Septin filaments assemble on the cortex, enriched in the T cell mid-zone
We first identified T cell expression of Sept9 in a gene-trap screen21
for molecules that polarized in T cells (data not shown). Analysis of septin expression in a murine T cell line (Fig. S1A
) by RT-PCR indicated expression of multiple Septins, comprising one from Group I (Septin 9), three from Group II (Septins 6,8,11), two from Group III (Septins 2 and 4) and one from Group IV (Septin 7). This RNA expression pattern is consistent with expression profiling of human septins22
. Using a panel of anti-septin specific polyclonal antibodies, we also detected Septin 1,6,7,8 and 9 protein. We did not have suitable reagents to assess Sept10 expression, and other septins were not present above our level of detection (Fig. S1B
and data not shown). As septins are known to form heteromeric complexes23–26
, we looked for septin complexes in lysates from D10 T cells. Immunoprecipitation of Septins 1,6,7 or 9 caused co-precipitation of complexes containing the majority of Septins 1,6,7,8, and 9 respectively ( and S5a
). Although this method does not reveal the diversity of filament compositions, individual septins were clearly involved in interactions with aggregates containing every other septin.
Septin Complexes form in T cells, assemble on the cortex and form filaments and puncta in the mid-zone
Four anti-septin antibodies proved suitable for widefield immunofluorescence. These grossly highlighted the cell cortex in crawling T cells and were frequently enriched through the mid-zone, forming an annular septin ‘corset’. This pattern was most prevalent for Septins 7 and 9 and less so for Septins 1 and 6, (). More rarely and only for Sept9, we observed staining inside the cell near the MTOC (data not shown). Similar distributions were observed in primary cells (Fig. S2e
Higher resolution confocal imaging revealed that the mid-zone distribution was highlighted by denser arrays and fibers that were partially perpendicular to the axis of travel ( and Movie SM1
). These septin densities appear similar to the stays of a corset or sail, latitudinally-oriented features in an otherwise uniform band. Total internal reflection fluorescence microscopy (TIRF) indicated that regions of adhesion also contained these fibrous arrays (). Fibers are near the limit of optical resolution (≤250nm) in width and vary from short puncta to filaments longer than 3 microns. These enrichments are not simply areas of higher membrane density, as septin staining did not colocalize with membrane dyes such as DiO (Fig. S2a
). In some cells, particularly for Septin6, distributions were predominantly punctate (Fig. S2b
). These may represent shorter fibers, loosely connected septin structures, subcortical vesicles or another subregion of the greater septin array. Septin puncta were also the dominant distribution in rounded, non-motile T cells (Fig. S2d
), suggesting a functional relationship of the Septin corset to T cell crawling. Finally, we occasionally (see ) observed cells with excess leading-edge protrusions in wild-type T cells, and Septins, notably Septin 6, were present in these protrusions (Fig. S2c
). In sum, a cortical array of septins decorates the surface of T lymphocytes.
Septin complexes are required for structural stability of the cell cortex and at the T cell mid-zone
Global Morphological and Cortical Defects in the Absence of Septins
To assess the function of septins in T cells, we targeted Septins 1, 6, 7 and 9 using plasmid-based short hairpin RNAs (shRNAs). We achieved significant knock-down (KD) of these Septins in D10 T cells when each was specifically targeted ( and S5b
). In addition, targeting of Sept7 eliminated the entire septin complex (). This result was specific to Sept7, as the reciprocal was not observed when other septins were targeted, supporting the hypothesis that Sept7 is critical to septin assembly 23–25,27
Septin knock-down in T cells results in augmented length and bending of uropods
To confirm that these results were specific to Sept7KD, we generated another shRNA targeting Sept7 (KD #2) at a distinct site. Expression of either shRNA resulted in similar reductions in all septins (). We reproducibly achieved greater than 80% KD of Septins 6,7,8, and 9, and over 70% of Sept1 using either shRNA. Although septin knockdown causes cell division defects in other systems 26–28
, Sept7KD in D10 T cells did not show such defects during the 72 hour period in which our observations were made (Fig. S3a
While there was no apparent phenotype of individual Septin 1, 6, or 9 deficiencies, the Sept7KD-induced loss of all septins caused profound disruption in cellular morphology and motility. The most obvious morphological disruption was in cell length (). The average uropod length in KD cells was 14.4 ± 4.3 μm for Sept7KD#1 (n = 107) and 15.1 ±3.7 μm for Sept7KD #2 (n = 108) versus 8.8 ±2.0 μm for control shRNA treated cells (n = 97). In contrast cell body length was unchanged in Sept7KD cells (), showing that Septins were not globally controlling cell size but were particularly necessary to regulate the mid-zone/uropod. The phenotype was independent of adhesion, as cells floating in suspension were also elongated (data not shown).
