The antizyme family for regulating polyamines

The polyamines spermidine, spermine, and their precursor putrescine are organic polycations involved in various cellular processes and are absolutely essential for cellular proliferation. Because of their crucial function in the cell, their intracellular concentration must be maintained at optimal levels. To a large extent, this regulation is achieved through the activity of an autoregulatory loop that involves two proteins, antizyme (Az) and antizyme inhibitor (AzI), that regulate the first enzyme in polyamine biosynthesis, ornithine decarboxylase (ODC), and polyamine uptake activity in response to intracellular polyamine levels. In this Minireview, I will discuss what has been learned about the mechanism of Az expression and its physical interaction with both ODC and AzI in the regulation of polyamines.

The polyamines spermidine, spermine, and their precursor putrescine perform essential cellular roles, and most notable is their role in supporting cellular proliferation (1). For maximal effectiveness, optimal levels of polyamines must be maintained under varying physiological conditions. This requirement is fulfilled by tight regulation of key enzymes of their biosynthesis and catabolic pathways (Fig. 1). The biosynthesis pathway begins with the conversion of ornithine to putrescine by the action of ornithine decarboxylase (2). Putrescine is then converted to spermidine and spermidine to spermine by the sequential action of the two aminopropyltransferases, spermidine synthase and spermine synthase (3)(4)(5). These two enzymes utilize decarboxylated SAM 2 that is generated by the action of adenosylmethionine decarboxylase on SAM as an aminopropyl donor (6). Polyamines can be interconverted through the direct action of spermine oxidase or indirectly by polyamine oxidase that acts on spermine and spermidine after their acetylation by spermine/spermidine-N 1 -acetyltransferase (7)(8)(9). Although the biosynthesis and interconversion pathways can be regulated at several steps, the most notable regulation in this pathway is that of ODC (Fig. 1).
ODC is active as a loose homodimer that exists in a constant state of association and dissociation (Fig. 2) (10 -13). The rapid exchange of ODC subunits is a key trait enabling its regulation by a unique polyamine-induced protein called antizyme (Az) (14 -16). Az, a small polyamine-induced protein, is a central player in an autoregulatory loop that adjusts cellular polyamines (Fig. 2) and is found in animalia, fungi, and protista (17). Az binds ODC subunits with higher affinity than these subunits display to each other. The formation of the ODC-Az heterodimers results not only in the obvious inactivation of ODC but also in targeting ODC to ubiquitin-independent degradation by the 26S proteasome ( Fig. 2) (18,19). Polyamines regulate the synthesis of Az by stimulating a ribosomal frameshifting (20,21) that serves as a sensor of cellular polyamines. As a central element of the autoregulatory loop regulating cellular polyamine, it is not surprising that Az also negatively regulates polyamine uptake activity (Fig. 2) (22)(23)(24) via a yet unresolved mechanism. The term Az relates to a family of three members and a putative fourth member, termed Az1, Az2, Az3, and Az4 (17,25). Interestingly, Az itself is regulated by another protein termed antizyme inhibitor (AzI) that is highly homologous to ODC but lacks ornithine-decarboxylating activity (Fig. 2) (26). A high concentration of polyamines represses AzI expression by promoting the expression of an upstream open reading frame (ORF) initiated at a noncognate codon (27,28).

Az is expressed via a polyamine-stimulated ribosomal frameshift
Az proteins are encoded by partially overlapping ORFs. A unique process of ribosomal frameshifting is required to fuse these two ORFs to produce a completely functional Az protein (20,21). The first short ORF is terminated by an in-frame termination codon, whereas the functional Az protein is encoded predominantly by the second long ORF. Therefore, the translating ribosomes must be subverted from the first to the second ORF via a rather inefficient frameshifting that is stimulated by polyamines. The frameshifting site is composed of an essential stop codon and a preceding triplet that in most cases is UCC. The ribosome encountering the UCC codon shifts to the ϩ1 frame ( Fig. 3) (21). Interestingly, although when tested all three stop codons are equally efficient (20,21,29), in the vast majority of cases UGA is the stop codon that ends ORF1. Frameshifting efficiency is enhanced by recoding signals that are part of the RNA located both 5Ј and 3Ј to the frameshifting site ( Fig. 3) (30). The 5Ј-stimulatory segment that is composed of about 50 nucleotides can be divided into three modules whose conservation and importance increase toward the stop codon ( Fig. 3) (17,20,21). 3Ј to the stop codon, one finds a stimulatory pseudoknot and a nucleotide supporting the least efficient termination located just 3Ј to the stop codon ( Fig. 3)

cro THEMATIC MINIREVIEW
Interestingly, in the budding yeast Saccharomyces cerevisiae, the recoding signals actually reduce the otherwise high-efficiency frameshifting (31). It was demonstrated that the nascent Az polypeptide acts in cis as a polyamine sensor that negatively regulates frameshifting. When the polyamine concentration is low, the growing polypeptide inhibits downstream translation, whereas the increased concentration permits completion of Az's translation (31). A combined regulation of frameshifting by the growing polypeptide and by the nucleotide sequence of a 3Ј-stimulator was also demonstrated (32).

