Optimization of mRNA Thermal Release on an Integrated Microfluidic Platform by a Complex System Response Method

Optimization of mRNA Thermal Release on an Integrated Microfluidic Platform by a Complex System Response Method | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Optimization of mRNA Thermal Release on an Integrated Microfluidic Platform by a Complex System Response Method Swati T. Gurme, Yi-Cheng Tsai, Bhushan Koparde, Da-Jeng Yao, Lily Hui-Ching Wang, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9465529/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 9 You are reading this latest preprint version Abstract Non-poly(A) messenger ribonucleic acid (mRNA) purification remains a critical bottleneck for cell-free translation platforms like tr anscription-translation coupled with the a ssociation of p uromycin linker (TRAP) display for peptide screening, where sub-optimal thermal release can compromise results. Herein, we demonstrate a “complex system response” approach for optimizing thermal elution from probe-coated magnetic beads. Using an orthogonal array composite design with two optimization rounds (17 three-factor + 9 two-factor experiments), we systematically explored temperature (60–95°C), time (5–30 min), pH (7–9), and their interactions, revealing an unanticipated optimum of 60°C for 23.5 min at pH 9 (vs. conventional protocols of 90°C for 10 min). These conditions were associated with a maximum release efficiency of 71 ± 7% with 10 pM RNA, sufficient for downstream TRAP library construction. complex system response microfluidics non-poly(A) mRNA extraction qRT-PCR thermal release optimization Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Messenger ribonucleic acid (mRNAs) link genomic information to protein synthesis, and their quantification has advanced a variety of applications ranging from gene‑expression profiling, both basic science (He et al. 2018 ) and diagnostics (Bustin 2002 ). Real‑time reverse transcription polymerase chain reaction (qRT‑PCR) is the primary method for quantifying mRNA because it is sensitive, specific, and high‑throughput (Han et al. 2014 ). Still, its accuracy depends on high‑quality, undegraded, and DNA-free mRNAs. Accordingly, robust mRNA isolation is a crucial first step in workflows ranging from cDNA library construction (Jiang and Harrison 2000 ) to “sample-to-answer” assays (Bustin 2002 , Nestorova et al. 2017 ). Conventional RNA extraction typically combines chaotropic lysis, organic extraction, and alcohol precipitation, followed by enrichment of polyadenylated transcripts (Nestorova et al. 2017 ); however, these procedures are laborious, use hazardous reagents (such as phenol), and are prone to RNA loss/degradation (Sarkar and Irudayaraj 2008 , Nestorova et al. 2017 ). Alternatively, solid-phase capture methods based on silica adsorption or oligo(dT)-mediated hybridization on magnetic beads (MBs) enable faster binding-wash-elution cycles and avoid organic solvents. The latter offer high surface area and straightforward magnetic manipulation, and carboxyl-coated magnetic nanoparticles functionalized with oligo(dT) 25 can provide µg-scale mRNA yields at reduced costs (Sarkar & Irudayaraj 2008 ). However, these methods require open-tube handling and multiple centrifugation or magnetization steps that introduce variability and ribonucleases (RNase) exposure. Flow-through porous polymer monoliths formed in capillaries or channels and functionalized with oligo(dT) or locked nucleic acid (LNA) substitutes offer high surface areas and purify mRNAs from total RNAs with yields and purities comparable to commercial column kits in < 20 µL. Satterfield et al. ( 2007 ) reported UV-polymerized oligo(dT) methacrylate monoliths in capillaries that achieved near-maximal binding in seconds, capturing ≥ 16 µg of mRNA from 315 µg of total RNA in 0.4 µL, with 15-fold rRNA enrichment in NaCl solution (up to 110-fold when using LNA probes or tetramethylammonium chloride). Solid-phase gene extraction further simplifies sampling by using dT(15)-modified steel or glass needles that briefly contact tissues or cells to hybridize poly(A) transcripts, which can then be released directly into tubes or chips. This eliminates the need for bulk lysis or centrifugation. Nesterova et al. (2017) used this approach to extract 100–300 pg of mRNA from glioblastoma spheroids. While oligo(dT)-based capture is highly effective, it inherently excludes non-poly(A) mRNAs, many non-coding RNAs, and bacterial or organellar transcripts that lack canonical poly(A) tails. For prokaryotic and mitochondrial transcriptomics (Filiatrault 2011 , Park et al. 2020 ) or in vitro display technologies that operate on synthetic mRNAs (Roberts & Szostak 1997 ), alternative strategies are required. These can include total RNA capture (Liu et al. 2009 ), sequence-specific hybridization (Adams et al. 2015 ), or ribosome-associated purification (Zeng et al. 2023 ). Classical mRNA display (Newton et al. 2020 ) platforms and newer variants like tr anscription-translation coupled with a ssociation of p uromycin linker (TRAP; Ishizawa et al. 2013 ) and click display (Zeng et al. 2024) covalently link polypeptides to their encoding mRNA or cDNA via puromycin-containing DNA linkers and rely on sequence-defined transcripts generated in vitro rather than on poly(A) tails; these methods enable rapid selection from 10 12 -10 13 peptides but are sensitive to RNases and involve complex chemistries. In addition, lack of reliable sequence information, as well as the complexity of reaction integration and handling, limit throughput, reproducibility, and performance. Recently, microfluidic lab-on-a-chip platforms can integrate these capture/release strategies with downstream enzymatic reactions while minimizing sample loss and RNase contamination (Han et al. 2014 ). Current devices can conduct lysis (Eastburn et al. 2013 ), mRNA capture (Gurme et al. 2026 ), cDNA synthesis, and amplification in closed microfluidic environments, enabling < µL assays and even single-cell analyses (Toriello et al. 2008 ). An example is the Y-channel glass microchip (Jiang & Harrison 2000 ), which used paramagnetic oligo(dT) 25 beads to capture poly(A) mRNA from total RNA under laminar flow. Beads and RNAs were introduced through separate inlets, enabling on-chip hybridization, followed by magnetic trapping of bead-mRNA complexes. The captured mRNA was subsequently washed off-chip for analysis by capillary gel electrophoresis or RT-PCR, preserving integrity of both abundant and low-copy transcripts. The system captured 3–34 ng of mRNA (26%) from 1–10 µg of total RNA, though off-chip pooling was required. Similarly, a PDMS-glass RT-PCR chip with permalloy wires was designed for lateral deflection of oligo(dT) bead-mRNA complexes from lysed samples, enabling solid-phase RT from 0.1 µL of blood, SKBR3 cells, or viral swabs (Hans 2014). These examples demonstrate potential, but limitations still persist. Furthermore, non-poly(A) mRNAs demand sequence-specific probes incompatible with oligo(dT), highlighting the need for fully automated, high-capacity platforms supporting diverse transcripts and downstream workflows. Achieving such performance requires precise control over multiple parameters- temperatures, buffers, flow rates, bead loading, and release conditions- that conventional trial-and-error optimization struggles to resolve efficiently. Statistical optimization frameworks are increasingly important for fine-tuning these multivariate workflows. “Complex system response” (CSR) approaches treat the device as a “black box” and fit low-order polynomial response surfaces that link a small number of carefully chosen experimental conditions to quantitative readouts (Wong