Degradation of Functional Triose Phosphate Isomerase Protein Underlies sugarkill Pathology

Abstract

Triose phosphate isomerase (TPI) deficiency glycolytic enzymopathy is a progressive neurodegenerative condition that remains poorly understood. The disease is caused exclusively by specific missense mutations affecting the TPI protein and clinically features hemolytic anemia, adult-onset neurological impairment, degeneration, and reduced longevity. TPI has a well-characterized role in glycolysis, catalyzing the isomerization of dihydroxyacetone phosphate (DHAP) to glyceraldehyde-3-phosphate (G3P); however, little is known mechanistically about the pathogenesis associated with specific recessive mutations that cause progressive neurodegeneration. Here, we describe key aspects of TPI pathogenesis identified using the TPI^sugarkill mutation, a Drosophila model of human TPI deficiency. Specifically, we demonstrate that the mutant protein is expressed, capable of forming a homodimer, and is functional. However, the mutant protein is degraded by the 20S proteasome core leading to loss-of-function pathogenesis.

GLYCOLYTIC enzymopathies are devastating human diseases that result from heritable mutations in genes encoding certain glycolytic enzymes. The underlying pathogenesis of these diseases remains poorly understood. Triose phosphate isomerase (TPI) deficiency is a severe glycolytic enzymopathy caused by specific missense mutations in the TPI gene that are associated with hemolytic anemia, progressive neuromuscular degeneration, increased susceptibility to infection, and premature death (Schneider et al. 1965; Valentine 1966). The TPI protein exists as a homodimer of ∼26.5-kDa monomers that functions to convert dihydroxyacetone phosphate (DHAP) to glyceraldehyde-3-phosphate (G3P) in glycolysis (Rieder and Rose 1959; Mande et al. 1994). Human TPI deficiency results when specific missense mutations are homozygous or are heterozygous with a null allele resulting in insufficient TPI function (Schneider 2000). TPI's well-characterized role in glycolysis suggests that bioenergetic impairment may underlie pathogenesis.

Alternatively, others have hypothesized that DHAP toxicity or its catabolite methylglyoxal may be integral to pathogenesis independent of a bioenergetic deficit (Karg et al. 2000; Ahmed et al. 2003; Ramasamy et al. 2005). Indeed, data are not consistent from existing reports to establish whether reduced protein levels, instability of the protein homodimer, aberrant protein complexes, and/or a decrease in isomerase activity contribute to pathogenesis of the disease (Olah et al. 2002; Orosz et al. 2006). In part, the difficulty arises from the fact that the disease has a strong neurological component, yet patients' erythrocytes are often used for studies due to their accessibility. Although these studies have produced important information about the disease, they have not provided much insight into neurological and degenerative aspects of TPI deficiency. Such studies have measured reduced isomerase activity from patients' erythrocytes, platelets, and lympocytes (Schneider et al. 1965; Zanella et al. 1985; Orosz et al. 2001; Olah et al. 2002, 2005), as well as brain and muscle tissue (Ationu et al. 1999). Studies have also documented markedly elevated DHAP levels, suggesting that the anemia may be caused by toxicity of DHAP or one of its catabolites (Zanella et al. 1985; Eber et al. 1991; Hollan et al. 1997; Karg et al. 2000; Olah et al. 2002, 2005). Interestingly, DHAP levels in nonerythrocytes are only modestly increased, presumably owing to it being a substrate for phospholipid biosynthesis in other cell types, and a yeast model of TPI deficiency reports that proteins bearing the human disease mutations maintain catalytic activity (Orosz et al. 2001, 2006; Ralser et al. 2006).

We have previously identified a recessive, hypomorphic, missense (M80T) mutation (sugarkill or sgk) in the TPI gene of Drosophila featuring reduced longevity, neurodegenration, and locomotor impairment (Celotto et al. 2006). These mutants, referred to as TPI^sugarkill, serve as an amendable genetic model for elucidating the mechanism of TPI deficiency in vivo (Celotto et al. 2006; Gnerer et al. 2006; Orosz et al. 2006). The TPI^sugarkill locomotor impairment is conditional with mechanical stress or elevated temperatures inducing acute paralysis. The mutation was originally identified as a temperature-sensitive paralytic mutant (Palladino et al. 2002). Each phenotype examined—including onset of neuropathology, locomotor impairment, and reduced longevity—is significantly worsened by modest elevations in temperature between 20° and 29° (Celotto et al. 2006). This suggests that TPI^sugarkill animals may serve as a useful model of TPI impairment where varying temperatures can be utilized to study a series of hypomorphic conditions from mild to severe. Furthermore, in vitro temperature sensitivity has also been reported for the most common homozygous human mutant allele, Glu104Asp (Daar et al. 1986; Arya et al. 1997), suggesting a shared mechanism of pathogenesis exists with at least one of the common disease-causing missense mutations.

