TIG3 interaction at the centrosome alters microtubule distribution and centrosome function

The TIG3 tumor suppressor protein was originally discovered as a growth suppressor in human keratinocytes (DiSepio et al., 1998). Also known as RIG1 and H-Rev107-2 (Huang et al., 2000; Jiang et al., 2005; Ou et al., 2008; Tsai et al., 2006; Tsai et al., 2007), TIG3 displays homology to the human, mouse and rat forms of H-rev107, and is a member of the H-rev family of class II tumor suppressors, and the NlpC and P60 superfamily (Deucher et al., 2000; DiSepio et al., 1998). These proteins include an N-terminal hydrophilic domain and a C-terminal membrane-anchoring domain (Deucher et al., 2000; DiSepio et al., 1998). The N-terminal domain encodes NCEHFV and LRYG sequence motifs that are conserved among family members (Anantharaman and Aravind, 2003). Consistent with a role in reducing cell survival, the level of TIG3 is reduced in hyperproliferative keratinocytes that are present in psoriatic lesions and skin tumor cells (Duvic et al., 1997; Duvic et al., 2000; Duvic et al., 2003). Treating psoriatic lesions with vitamin-A-related ligands increases the level of TIG3, which is associated with reduced disease severity (Duvic et al., 1997; Duvic et al., 2000; Duvic et al., 2003).

Absence of TIG3 expression in monolayer keratinocytes is thought to be necessary for the maintenance of proliferative potential (Sturniolo et al., 2003; Sturniolo et al., 2005), and TIG3 expression in monolayer cultures causes cell death (Sturniolo et al., 2003; Sturniolo et al., 2005). TIG3 is present at high levels in differentiated human keratinocytes in suprabasal epidermis and in raft cultures (Jans et al., 2008). TIG3-dependent death is associated with activation of selected differentiation-associated processes. For example, TIG3 localizes to the plasma membrane where it interacts with type I transglutaminase, an interaction that leads to increased transglutaminase activity and cornified envelope formation (Sturniolo et al., 2003; Sturniolo et al., 2005). Mutagenesis studies indicate that TIG3 mutants lacking the C-terminal membrane-anchoring domain are not active (Deucher et al., 2000; Sturniolo et al., 2003; Sturniolo et al., 2005). By contrast, N-terminal truncation converts TIG3 into a protein that causes keratinocyte apoptosis (Jans et al., 2008). TIG3 also suppresses the proliferation of keratinocytes (Sturniolo et al., 2003; Sturniolo et al., 2005), although very little is known about the mechanism of suppression.

The centrosome is an important organelle located adjacent to the nucleus, which serves to nucleate microtubule arrays that organize cytoplasmic organelles and primary cilia in interphase cells, and form the mitotic spindles during mitosis (Doxsey et al., 2005). It includes two perpendicularly-oriented barrel-shaped centrioles surrounded by pericentriolar material (PCM) (Doxsey et al., 2005; Kreitzer et al., 1999). Centrosome function is essential for cell survival and cell division. Our studies suggest that association of TIG3 with the centrosome alters microtubule distribution, increases microtubule stability, reduces microtubule anterograde elongation and suppresses daughter centrosome separation. We propose that TIG3 interaction at the centrosome is a crucial event in TIG3-dependent cessation of cell proliferation.