et al. 2008 , Kaladharan et al. 2024 , Tsai et al. 2025 ). CSR has been used with orthogonal array composite designs to optimize 10-drug cocktails and dose levels from 155 combinations in a 3-level search space, while in clinical settings, phenotypic response surface-based personalized medicine has reduced hospital stays for liver transplant recipients (Khong et al. 2025 ). In another case, a feedback-controlled platform was reported to navigate the vast combinatorial space of angiostatic drugs, identifying a potent four-drug synergy that effectively inhibits vessel growth at highly reduced doses (Weiss et al. 2015 ). We previously demonstrated that sequence-specific probe-coated MBs capture in vitro- transcribed (IVT), non-poly(A) mRNAs with 93% efficiency on-chip, but only 48% of bound transcripts are thermally released at 95°C (Gurme et al. 2026 ). Because IVT mRNAs used for TRAP display lack poly(A) tails, conventional oligo(dT) solid-phase extraction is not applicable, and low release efficiency can thwart the ability to generate high-diversity peptide-mRNA libraries. Herein, we adopted CSR to identify the combination of three reaction parameters-release temperature, buffer pH, and incubation time- that would lead to maximum mRNA release following a sequence-specific probe-coated MB protocol; this data-driven approach was hypothesized to preserve specificity at a significantly reduced workflow. 2. Materials and methods 2.1 Library preparation and probe design We employed the same library preparation and probe design protocols as in Gurme et al. (2026). A random dsDNA library was constructed for affinity peptide screening of kinesin family member 2C (KIF2C), which contributes to tumor growth and metastasis (Wang et al. 2012, Zuo et al. 2023). The 115-bp sequence (Genomics Biosci. & Tech, Taiwan) contained a T7 promoter, Shine-Dalgarno sequence, peptide-coding region (1 Met, 10 NNK [N-A/T/C/G & K-G/T]), random amino acids, 5 Gly spacer codons, 1 Ser (Kawakami et al. 2015), and an An21 sequence at the 3′ end (underlined) at a concentration of 386 ng μL -1 : 5′-ATACTAATACGACTCACTATAGGATTAAGGAGGTGATATTTATGNNKNNKNNKNNKNNKNNKNNKNNKG GTGGAGGAGGAGGTAGC TAGGACGGGGGGCGGGAGGCGGG -3′ 2.2 Probe synthesis and MB conjugation Complementary probes (5′‑amino C12 linker: 5′‑CATAAATATCACCTCCTTAA‑3′; 100 μM) were synthesized (Genomics Biosci. & Tech, Taiwan) and covalently bound to carboxylic acid MBs for mRNA capture from IVT reactions, following the established 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide/N-hydroxysuccinimide (EDC/ NHS) coupling protocol (Wang et al. 2024). Dynabeads™ MyOne™ carboxylic acid beads (2.8 µm in diameter, 2×10 9 beads/mL, Thermo-Fisher Scientific [TFS]) were vortexed (Vortex-Genie 2, Scientific Industries, USA) for 5 s in 100 µL, diluted in 900 µL of ddH 2 O, magnetically separated (DynaMag™-2 magnet), and washed twice. Beads were then incubated for 18 hr at room temperature in the dark (Intelli-mixer RM-26, ELMI, 20 RPM, C2 mode) with 950 µL of ddH 2 O containing 30 µL of 100 µM amino-modified probe and 20 µL of 120 mg mL -1 EDC (TFS). After incubation, beads underwent sequential washes with 0.02% Tween 20 (Merck, Germany, 2×), 0.1% SDS (Merck, 2×), and 0.1 M ethanolamine (Merck) blocking (1 hr, dark). The final probe-functionalized beads were resuspended in 1 mL of ddH 2 O and stored at 4°C. 2.3 Design, microfabrication, and operation of the microfluidic device An integrated microfluidic chip (46 × 53 mm; air-control/PDMS liquid-channel layers bonded to glass) microfabricated from PMMA molds with PDMS replication and O 2 plasma bonding adopted from Gurme et al. (2026) was used for mRNA capture and release experiments. Briefly, the device integrated 16 air inlets, 3 micromixers, 2 suction micropumps (D/E in Fig. 1: R=3 mm; H: R=1.5 mm), 10 microvalves (2.8 × 2.7 mm), and 6 chambers (A-J in Fig. 1, diameter: 2-7.6 mm), with three thermoelectric coolers (TECs; 9.5 W, 15×15×3.2 mm, 739-387004942, Mouser, Taiwan) for IVT/DNase (37°C) and reverse transcription (RT (42°C). The third was for thermal release at varying temperatures under Arduino feedback control (MAX6675, Maxim Integrated Products, USA). Samples/reagents were first loaded into chambers; pneumatic actuation by a compressor (DRS-210-22, SWAN, Taiwan) or vacuum pump (DF-506 K, Doctor's Friend, Taiwan) activated flow/mixing via an electromagnetic valve (EMV) module while TECs were operated at 6 V/3 A. DNase-treated IVT mixture (20 µL, 1 U µL -1 ) was hybridized on-chip with probe-coated MBs (20 µL, 10 8 beads/mL) in micromixer E. Complexes were washed twice with 25 µL of binding/washing buffer (15 mM sodium citrate, 150 mM NaCl, pH 6.8-7.0; F→D→E→W; Fig. 1). For the release buffer, we used 10 µL of 10 mM Tris-HCl (pH 7, 8, or 9; chamber G), splitting unbound mRNA into puromycin annealing (I) and RT (J) chambers. 2.4 On-chip thermal release We followed the same IVT, DNase treatment, hybridization, washing, and qRT‑PCR workflow as in Gurme et al. (2026), with modifications only to the thermal release step. Briefly, IVT was performed at 37°C for 60 min using the T7 RiboMAX™ Express RNAi system (Promega, USA), followed by DNase treatment (RQ1, 1 U µL -1 , 37°C, 15 min) (Fig. 2). The DNase‑treated IVT mixture (10 pM mRNA in all experiments) was hybridised on-chip with 20 µL of probe‑coated MBs (10 8 beads mL -1 ) for 15 min at room temperature. After magnetic separation, the supernatant was retained for unbound mRNA assessment. MB-mRNA complexes were washed twice (25 µL) with binding/washing buffer, and mRNA was eluted in 10 µL of 10 mM Tris‑HCl at different temperatures (according to CSR). TECs were used for precise temperature control. The eluted mRNA (8 µL) was annealed with puromycin linker (1.4 µL, 100 µM; 25°C for 30 min). The remaining 2 µL were used for qRT-PCR with 10 µL of KAPA SYBR FAST (Sigma-Aldrich), 7.15 µL ddH 2 O, 0.40 µL RT mix (Sigma-Aldrich), 0.20 µL of forward primer (T7SD8M2.F44, 10 µM, 5’-ATACTAATACGACTCACTATAGGATTAAGGAGGTGATATTTATG-3’), 0.25 µL of reverse primer (G5S-4.R20, 10 µM, 5’-CCACCTCCTCCTCCATCGAT-3’; Protech, Taiwan) on a StepOnePlus™ (Applied Biosystems). Positive (total mRNA), uncoated beads, and ddH 2 O negative controls were included. qRT-PCR cycling began with RT at 50°C for 5 min, then 42°C for 20 min. After a three-min hot-start at 95°C, 35 cycles of 95°C for 15 s (denaturation) and 65°C for 20 s (annealing/extension) were performed. 2.5 CSR optimization for thermal release of mRNA We used CSR‑based OACD to systematically optimize thermal release of mRNAs (Fig. 3). In the first run, three factors were allowed to vary over narrow ranges-release temperature (85-95°C), incubation time (5-15 min), and release buffer (10 mM of Tris‑HCl) pH (7-9)- using 10 pM of input mRNA and citrate/NaCl as the capture buffer. The design generated 17 conditions to test, far fewer than a full factorial; however, no result was satisfactory for subsequent analyses. We therefore broadened the dynamic range for a second optimization round, expanding temperature from 60 to 95°C and release time from 5 to 30 min, with a constant release buffer pH of 9 (based on the first OACD trends) and keeping all other conditions unchanged. The second OACD matrix generated a total of nine experimental conditions, which were subsequently tested on-chip. 3. Result and discussion 3.1 Temperature control TEC modules delivered precise, stable control across 60 to 95°C. Target temperatures of 60, 77, 85, 90, and 95°C were achieved within <1 min (& within 1°C) with minimal overshoot (<1°C) and long-term stability (<0.5°C standard deviation (std. dev.); Fig. 4a). This performance aligns with the <0.1-0.5°C precision demanded for lysis, RT, and elution (Ji et al. 2024). No recalibration was needed, confirming the robustness of temperature control. 