The major vault protein is related to the toxic anion resistance protein(TelA) family

Vaults are large, hollow ribonucleoprotein particles that are conserved from Dictyostelium to humans(Kedersha et al., 1986). To date, no prokaryote relatives of the vault particle have been identified. The barrel-shaped vault complex comprises the major vault protein (MVP),telomerase-associated protein-1 (TEP1) and the vault poly(ADP-ribose)polymerase (VPARP), and an untranslated small RNA(Suprenant, 2002). The major vault protein (MVP) comprises 75% of the mass of the vault particle and expression of MVP in baculovirus-infected insect cells is sufficient to self-assemble into a vault-like particle with the same overall shape and dimensions of the native vault structure(Stephen et al., 2001). Surprisingly, these vault shells are very dynamic and are capable of`breathing' to allow the incorporation of TEP1, VPARP and potential cargo molecules (Poderycki et al.,2006).

The function of vaults remains elusive. High levels of vaults are correlated with a poor prognosis in certain cancers and with some multi-drug resistant (MDR) cancer cell lines(Scheffer et al., 2000). Although most vaults reside in the cytosol, a small but detectable fraction(∼5%) is associated with the nucleus and it has been suggested that vaults may be involved in nucleocytoplasmic transport(Chugani et al., 1993; Hamill and Suprenant, 1997; Slesina et al., 2005; Stewart et al., 2005; van Zon et al., 2006). MVP may act as a scaffolding protein for the protein tyrosine phosphatase, SHP-2 and activated extracellular-regulated kinases (ERKs)(Kolli et al., 2004). In addition to interacting with the epidermal growth factor signaling pathways,MVP expression is upregulated through the JAK/STAT signaling pathway(Steiner et al., 2006).

From molecular analyses and high-resolution cryoelectron microscopy, we know that vault composition and structure are highly conserved(Kedersha et al., 1990; Kong et al., 1999; Mikyas et al., 2004; Stewart et al., 2005). The primary sequence of MVP comprises seven short repeats of ∼55 amino acids at the amino terminus, a central region and a coiled-coil interaction domain at the carboxyl terminus (van Zon et al.,2002), as depicted schematically in Fig. 1A. The repeat unit (aa 26–401) is highly conserved across eukaryotes and forms the basis for a characteristic protein sequence domain. The solution NMR structure of a two-repeat domain of the MVP reveals that the walls of the central barrel are built from repeating units of a unique three-stranded anti-parallel sheet(Kozlov et al., 2006). This structural repeat is the same length as the sequence repeat but is shifted by half of a sequence repeat (∼26 amino acids). The coiled-coil domain at the C terminus is thought to align the MVPs at the vault cap(Kozlov et al., 2006; Mikyas et al., 2004; van Zon et al., 2002).

Eukaryotic MVP coding sequences are well represented in the public databases. To date, MVP sequences have been identified in birds, mammals,amphibians, bony fishes, rays, echinoderms, mollusks, flatworms and trypanosomatids. Recognizable MVP sequences are notably absent in the genomes of C. elegans, D. melanogaster and S. cerevisiase. In this study, we initiated a sequence-based approach to understanding vault function. We report that the major vault protein (MVP) is related to the toxic anion-resistance protein, TelA, of Rhodobacter sphaeroides strain 2.4.1 and that mammalian MVP responds dynamically to tellurite, and other toxic oxyanions. We propose that vaults are an unrealized component of a stress response pathway that originated in bacteria.

Prospective homologs to MVP were initially identified by subjecting the MVP repeat sequence to an iterative search of the non-redundant protein sequence database (National Center for Biotechnology, NIH Bethesda, MD, USA) using the PSI-BLAST program (Altschul and Koonin,1998). The second search iteration yielded the tellurite resistance protein (TelA) from Rhodobacter sphaeroides (AF019377) as a plausible match (homologous alignment with E=6e–15). Based on the results of our initial BLAST search for homologs to the MVP repeat unit in which TelA was identified, we performed a reciprocal search based on the sequence of tellurite resistance protein, retrieving the eukaryotic MVP in a second round of iteration (E<3e–8). Further analysis of the eukaryotic MVP domain in common with the toxic ion resistance proteins revealed that the highest degree of similarity was found in a large internal repeat (residues 103–221; 156–276). An iterative search with this internal repeat (103–221) also identified the tellurite resistance protein (E=2e–14).