There are many proposed mechanisms of TPI deficiency pathogenesis, including (1) protein misfolding and toxic aggregates (Olah et al. 2002; Ovadi et al. 2004), (2) accumulation of a DHAP catabolite methylglyoxal, a potentially toxic metabolite (Karg et al. 2000; Ahmed et al. 2003; Ramasamy et al. 2005), (3) instability of the homodimer via thermoliablilty (Schneider 2000), and (4) bioenergetic defects resulting from reduced enzyme activity. Our model allows us to directly test the role of these mechanisms in pathogenesis resulting from a specific pathogenic missense mutation affecting TPI. Previously we have shown TPI^sugarkill markedly reduces glycolytic flux, but this does not result in a bioenergetic crisis, suggesting a compensatory mechanism and an alternative cause of pathogenesis (Celotto et al. 2006). Our data utilizing the TPI^sugarkill model of TPI deficiency reveal that the mutant protein is capable of forming a dimer and retains function; however, pathogenesis results when the protein is targeted for proteasome degradation.

MATERIALS AND METHODS

Drosophila stocks and culture:

The TPI^sugarkill mutation, unless noted otherwise, is maintained as a homozygous viable ve e TPI^sugarkill strain. The TPI^sugarkill mutation, also known as sugarkill, TPI^sgk, or TPI^sgk1, results from a missense mutation predicted to cause an M80T. This numbering uses the established nomenclature for TPI mutations, where the start methionine is assumed to be removed following translation (see the recent review by Orosz et al. 2006). TPI transgenic strains were previously published (Celotto et al. 2006). All other strains used were obtained from the Bloomington Stock Center. Wild-type controls are ve e TPI^sugarkill heterozygotes, ve e, Canton-S, or appropriate Gal4 controls, as noted.

Life span and behavioral analysis:

Life span analysis was performed as previously described (Palladino et al. 2003). Stress sensitivity (a.k.a. bang sensitivity) was assayed as previously described, and the time to recovery from paralysis was measured (Ganetzky and Wu 1982).

Size-exclusion chromatography:

Drosophila extract:

Two milliliters of flies were homogenized in 5 ml of mobile phase with an electric homogenizer (IKA Ultra-TURAX T8) on the highest setting. Protein complexes >100 kDa were removed from the extract using Amicon Ultra-4 Ultracel 100K columns (Millipore, Bedford, MA) to prevent damage to the HPLC column. The extract was then concentrated using an Amicon Ultra-4 Ultracel 5K column (Millipore). One hundred microliters was injected into the HPLC column, using an auto-injector. Each genotype was performed in duplicate at each temperature.

Nondenaturing size-exclusion chromatography:

Protein fractions were collected using a Shimadzu HPLC system and a Superdex 75 10/300 GL column at a flow rate of 0.4 ml/min. The temperature of the column was maintained at 22°, 30°, or 40°. The mobile phase consisted of 0.025 m NaH₂PO₄, 0.025 m Na₂HPO₄ pH 7.0, and 0.15 m NaCl. One hundred microliters of protein standard (Amersham Biosciences, Arlington Heights, IL) (at 5 mg/ml: ribonuclease A, 14.6 kDa; chymotrypsinogen A, 20.3 kDa; ovalbumin, 46.7 kDa; albumin, 62.9 kDa) or Drosophila extract were injected and monitored at 280 nm for 40 min. Four fractions were collected: (1) 1–23 min, >67-kDa proteins and complexes; (2) 23–28 min, corresponding to 31–67 kDa (53-kDa TPI dimer); (3) 28–34 min, corresponding to 12–31 kDa (26.5-kDa TPI monomer); and (4) 34–40 min, corresponding to 5- to 12-kDa proteins.

