Thiopurine S-methyltransferase (TPMT, S-adenosyl-l-methionine : thiopurine S-methyltransferase; EC 188.8.131.52) catalyzes the S-methylation of thiopurine drugs such as 6-mercaptopurine (6-MP) and azathioprine as well as other aromatic and heterocyclic sulfhydryl compounds [1,2].
Weinshilboum and Sladek  reported trimodality for level of red cell TPMT among 298 randomly selected Caucasians: 88.6% had high enzyme activity, 11.1% had intermediate activity, and 0.3% had undetectable activity. This distribution conformed to Hardy–Weinberg expectations for a pair of autosomal codominant alleles for low and high activity, TPMT-L and TPMT-H, with frequencies of 0.059 and 0.941, respectively. Segregation analysis was consistent with this hypothesis. This genetic polymorphism has been shown to be an important factor in individual variations in response to thiopurine drug therapy. 6-MP is inactivated, in part, by S-methylation, catalyzed by TPMT. An alternative ‘metabolic activation’ process leads to the formation of cytotoxic 6-thioguanine nucleotides (6-TGN). In addition, 6-MP is metabolized to methyl-thioinosine monophosphate, which inhibits de novo purine synthesis, adding another mechanism of cytotoxicity [4,5].
Lennard et al. [6,7] showed that, in the children with acute lymphoblastic leukemia who were treated with 6-MP, red cell 6-TGN correlated inversely with RBC (red blood cell) TPMT activity (i.e. the lower the level of S-methylation, the more drug would be available for metabolism to form the cytotoxic 6-TGNs). It was also shown that individuals with very low RBC TPMT (i.e. those homozygous for that trait) were at greatly increased risk for life-threatening myelosuppression when they were treated with ‘standard’ doses of thiopurine drugs [8,9]. Conversely, patients with acute lymphoblastic leukemia who had 6-TGN concentrations below the group mean had higher TPMT activities and a higher subsequent relapse rate. Individuals heterozygous for functional variants tolerate 6-MP intermediate between homozygous deficient and homozygous wild-type patients . RBC TPMT levels have been found to correlate with levels in other tissues such as liver, kidney, and lymphocytes [11,12]. Therefore, the TPMT genetic polymorphism is a significant factor responsible for serious adverse drug reactions (myelosuppression) in patients treated with thiopurines and may also contribute to individual variation in therapeutic efficacy . TPMT has become one of a small number of examples in pharmacogenomics to be ‘translated’ into routine clinical care.
The human TPMT gene is 34 kb in length, consists of 10 exons, maps to chromosome 6p22.3  and has a pseudo-gene located on chromosome 18 . Twenty-eight variant alleles have been identified , most of which have been associated with decreased activity in vitro . Most of these involve nonsynonymous single-nucleotide polymorphisms (SNPs) [17,18]. Among those, TPMT*2,*3A, *3B, and *3C have been intensively studied both with regard to their clinical implications and/or molecular mechanisms. Szumlanski et al.  and Tai et al.  described TPMT*3A (Online Mendelian Inheritance in Man 187680.0002, http://www.ncbi.nlm.nih.gov/omim/), the most common variant allele associated with low TPMT activity in Caucasians (frequency approximately 5%). TPMT*3A contains two nonsynonymous cSNPs, one in exon 7 and another in exon 10 that result in Ala154Thr and Tyr240Cys alterations in encoded amino acids. TPMT*3B occurs rarely and contains only the exon 7 SNP while TPMT*3C contains only the exon 10 SNP and is the most common variant allele in East Asian, African–American and some African populations (frequency approximately 2%) [20–23]. More extensive population testing shows that *8 also occurs at a frequency of approximately 2% in some African populations .
TPMT*2, the first variant allele described, results in an Ala80Pro amino acid substitution (*2; Online Mendelian Inheritance in Man 187680.0001) . This allele is much less common than either TPMT*3A or *3C. It was also shown that expression of TPMT*2 and *3A were comparable in wild-type and mutant cDNAs, but that wild-type had an approximately 100-fold higher enzymatic activity than mutant TPMT . Gene expression did not correlate with protein activity in TPMT*2 and *3 [25,26]. Tai et al.  showed that enhanced degradation of TPMT allozymes encoded by the TPMT*2 and *3A alleles is the mechanism for decreased levels of TPMT protein and catalytic activity inherited as a result of these alleles.
Subsequent studies performed by Wang et al. [27,28] demonstrated that the rapid degradation of TPMT*3A involves molecular chaperones such as the heat shock proteins hsp70 and hsp90 and that TPMT*3A can also form intracellular aggresomes; both processes contribute to the low levels of protein and activity observed in the tissues of individuals with this allele.
This series of observations suggested that a dynamic balance might exist among TPMT*3A protein folding, protein degradation, and protein aggregation and raised the question of the identity of the proteins involved in these cellular processes. In an effort to answer that question, a Saccharomyces cerevisiae yeast gene-deletion library was used to identify genes required for the degradation/aggregation of TPMT*3A (Li et al. ). Twenty-four genes that fell into several functionally related categories were identified. The classes of genes involved included those affecting ubiquitin-dependent protein degradation (E2 ubiquitin-conjugating enzymes, E3 ubiquitin ligases, and proteasome subunits), vesicle trafficking, and vacuolar (lysosomal) degradation. The presence of genes involved in vesicular transport and vacuolar degradation suggested a possible role for autophagy in TPMT*3A degradation. UBE2G2, the human homologue of the E2 ubiquitin-conjugating enzyme identified by the yeast genetic screen, was shown to be involved in the degradation of TPMT*3A in mammalian cells. Further, expression of TPMT*3A induced autophagy and small interfering RNA-mediated knockdown of the expression of ATG7, an autophagy-related gene, enhanced TPMT*3A aggregation in mammalian cells, indicating that autophagy is also involved in TPMT*3A degradation.
TPMT*4 and *15 involve alterations in canonical mRNA splice site sequences [30,31], resulting in alternative TPMT mRNA splicing and decreased enzyme expression.
Recently, trinucleotide repeat variants in the TPMT promoter region have been described which may explain the 1–2% of Caucasians who show ultrametabolizer phenotype . In addition, a promoter region variable number tandem repeat which may have functional significance has been investigated [33–37].
For detailed mapping information, see http://www.pharmgkb.org/search/annotatedGene/tpmt/variant.jsp
- TPMT: *2 (Ala80Pro; rs1800462)
- TPMT: *3B (Ala154Thr; rs1800460)
- TPMT: *3C (Tyr240Cys; rs1142345)
- TPMT: *4 (rs1800584).
For other variants of known or suspected functional importance, see Table 1.