Potential of the PowerSeq® 46GY System to Investigate Low Template Samples

Introduction

The increasing interest in massive parallel sequencing (MPS) in the forensic community called for the development of commercially available kits. With the PowerSeq® 46GY System, Promega offers a multiplex assay for the simultaneous amplification of 22 autosomal short tandem repeats (STR), 23 Y-STRs and amelogenin on the Illumina MiSeq™ System. There are several advantages of MPS methods over historically used capillary electrophoresis (CE) systems, including the ability to analyse more markers and the inclusion of sequence polymorphisms for better deconvolution of mixtures. The possibility to decrease the size of all amplicons ought to be of particularly great benefit for the analysis of degraded and low template samples. These sample types are commonly encountered during forensic casework in the form of bones or micro traces such as shed skin cells (“touch DNA”) and telogen hair roots and shafts. During this study, suitability of the PowerSeq® 46GY system for routine casework samples was tested. Particularly, low molecular traces, artificial mixtures as well as casework samples were examined and compared to CE results. In recognition of the need to lower expenses and minimize consumption of valuable sample extract, the practicality of a reduced reaction volume was tested. Furthermore, there is now a wide variety of library preparation kits available. In addition to the recommended reagents, the application of an alternative workflow was evaluated.

Materials and Methods

As mentioned before, the PowerSeq® 46GY System amplifies 46 targets in a single reaction. Fragment sizes between 140-300 base pairs are amplified in a multiplex PCR reaction in less than 60 minutes. Libraries are then prepared with the Illumina TruSeq® DNA PCR-free Library Preparation Kit.

Reference and mock samples were obtained from four individuals (two male, two female) with informed consent. The donors provided oral swabs and telogen hair samples. Mock touch DNA samples were created by handling a 15 ml tube followed by swabbing with a pre-moistened micro swab. Artificial mixtures were prepared from saliva DNA extracts with two male – female pairings. Ratios were 1:10 and 1:20 with a female major contributor. In addition, three casework samples from sexual assault cases as well as a bone sample were included in the study. DNA from saliva, hair, bone and casework samples was extracted with the Maxwell® FSC DNA IQ™ Casework Kit (Promega); the bone sample was digested with the Bone DNA Extraction Kit (Promega) first. Touch DNA samples were directly lysed with the Investigator Casework GO! Kit (Qiagen).

The PowerPlex® ESX 17 Fast System (Promega) and Investigator ESSplex SE QS Kit (Qiagen) were used for autosomal STR analysis, the PowerPlex® Y23 System (Promega) for Y-STR analysis, where applicable.

All samples were amplified with the PowerSeq® 46 GY system. Two of each sample type were selected for amplification in a halved reaction mix. Volumes were reduced for both the initial PowerSeq® amplification and the library preparation workflow.

As an alternative approach to the Illumina library preparation, the same small set of samples (except mixtures) was primed for MPS using the lower-priced GeneRead DNA Library I Core Kit (Qiagen). Adapters are also added through ligation, but this workflow requires one washing step less, which reduces hands-on time (see Figure 1).

Figure 1: Depiction of the difference between and changes to the two library preparation workflows. The Illumina TruSeq® workflow requires an additional washing step after the end-repair reaction. Furthermore, library amplification was necessary for Illumina samples due to low yields during quality control. Libraries were then loaded on a V2 rather than a V3 flow cell.

Final libraries were quantified with the NEBNext® Library Quantification Kit for Illumina (New England Biolabs) following manufacturer’s recommendations. Due to low concentrations in all TruSeq libraries, optional library amplification was performed using the GeneRead DNA I Amp Kit (Qiagen).

Samples prepared with the Illumina TruSeq® workflow were loaded on a V2 flow cell, whereas samples from the Qiagen workflow were loaded on a V2 micro flow cell. V2 chemistry is available at lower cost and in smaller sample sizes than V3 chemistry.

The web-based tool toaSTR [1] was used for data analysis. Analytical threshold was set to 10 reads, calling thresholds varied among sample types (reference: 10%; skin: 10%; hair: 10%; casework: 2%; mixtures: 1%). Only eight Y-STR markers (DYS19, DYS391, DYS392, DYS393, DYS438, DYS439, DYS448, DYS456) implemented in toaSTR at the time of analysis were included in this study.