A family of Az proteins
Mammalian cells contain a family of Az proteins. There are three well-characterized paralogs: Az1, Az2, and Az3 and a less characterized putative member termed Az4 (25). Two of these members, Az1 and Az2, are ubiquitously expressed; however, Az2 is expressed to much lower levels (33). Nevertheless, Az2 displays higher evolutionary conservation compared with Az1, perhaps suggesting added functional value. Although Az1 tar-  ODC is a loosely bound homodimer that easily dissociates into monomers. ODC is the first in a chain of enzymes involved in the synthesis of polyamines. Polyamines stimulate a ϩ1 frameshifting event leading to the synthesis of active Az. Az, which has a higher affinity for ODC monomers compared with the affinity ODC monomers have to each other, traps them and delivers them to the 26S proteasome where they are recognized and degraded without requiring ubiquitination. Az is not degraded while delivering ODC to the proteasome. Instead, it is released to support another round of ODC degradation or is independently degraded by the proteasome in a ubiquitin-dependent process. Az also inhibit polyamine uptake via a yet unresolved mechanism. Az functions are negated by AzI, a monomeric nonfunctional homologue of ODC, whose synthesis is repressed by polyamines that stimulate translation of an upstream ORF. Like Az, AzI is also subjected to ubiquitin-dependent proteasomal degradation. THEMATIC MINIREVIEW: The antizyme family for regulating polyamines gets ODC to degradation both in vivo and in vitro, Az2 inactivates ODC in both systems but fails to support its degradation in vitro (34,35). This difference was attributed to Arg 131 and Ala 145 , whose simultaneous conversion to Asp severely inhibited in vitro activity of Az1 (36). The third member, Az3 appears to be tissue-specific being expressed predominantly in testis during certain stages of spermatogenesis (37,38). Although the frameshifting sequence UCCUGA is present in all three Az proteins, the 3Ј-recoding pseudoknot is not found in Az3 (37). All three Az proteins inhibit ODC activity and polyamine uptake; however, Az3 fail to stimulate ODC degradation (35). However, this should be tested in cells that naturally express Az3, as it was suggested that the physiological role of Az3 is to prevent overaccumulation of ODC after the stage of spermatogenesis at which it is required (37).
Translation of Az1 is initiated at two in-frame initiation codons resulting in the synthesis of two isoforms (Fig. 3) (39). The second AUG is utilized more efficiently because it is located within an optimal consensus sequence for translation initiation. The longer form is localized in the mitochondria as the segment located between the two initiation codons contains a mitochondrial localization signal (Fig. 3) (39). The mitochondrial localization signal is missing in Az2 and in Az3 (37).

Antizyme inhibitors
Remarkably, Az is itself regulated by a protein termed AzI. AzI is highly homologous to ODC, but it retains no ornithinedecarboxylating activity (26,40). Several independent reasons were suggested to account for the lack of ornithine-decarboxylating activity. AzI fails to form homodimers under physiological conditions (41), most likely due to differences in four amino acids from the dimer interface (42). This prevents the formation of an active site that in the case of ODC is located at the interface between the interacting subunits (13). In addition, AzI does not bind pyridoxal phosphate, an essential co-factor of ornithine decarboxylase (41). The affinity of AzI to Az is significantly higher than that of ODC, and therefore it rescues ODC from interaction with Az and degradation (26,44,45). In addition to the monomeric nature of AzI that makes it more available for interaction with Az, it was suggested that residues 125 and 140 from the Az-binding site are responsible for its higher affinity to Az (46). The ability of AzI to save ODC from degradation is emphasized by the observed elevation of ODC and of polyamine uptake activity in AzI-overexpressing cells (47). Moreover, these changes result in increased growth rate, growth in low serum, anchorage-independent growth in soft agar, and tumor development when injected into nude mice.
A second ODC-related protein, initially called ODC-like or ODCp (48,49), was eventually characterized as AzI and named AzI2 (48 -52). The originally identified AzI, AzI1, is ubiquitously expressed in rapidly proliferating cells, and AzI2 is expressed predominantly in differentiated testicular and neural cells and in secretory cells of various tissues (48,(53)(54)(55)(56)(57). A third member of the ODC family that was recently suspected as an additional AzI was eventually demonstrated as a leucine decarboxylase (58).

Degradation of Az and AzI
Az arrives to the 26S proteasome together with ODC. However, although ODC is degraded, Az is spared and is recycled to support additional rounds of degradation (Fig. 2) (59), or it is subjected to ubiquitin-dependent degradation (60). In yeast, polyamines not only stimulate Az synthesis by promoting frameshifting, but they also inhibits its ubiquitin-dependent degradation (61).
As AzI is highly homologous to ODC, it is not surprising that it is also a rapidly degrading protein. However, like Az, its degradation is ubiquitin-dependent and is not stimulated by Az (62). Actually, interaction with Az stabilizes AzI, apparently by inhibiting its ubiquitination (Fig. 2) (62). The long-term destiny of the AzI-Az complex is not presently known.