Western:

Once fractions were collected, they were concentrated using Amicon Ultra-4 Ultracel 5K columns (Millipore) down to 100 μl, 1 ml H₂O was added, and the samples were reconcentrated to 50 μl. Sixteen microliters of loading dye were added to each fraction and were loaded onto a 12% protein gel. The gel was transferred to a nitrocellulose membrane and blocked with 1% milk. TPI primary antibody (Protein Tech rabbit anti-humanTPI) was added at a concentration of 1:1000 and incubated at room temperature overnight. The blot was washed extensively with PBST, secondary antibody (HRP goat anti-rabbit IgG) was added at a concentration of 1:4000, and the blot was incubated at room temperature for 1 hr. The blot was washed again with PBST and enhanced chemiluminescence (ECL) reaction (Pierce, Rockford, IL) was performed. Film was exposed for ∼1 min before developing.

Western blot analysis:

Fly lysates were obtained from flies aged for 24 or 48 hr as indicated and ground with a pestle in 100 μl of a standard SDS/Bromophenol Blue cell lysis buffer. After boiling for 1 min at 100°, 20 μl of extract were added per lane in a 12% SDS–PAGE gel at denaturing conditions. Signal detection was performed as above. ImageJ Version 10.2 software was used to analyze subsaturation images. Protein intensity represents the normalized TPI protein level (either to actin or β-tubulin, as indicated) in arbitrary pixel units (intensity). In all cases independent protein extracts were used (n ≥ 3).

RESULTS

Recessive nature of the sugarkill mutation:

The sugarkill mutation was identified as a recessive temperature-sensitive paralytic mutant that causes progressive neurodegeneration (Palladino et al. 2002). The mutation was positionally cloned and found to be a missense mutation in the TPI (triose phosphate isomerase) gene (Celotto et al. 2006). TPI^sugarkill (sgk) animals have progressive stress-sensitive locomotor impairment and reduced longevity. Surprisingly, given its well-characterized role in glycolysis, biochemical analyses revealed that bioenergic impairment was not a likely proximate cause of these phenotypes (Celotto et al. 2006; Gnerer et al. 2006). This surprising result led us to investigate alternative hypotheses to explain neuropathogenesis associated with TPI impairment, including the possibility that pathogenesis of the M80T missense mutation in TPI may in part result from a semidominant gain-of-function. Although previous data suggested that TPI^sugarkill was fully recessive (Celotto et al. 2006), we examined a transgenic overexpression assay to test for semidominance effects. Utilizing the UAS-GAL4 system, we generated animals that overexpressed either the TPI^sugarkill or the wild-type TPI protein in a wild-type genetic background and examined the effect on longevity and locomotor function. Specifically, we analyzed longevity of both UAS-TPI^sgk;actin∷GAL4;+/+ and UAS-TPI⁺;actin∷GAL4;+/+ animals compared to actin∷GAL4;+/+ controls at 25° (Figure 1A). There was no significant decrease in life span of the UAS-TPI^sgk;actin∷GAL4 animals compared to the control or the UAS-TPI⁺;actin∷GAL4 animals. There was a modest but significant decrease in median life span of the UAS-TPI⁺;actin∷GAL4 animals compared to that of the GAL4 controls. This is likely the result of transgenic overexpression toxicity or a background strain effect shared between the transgenic animal strains.

Figure 1.—

Life span and behavioral analysis of animals overexpressing wild-type and mutant TPI transgenes. (A) Survival of transgenic animals at 25° overexpressing mutant and wild-type TPI. A modest reduction in longevity is observed, associated with overexpression of wild-type TPI protein. Flies overexpressing the mutant TPI, UAS-TPI^[sgk];Actin∷GAL4;+/+, do not show a significant life span deficit relative to either the control strain or animals overexpressing wild-type TPI. Longevity of TPI^sgk mutant flies is shown for comparison and is significantly reduced compared to all genotypes. N = 80–120 animals in four to six independent populations. Locomotor function in response to stress was examined in transgenic flies at 29° (B) and room temperature (C). TPI^sugarkill mutants exhibit a significantly longer recovery rate from mechanical stress compared to controls when aged at 29° (day 6) or room temperature (day 18). Transgenic overexpression of wild-type or mutant TPI protein did not result in stress-induced locomotor impairment when aged at either temperature. A total of 10–20 animals per genotype were tested. In A–C the error given is SEM and statistics were calculated using Student's t-test (**P < 0.01 and ***P < 0.001).