3.2 Microfluidic component performance The large micropumps (D/E) reliably dispensed 25 µL (≤35 kPa negative/20 kPa positive gauge; Fig. 4b). Small micropumps (H) dispended 2 µL at 20 kPa, and micromixers achieved >90% homogenization in 14 s (20-35 kPa); complete mixing occurred in <16 s at 2 Hz, ideal for limiting shear damage. These metrics match PDMS micropump benchmarks (10-50 µL min -1 at 20-60 kPa; <5% CV), where pressure-dependent stroke volumes and minimal leakage (<1%) are recommended to enable precise, unidirectional reagent delivery (Ni et al. 2010, 2012). The planar PDMS design is attractive because it simplifies microfabrication while still allowing continuous pumping under pneumatic actuation and demonstrating robust, low-leakage, low-evaporation performance. 3.3 qRT-PCR dilution curve Please see Gurme et al. (2026) for details. Briefly, R 2 was 0.995, and no amplification was seen in negative controls. 3.4 CSR-based thermal release optimization 3.4.1. 17-experiment design Systematic optimization via CSR could significantly improve mRNA release from probe-coated MBs (Table S1). To optimize thermal release, we applied OACD via CSR modelling over two rounds. The initial study featured 17 experimental conditions (Table 1). qRT-PCR (n=3 technical replicates/condition) quantified absolute release rates via dilution curve-derived Ct values. Maximum release (54.4±4.6%) occurred at 90°C for 10 min at pH 9, while low release (1.3±1.8%) was observed at 95°C for 15 min at pH 9, suggesting that extended thermal stress caused mRNA degradation or dissociation of the probe-bead complexes. The release across all 17 conditions did not exceed the 48% from our prior work, and so a greater range of conditions was tested. Table 1: Matrix generated using optimal augmented composite design for evaluating mRNA thermal release efficiency from probe-coated MBs. Error terms represent std. dev. (n=3). Run no. Time (min) Temperature (°C) pH % release 1 5 85 7 6.9±4.4 2 5 85 9 2.0±1.4 3 5 95 7 9.2±0.2 4 5 95 9 1.5±0.6 5 15 85 7 3.4±1.2 6 15 85 9 1.9±1.0 7 15 95 7 2.5±1.3 8 15 95 9 1.2±1.7 9 5 85 7 3.0±1.7 10 5 90 8 21.1±1.7 11 5 95 9 2.7±1.8 12 10 85 8 24.0±3.0 13 10 90 9 54.3±4.6 14 10 95 7 12.6±6.2 15 15 85 9 3.5±2.1 16 15 90 7 12.3±0.4 17 15 95 8 2.4±1.3 OACD data further generated a quadratic response surface model (R 2 =0.89, Fig. 5), revealing strong interactions among factors. The global optimum predicted by the model (46% release) was achieved at 10.5 min and 90°C (pH 9). Fig. 5a instead illustrates a local optimum at intermediate temperatures and incubation durations, with reduced release at very short times and at the highest temperatures. Fig. 5b shows that, when time is fixed, elevated pH (≥8) consistently enhances release efficiency across tested temperatures. Fig. 5c instead predicts that maximum efficiencies are obtained at moderate incubation times and high pH values. The linear temperature term (f 2 ) accounted for the highest relative contribution (50%), followed by the quadratic temperature term (f 2 2 , 25%) and the intercept (25%) in the sensitivity analysis (Fig. 5d). In contrast, time (f 1 ), pH (f 3 ), and all interaction terms (f 1 f 2 , f 1 f 3 , f 2 f 3 ) were each <1%, with commensurately low coefficients in Eq. 1; their interactions with temperature were also weak. Eq. 1: Release efficiency=-6550.4+(14.84)f1+(-0.86)f1^2+(146.56)f2+(-0.82)f2^2+(-18.8)f3+(0.42)f3^2+(0.11)f2f3+(0.35)f1f3+(-0.01)f1f2 These results indicate that temperature is the dominant determinant of mRNA release within the design space. Its negative quadratic coefficient in Eq. 1 suggests that excessive heating leads to reduced efficiency, likely due to destabilization of mRNA structures or probe-bead interactions. The negative intercept reflects extrapolation of the model outside the parameter space. The system did not reach the >70% threshold required for robust puromycin-linker annealing and downstream TRAP display screening, warranting further optimization. 3.4.2 Second CSR round Informed by first-round trends, we fixed Tris-HCl at pH 9 and expanded the temperature (60-95°C) and time (5-30 min) ranges, generating a new 9-run matrix (Table 2). Lowest release (5.9±2.8%) occurred at 77.5°C for 5 min, likely reflecting insufficient thermal energy to disrupt mRNA-probe-bead hybridization. The highest release (74±12%) was achieved at 60°C for 30 min, suggesting that prolonged gentle heating permits better kinetic release without mRNA damage. Intermediate conditions (17.5 min at 77.5-95°C) showed moderate release (47-72%), with a local maximum at 77.5°C for 17.5 min (73±11%). A 3D response surface illustrating effects of time and temperature (Fig. 6a) on mRNA thermal release efficiency revealed a broad, high-efficiency region at 60-77.5°C at 20-30 min incubation. There was a sharp decline at higher temperatures, demonstrating that gentler but longer heating is superior. The optimal conditions, which led to a hypothetical 74% release, were 23.5 min at 60°C at pH 9 (vs. 90°C for 10 min for the standard protocol). The pie chart of Fig. 6b depicts that both temperature and time significantly affected release, though temperature slightly more. Under these conditions, a release of 71±7% was obtained, ~3% below the hypothetical max. Eq. 2: Release efficiency=(182.14)+(8.14) f1+(-0.10)f1^2 (-4.80)f2+(0.03)f2^2+(-0.06)f1f2 Eq. 2 confirms the positive effect of time (f 1 ), though its negative quadratic term (-0.10) suggests a diminishing return at longer durations; this could reflect a saturation effect or possible re-adsorption or degradation. In contrast, temperature (f 2 ) showed a negative linear coefficient (-4.80), coupled with a positive quadratic term (0.03), suggesting that while moderate increases may initially reduce release efficiency, further increases can partially compensate. Such behavior may arise from competing mechanisms, including enhanced thermal dissociation at higher temperatures and potential structural destabilization at intermediate ones. The interaction term (-0.06) indicates weak temperature-time coupling. The observation that 60°C for 23.5 min outperformed the standard protocol is worth discussing. First, 60°C only marginally exceeds the melting temperature (~58.6°C) of our 20-mer probe, permitting gradual strand separation via a kinetic route without thermal damage. The presence of 10 mM Tris-HCl at pH 9 helps maintain a stable alkaline environment that reduces acid-driven RNA degradation and permits longer incubations. Furthermore, our moderate optimized temperature of 60°C further preserves target-specific hybridization stringency while minimizing mRNA degradation. By restricting initial exploration to narrow ranges and then iteratively broadening high-value parameters, CSR mitigates overfitting and identifies potentially unexpected global optima. Table 2: CSR-predicted optimal parameter combinations used to guide experiments, along with corresponding measured mRNA release efficiencies. Error terms represent std. dev. (n=3). Run no. Time (min) Temperature (°C) % release 1 5.0 60.0 62.0±5.7 2 5.0 77.5 5.8±2.8 3 5.0 95.0 55.4±7.4 4 17.5 60.0 44.0±5.8 5 17.5 77.5 72.5±10.5 6 17.5 95.0 62.0±9.4 7 30.0 60.0 73.9±12.1 8 30.0 77.5 47.4±1.1 9 30.0 95.0 17.7±3.2 Our finding outperforms aggressive high-temperature/short-time elution. The 60°C marginally exceeds the melting temperature (Tm ~58.55°C) of our 20-mer complementary sequences, permitting gradual strand separation via a kinetic route without thermal damage. The presence of 10 mM Tris-HCl, pH 9, helps maintain a stable alkaline environment, which reduces acid-driven RNA degradation and allows longer incubation without significant loss of mRNA integrity. This dual-round CSR approach shows design-of-experiments efficiency in microfluidics. Comparable CSR/OACD methods in microfluidic drug screening achieved 50-fold precision improvements in dose-response curves via ~10,000-point droplet microfluidic assays, and AI-CSR optimization of 155 cocktail drug combinations on hydrogel chips in 2.5 hr demonstrated analogous rapid exploration of combinatorial landscapes (Kaladharan et al. 2024). By restricting initial exploration to narrow ranges and then iteratively broadening high-value parameters, CSR mitigates overfitting and identifies surprising global optima, here, the unexpected superiority of mild, prolonged elution. 