Evidence That a Mutation in the MLH1 3′-Untranslated Region Confers a Mutator Phenotype and Mismatch Repair Deficiency in Patients with Relapsed Leukemia*

Defects in DNA mismatch repair (MMR) are the molecular basis of certain cancers, including hematological malignancies. The defects are often caused by mutations in coding regions of MMR genes or promoter methylation of the genes. However, in many cases, despite that a hypermutable phenotype is detected in a patient, no mutations/hypermethylations of MMR genes can be detected. We report here a novel mechanism that a mutation in the MLH1 3′-untranslated region (3′-UTR) leads to MMR deficiency. A relapsed leukemia patient displayed microsatellite instability, but no genetic and epigenetic alterations in key MMR genes were identifiable. Instead, a 3-nucleotide (TTC) deletion in the MLH1 3′-UTR was found in the patient's blood sample. The mutant MLH1 3′-UTR was found to significantly reduce the expressions of both a firefly luciferase reporter gene and an ectopic MLH1 gene in model cell lines. Consistent with these observations, a significant reduction in the steady-state level of MLH1 mRNA was observed in white blood cells of the patient. These findings suggest that the mutant MLH1 3′-UTR can cause a severely reduced/defective MMR activity conferring leukemia relapse, likely by down-regulating MLH1 expression at the mRNA level. Although the exact mechanism by which the mutant 3′-UTR down-regulates the MLH1 mRNA is not known, our findings provide a novel marker for cancers with MMR defects.

Interestingly, despite that all above MSH and MLH genes are implicated in MMR, almost all MMR deficient cancers are associated with alterations in MSH2 and MLH1 (2-4) characterized by either genetic mutations in these genes or hypermethylation of the MLH1 promoter, which epigenetically silences the MLH1 expression (1). The selectively targeting on MSH2 and MLH1 for alterations in HNPCC and other MMR deficient cancers appears to be consistent with the fact MSH2 or MLH1 is an obligating subunit in each of the MutS and MutL heterodimers, whereas the other MSH and MLH components are functionally redundant (5). Tumor cells defective in MMR display a genome-wide mutator phenotype, e.g. frequent alterations in simple repetitive DNA sequences, a phenomenon called microsatellite instability (MSI) (2)(3)(4). However, not all cancer cells with MSI have identifiable mutations or epigenetic modifications in MMR genes (6 -8). For example, ϳ30% of MSI-positive HNPCC tumors are not associated with genetic or epigenetic alterations in any of the above MMR genes (2), suggesting that additional mechanisms are responsible for MMR deficiency in these MSI-positive tumors.
In this study, we demonstrate that a 3-nucleotide deletion in the 3Ј-untranslated region (3Ј-UTR) of MLH1 is associated with a relapsed leukemia patient displaying a mutator phenotype identical to cells defective in MMR. The mutant MLH1 3Ј-UTR appears to destabilize MLH1 mRNA and reduce MLH1 expression. Although exactly how the mutant MLH1 3Ј-UTR destabilizes MLH1 transcripts is unknown, the implication of this result for regulation of MMR in eukaryotic cells is discussed. Our findings here reveal a novel molecular mechanism by which the MMR system can be impaired.

EXPERIMENTAL PROCEDURES
Patient and Patient Samples-Patient AML0805 was diagnosed as acute myelogenous leukemia (AML) with cytogenetics showing a 45, XY, t(8, 21)(q22;q22), del(9) (q22q32) karyotype on August, 2005 at the University of Kentucky Hospital and treated with standard "7 ϩ 3" cytarabine plus idarubicin in which the patient went into complete remission (including negative FISH for t (8,21). This was followed by four courses of high dose cytarabine as postinduction therapy. However, the patient relapsed on June, 2006. Whole blood samples were obtained from the diagnostic (AML0805) and the relapsed patient (AML0606) according to an approved Institute Review Board protocol. Whole blood cells were fractionated by Ficoll-Paque (GE Healthcare) density gradient centrifugation, and the mononuclear white blood cell compartment was isolated and used to prepare genomic DNA and RNA using the FlexiGene DNA kit or QIAamp RNA blood mini kit (Qiagen), respectively, according to the manufacturer's instructions.