As shown in Fig. 1A, TIG3 distributes at three main locations in cells; the plasma membrane, punctate intracellular foci and at a perinuclear location. The perinuclear location is identified by arrows in Fig. 1A. We showed previously that the hydrophobic C-terminus of TIG3 is required for correct TIG3 subcellular localization (Deucher et al., 2000). Indeed, Fig. 1A shows that a mutant lacking the C-terminus, TIG3 (1–134), displays a diffuse cytoplasmic distribution and does not display perinuclear accumulation. We have not studied this mutant further in the present studies, because it is inactive and has no ability to modulate cell function (Jans et al., 2008; Sturniolo et al., 2003; Sturniolo et al., 2005). Because TIG3 suppresses cell proliferation, we are particularly interested in the intense staining that is observed adjacent to the nucleus (Fig. 1B, arrows). This location suggests that TIG3 interacts with the centrosome. To test this possibility, we stained cells to detect TIG3, the centriole protein, γ-tubulin and a centrosome marker, pericentrin. Fig. 1B shows that TIG3 staining surrounds γ-tubulin and pericentrin (arrows) (Nigg and Raff, 2009). Quantitative cell counts revealed centrosomal localization of TIG3 in 93±2% of TIG3-expressing cells. Additionally, cells were stained with the Golgi markers GM130 and mannose-6-phosphate receptor (M6PR), and calnexin, an endoplasmic reticulum marker. GM130, M6PR and calnexin stained structures in the vicinity of the centrosome, and some of this staining colocalized with TIG3 staining. The Manders' overlap coefficient was calculated to determine the extent of TIG3 localization to these organelle markers (Fig. 1C). TIG3 displays high localization to the centrosome markers and low localization to the nucleus. In addition, some TIG3 localizes to M6PR, GM130 and calnexin. Thus, TIG3 colocalizes mostly with the centrosome but some TIG3 also localizes with the Golgi complex and endoplasmic reticulum.

As part of the search for a centrosome-specific role for TIG3, we examined the TIG3 impact on microtubules. The centrosome is a crucial controller of microtubule function during mitosis and in interphase cells (Doxsey et al., 2005). Therefore, we assessed whether localization of TIG3 to the centrosome perturbs the distribution of microtubules. Fig. 2 compares the microtubule network in cells infected with an empty vector (EV) and cells infected with a TIG3 vector. Cells infected with an empty vector display a typical microtubule network, which includes a perinuclear halo organized at the centrosome that sweeps around the nucleus and out to the cell periphery (Fig. 2, left). By contrast, in TIG3-expressing cells, the microtubules distribute as a broad band at the cell periphery linked to the centrosome by thin microtubule threads (Fig. 2, right).

These findings indicate that TIG3 influences microtubule distribution, and suggest that it also influences other microtubule properties. We used four approaches to assess the impact of TIG3 on microtubule stability. First, we examined the effect of nocodazole on microtubule integrity. Microtubules in TIG3-negative (EV) cells are distributed throughout the cell (Fig. 3A). Nocodazole treatment produces diffuse high-intensity staining, which is typical of dissociated microtubules, and, as expected, withdrawal of nocodazole results in the formation of asters (sites of microtubule reassembly) at the centrosome (arrows) (Fig. 3A). By contrast, in TIG3-expressing cells, microtubules concentrate at the centrosome and in a thick band at the cell periphery with thin microtubule projections linking these locations (Fig. 3B). These microtubules display a different response to nocodazole. Nocodazole treatment eliminates the peripheral tubulin band in most cells, but the centrosome-localized asters (arrows) survive nocodazole treatment (Fig. 3B, right panel). The merged images show TIG3 (arrows) accumulation at the centrosome (Fig. 3B, left panel). To provide quantitative information, we infected cells with tAd5-EV or tAd5-TIG3 and after 24 hours counted the number of cells displaying a peripheral tubulin ring, tubulin at the centrosome or tubulin at both locations. In untreated cells, microtubules localized to the centrosome are visible in 19.8±3.1% of cells and no cells have a peripheral microtubule ring (Fig. 3C). This value is probably an underestimate, as the asters can be obscured in cells with robust microtubule staining. By contrast, 43.8±4.7% of TIG3-positive cells have centrosome-anchored tubulin filaments, and a microtubule peripheral ring is present in 45.9±5.5% of TIG3-positive cells. The majority of these cells (39.4±4.6%) are microtubule-positive at both locations. Few microtubule asters and no rings are detected in nocodazole-challenged EV cells, but 49.8±9.5% of TIG3 cells retain a microtubule aster (Fig. 3C). During recovery from nocodazole, asters are present in 47.0±3.4% of EV cells and 35.5±5.4% of TIG3 cells. In addition, the microtubule ring reforms in 31.2±2.2% of TIG3 cells. An interesting finding is that ~50% of all TIG3 cells lose immunologically detectable microtubules (Fig. 3B,C, Loss). To assess the reason for this loss, we prepared extracts for immunoblot detection of β-tubulin. These studies revealed no change in the level of β-tubulin, indicating that the actual level of β-tubulin per cell is not reduced (Fig. 3D). These findings suggest that microtubule status is very different in EV and TIG3 cells.