Further studies examined the effect of mechanical stress on locomotion of these transgenic animals at 29° or room temperature (∼22°). This revealed that overexpression of either mutant or wild-type TPI protein in a wild-type background does not result in a detectable stress-induced locomotor phenotype, such as is prominent in TPI^sugarkill animals (Figure 1, B and C; Celotto et al. 2006). Importantly, the transgenes used in these overexpression experiments are capable of rescuing TPI^sugarkill to a similar extent (see Figure 6). Together these transgenic experiments are consistent with the TPI^sugarkill mutation being fully recessive and suggest a simple loss-of-function model of TPI pathogenesis.

Figure 6.—

Rescue with TPI^sugarkill overexpression. Transgenic animals expressing wild-type and mutant TPI protein were examined for longevity at 25° in a TPI^sgk background. Both UAS-TPI⁺;Act∷Gal4; TPI^sgk/sgk and UAS-TPI^sgk;Act∷Gal4; TPI^sgk/sgk animals show a significant increase in life span compared to Act∷Gal4; TPI^sgk/sgk mutants (***P < 0.001, Student's t-test). Neither transgene was able to fully rescue the longevity to that of heterozygous control animals. Error given is SEM, and n = 120 animals in six independent populations.

Dimer instability of TPI protein in sugarkill mutants:

Several disease-associated TPI mutations and the TPI^sugarkill mutation result in an amino acid substitution near the dimer interface of the protein, which has led us and others to hypothesize that dimer instability may underlie pathogenesis (Celotto et al. 2006; Orosz et al. 2006). Human TPI deficiency has previously been reported to be associated with the destabilization of the dimer state (Olah et al. 2002). To determine if TPI^sugarkill pathogenesis may be associated with structural defects in the TPI protein that destabilize the dimer, we directly examined the stability of the TPI homodimer, using a nondenaturing size-exclusion chromatography assay (Figure 2). TPI is ∼26.5 kDa and exists physiologically as an ∼53-kDa homodimer (Rieder and Rose 1959; Mande et al. 1994). Native size-exclusion chromatography was standardized, whole fly extracts from wild-type controls and TPI^sugarkill were fractionated, and Westerns were performed to determine which fraction contained TPI. At both 22° and 40° the mutant and wild-type TPI were observed only in the dimer fraction (31–67 kDa). No protein was detected in monomer or large fractions corresponding to 12–31 kDa and >67 kDa, respectively. We have noted that increases in temperature from 20° to 37° exacerbate all described phenotypes in TPI^sugarkill (Celotto et al. 2006), yet even at 40° the dimer remained stable in this assay. These data demonstrate that TPI^sugarkill is capable of forming a dimer and that the protein is not observed as a monomer in fly extracts under native conditions. Additional studies were performed with more extensive fractionation, which corroborate these findings (supplemental Figure 1). The TPI^sugarkill protein exists as a dimer in the <100-kDa protein fraction, as is the case for wild-type TPI protein.

Figure 2.—

Size-exclusion chromatography: chromatography demonstrating that both the wild-type TPI and mutant TPI^sugarkill form a dimer. (A) HPLC chromatograph using four standards (albumin, ovalbumin, chymotrypsinogen, and ribonuclease A) to determine size retention. (B) Standard curve used to determine time of elution from the column: fraction L (>67-kDa molecules), fraction D (TPI dimer that is 53 kDa), fraction M (TPI monomer that is 26.5 kDa), and fraction C (molecules between 5 and 12 kDa). (C) Western blots using TPI antibody: each fraction from wild-type and TPI^sugarkill extracts was separated at 22° and 40°.

Protein stability and expression in sugarkill mutants:

We have examined TPI mRNA expression in TPI^sugarkill mutants at room temperature and at 29°. Expression of mRNA is not reduced, suggesting that the pathogenesis is associated with expression of TPI protein bearing the M80T missense mutation (Celotto et al. 2006 and data not shown). Surprisingly, we discovered that TPI protein in TPI^sugarkill mutants is reduced and that the extent of the reduction in protein is temperature dependent (Figure 3). At room temperature there is a significant (73%) reduction in total TPI protein, whereas age-matched flies shifted to 29° for 48 hr reveal a marked (98%) reduction in TPI protein (Figure 3). The severity of this reduction is antagonized by temperature, consistent with other phenotypes observed in TPI^sugarkill mutants. TPI protein being reduced in the absence of a reduction in TPI mRNA strongly suggests that TPI protein is unstable as a result of the TPI^sugarkill mutation.

