PolyQ and PolyQY Proteins Readily Form Polymers in Vivo
—Polyglutamine- and some glutamine-rich domains confer upon proteins an ability to polymerize in vivo
. Expansion of polyglutamine stretches in some human proteins causes their polymerization, aggregation, and disease, e.g.
Huntington disease (23
). Glutamine-rich domains define prion properties of yeast proteins. Here, we undertook a systematic study of polymerization of polyglutamine proteins to model both yeast prion formation and huntingtin polymerization. It was shown previously that polyglutamine can polymerize in yeast, but these polymers cannot propagate independently of [PIN+
] because of poor polymer fragmentation (10
). We proposed that efficient fragmentation requires the presence within the polymerization domain of hydrophobic residues, which could attract chaperone-fragmenting machinery. To test this, we created fusions of polyQ and polyQY tracts of different length to Sup35MC and studied their polymerization in yeast.
All hybrid proteins larger than 51Q and 46QY were able to polymerize. In contrast to Sup35 prion ([PSI+
]) polymers, which appear rarely and can be obtained only upon selection for the suppressor phenotype, the polymerization of polyQ/QY proteins started rapidly () and did not require cell selection. The ease of polymer appearance could be related to simplicity of the polyQ/QY primary structure. It is presumed that, in the Sup35 amyloid structure, the adjacent protein molecules are arranged as parallel in-register β-strands (22
). However, the restriction for register is likely to be inapplicable for the uniform polyQ/QY sequences. This increases greatly the number of possible variants of relative location of adjacent molecules in a polymer and is likely to cause (i) faster polymer appearance and (ii) less precise copying of the initial amyloid fold and its plasticity, i.e.
an ability to change gradually between different forms. In agreement with the latter, we failed to find stable variants differing by polymer size for any tested polyQ/QY polymers.
TABLE 1 Ability of polyQ/QY proteins to polymerize in vivo RNQ1 deletion was made either before the introduction of the polyQ/QY-encoding plasmids or after polyQ/QY polymer appearance (Δ after p.a.). Generations indicate the number of cell generations (more ...)
[PIN+] accelerated polymer appearance for polyQ and 50QY, but was not required for it. These polymers appeared at 100 generations in [pin-] cells, whereas in [PIN+] cells, polymers were observed at 30 generations () and possibly were present at much earlier points unavailable for analysis. The propensity to polymerize was higher for the polyQY proteins and increased with the size of the polyQ/QY tract, which was manifested in faster acquisition of the polymer state and lesser dependence on [PIN+] and deletions of RNQ1 and HSP104. For example, polymers of 76QY and 120QY were observed at 30 generations independently of the [PIN] status and were able to propagate in the Δhsp104 background.
Not only the prion form of Rnq1, but the very presence of this protein facilitated the appearance of polyQ and 50QY polymers because they appeared faster in the [pin-] cells than in the Δrnq1 cells. This observation suggests interaction, presumably co-polymerization of these proteins at initial steps of the polyQ and 50QY polymer appearance. As a likely scenario, the first appearing polyQ/Y polymers are poorly fragmented, and incorporation of Rnq1 improves their recognition by Hsp104. With time, the polymer fold evolves, through the imprecise copying mentioned above, into the forms better recognizable by Hsp104 and not requiring Rnq1.
Tyrosine Residues Enhance Polymer Fragmentation
—Incorporation of tyrosine residues into the polyQ domains greatly decreased the size of the respective polymers, which suggests their improved fragmentation. It may appear that the relation between polymer size and fragmentation is ambiguous because the size of polymers should also depend on the speed of polymerization. However, thorough mathematical analysis of prion polymerization in yeast by Tanaka et al.
) shows the lack of such dependence. This conclusion, although not directly formulated by the authors, easily follows from their model, which gives expressions for the number of polymers, y
/β, and the total number of molecules contained in polymers, z
/βγ. Here, α, β, γ, and R
are the rate constants for Sup35 synthesis, polymer growth, polymer division, and growth of yeast cells, respectively. The average size of polymers (number of Sup35 molecules/polymer) equals the quotient of these values, z
. Using the above expressions for z
, one obtains z
/γ. Thus, the size of polymers is defined solely by fragmentation (γ) and does not depend on polymerization (β). Then, a decreased size of polyQY polymers means their increased fragmentation. This confirms our starting assumption that hydrophobic residues serve as determinants for chaperone recognition and Hsp104-related fragmentation.
Hydrophobic residues themselves do not guarantee Hsp104 recognition because they usually fold inside of a protein structure. In line with this idea, we observed that most of the Sup35 polymers seeded by the prion form of Rnq1 are poorly fragmented and thus should have their hydrophobic residues hidden (10
). The content of such residues in the polyQY proteins (20% Tyr) is similar to that in the Sup35 N-terminal domain (16% Tyr and 8% others). The difference in fragmentation of the polyQ and polyQY polymers suggests that at least some tyrosines of the polyQY proteins are exposed. Tyrosines may be placed on the polymer surface via a self-selection (microevolution) process. Presumably, the polymer folds with exposed tyrosines appear with low frequency, inversely related to the proportion of exposed tyrosines. However, these folds have an increased propagation potential. When polymers with different folds are present in one cell, the one that allows faster multiplication of prion particles should increase its proportion and eventually displace others (20
). Faster multiplication requires reasonably high fragmentation efficiency. Thus, a feature that improves polymer fragmentation, such as exposed tyrosine residues, would be selected.
