In this study, we first established that the NGMC foci were the seeds responsible for propagating the [PSI+] prion. Specifically, regardless of whether Hsp104 was inactivated by guanidine or by overexpression of Hsp104-2KT, the fraction of [PSI+] yeast with NGMC foci directly correlated with the extent of prion curing as determined by the red/white colony assay. None of the yeast in growing culture showed large agglomerates of immobilized NGMC nor was there evidence for dead-end aggregates of NGMC; whenever cells contained visible NGMC foci, the cells and their progeny were [PSI+]. The one exception was yeast that were left with only one focus when [PSI+] was cured by overexpression Hsp104-2KT. This was not a dead-end focus because when a cell with a single focus was grown overnight in SD medium a small fraction of its daughter cells were [PSI+]. Evidently, when these cells were plated, high levels of Hsp104-2KT were still present for a time in the daughter cells resulting in the curing of most of these cells, and these cured daughter cells obscured the limited number of [PSI+] daughter cells present thus giving the colony a red appearance.
Second, we established that curing of [PSI+
] by Hsp104 inactivation is due to inhibition of severing of the prion seeds by Hsp104, in agreement with the model of Tuite et al. 
. During curing by overexpression of Hsp104-2KT, the number of foci halved each generation, which shows that severing was completely inhibited and thus the seeds were diluted out by cell division. During curing by guanidine treatment, when glucose deprivation was used to make the NGMC foci apparent, our data show that severing was not completely inhibited. This accounts for the extra 2–3 generations it takes to cure [PSI+
] by guanidine treatment compared to overexpression of Hsp104-2KT, as well as the difference between our data and the theoretical line obtained by halving the number of foci each generation. It is perhaps not surprising that there is residual severing activity in the presence of guanidine since in vitro experiments showed that the ATPase activity of Hsp104 was not completely inhibited by guanidine 
Previously, the Chernoff laboratory observed that NMG expressed from the SUP35
promoter formed large fluorescent agglomerates during curing of [PSI+
] by overexpression of Hsp104-2KT, as well as in a [PSI+
] yeast strain expressing low levels of Hsp104 
. They proposed that the formation of large agglomerates reduced the ability of the prion to propagate, perhaps because the agglomerate cannot pass to the daughter cells. In addition, the Taguchi laboratory reported a transmission defect when guanidine was used to cure [PSI+
] in yeast expressing NMG from the GAL1
, which probably caused considerable overexpression of the NMG. In this latter study, the daughter cells did not inherit the large NMG foci present in the mother cells, but instead, became enriched in monomeric NMG and were cured prior to the mother cells. In contrast, when NMG was expressed from the SUP35
promoter, which probably induced lower levels of NMG expression than the GAL1
promoter, the Chernoff laboratory observed diffuse NMG fluorescence during curing of [PSI+
] by guanidine 
, in agreement with our observations on NGMC. The fluorescent properties of NGMC in [PSI+
] yeast during curing by Hsp104 inactivation was also examined by the Serio laboratory. By expressing NGMC under the MFA1
promoter, they observed that the foci were largely immobilized when Hsp104 was inactivated either by guanidine treatment or by overexpression of Hsp104-2KT 
. The differences between these various studies may be due to experimental conditions including the time of curing, the amount of expression, and the construct that was GFP-labeled.
Studies overexpressing NMG suggest it forms large agglomerates that cannot be significantly reduced in size or transmitted to daughter cells. With regards to this study, we examined whether there was a defect in the transmission of the NGMC foci to daughter cells during curing of [PSI+
] by overexpression of Hsp104-2KT. Our data clearly show that the foci are transmitted based on their abundance in the daughter cells relative to the volume of the mother and daughter cells, as proposed by the Byrne et. al. model 
. On the other hand, even in yeast that are propagating the [PSI+
] prion, Serio’s laboratory has detected a defect in the transmission of seeds to daughter cells 
. Several factors might account for the fact that we do not detect a size bias in transmission of seeds during the curing that occurs when Hsp104 activity is reduced. First, in untreated [PSI+
] yeast, Hsp104 is actively remodeling the seeds, which could make the seed size heterogeneous, whereas when Hsp104-2KT is overexpressed, there is no remodeling of the seeds and all the seeds tend to be large 
. Just based on diffusion alone, if smaller foci were present, they would diffuse faster and be more likely to pass from mother to daughter than larger foci. In addition, the polarisome might play a role in causing the large foci to be retained in the mother cell. The polarisome is involved in an Hsp104-dependent mechanism for transporting misfolded proteins from the daughter to the mother along actin cables 
. If the polarisome plays an important role in yeast with active Hsp104 as opposed to yeast in which Hsp104 activity is inhibited, it could cause retention of large foci in the mother cells.
