GASTROENTEROLOGY 2008;135:499 –507
`
`ALIMENTARYTRACT
`
`BASIC–
`
`Characterization of Mutant MUTYH Proteins Associated With Familial
`Colorectal Cancer
`
`MOHSIN ALI,*,‡,§,储 HYEJA KIM,* SEAN CLEARY,* CLAIRE CUPPLES,¶ STEVEN GALLINGER,* and ROBERT BRISTOW‡,§,储
`
`*Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada; ‡Applied Molecular Oncology, Ontario Cancer Institute/Princess Margaret
`Hospital, University Health Network, Toronto, Ontario, Canada; §Department of Radiation Oncology, 储Department of Medical Biophysics, University of Toronto,
`Toronto, Ontario, Canada; and the ¶Department of Biochemistry and Microbiology, University of Victoria, Victoria, British Columbia, Canada
`
`Background & Aims: The human mutyh gene encodes
`a base excision repair protein that prevents G:C to T:A
`transversions in DNA. Biallelic mutations in this gene
`are associated with recessively inherited familial colo-
`rectal cancer. The aim of this study was to characterize
`the functional activity of mutant-MUTYH and single-
`nucleotide polymorphism (SNP)-MUTYH proteins in-
`volving familial colorectal cancer. Methods: MUTYH
`variants were cloned and assayed for their glycosylase
`and DNA binding activities using synthetic double-
`stranded oligonucleotide substrates by analyzing cleav-
`age products by polyacrylamide gel electrophoresis.
`Results: In this study, we have characterized 9 mis-
`sense/frameshift mutants and 2 SNPs for their DNA
`binding and repair activity in vitro. Two missense
`mutants (R260Q and G382D) were found to be
`partially active in both glycosylase and DNA bind-
`ing, whereas 3 other missense mutants (Y165C,
`R231H, and P281L) were severely defective in both
`activities. All of the frameshift mutants (Y90X, Q377X,
`E466X, and 1103delC) were completely devoid of both
`glycosylase and DNA binding activities. One SNP
`(V22M) showed the same activity as wild-type MUTYH
`protein, but the other SNP (Q324H) was partially im-
`paired in adenine removal. Conclusions: This study
`of MUTYH mutants suggests that certain SNPs may
`be as partially dysfunctional in base excision repair
`as missense-MUTYH mutants and lead to colorectal
`carcinogenesis.
`
`Endogenous reactive oxygen species (ROS) are pro-
`
`duced continually during normal cellular physiology
`owing to the production of metabolite by-products. ROS
`also are produced exogenously by either ionizing radia-
`tion or chemical carcinogens. ROS can cause a variety of
`DNA damage including double-strand breaks, single-
`strand breaks, and DNA base lesions1–3 such that the
`repair of ROS-induced DNA lesions is important for
`preventing mutations and maintaining the stability of
`the genome.
`One such ROS-induced base lesion is 8-hydroxy gua-
`nine (GO), which is generated from guanine (G) or de-
`oxyguanosine triphosphate and leads to G:C to T:A and
`
`T:A to G:C transversions, respectively.4,5 In human cells, 2
`base excision repair (BER) proteins, OGG1 and MUTYH,
`initiate the repair of GO lesions through their DNA
`glycosylase activity, whereas another protein, MTH1, hy-
`drolyses oxidized deoxyguanosine triphosphate. These
`glycosylase activities remove the damaged base and leave
`behind an apurinic/apyrimidinic site in the DNA. In
`subsequent processes, the long patch BER pathway com-
`pletes the repair process and this requires the APE1,
`PCNA, pol ␦(␧), FEN1, and DNA ligase I proteins.6 The
`OGG1, MTH1, and MUTYH enzymes work in concert:
`OGG1 removes GO from the DNA when paired with cytosine
`(C), whereas MTH1 prevents incorporation of GO in the DNA
`strand from the nucleotide pool by hydrolyzing 8-oxo-deox-
`yguanosine triphosphate to 8-oxo-deoxyguanosine mono-
`phosphate. If GO escapes from initial OGG1 action, DNA
`polymerase incorporates adenine (A) opposite to GO, allowing
`MUTYH to instead excise A from the A:GO pair, this
`allows for a secondary attempt by OGG1 to repair the
`GO lesion.7–9 MUTYH also can excise other DNA lesions
`including 8-hydroxadenine (from 8-hydroxyadenine:G)
`2-OH-A (from 2-OH-A:G) and A from A:G mispairs to a
`lesser extent.10 –14
`Recently, it was reported that germ-line mutations of
`the MUTYH gene are associated with familial colorectal
`cancer (CRC). While studying a British family affected
`with multiple colorectal polyposis, Al-Tassan et al15
`found 2 compound heterozygotes in the MUTYH gene
`in affected patients that substituted tyrosine at residue
`165 to cysteine, and glycine at residue 382 to aspartic
`acid. These investigators concluded that these somatic
`G:C to T:A transversions constitute a genetic signature
`of defective MUTYH protein activity in these CRC
`patients. Later, other groups confirmed the finding and
`concluded that
`in MUTYH-associated polyposis (ie,
`MAP), biallelic mutations of the MUTYH leading to G:C
`to T:A transversions in the adenomatous polyposis coli gene
`
`Abbreviations used in this paper: BER, base excision repair; CRC,
`colorectal cancer; GO, 8-hydroxyguanine; MAP, MUTYH-associated pol-
`yposis; ROS, reactive oxygen species; SNP, single-nucleotide polymor-
`phism; WT, wild type.