Having identified internally consistent evidence for homology between the eukaryote MVP repeat and TelA, a sequence alignment of two was carried out using the ClustalW program (Thompson et al., 1994) with the BLOSUM-30 substitution matrices (permitting both positive and negative match scoring)(Henikoff and Henikoff, 1992),a gap-opening penalty of 10 and a gap-extension penalty of 0.1. The portion of the sequence corresponding to the MVP domains 3 and 4 resolved viaNMR (Kozlov et al., 2006)yielded a net identity score of 31% relative to TelA, and a similarity score of 56%. These scores would normally suggest a probable homologous (potential superfamilial) relationship and a reasonable probability of a conserved domain 3/4 fold, but further scrutiny of the computed alignment relative to the three-dimensional structure revealed unphysical gaps within key MVP beta strands that would preclude common folding. Intuitive manipulation of the alignment to preferentially shrink gaps within strands at the expense of known coil regions, however, led to formulation of an alternative sequence alignment, reported in Fig. 1B,that retains 30% TelA identity and 55% similarity relative to the domain 3/4 region of MVP. This alignment meets the basic minimum criterion for meaningful homology modeling (∼30% sequence identity), and appears to be physically reasonable, thus we used it as the basis for comparative structure prediction for the portion of TelA in alignment with MVP domains 3 and 4. The TelA model was then constructed via the Modeller program(Marti-Renom et al., 2000). Default restraint settings were retained, and rigorous simulated annealing steps using the Chemistry at HARvard Molecular Mechanics (CHARMM) force field and charges (MacKerell, Jr et al.,1998) were employed in order to permit reasonable relaxation of the 12-residue TelA loop (MQSLPSIRLVQE), for which no MVP template residues exist. Specifically, five simulated annealing cycles of 4.4 ps stepwise warming (0 K→150 K→250 K→400 K→700 K→1000 K) followed by 19.2 ps stepwise cooling (1000 K→800 K→600 K→500 K→400 K→300 K) were performed. The resulting structure was then examined for unrealistic contacts and geometry via the ProCheck program(Laskowski et al., 1993). Further structural analysis on the proteins was carried out using the SYBYL suite of programs (Tripos,2006).

The primary antibodies used in this study were: anti-β-tubulin antibody Tub2.1 (Sigma-Aldrich, St Louis, MO, USA), anti-MVP antibody LDQN(Eichenmuller et al., 2003),anti-VPARP antibody ACE (a kind gift of Valerie Kickhoefer, David Geffen School of Medicine at UCLA, Los Angeles, CA, USA)(Kickhoefer et al., 1999),anti-PABP antibody 10E10 (a kind gift of Gideon Dreyfuss, University of Pennsylvania, School of Medicine, Philadelphia, PA, USA)(Gorlach et al., 1994),anti-HnRNPA2 antibody EF67 (a kind gift of William Rigby, Dartmouth Medical School, Lebanon, NH, USA) (Nichols et al.,2000), anti-TIA-1 3E6 (a kind gift of Nancy Kedersha, Brigham and Women's Hospital, Boston, MA, USA) (Taupin et al., 1995) and anti-ubiquitin antibody FK-2 (StressGen Biotechnologies, Ann Arbor, MI, USA).

HeLa cervical cancer cells (ATTC CCL-2) and human umbilical vascular endothelial cells (HUVEC; ATCC CRL-1730), were grown in T-75 tissue culture flasks in Dubecco's Modified Eagle's Medium (DMEM)/Hams F12 supplemented with 2 mmol l^–1 L-glutamine, 10% fetal bovine serum, 100 units ml^–1 penicillin and 100 μg ml^–1streptomycin. HeLa cells used in MVP siRNA knockdown assays were cultured in DMEM containing 4 g l^–1 glucose and supplemented with 4 mmol l^–1 glutamine, 10% fetal bovine serum (FBS), 500 units ml^–1 penicillin, 0.1 mg ml^–1 streptomycin and 0.01 mg ml^–1 of Tylosin. HUVEC culture medium additionally contained 15 μg ml^–1 endothelial cell growth supplement (Biomedical Technologies, Inc., Stoughton, MA, USA) and 100 μg ml^–1 heparin. The culture flasks for the HUVECs were coated with type-1 rat-tail collagen. All cells were maintained in an atmosphere with 5% CO₂ at 37°C, unless otherwise noted.