The long uropods of Sept7KD cells also frequently appeared bent (). This was observed in 32% of Sept7KD#1 (n=107) and 34% of Sept7KD#2 (n=108) cells, but only 10% of control cells (n=97). This further suggests a stabilizing role for septins in maintaining and coordinating the morphology of motile cells.
Defects due to septin disruption were not confined to the uropod and also suggested poor structural integrity of the cortex. By analyzing time-lapse images ( and Movies SM2, SM3 and SM4
), we quantified two defects in the leading edge region: membrane blebbing and excess protrusions, both of which were uncommon in wild-type T cells. As shown in , membrane blebbing was very dynamic and mainly confined to the cell body. Blebbing was observed in 29.7% (KD #1, n = 118) and 37.1% (KD #2, n = 140) of Sept7KD T cells versus 9.9% of the control sample (n = 151) (). As visible in and movie SM3
, cells frequently blebbed during pauses in motility, perhaps due to poor coordination of compression and elongation. Though blebbing is often associated with apoptosis, Sept7KD and control cells had similar Annexin V staining (Fig. S3b
), suggesting that blebs associated with septin-deficiency arose through another pathway. Time-lapse analysis also demonstrated the formation of long, thin appendages that emanated from both the leading edge and the mid-zone in Sept7KD cells before eventually being re-absorbed. Although these protrusions occurred less frequently than membrane blebbing (), there were significantly more excess protrusions in Sept7KD T cells (KD #1 = 9.3%, KD #2 = 15.0%) than in the control treated population (2.0%).
Normal Tubulin and Actin Cytoskeletal Activities in the Absence of Septins
Since many of the defects in Sept7KD cells resemble those of an overactive cytoskeleton, we analyzed the localization and activity of a complement of tubulin and actomyosin-related proteins. Tubulin and pericentrin staining were grossly normal in Sept7KD cells, though bundled microtubules were elongated through their uropods (). Staining for pericentrin revealed normal positioning of the microtubule-organizing center (MTOC), directly behind the nucleus (). These data indicate that the Sept7KD phenotype was not due to disorganization of microtubules.
Integrity of Tubulin and Actomyosin Cytoskeletons in Septin-deficient T cells
Though an overactive actomyosin cytoskeleton could cause the blebbing, elongation and protrusions observed in septin-deficient T cells, we observed a 10% to 30% reduction in the levels of F-actin in septin-deficient cells, as assessed by phalloidin staining (), which was not due to changes in total actin protein level ( and S5c
). Phalloidin staining of T cells was grossly normal, with polymerized actin lining the cortex and mildly enriched at the mid-zone in both control and Sept7KD cells (). There were also no detectable differences in the phosphorylation state of myosin light chain (MLC) or myosin heavy chain IIA (Myh9) or changes in the overall levels of myosin II in Sept7KD cells ( and S5c). Likewise, in both control and Sept7KD cells, phospho-MLC staining was uniformly distributed, and mildly enriched in the cell body relative to the uropod (). We would have expected additional enrichment of phospho-MLC in the mid-zone/uropod if Septins directly regulated MLC activity there. Despite these results, the blebbing, elongation and protrusions of Sept7KD cells strongly required myosin II activity. Treatment of Sept7KD and control cells with the myosin II inhibitor Blebbistatin or a Rho Kinase inhibitor Y-27632, essential for MLC activity in the uropod, resulted in complete loss of the uropod and existing blebs (, and Movies SM5-SM6
Defective Motility in the Absence of Septins
Sept7KD T cells exhibited reduced instantaneous crawling velocities in vitro
(). Sept7KD T cells crawled 7.5 ±1.9 μm/min (KD #1, n = 36) and 7.0 ±1.8 μm/min (KD #2, n = 35) compared to 8.5 ±2.6 μm/min for control cells (n = 30). More significantly, they had shorter net displacements than control cells (Control: 15.0 ± 6.1μm/min1/2
, KD #1:10.81 ± 5.8 μm/min1/2
, KD #2: 10.85 ± 4.5 μm/min1/2
). This result is surprising with respect to the findings of and demonstrate that despite functional actomyosin contraction, septin-deficient cells lack normal processivity. These defects were not due to decreased integrin binding by Sept7KD cells (Fig. S4a
) and primary Sept7KD T cells exhibited similar defects under shear flow conditions (Fig. S4b
). Since Sept7KD cells were more protrusive overall than control cells, we wondered whether Sept7KD cells were making non-productive protrusions outside the direction of motility that impeded their processivity. To address this, we measured the volume of motile cells that extended outside the path of motility and, therefore, did not propel the cell forward. We labeled the nuclei of GFP-transfected cells with Hoechst 33342 and acquired confocal three-dimensional time-lapses of crawling cells. We then created volumes representing the paths occupied by the nuclei during 15-minute movies (, blue), showing the overall directions of the cells. The GFP volumes were overlaid on the nuclear path volumes at each time point and the total volumes (, red outline) and the volumes extending outside the nuclear path were measured (, green). The total volumes of control and Sept7KD cells were similar, but on average Sept7KD cells had about 1.5 times the extra-nuclear path volume of control cells (median volumes of 167.8 μm3
and 249.5 μm3
for control and Sept7KD, respectively), supporting the hypothesis that their excess protrusions do not contribute to processive motion and, therefore, contribute to diminshed motility in Sept7KD cells.
