If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Determining whether the source tissue of biological material is semen is important in confirming sexual assaults, which account for a considerable percentage of crime cases. The gold standard for confirming the presence of semen is microscopic identification of sperm cells, however, this method is labor intensive and operator-dependent. Protein-based immunologic assays, such as PSA, are highly sensitive and relatively fast, but suffer from low specificity in some situations. In addition, proteins are less stable than DNA under most environmental insults. Recently, forensic tissue identification advanced with the development of several approaches based on mRNA and miRNA for identification of various body fluids. Herein is described DNA source identifier (DSI)-semen, a DNA-based assay that determines whether the source tissue of a sample is semen based on detection of semen-specific methylation patterns in five genomic loci. The assay is comprised of a simple single tube biochemical procedure, similar to DNA profiling, followed by automatic software analysis, yielding the identification (semen/non-semen) accompanied by a statistical confidence level. Three additional internal control loci are used to ascertain the reliability of the results. The assay, which aims to replace microscopic examination, can easily be integrated by forensic laboratories and is automatable. The kit was tested on 135 samples of semen, saliva, venous blood, menstrual blood, urine, and vaginal swabs and the identification of semen vs. non-semen was correct in all cases. In order to test the assay's applicability in “real-life” situations, 33 actual casework samples from the forensic biological lab of the Israeli police were analyzed, and the results were compared with microscopic examination performed by Israeli police personnel. There was complete concordance between both analyses except for one sample, in which the assay identified semen whereas no sperm was seen in the microscope. This sample likely represents true semen because sperm cells were detected from an adjacent sample from the same garment, therefore in this case the assay appears to be more sensitive than the microscopic examination. These results demonstrate that this assay is a bona fide confirmatory test for semen.
Relationship of spermatoscopy, prostatic acid phosphatase activity and prostate-specific antigen (p30) assays with further DNA typing in forensic samples from rape cases.
Relationship of spermatoscopy, prostatic acid phosphatase activity and prostate-specific antigen (p30) assays with further DNA typing in forensic samples from rape cases.
], which is often used following a positive result in a presumptive test. In this technique, evidence is stained (e.g. with the Christmas tree method) and scanned for presence of spermatozoa, which have a distinct morphology and staining pattern. While microscopic examination can be specific, it might be difficult or sometimes impossible to identify sperm cells in cases where the evidence is in low amount and/or quality (e.g. due to degradation). It might suffer from low specificity when the sample is contaminated (e.g. by microorganisms). Microscopy also suffers from carrying a high cost because it requires skilled personnel to perform a process that can be time-consuming, and which is not amenable to automation. There are also several immunological confirmatory assays for semen detection (reviewed in [
]), which detect the presence of specific proteins found in semen and/or sperm. Among these, the most popular is the prostate-specific antigen (PSA) test, which detects the presence of the prostate-derived protein, present in semen of all men, including azoospermic [
]. In addition, protein-based assays might suffer from low sensitivity in relation to DNA-based assays because proteins are generally not as stable as DNA under most environmental conditions. Therefore, for example, a negative PSA result might be due to degradation of the protein, and not necessarily indicate absence of semen.
Alternative approaches that were developed for semen detection include mRNA/miRNA profiling [
]. In this approach, DNA extracted from forensic evidence is analyzed at specific genomic loci, which are known to be differentially methylated among different tissues, and the resulting “methylation profile” of the sample is indicative of its biological tissue source. Frumkin et al. implemented this approach by an assay based on methylation-sensitive restriction endonuclease digestion of the DNA followed by fluorescent PCR and capillary electrophoresis of the amplification products. It was used to differentiate between samples of blood, saliva, semen, and skin epidermis.
Methylation-based forensic tissue identification was also demonstrated by another group, who used methylation specific PCR of five tissue-specific loci to differentiate between several forensically relevant tissues, including semen [
Here the approach employed by Frumkin et al. has advanced and a kit for forensic semen identification was developed, which aims to replace microscopic examination of casework samples. We describe the kit's setup and demonstrate its performance on 135 samples of semen, saliva, venous blood, menstrual blood, urine, and vaginal swabs, as well as it performance on actual casework samples from the Israeli police.
2. Materials and methods
2.1 Collection of tissue samples from donors
Blood (venous/menstrual), saliva, semen, urine, and vaginal swabs were collected from volunteers. Informed consent was obtained from all participants recruited into the study.
