|Home | About | Journals | Submit | Contact Us | Français|
The endoplasmic reticulum (ER) undergoes significant reorganization between interphase and mitosis, but the underlying mechanisms are unknown . STIM1 is an ER Ca2+ sensor that activates store-operated Ca2+ entry (SOCE) [2, 3], and also functions in ER morphogenesis through its interaction with the microtubule +TIP protein EB1 . We previously demonstrated that phosphorylation of STIM1 during mitosis suppresses SOCE . We now show that STIM1 phosphorylation is a major regulatory mechanism that excludes ER from the mitotic spindle. In mitotic HeLa cells, the ER forms concentric sheets largely excluded from the mitotic spindle. We show that STIM1 dissociates from EB1 in mitosis and localizes to the concentric ER sheets. However, a non-phosphorylatable STIM1 mutant (STIM110A) co-localized extensively with EB1 and drove ER mislocalization by pulling ER tubules into the spindle. This effect was rescued by mutating the EB1 interaction site of STIM110A, demonstrating that aberrant association of STIM110A with EB1 is responsible for the ER mislocalization. A STIM1 phospho-mimetic exhibited significantly impaired +TIP tracking in interphase, but was ineffective at inhibiting SOCE, suggesting different mechanisms of regulation of these two STIM1 functions by phosphorylation. Thus, ER spindle exclusion and ER-dependent Ca2+ signaling during mitosis require multi-modal STIM1 regulation by phosphorylation.
During interphase the endoplasmic reticulum (ER) is organized in interconnected tubules and sheets and associates extensively with microtubules (Figure 1A, A′) [1, 6], whereas the ER during mitosis consists predominantly of concentric sheets  and is largely excluded from the mitotic spindle in HeLa cells (Figure 1B, B′) [7, 8]. Mechanisms that underlie this remodeling during mitosis are unknown; however, exclusion from the spindle suggests that mechanisms that associate the ER with microtubules are negatively regulated. One mechanism of ER-microtubule association involves the ER Ca2+ sensor STIM1. During interphase, STIM1 mediates ER tubule extension through interaction with the microtubule +TIP protein EB1, which decorates microtubules in short linear structures known as comets . This interphase distribution can be seen using an enhanced yellow fluorescent protein-tagged STIM1 (STIM1WT) and EB1 staining (Figure 1C, C′). In mitosis, however, STIM1WT is largely absent from the EB1-dense mitotic spindle (Figure 1D), similar to the ER. We therefore hypothesized that ER exclusion from the mitotic spindle depends on inhibition of the interaction of STIM1 with EB1. In support of this, higher magnification of mitotic cells revealed no evidence of STIM1WT-EB1 colocalization (Figure 1D′), including at the spindle poles where a small accumulation of STIM1WT resides (Figure 1D′, arrow). Examination of subsequent stages of mitosis in fixed (Figure S1A) and live cells (Figure 2F and Movie S1) revealed that dissociation of STIM1WT from EB1 persisted through cytokinesis.
One mechanism for dissociation of +TIP proteins from EB1 is phosphorylation near the EB1 interaction site [10, 11]. We therefore hypothesized that the recently-described phosphorylation of STIM1 during mitosis  inhibits the interaction of STIM1 with EB1. Phosphorylation of STIM1 during mitosis occurs at S/T-P sites that are recognized by the MPM-2 antibody [12, 13], and several are located near the EB1 interaction site (Figure 2A and S1D). To inhibit mitosis-specific STIM1 phosphorylation, we mutated to alanine all 10 of the serines or threonines within candidate MPM-2 recognition sites (STIM110A). STIM110A immunoprecipitated from mitotic HeLa extracts failed to exhibit the molecular weight shift seen with STIM1WT and was not recognized by MPM-2 (Figure 2A), consistent with loss of mitosis-specific phosphorylation. We next analyzed the effect of STIM1 phosphorylation on EB1 interaction using a GST-tagged EB1 (GST-EB1) pull-down assay . Phosphorylated STIM1WT from mitotic cells interacted very weakly with GST-EB1 compared to STIM1WT from asynchronous cells, and this interaction was fully restored with non-phosphorylatable STIM110A from mitotic cells (Figure 2B). Thus, phosphorylation negatively regulates the STIM1-EB1 interaction. Accordingly, and in striking contrast to STIM1WT, STIM110A localized extensively to the spindle, both in interpolar and astral regions (Figure 2D). Higher magnification revealed STIM110A comets associated with EB1 comets (Figure 2D′), suggesting that spindle localization of STIM110A is due to EB1 association. Further, STIM110A showed +TIP tracking behavior during metaphase (Figure 2G, G′ and Movie S2), again suggesting association with and transport by EB1. To determine whether the altered mitotic localization of STIM110A is specifically due to EB1 association, we inhibited EB1 association by mutating isoleucine 644 and proline 645 of the T-R-I-P EB1 interaction site of STIM110A to asparagines (STIM110A_TRNN) . As expected, STIM110A_TRNN from mitotic cell extracts failed to interact with GST-EB1 (Figure 2B), and STIM110A_TRNN also failed to co-localize with EB1 in the mitotic spindle (Figure 2E). Quantitation revealed that greater than 60% of metaphase cells exhibited STIM110A localization in the spindle compared to less than 20% for STIM1WT or STIM110A_TRNN (Figure 2H), showing consistent mislocalization of STIM110A and rescue by STIM110A_TRNN. Examination of subsequent mitotic stages in fixed (Figure S1A) and live cells (Figure 2F and Movie S1) revealed STIM110A association with the central spindle through abscission. STIM110A also localized to the mitotic spindle in metaphase HEK293 cells (Figure S1B), suggesting similar regulation of STIM1 during mitosis in several cell types. These data collectively indicate that phosphorylation is necessary to dissociate STIM1 from EB1 during mitosis.