Polyglutamine Polymers Are Fragmented and Show Prion-like Properties
—Here, we have shown that polyQ stretches longer than 70 residues allow proteins to form polymers able to propagate in the absence of Rnq1. Propagation of such polymers suggests that they are fragmented. This contradicts our assumption that polyQ polymers should not be recognized by chaperones because their polymerization domains lack hydrophobic residues. This contradiction may be solved by proposing the following mechanisms for fragmentation of the polyQ polymers. (i) Chaperones can recognize, although inefficiently, the amyloid structures lacking hydrophobic residues; (ii) polymerization of the polyglutamine region interferes with proper folding of the adjacent Sup35 middle domain (referred to as Sup35M), which becomes a target for chaperone(s); and (iii) other cellular glutamine-rich proteins that contain hydrophobic residues co-polymerize with polyglutamine domains and attract fragmentation. In support of the latter opportunity, we have observed that Rnq1 and Sup35 co-polymerize efficiently with 70Q and 85Q.4
Previously, we described two categories of yeast amyloids, heritable (prion) and non-heritable, the propagation of which depends on [PIN+
). The polymers of 70Q and 85Q proteins should belong to the heritable category, being [PIN+
]-independent, but formally they are not prions because in wild-type cells they lack a stable non-polymerizing state, which could be infected. This formal restriction is essentially removed in the Δrnq1
cells, in which 70Q and 85Q proteins are relatively stable in the non-polymerizing form. The polymer state of these proteins was curable by cell growth in the presence of GdnHCl, and the polymers of 85Q were able to infect cells producing non-polymerizing 85Q. Thus, with some reservations, the 85Q polymers may be considered as prions. PolyQY polymers are efficiently fragmented and heritable, but again they may not be considered as prions because of the lack of a stable monomer state. Such polymers should be categorized as replicating amyloids, which are close to, but formally distinct from, prions.
Polymer Fragmentation in the Absence of Hsp104—Propagation of natural yeast prions, i.e. [PSI+], [URE3], and [PIN+], strictly requires Hsp104 because the propagation involves polymer fragmentation performed by this chaperone. However, polymers of 76QY and 120QY propagated in the absence of Hsp104. This may be due to either of two mechanisms: (i) polymer fragmentation independent of Hsp104 and (ii) highly efficient formation of these polymers de novo. However, the second mechanism may be excluded. With it, the efficiency of polymer formation should not depend on time, which contradicts the data. The deletion of HSP104 almost eliminated 76QY polymers, as observed at 30 cell generations after deletion, but polymerization was restored to nearly the original level at 100 generations. Apparently, the original 76QY polymer fold could not propagate without Hsp104 due to poor fragmentation but, after some time, rearranged to allow better fragmentation. Then, two opportunities are available: the fragmentation is performed by a protein, presumably, chaperone, or it occurs because of intracellular mechanic shearing forces. In the first case, it may be proposed that the new amyloid fold was distinguished by further increase in the proportion of exposed tyrosines. In the second, we may expect self-selection for the most fragile amyloid fold. The feasibility of a mechanic shearing mechanism is difficult to estimate because it is unknown whether the shearing forces are sufficient to break amyloids. Two following observations are more compatible with Hsp104-independent fragmentation performed by a chaperone. Such fragmentation does not occur or is very inefficient for polyQ polymers, which may be related to tyrosines being a chaperone recognition target. Fragmentation improves with the length of polyQY stretch, which should alleviate the access of chaperones to the amyloid stem, but should not affect its fragility.
The size of 76QY polymers in the absence of Hsp104 was increased by 4–6-fold, which suggests that fragmentation was several times less efficient than in the presence of Hsp104. In contrast to Hsp104, the additional fragmentation was insensitive to GdnHCl, which suggests that Hsp104 is the only target inhibited by GdnHCl related to prion polymer fragmentation.
Implications for Huntington and Related Diseases—In human diseases related to expansion of polyglutamine tracts in some proteins, longer tracts define earlier onset and faster progression of these diseases. The results of this work illuminate the reasons for this. We observed that shorter polyQ stretches (51Q and 65Q) allow polymerization only in a non-heritable, [PIN+]-dependent mode, whereas 70Q and longer stretches can polymerize in a prion mode. The latter means fundamentally faster kinetics of polymerization because fragmentation multiplies amyloid polymerization seeds. Also, the ability of polyQ polymers to appear independently of other amyloids and the ease of their appearance increased with polyQ size. For human diseases, such properties should define earlier disease onset for longer polyQ stretches. For both yeast Sup35 and animal PrP prions, the initial appearance of prion is a very infrequent event. The polyQ polymers appear incomparably faster than Sup35 prion polymers. This suggests that the appearance of polyQ polymers in human cells should be restricted not by the primary seed formation, but by the ability of a cell to counteract the amyloid polymerization. It should be noted that, in contrast to the PrP protein, which forms polymers extracellularly, the intracellular huntingtin polymers replicating in the “prion” mode are not infectious because the spread of polymerization is restricted by cell borders. Such polymers may be regarded as “intracellular prions.”