Figure 3.—

Western blot analysis of TPI protein levels. (A) TPI^sugarkill mutants show a reduced level of TPI protein when aged for 48 hr at room temperature and 29° compared to actin controls. (B) Quantification of Western blot shows an ∼3.5-fold decrease in TPI protein levels at room temperature and an ∼48-fold decrease at 29°. *P < 0.05 (Student's t-test). Error reported is SEM. N = 3 at each temperature. Intensity of TPI was measured from a subsaturation blot and normalized to the intensity of the actin control using ImageJ 10.2.

Proteasomal degradation of TPI^sugarkill protein:

Pros26¹ is a dominant, temperature-sensitive lethal mutation, formerly known as DTS5, which results in larval stage-three lethality at nonpermissive temperatures ≥25° (Saville and Belote 1993). Pros26¹ was shown to be a hypomorphic missense mutation in the Pros26 gene that encodes an altered B6 subunit of the 20S proteasome core (Saville and Belote 1993; Smyth and Belote 1999). In Drosophila, this mutation has been shown to be a useful genetic inhibitor of proteasome function (Fernandez-Funez et al. 2000; Shulman and Feany 2003; Speese et al. 2003; Neuburger et al. 2006). We generated Pros26¹ TPI^sgk animals to investigate the effect of proteasome inhibition on TPI^sugarkill mutants, utilizing longevity as our most sensitive and quantitative phenotype. Proteasome impairment is itself toxic with respect to aging (Keller et al. 2002; Chondrogianni and Gonos 2005). Impairment also dramatically affects protein synthesis (Ding et al. 2006) and has been implicated in numerous disease states, including Alzheimer's (Keller et al. 2000), Parkinson's (McNaught and Olanow 2006; Olanow and Mcnaught 2006), and Huntington's (Seo et al. 2004) diseases. Surprisingly, Pros26¹ TPI^sgk double mutants did not have reduced longevity compared to that of TPI^sugarkill mutants. In fact, there is a modest but significant increase in longevity in the double-mutant strain (Figure 4). These data suggest that the TPI^sugarkill protein is targeted for degradation by the 20S proteasome core and that impairment of the proteasome, though normally toxic, can partially suppress the TPI^sugarkill life span phenotype.

Figure 4.—

Survivorship of TPI^sugarkill Pros26¹ double mutants: statistical analysis of median life span of mutants aged at (A) 29° and (B) 25°. (A) Pros26¹ TPI^sgk/sgk have a modest but significant increase in longevity compared to TPI^sgk/sgk animals at 29° (P < 0.05). (B) Flies aged at 25° show a nonsignificant improvement in longevity. The median life span of Pros26¹ TPI^sgk/sgk double mutants is still much reduced from that of heterozygous controls at both temperatures examined (P < 0.0001). In A and B, the error given is SEM, statistical comparison uses Student's t-test, and n = 120 animals in six independent populations for each genotype.

To examine this more directly, we measured TPI protein levels from the double mutant compared to control animals (Figure 5). This analysis revealed a significant increase in TPI^sugarkill protein levels in the proteasome-inhibited double mutant compared to TPI^sugarkill animals. Partial rescue of TPI^sugarkill protein levels is consistent with the modest improvement observed in longevity (Figure 4). This also indicates that the 20S proteasome core is implicated in TPI^sugarkill degradation. Further experiments have shown that the Pros26¹ mutation does not have a significant effect on wild-type TPI protein levels (supplemental Figure 2). These data imply either that the proteasome does not regulate wild-type TPI protein or that wild-type protein turnover is insufficient to detect a difference with these methods. The finding that a loss-of-function proteasome mutation reduces TPI^sugarkill degradation and improves the longevity phenotype also strongly suggests that the TPI^sugarkill protein is functional and that degradation of functional protein underlies pathogenesis in TPI^sugarkill mutants.