`
`© 2008 by the AGA Institute
`0016-5085/08/$34.00
`doi:10.1053/j.gastro.2008.04.035
`
`GDX 1034
`
`

`

`500 ALI ET AL
`
`GASTROENTEROLOGY Vol. 135, No. 2
`
`could drive colorectal epithelial cell genomic instability
`and increased cell proliferation in the epithelium of the
`colon.16 –24 To date, more than 80 mutations have been
`described in the MUTYH gene within MAP patients, with
`the Y165C and G382D variants as the most common
`documented mutations in Caucasian populations.25
`One method to study the function of CRC-associated
`MUTYH variant proteins mutations is to express them as
`recombinant proteins and assay for BER activity in vitro
`using synthetic DNA substrates containing A:GO and
`8-hydroxyadenine:G base mismatches. This has led to
`biochemical characterization of a number of CRC-related
`MUTYH variant proteins including Y165C, G382D,
`R227W, V232F, and R231L. For example, Al-Tassan et
`al15 showed that the Y82C and G253D MutY proteins of
`Escherichia coli (analogous to Y165C and G382D variants
`of human MUTYH, respectively) were partially defective
`in removing adenine from A:GO pairs. Wooden et al26
`reported that bacterially expressed recombinant Y165C
`and G382D MUTYH proteins were completely devoid of
`glycosylase activity. In another study, the R227W and
`V232F MUTYH mutants also were found to be partially
`or severely defective in both DNA substrate binding and
`glycosylase activity. Yet, the latter mutants still bound to
`hMutS␣ (a heterodimer of the human mismatch repair
`proteins hMSH2 and hMSH6 and a complex that can
`enhance MUTYH activity). In addition, both of these
`mutants failed to complement bacterial MutY deficiency
`when expressed in E coli cells in vivo.27
`Very recently, the R231L MUTYH mutant was shown to
`be severely defective in DNA substrate binding and in ade-
`nine removal activity. Although this variant showed intact
`binding activity with hMutS␣, it did not complement MutY
`deficiency in E coli.28 Parker et al10 reported that lysates
`derived from lymphoblastoid cell lines from MAP pa-
`tients who expressed the mutants Y165C, G382D, and
`1103delC had lowered DNA binding and adenine cleav-
`age.
`We recently completed a large multisite, population-
`based, CRC case-control study (Cleary et al, unpublished
`data) on a total of 3835 CRC cases and 2889 controls in
`which all subjects were screened for 9 known germline
`MUTYH mutations using mass spectrometry. DNA from
`subjects with at least one mutation was screened further
`by denaturing high performance chromatography/WAVE
`and sequencing analysis of WAVE variance. We found
`that 27 cases and 1 control subject carried either homozy-
`gous or compound heterozygous MUTYH mutations;
`carriers were at increased CRC risk (adjusted odds ratio,
`18.3; 95% confidence interval, 2.5–133.6). Heterozygous
`MUTYH mutations were identified in 88 CRC cases and
`44 controls; carriers were at increased risk of CRC (ad-
`justed odds ratio, 1.47; 95% confidence interval, 1.02–
`2.13).
`Herein, we have conducted an in vitro study of these
`clinically relevant MUTYH mutations as they directly re-
`
`late to human cancer risk. We selected 5 missense, 4
`nonsense or frameshift mutations, and 2 single-nucle-
`otide polymorphisms (SNPs) of MUTYH protein for in
`vitro characterization of their glycosylase and DNA bind-
`ing activities. We show that 2 frameshift mutants are
`partially active in DNA glycosylase and binding activities
`whereas the other 7 variants are totally devoid of both
`activities.