Adherent cells were detached from the flasks by a 5 min incubation(37°C) with a trypsin-EDTA solution (0.01% crystalline trypsin, 0.1% EDTA in divalent-cation-free Dulbecco's PBS). The cells were washed once with medium containing 5% FBS, diluted, seeded onto coverslips and grown for 24 h at 37°C. Cells were treated with either potassium tellurite(K₂TeO₃), sodium selenite(Na₂SeO₃), sodium selenate(Na₂SeO₄), sodium arsenite(Na₃AsO₃), sodium arsenate(Na₃AsO₄) or sodium orthovandate(Na₃VO₄), as described in the Results, and fixed with either –20°C methanol for 5–7 min or 2–4%paraformaldehyde for 15 min. No differences were observed in the staining patterns for the anti-vault antibodies with either fixation procedure so methanol fixation was used throughout the studies reported. The cells were rinsed three times with PBS and incubated with the primary antibodies for 60 min at 37°C. After rinsing the coverslips four times (10 min each rinse)with PBS, the cells were incubated with either Cy3- or Cy2-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA,USA). The coverslips were rinsed three times with PBS and once with 300 nmol l^–1 4′,6-diamidino-2-phenylindole (DAPI) in PBS to counterstain the nuclei. The coverslips were mounted in 90% glycerol and 10%PBS, pH 9.0 containing 4% (w/v) n-propyl gallate. Images were acquired by using a Nikon Optiphot microscope equipped with a Zeiss 63X, 1.3 NA objective lens and an Orca ER cooled CCD camera (Hamamatsu, Bridgewater,NJ, USA). Image acquisition was controlled by OpenLab software (Improvision,Inc., Lexington, MA, USA) and compiled using Adobe Photoshop.

In addition to verifying a potential homologous relationship between the MVP repeat unit and TelA, our reciprocal BLAST search retrieved several toxic ion resistance proteins from Staphlococcus aureus, Bacillus subtilis,Listeria monocytogenes, Clostridium acetobutylicum and Streptococcus pyogenes (first round of iteration, E<2e–6). As well, a detailed sequence comparison via the PROSITE database(Hulo et al., 2004) indicates that both the MVP repeat segment and its putative TelA homolog display extensive and reasonably strong similarity (greater than 60%) with S-layer paracrystalline bacterial surface coatings. These findings strongly suggest that aspects of mammalian MVP are more closely related to a variety of prokaryotic proteins, strongly bolstering the plausibility for a common ancestry between the eukaryotic MVP and TelA.

The highly similar spatial alignment of TelA and MVP shown in Fig. 2A, even after fairly thorough simulated annealing of the former, suggests that the fold exhibited by MVP is likely a viable conformation for TelA. Incorporation of an additional 12 residues of TelA (MQSLPSIRLVQE) into the coil between the 4th and 5th strands of the MVP sheet structure does not appear to impose substantial new strain within the TelA structure. In fact, ProCheck analysis(Laskowski et al., 1993)suggests that our predicted TelA structure may be under less strain (only nine bad atomic contacts flagged) than the MVP template itself (20 bad contacts) on which the TelA model was based, perhaps due to explicit solvent effects on the latter NMR structure. Otherwise, ProCheck analysis provided a comparable assessment of the physical soundness of both structures, determining an average G-factor score (accounting for all torsions, bond angles and bond distances) of –0.50 for MVP and –0.48 for TelA, both of which are well within the acceptable range.