4. Conclusion We showcased CSR-based optimization as a powerful methodology for microfluidic bioassay development, delivering an mRNA thermal release efficiency of 71% at 60°C for 23.45 min (pH 9); gentle, prolonged heating outperformed conventional high-temperature elution. This release efficiency is high enough for non-poly(A) mRNA processing for synthetic biology and therapeutic peptide discovery. Integration with on-chip puromycin annealing and automated TRAP library screening will establish a complete sample-to-peptide pipeline for high-throughput functional screening. Declarations CRediT authorship contribution statement Swati T. Gurme designed and conducted the experiments, analysed the results, and wrote the manuscript; Yi-Cheng Tsai performed the CSR analysis; Bhushan Koparde performed microfluidic analyses and chip operations; Da-Jeng Yao provided CSR and OACD algorithms; Lily Hui-Ching Wang validated the experimental data and provided the mRNA extraction protocol. Gwo-Bin Lee supervised, administered the work, and proofread the manuscript. Declaration of competing interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability Data will be made available upon request. Funding Declaration The authors thank the financial support from the National Science and Technology Council (NSTC) of Taiwan (NSTC 114-2221-E-007-009-MY3, NSTC 112-2221-E-007-077-MY3, MOST 111-2221-E-007-062-MY3, and NSTC 113-2221-E-007-144-MY3). Acknowledgements The authors thank the financial support from the National Science and Technology Council (NSTC) of Taiwan (NSTC 114-2221-E-007-009-MY3, NSTC 112-2221-E-007-077-MY3, MOST 111-2221-E-007-062-MY3, and NSTC 113-2221-E-007-144-MY3). We sincerely acknowledge Dr. Chih-Ming Ho from NTHU for providing valuable discussion on CSR and the OACD algorithm. References Adams NM, Bordelon H, Wang KK, Albert LE, Wright DW and Haselton FR (2015) Comparison of three magnetic bead surface functionalities for RNA extraction and detection. ACS Applied Materials & Interfaces 7(11):6062-6069. DOI: 10.1021/am506374t. Bustin SA (2002) Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): trends and problems. J Mol Endocrinol 29(1): 23-39. DOI: 10.1677/jme.0.0290023. Eastburn DJ, Sciambi A and Abate AR (2013) Ultrahigh-throughput mammalian single-cell reverse-transcriptase polymerase chain reaction in microfluidic drops. Analytical Chemistry 85(16): 8016-8021. DOI: 10.1021/ac402057q. Filiatrault MJ (2011) Progress in prokaryotic transcriptomics. Current Opinion in Microbiology 14(5):579-586. DOI:10.1016/j.mib.2011.07.023. Gurme ST, Su YT, Koparde B, Wang LHC and Lee GB (2026) An integrated microfluidic system for automated extraction of in vitro transcribed mRNAs via probe-coated magnetic beads. Lab on a Chip 26:471-480. DOI: 10.1039/D5LC00938C. Han N, Shin JH and Han KH (2014) An on-chip RT-PCR microfluidic device, that integrates mRNA extraction, cDNA synthesis, and gene amplification. RSC Advances 4(18): 9160-9165. DOI: 10.1039/C3RA47980C. He X, Zhang C, Shi C and Lu Q (2018) Meta-analysis of mRNA expression profiles to identify differentially expressed genes in lung adenocarcinoma tissue from smokers and non-smokers. Oncology Reports 39(3): 929-938. DOI: 10.3892/or.2018.6197. Ishizawa T, Kawakami T, Reid PC and Murakami H (2013) TRAP Display: A High-Speed Selection Method for the Generation of Functional Polypeptides. Journal of the American Chemical Society 135(14): 5433-5440. DOI: 10.1021/ja312579u. Ji J, Hu C, Pang X, Liang J, Huang Q, Hu S, Mei Q and Ma H (2024) Open thermal control system for stable polymerase chain reaction on a digital microfluidic chip. ACS Omega 9(9): 10937-10944. DOI: 10.1021/acsomega.3c10312. Jiang G and Harrison DJ (2000) mRNA isolation in a microfluidic device for eventual integration of cDNA library construction. Analyst 125(12):2176-2179. DOI: 10.1039/B005999O. Kaladharan K, Ouyang CH, Yang HY and Tseng FG (2024) Selectively cross-linked hydrogel-based cocktail drug delivery micro-chip for colon cancer combinatorial drug screening using AI-CSR platform for precision medicine. Lab on a Chip 24(20):4766-4777. DOI: 10.1039/D4LC00520A. Kawakami T, Ogawa K, Goshima N and Natsume T (2015) DIVERSE system: de novo creation of peptide tags for non-enzymatic covalent labeling by in vitro evolution for protein imaging inside living cells. Chemistry & Biology 22(12):1671-1679. DOI: 10.1016/j.chembiol.2015.10.016. Khong J, Lee M, Warren C, Kim UB, Duarte S, Andreoni KA, Shrestha S, Johnson MW, Battula NR and McKimmy DM (2025) Tacrolimus dosing in liver transplant recipients using phenotypic personalized medicine: A phase 2 randomized clinical trial. Nature Communications 16(1):4558.DOI: 10.1038/s41467-025-59739-6. Liu CJ, Lien KY, Weng CY, Shin JW, Chang TY and Lee GB (2009) Magnetic-bead-based microfluidic system for ribonucleic acid extraction and reverse transcription processes. Biomedical Microdevices 11:339-350. DOI: 10.1007/s10544-008-9240-1. Nestorova GG, Hasenstein K, Nguyen N, DeCoster MA and Crews ND (2017) Lab-on-a-chip mRNA purification and reverse transcription via a solid-phase gene extraction technique. Lab on a Chip 17(6):1128-1136. DOI: 10.1039/C6LC01421F. Newton MS, Cabezas-Perusse Y, Tong CL and Seelig B (2020) in vitro selection of peptides and proteins-advantages of mRNA display. ACS Synth. Biol 9:181-190. DOI: 10.1021/acssynbio.9b00419. Ni J, Huang F, Wang B, Li B and Lin Q (2010) A planar PDMS micropump using in-contact minimized-leakage check valves. J Micromech Microeng 20(9):095033. DOI: 10.1088/0960-1317/20/9/095033. Ni J, Li B and Yang JG (2012) A pneumatic PDMS micropump with in-plane check valves for disposable microfluidic systems. Microelectronic Engineering 99:28-32. DOI: 10.1016/j.mee.2012.04.002. Park D, Lee S and Min KT (2020) Techniques for investigating mitochondrial gene expression. BMB Reports 53(1):3-9. DOI: 10.5483/BMBRep.2020.53.1.272. Roberts RW and Szostak JW (1997) RNA-peptide fusions for the in vitro selection of peptides and proteins. Proceedings of the National Academy of Sciences USA 94(23):12297-12302. DOI: 10.1073/pnas.94.23.12297. Sarkar TR and Irudayaraj J (2008) Carboxyl-coated magnetic nanoparticles for mRNA isolation and extraction of supercoiled plasmid DNA. Analytical Biochemistry 379(1): 130-132. DOI: 10.1016/j.ab.2008.04.016. Satterfield BC, Stern S, Caplan MR, Hukari KW and West JAA (2007) Microfluidic purification and preconcentration of mRNA by flow-through polymeric monolith. Analytical Chemistry 79(16): 6230-6235. DOI: 10.1021/ac0709201. Thompson AM, Gansen A, Paguirigan AL, Kreutz JE, Radich JP and Chiu DT (2014) Self-digitization microfluidic chip for absolute quantification of mRNA in single cells. Analytical Chemistry 86(24):12308-12314. DOI: 10.1021/ac5035924. Toriello NM, Douglas ES, Thaitrong N, Hsiao SC, Francis MB, Bertozzi CR and Mathies RA (2008) Integrated microfluidic bioprocessor for single-cell gene expression analysis. Proc Natl Acad Sci USA 105(51):20173-20178. DOI: 10.1073/pnas.0806355106. Tsai YC, Patil PJ, Lin CN, Chen WJ, Hsu KF, Shan YS, Yao DJ, Ho CM and Lee GB (2025) Optimization of drug-dosage combination of cancer therapy for ovarian cancer and cholangiocarcinoma by using complex system response technology. 