`
`Materials and Methods
`Construction of Vectors of N-Terminal
`Glutathione S-Transferase (GST)-Tagged
`MUTYH Proteins
`The MUTYH gene was amplified using polymerase
`chain reaction from a Hela cell complementary DNA library
`(Stratagene, La Jolla, CA) using the primers: 5=CATATTGAAT-
`TCATGACACCGCTCGTCTCC3= and 5=CATACGTCGACT-
`CACTGGGCTGCACTGTTGA3= with Phusion high-fidelity
`DNA polymerase (Invitrogen, Carlsbad, CA). Polyacrylamide
`gel electrophoresis (PAGE)-purified oligonucleotides were pur-
`chased from Operon (Huntsville, AL). Gel-purified products
`were digested with EcoRI and SalI restriction enzymes (Invitro-
`gen) and purified using nucleotide purification spin columns
`(Qiagen, Valencia, CA). The doubly digested products were
`ligated into the pGEX-4T-1 vector (GE Health Sciences, Pisca-
`taway, NJ) between EcoRI and SalI sites using T4 DNA ligase
`(Invitrogen). A final pGEX-4T-1-MUTYH (wild-type [WT])
`construct was transformed and grown in DH5␣ cells. Both
`strands of the extracted plasmid were sequenced for the entire
`open reading frame of the cloned MUTYH gene and the DNA
`sequences were confirmed to be the same as previously pub-
`lished sequences.7
`Mutations were generated in the cloned WT MUTYH
`gene in pGEX-4T-1 using a site-directed mutagenesis kit
`as per the manufacturer’s instructions (Promega, San
`Luis Obispo, CA). The base sequences on both strands of
`each mutant were confirmed by DNA sequence analysis.
`
`Protein Induction and Purification
`Expression vectors were transformed into the
`BL21 CodonPlus (Stratagene, La Jolla, CA) (DE3) RIL E
`coli strain. Cells were grown in luria broth medium with
`100 ␮g/mL of chloramphenicol and ampicillin at 37°C
`to 0.6 optical density at 600 nm and cooled on ice.
`Protein production was initiated by adding 0.4 mmol/L
`isopropyl-beta-D-thiogalactopyranoside (IPTG)
`to the
`cells and continuing incubation at room temperature for
`2 hours. Cells were harvested, washed with ice-cold phos-
`phate-buffered saline (PBS), and stored at ⫺80°C until
`protein purification.
`Cell pellets were resuspended into ice-cold PBS (1/40th
`of culture volume). Lysozyme (1 mg/mL), dithiothreitol
`(5 mmol/L), and protease inhibitor were added to the cell
`suspension and incubated on ice for 30 minutes followed
`by 3 freeze/thaw cycles and ultrasonic disruption. Cell
`lysates then were centrifuged at 15,000 rpm for 15 min-
`
`ALIMENTARY TRACT
`
`BASIC–
`
`GDX 1034
`
`

`

`August 2008
`
`CHARACTERIZATION OF MUTANT MUTYH 501
`
`ALIMENTARYTRACT
`
`BASIC–
`
`other bands in the lane in the scanned image by Image-
`Quant (GE Health Sciences) software (Figure 1A, lane 3, and
`supplementary Table 1; see supplementary material online
`at www.gastrojournal.org).
`
`MUTYH Glycosylase Assay
`A 5=-Cy5–labeled 39-mer oligonucleotide containing A
`at the 21-mer position from the 5=-end was hybridized with its
`complementary strand in a 100 ␮L buffer containing 20
`mmol/L Tris-HCl (pH 8), 10 mmol/L ethylenediaminetetraace-
`tic acid (EDTA) (pH 8), and 150 mmol/L NaCl to make the
`following duplex DNA substrate containing an A:GO mis-
`match:
`5=cyTGAGACTGGCCAGCTAACTGAACTGATCATGCCTAGCGT
`ACTCTGACCGGTCGATTGACXTGACTAGTACGGATCGCA,
`
`Where, X ⫽ GO.
`The glycosylase assay was performed according to the
`procedure used by Bai et al27 with slight modification.
`Briefly, 100 fmol 5=-Cy5–labeled duplex was incubated
`with 15 pmol wild-type or mutant MUTYH at 37°C for
`30 minutes in 10 ␮L buffer containing 50 mmol/L EDTA
`(pH 8), 500 ␮mol/L ZnCl2, 250 mmol/L HEPES (pH 7),
`and 1.5% glycerol. The reaction was stopped by adding 10
`␮L denaturing PAGE gel loading buffer containing 10
`mmol/L EDTA (pH 8), 98% formamide, 10 mg/mL blue
`dextran, and 200 mmol/L NaOH followed by heating at
`90°C for 30 minutes. Cleavage products were separated
`using a 14% denaturing PAGE gel running at 1400 V and
`100 W for 1 hour and fluorescent bands on the gel were
`visualized using a Typhoon Variable Imager (Amersham
`Biosciences, Piscataway, NJ). A schematic of this assay is
`shown in supplementary Figure 1A (see supplementary
`material online at www.gastrojournal.org). Enzymatic re-
`action rates were calculated at the 0–4 minute time points
`by linear regression.
