Studies over many years have suggested that increased polyamine synthesis may be necessary for neoplastic growth. This review summarizes recent work on the regulation of putrescine production both de novo and via the degradation of higher polyamines and provides a summary of studies using transgenic mice in which the levels of proteins that regulate these processes (L-ornithine decarboxylase, antizyme and spermidine/spermine-N1-acetyltransferase) are altered.
Introduction and putrescine synthesis
The only proven route in mammals for the de novo formation of putrescine is via the activity of ODC (L-ornithine decarboxylase) [1,2]. Putrescine can also be made from arginine via an ADC (L-arginine decarboxylase) and subsequent conversion of the agmatine into putrescine. This is a major pathway in plants and some micro-organisms but there is no convincing evidence for a mammalian ADC. Reports of a mitochondrial ADC and of a rat kidney ADC sequence could not be confirmed . A recent report that ODCp, a protein with sequence similarity to ODC , had ADC activity  was also not confirmed , and this protein has now been conclusively identified as an AzI (antizyme inhibitor) , as suggested on the basis of its amino acid sequence [3,4].
It is also possible for mammalian cells to take up either putrescine or agmatine via the polyamine transport system. Both amines are available in the diet and putrescine can also be generated from higher polyamines via the SSAT (spermidine/spermine N1-acetyltransferase)/AcPAO (acetylpolyamine oxidase) system, excreted from donor cells and then taken back up by other cells. However, it appears that de novo synthesis of putrescine via ODC is needed for neoplastic growth. This brief review summarizes some of the studies leading to this conclusion. Only recent work is covered along with some references to reviews summarizing many important earlier studies.
Control of ODC content
ODC is very highly regulated at levels of transcription, translation and protein turnover [2,7,8]. Increased synthesis of ODC occurs under the influence of oncogenes such as Myc and Ras. The most important factor limiting ODC content appears to be AZ (antizyme) (Figure 1). This protein binds to ODC and increases its binding to the 26S proteasome [2,9–11]. ODC is then degraded and the AZ is released to bring about further degradation of ODC. The rapid turnover of ODC also requires the C-terminal 37-amino-acid domain of the protein. Removal of this sequence or mutation of Cys441, which is located in it, causes ODC to be stable even in the presence of AZ. However, binding of AZ inactivates ODC since AZ binds to the ODC monomer and active ODC is an obligate homodimer. AZ may itself be regulated by the binding of a protein AzI (Figure 1), which binds to antizyme more tightly than ODC and thus releases ODC from the AZ–ODC complex [2,6,11,12]. AzI has no decarboxylase activity. AzI itself turns over rapidly but the AZ–AzI complex is stable . The physiological role of AzI is not yet fully established since it is present in very low amounts and results with knockout mice are not yet available. However, its mRNA increases when growth is induced and siRNA (small interfering RNA)-mediated down-regulation of AzI reduces polyamine content . AzI overproduction clearly influences polyamine content and cell proliferation. Recently, it has been shown that expression of AzI in 3T3 cells increases their proliferation and anchorage-independent growth [15,16]. These cells produced tumours when injected into nude mice.
Schematic diagram of the regulation of putrescine content
There is only one active ODC gene in mammals but both AZ and AzI occur as protein families. There are two established forms of mammalian AzI  and four of AZ itself [10,11]. There are also multiple forms of AZ-1 due to the presence of two potential start codons leading to 29 and 24.5 kDa forms. The content of AZ is increased in response to increased polyamines, with spermidine and spermine being more effective than putrescine. It is generally accepted that the major factor in this regulation is the ability of polyamines to increase productive translation of the AZ mRNA [10,11,17]. A+1 frameshift during translation of the AZ mRNA is needed to produce AZ and this is stimulated by polyamines. There may also be other effects contributing to polyamine-mediated AZ content. Transcriptional up-regulation of AZ by polyamines has been reported and AZ degradation, which occurs via the 26S proteasome after ubiquitination, is inhibited by polyamines [11,18].
Many ODC inhibitors are available  but the most widely used is DFMO (α-difluoromethylornithine) which acts as a suicide substrate for the enzyme and is highly specific but not very potent in vivo due to poor uptake and pharmacokinetics.