The ten putative phosphorylation sites mutated in STIM110A span nearly 200 amino acids, but only those closest to the EB1 interaction domain may actually regulate EB1 association. To begin to address this, we mutated to alanine serine 668, the site closest to the EB1 domain that’s clearly phosphorylated during mitosis , and serine 628, the site closest to the EB1 domain with unknown phosphorylation status (STIM12A) (Figure S1D). STIM12A partially localized to the mitotic spindle (Figure S1C), whereas mutation of either site alone did not cause spindle localization (not shown). Therefore, S628 and S668 are likely both phosphorylated during mitosis and coordinately regulate EB1 interaction. However, STIM12A localization to the spindle was not as extensive as STIM110A (Figure S1C), suggesting that additional sites are likely phosphorylated and contribute to regulation of the EB1 interaction.
We next asked whether phosphorylation-dependent dissociation of STIM1 from EB1 is an important determinant of spindle exclusion of the ER. Consistent with previous studies [7, 8], the ER was largely excluded from the spindle from metaphase through telophase in wildtype HeLa cells based on localization of the ER marker GFP-Sec61β (Figure S2A; also Figure 1B). Small clusters of ER often emanated from the spindle poles, but this represents only a very small fraction of total ER . STIM1WT expression did not alter mitotic ER organization based on localization of the ER marker calreticulin in fixed cells (Figure 3A) and ER-targeted dsRed in live cells (Figure 3B and Movie S3). Further, siRNA depletion of STIM1 had no discernable effect on mitotic ER localization (Figure S2B, B′). Thus, ER localization during mitosis, including the small clusters at spindle poles, is independent of STIM1, suggesting that phosphorylation of STIM1 in wildtype mitotic cells completely suppresses its function.
In contrast, STIM110A-expressing cells exhibited extensive ingression of ER tubules into the metaphase spindle (Figure 3A, C, and Movie S4). Spindle-localized ER tubules were seen in 64% of STIM110A-expressing metaphase cells, compared to 18% for STIM1WT (Figure 3D). In live imaging of STIM110A-expressing cells, numerous ER tubules extended through the contractile ring during abscission, in contrast to the limited localization of ER inside the contractile ring with STIM1WT (Figure 3B, C, arrows). Thus, phosphorylation of STIM1 is necessary for exclusion of the ER from the mitotic spindle through cytokinesis. Furthermore, STIM110A_TRNN rescued spindle exclusion compared to STIM110A (Figure 3A, D), demonstrating that altered ER distribution caused by STIM110A is due to association with EB1.
Significantly, STIM110A provides the first opportunity to examine whether ER mislocalization during mitosis affects cell division. One possibility is that substantial augmentation of spindle-localized ER tubules affects spindle organization or function. However, STIM110A did not alter spindle morphology (Figure 3A) or DNA congression to the metaphase plate (Figure 2F and Movie S1). Further, despite accumulation of ER tubules through the contractile ring in STIM110A-expressing cells, cytokinesis was not visibly impaired (Movies S3 and S4). Alterations to the timing of mitosis, if observed, would also suggest mitotic defects. However, there was no difference in the time of progression from metaphase through telophase in cells expressing STIM110A compared to STIM1WT (Figures 2F and 3B, C). To analyze the timing of mitosis more robustly, we calculated mitotic indices of asynchronous cells. Mitotic index reflects the average time spent in mitosis of a cell population. Consistent with the live imaging results, the mitotic index of STIM1WT-expressing HeLa cells was 5.74 ± 0.61% compared to 4.11 ± 0.49% for STIM110A, not a statistically significant difference (P>0.05, Student’s t-test). There was also no significant difference in the mitotic index of BJ cells, a non-transformed, primary human fibroblast cell line (0.91 ± 0.40% for STIM1WT versus 0.72 ± 0.14% for STIM110A; P>0.05, Student’s t-test). Thus, ER mislocalization driven by STIM110A did not cause obvious mitotic defects in the cell types tested. However, changes in ER localization may be detrimental in other physiological contexts or cell types. Notably, cell division in 2-dimensional cell culture may not recapitulate the process as it occurs in 3-dimensional tissues, particularly with respect to the structure and partitioning of cellular components. Further, cellular defects, such as chromosome instability or altered organelle inheritance, may accumulate over multiple divisions due to ER disorganization, with dire consequences that may not have been apparent here. These possibilities can be addressed by generating STIM110A-expressing transgenic animals.