Figure 5.—

Proteasome mutation increases TPI[sgk] protein levels. (A) Western blot of TPI and β-tubulin proteins from flies aged at 29° for 24 hr. Pros26¹ TPI^sgk/sgk animals show an increase in TPI protein compared to TPI^sgk/sgk mutants. (B) Quantitative analysis of TPI protein normalized to the tubulin loading control indicates a greater than twofold increase in TPI protein in the Pros26¹ TPI^sgk/sgk double mutants (*P < 0.05, n = 3 each genotype, Student's t-test). Both TPI^sgk/sgk and Pros26¹ TPI^sgk/sgk animals show a marked reduction in TPI protein when compared to wild-type controls, consistent with previous data. Error reported is SEM. Protein intensity is measured from a subsaturation image with ImageJ 10.2.

Transgenic rescue with overexpression of TPI^sugarkill protein:

If pathogenesis in TPI^sugarkill results from a loss-of-function due to enhanced proteasomal degradation of functional protein, one would predict that overexpression of the mutant protein may rescue or partially rescue the TPI^sugarkill mutation. Previous data demonstrated that this is not the case at 29° (Celotto et al. 2006). This is likely due to the marked instability of TPI^sugarkill protein at 29° (Figure 3). We examined whether transgenic overexpression of TPI^sugarkill protein could rescue the TPI^sugarkill mutation at 25°, where the reduction in protein stability was not as severe. Both UAS-TPI^sgk;actin∷GAL4; TPI^sgk/sgk and UAS-TPI⁺;actin∷GAL4; TPI^sgk/sgk animals were examined for longevity and found to dramatically improve longevity of TPI^sugarkill mutants (Figure 6). Interestingly, the improvement was similar for each transgene, and neither transgenic strain fully rescued longevity compared to heterozygous controls. These data demonstrate that proteasome degradation of TPI^sugarkill protein underlies pathogenesis and opens up the possibility that one or more of the human pathogenic mutations that cause TPI deficiency may result from a decrease in protein stability.

DISCUSSION

Despite the well-characterized role of TPI in glycolysis, TPI-deficiency pathogenesis remains poorly understood. Considerable attention has been focused on the fact that only specific missense mutations are pathogenic; TPI enzymopathy results when these missense mutations are homozygous or are heterozygous with a known null allele (Schneider 2000; Orosz et al. 2006). Numerous lines of evidence demonstrate that homozygous TPI null mutations do not result in disease and are developmentally lethal. Most notably, the high frequency of heterozygous null mutations in the population and lack of homozygous null observations, as well as the stage of lethality observed in mouse and fly null animals, are all consistent with this assertion (Mohrenweiser and Fielek 1982; Mohrenweiser 1987; Merkle and Pretsch 1989; Celotto et al. 2006). Several hypotheses have been proposed to explain TPI-deficiency pathogenesis associated with specific missense mutations. We have utilized a Drosophila model of human TPI-deficiency disease to examine several of these hypotheses in vivo. The Drosophila TPI^sugarkill missense mutation recapitulates numerous key features of this progressive neurodegenerative disease and provides a unique system for investigating pathogenesis (Celotto et al. 2006; Orosz et al. 2006).

Independently, two laboratories studying the same genetic model of glycolytic enzymopathy demonstrated that TPI pathogenesis was not simply the result of bioenergetic impairment (Celotto et al. 2006; Gnerer et al. 2006). TPI-deficiency pathogenesis has also been proposed to result from protein unfolding, which could result from a loss-of-function or a gain-of-function toxicity (Olah et al. 2002). We investigated the possibility of TPI^sugarkill semidominance that may result from a gain-of-function protein toxicity. The TPI^sugarkill mutation is completely recessive and expression of transgenic TPI^sugarkill protein does not result in any detectable toxicity beyond that of wild-type overexpression. Native HPLC size exclusion chromatography assays demonstrate that TPI^sugarkill protein is capable of forming a dimer, suggesting the protein is properly folded.

The TPI^sugarkill strain has reduced life span, progressive locomotor impairment, and neurodegeneration at room temperature, 25°, and 29°, where phenotypic onset is accelerated with increasing temperature (Celotto et al. 2006). These data suggest that the dysfunction associated with the mutation is temperature sensitive and the extent of dysfunction increases within the physiologically relevant range for Drosophila (∼18°–30°). This interpretation is consistent with the fact that the mutation was initialy isolated as a temperature-sensitive paralytic mutant (Palladino et al. 2002). Western blot analysis has demonstrated that the TPI^sugarkill protein is reduced in the mutants and that this decrease is enhanced with increases in temperature within the normal physiological temperatures for Drosophila. These findings suggest that temperature can be used to antagonize TPI^sugarkill function and allow a range of phenotypic severity to be studied, analogous to an allelic series of mutations.

