full-sized Girdin DNA fragments were derived from our Jurkat cDNA library using PCR. A C-terminal eGFP-tagged Girdin construct was made by first adding XhoI and XmaI restriction sites to the ends of Girdin DNA fragments. These fragments were then inserted into a peGFP-N1 plasmid. Both peGFP-C1 and -N1 plasmids contain a monomeric eGFP sequence. PCR point editing was used to change amino acid 206 of the eGFP sequence from alanine to lysine.

The NDE1-mCherry construct was made by creating DNA fragments of the NDE1 coding sequence through PCR amplification of a pmeGFP-N1-NDE1 template, as described by Nath et al. (2016). DNA fragments contained AgeI and XbaI restriction sites on their 5′ and 3′ ends, which were used to insert the sequence into the pIRES-PURO3-mCherry vector. The pIRES-PURO3-mCherry vector was obtained from Dr Roger Tsien (University of California San Diego), and subsequently modified using a standard PCR protocol to insert several additional restriction sites into the multicloning site.

All DNA constructs were transformed from frozen aliquots of competent bacteria suspended in CaCl₂ solution. These aliquots were thawed and mixed with DNA constructs, then put through heat shock at 43°C. Depending on the concentrations needed, DNA constructs were isolated using the Mini Plasmid or Midi Fast Ion Plasmid Kits and plasmid sequences were verified by Sanger sequencing. DNA was introduced into Jurkat cells through electroporation. For the transformation, Jurkat cells were washed and resuspended in Opti-MEM reduced serum medium at a concentration of 2×10⁷ cells/ml and incubated with 10 µg of plasmid DNA for 15 min at 37°C. Cells were placed in 4 mm gap transfection cuvettes and pulsed at 250 V (950 µF) using the Gene Pulser Electroporation System (Bio-Rad). After electroporation, cells were resuspended in fresh ACC growth media. Cells containing DISC1 constructs were grown under selection with 1 mg/ml G418, whereas cells containing NDE1 constructs were grown under selection with 2 µg/ml puromycin. Selection began 24 h post-transfection and continued for two weeks. Afterwards, cells were sorted for the expression of fluorescent proteins using the FACSAria cell sorter.

A DISC1 or Girdin sgRNA plasmid and a Cas9 plasmid were transfected into the Gryphon viral packaging cell line using the Xfect transfection reagent. Fresh growth medium was added 4 h post-transfection. At 48 h after transfection, supernatants containing the viral particles were collected. For transduction, 2×10⁶ Jurkat cells in six-well plates were spinfected at 500×g and 30°C for 1 h with medium containing viral particles and 8 µg/ml polybrene. Spinfection through centrifugation was repeated every 12 h for 36 h. After the final centrifugation, the medium was replaced with fresh growth medium. Successful transduction was confirmed through observation of eGFP and mCherry fluorescent proteins expressed by the Cas9 and sgRNA plasmids. Complete knockout of DISC1 in culture was achieved through FACSAria sorting of eGFP- and mCherry-expressing cells, followed by verification through DISC1 western blotting of the resultant cells. Cells were grown under selection with 1 mg/ml G418 sulfate and 2 µg/ml puromycin. Selection began 36 h post-transduction and continued for two weeks, until sorting was conducted with a FACSAria cell sorter. After sorting, selection was stopped. Cells were finally used for immunostaining or transfected with a new construct, after the loss of eGFP and mCherry fluorescence had been observed.

To prepare coverslips for cell staining and fluorescence microscopy, coverslips were first cleaned with a 9:1 mixture of ethanol and 1 M KOH for 1 h. They were then washed in dH₂O and coated with an aqueous solution of 0.1 µg/ml 30,000-70,000 kDa poly-L-lysine. These were rinsed again in dH₂O and left to dry for 30 min at room temperature.

To prepare Jurkat–Raji cell conjugates, Raji cells were suspended at a concentration of 1×10⁶ cells/ml in serum-free1 RPMI 1640 and treated with SEE at 1 µg/ml for 1 h at 37°C. They were subsequently stained with 10 µM Cell Tracker Blue for 15 min at 37°C in order to identify Raji cells from Jurkat cells during imaging. In experiments using CytB or LatB, Jurkat cells were treated with 20 μg/ml CytB or 10 μM LatB for 30 min. Both Jurkat and Raji cells were washed with ACC media, paired at a ratio of 3:2 Jurkat to Raji cells and centrifuged at a light speed (500×g) for 5 min. Cells were then washed again with ACC and settled on poly-L-lysine-coated coverslips at a total concentration of 1×10⁶ cells/ml for 15 min.

To prepare TCR antibody-coated coverslips, a 10 µg/ml solution of mouse Vβ8 anti-TCR antibody was coated over dried poly-L-lysine-treated coverslips for 3 h. Coverslips were then washed in three 10 min intervals with phosphate-buffered saline (PBS) and then either used directly or stored for up to 24 h at 4°C. Jurkat cells in ACC growth medium at a cellular concentration of 1×10⁶ cells/ml were settled onto antibody-coated coverslips and used for imaging studies.