To determine whether phosphorylation is sufficient to dissociate STIM1 from EB1, we created a putative phospho-mimetic STIM1 by mutating all 10 MPM-2-directed serines and threonines to glutamates (STIM110E) . We hypothesized that STIM110E would not +TIP tracking during interphase, indicative of loss of EB1 interaction. Consistent with this, the robust +TIP tracking seen with STIM1WT was nearly completely absent with STIM110E; only a limited number of short-lived STIM110E comets were visible (Figure 4A and Movie S5). Loss of +TIP tracking of STIM110E, however, was not as complete as with STIM1TRNN, wherein the EB1 interaction site is mutated (Figure 4A and Movie S5). Therefore, factors in addition to phosphorylation, such as alterations to microtubule dynamics or cellular geometry, may also contribute to STIM1 dissociation from EB1 during mitosis to a limited degree. Alternatively, glutamates may not fully mimic phosphorylation , and this may also account for the remaining +TIP tracking of STIM110E. Nonetheless, the significant abrogation of +TIP tracking with STIM110E, in combination with the strong association of STIM110A with EB1 during mitosis, demonstrates that phosphorylation is the primary mechanism that dissociates STIM1 from EB1 during mitosis.
The effects of mitosis-specific phosphorylation on STIM1 function are not limited to EB1 dissociation, as phosphorylation has also been suggested, based on a STIM1 truncation mutant, to suppress SOCE activation by STIM1 during mitosis; the functional significance of this remains unknown [5, 17]. In agreement with this, we found that our non-phosphorylatable STIM110A fully supported SOCE in mitotic cells whereas STIM1WT did not (Figure S3). This is likely not due to restoration of the EB1 interaction by STIM110A, as STIM1 activation of SOCE is independent of its EB1 interaction . It is possible, therefore, that phosphorylation differentially regulates these two independent STIM1 functions. In support of this, we show that STIM110E is a poor phospho-mimetic for inhibition of SOCE activation, as STIM110E activates SOCE responses in interphase cells that are not significantly different from those activated by STIM1WT (Figure 4B, C). Therefore, glutamates more effectively recapitulate the effects of phosphorylation on the STIM1-EB1 interaction than on SOCE activation, suggesting that the mechanisms by which phosphorylation regulates these two functions, such as conformational changes or altered oligomerization, likely differ. Future work will focus on delineating which phosphorylation sites regulate the STIM1-EB1 interaction and which regulate SOCE activation.
The association of the ER with microtubules is remodeled during mitosis, although the extent of ER exclusion from the mitotic spindle varies by cell type [7, 18–23]. Mechanisms that underlie this remodeling are unknown and accordingly, the functional significance is also unclear. Our finding that phosphorylation of STIM1 is an important determinant of ER reorganization during mitosis is therefore a significant advance, as it provides the first mechanistic insight into mitotic ER structural remodeling. Further, STIM110A, which results in ectopic localization of ER tubules in the mitotic spindle, offers a molecular tool with which to probe the functional consequences of ER mislocalization during mitosis. Future studies will use STIM110A to address the importance of proper mitotic ER partitioning and function for normal organismal development and physiology. It will also be important to understand how other associations of the ER with microtubules are modified during mitosis, as STIM1 phosphorylation is likely only one of many mechanisms that regulate mitotic ER morphology.
STIM1 now joins a growing list of +TIP proteins whose interactions with EB1 are regulated by phosphorylation [11, 24], suggesting a paradigm of functional remodeling of microtubule tip-associated complexes during specific cellular processes. Interestingly, +TIP tracking by the EB1-interactors CLASP2 and SLAIN2 is also inhibited during mitosis by multi-site phosphorylation [24, 25]. Thus, remodeling of the EB1-associated complex of +TIPS may be necessary during mitosis, and the functional significance of this remodeling requires significant future investigation.
This research was supported by the Intramural Research Program of the NIH, National Heart, Lung and Blood Institute and National Institute of Environmental Health Sciences. We acknowledge Dr. Christian Combs in the NHLBI Light Microscopy Core Facility, Jeff Tucker and Dr. Agnes Janoshazi in the NIEHS Confocal Core Facility, and Leigh Samsel and Dr. Pradeep Dagur in the NHLBI Flow Cytometry Core Facility for assistance with experiments. We also thank Drs. Brian Galletta, Dorothy Lerit, Stephen Shears, and Carmen Williams for helpful comments on the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.