Transglutaminases and neurodegeneration

Abstract

Transglutaminases (TGs) are Ca²⁺-dependent enzymes that catalyze a variety of modifications of glutaminyl (Q) residues. In the brain, these modifications include the covalent attachment of a number of amine-bearing compounds, including lysyl (K) residues and polyamines, which serve to either regulate enzyme activity or attach the TG substrates to biological matrices. Aberrant TG activity is thought to contribute to Alzheimer disease, Parkinson disease, Huntington disease, and supranuclear palsy. Strategies designed to interfere with TG activity have some benefit in animal models of Huntington and Parkinson diseases. The following review summarizes the involvement of TGs in neurodegenerative diseases and discusses the possible use of selective inhibitors as therapeutic agents in these diseases.

Cerebral transglutaminases

Eight active transglutaminases (TGs) (TGs 1–7 and factor XIIIa) are expressed in mammals, of which TGs 1–3 (Kim et al. 1999)¹ and 6 (Hadjivassilou et al. 2008) are present in human brain. The major reaction thus far attributed to the cerebral TGs is transamidation. In this reaction the carboxamide moiety of a Q residue [-C(O)NH₂] is converted to a substituted carboxamide [-C(O)NHR] by nucleophilic attack of an amine [RNH₂] such as various mono-, di-, and polyamines or the ε amino group of a K residue (Lorand and Graham 2003). Of the possible transamidation linkages, the γ-glutamyl-ε-lysine [N^ε-(γ-l-glutamyl)-l-lysine] (GGEL) isopeptide linkage formed between Q and K resides, is the most commonly studied. GGEL bonds occur both within and between polypeptide chains, and thereby contribute to the formation of stable soluble and insoluble polymers. TGs also cross-link proteins via bis-γ-glutamylpolyamine bridges between Q residues (Piacentini et al. 1988). These linkages are formed by two successive transamidations: the first utilizes a free polyamine to generate a γ-glutamylpolyamine residue, which becomes the amine-bearing substrate for a second transamidation. bis-γ-Glutamylpolyamine cross-links are formed at least as frequently as those involving GGEL (Piacentini et al. 1988). Under physiological conditions, the majority of cross-linking bonds are generated outside of cells where the concentrations of Ca²⁺ are sufficiently high enough to stimulate the catalysis of these bonds by TGs. TG 2 and 3 have three Ca²⁺ binding sites and the current indications are that the catalysis of GGEL bonds requires the occupation of all three sites (Datta et al. 2006).

Intracellular Ca²⁺ concentrations rarely match the extracellular concentrations. Moreover, GTP acts in cells as an endogenous inhibitor of TGs (Bergamini et al. 1987). Nevertheless, intracellular TG-catalyzed reaction products can be detected in normal cells, especially those products related to polyamination (Piacentini et al. 1988). The consequences of polyamination in the brain, however, are poorly understood as only a limited number of polyaminated proteins have been identified (Tucholski et al. 1999). Of these, phospholipase A₂ (PLA₂) is especially interesting since the polyamination of this enzyme may contribute to the inflammation associated with neurodegeneration. PLA₂ produces two groups of pro-inflammatory mediators: leukotrienes and prostaglandins. Polyamination of PLA₂ results in a 3-fold increase in activity (Cordella-Miele et al. 1993) that may persist for the life of the protein given the inability of most peptidases to hydrolyze γ-glutamylamine (GGEL, γ-glutamylpolyamine and bis-γ-glutamylpolyamine) linkages (Fink and Folk 1981). Thus, polyamination may represent a unique post-translational modification of enzymes that permanently affects activity. This situation contrasts with other types of covalent post-translational modifications, such as phosphorylation that typically are transient.