`
`4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™
`Figure 1. Analysis of induction and purification of recombinant N-
`termini tagged GST-MUTYH mutant. (A) Cell lysates were resolved by a
`10% sodium dodecyl sulfate–PAGE followed by Coomassie-blue stain-
`ing to visualize expressed proteins. For each MUTYH protein: lane 1,
`lysate from uninduced cells; lane 2, lysate from IPTG-induced cells; lane
`3, partially purified GST-tagged protein. (B) Western blot analysis of
`purified recombinant GST-tagged MUTYH and its mutants using rabbit
`polyclonal antibody against MUTYH (C-termini–truncated variants
`Y90X, Q377X, E466X, and 1103delC were not detected because the
`antibody corresponded to amino acids 513-546 at the C-terminal of the
`MUTYH protein) or (C) mouse monoclonal antibody against GST. Fluor-
`conjugated secondary antibody bound to the antibody-antigen com-
`plex was detected by infrared image analyzer (LICOR, Lincoln, NE). (B
`and C) Lane assignment is as follows: lane 1, WT; lane 2, V22M; lane 3,
`D222N; lane 4, Q324H; lane 5, Y90X; lane 6, Q377X; lane 7, 1103delC;
`lane 8, Y165C; lane 9, R231H; lane 10, P281L; lane 11, R260Q; lane
`12, G382D; and lane 13, E466X; arrows indicate positions of the in-
`duced protein bands.
`
`utes and the clear supernatant was saved. A 50% slurry of
`glutathione sepharose beads (GE Health Sciences) was
`added to the clear lysate and rocked overnight at 4°C.
`Beads were pelleted by centrifugation at 2000 rpm for 5
`minutes and washed 4 times with ice-cold PBS. Finally,
`protein-bound beads were diluted with ice-cold PBS to
`make a 75% slurry. The concentrations of the partially
`purified proteins were estimated by the bicinchoninic acid
`method and proteins were stored at ⫺80°C until use. The
`purity of induced proteins was estimated by quantifying the
`band at the relevant molecular weight compared with all
`
`GDX 1034
`
`

`

`502 ALI ET AL
`
`GASTROENTEROLOGY Vol. 135, No. 2
`
`DNA Binding Assay
`For the DNA binding assay, GST-tag MUTYH
`proteins were eluted from the glutathione sepharose
`beads by an elution buffer containing 10 mmol/L re-
`duced glutathione and 50 mmol/L Tris-HCl, pH 8, and
`concentrated using a centrifugal filter (Millipore, Bed-
`ford, MA). A total of 100 fmol of Cy5-labeled 39-mer
`duplex DNA substrate containing A:GO base pair was
`incubated with 15 pmol partially purified, beads-free pro-
`teins at 37°C for 30 minutes in 18 ␮L buffer containing
`10 mmol/L Tris-HCl (pH 7.6), 0.5 mmol/L dithiothreitol,
`0.5 mmol/L EDTA, and 1.5% glycerol. The reaction mix-
`ture was supplemented with 2 ␮L of loading buffer (50%
`glycerol and 10 ␮g/␮L blue dextran) and analyzed by 6%
`nondenaturing PAGE gel
`in Tris-Borate-EDTA buffer
`running at 100 V at 4°C. The fluorescent bands on the
`gel were visualized using a Typhoon Variable Imager
`(Amersham Biosciences).
`
`Results
`Expression and Purification
`MUTYH is a human DNA glycosylase that re-
`moves A preferentially from A:GO pairs in DNA to pre-
`vent G:C to T:A transversions. In this study we charac-
`terized the adenine removal and DNA substrate binding
`activities of a series of MUTYH variants (ie, Y90X, Y165C,
`R231H, R260Q, P281L, Q377X, G382D, E466X, and
`1103delC), which are derived from MAP-phenotype pa-
`tients.
`Initially, we expressed these mutant proteins in vitro
`from a pTNT expression vector (Promega) using the TNT
`SP6 Quick Coupled Transcription/Translation system
`(Promega) and a significant amount of proteins were
`found to be induced. However, these proteins were found
`to be inactive. However, in subsequent studies using an in
`vivo approach, we were able to successfully induce N-
`termini GST-tag proteins from the pGEX-4T-1 expres-
`sion vectors in BL21 CodonPlus RIL E coli host cells and
`partially purify them with glutathione sepharose beads.
`We therefore expressed mutants and WT proteins, 2
`SNPs (V22M and Q324H), and enzyme-active center mu-
`tant (D222N) as a positive control, a pseudopositive
`control, and a negative control, respectively.