Spermidine and spermine are acetylated by SSAT, a highly inducible enzyme, which is normally present at very low levels. These levels are very hard to evaluate accurately and basal SSAT is almost invariably overestimated by the use of assays that merely measure the production of acetylated polyamines from acetyl-CoA. Enzymes other than SSAT can carry out this reaction and non-enzymatic acetylation also occurs. Definitive measurement of SSAT activity requires identification of the product, which should be exclusively at the N1 (aminopropyl end) of spermidine, and the use of a specific and potent antiserum to SSAT to show that all measured activity is lost when the SSAT protein is removed by immunoprecipitation. Unfortunately, no specific inhibitors of SSAT are available. SSAT is highly inducible by polyamines, polyamine analogues and some other stimuli including toxins and stress pathways [20,21]. Recently it was shown that stress-related induction of SSAT involves TNFα (tumour necrosis factor α) and NF-κB (nuclear factor κB)  and that induction via non-steroidal anti-inflammatory agents is mediated via the peroxisomal-proliferator-response element and PPARγ (peroxisome-proliferator-activated receptor γ) .
SSAT regulation by polyamines and analogues occurs at multiple levels including transcription, mRNA processing and protein stabilization. Increased transcription occurs via a polyamine-responsive element . The SSAT pre-mRNA can undergo alternative splicing by inclusion of an exon that contains premature termination codons. This is suppressed by polyamines . SSAT normally turns over very rapidly due to polyubiquitination and degradation via the 26S proteasome. Binding of polyamines or polyamine analogues prevents this [24,25] and leads to accumulation of SSAT protein, providing a dramatic amplification of the signal provided by increased mRNA production. Such accumulation of SSAT protein can increase SSAT activity and acetylation of polyamines substantially but it should be noted that binding of polyamine analogues also inhibits the enzymatic activity so the very high values for SSAT activity obtained by in vitro assays of extracts in which the analogue is diluted may be overestimates of the in vivo activity. The N1-acetylated forms of the polyamines formed by SSAT activity are degraded by AcPAO, a peroxisomal enzyme [20,26,27]. The SSAT/AcPAO system therefore forms putrescine from spermidine.
SSAT also acts well on spermine and AcPAO can convert this acetylation product into spermidine. However, another enzyme SMO (spermine oxidase) [20,28,29] can directly convert spermine into spermidine. SMO is not peroxisomal, and the H2O2 formed in its reaction may be more toxic than that produced by AcPAO. It should be noted that both pathways produce the potentially toxic aldehydes as well as H2O2. SMO is also inducible by polyamine analogues  and it is therefore difficult to identify whether the SSAT/AcPAO or the SMO system is responsible for toxic effects when these systems are activated. There are no specific inhibitors of AcPAO or SMO. Inhibitors based on allenic amines such as MDL72527, which have been used to block AcPAO, also inactivate SMO [26,29].
Polyamine excretion and re-uptake
Although many details remain unclear, there is a mammalian cellular uptake system for polyamines, including agmatine, and AZ blocks this system [1,2,8,30]. It also appears that there is an export system for putrescine and N1-acetylspermidine [1,8]. This system is even less well understood, but it clearly provides a mechanism by which putrescine and N1-acetylspermidine could be generated in one tissue or cell type via the SSAT/AcPAO pathway and then be available for uptake in other cells.
Role of ODC and AZ in carcinogenesis
Previous studies mainly by O'Brien and Gilmour and their collaborators have shown that a high level of ODC greatly facilitates tumour development in transgenic mice (reviewed [31–33]). In these experiments, elevated ODC expression was achieved in the skin using a stable C-terminally truncated form of ODC driven by a keratin 6 (K6) promoter. Tumour development was increased in these mice after a variety of stimuli including chemical carcinogens, UV radiation and an activated Ras oncogene [31–33]. The K6/ODC mice were much more susceptible than controls to tumour development after treatment with an initiating dose of a carcinogen such as DMBA (7,12-dimethylbenz[a]anthracene) and treatment with a tumour promoter such as PMA was not required for tumour development. Treatment with DFMO blocked the appearance of papillomas and caused a rapid regression of existing tumours in K6/ODC mice.
