In addition to its function in spindle formation and cytokinesis Plk1 has a well established role in centrosome maturation at the G2/M transition, which involves the recruitment of additional γTuRCs from the surrounding cytoplasm. A longstanding goal in the analysis of this pathway has been the identification of a Plk1 substrate that regulates this process. Our results suggest that Plk1 regulates γ-tubulin recruitment to mitotic centrosomes by controlling the centrosomal levels of the γ-tubulin targeting factor GCP-WD, and that GCP-WD itself is a Plk1 substrate. However, by analyzing GCP-WD mutations that disrupt interaction with Plk1 (S418A, T557A) and mutations in potential Plk1 phosphorylation sites (S433A, T487A) we find no evidence for defects in centrosome targeting.
What is the role Plk1 in the regulation of GCP-WD? One possibility is that Plk1 regulates the interaction of GCP-WD with the mitotic spindle, which depends on phosphorylation of serine S418 and is critical for spindle-targeting of the γTuRC and the formation of additional microtubules within the spindle 
. Might Plk1 be responsible for serine 418 phosphorylation? We consider this unlikely because serine 418 does not correspond to a Plk1 consensus phosphorylation site, and because the S418A mutation strongly reduces the global phosphorylation of GCP-WD, whereas Plk1 inhibition has only mild effects (as judged by changes in GCP-WD mobility in SDS-PAGE). However, serine 418 phosphorylation could promote additional phosphorylation by Plk1 at a different site, which could contribute to the interaction of GCP-WD with the spindle. This is supported by our observation that Plk1 inhibition not only prevents accumulation of GCP-WD and γ-tubulin at centrosomes, but also reduces their spindle localization. This would impair microtubule formation within the spindle and could contribute to the defects in kinetochore fiber assembly after Plk1 inhibition 
. However, our results for the T557A mutant, which is defective in Plk1-binding and phosphorylation but does not show any localization defects, suggest that regulation of GCP-WD by Plk1 is not essential for spindle localization of γTuRCs. Instead, Plk1 might regulate the augmin complex, which mediates spindle interaction of GCP-WD/γTuRC without affecting centrosome targeting 
. Interestingly, FAM29A/hDgt6, a subunit of the human augmin complex, was recently identified as a Plk1 substrate 
Our data show that in addition to GCP-WD several other centrosome proteins with a known role in γTuRC recruitment are also mislocalized after inhibition of Plk1. One of these proteins, Cep192, has recently been shown to function upstream of GCP-WD in centrosome targeting of γ-tubulin 
. Cep192 is also required for the recruitment of pericentrin to mitotic centrosomes. These results indicate that at least one Plk1-dependent step in γTuRC recruitment takes place upstream of GCP-WD and might promote the recruitment of other γTuRC attachment factors. Several lines of evidence suggest that γTuRC recruitment to mitotic centrosomes might not follow a simple pathway based on a single adaptor protein, but rather be the result of complex interactions between several PCM proteins. For example, pericentrin has been suggested to directly interact with γTuRC subunits and recruit γTuRCs to mitotic centrosomes 
, but compared to Cep192 the depletion of pericentrin causes only a moderate loss of centrosomal γ-tubulin 
(this study). However, there is also some mutual dependency between Cep192 and pericentrin for their localization to mitotic centrosomes, which complicates the interpretation of depletion phenotypes 
. Cep215/Cdk5Rap2 is yet another protein suggested to function in the recruitment of γTuRCs to centrosomes 
. Similar to the homologous proteins Mto1 in yeast and Cnn in flies 
, Cep215 was shown to interact with γ-tubulin complexes, but its relationship with GCP-WD was not investigated. Cep215 also associates with pericentrin to promote PCM cohesion at interphase centrosomes 
. We have shown that there is interdependency between pericentrin and Cep215 for their localization to centrosomes in mitosis and depletion of either protein causes a similar partial loss of both GCP-WD and γ-tubulin from mitotic centrosomes. GCP-WD depletion only weakly affects centrosomal accumulation of these proteins and this effect might be the result of PCM fragmentation after mitotic arrest of GCP-WD depleted cells 
rather than a direct effect on Cep215/pericentrin. We propose to place pericentrin and Cep215 together with Cep192 upstream of GCP-WD in the recruitment pathway ().
Interdependencies of components involved in the recruitment of γ-tubulin complexes to mitotic centrosomes.
Our finding that Cep215 depletion results in a much weaker effect than Plk1 depletion or inhibition contrasts with data from a recent genome-wide RNAi screen in flies. Plk1 and the Cep215 homolog Cnn were identified as the genes with the strongest impact on centrosome maturation and Cnn was shown to be phosphorylated in a Plk1-dependent manner 
. Most importantly, however, homologs of GCP-WD, Cep192, pericentrin were also identified as components involved in centrosome maturation 
. In addition, myomegalin, another mammalian protein with homology to Cnn, might participate in centrosome maturation in the absence of Cep215
It is important to note that the recruitment of γ-tubulin to mitotic centrosomes might be mechanistically different from its interaction with interphase centrosomes. In interphase γTuRC binding to centrosomes involves Ninein-like protein (Nlp). However, at the onset of mitosis, Plk1 phosphorylates Nlp and abolishes its interaction with centrosomes and with the γTuRC 
. This might clear the way for the involvement of mitosis-specific recruitment factors.
To understand the complex interplay of different PCM proteins in mitosis and their impact on centrosome maturation in mammalian cells it will be important to test proteins such as Cep192, Cep215, and pericentrin for mitotic phosphorylation and determine the phosphorylation sites. Phosphorylation mutant analysis will reveal whether there is indeed one key substrate that initiates centrosome maturation or, alternatively, Plk1-dependent phosphorylation of several proteins, acting in parallel or complementary pathways, is required for efficient centrosome maturation and accumulation of γTuRCs.