As a second method to assess microtubule status, we measured the amount of α-tubulin in the insoluble (pellet) fraction. Previous studies indicate that polymerized or stabilized α-tubulin distributes in the 15,000 g pellet fraction (Onishi et al., 2007). At 24 hours after tAd5-EV or tAd5-TIG3 infection, keratinocytes were harvested, total extract and pellet fraction were prepared, and α-tubulin level was monitored in each fraction. These experiments show a substantial increase in the level of α-tubulin that is present in the pellet fraction in TIG3-positive cells (Fig. 4A). As a third method, we measured the effect of TIG3 on α-tubulin acetylation and detyrosination. Detyrosination of α-tubulin, to form glu-α-tubulin, is associated with increased microtubule stability, as is acetylation of α-tubulin (Bulinski and Gundersen, 1991; Kreitzer et al., 1999; Maruta et al., 1986; Thyberg and Moskalewski, 1999). Fig. 4B shows that TIG3 expression leads to increased levels of acetylated α-tubulin and glu-α-tubulin. To determine whether the acetyl-α-tubulin is localized to a particular region of the microtubule network, we stained EV and TIG3 cells with antibodies against acetyl-α-tubulin. Fig. 4C shows that acetyl-α-tubulin is distributed throughout the cell in EV cells. In TIG3-positive cells it is distributed in a ring at the cell periphery and at the centrosome. Anti-β-tubulin staining is included to confirm microtubule distribution (Fig. 4C). Monitoring acetyl-α-tubulin distribution in individual cells (Fig. 4D) reveals that acetyl-α-tubulin distributes at the cell periphery and centrosome in TIG3-positive cells. Thus, the level of acetylated-α-tubulin is increased in TIG3-positive cells and is present in the tubulin network at both the centrosome and peripheral ring, and the level of glu-α-tubulin is also increased. These studies suggest that microtubules are stabilized in cells that express TIG3.

As a fourth approach, we determined whether TIG3 affects microtubule growth using the microtubule plus end binding protein EB1–GFP to monitor anterograde microtubule extension (Dixit et al., 2009; Piehl et al., 2004; Piehl and Cassimeris, 2003). EB1–GFP binds specifically to the growing plus end of microtubules and can be used to trace movement of the leading tip of the microtubule as it grows towards the cell periphery (Dixit et al., 2009; Piehl et al., 2004). Keratinocytes were transfected with pEB1-GFP in the presence of pcDNA3 or pcDNA3-TIG3. At 18 hours post-transfection, the cells were monitored for EB1–GFP distribution by fluorescence confocal microscopy. EV cells display robust plus end microtubule growth (Fig. 5, EV). By contrast, TIG3-expressing cells display substantial EB1–GFP accumulation in the vicinity of the centrosome (arrows) with reduced plus-end growth towards the cell periphery. These results suggest that TIG3 reduces anterograde microtubule extension and that extension of many microtubules is halted before extension is complete. In addition, EB1–GFP appears to label multiple foci in the vicinity of the centrosome, suggesting that the structure of the centrosome nucleation site(s) have changed.