Conversely, there is convincing evidence that a low level of ODC reduces carcinogenesis. Cells and tissues derived from Odc+/− mice have a reduction in polyamine content and a 50% reduction in ODC activity in parallel to the reduced gene copy number [34,35]. Such Odc+/− mice developed substantially fewer skin papillomas than Odc+/+ littermates when exposed to a two-stage carcinogenesis protocol. In another model of tumour development, lymphoma development was strongly retarded in Eμ-Myc transgenic Odc+/− mice .
Transgenic overexpression of AZ also reduces carcinogenesis [36–40]. These experiments were carried out using an AZ cDNA construct with a single nucleotide deletion (T205) under the control of the K5 (keratin 5) or K6 promoters. This deletion removes the requirement for polyamine-stimulated frameshifting in the translation of AZ mRNA, but it is not totally ruled out that other post-transcriptional regulation of AZ synthesis can occur in the K5/AZ and K6/AZ mice. There was a reduced skin tumour incidence in these mice when subjected to two-stage initiation–promotion protocols, activated oncogenes and reduced tumour-suppressor genes combined with exposure to UV radiation [31,37,39].
Similarly, the incidence of tumours after exposure to chemical carcinogens was also reduced in other epithelial cell targets where the K5/AZ transgene is expressed, including tongue, oesophagus and forestomach ([31,38,40], and D.J. Feith, L.Y.Y. Fong and A.E. Pegg, unpublished work). Tumours in the oesophagus and forestomach can be induced in mice using a zinc-deficient diet to induce hyperproliferation followed by a single dose of NMBA (N-nitrosomethylbenzylamine). Tumour incidence and progression in this model is increased by inactivation of one or both p53 alleles. AZ expression suppressed forestomach tumour induction with marked reduction in both tumour multiplicity and size in both p53+/− and p53−/− mice . Chronic exposure to 4-nitroquinoline 1-oxide in the drinking water of zinc-deficient p53+/− mice results in tumours of the tongue in addition to the forestomach and oesophagus. AZ expression from K5/AZ in these mice greatly reduced the tumour incidence in all sites (D.J. Feith, L.Y.Y. Fong and A.E. Pegg, unpublished work). In the tongue, there was a large reduction in the incidence, multiplicity and size of the tumours as well as reduced progression to more aggressive invasive tumours (D.J. Feith, L.Y.Y. Fong and A.E. Pegg, unpublished work). These studies demonstrate that the tumour suppressive effects of AZ are not dependent on wild-type p53 function.
The mechanism by which increased AZ expression reduces tumour formation appears to be tissue dependent. In the NMBA-treated forestomach, K5/AZ mice exhibit both a decreased rate of cell proliferation and an increased rate of apoptosis . Interestingly, the increased rate of apoptosis in the presence of the K5/AZ transgene was maintained in the p53−/− background . However, little effect on apoptosis was seen in the skin in a model where transgenic AZ expression reduced tumour incidence in mice with an activated Mek oncogene  or in response to DMBA/PMA treatment. A detailed study showing that AZ expression in haematopoietic cells causes apoptosis through a mitochondrial-mediated pathway has appeared recently .
All of these studies are consistent with the stimulatory role of ODC and inhibitory role of AZ in tumour development described above. The simplest unifying hypothesis to explain these results would be that the production of putrescine by ODC is a critical and rather general factor in tumour development and the effects of AZ and AzI are mediated via the changes they produce in ODC. However, other roles, which are not related to ODC or polyamines, have been postulated for both AZ and AzI (see [2,12,42] and references therein). The evidence for these effects is currently scarce but it is hard to definitively rule out such effects. The general similarity in consequences between AZ expression and treatment with DFMO supports the ‘polyamine-mediated’ interpretation [38,39]. The extent to which AZ's ability to block polyamine uptake increases its effects is currently unknown. However, transgenic expression of a stable dominant negative form of ODC was not effective in blocking carcinogenesis in mouse skin . This protein would be expected to increase transport by binding and sequestering AZ.
