ML: “Is it possible that PIAS-g depletion is interfering with tension sensing, thereby activating aurora-B kinase, inducing a futile microtubule attachment-detachment cycle and consequent unattached kinteochore (a.k.a. spindle) checkpoint? This would account for the suppression of PIAS-g depletion by aurora-B or CDK inhibition.”

DJC: Yes – our working model is that PIASy depletion limits Topo II activity at the centromere resulting in an overabundance of catenations. This could very well result in modified fine structure at centromere regions, leading to differences in tension that are picked up by the spindle checkpoint. It is formally possible that PIASy depletion interferes with microtubule dynamics at the kinetochore more directly. However, some facts argue against this idea, and also suggest that the tension on sister kinetochores in the absence of PIASy is sufficiently balanced to accommodate proper positioning of chromosomes at the metaphase plate.

(1) Our analysis by time-lapse microscopy of PIASy-depleted cells shows that: (a) Chromosomes usually biorient successfully and reach the metaphase plate and they do so with normal kinetics; (b) The wave-like movements of metaphase plates are similar to controls; (c) Some metaphases progress to a fairly normal anaphase after a prolonged delay in metaphase; (d) The cells that had metaphase plates with a few chromosomes laying off the plate (de-congressed metaphases) were derived from those cells that arrested in metaphase for a considerable time (i.e. those few chromosomes are likely to have been lost from the plate subsequent to metaphase arrest, rather being the result of failed congression). These observations indicate that tension on each sister kinetochore is normal enough to promote proper biorientation and chromosome dynamics on the spindle. However, it is still possible that there are subtle defects in tension sensing that activate the checkpoint.

(2) It has been proposed that Aurora B functions to correct monopolar attachments. Our understanding of the current model is that tension defects are converted to occupancy defects (i.e. kinetochore-microtubule attachments are severed) by Aurora B, and then the lack of occupancy activates the checkpoint cascade. But, the process of chromosome biorientation seems to take place normally in PIASy-depleted cells. That is, the chromosomes attach and congress to the plate with normal kinetics. Based on this, we think that at least initially there are no attachments defects. We assume that Aurora B is being activated by a different mechanism that, as described above, might be due an overabundance of catenations at the centromere regions (possibly leading to small changes in tension, or perhaps catenations are monitored more directly). Several pieces of data argue against a defect in attachment. First, sister chromatids segregate to the poles when the spindle checkpoint is overridden (either by inhibition of Aurora B or by Cdk inhibition). This says that the spindle is functional. Chromatin bridges are usually seen in these anaphase attempts that indicate residual catenations between sister chromatids. However, laggard chromosomes are more rare. This could be important as laggard chromosomes indicate merotelic attachment, which might be expected to occur if kinetochore-microtubule attachments are occurring aberrantly.

(3) CENP-E largely disappears from kinetochores in PIASy-depleted metaphase cells. Though the significance of CENP-E localization at kinetochores may be a point of further discussion, it is our understanding that CENP-E at kinetochores is an upstream marker for spindle checkpoint inactivation. This hints that the checkpoint that is activated may not be a conventional spindle checkpoint.


ML: “The one finding arguing about this is the blockage of division by ICRF-193 in PIAS-g siRNA ZM447439-treated cells, but ICRF-193 is a somewhat controversial drug and it has been argued that it induces a DNA damage checkpoint or some other regulatory checkpoint.”

DJC: When ZM447439 is added in combination with ICRF-193 to PIASy-depleted cells, the cells exit mitosis with no delay with respect to controls. Thus, the observed block in chromosome segregation is independent of a checkpoint response and must be due to topoisomerase II inhibition.

However, in the absence of checkpoint overriding agents ICRF-193 triggers a metaphase checkpoint (Skoufias et al., Mol Cell. 2004, 15(6):977-90; Clarke et al., Cell Cycle. 2006, 5(17):1925-8; Mikhailov et al., Curr Biol. 2002, 12(21):1797-806). We think that the metaphase delays observed after PIASy depletion may correspond to the same checkpoint. There is a difference of opinion as to whether this checkpoint is triggered by DNA damage (Mikhailov et al.) or in the absence of damage (Skoufias et al.).

ML: “It would be nice to see more documentation for chromatin bridges in PIAS-g depleted/ZM447439-treated cells--this would really support the intriguing suggestion that the chromatid "cohesion" is really occurring by catenanes.”

DJC: In the micrographs shown, we have tried to select cells that represent an “average” phenotype to illustrate the experiment. However, anaphases occurred ranging from “close to normal” anaphases to more bizarre segregation attempts. We can send, or upload if possible, a greater selection of pictures. However, we can say that almost 100 % of the anaphases have at least one chromosome bridge. However, it is difficult to conclude with absolute certainty that a persistence of catenations is the cause of all the of the segregation defects observed.

Based on our findings we think that, after checkpoint override, a mechanism must exist to allow decatenation, to a large extent, even in the absence of PIASy. Otherwise, separation of every centromere would be blocked and there would be a complete failure in chromosome segregation. However, since almost every cell has at least one chromatin bridge, we think that PIASy is needed to make sure that Topo II is correctly targeted to centromere regions and as a result that every last catenation is removed. In other words, there may be a base-line Topo II activity at centromere regions that is sufficient to resolve most catenations, but PIASy is required to ensure fidelity.

The phenotype observed when Topo II is inhibited is different and a complete block of chromosome segregation occurs (Fig. 6 A-D).

ML: “Even if this is the case, the question remains--why so many more cats in the cen than on the arms?”

DJC: Regulated loss of cohesion between arms must be crucial for a successful meiotic first anaphase. However, in mitosis it seems to be partly dispensable. For example, cells arrested in nocodazole lose cohesion between arms within 2 hours and are still able to segregate successfully upon nocodazole release.

We can conceive of three possible mechanisms that would result in “more” catenations at the centromere regions than at the arms during metaphase of mitosis: (1) It could be that the cells posses a mechanism to maintain catenations at the centromere until the onset of anaphase (as we suggested in Eur J Cell Biol. 2002, 81(1):9-16) or, (2) It could be that from the moment of replication, centromeres possess more catenations than arms, or (3) Catenations might be actively introduced at centromeres. We favour the idea of a combination of some of these factors.