Results

Of the 24 corresponding loci, all reference samples showed concordant results between CE and MPS. However, in accordance to the larger number of targets in the MPS panel, the information obtained from a single analysis is greater. Female samples yielded higher mean read counts than male samples due to absent Y-STR markers.

Analysis of standard and half reaction volume showed comparable performances. Some of the low molecular samples from the half volume set showed 1-2 alleles less, which might be due to stochastic effects within the small sample set. Likewise, there was no trend towards enhanced read coverage in either of the sample sets (Figure 2). Based on these findings, the Qiagen workflow was also conducted in half reaction volume.

Figure 2: Comparison of the Illumina TruSeq® workflow with standard or halved reaction volume (n=2 per sample type/volume). (A) Relation of allele numbers between reaction volumes (standard reaction volume was assumed to show 100% of alleles). Absence of 1-2 alleles in the reduced reaction volume could be owed to stochastic effects. (B) Average read count per sample type. A tendency towards either of the reaction volumes cannot be deduced from this dataset.

Comparison of the Illumina and Qiagen library preparation workflow also showed no notable tendency toward either of the methods regarding number of detected alleles. However, there was a contrast in terms of read coverage (Figure 3). Most likely, the different loading capacities of the two separate flow cells caused this difference. Nonetheless, all samples reached the minimum of 4500 reads per marker as calculated from the number of samples recommended for loading on a V3 flow cell.

It seemed like library amplification of Illumina samples introduced a great number of artefacts with sequential errors. Even though fragment lengths were identical to regular alleles, each of these minimally different sequences was called as individual artefact. Their impact on profile interpretation should be considered before integrating library amplification into a workflow.

Figure 3: Comparison of the two library preparation workflows (n=2 per sample type/volume). (A) Number of alleles obtained was similar for both methods (Illumina standard reaction volume was assumed to show 100% of alleles). (B) Samples prepared with the Illumina workflow showed higher read counts throughout all sample types. Loading on different flow cells might have caused this difference.

Comparison of informative value for routine casework involved three samples from actual sexual assault cases (tapings from clothing). Figure 4 shows the number of typed alleles belonging to either victim, offender or partner of the victim as well as alleles that could not be assigned to a known individual. Additional alleles provided only by MPS analysis are stacked on top of MPS bars. As expected, all reference alleles were in accordance with CE results. An increase of unknown alleles through MPS was only observed in casework sample 2 (within corresponding CE and MPS markers). However, owing to the greater marker set, additional information was gained for all samples with the single MPS run.

Figure 4: Comparison of informational value from three casework samples with MPS (1 run) and CE (2 runs). Complete profiles of persons of interest (victim, suspect or partner) were detected within the mixtures. MPS analysis detected an identical or higher number of unknown alleles and added information to all samples due to the larger panel.

To our contentment, MPS analysis of a bone sample highly increased evidential value. Applying our regular casework analysis, three separate PCR and CE runs were necessary to obtain a full composite DNA-profile. Drop-in and drop-out events as well as unbalanced peaks complicated profile interpretation. In contrast, the profile from the single MPS run not only showed few stochastic effects and matched the derived CE profile but also provided valuable information from additional markers (Figure 5).

Figure 5: Comparison of detected alleles from the bone sample with MPS (1 run) and CE (3 runs). MPS loci were less affected by artefacts. Successful amplification of further markers in the larger MPS panel provided additional information.

Conclusion

Despite the rather small set of samples that limit statistical analyses, some useful insights were gained through this study. On the technical side, the reaction volume of the PowerSeq® 46GY System can be halved to save valuable sample and costs per analysis. Furthermore, the Qiagen GeneRead workflow produces comparable results to the Illumina TruSeq® workflow. In combination with the V2 instead of the recommended V3 flow cell expanses may be further decreased while maintaining satisfactory STR-typing results. However, the V3 chemistry ought to enhance sequencing quality and allows higher throughput.

Overall, the PowerSeq® 46GY System is suitable for sensitive casework samples such as micro traces and bone whilst increasing informational value from a single analysis compared to traditional CE analysis.