`The sodium dodecyl sulfate–PAGE analysis of the
`GST-tag recombinant MUTYH wild-type protein and its
`variants showed that the full-length (535 amino acid)
`proteins WT, V22M, D222N, Q324H, Y165C, R231H,
`R260Q, P281L, and G382D all had the expected and
`similar molecular weights based on the number of ex-
`pressed codons. The C-termini–truncated Y90X, Q377X,
`E466X, and 1103delC MUTYH mutant proteins were
`expressed as proteins with expected lower molecular
`weights of 37, 68, 78, and 70 kilodaltons, respectively.
`The specificity of protein induction was confirmed by
`
`Western blot analyses using polyclonal antibodies against
`MUTYH or GST (Figure 1).
`To ensure that the exogenous partially purified pro-
`teins were not contaminated with endogenous bacterial
`homolog MutY, we purified the WT, V22M, and D222N
`MUTYH proteins from uninduced and IPTG-induced
`cells and assayed their activities on duplex DNA sub-
`strates containing an A:GO mismatch. We observed
`that the proteins from uninduced cells harboring the plas-
`mids pGEX-4T-1-MUTYH (WT) and pGEX-4T-1-MUTYH
`(V22M) possessed slight adenine removal activity (compare
`lanes 4 and 6 in supplementary Figure 1B; see supplemen-
`tary material online at www.gastrojournal.org). This ac-
`tivity could be from the co-eluted bacterial MutY protein
`from the E coli host cells (ie, the BL21 CodonPlus RIL
`strain is not MutY-deficient) or from leaky MUTYH pro-
`tein. However, the D222N variant is an inactive protein
`because of a mutated enzyme-active site26 and it should
`not show any glycosylase activity. Indeed, the D222N
`variant partially purified from uninduced or IPTG-in-
`duced cells was found to be completely inactive (compare
`lanes 2 and 3 in supplementary Figure 1B; see supplemen-
`tary material online at www.gastrojournal.org). This sug-
`gests that the glycosylase activity detected in the unin-
`duced cells was from the leaky WT or V22M protein.
`Moreover, sodium dodecyl sulfate–PAGE analysis of the
`partially purified proteins did not show any band at 39
`kilodaltons that corresponded to the molecular weight of
`MutY. Therefore, we conclude that the N-termini GST-tag
`recombinant WT and mutant MUTYH proteins partially
`purified from the BL21 CodonPlus (DE3) RIL host cells
`are free from MutY contamination. This GST-tagged
`recombinant WT type MUTYH protein was almost as
`active as the standard bacterial homolog MutY protein.
`
`Adenine Removal Activity (Glycosylase
`Activity) Assay
`Figure 2 shows a typical in vitro glycosylase assay
`profile of wild-type or mutant MUTYH proteins on the
`synthetic duplex DNA substrates containing an A:GO
`pair. Similar to the bacterial homolog MutY control, the
`WT, V222M, and Q324H MUTYH proteins all were able to
`cleave substrates containing an A:GO mismatch (compare
`lanes C4, C5, and C6, respectively, with lane C3 in Figure 2).
`The major slower migrating band is a 20-mer ␣, ␤-unsat-
`urated aldehyde generated by ␤-elimination of
`the
`apurinic/apyrimidinic site by NaOH treatment. The mi-
`nor product migrating faster than the major product is a
`20-mer product with 5=-phosphate and produced from
`the ␤-elimination product by (cid:1)-elimination. Ohtsubo et
`al12 reported similar products from the substrates incu-
`bated with wild-type MUTYH.
`The intensity of the bands from the cleavage products
`generated by 2 missense MUTYH mutants (ie, R260Q
`and G382D) was much less compared with the bands
`from wild-type MUTYH protein (compare lanes C4 with
`
`ALIMENTARY TRACT
`
`BASIC–
`
`GDX 1034
`
`

`

`August 2008
`
`CHARACTERIZATION OF MUTANT MUTYH 503
`
`ALIMENTARYTRACT
`
`BASIC–
`
`Figure 2. Enzymatic activity of MUTYH mutants on 39-mer duplex DNA substrate containing A:GO mismatch. A total of 100 fmol 5=-Cy5–labeled
`duplex substrate was incubated with 15 pmol WT or mutants MUTYH at 37°C for 30 minutes and analyzed in a denaturing PAGE to detect cleavage
`products. Lane assignment is as follows: lane S, 20p-mer standard; lane C1, substrate only; lane C2, GST beads; lane C3, MutY (positive control);
`lane C4, WT (positive control); lane C5, V22M (pseudopositive control); C6, Q324H (pseudopositive control); C7, D222N (negative control); lane 1,
`Y90X; lane 2, Y165C; lane 3, R231H; lane 4, R260Q; lane 5, P281L; lane 6, Q377X; lane 7, G382D; lane 8, E466X; and lane 9, 1103delC.