Increased TG(s) in neurodegenerative diseases

Transglutaminase activity is widespread in brain (Kim et al. 1999) and is present in primary cultures of neurons and astrocytes (Perry et al. 1995; Caccamo et al. 2004). TG 2 is associated with the extracellular matrix, cell membranes and cytosol of neurons, and TG activity has been identified in synaptosomes (Pastuszko et al. 1986), mitochondria (Krasnikov et al. 2005), and nuclei (Lesort et al. 1998). The activity, expression and amounts of individual TG enzymes are increased in a variety of neurodegenerative diseases. TG activity is significantly elevated in the affected cerebral regions in Alzheimer disease (AD) (Johnson et al. 1997; Kim et al. 1999), Huntington disease (HD) (Karpuj et al. 1999; Lesort et al. 1999), and supranuclear palsy (Zemaitaitis et al. 2003). These increases in activity are accompanied by gains in the amount of TG 1 and TG 2 proteins in AD brain (Kim et al. 1999; Bonelli et al. 2002), and also of TG 2 protein in the brains of HD (Lesort et al. 1999) and supranuclear palsy (Zemaitaitis et al. 2003) patients. Increased TG 2 protein is also found in the CSF of AD (Bonelli et al. 2002) and Parkinson disease (PD) (Vermes et al. 2004) patients.

Not only are the amounts of TG increased in AD, HD and supranuclear palsy, but the conditions favoring the activation of these enzymes are also enhanced in these diseases. These conditions include elevations in intracellular Ca²⁺ due to glutamate-mediated excitotoxity (Caccamo et al. 2004) and other perturbations in Ca²⁺ homeostasis (Mattson 2007) as well as decreases in GTP concentrations following from losses in energy production (Lin and Beal 2006).

The number of TG 2 transcripts is also increased in HD (Lesort et al. 1999) and supranuclear palsy (Zemaitaitis et al. 2003), and a shortened alternate transcript of TG 2 encoding a form of enzyme missing the GTP binding domain is expressed in AD brain (Festoff et al. 2002). A number of mechanisms may account for the increased transcription and translation of TGs in neurodegenerative disorders. The TG 1 promoter has Ca²⁺ (Kawabe et al. 1998), retinoid (Polakowska et al. 1999), cAMP, Sp1, and AP1 responsive elements (Medvedev et al. 1999), while the TG 2 promoter contains elements that respond to retinoids (Nagy et al. 1996; Yan et al. 1996), interleukin 6, transforming growth factor β1 (Ritter and Davies 1998), and tumor necrosis factor-α (Kuncio et al. 1998). The inflammatory mediators are likely to act via the NF-κB (Nuclear factor κB) binding region in the TG 2 promoter (Kuncio et al. 1998; Kim et al. 2008). NF-κB translocation and DNA binding are stimulated by tumor necrosis factor-α and glutamate, both of which have been shown to increase TG 2 expression in microglia and astrocytes (Campisi et al. 2004; Park et al. 2004). As noted earlier, inflammation accompanies neurodegeneration and TGs may contribute to this response via the sustained activation of polyaminated PLA₂. The possibility that TGs may contribute to their continued activation highlights the necessity of limiting the activity of these enzymes in neurodegenerative disorders.

Increased TG-catalyzed products in neurodegenerative diseases

Increased TG activity in neurodegenerative disorders is accompanied by an increase in TG-catalyzed products. Selkoe et al. (1982a,b) demonstrated that cerebral TGs catalyze the in vitro polymerization of cytoskeletal elements, and hypothesized that TGs might facilitate paired helical formation in AD tangles. TG 2 and GGEL cross-links were subsequently shown to co-localize with the tangles (Miller and Anderton 1986; Johnson et al. 1997) and TG2 was shown to co-localize with plaques (Zhang et al. 1998) in AD brain. Components of the plaques or tangles, including β-amyloid (Aβ) (Ikura et al. 1993; Dudek and Johnson 1994; Ho et al. 1994; Rasmussen et al. 1994), the Dutch mutation of Aβ (Q²² → E²²) (Dudek and Johnson 1994), tau (Miller and Anderton 1986; Dudek and Johnson 1993; Miller and Johnson 1995; Appelt and Balin 1997; Murthy et al. 1998; Tucholski et al. 1999), and the non-Aβ component derived from α-synuclein (Jensen et al. 1995) are TG substrates. The in vitro products of the reaction of these substrates with TG bear a striking resemblance to the insoluble polymers found in AD brain (Jensen et al. 1995; Appelt and Balin 1997; Hartley et al. 2008).