The centriole and centrosome play a crucial role at all stages of the cell cycle (Lim et al., 2009; Loncarek et al., 2008; Sekine-Suzuki et al., 2008). Centrosomes replicate simultaneously with nuclear DNA during S phase (Doxsey et al., 2005) and during prophase of mitosis, and the daughter centrosomes separate and move to opposite poles of the mitotic cell (Doxsey et al., 2005; Lim et al., 2009; Loncarek et al., 2008). Because of the role of centrosomes and microtubules in this process, an obvious expectation is that TIG3 might impede these processes. Indeed, our studies suggest that TIG3 interferes with centrosome separation. Keratinocytes were infected with TIG3-expressing virus, and after 24 hours they were stained with anti-TIG3 and anti-γ-tubulin antibodies. Fig. 6A shows that centrosomes separate (rectangle) in TIG3-negative cells. By contrast, daughter centrosomes appear to be closely spaced and not separated in TIG3-positive cells (arrows). In fact, centrosome separation is rarely observed in TIG3-positive cells. Cell counting of centrosome status in EV and TIG3 cells reveals centrosome separation in 16±2% of TIG3-negative cells, but in <1% of cells that express TIG3 (Fig. 6A).

TIG3 could either delay or prevent centrosome separation. To distinguish between these possibilities, we monitored the impact of TIG3 on centrosome separation as a function of time. In EV-infected cells, separated daughter centrosomes are observed at time points where cells are undergoing division, including 16, 24 and 48 hours (Fig. 6B). The number of cells displaying separated centrosomes ranges from 23.8±4% to 10.3±5% in EV-infected cells. The number is slightly lower in cells observed at 72 hours, as these cells are nearly confluent. By contrast, the number of cells with separated centrosomes is markedly reduced in cells that express TIG3 (Fig. 6C). These findings suggest that TIG3 does not delay centrosome separation, but prevents it. We also measured whether TIG3 has an impact on centrosome duplication by counting the number of EV cells and TIG3-expressing cells that contain duplicated centrosomes. Fig. 7A shows that centrosome duplication is not significantly altered in cells that express TIG3. Fig. 7B shows a TIG3-positive cell displaying centrosome localization of TIG3 (left panel), and an expanded higher magnification image shows the duplicated centrosomes (right panel).

Next, we assessed the role of microtubules in regulating centrosome separation. Our goal was to assess whether TIG3 suppresses centrosome separation through an impact on microtubules. To investigate this, we monitored the impact of the microtubule disruptor, nocodazole, on centrosome separation in control (EV) and TIG3-expressing cells. In EV-infected cells, nocodazole treatment decreased the number of cells that displayed separated centrosomes from 20±7% to 14±8%, a number that returned to 27±6% following the removal of nocodazole (Fig. 8, recovery). These findings indicate that an intact microtubule network is necessary for efficient centrosome separation, and that microtubule status influences centrosome function. By contrast, centrosome separation was observed in <2% of TIG3-expressing cells and this was not reduced further by treatment with nocodazole.

Suppressing centrosome separation is expected to impede cell cycle progression. To assess this, keratinocytes were infected with tAd5-TIG3 or tAd5-EV, and after 24 hours they were incubated for 4 hours with bromodeoxyuridine (BrdU). BrdU is incorporated into DNA specifically during new DNA synthesis in S phase. Parallel cultures were stained to detect phosphorylated histone H3, an M phase marker. Increased histone H3 phosphorylation is observed in mitosis (Goto et al., 1999; Tapia et al., 2006). Fig. 9A shows the reduction in the number of BrdU-positive nuclei (green) from 14±1% in TIG3-negative cells to 1.1±0.3% in TIG3-positive cells. Similarly, TIG3-expressing cells display parallel reduction in the mitosis marker, phosphorylated histone H3 (Fig. 9A). This suggests that TIG3 reduces both S and M phase cell cycle progression. We performed immunoblot analysis to determine if these changes are due to a change in the level of cell cycle control proteins. No change in the level of cell cycle regulatory proteins was observed in TIG3-expressing cells (Fig. 9B), suggesting that the reduced cell proliferation is not due to changes in the expression of cell cycle regulators.