An indication of the relative size, shape and surface charge distribution of our predicted TelA model and its MVP template is given in Fig. 2B. Given the fact that the portion of the TelA sequence corresponding to domains 3,4 of MVP resolved via NMR is longer by 22 amino acids (135 vs 113), it is not surprising that the predicted TelA volume (28 212 ų) and surface area (9083 Ų) are somewhat larger than that computed for MVP (24 096 ų and 7954 Ų,respectively). This is reflected in a marginally larger appearance in the rendered versions in Fig. 2B. The shape of the two proteins, however, is very similar, with the only major difference between the two being an additional lobe just below the upper right corner of the TelA structure, corresponding to the aforementioned unmatched 12 residues between the 4th and 5th strands of the sheet structure. This extended coil is located mostly on the surface of the TelA structure, which suggests that it is unlikely to significantly perturb the basic fold of the protein relative to MVP, but may yield modest differences in protein-protein associative tendencies, environmental preferences, and functionality. Beyond this, however, the two systems have very similar electrostatic/hydrophobic profiles, with polar residues constituting 41.7% of the MVP and 40.5% of the TelA surface areas. It is interesting to note, however, that while MVP has only a modest difference in the relative surface areas of cationic and anionic residues (1264 and 1185 Ų, respectively), the ratio is significantly more slanted toward surface cations in the case of TelA (1531 vs 1047 Ų).

We investigated whether MVP abundance or cellular location changed in response to tellurite treatment. Previously, we generated an antipeptide antibody (anti-LDQN), against conserved sequences within the amino terminus of the MVP, and showed that MVP localizes predominantly to the cytosol in a punctate staining pattern (Eichenmuller et al., 2003). In the HeLa cytosol, MVP is concentrated near the nucleus where the cell is the thickest(Fig. 3A). In addition,unidentified intranuclear speckles are also observed with this anti-MVP antibody. As you move outwards towards the cell periphery the number of MVP puncta diminishes. A similar staining pattern has been described with other MVP antibodies (Kedersha et al.,1990; Kickhoefer et al.,1999; van Zon et al.,2003). When potassium tellurite (K₂TeO₃) is added to the culture medium, MVP is no longer concentrated near the nucleus. In contrast, MVP is found in large, irregularly shaped cytoplasmic aggregates. In some cases, the aggregates are as large as 7 μm in diameter. The MVP aggregates are found predominantly at the cell margins where the cell makes contact with the substrate. Aggregate formation is temperature sensitive and can occur in 15 min or less and at tellurite concentrations as low as 5μm.

As mentioned above, nearly all the MVP in the cell is assembled into a vault particle (Kickhoefer et al.,1999; Siva et al.,2001). Therefore it is likely that vault particles move to the cell margins in response to tellurite treatment. To be certain that this is indeed the case, tellurite-treated cells were immunostained with a polyclonal antibody against the vault-associated poly(ADP-ribose) polymerase (VPARP). VPARP catalytically adds ADP-ribose polymers to itself and MVP and is located in three stave-like bands on the inside of the vault barrel(Kickhoefer et al., 1999; Mikyas et al., 2004). In HeLa cells, anti-VPARP antibodies stain peri-nuclear cytoplasmic puncta as well as intranuclear speckles (Kickhoefer et al.,1999) (Fig. 3F). In the presence of tellurite, VPARP staining occurs in large aggregates at the cell periphery (Fig. 3G). Since MVP and VPARP are integral components of vaults, it is likely that the peripheral accumulations of VPARP and MVP occur together in an intact vault particle. We refer to these large vault aggregates as vaultosomes.

Since HeLa cells are derived from an aggressive glandular cervical cancer,it is important to determine whether vaults respond to tellurite treatment in a non-transformed cell line. For these experiments, we examined vault location in HUVEC cells, an endothelial cell line derived from normal human umbilical cord. In untreated HUVEC cells, MVP is concentrated in the cytosol near the nucleus in a pattern that is indistinguishable from that seen in the aneuploid HeLa cell line (Fig. 4A). Upon tellurite exposure, MVP staining appears to be concentrated at the cell periphery in irregularly shaped patches or aggregates(Fig. 4B). Thus, vaults respond to tellurite in a similar manner in these two different human cell lines.

