This study shows that phosphatase inhibitor 2 (I-2) is concentrated in the primary cilium of human epithelial cells. We found endogenous I-2 localizes to the primary cilium prior to acetylation of the axonemal tubulin that serves as a marker for the primary cilium. Knockdown of I-2 by siRNA significantly suppressed the full acetylation of tubulin in the axoneme of the primary cilium, and reduced the percentage of cells with a primary cilium. These effects were rescued by pharmacological inhibition of PP1 or HDACs. Our results link I-2 and PP1, proteins involved in control of protein Ser/Thr phosphorylation, to the acetylation of tubulin in the primary cilium. This links two different systems of protein post-translational modification in the primary cilium.
Tubulin acetylation and deacetylation are catalyzed by a tubulin acetyltransferase and a specific histone/tubulin deacetylase (HDAC). Acetylation stabilizes microtubules [23
], and the modification has been mapped to a single residue K40 of alpha-tubulin [24
]. However, the identity of the acetyltransferase reactive with this site is still unknown. Among the multiple histone deacetylases, HDAC-6 has been shown to interact with and catalyze alpha tubulin K40 deacetylation [26
]. HDAC-6 has been proposed to destabilize microtubules of the axoneme in the primary cilium and be regulated through phosphorylation involving Aurora A [9
]. Previous results have shown that histone deacetylases, such as HDAC-1, HDAC-6, and HDAC-10 bind directly to the PP1 catalytic subunit [28
]. Inhibition of HDACs by TSA causes dissociation of these HDAC-PP1 complexes, presumably through a conformational change. [28
]. On the other hand, ATM dependent activation of PP1 by ionizing radiation led to dissociation of HDAC-PP1 complexes and dephosphorylation of HDAC-1, with an increase of HDAC activity [29
]. Phosphorylation of I-2 directly by ATM is proposed to cause dissociation from PP1, accounting for PP1 activation by ionizing radiation [30
]. Previous results therefore established functional links between PP1, HDAC-6 and I-2 and we suggest these are related to acetylation of tubulin in the primary cilium of epithelial cells.
PP1 was found in proteomic analysis of purified flagella from Chlamydomonas [31
] and isolated human ciliary axonemes [32
]. This supports the idea that PP1 regulates Ser/Thr phosphorylation of proteins in flagella and cilia. In addition, RT-PCR analysis showed that the levels of PP1 mRNA increased by 1.9-fold upon deflagellation, considered as evidence that PP1 is involved in flagellum function in Chlamydomonas. PP1 co-purifies with microtubules (MT) and binding to microtubules is mediated by the MT-associated protein Tau [33
]. MT-associated phosphoproteins that might be PP1 substrates include kinesin and dynein motor complexes that are responsible for intraflagellar transport (IFT). However, the lack of change in size of the primary cilium in I-2 knockdown cells argues that I-2 does not regulate PP1 holoenzymes that control IFT. Altogether, various results point to some regulation of microtubule function and axoneme organization by PP1. In our experiments, inhibition of PP1 by 0.5 nM calyculin A rescued the reduction of alpha-tubulin acetylation in response to I-2 knockdown. Low doses of calyculin A are known to selectively inhibit PP2A [22
], and we observed that they did not affect tubulin acetylation in the primary cilium in ARPE-19 cells (not shown). High doses of calyculin A were lethal to ARPE-19 cells. We surmised that the intermediate dose of calyculin A we used was producing selective but probably incomplete inhibition of PP1. We propose that I-2 affects tubulin acetylation and stabilization of the axoneme in the primary cilium by inhibiting the activity of specific PP1 holoenzymes. I-2 knockdown did not affect overall acetylation of tubulin, as detected by Western blotting whole cell extracts, or tubulin acetylation in either cytoplasm or midbody, as detected by immunostaining. We suspect that I-2 sensitive PP1 holoenzymes are specifically concentrated in the axoneme of the primary cilium to regulate ciliary alpha-tubulin acetylation, possibly by control of HDAC-6 activity. Previous work indicated that PP1 is targeted to microtubules by the protein Tau, however, Tau has not been identified by proteomics in flagella from Chlamydomonas or in axonemal fraction from cilia of human cells. This suggests that PP1 and therefore I-2 might be targeted to the primary cilium by some other PP1 regulatory subunit yet to be identified.
Taken in sum, our data suggest a model (Fig. ) for the regulation of tubulin acetylation by I-2 in the primary cilium of human epithelial cells. Axonemal tubulin acetylation is a balance between tubulin acetyltransferase and deacetylase. We propose that a HDAC forms a complex with a PP1 holoenzyme that binds I-2. I-2 inhibits the PP1 activity, keeping HDAC in an inactive, phosphorylated state. When I-2 is knocked down or dissociated from the PP1-HDAC complex, an increase of PP1 activity leads to dephosphorylation and activation of HDAC, favoring deacetylation of tubulin and destabilization of the primary cilium axoneme. This model invokes phosphorylation to negatively regulate HDAC-6, but the opposite has been suggested, i.e. HDAC-6 activation by phosphorylation. An alternative could be regulation of HDAC-6 by binding to PP1 with I-2 acting as an allosteric modifier of PP1 inhibition of HDAC-6. More detailed study of HDAC6-PP1 complexes should distinguish between these models.
Figure 9 Model for regulation of axonemal tubulin acetylation in the primary cilium by PP1 and I-2. Axonemal tubulin acetylation is regulated by the balance of acetyltransferase (HAT) and deacetylases (HDACs). The tubulin-localized HDAC binds and interacts with (more ...)
Lastly, we have demonstrated unanticipated multifunctionality of I-2 that will need to be reconciled into more complex models. I-2 is a mitotic phosphoprotein substrate of CDK1-cyclin B1 [17
]. In addition, I-2 binds and regulates the Pin1 prolyl isomerase [36
]. Both Nek2A and Aurora A kinases are activated by I-2, the former indirectly by PP1 inhibition, the latter directly by protein-protein interaction [37
]. In separate studies we have found that I-2 is required for proper chromosome segregation and cytokinesis, probably by indirect control of Aurora B [19
]. Thus, I-2 has proved to be quite a versatile protein. The primary cilium may take advantage of I-2 to connect and coordinate different signaling pathways.