Septins regulate motility and transmigration
Enhanced Transmigration in the Absence of Septins
Although specific chemotaxis of D10 cells to MIP-1α is poor, control and Sept7KD T cells responded equivalently to this chemokine (). However, Sept7KD exhibited two-fold higher transmigration than control cells through 8 μm pores in the absence of chemokine. Since increased motility cannot account for this greater background transmigration (), we hypothesized that these cells passed through the pores, which are similar in dimension to the cells girth, more efficiently than control cells. When we challenged the cells with transwell pores considerably smaller (3 μm and 5μm) than their smallest dimension, control treated D10 cells were essentially unable to pass through the small pores (2.0% and 3.7% migration, respectively). However, Sept7KD cells migrated through 3 and 5 μm pores approximately fivefold better than control cells (Sept7KD#1: 10.7% and 15.7% migration, respectively and, Sept7KD#2: 8.8% and 14.1% migration, respectively) and comparably to wild-type cells facing nearly 7× larger openings (). This shows that septins restrict chemokine-independent migration and suggests a cell-intrinsic barrier function of the septin cytoskeleton for cells encountering small junctions.
Because D10 T cells exhibit little specific chemotaxis, we expressed a Septin7 shRNA in primary T cell blasts, which are more suitable to chemotaxis assays. With knockdown, these cells adopted the same extended-uropod phenotype as septin-deficient D10 cells (.) In a chemotaxis assay using 5 μm pores, we found that septin-deficient primary cells transmigrated at a rate double that of control cells (). This was accompanied by a small increase in their transmigration in the absence of chemokine.
Transmigration Efficiency is Correlated with Cortical Rigidity
Because of the extreme compression of large cells migrating through small pores, we hypothesized that a loss of cortical rigidity in septin-deficient cells allowed them to transmigrate more efficiently. To investigate this possibility, we sought to pharmacologically mimic or counteract the septin-based loss of rigidity. We treated D10 cells with nocodazole to inhibit microtubule polymerization and relax the cell cortex or with taxol to stabilize microtubules and rigidify the cortex. Interestingly, taxol treatment has previously been shown to increase cell-permeation of collagen gels29
. After nocodazole treatment, both control and Sept7KD cells maintained their uropods, but appeared less compact and often generated multiple leading edges (.) Conversely, when treated with taxol, both groups had few protrusions, and most became rounded (.)
Septins and microtubules regulate rigidity and transmigration
Time-lapse images of nocodazole-treated cells indicated that they exhibit a similar protrusive phenotype to Sept7KD cells. While control cells generally made one or two small protrusions in a 10-minute period, Sept7KD and nocodazole-treated cells frequently made four or five (), and the persistence of each protrusion was significantly longer for Sept7KD and nocodazole-treated than control cells (). Like Sept7KD cells, protrusions in nocodazole-treated cells were uncoupled from the processive motion of the cell. While protrusions frequently arose at the leading edge, their growth was not restricted to that region and they extended without adhesion in many directions, sometimes reaching so far back before being retracted that the uropod of the cell took on a forked appearance (). On average, these protrusions did not extended farther from the cell body than the less frequent protrusions in control cells, though occasionally, very long (>15 μm) protrusions arose from Sept7KD and nocodazole-treated cells, but not control cells (). Additionally, the means shown in may underestimate the true population means for Sept7 KD and nocodazole-treated cells because the most severely abnormal protrusions often extended out of the imaging plane, excluding them from this analysis.
Since the cortical phenotype after nocodazole treatment was like that of SeptKD cells, we compared the migration of Sept7KD and nocodazole or taxol-treated D10 cells through 3 μm transwell pores in the absence of chemokine (). Interestingly, nocodazole-treated control D10 cells transmigrated comparably to untreated Sept7KD cells (8.2 ± 0.7% versus 11.8 ± 0.3% and 12.5 ± 1.8% for KD #1 and KD #2, respectively) while few (1.9 ± 0.1%) control cells migrated. When Sept7KD cells were treated with nocodazole, the effect was greater than additive, with (35.1 ± 1.9% of KD #1 and 30.9 ± 7.7% of KD #2 cells). Conversely, when cells were stiffened with taxol, less than 1.5% passed through the pores, even among Sept7KDs. These data support the hypothesis that septin depletion relaxes the cell cortex, allowing highly efficient movement through spaces much smaller than the resting cell diameter.