Exposure to 0.5 mmol l^–1 tellurite for 30 min does not affect cell viability (Ding et al.,2002). HeLa cells were exposed to 0.5 mmol l^–1tellurite for 30 min, washed with fresh medium and allowed to recover in tellurite-free medium overnight. Over a24 h period, the cell number had doubled in both the tellurite-treated and untreated cultures (compare Fig. 4E with H). In addition,the vaultosomes were no longer present at the cell margins. The MVP staining pattern in the tellurite-treated and recovered cells was identical to the untreated cells (compare Fig. 4C with E). These experiments lead us to conclude that vaultosome formation is a dynamic and reversible process.

Previous studies have shown that vaults bind directly to microtubules in vivo and in vitro(Eichenmuller et al., 2003; Hamill and Suprenant, 1997). To be certain that vault aggregation at the cell margin was not dependent upon microtubule reorganization or depolymerization, the cells shown in Fig. 4 were double stained with antibodies against tubulin and MVP. In untreated HeLa and HUVEC cells, the microtubules extend radially from the mitotic organizing center near the nucleus to the cell periphery. Exposure to tellurite did not have a great effect on the organization of the HeLa cell microtubules (compare Fig. 4F with G). In the HUVECs,the microtubule array also extends radially from the nucleus to the cell surface although the microtubules are somewhat collapsed in regions where the cells lost their attachment to the coverslips. Tellurite appears to be more toxic to the HUVECs. Even so, the vaults were still aggregated at the cell periphery (Fig. 4B). These experiments indicate that a catastrophic change in the microtubule array was not responsible for the tellurite-induced movement of vaults to the cell margin.

To determine whether the response was specific to tellurite, HeLa cells were treated with sodium arsenite, sodium arsenate, sodium orthovanadate,sodium selenite, sodium selenate or potassium tellurite (all at 0.5 mmol l^–1, 15 min, 37°C), fixed and examined by immunofluorescence. Vaultosomes formed in the presence of all of these oxyanions, albeit to different extents ranging from 25 to 40% of the cells examined (Fig. 5). Cells treated with arsenite are more likely to contain vault aggregates although differences amongst the various ions are not statistically significant. Moreover, the MVP aggregates found in cells treated with arsenite or tellurite appear brighter by immunofluorescence than the MVP aggregates found in cells treated with selenite, selenate or vanadate, for example.

Vaultosomes do not form during a general stress response. To determine whether vault aggregation was a general stress response, HeLa cells were heat-shocked (44°C, 60 min) or UV irradiated (100 μJ) and the location of the MVP was monitored by immunofluorescence. Similar to a previous study(Kickhoefer et al., 1999), no changes in the distribution of the MVP were observed with either treatment(data not shown).

Arsenite is known to inhibit protein synthesis and to accumulate untranslated RNAs into dense stress granules(Anderson and Kedersha, 2002). Since the vaultosomes also formed under toxic ion stress, we reasoned that the vaultosomes might be large enough to harbor the stress granules. To examine whether stress granules were associated with vaultosomes, we treated cells with arsenite and immunostained the cells with antibodies against vaults and the RNA-binding protein TIA-1, a robust marker of mammalian stress granules(Kedersha et al., 1999). Arsenite treatment induced the formation of vaultosomes and the smaller stress granules; however, neither particle colocalized with each other in the cell(Fig. 6). Furthermore,vaultosome formation was not affected by drugs (emetine, puromycin) that affect the dynamic equilibrium of RNAs in the stress granules(Kedersha et al., 2000) (data not shown). These results indicate that vaultosomes form independently of stress granules.

To examine whether other RNPs were associated with the vaultosomes, we treated cells with arsenite and immunostained the cells with antibodies against either the mRNA polyA-binding proteins-I (PABP-I) or the heterogeneous nuclear ribonucleoprotein A2, a protein implicated in translational regulation, RNA processing and RNA transport(Dreyfuss et al., 1993; Hamilton et al., 1999; Kwon et al., 1999). Neither RNA binding protein, PABP-I or hnRNP A2 appear to colocalize to a significant extent with the vaults in untreated cells or arsenite-induced vaultosomes(data not shown).