We previously reported that TIG3 reduces cell survival (Jans et al., 2008; Scharadin et al., 2011; Sturniolo et al., 2003); however, little is known about the mechanism of growth suppression. In the present study, we demonstrate that TIG3 localizes to the centrosome, as shown by colocalization with the centriole and centrosome markers, γ-tubulin and pericentrin. The centrosome is a 1–2 μm diameter organelle located adjacent to the nucleus – it includes two perpendicularly oriented barrel-shaped centrioles surrounded by the PCM. The staining pattern of TIG3 suggests that it interacts with the PCM; moreover, the interaction is relatively specific. Some TIG3 localizes to the Golgi complex (GM130, M6PR) and endoplasmic reticulum (calnexin), but the main site of TIG3 interaction is at the centrosome. The centrosome is required for cell division and cell survival, as it serves to nucleate polarized microtubule arrays, which organize cytoplasmic organelles and primary cilia in interphase cells, and it also forms the mitotic spindles during mitosis (Doxsey et al., 2005). The PCM includes hundreds of proteins, including many large scaffold proteins that function as regulatory-protein docking sites (Doxsey et al., 2005), and the γ-tubulin ring complexes that are responsible for microtubule nucleation (Doxsey et al., 2005; Kreitzer et al., 1999). The centrosome and the microtubule system are intimately connected, and deficiencies in either microtubule or centrosome function can reduce cell survival (Mazzorana et al., 2011; Rusan and Rogers, 2009). Moreover, because the centrosome and microtubules reciprocally influence the function of each other, we studied the impact of TIG3 on both the centrosome and the microtubules.

One feature we observe is a striking impact of TIG3 on microtubule nucleation. EB1–GFP is a probe that labels the growing ends of microtubules and is useful for studying nucleation (Piehl et al., 2004; Piehl and Cassimeris, 2003). In TIG3-positive cells, EB1–GFP intensely labels the microtubule-organizing center (MTOC), which suggests that TIG3 does not inhibit microtubule nucleation. However, differences are observed. First, there appear to be multiple sites of nucleation, suggesting that the nucleation sites have been rearranged, and, second, the nucleated microtubules do not appear to elongate much beyond the region surrounding the centrosome. This suggests that although nucleation is ongoing, the fate of the nucleated microtubules is different from that of microtubules in control cells.

A second feature is that the microtubules show an unusual distribution in TIG3-positive cells. In control cells the microtubules extend around the nucleus and project in a continuous network to the cell periphery. In TIG3-expressing cells, microtubules accumulate as a ring at the cell periphery, leaving thin microtubule threads to connect this structure to the centrosome. It has been shown that the microtubule network is reoriented in suprabasal keratinocytes such that it distributes to the cell periphery (Lechler and Fuchs, 2007) and it is at least possible that TIG3 drives this rearrangement, because it is expressed in suprabasal epidermis (Jans et al., 2008; Sturniolo et al., 2003; Sturniolo et al., 2005).

Microtubule accumulation at the cell periphery could also be due to microtubule nucleation at alternate sites (e.g. at the cell periphery). However, our experiments using EB1–GFP suggest that the centrosome is maintained as the primary nucleation site in TIG3-positive cells. Alternatively, this change in microtubule distribution could be the result of differences in microtubule anchoring. A specific set of proteins is involved in anchoring the negative end of microtubules to the centrosome. Ninein is a centrosome protein involved in this process (Dammermann and Merdes, 2002; Delgehyr et al., 2005; Mogensen et al., 2000), and is shown to redistribute to the plasma membrane in differentiated keratinocytes (Lechler and Fuchs, 2007). This redistribution results in the formation of a cortical microtubule ring in these cells, which is reminiscent of the peripheral ring that we observe. It is also possible that TIG3 catalyzes release of the microtubule minus end from the centrosome, leading to peripheral accumulation. Although defining a precise mechanism will require further investigation, the present studies show that TIG3 alters the distribution of microtubules.

A third feature is the impact of TIG3 on microtubule response to nocodazole. In control cells, microtubule asters completely dissociate following treatment with nocodazole. By contrast, centrosome-associated asters persist in TIG3-expressing cells treated with nocodazole. The possibility that this is due to a high rate of nucleation is supported by the EB1–GFP labeling studies, which show a high rate of nucleation, and by studies showing rapid expansion of asters upon removal of nocodazole. However, increased maintenance of asters could also be the result of enhanced microtubule stability.