`
`lanes 4 and 7 in Figure 2). The other missense mutants
`(Y165C, R231H, and P281L) and the frameshift mutants
`(Y90X, Q377X, E466X, and 1103delC) generated no
`products (Figure 2, lanes 1, 2, 3, 4, 5, 8, and 9) even after
`2 hours of incubation.
`The time-course assay of A:GO repair activity of WT and
`SNP-V22M MUTYH protein indicated that the generated
`products reached a maximum within 16 minutes and pla-
`teaued at periods of up to 180 minutes (Figure 3A). The rate
`constants k2 for SNP-Q324H MUTYH and WT were found
`to be 4.288 ⫾ 0.7831 min⫺1 and 4.507 ⫾ 0.5812 min⫺1,
`respectively (Figure 3B). We reproducibly detected only
`66% activity in the SNP-Q324H MUTYH compared with
`the WT at 2 minutes. The glycosylase activities of both
`missense mutants (R260Q and G382D) were 36% (k2 ⫽
`1.28 ⫾ 0.122 min⫺1) and 7% (k2 ⫽ 1.033 ⫾ 0.0979 min⫺1),
`respectively of the WI at 2 minutes (Figure 3). In contrast,
`the missense mutants (Y165C, R231H, and P281L) and
`frameshift mutants (Y90X, Q377X, E466X, and 1103delC)
`produced no products, even after 10 hours of incubation.
`
`DNA Binding Activity Assay
`The binding of repair proteins to DNA substrates
`is a crucial step in the repair process. Human cells recruit
`MUTYH protein to the A:GO site where it binds tightly
`to DNA and catalyzes the removal of A from A:GO
`mismatch. MUTYH and its bacterial homolog MutY re-
`main bound to the substrate even after removal of A,
`until displaced by other proteins that subsequently are
`recruited to complete the repair process. In this work, we
`also examined the effect of amino acid substitution in
`mutant MUTYH on DNA binding activity. Figure 4A
`shows a typical gel shift assay for wild-type and mutants
`of MUTYH.
`The WT and SNP MUTYH proteins formed 2 com-
`plexes with the duplex DNA substrates containing A:GO
`mismatch, whereas the MutY bacterial homolog formed
`only 1 complex (Figure 4A, lanes 2, 3, and 4). The sub-
`
`strates bound to WT MUTYH protein as complexes I and
`II were found to be 52% (26 fmol/pmol protein) and 22%
`(11 fmol/pmol protein), respectively. The missense mu-
`tants (R260Q and G382D) also formed complexes I and
`II with the DNA substrates. The DNA substrates bound
`to R260Q protein were estimated to be 18% (9 fmol/pmol
`protein) and 8% (4 fmol/pmol protein), respectively. On
`the other hand, 22% (11 fmol/pmol protein) and 14% (7
`fmol/pmol protein) substrates were bound with G382D
`as complexes I and II, respectively (Figure 4, lanes 9 and
`12). No such complexes were formed with the missense
`mutants (Y165C, R231H, and P281L) or with the frame-
`shift mutants (Y90X, Q377X, E466X, and 1103delC) (Fig-
`ure 4A, lanes 6 – 8, 10, 11, 13, and 14).
`
`Discussion
`Compound heterozygotes in the mutyh gene have
`been shown to be associated with familial CRC in human
`beings. In this work, we studied 9 bacterially expressed
`mutant MUTYH proteins for their DNA glycosylase and
`binding activities. In vitro assay using synthetic DNA
`substrates revealed that missense mutants (R260Q and
`G382D) are partially active in glycosylase activity (rate
`constants k2, 1.28 ⫾ 0.122 min⫺1 and 1.033 ⫾ 0.0979
`min⫺1, respectively, compared with 4.507 ⫾ 0.5812 min⫺1
`of WT) and DNA binding activity (26% and 36% of the
`substrates as complex I and II with R260Q and G382D,
`respectively) whereas mutants (Y90X, Y165C, R231H,
`P281L, Q377X, E466X, and 1103delC) are unable to
`generate any detectable cleavage products from the sub-
`strates containing A:GO mismatch or to bind to the
`substrates (Figures 2– 4).
`Previously, Al-Tassan et al15 showed that the E coli mu-
`tant Y82C (analog to human Y165C) shows barely detect-
`able glycosylase activity, whereas the mutant G253D (anal-
`ogous to the human G382D variant) cleaves adenine from
`an A:GO mismatch almost as efficiently as WT protein. And
`
`GDX 1034
`
`

`

`504 ALI ET AL
`
`GASTROENTEROLOGY Vol. 135, No. 2
`
`for the prediction of activity of clinical mutations in
`MUTYH based on sequence data.