2025 IEEE 4th International Conference on Micro/Nano Sensors for AI, Healthcare, and Robotics (NSENS). Wang CH, Shao YS, Hsu KF and Lee GB (2024) Detection of dual-methylated BRCA1/BRCA2 cell-free DNA for ovarian cancer on an aptamer-based integrated microfluidic system. Chemical Engineering Journal 495:153478. DOI: 10.1016/j.cej.2024.153478. Wang W, Jiang Q, Argentini M, Cornu D, Gigant B, Knossow M and Wang C (2012) Kif2C minimal functional domain has unusual nucleotide binding properties that are adapted to microtubule depolymerization. Journal of Biological Chemistry 287(18):15143-15153. DOI: 10.1074/jbc.M111.317859. Weiss A, Ding X, van Beijnum JR, Wong I, Wong TJ, Berndsen RH, et al. (2015) Rapid optimization of drug combinations for the optimal angiostatic treatment of cancer. Angiogenesis 18(3):233-244. DOI: 10.1007/s10456-015-9462-9. White AK, VanInsberghe M, Petriv OI, Hamidi M, Sikorski D, Marra MA, Piret J, Aparicio S and Hansen CL (2011) High-throughput microfluidic single-cell RT-qPCR. Proceedings of the National Academy of Sciences 108(34):13999-14004. DOI: 10.1073/pnas.1019446108. Wong PK, Yu F, Shahangian A, Cheng G, Sun R and Ho CM (2008) Closed-loop control of cellular functions using combinatory drugs guided by a stochastic search algorithm. Proc Natl Acad Sci USA 105(13):5105-5110. DOI: 10.1073/pnas.0800823105. Yu T, Tang C, Zhang Y, Zhang R and Yan W (2017) Microfluidics-based digital quantitative PCR for single-cell small RNA quantification. Biol Reprod 97(3): 490-496. DOI: 10.1093/biolre/iox102. Zeng Y, Woolley M, Chockalingam K, Thomas B, Arora S, Hook M and Chen Z (2023) Click display: a rapid and efficient in vitro protein display method for directed evolution. Nucleic Acids Research 51(16): e89-e89. DOI: 10.1093/nar/gkad643. Zuo X, Meng P, Bao Y, Tao C, Wang Y, Liu X, Bu Y and Zhu J (2023) Cell cycle dysregulation with overexpression of KIF2C/MCAK is a critical event in nasopharyngeal carcinoma. Genes & Diseases 10(1):212-227. DOI: 10.1016/j.gendis.2021.05.003. Additional Declarations No competing interests reported. Supplementary Files FinalCSRSupplementaryInformation20260415.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 16 May, 2026 Reviewers agreed at journal 12 May, 2026 Reviewers agreed at journal 11 May, 2026 Reviewers agreed at journal 11 May, 2026 Reviewers agreed at journal 07 May, 2026 Reviewers invited by journal 07 May, 2026 Editor assigned by journal 07 May, 2026 Submission checks completed at journal 05 May, 2026 First submitted to journal 19 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9465529","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":641483984,"identity":"f8bc0f58-6820-471c-922d-a02ea4894a6c","order_by":0,"name":"Swati T. Gurme","email":"","orcid":"","institution":"National Tsing Hua University","correspondingAuthor":false,"prefix":"","firstName":"Swati","middleName":"T.","lastName":"Gurme","suffix":""},{"id":641483985,"identity":"fb2cce9b-77e8-4991-9c22-bdad17033a3e","order_by":1,"name":"Yi-Cheng Tsai","email":"","orcid":"","institution":"National Tsing Hua University","correspondingAuthor":false,"prefix":"","firstName":"Yi-Cheng","middleName":"","lastName":"Tsai","suffix":""},{"id":641483986,"identity":"c80f8f31-634c-498e-b7a1-d6855a172609","order_by":2,"name":"Bhushan Koparde","email":"","orcid":"","institution":"National Tsing Hua University","correspondingAuthor":false,"prefix":"","firstName":"Bhushan","middleName":"","lastName":"Koparde","suffix":""},{"id":641483987,"identity":"0a5b38c7-9915-4ab9-b37e-ebea48271d2f","order_by":3,"name":"Da-Jeng Yao","email":"","orcid":"","institution":"National Tsing Hua University","correspondingAuthor":false,"prefix":"","firstName":"Da-Jeng","middleName":"","lastName":"Yao","suffix":""},{"id":641483988,"identity":"3cc21617-4fd1-40d5-a5c2-fe5e16720a70","order_by":4,"name":"Lily Hui-Ching Wang","email":"","orcid":"","institution":"National Tsing Hua University","correspondingAuthor":false,"prefix":"","firstName":"Lily","middleName":"Hui-Ching","lastName":"Wang","suffix":""},{"id":641483989,"identity":"ee77b0db-d06f-4513-aae4-d7da6488d361","order_by":5,"name":"Gwo-Bin Lee","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6ElEQVRIiWNgGAWjYHACNoYEBhsIk7EBRCYQpSUNwjwI03KAkBYGhsMkaJGPSD724OGO83m67Q1sjz/uOMzAz55jwPyxDbcWwxtp6QaJZ24Xm505wG5w8MxhBsmeNwYMB/FpmZ1jJpHYdjtx240ENomDbYcZDG7kALVsI6jlXOK2+w8gWuwJaZGXBms5ALSFAWqLBAEtBvLPgH5pSwb6JbFN4mxbOo/EmWcFB87+w2NLz+FjD3+22eWZHT98TKKyzVqOvz1544OKM3hsOQChE2BRzwMiDuDWALSlAa5lFIyCUTAKRgEOAACJU1u3wNuPuQAAAABJRU5ErkJggg==","orcid":"","institution":"National Tsing Hua University","correspondingAuthor":true,"prefix":"","firstName":"Gwo-Bin","middleName":"","lastName":"Lee","suffix":""}],"badges":[],"createdAt":"2026-04-20 01:53:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9465529/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9465529/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109474287,"identity":"cb5abecb-b0cf-43fd-bc4e-e5a1a1b883f7","added_by":"auto","created_at":"2026-05-18 13:48:54","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":347419,"visible":true,"origin":"","legend":"\u003cp\u003e(a) Layout of the microfluidic device highlighting microchannel architecture, circular reservoirs, micropumps, micromixers, and three thermoelectric coolers (TEC); (b) enlarged photographic view of the thermal release region with the associated micropump.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-9465529/v1/45b1c4d8e8282e0c572e923b.png"},{"id":109760303,"identity":"e7e3d17d-9ad7-4268-9534-d0a5eddbe61e","added_by":"auto","created_at":"2026-05-22 07:28:30","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":354571,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic of the IVT-translation system coupled with puromycin linker annealing for peptide synthesis. The workflow includes probe-coated MB-based mRNA isolation and subsequent thermal release. (a) IVT; (b) DNase-mediated degradation of residual DNA; (c) hybridization-based capture of mRNA onto probe-coated MBs; (d) sequential washing and thermal release of bound mRNA; (e) puromycin linker annealing; and (f) RT.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-9465529/v1/eb8900702931473954f2c517.png"},{"id":109474289,"identity":"3fadea8d-aef0-4ffa-9824-32d793978737","added_by":"auto","created_at":"2026-05-18 13:48:54","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":218576,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic of the CSR-based workflow employed for parameter optimization for mRNA thermal release on the microfluidic platform, with a 3D model of factor interactions shown as a CSR surface.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-9465529/v1/3150975d71872aea725d6a37.png"},{"id":109474291,"identity":"32b2e755-ef15-4f4a-8426-b08007421912","added_by":"auto","created_at":"2026-05-18 13:48:54","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":82399,"visible":true,"origin":"","legend":"\u003cp\u003e(a) Temperature profiles demonstrating the stability and performance of individual TEC modules, monitored with a thermocouple; (b) characterization of pumping volume for micropumps (6-mm diameter). Error bars and red box in panel b represent std. dev. and optimal condition, respectively.