Huntington disease is caused by a CAG expansion in the huntingtin (htt) gene that encodes a length of contiguous Q residues [polyglutamine (Q_n)] in the N-terminus of the expressed protein. Green (1993) hypothesized that the expanded Q_n region would favor the formation of TG-catalyzed GGEL linkages and lead to the formation of htt-containing aggregates. In support of this hypothesis, expanded Q_n domains are excellent TG substrates (Kahlem et al. 1996; Cooper et al. 1997a; Gentile et al. 1998; Lesort et al. 1999; Zainelli et al. 2005) and mutant htt is present in HD aggregates (DiFiglia et al. 1995) as are GGEL crosslinks (Zainelli et al. 2003).

The increased cerebral aggregation seen in HD, PD, and supranuclear palsy is also associated with a comparable increase in GGEL immunoreactivity within the polymers (Zemaitaitis et al. 2000; Zainelli et al. 2003; Andringa et al. 2004). Although some concerns have been raised about the specificity of GGEL antibodies in immunoblots (Johnson and LeShoure 2004), the increase in protein-associated GGEL in AD brain has been unequivocally confirmed using mass spectrometric techniques (Kim et al. 1999; Nemes et al. 2004).

As noted earlier, the isopeptide bonds in γ-glutamylamine linkages are resistant to proteolysis (Fink and Folk 1981). Moreover, the ability to metabolize free γ-glutamylamines in brain is limited. Consequently, γ-glutamylamines are excised intact during proteolysis and are present in brain and CSF (Jeitner et al. 2001, 2008; Dedeoglu et al. 2002). A several-fold increase in free GGEL has been measured in the brains of HD patients (Dedeoglu et al. 2002), and the amount of GGEL in the CSF of patients with AD, PD (Sárvari et al. 2002), or HD (Jeitner et al. 2001, 2008; Dedeoglu et al. 2002) is also increased relative to control CSF. The increase in CSF GGEL reported in HD is also matched by comparable increases in the amounts of CSF γ-glutamylspermidine, γ-glutamylputrescine and bis-γ-glutamylputrescine (Jeitner et al. 2008). CSF contains higher (μM) quantities of γ-glutamylspermidine than GGEL (< μM) in accord with the suggestion noted above that TGs predominantly catalyze polyamination over Q → K cross-linking within the brain.

Possible mechanisms for TG-mediated neurotoxicity

Although the formation of insoluble protein aggregates has been proposed to account for the toxic actions of TGs, the role of such aggregates in the etiology of diseases such as HD is controversial (Kuemmerle et al. 1999; Sieradzan and Mann 2001). Indeed, mice that lacked TG 2 and over-express mutant htt had 30% more brain aggregates and still lived longer than their TG 2- and mutant htt-expressing littermates (Mastroberardino et al. 2002). The increased aggregation is unlikely to have been due to compensatory cross-linking by TG 1 and 3, since the TG 2-deficient mice also exhibited a 10-fold decrease in the number of GGEL linkages. It was subsequently shown that GGEL cross-links serve to produce soluble Q_n aggregates, whereas polyaminated or unmodified Q_n domains spontaneously aggregate to form insoluble polymers (Lai et al. 2004; Konno et al. 2005). These observations suggest that TG 2-catalyzed GGEL bond formation generates soluble aggregates that may be neurotoxic.

Another possibility is that rather than being toxic per se, the soluble and insoluble polymers cause neuronal death by sequestering critical proteins within the aggregates. These proteins could include, for example, glyceraldehyde 3-phosphate dehydrogenase, α-ketoglutarate dehydrogenase, and histones (Cooper et al. 1997b, 2000; Gentile et al. 1998) in HD, and ubiquitin, HSP27, parkin, and α-synuclein in AD (Nemes et al. 2004). It has also been suggested that congestion of proteasomes may contribute to CAG-expansion and other neurodegenerative diseases (Cooper et al. 2002; Wang et al. 2008). In support of this hypothesis, components of the ubiquitin proteasome system are found in HD aggregates (Bennett et al. 2007).

The above observations suggest the following model for the contribution of TGs to neurodegenerative diseases. Early in these diseases, TGs predominantly catalyze polyamination reactions. As these diseases progress, TGs begin to form more GGEL cross-links, which stabilize soluble toxic protein aggregates, eventually leading to removal of key proteins and to a fatal congestion of proteasomes.

