Friday, June 8, 2012

Detecting Cancer by MSP

MSP (methylation-specific PCR) detects hyper and hypo methylation of genes through a bisulfite modification. As a review CpG island promoters that are methylated are silenced. Cancerous cells are usually observed to have hypermethylation of genes that result in silenced tumor suppressor genes. Usually, the hypermethylation is due to an excess expression of DNA methyltranserase which are responsible for methylating the promoters by attaching a methyl group to a cytosine base. Rather than using traditional restriction enzymes to cleave distinguish methylated from unmethylated DNA for PCR, MSP (Methylation specific PCR) modifies DNA with sodium bisulfate which converts all unmethylated cytosines to uracil (8). The sodium bisulfate does not affect methylated cytosines, allowing clear distinct comparisons of methylated vs unmethylated DNA.

By converting the cytosine base, we can distinguish the methylated and the unmethylated by PCR. The conversion of unmethlyated DNA base cytosines to uracil are then amplified and copied through PCR. The DNA is then ran on gel. On the gel, one would distinguish methylated genes by the presence of cytosine bases and unmethylated genes by the presence of thymidine (after PCR, uracil is read as thymidine). Since there are only specific CpG island promoters in the genome, one doesn’t have to worry if the comparison of thymidine will be invalid with unrelated thymidine bases. Another benefit of using MSP is that you don’t need a lot of DNA sample. As less as 1µg of DNA would work for MSP (8). By testing the DNA sample directly such as stool testing, biomarkers are not introduced to the body through invasive means (penetrating or entering the body). This also means that the body is not given a chance to adapt to the biomarkers that may interfere with testing such as producing false positives after a number of certain trials. However with stool testing, there are limitations such as the types of cancer. Stool samples only apply to cancers that are involved in the digestive system. Brain cancer MSP testing must involve other methods to obtain DNA samples.

bisulfite final.jpg

Traditionally, researchers would distinguish hypermethylation through Southern hybridization approaches. Southern hybridization uses restriction enzymes to cut the DNA and the sample would be ran on a gel. However, the restriction enzyme must be recognized as a methylation sensitive enzyme and is limited to the known methylated gene.


With MSP, all CpG island promoters are examined, not just within regions where the restriction enzymes recognize in the DNA (8). Besides the utilization of restriction enzymes (which is not 100% effective), it requires large amount of DNA (5 ug or more) (8). Since PCR with restriction enzymes are not perfect, the uncleaved DNA will turn up as false negatives for methylation. An example is when the DNA is really methylated but since it was not cut, we would have a false negative. MSP does use restriction enzymes however. The difference is the conversion of all unmethylated cytosine to uracil. There are special primers that will not recognize "unmodified" DNA, so there are no false positives for methylation.

MSP allows for specific probing for a certain methylated gene that is usually abundant in the cancerous cells. The process does not rely on restriction enzymes. The methyl group on methylated DNA prevents the conversion of cytosine to uracil. After conversion of uracil, DNA is amplified through PCR (in this process, uracil is read as thymidine in DNA). An experiment by Herman et al. (1996) tests how effective MSP process was in distinguishing methylated DNA and MSP analysis for a tumor suppressor gene p16 (8). Researchers obtained DNA sample from renal carcinoma cell line and normal tissue as a control. Bisulfite Modification was introduced to DNA which was later re-suspended in water. For means of comparing accuracy, genomic sequencing of both renal carcinoma DNA and normal tissue DNA was performed after the bisulfite modification. The absence of cytosine represents unmethylated cytosines in CpG islands in the DNA. Figure 3 is the genomic sequencing of p16. On the gel, H157 has cytosine bases, indicating the presence of methylated DNA, as expected from the renal carcinoma DNA. H249 is the normal tissue control and shows no methylation on DNA due to the absence of cytosine bases (converted into uracil and read as thymidine). Instead, it is observed to have more thymidine bases than H157. The genomic sequencing allows for the comparison of accuracy of MSP now that expected methylation is obtained. The brackets on the gel indicate a BstUI site where the enzyme will cut for p16 (specific gene probing). The DNA samples were amplified by PCR, were cut by BstUI restriction enzyme and ran on a gel.


For simplicity, we will concentrate on Figure 6a and d. Figure 6a shows MSP results of H157 (renal carcinoma), untreated DNA (without bisulfite modification, normal lymphocytes and H249 (normal tissue control). The DNA in 4a were treated with bisulfite modification and were amplified with PCR. The DNA was not cut with BstUI. Due to modifications of the DNA bases, there are two primers used: methylated & unmethylated specific primers and primers specific to wildtype p16 sequence. The methylated & unmethylated specific primers will only interact with modified DNA (DNA reaction in the presence of bisulfite, even if the DNA is not modified to a uracil). The specific primers for p16 sequence will interact with unmodified DNA (DNA with no bisulfite modification). The primers are selective and will only interact with their specific DNA sequences conditions. The reason the experiment had both methylated and unmethylated specific primers is due to both hyper and hypo methylations in DNA found in cancerous cells. U1752 is lung cancer cell line methylated control; it was a known hypermethylated gene of p16. We see two samples that indicate methylation (under M). H157 is the experimental renal carcinoma, from genomic sequencing we know that it is methylated. We see two bands that indicate methylation. There is also a weak amplification under wild type (under W), indicating incomplete bisulfite reaction. However, this is not a false positive for methylation because the DNA in H157 that has not been modified will not be recognized by methylated specific primers and will not interfere with methylation sensitivity of MSP. Untreated DNA is DNA without bisulfite modification to test if MSP is methylation sensitive by interaction with the modified DNA (to assure no false negatives of methylation by incomplete reaction of bisulfite). We see one band under wild type, indicating that MSP is methylation sensitive. H209 is a unmethylated control (a case of hypomethylated cancer cell line), where we see two unmethylated bands (under U). Lastly, H249 is out control tissue which yield similar results as normal lymphocytes, indicating p16 normal methylation sequence (8).