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-9465529/v1/bfdb9dbd4184e87eb7c66367.png"},{"id":109759695,"identity":"e75feb0d-1deb-49b5-b789-5600b0bebb13","added_by":"auto","created_at":"2026-05-22 07:27:33","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":429581,"visible":true,"origin":"","legend":"\u003cp\u003e3D response surface plots generated by CSR for three-factor interactions, illustrating the relationships among time (factor 1), temperature (factor 2), and pH (factor 3) on mRNA thermal release. (a) Interaction between time and temperature, (b) interaction between pH and temperature, (c) interaction between pH and time, and (d) pie chart showing sensitivity analysis.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-9465529/v1/88ab6422a9264fde4a4597fe.png"},{"id":109474293,"identity":"71c8a89f-9fdf-4cf6-a349-0bb69842996b","added_by":"auto","created_at":"2026-05-18 13:48:54","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":154729,"visible":true,"origin":"","legend":"\u003cp\u003e3D response surface plot generated by CSR illustrating the relationship among time (factor 1), temperature (factor 2), and mRNA thermal release. (a) Interaction between time and temperature, and (b) pie chart showing results of sensitivity analysis.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-9465529/v1/fef2880f8dbcdaa97e67280c.png"},{"id":109765087,"identity":"3909a66a-5fd5-436d-9e61-59470958405b","added_by":"auto","created_at":"2026-05-22 07:39:23","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1620778,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9465529/v1/ce71d4b8-da64-48af-8280-2d798970a08c.pdf"},{"id":109759715,"identity":"f68ed1ec-cd11-488f-bd6e-7cf38c0781fa","added_by":"auto","created_at":"2026-05-22 07:27:35","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":16743,"visible":true,"origin":"","legend":"","description":"","filename":"FinalCSRSupplementaryInformation20260415.docx","url":"https://assets-eu.researchsquare.com/files/rs-9465529/v1/dac1f33df6e4a1086b876776.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Optimization of mRNA Thermal Release on an Integrated Microfluidic Platform by a Complex System Response Method","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eMessenger ribonucleic acid (mRNAs) link genomic information to protein synthesis, and their quantification has advanced a variety of applications ranging from gene‑expression profiling, both basic science (He et al. \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) and diagnostics (Bustin \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2002\u003c/span\u003e). Real‑time reverse transcription polymerase chain reaction (qRT‑PCR) is the primary method for quantifying mRNA because it is sensitive, specific, and high‑throughput (Han et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Still, its accuracy depends on high‑quality, undegraded, and DNA-free mRNAs. Accordingly, robust mRNA isolation is a crucial first step in workflows ranging from cDNA library construction (Jiang and Harrison \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2000\u003c/span\u003e) to \u0026ldquo;sample-to-answer\u0026rdquo; assays (Bustin \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2002\u003c/span\u003e, Nestorova et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Conventional RNA extraction typically combines chaotropic lysis, organic extraction, and alcohol precipitation, followed by enrichment of polyadenylated transcripts (Nestorova et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2017\u003c/span\u003e); however, these procedures are laborious, use hazardous reagents (such as phenol), and are prone to RNA loss/degradation (Sarkar and Irudayaraj \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2008\u003c/span\u003e, Nestorova et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Alternatively, solid-phase capture methods based on silica adsorption or oligo(dT)-mediated hybridization on magnetic beads (MBs) enable faster binding-wash-elution cycles and avoid organic solvents. The latter offer high surface area and straightforward magnetic manipulation, and carboxyl-coated magnetic nanoparticles functionalized with oligo(dT)\u003csub\u003e25\u003c/sub\u003e can provide \u0026micro;g-scale mRNA yields at reduced costs (Sarkar \u0026amp; Irudayaraj \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). However, these methods require open-tube handling and multiple centrifugation or magnetization steps that introduce variability and ribonucleases (RNase) exposure.\u003c/p\u003e \u003cp\u003eFlow-through porous polymer monoliths formed in capillaries or channels and functionalized with oligo(dT) or locked nucleic acid (LNA) substitutes offer high surface areas and purify mRNAs from total RNAs with yields and purities comparable to commercial column kits in \u0026lt;\u0026thinsp;20 \u0026micro;L. Satterfield et al. (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2007\u003c/span\u003e) reported UV-polymerized oligo(dT) methacrylate monoliths in capillaries that achieved near-maximal binding in seconds, capturing\u0026thinsp;\u0026ge;\u0026thinsp;16 \u0026micro;g of mRNA from 315 \u0026micro;g of total RNA in 0.4 \u0026micro;L, with 15-fold rRNA enrichment in NaCl solution (up to 110-fold when using LNA probes or tetramethylammonium chloride). Solid-phase gene extraction further simplifies sampling by using dT(15)-modified steel or glass needles that briefly contact tissues or cells to hybridize poly(A) transcripts, which can then be released directly into tubes or chips. This eliminates the need for bulk lysis or centrifugation. Nesterova et al. (2017) used this approach to extract 100\u0026ndash;300 pg of mRNA from glioblastoma spheroids.\u003c/p\u003e \u003cp\u003eWhile oligo(dT)-based capture is highly effective, it inherently excludes non-poly(A) mRNAs, many non-coding RNAs, and bacterial or organellar transcripts that lack canonical poly(A) tails. For prokaryotic and mitochondrial transcriptomics (Filiatrault \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2011\u003c/span\u003e, Park et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) or \u003cem\u003ein vitro\u003c/em\u003e display technologies that operate on synthetic mRNAs (Roberts \u0026amp; Szostak \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e1997\u003c/span\u003e), alternative strategies are required. These can include total RNA capture (Liu et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2009\u003c/span\u003e), sequence-specific hybridization (Adams et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), or ribosome-associated purification (Zeng et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Classical mRNA display (Newton et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) platforms and newer variants like \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003etr\u003c/span\u003eanscription-translation coupled with \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ea\u003c/span\u003essociation of \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ep\u003c/span\u003euromycin linker (TRAP; Ishizawa et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2013\u003c/span\u003e) and click display (Zeng et al. 2024) covalently link polypeptides to their encoding mRNA or cDNA via puromycin-containing DNA linkers and rely on sequence-defined transcripts generated \u003cem\u003ein vitro\u003c/em\u003e rather than on poly(A) tails; these methods enable rapid selection from 10\u003csup\u003e12\u003c/sup\u003e-10\u003csup\u003e13\u003c/sup\u003e peptides but are sensitive to RNases and involve complex chemistries. In addition, lack of reliable sequence information, as well as the complexity of reaction integration and handling, limit throughput, reproducibility, and performance.