For immunostaining, cells settled on coverslips were fixed with PBS containing 1% paraformaldehyde for 30 min before being washed with PBS. Cells were then permeabilized with a 1:1 solution of ice-cold methanol and acetone for 15 min. Cells were washed again with PBS and blocked with PBS containing 5% goat serum and 0.5% Tween-20 for 30 min. Cells were then incubated with a 1:50 solution of primary antibody in blocking solution for 1 h. After primary antibody staining, the cells were then washed again with PBS and incubated with a 1:100 solution of secondary antibody or phalloidin-TRITC in blocking solution for another hour. After a final wash with PBS, the coverslips were mounted onto slides with ProLong Gold antifade reagent overnight at room temperature before being stored long-term at −20°C.

Jurkat cells meant to be stimulated prior to lysis and immunoprecipitation were pelleted at 750×g and resuspended in RPMI at a cellular concentration of 1×10⁶ cells/ml. Cells were treated with 500 ng/ml Vβ8 anti-TCR antibody and incubated at 37°C for 30 min. After that, cells were pelleted at 750×g and resuspended in lysis buffer containing 200 mM NaCl, 50 mM Tris pH 8, 2 mM EDTA, 2 mM NaVO₄, 20 mM NaF, 3 mM PMSF, 2 mM imidazole, 1 mM Na-β-glycerophosphate and 1% Triton X-100. The suspension was then passed through a 21-gage needle repeatedly to homogenize it and then clarified at 16,000×g and 4°C for 10 min to remove cell debris.

To prepare beads for immunoprecipitation, 5 µg antibody was added to 300 µl PBS and 80 µl of a 50% slurry of Protein A agarose beads. This solution was mixed gently on a rotator overnight at 4°C. The next day, beads were washed with cell lysis buffer for 15 min and added to cell lysate made from 1×10⁷ Jurkat cells as previously described. Cells were then incubated on a rotator at 4°C for 2 h. Afterwards, the beads were washed four times with PBS, diluted with SDS-PAGE loading buffer and boiled for 5 min. The lysate was then adjusted with SDS-PAGE loading buffer to a final concentration of 2% (w/v) SDS and 5% (v/v) β-mercaptoethanol.

Samples prepared in SDS-PAGE sample buffer were run through SDS-PAGE and transferred to nitrocellulose paper for western blotting. Samples were blocked in a blocking solution of Tris-buffered saline with 0.1% Tween and 5% BSA. Primary antibody was then added, diluted to a concentration of 1 µg/ml in blocking solution. Primary antibodies were tagged with HRP using a goat anti-rabbit or goat anti-mouse HRP-conjugated secondary antibody and treated with SuperSignal West Pico chemiluminescent substrate solution before being exposed to X-ray film.

Primary CD4⁺ T cells were derived from peripheral blood mononuclear cells and monitored on lipid bilayers, as described by Steblyanko et al. (2018). The procedure was adjusted by preparing bilayers with Cy5-ICAM1-His6 at 1 µg/ml in order to eliminate the high background fluorescence of Cy5 caused by the relatively low levels of LFA-1 expressed on the surface of Jurkat cells.

Images were viewed using a Nikon inverted microscope and captured using a CMOS camera (Andor). The images were processed and analyzed using the ImageJ processing software. To determine protein accumulation at the synapse, a line of length 230 pixels and width 70 pixels was drawn on the Jurkat–Raji cell pairs such that the mid-point of the line (pixel 115) was on the synapse. Background intensity was obtained from a region outside the cell pairs and subtracted from the fluorescence intensity. The fluorescence was normalized by dividing all the pixel measurements by the average intensity of the row furthest away from the synapse. Beginning at the synapse (row 115) and moving toward the opposite edge of the Jurkat cell (row 1), intensity values for 5 pixel groups were treated as one increment and used for statistical analysis (mean±s.e.m.). The compound mean and standard error for the increments was plotted against the mean intensity of fluorescence. A one-tailed t-test with independent variance was performed for the first five increments starting from the synapse and moving to the back of the Jurkat cell.

To determine the distance of the MTOC to the immunological synapse, the position of the MTOC was first determined by finding the signal maxima of fluorescence intensity within Jurkat cells immunostained for β-tubulin. Then, we measured the distance of the MTOC to the edge of the Raji cell, as determined through Cell Tracker Blue staining. This distance was recorded as micrometers converted from pixels (5 pixels=0.33 μm). A two-tailed t-test with independent variance was then performed comparing the MTOC polarization of 33 cell pairs from each cell type. Cell pairs of each cell type were grouped and plotted on a graph depending on the degree of MTOC polarization.

Cross-sectional imaging of the immunological synapse was taken with a Zeiss LSM 710 confocal microscope from the University of Texas ICMB core facilities. Image processing was carried out by reslicing z-stacks of images using ImageJ.

We thank Dr Jeffrey Kuhn (currently Scientific Director, Microscopy Core Facility, Koch Institute, Massachusetts Institute of Technology) for consultation and feedback given during the course of this project. We would also like to thank Dr Lauren Ehrlich (Molecular Biosciences, University of Texas at Austin) for providing OT-1 splenocytes, as well as Dr Jessica Lancaster (currently of the Mayo Clinic, Phoenix, AZ) for extracting splenocytes and providing consultation for cell culture methods.