TGs as potential therapeutic targets in neurodegenerative diseases

Several authors have raised the possibility that TG inhibitors may be of therapeutic benefit in neurodegenerative diseases (e.g. Cooper et al. 2002; Gentile and Cooper 2004), and one such in vitro inhibitor – cystamine – is beneficial in murine models of HD and PD (e.g. Dedeoglu et al. 2002; Van Raamsdonk et al. 2005; Stack et al. 2008). We have shown that cysteamine, the reduced form of cystamine, is a competitive inhibitor/alternative substrate of TG 2 (Jeitner et al. 2005). Cysteamine attenuates polyamination by acting as an alternative TG 2 substrate, presumably forming N^β-(γ-l-glutamyl)-cysteamine linkages (Jeitner et al. 2005). In addition to cysteamine, cystamine is metabolized to hypotaurine and taurine, and cystamine treatment in mice leads to increased brain cysteine levels (Fox et al. 2004; Pinto et al. 2005). We tested the ability of hypotaurine, taurine, cysteine, and cysteamine to inhibit TG 2. Of the tested compounds, only cysteamine was able to inhibit TG 2-catalyzed polyamination (Fig. 1). As neither cystamine nor cysteamine can be detected (detection limit ≤ 20 μM) in the brains of cystamine-treated YAC128 (HD) mice (Pinto et al. 2005), the conversion of cystamine to cysteamine, and then to N^β-(γ-l-glutamyl)-cysteamine, is likely to be rapid.

Fig. 1.

Effect of cysteamine, cysteine, hypotaurine, and taurine on TG activity. TG activity in the presence of cysteamine (□), cysteine (■) hypotaurine (○), or taurine (●) was determined by measuring the incorporation of radiolabeled putrescine into N,N-dimethylcasein as described by Jeitner et al. (2005) using tritiated putrescine and his-tagged guinea pig TG 2 (Gillet et al. 2004). The data are depicted as percent of the control (21 910 ± 1731 dpm per tube) and represent the mean ± SEM of four separate experiments. The data with cysteamine at concentrations ≥ 2.5 mM and taurine at 5 and 10 mM were significantly different from that of the control (p < 0.05, paired t-test).

Prolonged cystamine treatment results in decreased TG activity (Dedeoglu et al. 2002; Van Raamsdonk et al. 2005), even though cysteamine is not an irreversible TG inhibitor (Jeitner et al. 2005). Thus, another mechanism must account for the diminished TG activity. We hypothesize that cystamine-derived cysteamine inhibits the binding of transcription factors to TG promoters and thereby limits the transcription of TGs. In support of this hypothesis, cysteamine attenuates the DNA binding of AP1 and NF-κB (Goldstone et al. 1995), and binding sites for these factors are present in the TG 1 and TG 2 promoters (Kuncio et al. 1998; Medvedev et al. 1999; Kim et al. 2008).

As indicated above, cystamine has multiple biological actions in the brain, including raising the levels of cysteine (Fox et al. 2004; Pinto et al. 2005). Cystamine also causes elevation of brain-derived neurotropic factor (Borrell-Pages et al. 2006) and possibly attenuates apoptosis through inhibition of caspase 3 activity (Lesort et al. 2003). Recently, we discovered another potentially beneficial property of cystamine. Cystamine, at concentrations as low as 15 μM, significantly attenuated dopamine-induced macroautophagy in SH SY5Y cells, whereas cysteamine had no effect (Fig. 2). Dopamine reduced the viability of SH SY5Y cells by 48 ± 3% (mean ± SEM, n = 11) after 24 h, while the combination of dopamine and cystamine (15 μM) only reduced viability by 24 ± 2% (p < 0.05, paired t-test). In these experiments, the cells were pre-treated with cystamine for 2 h, and then treated for a further 24 h (i.e. 26 h). Importantly, the treatment of cells with cystamine for only 4 h prior to the application of dopamine was as effective as the 26-h cystamine treatment (Fig. 2). This observation suggests that cystamine primes the cells against the induction of macroautophagy.

Fig. 2.