2.2 DNA extraction and quantification
DNA was extracted from donor samples using the organic extraction protocol of Fritsch et al. [
]. Differential extraction from police casework samples was performed using the following protocol: 0.4 μg proteinase K (Qiagen, Germany) and 400 μl extraction buffer (100 mM Tris pH 8.0, 10 mM EDTA, 100 mM NaCl, and 2% SDS) were added to each evidence (piece of garment or swab) in the original tube, and incubated at 37 °C for 30 min. Following incubation, a small hole was pierced in the bottom of each tube (using a sterile syringe), and the tubes were placed on top of new microcentrifuge tubes and centrifuged at 1500 × g for 5 min. Following centrifugation, the upper tube containing the evidence was discarded and the bottom tube was centrifuged at 18,000 × g for 5 min. Following the second centrifugation the supernatant phase was transferred into a fresh tube labeled “supernatant”, and the pellet containing tube was labeled “pellet”.
DNA was extracted from both pellet and supernatant fractions of differentially extracted samples, as well as from casework samples that did not require differential extraction, using the QIAamp DNA Blood Mini kit (Qiagen, Germany), according to the kit instructions, using the “DNA purification from buccal swabs” protocol. The quantity of recovered DNA was determined using the Quantifiler® Human DNA quantification kit (Applied Biosystems) and the 7300 Real-Time PCR system (Applied Biosystems). Casework samples were profiled using the SGM+ kit (Applied Biosystems) using the same amount of DNA as used for the assay (see Table 3).
Table 3Results of analysis of real casework samples.
For each casework stain/swab, 10% of the extract (e.g. 10 μl out of 100 μl extract) was prepared for microscopic examination by hematoxylin and eosin staining [
], and manually viewed under the microscope using 250× and 500× magnification. Sperm cells were identified based on morphology. The presence of three or more sperm cells across the whole of the slide was considered positive for semen.
2.4 Selection of genetic loci and primer design
Random CpG islands (defined as a region ≥200 bp and with a GC content ≥0.5 and with at least 8 CGs; corresponding to observed/expected CpG ratio > 0.6) were searched using a software program (implemented in MATLAB, MathWorks) developed for this task. From these, CpG islands that contained a HhaI recognition sequence were selected for initial screening. Primers (18–28 bps) flanking the HhaI recognition sequence were designed with a Tm of 64–66 °C and for amplicon sizes of 80–200 bps. Fluorescently labeled forward primers (FAM) were purchased from Integrated DNA Technologies. Loci that showed significant differences between semen and non-semen samples were selected. In total, 354 genomic loci were screened. The 5 loci showing the greatest difference between semen and non-semen samples were selected.
2.5 Semen identification assay
A full description of the assay is provided in supplementary text S1. Five hundred pictograms of DNA extracted from a forensic sample undergo a single reaction (performed in a GeneAmp® PCR System 9700, Applied Biosystems) in which digestion by the methylation-sensitive restriction enzyme HhaI is immediately followed by PCR amplification of a panel of 8 genomic loci (L1–L8; Table 1). Amplified products are then separated and detected by capillary electrophoresis (CE; performed using an ABI 310 Genetic Analyzer, Applied Biosystems). The output file is analyzed by the assay software that determines whether the source of the sample is semen or not, as well as the statistical confidence level of the identification. L1 and L8 are control loci used to verify that PCR amplification was successful. L2 is a control locus used to verify that digestion was successful. The software will output identification if PCR amplification and HhaI digestion were successful, and if the methylation pattern of the semen identification loci is compatible with either semen or non-semen. If the software detects PCR failure (due to polymerase failure and/or low template concentration) it will output: “Aborted (Amplification-control too low)”. If the software detects template overload it will output: “Aborted (Amplification-control too high)”. If the software detects incomplete HhaI digestion it will output: “Aborted (Incomplete digestion)”. If the software detects a pattern which is not typical of semen or non-semen it will output: “Aborted (Inconclusive)”. Therefore, the software will output identification if all controls performed properly and the methylation pattern is typical of semen or non-semen.
Table 1Panel of loci used for the DSI-semen assay.
A scheme of the assay is shown in Fig. 1A , and the full description is provided in Section 2 and supplementary Text S1. Briefly, 8 genomic loci (L1–L8; Table 1) are amplified immediately following digestion of the template DNA. Three loci (L1, L2, and L8) are controls used to verify the success of the assay. Five loci (L3–L7) are differentially methylated between semen and non-semen and hence provide a unique pattern allowing identification. The relative peak heights of these loci correlate to their methylation levels. Typically, L3 and L4 are low or absent in semen and high in non-semen, whereas the reverse is true for L5, L6, and L7 (Fig. 1B). The software analyzes the methylation pattern and determines whether the observed methylation pattern is compatible with semen or with other body fluids.