Our studies demonstrate a temperature-dependent reduction in TPI protein: even an acute temperature shift dramatically reduces cellular TPI protein (Figure 3). However, mRNA levels are not reduced in TPI^sugarkill animals at any temperature examined (Celotto et al. 2006 and data not shown). These findings suggested active proteolytic degradation of TPI^sugarkill protein. Indeed, genetic inhibition of the 20S proteasome core with the Pros26¹ mutation results in modest increases in TPI^sugarkill protein levels (Figure 5). These data suggest the TPI^sugarkill protein is functional. Transgenic overexpression of TPI^sugarkill protein was able to partially rescue the longevity defect associated with TPI^sugarkill, further supporting the assertion that TPI^sugarkill protein is functional. The data are consistent with a model whereby the mutation results in the degradation of functional TPI protein. Further experiments will be needed to determine whether the protein is fully functional and to assess the relative contributions of protein degradation and hypomorphic function.

It is possible that the mutant protein has an essentially normal structure and retains function but that the methionine-to-threonine missense mutation destabilizes the protein by a mechanism independent of protein function and structure. This could occur if the mutation created a signal stimulating the protein's degradation via a regulatory or a cellular quality-control mechanism. Such signals could include (1) an exposed hydrophobic patch in the protein, (2) a degradation signal for an E3 ubiquitin ligase subunit, (3) a post-translational covalent modification, or (4) a functional PEST domain. Although the protein retains function and the ability to form a dimer, it remains possible that it is partially misfolded, stimulating known quality-control pathways and its degradation. Alternatively, the mutation could create a degradation signal. One such pathway known to increase the rate of proteolytic degradation is the PEST pathway. Amino acid sequences rich in proline (P), glutamic acid (E), serine (S), and threonine (T) residues form functional PEST domains that can result in rapid degradation of proteins in vivo (Rogers et al. 1986; Rechsteiner and Rogers 1996). These are assigned PEST-FIND scores on the basis of an established algorithm (Rogers et al. 1986). TPI proteins are not known to be regulated by the PEST pathway—consistent with this, the PEST algorithm does not find any PEST sequences in the protein (http://www.at.embnet.org/toolbox/pestfind/). The sequence KGAFTGEISPAMLK is identified as a “poor PEST domain” with a score of −22.8. Although the TPI^sugarkill mutant adds an additional T within this sequence (KGAFTGEISPATLK), the algorithm still rates the region as a “poor PEST domain” with a marginally improved score of −16.55 (Rogers et al. 1986).

Although it is not clear what stimulates the degradation of TPI^sugarkill protein, the efficient degradation of mutant protein suggests a regulated quality-control mechanism. Quality-control pathways that degrade nonfunctional, misfolded, and unfolded proteins are extensive and required for the health and longevity of the cell (Hartl and Hayer-Hartl 2002; Barral et al. 2004; Young et al. 2004; Zhang and Kaufman 2006; Outeiro and Tetzlaff 2007). Cytosolic protein quality-control pathways are often mediated by Hsp90 and various cochaperones that form extensive networks with chaperones specific for individual or groups of substrates (Zhao et al. 2005). Further experimentation will be needed to determine the basis of the TPI^sugarkill protein degradation and whether the mutant protein is degraded in an Hsp90 and/or ubiquitin-dependent manner. The TPI^sugarkill mutant may prove useful at elucidating cytosolic quality-control mechanisms required to prevent premature senescence and progressive neurodegenerative disease.

Together, the data suggest that TPI^sugarkill pathogenesis does not result from bioenergetic impairment, the encoded protein forms a stable dimer and retains sufficient function to rescue the mutation upon transgenic overexpression, and that pathogenesis results from proteasomal degradation of the apparently functional protein. Alternatively, it is possible that mutations abrogating isomerase activity are lethal (do not cause disease) and that the specific mutations that result in disease disrupt a hitherto novel function of TPI. This interpretation is consistent with recent findings in a yeast TPI model, where transgenic expression of multiple human mutant TPI proteins in vivo demonstrated catalytic isomerase activity associated with the pathogenic mutant proteins (Ralser et al. 2006).