Effect of cyst(e)amine on dopamine-induced macroautophagy. SH SY5Y cells at 70% confluence in a 75 cm² flask were detached with 0.05% trypsin : verscene (1 : 1), then collected into 50 mL 10% fetal bovine serum, 90% Dulbecco's modified eagle medium prior to seeding onto 24 multi-well plates at 1 mL per well. Twenty-four hours later the cells were treated with either cystamine (■) or cysteamine (□) for 2 h. The cells were then incubated for an additional 24 h together with 10⁻⁴ M dopamine to induce macroautophagy as described by Gomez-Santos et al. (2003). The data from these studies are depicted as the mean ± SEM of four separate experiments. The cells were also treated with cystamine for 4 h then washed three times with Hank's balanced salt solution (HBBS) at 37°C, followed by 24 h incubation with 100 μM dopamine (gray open circles). The data from these studies represent the mean of two individual experiments that did not vary by more than 5% of the mean. At the end of the incubations, the cells were washed twice with HBSS at 37°C then incubated with 250 μL of 1 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide in HBSS for 30 min at 37°C, for the determination of viability. The resulting formazan precipitates were then solubilized with 200 μL DMSO. The viability of cells treated with cystamine at 15 μM and 100 μM dopamine (■) was significantly different than that of cells treated with dopamine alone (p < 0.05, paired t-test.

Given the multiplicity of biological activities attributed to cyst(e)amine, it is difficult to assign the therapeutic benefit of this agent to the inhibition of TGs per se. Moreover, excess cysteamine has been reported to be harmful in at least one setting. Thus, Frankel and Schipper (1999) noted that cysteamine induces the appearance of iron-rich (peroxidase-positive) cytoplasmic inclusions in cultured rat astroglia, which are identical to glial inclusions that progressively accumulate in substantia nigra and other subcortical brain regions with advancing age.

The positive results obtained with a mouse model of HD in which the animals lacked TG 2 are a more compelling argument for the involvement of TGs in neurodegeneration (Mastroberardino et al. 2002) than the results obtained with cystamine. In this regard, several groups are actively synthesizing more selective TG inhibitors than cystamine as possible therapeutic agents. These inhibitors include dihydroisoxazole derivatives, peptide-bound 1,2,4-thiadiazoles, peptides containing diazo-5-oxo-l-norleucine in place of glutamine, α,β-unsaturated amides and epoxides (Pardin et al. 2008a,b). Pardin et al. have recently focused their attention on trans-cinnamoyl benzotriazole amides and 3-(substituted cinnamoyl)pyridines [azachalcones], which are potent reversible TG inhibitors (Pardin et al. 2008a,b), and have discovered a triazole compound that inhibits guinea pig TG 2 with a K_i value of ∼170 nM (Pardin et al. 2008b). Stein and colleagues have discovered another series of reversible TG inhibitors that are thieno[2,3-d]pyrimidin-4-one acylhyd-razide derivatives (Duval et al. 2005; Case and Stein 2007).

Finally, Sohn et al. (2003) have developed a novel strategy that blocks the polyamination of PLA₂. This group noted that uteroglobin and lipocortin-1 contain common sequences that antagonize the interaction of TG 2 with PLA₂. Synthesized variants of these sequences prevented both the polyamination of PLA₂ and experimentally-induced allergic conjunctivitis in experimental animals. These results suggest that rather than inhibiting TG 2 directly, some benefit may be derived from targeting the interaction of TGs and specific pathogenic TG substrates. Polyaminated PLA₂ may be one such target in neurodegenerative disorders.

Conclusion and future prospects

Although TGs do not cause neurodegenerative diseases directly, the current evidence suggests that this family of enzymes contributes to the neuropathology once the disease process has begun. It is anticipated that potent TG inhibitors will soon be evaluated for their therapeutic potential in cellular and animal models of HD and other neurodegenerative diseases. Care will be required to ensure that these inhibitors are sufficiently selective so as not to affect crucial TG reactions critical to normal metabolic processes or to inhibit blood clot formation.

Abbreviations

AD

Alzheimer disease

Aβ

β-amyloid

GGEL

γ-glutamyl-ε-lysine [N^ε-(γ-l-glutamyl)-l-lysine]

HD

Huntington disease

htt

huntingtin

PD

Parkinson disease

PLA2

phospholipase A₂

Q_n

polyglutamine

TG

transglutaminase