Effects of SSAT on carcinogenesis
K6/SSAT transgenic mice were constructed in which a wild-type SSAT cDNA was driven by the K6 promoter. These mice have only a very small overexpression of SSAT, presumably because of the rapid turnover of SSAT described above and possibly the limited expression from this promoter. However, SSAT activity was increased when the mice were subjected to a two-stage tumorigenesis protocol . This may be due to either an increased expression from the K6 promoter or stabilization of the SSAT protein by increased polyamines. Remarkably, there was a striking increase in the treated K6/SSAT mice in both the number and size of skin tumours, and in their progression to carcinomas . Intestinal tumorigenesis in ApcMin/+ mice was also increased by transgenic expression of SSAT and decreased by removing SSAT via a gene deletion .
Widespread transgenic expression of SSAT using its own promoter or a metallothionein promoter led to a substantial reduction in spermidine/spermine pools, with a large increase in putrescine and in N1-acetylspermidine . These mice also showed a wide variety of other defects including hair loss, female infertility, weight loss, CNS effects and altered lipid metabolism and a tendency to develop pancreatitis. The pleiotropic effects of these changes complicate the use of these animals for specific studies in carcinogenesis but it was reported that they had a reduced development of papillomas in response to a two-stage skin tumorigenesis protocol . Similarly, when these mice were bred with TRAMP mice that develop tumours in the prostate due to an expression of SV40 T antigens from the probasin promoter, the incidence of prostate tumours was reduced . It is therefore apparent that the effects of perturbing polyamine catabolism are system dependent.
Further studies of the induction of skin tumours in K6/SSAT mice have evaluated the role of altered levels of polyamines and oxidative stress due to the SSAT/AcPAO pathway (X. Wang. D.J. Feith, A.E. Pegg and P.M. Woster, unpublished work). The increased tumour incidence was partially prevented by treatment with MDL72527, which prevented the increase in putrescine and the oxidation of acetylated polyamines causing an increase in N1-acetylspermidine. This result suggests that toxic products such as reactive oxygen species and aldehydes may enhance tumour development and rules out a direct effect of the increase in N1-acetylspermidine. As described above, MDL72527 also inhibits SMO. There was no detectable N1-acetylspermine, even in the presence of MDL72527, but there was a significant decline in spermine in the SSAT transgenic mice and a role for SMO cannot be ruled out. However, treatment with MDL72527 did not affect tumour development in the control mice. Therefore, the effect in K6/SSAT mice seems more likely to be mediated through prevention of the degradation of acetylated polyamines (X. Wang. D.J. Feith, A.E. Pegg and P.M. Woster, unpublished work).
Breeding of the K6/SSAT mice with K6/AZ mice blocked the development of tumours, and treatment of tumour-bearing K6/SSAT mice with DFMO resulted in the complete regression of established tumours, confirming the importance of de novo putrescine in the carcinogenic process . Treatment with a polyamine analogue inducer of SSAT, which substantially increased SSAT activity in the tumours, did not enhance regression. These results indicate that the tumour progression in K6/SSAT mice is dependent on elevated ODC activity and increased putrescine levels and may be further enhanced by oxidative stress. They support the use of cancer chemoprevention strategies that modulate polyamine levels through the inhibition of ODC activity or polyamine uptake, but not by increased SSAT expression.
Increased susceptibility to carcinogenic stimuli in mice overexpressing SSAT may be due to a compensatory increase in the biosynthesis of polyamines. In both the skin and the intestinal tumour models described above [44,45], transgenic SSAT expression led to increased tumour development and produced a large increase in putrescine but a small decline in spermidine and spermine. This is consistent with a general up-regulation of polyamine biosynthesis to replace those polyamines degraded via the SSAT/AcPAO pathway. Both the increase in putrescine and the metabolic consequences of the SSAT/PAO pathway may influence carcinogenesis.
Health Implications of Dietary Amines: A joint COST Action 922 and Biochemical Society Focused Meeting held at Medico-Chirurgical Hall, University of Aberdeen, U.K., 19–21 October 2006. Organized and Edited by H.M. Wallace (Aberdeen, U.K.).
We thank Dr L.M. Shantz, Dr L.Y.Y. Fong and Dr T.G. O'Brien for helpful discussion and comments. We apologize to the many contributors to this field whose work is not cited due to space constraints. Research on polyamines in the authors' laboratory is supported in part by grants CA-18138 and GM-26290 from the National Institutes of Health and a TSF award from Pennsylvania State University.