Figure 6d is the bisulfite modification with restriction enzyme BstUI digest. The purpose of 6d is to indicate that the specific restriction enzyme is only sensitive to methylated promoters. The (+) indicates the presence of BstUI enzyme, while the (-) indicates the absence. From H249 sample which is unmethylated, the presence of the enzyme does not show any differences. However, for H157 which was established to by hypermethylated, the presence of the enzyme is depicted through the result of two small bands. The two bands are the product of the recognition of the BstUI site CGCG. Recall that if the DNA is unmethylated, then the site would be converted into TGTG after PCR. BstUI will not longer recognize the site and the DNA will not be cleaved into two. As for the primary lung cancer sample, there is no difference under unmethylated regions but under methylated regions, two bands are observed. With the restriction enzymes that are sensitive to methylation, CpG islands that are embedded in DNA can also be detected after digest. A final note is MSP does not need genomic sequencing. The only reason the researchers who test this method had done genomic sequencing was to use as a control to compare MSP results. One benefit of MSP is that it does not use radioactivity and its not expensive as genomic sequencing (8).

Utilizing MSP
In regards to detecting DNA methylation through noninvasive means, DNA methylation patterns can aid in diagnosis of cancer in external samples such as stool. An experiment done by Kisel and colleagues assessed DNA methylation marker candidates that will aid in diagnosis of pancreatic cancer (12) . The target genes are EYA4, MDF1, UCHL1 and BMP3. Stool and tissue sample was obtained from patients with pancreatic cancer. These samples were run through a gel by MSP and compared with normal controls.

In the figure, EYA4 had a higher concentration of methylation in pancreatic cancer stool sample than the control. However, due to a lot of noise, there was no significant difference to indicate whether the sample is cancerous. For MDF1, there is no significant difference. In UCHL1, there is a hypomethylation of the target gene in the pancreatic stool sample. But this target gene could not be used as a biomarker because the amount of methylation did not match the tissue sample. The tissue sample (from the source) is more accurate than the stool sample in terms of the presence of these target genes. For BMP3, there is a significant difference in hypermethylation in pancreatic stool sample than the control. The tissue sample also shows a significant difference.

Stool markers.jpg

BMP3 is bone morphogenic protein 3 that plays a role in tumor suppressor genes. From this experiment, BMP3 is hypermethylated in pancreatic cancer samples (tissue and stool). By targeting this gene alone for hypermethylation, 51% of pancreatic cancer patients were detected. When used in combination of target mutant KRAS (oncoprotein that is hypomethylated), gives a more accurate diagnosis of the pancreatic cancer. Hypomethylation of KRAS detected 50% of pancreatic cancer patients. Together, these biomarkers detected 67% of pancreatic cancer patients. The take home message is that these biomarkers that detect hypo or hyper methylation can aid in diagnosis in combination of other means of detection such as clinical scans. 

DNA methylation is relatively stable and will not degrade in or outside the body. Since the DNA methyl groups are already in place, there is no need for an introduction of a fluorescing molecule that will need to bind to certain proteins. Most biomarkers would be introduced to the body and will need to bind to certain target locations. Over time, the body may either adapt to these biomarkers through immune response or these biomarkers may degrade quickly, resulting in false positives and negatives. DNA methyl groups can be sensitively detected by Methylation-Specific PCR (MSP) with no false positives for methylation due to the bisulfite modification and special primers. With MSP, we can also detect DNA methylation outside the body through the means of noninvasive sampling such as stool. Monitoring the increase of hypermethylation or hypomethylation of certain target genes allow predicitive prognosis of the tumor and how the patient is responding to chemotherapy treatment.

There are some concerns using DNA methylation as biomarkers. For diagnosis of cancer, certain DNA abnormal methylation patterns must be known and identified in both normal and cancerous cells. In stool samples, the combination of the two target genes BMP3 and KRAS only allows 67% of correct diagnosis of pancreatic cancer samples. This is due to pancreatic cancer arising from many different types of mutations and causes. It is the same as treating a breast cancer patient with a drug to inhibit a specific gene, when the patient's cancer is not associated with that specific gene. If we use BMP3 and KRAS as target methylated genes alone, we may be wasting our time and resources. More research about specific DNA methylation patterns is needed. However, one should not throw the baby out with the bath water. Instead, we should look for more methylation markers that associate with the type of cancer and stool sampling can be used as a "pretest" for pancreatic cancer before any imaging or invasive endoscopy is done on the patient.


Herman, J.G., Graff, J.R., Myohanen, S., Nelkin, B.D. & Baylin, S.B (1996) “Methylation-specific PCR: A novel PCR assay for methylation status of CpG islands.” Proc. Natl, Acad. Sci. USA, 93:9821-9826.

Kisiel, J.B., Yab, T.C., Taylor, W.R., Chari, S.T., Petersen, G.M., Mahoney, D.W., & Ahlquist, D.A. “Stool DNA Testing for the Detection of Pancreatic Cancer.” Cancer. (2011) DOI: 10.1002/cncr.26558.