Cell Culture and Transfection-HeLa-S3 and 293T cells were grown in RPMI 1640 or Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Sigma) and 100 units/ml penicillin/streptomycin (Invitrogen). Cells were maintained in a humidified atmosphere of 5% CO 2 and 95% balanced air at 37°C and transfected using FuGENE HD Transfection Reagent (Roche Applied Science) at a 3:2 ratio of transfection reagent (l) to DNA (g).
Construction and Expression of the Luciferase Reporter with Wild-type and Mutant MLH1 3Ј-Wild-type (WT) and mutant (MT) MLH1 3Ј-UTR were cloned downstream of the firefly luciferase gene in the pGL3 vector to produce pLucϩ WT-3Ј-UTR and pLucϩMT-3Ј-UTR, respectively. The SV40 late poly(A) signal and SV40 enhancer in the pGL3 vector were deleted during this cloning step. The constructs were co-transfected with a Renilla luciferase (pRL-TK; Promega Corp.) into HeLa-S3 cells (ϳ2.5 ϫ 10 5 ), and the transfected cells were allowed to grow for 48 h prior to harvest. Luciferase expression was quantified using the Dual-Glo ™ luciferase assay system and a Lumat luminometer (Lumat; EG&G Berthold). A relative luciferase activity was calculated by normalizing the firefly luminescence to the Renilla luminescence. The pGL3 vector was used as a positive control, and the promoterless pGL3 vector was used as a negative control.
Construction and Expression of MLH1 with WT and MT 3Ј-UTR-The MLH1 gene was cloned into the BamHI and XhoI sites of pcDNA3.1/His C vector (Invitrogen), generating pcDNA3.1/HisC-MLH1. The WT and MT MLH1 3Ј-UTR amplified from genomic DNA derived from the patient's diagnostic or relapsed blood samples, respectively, were cloned downstream of the MLH1 gene in pcDNA3.1/HisC-MLH1 to generate pc-WT MLH1 3Ј-UTR and pc-MT MLH1 3Ј-UTR. The bovine growth hormone poly(A) sequence was deleted from the plasmid during this cloning step. The constructs were transfected into 293T cells, which were defective in MLH1 because of promoter hypermethylation, and MLH1 expression in each vector was determined by Western blot using a chemiluminescent system (GE Healthcare).
Analysis of mRNA Stability-293T cells were transiently transfected with pGL3 containing WT/MT 3Ј-UTR in 24-well plates. 24 h later, the cells were treated with 1 g/ml of actinomycin D (Sigma) to stop the transcription of the reporter gene. Cell aliquots were harvested at 0, 2, 4, 6, or 8 h after the treatment. Total RNA was extracted from the treated cells and employed to prepare the first-strand cDNA using TaqMan reverse transcription regents (Applied Biosystems) and random hexamer primers. Both the luciferase mRNA and ␤-actin mRNA were amplified from the cDNA using regular and realtime quantitative RT-PCR in a 7300 Real-Time PCR system (Applied Biosystems). The mRNA level in each case was measured using the power SYBR Green PCR Master Mix (Applied Biosystems). The luciferase mRNA level was normalized with that of ␤-actin mRNA by dividing the fluorescence emission intensity of the luciferase reporter with that of ␤-actin at the threshold cycle (C T ). The stability of MLH1 mRNA was determined similarly using total RNA (1 g) derived from patient samples after treatment with RQ1 RNase-free DNase (Promega) at 37°C for 30 min.
MSI Analysis-Five standard microsatellite markers from five different chromosomes (CHLC.GGAA4D07, D8S135, MfD257, ACTC, GGAA2E02) were used to determine MSI in AML0805 and AML0606 as described (9,10). Similarly, frameshift mutations in mononucleotide runs in exon 2 of CASPASE-5 and intron 5 of FANCD2 were investigated using the primers and conditions reported elsewhere (11) with minor modifications.
Statistical Analysis-Data from dual-luciferase assays and real-time quantitative PCR were analyzed using GraphPad Prism 4.0 software (GraphPad Software). Mean values (ϮS.D.) were calculated. Statistical significance was determined using Student's t test or one-way analysis of variance, and post-test was processed using Newman-Keuls multiple comparison tests. Data were considered statistically significant when p Ͻ 0.05.