An additional interesting response to nocodazole treatment is the preferential loss of the peripheral tubulin ring, a structure specifically present in TIG3 positive cells. Treatment with nocodazole results in a complete loss of this structure. It is not clear why this ring structure forms in TIG3-positive cells because microtubule elongation appears to cease close to the centrosome. However, these rings reappear during recovery from nocodazole treatment, suggesting that ring formation is not a one-time response to TIG3. An additional feature is that microtubules are not visible by immunostaining in ~50% of TIG3-positive cells. This is despite the fact that the level of tubulin has not decreased, as measured by immunoblot analysis. The absence of signal is not likely to be due to epitope masking, as the reduction is observed with each of three antibodies that detect different tubulin epitopes (rabbit anti-β-tubulin, mouse anti-β-tubulin and mouse anti-acetyl-α-tubulin antibodies).

We also investigated whether TIG3 expression covalently modifies microtubules to alter stability. Microtubule stability is regulated by the post-translational modification of α-tubulin. One mechanism is the removal of the α-tubulin C-terminal tyrosine (Kreitzer et al., 1999). This process is controlled by an unidentified tubulin carboxypeptidase, which removes tyrosine to expose glutamic acid; the modified product is called glu-α-tubulin (Kreitzer et al., 1999). Tubulin tyrosine ligase is an enzyme that catalyzes tyrosine replacement to regenerate tyrosinated α-tubulin. Tyrosinated α-tubulin subunits turn over in 3 to 5 minutes; detyrosinated tubulin (glu-α-tubulin) turns over in 3 to 5 hours (Kreitzer et al., 1999). Thus, microtubules that are enriched with glu-α-tubulin are highly stable. Acetylation of α-tubulin is also associated with enhanced stability (Hammond et al., 2008), as is the accumulation of microtubules in the insoluble fraction in lysates. One possibility is that microtubule accumulation at the cell periphery in TIG3-positive cells is due to enhanced microtubule stability, perhaps associated with increased acetylation of α-tubulin and increased formation of glu-α-tubulin. We detected glu-α-tubulin and acetyl-α-tubulin in control cells and increased levels in TIG3-positive cells. Acetyl-α-tubulin was detected in TIG3-positive cells at the centrosome and in the peripheral tubulin ring. Moreover, we observed increased levels of insoluble tubulin in TIG3-positive cells, which is also an indicator of increased microtubule stability; thus, TIG3 appears to increase microtubule stability. An important question is how the impact of TIG3 on microtubules might influence cell survival and centrosome function. In this respect, centrosome and microtubule functions are linked together. It is known that the centrosome influences microtubule function, and it is also known that agents that influence microtubule turnover or stability, such as paclitaxel, influence centrosome function (Lanzi et al., 2001; Mazzorana et al., 2011). Thus, we propose that the one way that TIG3 reduces cell survival and halts proliferation is through its impact on microtubule distribution and stability, and that this also impacts the centrosome.

The localization of TIG3 to the centrosome could be expected to impact on centrosome separation. Indeed, centrosome separation is impeded in TIG3-positive cells, and this is associated with reduced cell proliferation as measured by reduced incorporation of BrdU into DNA (an S-phase event), and reduced levels of phosphorylated histone H3 (an M-phase marker). In control cultures, 15% of cells are BrdU positive and 10% display high-level phosphorylated histone H3 staining. By contrast, in TIG3-expressing cultures, these levels are suppressed to <1%, suggesting an impact of TIG3 on cell cycle progression. This is not associated with changes in expression of cell cycle control proteins associated with G1, S or G2/M, suggesting that it uniformly impacts all phases of the cell cycle.