Fig. 1Semen identification assay. (A) Schematic overview of the assay. (B) Typical semen and non-semen electropherograms following analysis by the assay software. The software determines whether the source of each analyzed sample is semen or not, as well as the statistical confidence level of the identification.
The semen identification assay was tested on 135 pure tissue samples collected from 92 volunteers (some individuals donated more than one sample; in these cases all the samples from the same individual were from different tissues—i.e. no duplicates were used). Of the 135 samples, 29 were semen (donor ages ranging from 19 to 61 years), and 106 were non-semen (30 saliva, 18 venous blood, 11 menstrual blood, 34 urine, and 13 vaginal swab). The software identified the source tissue of 129 of the 135 samples (96%) and aborted source identification in the remaining 6 samples due to PCR failure (2 samples), incomplete digestion (2 samples) and inconclusive identification (2 samples). All 129 identified sources were correct (there were no false identifications) with confidence levels typically above 0.9999 (Table 2). Fig. 2 depicts electropherograms of 10 of these samples. Although the typical methylation pattern of semen vs. non-semen is readily apparent, there is slight variability among individual samples. This variability reflects natural inter-individual variability of methylation levels. Integration of data from the five semen identification loci allows the software to overcome the “noise” stemming from inter-individual variability and achieve identification with high confidence levels.
Table 2Results of analysis of pure tissue samples collected from volunteers.
To test the assay's applicability in “real-life” situations, 33 samples from the Israeli Police Biological Lab were analyzed. These samples were from sexual assault cases comprised of stains on various garments and swabs obtained from victims (Table 3). Each sample was divided into two portions—one portion was used for microscopic examination performed by Israeli police personnel, and the other portion was used for DNA extraction and subsequent analysis. Fifteen of the 33 samples, which were suspected of containing a mixture of DNA, underwent differential extraction, and subsequent analysis was performed on both fractions. Analysis was performed by the developers in a blinded fashion (numbered tubes with no additional information were received from the Israeli Police). The results of both analyses were compared, and there was complete concordance of results, except for sample 9a (see below). Microscopic examination found sperm cells in 5 of the samples. All of these were also detected as “semen” by the assay. In all other 28 samples, microscopic examination failed to find any sperm cells. Twenty-one of these 28 samples were detected as “non-semen”, 6 had no DNA (analysis aborted due to “amplification control too low”), and in one sample, 9a, the pellet fraction was detected as “semen” and the supernatant as “non-semen”. Sample 9a was the only one for which there was a discrepancy between the results of the microscopic examination and the semen identification assay. However, microscopic examination by the Israeli police of an adjacent sample from the same garment as 9a, did reveal sperm cells, suggesting that the assay performed better than microscopic analysis for this sample. Fig. 3 depicts 6 electropherograms from 5 casework samples.
Fig. 3Electropherograms from real casework samples. The assay software analyzed the pellet fraction of sample 9a as semen, whereas microscopic examination did not find any sperm cells in this sample. Microscopic examination of an adjacent sample from the same garment as 9a, did reveal sperm cells, suggesting that this assay may be more sensitive than microscopic analysis in this case.
We also wanted to assess the assay on semen-containing vaginal swabs (the casework vaginal swabs did not contain semen). Therefore, we created mock victim swab samples consisting of internal vaginal swabs to which different quantities of semen (20 μl, 10 μl, 5 μl, 2 μl, and 1 μl) were applied. The mock samples underwent differential extraction, and subsequent analysis was performed on both fractions of each sample. In all samples the pellet fraction was detected as “semen” and the supernatant fraction was detected as “non-semen” (supplementary Fig. S2). This demonstrates the applicability of the assay for analysis of internal vaginal victim swab samples.