Identification of a Mutator Phenotype in a Relapsed
Leukemia Patient-We recently investigated the impact of MMR deficiency on diagnostic and relapsed leukemia and provided compelling evidence showing that loss of MMR function is closely associated with leukemia relapse (12). Although the vast majority of the cases defective in MMR in the aforementioned study were because of either mutations in key MMR genes MSH2 and MLH1 or epigenetic modifications in the promoter region of MLH1, a patient, as described below, was found to carry a novel mutation that also lead to a mutator phenotype. This patient was diagnosed with AML on August 2005 and received standard chemotherapy (see "Experimental Procedures"). After achieving complete remission, the patient underwent relapse on June 2006. Blood samples collected from the patient at both the initial diagnosis and the relapse were designated as AML0805 and AML0606, respectively (12).
To determine whether the patient was associated with MMR defects, genomic DNA from blood samples of the patient was tested for MSI. Among five standard microsatellite markers examined, polymorphic PCR products were detected in a dinucleotide repeat marker ACTC between the patient's diagnostic and relapsed blood cells (Fig. 1A), indicative of MSI (13). Instability was also examined in mononucleotide repeat markers CASPASE-5(A) 10 (Fig. 1B) and FANCD2(T) 10 (Fig. 1C), where a run of 10 adenines and 10 thymines were present in exon 2 of CASPASE-5 and intron 5 of FANCD2, respectively. As shown in Fig. 1 (D and E), frameshift mutations of these mononucleotide repeats were observed in the patient's relapsed sample (see MT sequence in Fig. 1, D and E). These observations strong suggest that the relapsed leukemia adopted a hypermutable phenotype during chemotherapy.
Identification of a 3-Nucleotide Deletion in the 3Ј-UTR of MLH1 in the Relapsed Sample of the Patient-The mutator phenotype in the relapsed sample of the leukemia patient prompted us to search for alterations in key MMR genes MSH2 and MLH1 using PCR-based SSCP analysis, followed by DNA sequencing. This analysis did not reveal any mutations in exons and exon-intron junctions of either gene (data not shown). Hypermethylation analysis was also performed, but no such epigenetic modification was detected in the promoter region of MLH1 or MSH2 (data not shown). However, we did observe a deletion mutation in the MLH1 3Ј-UTR (Fig. 2). Two sets of primers were used to amplify the 3Ј-UTR and adjacent DNA sequence of MLH1. One primer set amplified exon 19 and the 3Ј-UTR, and the other set amplified the 3Ј-UTR and downstream untranscribed DNA sequences ( Fig. 2A). Both primer pairs generated one set of PCR products common to AML0805 (diagnostic sample) and AML0606 (relapsed sample), and one set of PCR products unique to AML0606. These bands were designated 1-4 and 1Ј-3Ј, as shown in Fig. 2, B and C. DNA sequencing analysis revealed that bands 1-4 corresponded to the wild-type MLH1 3Ј-UTR sequence (Fig. 2D, left). In contrast, DNA sequencing of bands 1Ј and 2Ј (products of primers ex19F and ex19R) and 3Ј (a product of primers 3ЈutrF and 3ЈutrR) revealed the presence of a 3-nucleotide (TTC) deletion in the 3Ј-UTR of MLH1 in AML0606 (Fig. 2D, center and right).
The observation of both wild-type and mutated alleles of the 3Ј-UTR seems to suggest a heterozygous mutation. However, based on the information from this and previous studies, we believe that the mutation is a homozygous one. First, the DNA used in this experiment was extracted from a heterogeneous population of white blood cells, which was likely to include leukemic and non-leukemic cells. Secondly, previous studies have revealed that individuals with heterozygous defects (e.g. germline mutations in HNPCC or heterozygous knockouts in mice) of an MMR gene generally possess a functional MMR system and do not display MSI (14). The identification of MSI in the patient relapsed blood sample suggests a complete loss of MMR function in a significant fraction of leukemic cells. Therefore, the detection of both the wild-type and mutant MLH1 3Ј-UTR is likely caused by the presence of both MMR proficient and deficient cells in the samples analyzed.