Interaction of Az with ODC and AzI
Two elements of a protein are required for its proteasomal degradation. The first is a segment mediating its recognition by the proteasome. Usually such segment mediates polyubiquitination of the target protein mediating its recognition by the proteasome. The second is an unstructured region that enables the initial penetration of the protein into the proteolytic cavity of the proteasome (63,64). Two segments of ODC were demonstrated to be required for its ubiquitin-independent degradation (65)(66)(67)(68). One region containing amino acids 117-140 serves as the Az-binding site. The second is the most C-terminal segment encompassing amino acids 424 -461 that appear to act as a degron, as it can be appended to stable proteins imposing on them rapid degradation (65, 66, 69 -71). In most rapidly degraded proteins, the recognition and infiltration regions are physically separated. In the case of ODC, it was originally thought that both are harbored within the unstructured C-terminal segment (72,73). Interestingly, in its native context the C-terminal segment functions rather inefficiently unless ODC interact with Az. It was suggested that the C-terminal region is buried within the structure of the native dimeric ODC and becomes exposed upon interaction with Az (74,75). However, more recent structural studies suggested that interaction with Az does not alter the structure of the most C-terminal segment of ODC, but rather it changes the structure of an adjacent segment encompassing amino acids 395-421 that might be involved in mediating the recognition of the ODC-Az complex by the proteasome (76,77). Compatible with this possibility is the demonstration that the ODC mutant lacking amino acids 424 -461 interacts efficiently with the proteasome but remains stable (77). It was therefore suggested that amino acids 395-421 mediate recognition by the proteasome, whereas amino acids 424 -461 initiate the entrance into the proteasome.
Like ODC, Az also contains two segments required for stimulating ODC degradation: a large C-terminal segment that mediates binding to ODC, and an N-terminal segment that stimulates the degradation (19,34,78). The close proximity of the proteasome recognition segment of ODC to Az (77) may suggest that segment of Az together with a segment of ODC constitute a complete recognition signal.
The C-terminal segment of AzI does not display significant homology to the C-terminal segment of ODC. Nevertheless, AzI is rapidly degraded in a ubiquitin-dependent manner.

THEMATIC MINIREVIEW: The antizyme family for regulating polyamines
Because interaction with Az interferes with its ubiquitination (62), it was suggested that this interaction imposes conformational changes that interfere with the ubiquitination process. Because the potentially ubiquitinated lysine residues of AzI are not masked by interacting with Az, it was suggested that this interaction may interfere with the recognition of AzI by the relevant E3 enzyme (77).

Az and AzI and cellular proliferation
ODC was suggested to act as an oncogene as its forced expression leads to cellular transformation (79). Az was suggested to function as a tumor suppressor (80). Because the synthesis of Az is completely dependent on the cellular concentration of polyamine, its involvement in regulating components of the polyamine biosynthesis pathway is rather obvious. However, several studies raised the possibility that Az may target the degradation of some proliferation-related proteins that are alien to the polyamine metabolic pathway. This includes cyclin D1, Smad1, Aurora-A, and a p73 anti-apoptotic variant (81)(82)(83)(84). However, the observation that massive overexpression of Az does not inhibit proliferation of cells expressing trypanosomal ODC, a form of ODC that is refractory to Az action, suggests that these proteins are not efficient targets of Az (85). Furthermore, some of these proteins remained stable in an in vitro degradation assay, whereas Az stimulated the degradation of co-incubated ODC. Recently, it was reported that under specific conditions Az2 stimulates nuclear ubiquitin-independent degradation of c-Myc (86).
Various studies suggested that AzI expression is elevated in several types of tumors (87)(88)(89)(90), suggesting that AzI may act as an oncogene (80). Compatible with this suggestion are the observations that overexpression of AzI in NIH3T3 cells resulted in their transformation (47), and suppression of AzI levels decreased cell proliferation (91,92). AzI can be even more effective than ODC in stimulating cellular proliferation and transformation as its overexpression stimulates both ODC and polyamine uptake activity. Furthermore, recent studies have demonstrated that increased A-to-I RNA editing of AzI leads to a gain-of-function phenotype leading to more aggressive tumor behavior (43,(93)(94)(95)(96). In one of these cases, the reported editing resulted in amino acid substitution yielding a protein with enhanced ability of neutralizing Az functions (93).

Conclusions
The recent characterization of the physical interactions of Az with ODC and AzI emphasized the need to further investigate the interaction of Az with ODC and AzI and the mechanism of recognition of ODC by the proteasome, including the identification of the proteasomal subunits involved in the recognition of ODC. Attention should be also given to resolving the mechanism by which Az and AzI are recognized by the ubiquitin system and to the mechanism by which Az regulates polyamine uptake. Of great interest will be the determination of the structural reason that despite their similarity drives AzI and ODC to ubiquitin-dependent and -independent degradation, respectively.