\u003c/p\u003e \u003cp\u003eRecently, microfluidic lab-on-a-chip platforms can integrate these capture/release strategies with downstream enzymatic reactions while minimizing sample loss and RNase contamination (Han et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Current devices can conduct lysis (Eastburn et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2013\u003c/span\u003e), mRNA capture (Gurme et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2026\u003c/span\u003e), cDNA synthesis, and amplification in closed microfluidic environments, enabling\u0026thinsp;\u0026lt;\u0026thinsp;\u0026micro;L assays and even single-cell analyses (Toriello et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). An example is the Y-channel glass microchip (Jiang \u0026amp; Harrison \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2000\u003c/span\u003e), which used paramagnetic oligo(dT)\u003csub\u003e25\u003c/sub\u003e beads to capture poly(A) mRNA from total RNA under laminar flow. Beads and RNAs were introduced through separate inlets, enabling on-chip hybridization, followed by magnetic trapping of bead-mRNA complexes. The captured mRNA was subsequently washed off-chip for analysis by capillary gel electrophoresis or RT-PCR, preserving integrity of both abundant and low-copy transcripts. The system captured 3\u0026ndash;34 ng of mRNA (26%) from 1\u0026ndash;10 \u0026micro;g of total RNA, though off-chip pooling was required. Similarly, a PDMS-glass RT-PCR chip with permalloy wires was designed for lateral deflection of oligo(dT) bead-mRNA complexes from lysed samples, enabling solid-phase RT from 0.1 \u0026micro;L of blood, SKBR3 cells, or viral swabs (Hans 2014).\u003c/p\u003e \u003cp\u003eThese examples demonstrate potential, but limitations still persist. Furthermore, non-poly(A) mRNAs demand sequence-specific probes incompatible with oligo(dT), highlighting the need for fully automated, high-capacity platforms supporting diverse transcripts and downstream workflows. Achieving such performance requires precise control over multiple parameters- temperatures, buffers, flow rates, bead loading, and release conditions- that conventional trial-and-error optimization struggles to resolve efficiently. Statistical optimization frameworks are increasingly important for fine-tuning these multivariate workflows. \u0026ldquo;Complex system response\u0026rdquo; (CSR) approaches treat the device as a \u0026ldquo;black box\u0026rdquo; and fit low-order polynomial response surfaces that link a small number of carefully chosen experimental conditions to quantitative readouts (Wong et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2008\u003c/span\u003e, Kaladharan et al. \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2024\u003c/span\u003e, Tsai et al. \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). CSR has been used with orthogonal array composite designs to optimize 10-drug cocktails and dose levels from 155 combinations in a 3-level search space, while in clinical settings, phenotypic response surface-based personalized medicine has reduced hospital stays for liver transplant recipients (Khong et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). In another case, a feedback-controlled platform was reported to navigate the vast combinatorial space of angiostatic drugs, identifying a potent four-drug synergy that effectively inhibits vessel growth at highly reduced doses (Weiss et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2015\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWe previously demonstrated that sequence-specific probe-coated MBs capture \u003cem\u003ein vitro-\u003c/em\u003e transcribed (IVT), non-poly(A) mRNAs with 93% efficiency on-chip, but only 48% of bound transcripts are thermally released at 95\u0026deg;C (Gurme et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). Because IVT mRNAs used for TRAP display lack poly(A) tails, conventional oligo(dT) solid-phase extraction is not applicable, and low release efficiency can thwart the ability to generate high-diversity peptide-mRNA libraries. Herein, we adopted CSR to identify the combination of three reaction parameters-release temperature, buffer pH, and incubation time- that would lead to maximum mRNA release following a sequence-specific probe-coated MB protocol; this data-driven approach was hypothesized to preserve specificity at a significantly reduced workflow.\u003c/p\u003e"},{"header":"2. Materials and methods","content":"\u003cp\u003e\u003cstrong\u003e2.1 Library preparation and probe design\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;We employed the same library preparation and probe design protocols as in Gurme et al. (2026). A random dsDNA library was constructed for affinity peptide screening of kinesin family member 2C (KIF2C), which contributes to tumor growth and metastasis (Wang et al. 2012, Zuo et al. 2023). The 115-bp sequence (Genomics Biosci. \u0026amp; Tech, Taiwan) contained a T7 promoter, Shine-Dalgarno sequence, peptide-coding region (1 Met, 10 NNK [N-A/T/C/G \u0026amp; K-G/T]), random amino acids, 5 Gly spacer codons, 1 Ser (Kawakami et al. 2015), and an An21 sequence at the 3\u0026prime; end (underlined) at a concentration of 386 ng \u0026mu;L\u003csup\u003e-1\u003c/sup\u003e:\u003c/p\u003e\n\u003cp\u003e5\u0026prime;-ATACTAATACGACTCACTATAGGATTAAGGAGGTGATATTTATGNNKNNKNNKNNKNNKNNKNNKNNKG\u003cbr\u003eGTGGAGGAGGAGGTAGC\u003cu\u003eTAGGACGGGGGGCGGGAGGCGGG\u003c/u\u003e-3\u0026prime;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 Probe synthesis and MB conjugation\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eComplementary probes (5\u0026prime;‑amino C12 linker: 5\u0026prime;‑CATAAATATCACCTCCTTAA‑3\u0026prime;; 100 \u0026mu;M) were synthesized (Genomics Biosci. \u0026amp; Tech, Taiwan) and covalently bound to carboxylic acid MBs for mRNA capture from IVT reactions, following the established 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide/N-hydroxysuccinimide (EDC/ NHS) coupling protocol (Wang et al. 2024). Dynabeads\u0026trade; MyOne\u0026trade; carboxylic acid beads (2.8 \u0026micro;m in diameter, 2\u0026times;10\u003csup\u003e9\u003c/sup\u003e beads/mL, Thermo-Fisher Scientific [TFS]) were vortexed (Vortex-Genie 2, Scientific Industries, USA) for 5 s in 100 \u0026micro;L, diluted in 900 \u0026micro;L of ddH\u003csub\u003e2\u003c/sub\u003eO, magnetically separated (DynaMag\u0026trade;-2 magnet), and washed twice. Beads were then incubated for 18 hr at room temperature in the dark (Intelli-mixer RM-26, ELMI, 20 RPM, C2 mode) with 950 \u0026micro;L of ddH\u003csub\u003e2\u003c/sub\u003eO containing 30 \u0026micro;L of 100 \u0026micro;M amino-modified probe and 20 \u0026micro;L of 120 mg mL\u003csup\u003e-1\u0026nbsp;\u003c/sup\u003eEDC (TFS). After incubation, beads underwent sequential washes with 0.02% Tween 20 (Merck, Germany, 2\u0026times;), 0.1% SDS (Merck, 2\u0026times;), and 0.1 M ethanolamine (Merck) blocking (1 hr, dark). The final probe-functionalized beads were resuspended in 1 mL of ddH\u003csub\u003e2\u003c/sub\u003eO and stored at 4\u0026deg;C.