Methylation Markers for the Identification of Body Fluids and Tissues from Forensic Trace Evidence

Abstract

The identification of body fluids is an essential tool for clarifying the course of events at a criminal site. The analytical problem is the fact that the biological material has been very often exposed to detrimental exogenous influences. Thereby, the molecular substrates used for the identification of the traces may become degraded. So far, most protocols utilize cell specific proteins or RNAs. Instead of measuring these more sensitive compounds this paper describes the application of the differential DNA-methylation. As a result of two genome wide screenings with the Illumina HumanMethylation BeadChips 27 and 450k we identified 150 candidate loci revealing differential methylation with regard to the body fluids venous blood, menstrual blood, vaginal fluid, saliva and sperm. Among them we selected 9 loci as the most promising markers. For the final determination of the methylation degree we applied the SNuPE-method. Because the degree of methylation might be modified by various endogenous and exogenous factors, we tested each marker with approximately 100 samples of each target fluid in a validation study. The stability of the detection procedure is proved in various simulated forensic surroundings according to standardized conditions. We studied the potential influence of 12 relatively common tumors on the methylation of the 9 markers. For this purpose the target fluids of 34 patients have been analysed. Only the cervix carcinoma might have an remarkable effect because impairing the signal of both vaginal markers. Using the Illumina MiSeq device we tested the potential influence of cis acting sequence variants on the methylation degree of the 9 markers in the specific body fluid DNA of 50 individuals. For 4 marker loci we observed such an influence either by sole SNPs or haplotypes. The identification of each target fluid is possible in arbitrary mixtures with the remaining four body fluids. The sensitivity of the individual body fluid tests is in the same range as for the forensic STR-analysis. It is the first forensic body fluid protocol which considers the exogenic and endogenic parameters potentially interfering with the true results.

Fig 1. Boxplot diagrams showing the discriminant power of the 9 markers.

They present the methylation rates for 20–22 samples per body fluid. Menstrual blood and vaginal fluid samples were collected by 22 healthy women (age 15 to 41 y, mean 30 y); the peripheral blood samples were taken from 9 healthy women (age 22–68 y, mean 38 y) and 11 healthy men (age 28–52 y, mean 39 y); sperm samples were given by 20 healthy men (age 20–42 y, mean 32 y); the saliva samples were collected from 12 healthy women (age 28–66 y, mean 35 y) and 8 healthy men (age 21–48 y, mean 39 y). Methylation values identified as outliers are marked with an asterisk. Only the lowest and highest outliers are shown.

Fig 2. Results of mixture experiments for both peripheral blood markers.

The specific reaction of Blut1 (left) and Blut2 (right) with blood DNA in a mixture with DNA of the remaining 4 fluids is shown. Both markers are diplex assays. 100% mixture + 0% blood (I), 80% mixture + 20% blood (II), 60% mixture + 40% blood (III), 40% mixture + 60% blood (IV), 20% mixture + 80% blood (V), 0% mixture + 100% blood (VI). Blut1 is umethylated in pure peripheral blood (green and red signal). Blut 2 is methylated in blood (blue and black signal).

Fig 3. Comparison of the different methylation quantifying methods.

Each marker was analyzed in one sample of the target fluid by bisulfite sequencing, SNaPshot reaction, Roche 454 and Illumina MiSeq NGS respectively.

Fig 4. Methylation values in tumor related body fluids.

(A) Venous blood of CLL, CML, MPS and B-cell lymphoma patients. The effect of CLL (chronic lymphocytic leukemia), CML (chronic myelocytic leukemia), MPS (myeloproliferative syndrome) and B-cell lymphoma was tested in venous blood of 13 patients. The obtained methylation values are shown as circles beside the boxplot diagrams from Fig 1. Each of the three markers relevant for venous and menstrual blood detection produced methylation signals in the range of normal controls except for one CLL and one CML sample with Blut2 in peripheral blood here revealing some overlap with the menstrual blood distribution. The remaining Blut2-values rather increased the discrimination power. Each of the 13 samples showed a zero signal with Blut1. One CLL sample showed a slightly elevated signal with Mens1.(B)Venous blood of breast cancer and colorectal carcinoma patients. The effect of breast cancer and colorectal carcinoma was tested in venous blood of 12 patients. The obtained methylation values are shown as rhombs beside the boxplot diagrams from Fig 1. All of the three markers relevant for venous and menstrual blood detection showed methylation values in the normal range except for one blood sample of a patient with breast cancer and one blood sample of a patient with colorectal carcinoma (HNPCC hereditary nonpolyposis colorectal cancer). Nevertheless, they still discriminate between the bodyfluids although overlapping with the menstrual blood distribution. Each of the 12 samples showed a zero signal with Blut1 and Mens1. (C) Vaginal fluid of endometrial cancer, cervix carcinoma, ovarian carcinoma and endometriosis patients. The influence of endometrial carcinoma, cervix carcinoma, ovarian carcinoma and endometriosis was analysed in 6 vaginal fluid samples. The obtained methylation values are shown as triangles beside the boxplot diagrams from Fig 1. Both Cervix carcinoma samples showed with both vaginal fluid specific markers methylation values at the lower part of the distribution thus overlapping with the menstrual blood distribution. The remaining values did not compromise a decision for vaginal fluid. (D) Saliva of an esophageal carcinoma patient. In one case of esophageal carcinoma we analysed the methylation with both saliva markers in saliva. The obtained methylation value is shown as a square beside the boxplot diagrams from Fig 1. The Spei1 value was lower than normal; the Spei2 value was in the higher normal range thereby maintaining the discriminant power. (E) Sperm of testicular carcinoma patients. We studied the potential impact of testicular carcinoma in the sperm of two affected patients.The obtained methylation values are shown as rectangles beside the boxplot diagrams from Fig 1. The signals of Sperm1 were slightly above the normal controls, whereas the Sperm2 values corresponded to normal controls.

Fig 5. Marker loci influenced by single nucleotide variants (SNV).

Only sequence variants with influence on the methylation of a marker locus are shown. Left vertical column: Type and position of variant. The upper horizontal line: position (number) of the CpG. The second line: normal methylation state without influence. Colours reflect the methylation degree of the CpG, increasing from yellow (0% methylated) to blue (100% methylated). CG positions marked with a star represent the analytical CGs where the primer extension takes place. Mens1: (A), Spei1: (B), Spei2: (C). Frequency of reads containing the SNV is 2,6% (A), 2,8% (B) and 14,7% (C).

Fig 6. Marker loci influenced by not converted cytosines (n.c. C).

Only sequence variants with influence on the methylation of a marker locus are shown. Left vertical column: Type and position of variant (H means haplotype of several variants). The upper horizontal line: position (number) of the CpG. The second line: normal methylation state without influence. Colours reflect the methylation degree of the CpG, increasing from yellow to blue. CG positions marked with a star represent the analytical CGs where the primer extension takes place. Mens1: (A), Vag2: (B). Frequency of reads containing the not converted cytosine is 1,8% for C157 (A) and 5,0% for C143 (B).