In response to proteosome inhibition, aggregates of ubiquinated proteins and proteosome subunits will accumulate around the microtubule organizing center of HeLa cells (Wojcik et al.,1996). These `aggresomes' form when the synthesis of misfolded proteins exceeds the capacity of the chaperones and the proteosomes(Johnston et al., 1998). We examined whether vaultosomes colocalized with ubiquinated proteins in the aggresomes when cells are treated with the proteosome inhibitor,N-acetyl-Leu-Leu-Norleucinal (ALLN). Fig. 7 shows that aggregates of ubiquinated proteins are found near the nucleus of the HeLa cells after treatment with ALLN. We were surprised to find that vaults also aggregated when the HeLa cells were treated with ALLN,however the vaultosomes did not colocalize with the ubiquinated proteins in the aggresomes. These results show that vaultosomes form in response to proteosome inhibition; however, vaultosomes do not colocalize with the aggresomes.

Elemental tellurium (Te), named from the Latin `tellus' meaning `earth',was discovered by F. J. Mueller von Reichenstein in 1782 from ores mined in the gold districts of Transylvania (Cooper,1971). In humans, elemental tellurium is very toxic and induces both chronic and acute toxicity, with death being the most extreme outcome. Although the toxic metalloid tellurium is relatively rare in the natural environment, several electrical, metallurgical and chemical manufacturing applications generate elevated levels of tellurium in the environment.

The toxicology of tellurium compounds on humans and other animals has been studied in detail, but the basic mechanism(s) of cellular toxicity is not known. For example, one of the earliest documented biological effects of tellurium poisoning is the demyelination of the sciatic nerve due to the inhibition of cholesterol biosynthesis and the accumulation of squalene in Schwann cells (Harry et al.,1989; Lampert et al.,1970). In bacteria, the toxicity of the soluble oxyanion tellurite, TeO ^–2₃, is thought to be due to its strong oxidizing effects on many enzymatic processes(Summers and Jacoby, 1977) and its ability to replace sulfur in metabolic reactions(Turner et al., 1995).

Although bacteria rarely come into contact with tellurium, with the exception of highly polluted environments, there are several different and seemingly unrelated tellurite resistance (Te^R) determinants that are found both on plasmids and in the genomes of unrelated bacteria(Taylor, 1999). In general,bacterial Te^R does not involve active efflux or reduced uptake of tellurium compounds, but rather the conversion of tellurite to a form that is less toxic to the bacteria. Some determinants involve basic cysteine metabolism (O'Gara et al.,1997) or methyl group transfer(Cournoyer et al., 1998) to detoxify tellurite. In E. coli, a basal level of tellurite resistance is brought about by nitrate reductase enzymes that are able to reduce tellurite and selenate at the cell membrane(Avazeri et al., 1997). In addition, the thiol:redox buffering system (glutathione and thioredoxin)appears to play an important role. One of the targets for tellurite toxicity in E. coli as well as erythrocyes is glutathione(Deuticke et al., 1992; Turner et al., 2001; Young et al., 1981). However,at this time there is no unifying theory of tellurite resistance. In this regard, tellurite resistance is thought to be a secondary effect of a normal metabolic or stress response pathway(Taylor, 1999).

The photosynthetic bacteria, Rhodobacter sphaeroides strain 2.4.1,has a surprising metabolic repertoire that includes anaerobic photosynthesis,aerobic and anaerobic respiration, as well as nitrogen and carbon fixation. R. sphaeroides survives in high levels of rare-earth oxides and oxyanions such as tellurite and selenite(Moore and Kaplan, 1992). Te^R in R. sphaeroides 2.4. is associated with two loci, trgAB and telA (O'Gara et al., 1997). The trgAB genes confer Te^R when introduced into the related bacterium Paracoccus denitrificans. In addition, a TelA null strain shows a partial decrease in resistance when compared to a wild-type strain. The mechanism of TrgAB and TelA associated resistance remains unknown.

The toxic anion resistance protein (TelA) family consists of several prokaryotic TelA-like proteins including the TelA protein of R. sphaeroides, D. radiodurans, B. subtilis and Y. pestis(O'Gara et al., 1997), the KIA protein associated with plasmid fertility inhibition(Whelan et al., 1995) and the KlaA protein from the methanomicrobe, Methanosarcina acetivorans.