Although the mechanism of curing by inactivation of Hsp104 by guanidine treatment and by overexpression of Hsp104-2KT was similar, our data show that the NGMC foci became undetectable during curing by guanidine treatment. The fact that the NGMC appeared diffusive during curing by guanidine even though the cells were still [PSI+
], led us to incorrectly propose that cell division was not essential for the curing of [PSI+
. However, even though the seeds were not visible, we have now shown that the NGMC foci became readily apparent when the [PSI+
] yeast were either starved or Hsp104-2KT was overexpressed. Since severing activity is 80% inhibited in 5 mM guanidine, severing activity cannot account for all of the seeds becoming undetectable. Furthermore, when [PSI+
] was cured by overexpression of Hsp104(K218T) in the presence of guanidine, the curing kinetics showed severing was completely inhibited, but still the seeds became undetectable. This again shows that severing is not causing the absence of detectable foci.
FCS microscopy observations suggested that the loss of detectable foci in [PSI+
] yeast during curing by guanidine treatment is due to a reduction in the size of the foci. This reduction in size is dependent on Hsp104 since yeast in which the HSP104
gene was conditionally deleted showed prominent NGMC foci during curing of [PSI+
] both in the absence and presence of guanidine. Therefore, our data are consistent with Hsp104 having an activity in the presence of guanidine that reduces or “trims” the size of the prion seeds without severing them into new seeds. This trimming activity was also observed when [PSI+
] was cured by overexpression of Hsp104(K218T), which contains only a point mutation at the ATPase site of the NBD1 domain of Hsp104. Apparently, trimming activity removes Sup35 molecules from the prion seeds, but these released Sup35 molecules do not form a new seed. In fact, a size threshold for misfolded Sup35 oligomers to propagate new seeds rather than to dissolve into monomer has been incorporated into previous models of prion propagation 
. One possibility is that during trimming, Hsp104 extracts Sup35 molecules from the ends of the prion fiber, whereas during severing, it extracts Sup35 molecules from the interior of the prion fiber. Therefore, during trimming from the ends of the amyloid fiber, the released Sup35 oligomers may dissolve into monomer, whereas during severing a new seed is formed.
Not only does trimming by Hsp104 occur during curing of [PSI+] by guanidine, but there is also evidence that trimming occurs in propagating yeast. When Ssa1 was overexpressed, the time course of [PSI+] curing by guanidine treatment was not affected, but the NGMC foci remained prominent until the yeast were cured. This shows that SSa1 does not affect severing in the presence of guanidine even though it markedly reduces trimming. If trimming also occurs in propagating [PSI+] yeast, Ssa1 should have the same effect here, that is, without affecting severing it should increase the size of the foci. Indeed this is what occurred in the 1074 yeast strain of propagating [PSI+] yeast. Ssa1 markedly increased the size of the foci without curing [PSI+] even after 20 generations of Ssa1 overexpression. These results are consistent with inhibition of trimming of prion seeds in [PSI+] yeast in the presence of excess Ssa1. As for why the foci are larger in [PSI+] yeast with active Hsp104 than in [PSI+] yeast partially cured by guanidine, this may occur because as Hsp104 severs the prion seeds, there may be less Hsp104 available to trim the seeds.
Our proposal that Hsp104 both severs and trims seeds is in agreement with observations that have been made on other triple-A ATPases. In mammalian cells, two different triple-A ATPase proteins, spastin and katinin, are able to both trim and sever microtubules 
. However, as part of their trimming activity these proteins are able to completely depolymerize microtubules, and if trimming alone were able to completely depolymerize prion seeds, incubation of [PSI+
] cells in the presence of guanidine would be expected to cure yeast even if cell division were prevented. Of course, as we pointed out above, this did not occur in our previous experiments, suggesting that perhaps the prion core has a special structure that makes it resistant to trimming. Finally, not only inactivation of Hsp104 cures [PSI+
] prion, but overexpression of Hsp104 also cures 
. Since it has been previously been observed that during the curing of [PSI+
] by overexpression of Hsp104, Hsp104 partially solubilizes Sup35 aggregates 
, it may turn out that the trimming activity of Hsp104 is important for curing by overexpression of Hsp104.