Centrosomes are duplicated during the S-phase of the cell cycle in synchrony with nuclear DNA replication (Doxsey et al., 2005; Hinchcliffe and Sluder, 2001). During early prophase, the centrosomes separate and begin the process of migration to opposite poles of the dividing cell (Doxsey et al., 2005; Hinchcliffe and Sluder, 2001). Our studies suggest that TIG3 does not inhibit centrosome duplication, but rather inhibits centrosome separation and thereby influences cell division. Indeed, our findings show that the daughter centrosomes are separated by more than 1.5 μm in 15.9% of control cells, a number that is consistent with the number of cells in mitosis (10.2%). By contrast, centrosome separation is observed in <1% of TIG3-positive cells, suggesting that TIG3 inhibits centrosome separation and/or centrosome migration. Time course studies reveal that TIG3 does not delay but actually halts separation. Proteins that link the mother and daughter centrosomes during replication have been identified – these proteins are cleaved to permit centrosome separation at the appropriate stage of the cell cycle (Bahe et al., 2005). It is possible that TIG3 impedes the cleavage of these proteins and, thereby, inhibits centrosome separation. TIG3 might also inhibit centrosome replication. However, the number of cells that contained duplicated centrosomes was not significantly reduced (P<0.41) in cells expressing TIG3. Thus, TIG3 does not inhibit centrosome replication.

In summary, our studies show that the centrosome is a major site of TIG3 localization in normal human keratinocytes, and that this is associated with changes in microtubule distribution, elongation and covalent modification, and that TIG3 association at the centrosome suppresses centrosome separation (Fig. 10). In addition, we suggest that the altered microtubule environment can feedback to further alter centrosome function. Ultimately, we argue that these events lead to cessation of cell proliferation and reduce cell survival.

Primary cultures of human foreskin keratinocytes were cultured in 0.09 mM Ca²⁺-containing keratinocyte serum-free medium (KSFM) (Sturniolo et al., 2005). The rabbit polyclonal antibody against TIG3 has been described (Deucher et al., 2000). Mouse monoclonal anti-γ-tubulin (catalog number sc-17788), rabbit polyclonal anti-β-tubulin (catalog number sc-9104), rabbit polyclonal anti-cyclin E (catalog number sc-481), rabbit polyclonal anti-CDK2 (catalog number sc-163), mouse monoclonal anti-cyclin A (catalog number sc-239), mouse monoclonal anti-cyclin B1 (catalog number sc245), mouse monoclonal anti-CDK1 (catalog number sc-54), and rabbit polyclonal anti-CDK4 (catalog number sc-601) antibodies were from Santa Cruz (Santa Cruz, CA). Mouse monoclonal anti-β-actin (catalog number A5441), mouse monoclonal anti-BrdU (catalog number B8424) and mouse monoclonal anti-acetyl-α-tubulin (catalog number T7451) antibodies were obtained from Sigma (St. Louis, MO). Rabbit polyclonal anti-phosphorylated histone H3 (catalog number 06–570) and rabbit polyclonal anti-detyrosinated α-tubulin (Glu-tubulin) (catalog number AB3201) antibodies were obtained from Millipore (Billerica, MA). Mouse monoclonal anti-β-tubulin (catalog number ab11311), mouse monoclonal anti-M6PR (catalog number ab2733), and mouse monoclonal anti-pericentrin (catalog number ab28144) antibodies were from Abcam (Cambridge, MA). Rabbit polyclonal anti-α-tubulin (catalog number 2144) and mouse monoclonal anti-p21 (catalog number 2947S) antibodies were from Cell Signaling Technology (Danvers, MA). Alexa 488-conjugated goat anti-rabbit IgG (catalog number A11008), goat anti-mouse IgG (catalog number A11029), Alexa 555-conjugated goat anti-rabbit IgG (catalog number A21429), goat anti-mouse IgG (catalog number A21424) antibodies and Hoechst 33258 were from Invitrogen (Carlsbad, CA). Peroxidase-conjugated donkey anti-rabbit IgG (catalog number NA934) and peroxidase-conjugated sheep anti-mouse IgG (catalog number NA931) antibodies were obtained from GE Healthcare (Piscataway, NJ). Nocodazole was purchased from Calbiochem (Gibbstown, NJ). Mouse monoclonal anti-GM130 (catalog number 610822), mouse monoclonal anti-calnexin (catalog number 610524) and mouse monoclonal anti-cyclin D1 (catalog number 554180) antibodies, and BrdU were purchased from BD Biosciences (Rockville, MD). Plasmid pEB1-GFP was purchased from Addgene (Cambridge, MA).

