Molecular diagnostics based on circulating biomarkers have become increasingly important in precision medicine, especially in oncology, where there is an ongoing need for minimally invasive methods of disease detection and monitoring. In this context, liquid biopsy has gained significant attention as a technique that enables analysis of tumor-related material directly from body fluids. Unlike conventional tissue biopsies, it allows repeated sampling and facilitates the monitoring of disease progression and treatment response over time. Among the accessible analytes, circulating cell-free DNA (cfDNA) is particularly relevant, as it reflects tumor-associated genetic alterations and supports various clinical applications, including early diagnosis, therapeutic decision-making, and longitudinal disease monitoring. In addition, its rapid turnover in circulation provides a dynamic snapshot of disease status, making it suitable for real-time clinical assessment. However, the analytical use of cfDNA is associated with several challenges. The concentration of cfDNA in plasma is typically low, rarely exceeding a few nanograms per milliliter, and varies by cancer type and stage. ... mehrFurthermore, cfDNA consists of short, fragmented molecules with a characteristic length of approximately 167 bp and often exhibits a nucleosomal structure. These molecules are found within a complex biological environment with a high abundance of proteins, as well as high-molecular-weight genomic DNA (gDNA) potentially released during blood processing. These factors complicate sample preparation for sequencing-based analytical methods, such as droplet digital PCR and next-generation sequencing (NGS), which require highly efficient and selective extraction techniques. Therefore, the isolation of cfDNA is considered a crucial step in the pre-analytical phase of liquid biopsy procedures. Current state-of-the-art cfDNA extraction predominantly relies on solid-phase adsorption to silica membranes or functionalized magnetic beads using strong chaotropes. Although these methods are widely adopted for their robustness and high purity, they present several drawbacks. Most kits involve time-consuming bind-wash-elute protocols with repeated manual handling and vacuum or centrifugation steps, which can increase the risk of sample loss and cross-contamination. Automated platforms can mitigate these risks and improve throughput and ease of use, but high reagent costs and the need for specialized equipment remain limitations that may restrict flexibility and scalability. Furthermore, incomplete adsorption or elution and contamination with gDNA can reduce the detection of rare tumor-derived variants. Short DNA fragments may also be recovered less efficiently, introducing a size bias that can affect downstream analysis. These issues highlight the need for alternative cfDNA extraction methods that can simplify workflows while maintaining high performance. Aqueous two-phase systems (ATPS) represent a promising liquid–liquid extraction method for biomolecule separation and purification. These systems consist of two immiscible aqueous phases formed by mixing polymers and salts above a critical concentration threshold, within which biomolecules distribute according to their physicochemical properties. This enables selective separation without the need for denaturing agents. Besides their mild operating conditions and cost-effectiveness, ATPS offer advantages in scalability and process simplicity, as phase separation can be achieved quickly by centrifugation. Although ATPS are well established for a variety of biomolecules, including proteins and nucleic acids, their application to short DNA fragments relevant to liquid biopsy has not been systematically studied. In particular, the partitioning behavior of cfDNA-like fragments in polymer–salt systems remains poorly understood, especially in the presence of plasma, a complex biological matrix that may influence phase properties. Furthermore, the integration of ATPS capture into a complete extraction workflow compatible with downstream molecular analyses has only been superficially addressed to date. Therefore, linking partitioning behavior to process design is essential for the development of an efficient ATPS-based separation strategy. Supported by relevant literature, Chapter 1 offers a comprehensive overview of the theoretical background, current research status, and remaining challenges within the field. As outlined in Chapter 2, this thesis presents the development of an ATPS-based extraction workflow for cfDNA-like fragments from plasma, integrating fundamental insights into phase partitioning with a practical, analytically compatible process. The three presented studies progress from investigations in well-defined model systems to application in a complex biological matrix and, ultimately, the creation of an integrated workflow. To enable systematic investigation, PCR-generated 160 bp DNA fragments were used as cfDNA-like surrogates in most experiments. The results provide new insights into the partitioning behavior of short DNA fragments in ATPS and demonstrate that the developed process yields sequencing-compatible DNA. The first study (Chapter 3) examined the partitioning behavior of 160 bp DNA fragments in well-defined polymer–salt ATPS, addressing the limited availability of systematic data for molecules in the size range relevant to cfDNA. Considering the complex interaction of factors affecting solute distribution, a comprehensive experimental approach was used to investigate how key system parameters, including polymer molecular weight, salt type, tie-line length, phase volume ratio, and the addition of neutral salts, influence DNA partitioning and recovery. High-throughput screening with a liquid-handling platform and automated routines for binodal and tie-line determination enabled efficient testing of many system compositions. The results revealed consistent trends in DNA distribution and identified conditions that allowed high recovery. More generally, it was demonstrated that short DNA behaves similarly to plasmid DNA in PEG/phosphate ATPS, with overall partitioning trends mainly driven by PEG molecular weight. For example, DNA fragments were primarily found in the top phase when PEG was below 600 g/mol, whereas they accumulated in the bottom phase with PEG 600 and 