Functional Effect of the MT 3Ј-UTR on Reporter Gene Expression and mRNA Stability-To determine whether the 3-nucleotide deletion in the MLH1 3Ј-UTR altered expression of  10 (C) were amplified from DNA derived from the patient's diagnostic (AML0805) and relapsed (AML0606) white blood cells and analyzed in a 6% sequencing gel (A) or SSCP gels (B and C). Both normal and abnormal SSCP (arrows) products in B and C were cut and re-amplified followed by DNA sequencing analysis. Frameshift mutations (1-base deletion) were detected in both the CASPASE-5(A) 10 (D) and FANCD2(T) 10 (E) markers. WT, wild-type sequence (from normal SSCP products) and MT, mutant DNA sequence (from abnormal SSCP products). FEBRUARY 8, 2008 • VOLUME 283 • NUMBER 6 upstream coding sequences, the WT or MT MLH1 3Ј-UTR was cloned downstream of the firefly luciferase reporter gene (Fig.  3A). The reporter gene constructs were transfected into HeLa-S3 cells, and the expression of the luciferase and the stability of the reporter transcript were quantified in transfected cells. The results showed that the reporter construct with the MT 3Ј-UTR reproducibly produced ϳ2-fold lower luciferase activity than the construct with the WT 3Ј-UTR (Fig. 3B). Quantification of luciferase mRNA by real-time RT-PCR also showed a statistically significant decrease (p Ͻ 0.01) in the stability of the firefly luciferase transcript with the MT 3Ј-UTR. These results clearly indicate that the 3-nucleotide deletion in the MLH1 3Ј-UTR has significant functional effects on reporter gene expression and stability of transcripts from an upstream reporter gene.

Mutator Phenotype Associated with MLH1 3-UTR Mutation
Functional Effect of the MT 3Ј-UTR on MLH1 Expression-The effect of the MT MLH1 3Ј-UTR on expression of an ectopic copy of MLH1 was also examined. For this experiment, the MLH1 gene with WT or MT 3Ј-UTR was cloned downstream of the cytomegalovirus promoter (P CMV ) in the vector pcDNA3.1. MLH1 was expressed from this construct with an N-terminal His 6 tag (Fig. 4A). The constructs also included the gene encoding neomycin resistance and expressed the corresponding polypeptide, which conferred neomycin resistance to transfected cells. These expression constructs were transiently transfected into human kidney cell line 293T, in which the endogenous MLH1 gene was epigenetically silenced by promoter methylation (15), and the expression of ectopic MLH1 was monitored by Western blot. The neomycin resistance protein was used as an internal control. The results show that a much lower level of MLH1 was produced by the expression construct with the MT MLH1 3Ј-UTR (Fig. 4B). Thus, the 3-nucleotide deletion in the MLH1 3Ј-UTR reduces expression of either MLH1 or the luciferase reporter gene. These results suggest that the MT 3Ј-UTR may non-specifically decrease the stability and/or expression of co-linear RNA transcripts in cis.
To determine the effect of the MT 3Ј-UTR on the stability and/or steady-state level of MLH1 mRNA in relapsed leukemia of the patient, mRNAs isolated from the patient's relapsed (AML0606), and diagnostic (AML0805) white blood cells were analyzed for the steady-state level of MLH1 by RT-PCR. As shown in Fig. 4 (C and D), the MLH1 cDNA in AML0606 was at least 33% less abundant than in AML0805. The reduction in mRNA levels in AML0606 is statistically significant (p Ͻ 0.05). Nevertheless, it is worth noting that the impact of the 3Ј-UTR mutation on MLH1 cDNA expression could be underestimated by these data because the blood sample used for this experiment likely included a heterogenous mixture of cells with different MLH1 genotypes. These observations support a role for the MT 3Ј-UTR in leukemia relapse in AML0606. However, we cannot rule out the possibility that other factors also contribute to the relapse.

DISCUSSION
This study characterizes the MLH1 genotype in a relapsed leukemia patient, who was previously diagnosed with AML. The results indicate that leukemia relapse was associated with a FIGURE 2. Identification of a 3-nucleotide deletion in the MLH1 3-UTR in AML0606. DNA samples from white blood cells of AML0805 and AML0606 were used to amplify the MLH1 3Ј-UTR using the primers indicated in (A), and PCR products were analyzed by SSCP gels. B, SSCP gel of PCR products from primer pair ex19F/R. C, SSCP gel of PCR products from primer pair 3ЈutrF/R. D, DNA sequencing analysis of PCR products. Nucleotides TTC (indicated by a box in band 1) were missing in bands 2Ј and 3Ј, indicative of a 3-nucleotide deletion in the MLH1 3Ј-UTR in sample from AML0606. 3-nucleotide deletion in the 3Ј-UTR of MLH1 and that the MT MLH1 3Ј-UTR reduced MLH1 expression in vivo. In particular, the steady-state level of the MT MLH1 transcript in the patient's relapsed blood (AML0606) was much lower than the steady-state level of the WT MLH1 transcript in AML0805 (Fig.  4). This is likely due to the fact that the MT MLH1 3Ј-UTR destabilizes mRNA transcripts in cis, as shown here by its effects on transcripts from a proximal firefly luciferase reporter gene (Fig. 2).