`Importantly, our DNA binding results are consistent
`with the relative glycosylase activity of mutant MUTYH
`proteins. Frameshift mutants (Y90X, Q377X, E466X, and
`1103delC) were unable to remove A from the substrates
`(although they possess intact catalytic domains), proba-
`bly because of defective DNA-binding activity. Why the
`DNA-binding activity of the full-length (535 amino acid)
`missense MUTYH variants (Y165C, R231H, and P281L)
`was severely defective is unclear at this point and this
`deserves further study.
`We observed that despite similar substrate binding
`activity (74% and 68%, WT and Q324H, respectively), the
`SNP Q324H is only 66% active at 2 minutes in excising A
`(k2 ⫽ 2.971 ⫾ 0.2172 min⫺1) from the substrates com-
`pared with the WT (k2 ⫽ 4.507 ⫾ 0.5812 min⫺1). In
`contrast, Shinmura et al29 found Q324H to be fully active
`as WT. Interestingly, Yuan et al30 just reported that SNP
`Q324H is associated strongly with familial CRC among
`
`Figure 4. Binding of WT and mutant MUTYH proteins with DNA sub-
`strates containing an A:GO base pair. A total of 100 fmol of 5=-Cy5–
`labeled 39-mer substrates were incubated with 15 pmol partially puri-
`fied bead-free WT or mutant MUTYH proteins at 37°C for 30 minutes in
`binding buffer. (A) DNA-protein complexes were analyzed by 6% non-
`denaturing PAGE. Lane 1, substrate only; lane 2, MutY; lane 3, WT; lane
`4, Q324H; lane 5, D222N; lane 6, Y90X; lane 7, Y165C; lane 8, R231H;
`lane 9, R260Q; lane 10, P281L; lane 11, Q377X; lane 12, G382D; lane
`13, E466X; and lane 14, 1103delC. (B) DNA-protein complexes I and II
`were quantified by ImageQuant software. Product (mean ⫾ SD) of WT
`and mutant MUTYH was determined from 3 independent experiments.
`□, complex I; , complex II.
`
`Figure 3. Time-course activity of the mutant MUTYH proteins was
`assayed to compare their glycosylase activity with WT. A total of 1000
`fmol 5=-Cy5–labeled duplex was incubated with 150 pmol WT or mu-
`tant MUTYH in 100 ␮L buffer at 37°C, 10 ␮L was removed at 0, 1, 2, 4,
`8, 16, 32, 60, 120, and 180 minutes, and immediately stopped the
`reaction by adding 10 ␮L gel loading buffer with 200 mmol/L NaOH
`followed by heating at 90°C for 30 minutes. The reaction mix was
`fractionated and visualized by a Typhoon variable imager. (A) Major
`product I was quantified by ImageQuant software. Product (mean ⫾
`SD) of WT and mutant MUTYH were determined from 3 independent
`experiments. , WT; ‘, V22M; ’, Q324H; ⽧, R26OQ; ●, G382D; □,
`Y90X; ⌬, Y165C; , R231H; 〫, P281L; Œ, Q377X; ⫻, E466X; ⫹,
`1103⌬C. (B) Rate constants (mean ⫾ SD) were determined by linear
`regression.
`
`yet, in another study, the murine mutant G365D protein
`(also corresponding to the human G382D MUTYH variant)
`was found to be fully active in removing A from an A:GO
`pair.13 Wooden et al26 characterized bacterially expressed
`GST-tag mutants (Y165C and G382D) and found them
`completely devoid of glycosylase activity. When taken
`together, these previous data and our current data
`strongly suggest that frameshift mutants are completely
`defective in enzymatic activities because of loss of the
`C-terminal domain. Our study also shows that a muta-
`tion anywhere in the catalytic domain can impair enzyme
`activity (even at a distance from residue 222). This allows
`
`ALIMENTARY TRACT
`
`BASIC–
`
`GDX 1034
`
`

`

`August 2008
`
`CHARACTERIZATION OF MUTANT MUTYH 505
`
`Figure 5. Alignment of MUTYH
`and MutY amino acid se-
`quences. Unshaded, catalytic
`domain; shaded, C-terminal
`domain; red letter, conserved
`amino acid residue.
`
`ALIMENTARYTRACT
`
`BASIC–
`
`African Americans. Therefore, on the basis of our results,
`we concluded in addition to the germ-line mutations,
`SNPs also should be studied for their possible involve-
`ment in MAP.
`The MUTYH crystal structure has not yet been solved.