1000. Additionally, it was confirmed that adding a neutral salt, such as NaCl, to a PEG 400 system shifted the DNA distribution from the top to the bottom phase. Preliminary experiments in a simplified synthetic plasma matrix containing human serum albumin and immunoglobulin G showed that partitioning behavior could, to some extent, be transferred to more complex systems. These experiments also indicated that the NaCl-induced phase shift is selective for DNA, achieving 88% recovery while removing 99% of proteins. These findings improved the understanding of short DNA partitioning in ATPS and laid the groundwork for selecting optimal conditions for future extraction studies. Building on this foundation, the second study (Chapter 4) investigated the use of ATPS to capture cfDNA-like fragments from human plasma. Unlike the model systems described in Chapter 3, the presence of proteins and other plasma components introduces additional interactions that can affect phase behavior and solute partitioning. Therefore, the study assessed the recovery and distribution of these fragments, as well as phase formation, in plasma-containing PEG/phosphate ATPS. The selective NaCl-induced shift observed in PEG 400 systems proved less effective due to the formation of an interphase consisting of reversibly aggregated plasma proteins present in excess. Consequently, PEG 1000 ATPS were preferred, enabling DNA recoveries of up to 90% in the salt-rich bottom phase while reducing protein concentrations from 60–80 mg/mL in plasma to below 3 mg/mL, corresponding to over 95% removal. This study also confirmed the robustness of the ATPS capture step across plasma samples from different donors and collection tubes, both of which introduce variability. To address practical limitations caused by sample dilution from phase-forming components pre-dissolved in water, a system was designed in which solid PEG and salts were added directly to the plasma. This modification increased the plasma input from 38% (w/w) to 67% (w/w) without affecting DNA or protein partitioning. The scalability of this approach was verified by processing up to 4 mL of plasma at once, which is equivalent to the volume obtainable from a single blood tube. Additionally, reverse elution was implemented as a purification step to desalt and increase protein removal to above 99%, while reducing flow-through volume and thereby concentrating the DNA-rich extract, ultimately improving suitability for downstream sequencing-based analyses. The results confirmed that DNA can be efficiently recovered from plasma while separating it from the main protein fraction, yielding extracts with substantially reduced matrix complexity. This work established key process elements required for an integrated workflow and provides a basis for further process optimization. The third study (Chapter 5) focused on integrating individual process steps into a cohesive, optimized ATPS-based extraction workflow and evaluating its compatibility with downstream analyses. First, the effect of different lysis methods on DNA recovery and protein removal was examined. The results indicated that a lysis step might not be necessary, as the high salt content of the DNA-containing bottom phase may be sufficient to separate DNA from most proteins and possibly dissociate the nucleosomal complex, streamlining the entire process. The system composition was then optimized to increase DNA concentration while minimizing losses by adjusting the phase volume ratio and tie-line length. In addition, the potential of serial ATPS steps for further enrichment was evaluated. Although each ATPS step increased DNA concentration and allowed additional plasma to be added to the previous bottom phase, it also led to cumulative DNA losses, highlighting a trade-off between concentration and overall yield. Practical process parameters such as centrifugation time, mixing temperature, PEG form, and tube filling ratio were evaluated and carefully optimized for simplicity, speed, and robustness, revealing that proper mixing and complete dissolution of the components before phase separation are key to reproducible DNA recovery. The performance of the developed workflow was evaluated by its ability to recover a wide range of input DNA, with as little as 2 ng/mL input detectable by quantitative PCR, even in a 1000-fold-diluted extract. Fragment-size selectivity and potential PCR inhibition were tested and compared with a widely used adsorption-based extraction method. The extracts showed no signs of inhibition and exhibited a size cutoff of ∼750 bp, indicating selective enrichment of shorter fragments. This feature may benefit liquid biopsy applications by separating contaminating high-molecular-weight gDNA from the cfDNA fraction. Although the total yield of 65% sequencing-ready DNA was slightly lower than that of the reference adsorption-based kit, the ATPS workflow significantly shortened processing time to approximately 10 minutes while reducing equipment requirements and costs. Finally, the applicability of the extraction process was demonstrated by isolating cfDNA reference material from DNA-free plasma and subjecting it to NGS analysis. The successful generation of a sequencing library and the detection of the spiked mutations confirmed that the extracted DNA meets the requirements of downstream molecular assays, thereby providing proof of concept for a simplified ATPS-based cfDNA extraction approach. In summary, this thesis presents a systematic study of ATPS for extracting cfDNA-like fragments from plasma, combining mechanistic research with process development and evaluation. The results indicate that ATPS are an effective and simple platform for capturing and concentrating these fragments. Additionally, this work offers new insights into the factors governing short DNA partitioning in polymer–salt systems, particularly in the presence of a complex biological matrix. Although further studies with clinical samples are needed to fully assess its applicability in a diagnostic setting, the presented approach establishes the foundation of an alternative cfDNA extraction method. Overall, these findings highlight the potential of liquid–liquid extraction strategies to complement existing methods and support the development of more efficient and scalable sample preparation workflows for liquid biopsy applications.
1