4. Discussion
Our results demonstrate that the assay is a useful and reliable tool for forensic semen identification. The assay is highly sensitive and in the analysis of real casework samples performed at least as good as the microscopic examination, the current gold standard for semen identification. It was also highly specific for semen and did not produce false positive results from any forensically relevant body fluid/tissue, including male urine. In contrast to microscopic examination, which is labor intensive, requires specific expertise, and produces user dependant results, the kit is a simple and straightforward assay with a setup identical to standard DNA profiling, is fully automatable, does not require special expertise, and can produce user-independent results with reduced costs. The assay works on DNA rather than directly on evidence. DNA is often extracted from evidence for identification purposes, and a small portion of this DNA can be used for semen identification. The assay therefore does not consume additional evidence beyond what is required for identification. It is important to note that the assay does not perform exactly the same function as microscopic examination. Whereas microscopic examination detects the presence of semen even if it is a minor component of the sample, this assay is designed to determine the major component of samples that are either pure or contain a clear major component. The main difference between these two methods is in mixed samples where semen is a minor component: in this case microscopic examination will detect semen, whereas this assay will output non-semen, as it is referring to the major component of the mixture.
This unique design of the assay allows it to positively link between a specific DNA profile and the source tissue (semen) corresponding to that profile. This is in contrast to microscopic examination and immunological based assays. This is so because even when microscopic examination or an immunological assay detects the presence of semen in a casework sample, that does not exclude the presence of other tissues. Therefore the DNA profile obtained from that sample might come from a different tissue source. For example, a sample obtained from a victim's clothing is found positive for semen (by any conventional assay), and the DNA profile obtained from that sample is that of a single contributor “matching” a certain suspect. In this case it is tempting to assume that the evidence is a semen stain originating from the suspect. However, it is possible that the stain is in fact mostly the suspect's saliva or other body fluid, with a minute amount of semen from an unrelated person. These are typical considerations in evaluating mixture profiles that are not homogeneous. This possible ambiguity is eliminated by using the assay. Because it works on DNA in a setup that is similar to DNA profiling, the tissue-identifying signal can be positively linked to the major contributor of the DNA profile obtained from the same sample. In the example outlined above the assay would output “non-semen” because the major contributor to the DNA sample is non-semen. Moreover, it is feasible to multiplex the semen detection assay with STR typing.
In this work 500 pg were generally used as input for the assay, except for some casework samples in which there was insufficient DNA. In these cases, lower amounts of DNA were used (down to 95 pg), and the assay was successful (outputting semen or non-semen) on as little as 148 pg. Although no drop out of loci was observed, such drop out is expected to occur in samples with extremely low DNA amounts, as occurs in DNA profiling. However, in contrast to DNA profiling, in which most peaks represent single alleles, in the semen identification assay all peaks represent two alleles (“homozygous” locus), and therefore drop out in the semen identification assay is generally expected to occur in lower amounts of DNA as compared to profiling. In addition, the assay contains an internal control for DNA quantity (loci L1 and L8) causing the software to automatically abort analysis of samples with low DNA quantities. This assay represents the first DNA methylation-based forensic tissue identification kit. Using a similar approach, additional forensic tissue identification assays can be developed (e.g. for blood, saliva, and menstrual blood). Such assays could use the same setup, reagents, and software as the present assay, requiring only a different primer mix. As Frumkin et al. previously demonstrated [
], it is also possible to use the approach for creating multiple-tissue identification assays, and/or to combine tissue identification with DNA profiling in the same reaction. Regarding the latter possibility, the loci used for identification should not contain HhaI recognition sequences, so that they would not be cleaved in the initial step of the assay. We performed an in silico assessment and found that none of the STR loci used in the UK, European, and CODIS STR marker sets (amelogenin, CSF1PO, D1S1656, D2S441, D2S1338, D3S1358, D5S818, D7S820, D8S1179, D10S1248, D12S391, D13S317, D16S539, D18S51, D19S433, D21S11, D22S1045, FGA, SE33, THO1, TPOX, and vWA) contain HhaI sites within 65 bps of their repetitive sequences. Therefore, in principle, any of these loci can be used together with tissue identification loci in combined assays. In such combined assays the size of the tissue identification amplicons should be smaller than ∼100 bps, so that they do not overlap with the identification STRs. This can easily be achieved because tissue identification loci do not contain repetitive sequences, and therefore there is little constraint in designing their primers, which only need to span the HhaI recognition sequence. Moreover, as demonstrated here, tissue identification can be achieved only one fluorescent dye (FAM in this case). The availability of at least two other colors means the expansion of this test to cover other tissue types is realistic in the future, or for multiplexing with existing STR kits.
Appendix A. Supplementary data
The following are the supplementary data to this article:
Relationship of spermatoscopy, prostatic acid phosphatase activity and prostate-specific antigen (p30) assays with further DNA typing in forensic samples from rape cases.
00193-7&id=gr1.jpg){kind=link}
00193-7&id=gr2.jpg){kind=link}
00193-7&id=gr3.jpg){kind=link}