This paper reports the novel observation that a 3-nucleotide deletion in the MLH1 3Ј-UTR reduces MLH1 expression in vivo, which could in turn reduce MMR efficiency. Although there is ample precedent for 3Ј-UTRs regulating expression of the proximal upstream gene (16 -18), to our knowledge, this is the first report of a functional regulatory role for the 3Ј-UTR of a human MMR gene. It will be of interest to determine the frequency at which this mutation occurs in leukemia patients treated with or without chemotherapy, in patients with other types of cancer (including HNPCC), and in healthy individuals. Although this study only reports one patient, we actually identified the same mutation in 3 out of 53 AML patients (12). Because blood samples available from the other two patients were collected during the course of their chemotherapy, it is difficult to perform similar analyses as described for patient AML0606. In addition, we have no clue if the mutation is therapy-induced or pre-existing at the diagnosis. Although these data must be interpreted with caution due to small sample number, they suggest a higher than expected frequency of this mutation in AML patients. Based on the fact that the mutation was detected in relapsed but not diagnostic cells of the AML patient, it is possible that the MT 3Ј-UTR could play a role in disease relapse. Because the MT MLH1 3Ј-UTR appears to destabilize the MLH1 transcript in white blood cells from AML0606 ( Fig. 4C and D), it is predicted that these cells could be hypomorphic for MLH1 and therefore functionally deficient in MMR. This possibility is consistent with the observation that AML0606 cells are MSI-positive in one of five microsatellite markers tested and contain frameshift mutations in mononucleotide repeat sequences of CASPASE-5 and FANCD2 genes (Fig. 1), both of which have been shown to be important targets for the mutation in MSIpositive cancers (11).
The mechanism by which the MT MLH1 3Ј-UTR destabilizes proximal RNA transcripts is not known. Previous studies show that cis-acting DNA sequence elements in 3Ј-UTRs and trans-acting protein factors influence mRNA stability. For example, the AU-rich element is an extensively studied cis-acting element commonly found in eukaryotic 3Ј-UTRs that binds to a cognate protein, AU-rich element binding protein, and their interaction stimulates mRNA turnover (19,20). Sequence analysis shows that the 3Ј-UTR of MLH1 does not contain any AU-rich element. However, secondary structure analysis (21) of the WT and MT 3Ј-UTR of MLH1 showed that the TTC-deletion identified in this study lies in a large stem-loop structure. The deletion of 3-nucleotide in the region reduced the size of stem-loop and created a new small stem-loop (supplemental Fig. 1). The impact of altering the size of this stem-loop and loop numbers on MLH1 mRNA structure, stability and/or capacity to bind trans-acting protein factors is not known. Recent studies implicate microRNA molecules as important regulators of gene expression, whose effects are often mediated through interactions with 3Ј-UTRs. It is highly possible that the microRNA structure change caused by the 3Ј-UTR mutation may allow different microRNAs to bind, which may have a negative effect on the mRNA stability.
It is noteworthy that the 3Ј-UTR mutation seems to have more negative effect on the protein level than the mRNA level (compare results in Figs. 3 and 4), suggesting that the mutation has an impact on both the transcription and translation of MLH1. However, the molecular basis by which the 3Ј-UTR 293T cells were transfected with the indicated plasmids, and cell extracts of the treated cells were subjected to Western blot analysis using MLH1 antibody. Expression of the neomycin resistance gene product, Neo-r, in the transfected cells was used as an internal (positive) control. Non-transfected cells were used as a negative control. Values in the bottom of the panel are the amount of MLH1 protein relative to the amount of Neo-r protein in each case. C and D, quantification of MLH1 mRNA in white blood cells from AML0805 and AML0606 analyzed by RT-PCR and real-time RT-PCR, respectively, using ␤-actin mRNA as an reference. Values shown in the bottom of panel C are the fractions of MLH1 mRNA relative to ␤-actin mRNA. *, the difference in steady-state mRNA levels between two samples is statistically significant (p Ͻ 0.05) as judged by Student's t-test. mutation down-regulates MLH1 expression at both the mRNA and protein levels remains to be investigated.