In this report, we have shown that the eukaryotic major vault protein also shares significant sequence similarity with the TelA protein of Rhodobacter sphaeroides, and make a reasonable argument that the two share a common fold, spatial size and shape, and electrostatic profile. A solution NMR structure reveals a unique, novel fold consisting of a three-stranded antiparallel β-sheet(Kozlov et al., 2006). Our molecular modeling of the bacterial TelA protein reveals a common fold. Further evidence for a common MVP and TelA ancestry was obtained when it was shown that the MVP repeat segment and the TelA protein display greater than 60% similarity with S-layer paracrystalline bacterial surface coatings. These results indicate that the TelA protein may be a distant relative of the major vault protein.

In bacterial cells, it has been speculated that tellurite resistance is not the primary property of any of the unrelated tellurite resistance determinants(Taylor, 1999). The lack of similarity between any of the proteins associated with tellurite resistance has confounded their functional analysis. In addition, most bacteria rarely encounter high levels of tellurite in their environment so it is difficult to understand why so many bacteria maintain these genes in their genomes and on plasmids. If the primary function of the Te^R proteins is not to confer tellurite resistance why have they been conserved amongst so many divergent bacteria?

An intriguing hypothesis was put forth(Taylor, 1999) that the primary role of the Te^R determinants might be to protect the bacterium from mammalian host defenses. Many pathogenic bacteria are resistant to tellurite and one or more Te^R determinants have been identified in Y. pestis, M. tuberculosis and N. meningitidis, for example. In Taylor's model, the products of the Te^R genes are able to counterattack toxic substances produced by macrophages and other human cellular defenses. It is curious that mammalian vaults are very abundant in macrophages and in the dendritic cells of the immune system(Mossink et al., 2003; Mossink et al., 2002; Schroeijers et al., 2002). It is intriguing to speculate that macrophages inherited a tellurite-resistant gene from an ancestral pathogen and that vaults are an unrealized component of a metabolic pathway that originated in bacteria.

Movement of vaults to the cell surface in response to tellurite suggests that there is a link between MVP and the TelA protein. In response to a sublethal dose of K₂TeO₃, large vault aggregates appear at the cell margins. Aggregate formation is dynamic and reversible and occurs in normal cells grown in culture as well as cancer cell lines. Vaultosomes form independently of stress granules and from aggresomes.

Curiously, vaultosomes were induced by treatment of cells with ALLN, which is an inhibitor of neutral cysteine proteases. This result suggests that thiol-modified proteins may accumulate in the vicinity of vaultosomes and that vaults may be involved in the complex biochemical sensing and response to perturbations in the thiol:redox buffering system. It has been known for quite some time that selenium and tellurium compounds are thiol reactive reagents and all of these oxyanions perturb the thiol:redox buffering system(Bersin and Logemann, 1933; Valko et al., 2006). Our study shows that vaults respond dynamically to these toxic oxyanions. The large size of the vault interior indicates that vaults could harbor macromolecular complexes at the cell surface in response to tellurite uptake. Recent work shows that the vault shell can open and close transiently(Poderycki et al., 2006). Whether vaults contain complexes of organic telluroproteins(Ramadan et al., 1989),macromolecular complexes involved in the reduction of tellurite to elemental tellurium (Deuticke et al.,1992; Turner et al.,1992), or components of the thiol:redox system remains to be determined.

 
• ALLN

  N-acetyl-Leu-Leu-Norleucinal

 
• CHARMM

  Chemistry at HARvard Molecular Mechanics

 
• ERKs

  extracellular regulated kinases

 
• HUVEC

  human umbilical vascular endothelial

 
• MDR

  multidrug resistant

 
• MVP

  major vault protein

 
• NMR

  nuclear magnetic resonance

 
• PSI-BLAST

  position specific iterated basic local alignment search tool

 
• TelA

  toxic anion (tellurite) resistance protein

 
• TEP1

  telomerase-associated protein

 
• VPARP

  vault poly(ADP-ribose) polymerase

This project was supported by NSF MCB-9982377 and NIH Grants Numbers P20 RR015563 and P20 RR016475 from the COBRE and INBRE Programs of the National Center for Research Resources, with matching support from the State of Kansas,and the University of Kansas. N.B. was supported by the NSF Research Experiences for Undergraduates Program.