`However, the observed biochemical activity of the MUTYH
`variants in this study may be explained by the study of the
`amino acid sequence of the E coli MutY protein, which is
`41% identical with MUTYH.7,26 –28 MutY has a catalytic
`domain consisting of helix-hairpin-helix, pseudo helix-
`hairpin-helix and iron-sulfur cluster [4Fe-4S], and a char-
`acteristic C-terminal domain, which is not found in other
`helix-hairpin-helix-superfamily BER proteins.31–33
`Positions of the amino acid substitution of the mu-
`tants studied in this work are shown in Figure 5. All the
`missense mutants, except G382D, would lie within this
`catalytic domain. These amino acid residues are highly
`conserved among the human, E coli, murine, and Schizo-
`saccharomyces pombe MutY homologs. Although the C-
`terminal domain of E coli MutY protein is not needed for
`its adenine removal activity, it strongly interacts with the
`GO-containing strand to flip the target adenine out of
`the helix. In this way, A is buried into a pocket formed
`between the 6-helix barrel module and the [4Fe-4S] mod-
`ule and subsequently the ␤-glycosidic bond between A
`and pentose sugar moiety is hydrolyzed by the protein’s
`glycosylase activity.31,32 The mutation in the G382D vari-
`ant lies within the C-terminal domain, which is required
`for GO recognition. In support of this theory, it has been
`shown previously that the removal of this domain from
`MutY protein drastically reduces its adenine incision
`activity from A:GO pair, but not from A:G mismatch.33
`Therefore, the inactivity of our frameshift variants can be
`explained by the fact that the Y90X mutant loses both
`catalytic domain and the C-terminal domain whereas the
`other mutants (Q377X, E466X, and 1103delC) lost only
`the catalytic domain (Figure 5). Indeed, it was previously
`
`shown that the C-terminal domain truncated MutY is
`not only defective in removing A, but also in binding
`substrate containing A:GO.33,34
`A mutation in the catalytic domain might render the
`protein partially or fully inactive whereas the loss of
`C-terminal domain could impair binding to the substrate
`containing A:GO. Protein products of SNPs are generally
`active as WT, but we observed that the SNP Q324H is
`34% less active than WT. This finding is consistent
`with the recent observation by Yuan et al,30 who re-
`ported that the Q342H variant SNP is associated
`strongly with familial CRC among the African-Ameri-
`can population. The second SNP V22M was found
`to be as active as the WT (k2 ⫽ 4.288 ⫾ 0.7831 and
`4.507 ⫾ 0.5812 min⫺1, respectively). In summary, we
`have characterized in vitro a large series of frameshift,
`missense mutants, and SNP of MUTYH in one labora-
`tory setting that are associated clinically with increased
`CRC risk. The results are summarized in Table 1. The
`2 missense variants (R260Q and G382D) were partially
`active in DNA binding and BER activities, whereas 3
`missense variants (Y165C, R231H, and P281L) and all 4
`frameshift variants (Y90X Q377X, E466X, and 1103delC)
`were dysfunctional in both activities.
`Adding further complexity is that there may be cross-
`talk between repair pathways in preventing colorectal
`carcinogenesis. In human cells, 3 BER proteins (OGG1,
`MTH1, and MUTYH) and 3 MMR proteins (MSH2,
`MSH6, and MLH1) guard genomes from mutagenic
`DNA base lesion 8-oxoG.35 Their functions have been
`well studied both in vitro and in vivo and indeed MMR
`and MUTYH proteins interact biochemically.12,13,29,36 – 40
`As such, mutations in the MUTYH catalytic domain that
`render the protein fully or partially inactive could abro-
`gate normal interactions between MUTYH and MMR
`proteins. In this study, we further confirmed that even
`mutations outside the catalytic domain can deactivate
`
`GDX 1034
`
`

`

`506 ALI ET AL
`
`GASTROENTEROLOGY Vol. 135, No. 2
`
`Table 1. Summary of Activities of WT and Variant MUTYH Proteins
`
`Mutant
`
`Y90X
`Q377X
`E466X
`1103⌬C
`R260Q
`G382D
`Y165C
`R231H
`P281L
`Q324H
`V22M
`WT
`
`Type of mutation
`
`Glycosylase activity
`
`DNA binding activity
`
`Frameshift
`Frameshift
`Frameshift
`Frameshift
`Missense
`Missense
`Missense
`Missense
`Missense
`SNP
`SNP
`
`Severely defective
`Severely defective
`Severely defective
`Severely defective
`Partial (k2 ⫽ 1.28 ⫾ 0.12 min⫺1)
`Partial (k2 ⫽ 1.033 ⫾ 0.098 min⫺1)
`Severely defective
`Severely defective
`Severely defective
`Partial (k2 ⫽ 2.97 ⫾ 0.22 min