tAd5-EV, tAd5-TIG3(1–164) and tAd5-TIG3(1–134) adenoviruses have been described (Jans et al., 2008; Sturniolo et al., 2003; Sturniolo et al., 2005). The tAd5 adenovirus encodes the tetracycline operator element linked to the cytomegalovirus promoter. This promoter is active in the presence of a transactivator (TA) protein, which is provided by co-infection with an Ad5-TA adenovirus (Jans et al., 2008; Sturniolo et al., 2005). tAd5-EV is an empty adenovirus and tAd5-TIG3 encodes the full-length 164 amino acid TIG3 protein (Jans et al., 2008; Sturniolo et al., 2005). TIG3(1–134) encodes an inactive mutant that lacks the C-terminal membrane-anchoring domain (Jans et al., 2008; Sturniolo et al., 2005). Keratinocyte cultures were incubated with a multiplicity of infection (MOI) of 10 of tAd5-EV, tAd5-TIG3 or tAd5-TIG3(1–134) in the presence of 5 MOI of Ad5-TA in KSFM containing 6 μg/ml polybrene (catalog number H9268; Sigma). Cells were fixed and stained for immunofluorescence or extracts were prepared for immunoblot analysis at 24 or 48 hours post-infection. The Manders' overlap coefficient was used to measure overlap of TIG3 staining with other organelle marker-proteins using the JACoP plugin for ImageJ (Bolte and Cordelières, 2006).

Keratinocytes growing on coverslips were infected with adenovirus, and after 24 or 48 hours they were washed, fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 30 minutes, and permeabilized with methanol for 10 minutes at −20°C. The coverslips were then incubated for 1 hour with appropriate primary and secondary antibodies. After washing, the cells were fixed to slides by using Mowiol 4–88 (Calbiochem, Gibbstown, NJ), and fluorescence was visualized using an Olympus OX81 spinning-disc confocal microscope.

Keratinocytes were infected with 10 MOI of tAd5-EV or tAd5-TIG3(1–164). After 24 or 48 hours the cells were collected in PBS and centrifuged at 500 g for 5 minutes. The cell pellet was dissolved in lysis buffer (Cell Signaling Technology) supplemented with protease inhibitor cocktail (Calbiochem) and centrifuged at 20,000 g for 20 minutes. The pellet was washed twice with PBS, and then re-suspended and boiled in 4% SDS. Electrophoresis was performed on an equal number of cell equivalents of supernatant (soluble) and pellet (insoluble) fractions on denaturing and reducing 4–15% polyacrylamide gels for immunoblot analysis.

Keratinocytes on coverslips were infected with 10 MOI of tAd5-EV or tAd5-TIG3. At 24 hours the cells were incubated with 10 μM of BrdU for 4 hours. The cells were fixed with 4% paraformaldehyde for 30 minutes at 4°C, washed in PBS containing 1% Triton X-100, incubated in 1 M HCl for 10 minutes on ice, in 2 M HCl for 10 minutes at room temperature and 20 minutes at 37°C. Coverslips were incubated in 0.1 M borate buffer for 12 minutes at 25°C, washed with PBS and stained with antibodies against BrdU or TIG3.

EB1 is a microtubule plus-end binding protein that exchanges continuously and labels the plus end of microtubule (Piehl et al., 2004; Piehl and Cassimeris, 2003). Normal keratinocytes, growing in glass bottom dishes, were transfected with 1 μg of plasmid encoding EB1-GFP fusion protein (EB1–GFP) (Piehl et al., 2004; Piehl and Cassimeris, 2003) in the presence of 2 μg of pcDNA3 or pcDNA3-TIG3(1–164) (Jans et al., 2008; Sturniolo et al., 2003; Sturniolo et al., 2005). After 18 hours, EB1–GFP was detected using an Olympus FluoView FV1000 laser confocal microscope and a 60× objective with 3-second intervals between image capture.