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Design, microfabrication, and operation of the microfluidic device\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAn integrated microfluidic chip (46 \u0026times; 53 mm; air-control/PDMS liquid-channel layers bonded to glass) microfabricated from PMMA molds with PDMS replication and O\u003csub\u003e2\u003c/sub\u003e plasma bonding adopted from Gurme et al. (2026) was used for mRNA capture and release experiments. Briefly, the device integrated 16 air inlets, 3 micromixers, 2 suction micropumps (D/E in Fig. 1: R=3 mm; H: R=1.5 mm), 10 microvalves (2.8 \u0026times; 2.7 mm), and 6 chambers (A-J in Fig. 1, diameter: 2-7.6 mm), with three thermoelectric coolers (TECs; 9.5 W, 15\u0026times;15\u0026times;3.2 mm, 739-387004942, Mouser, Taiwan) for IVT/DNase (37\u0026deg;C) and reverse transcription (RT (42\u0026deg;C). The third was for thermal release at varying temperatures under Arduino feedback control (MAX6675, Maxim Integrated Products, USA). Samples/reagents were first loaded into chambers; pneumatic actuation by a compressor (DRS-210-22, SWAN, Taiwan) or vacuum pump (DF-506 K, Doctor\u0026apos;s Friend, Taiwan) activated flow/mixing via an electromagnetic valve (EMV) module while TECs were operated at 6 V/3 A. DNase-treated IVT mixture (20 \u0026micro;L, 1 U \u0026micro;L\u003csup\u003e-1\u003c/sup\u003e) was hybridized on-chip with probe-coated MBs (20 \u0026micro;L, 10\u003csup\u003e8\u003c/sup\u003e beads/mL) in micromixer E. Complexes were washed twice with 25 \u0026micro;L of binding/washing buffer (15 mM sodium citrate, 150 mM NaCl, pH 6.8-7.0; F\u0026rarr;D\u0026rarr;E\u0026rarr;W; \u0026nbsp;Fig. 1). For the release buffer, we used 10 \u0026micro;L of 10 mM Tris-HCl (pH 7, 8, or 9; chamber G), splitting unbound mRNA into puromycin annealing (I) and RT (J) chambers.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.4 On-chip thermal release\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; We followed the same IVT, DNase treatment, hybridization, washing, and qRT‑PCR workflow as in Gurme et al. (2026), with modifications only to the thermal release step. Briefly, IVT was performed at 37\u0026deg;C for 60 min using the T7 RiboMAX\u0026trade; Express RNAi system (Promega, USA), followed by DNase treatment (RQ1, 1 U \u0026micro;L\u003csup\u003e-1\u003c/sup\u003e, 37\u0026deg;C, 15 min) (Fig. 2). The DNase‑treated IVT mixture (10 pM mRNA in all experiments) was hybridised on-chip with 20 \u0026micro;L of probe‑coated MBs (10\u003csup\u003e8\u003c/sup\u003e beads mL\u003csup\u003e-1\u003c/sup\u003e) for 15 min at room temperature. After magnetic separation, the supernatant was retained for unbound mRNA assessment. MB-mRNA complexes were washed twice (25 \u0026micro;L) with binding/washing buffer, and mRNA was eluted in 10 \u0026micro;L of 10 mM Tris‑HCl at different temperatures (according to CSR). TECs were used for precise temperature control. The eluted mRNA (8 \u0026micro;L) was annealed with puromycin linker (1.4 \u0026micro;L, 100 \u0026micro;M; 25\u0026deg;C for 30 min). The remaining 2 \u0026micro;L were used for qRT-PCR with 10 \u0026micro;L of KAPA SYBR FAST (Sigma-Aldrich), 7.15 \u0026micro;L ddH\u003csub\u003e2\u003c/sub\u003eO, 0.40 \u0026micro;L RT mix (Sigma-Aldrich), 0.20 \u0026micro;L of forward primer (T7SD8M2.F44, 10 \u0026micro;M, 5\u0026rsquo;-ATACTAATACGACTCACTATAGGATTAAGGAGGTGATATTTATG-3\u0026rsquo;), 0.25 \u0026micro;L of reverse primer (G5S-4.R20, 10 \u0026micro;M, 5\u0026rsquo;-CCACCTCCTCCTCCATCGAT-3\u0026rsquo;; Protech, Taiwan) on a StepOnePlus\u0026trade; (Applied Biosystems). Positive (total mRNA), uncoated beads, and ddH\u003csub\u003e2\u003c/sub\u003eO negative controls were included. qRT-PCR cycling began with RT at 50\u0026deg;C for 5 min, then 42\u0026deg;C for 20 min. After a three-min hot-start at 95\u0026deg;C, 35 cycles of 95\u0026deg;C for 15 s (denaturation) and 65\u0026deg;C for 20 s (annealing/extension) were performed.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.5 CSR optimization for thermal release of mRNA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe used CSR‑based OACD to systematically optimize thermal release of mRNAs (Fig. 3). In the first run, three factors were allowed to vary over narrow ranges-release temperature (85-95\u0026deg;C), incubation time (5-15 min), and release buffer (10 mM of Tris‑HCl) pH (7-9)- using 10 pM of input mRNA and citrate/NaCl as the capture buffer. The design generated 17 conditions to test, far fewer than a full factorial; however, no result was satisfactory for subsequent analyses. We therefore broadened the dynamic range for a second optimization round, expanding temperature from 60 to 95\u0026deg;C and release time from 5 to 30 min, with a constant release buffer pH of 9 (based on the first OACD trends) and keeping all other conditions unchanged. The second OACD matrix generated a total of nine experimental conditions, which were subsequently tested on-chip. \u0026nbsp;\u003c/p\u003e"},{"header":"3. Result and discussion","content":"\u003cp\u003e\u003cstrong\u003e3.1 Temperature control\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTEC modules delivered precise, stable control across 60 to 95\u0026deg;C. Target temperatures of 60, 77, 85, 90, and 95\u0026deg;C were achieved within \u0026lt;1 min (\u0026amp; within 1\u0026deg;C) with minimal overshoot (\u0026lt;1\u0026deg;C) and long-term stability (\u0026lt;0.5\u0026deg;C standard deviation (std. dev.); Fig. 4a). This performance aligns with the \u0026lt;0.1-0.5\u0026deg;C precision demanded for lysis, RT, and elution (Ji et al. 2024). No recalibration was needed, confirming the robustness of temperature control.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 Microfluidic component performance\u003c/strong\u003e\u003cbr clear=\"ALL\"\u003eThe large micropumps (D/E) reliably dispensed 25 \u0026micro;L (\u0026le;35 kPa negative/20 kPa positive gauge; Fig. 4b). Small micropumps (H) dispended 2 \u0026micro;L at 20 kPa, and micromixers achieved \u0026gt;90% homogenization in 14 s (20-35 kPa); complete mixing occurred in \u0026lt;16 s at 2 Hz, ideal for limiting shear damage. These metrics match PDMS micropump benchmarks (10-50 \u0026micro;L min\u003csup\u003e-1\u003c/sup\u003e at 20-60 kPa; \u0026lt;5% CV), where pressure-dependent stroke volumes and minimal leakage (\u0026lt;1%) are recommended to enable precise, unidirectional reagent delivery (Ni et al. 2010, 2012). The planar PDMS design is attractive because it simplifies microfabrication while still allowing continuous pumping under pneumatic actuation and demonstrating robust, low-leakage, low-evaporation performance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 qRT-PCR dilution curve\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/strong\u003ePlease see Gurme et al. (2026) for details. Briefly, R\u003csup\u003e2\u003c/sup\u003e was 0.995, and no amplification was seen in negative controls.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4 CSR-based thermal release optimization\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4.1. 17-experiment design\u003c/strong\u003e \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Systematic optimization via CSR could significantly improve mRNA release from probe-coated MBs (Table S1). To optimize thermal release, we applied OACD via CSR modelling over two rounds. The initial study featured 17 experimental conditions (Table 1). qRT-PCR (n=3 technical replicates/condition) quantified absolute release rates via dilution curve-derived Ct values. Maximum release (54.4\u0026plusmn;4.6%) occurred at 90\u0026deg;C for 10 min at pH 9, while low release (1.3\u0026plusmn;1.8%) was observed at 95\u0026deg;C for 15 min at pH 9, suggesting that extended thermal stress caused mRNA degradation or dissociation of the probe-bead complexes. The release across all 17 conditions did not exceed the 48% from our prior work, and so a greater range of conditions was tested.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1:\u003c/strong\u003e Matrix generated using optimal augmented composite design for evaluating mRNA thermal release efficiency from probe-coated MBs. Error terms represent std. dev. (n=3).\u0026nbsp;\u003c/p\u003e\n\u003cdiv align=\"center\"\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"492\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eRun no.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTime (min)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTemperature (\u0026deg;C)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003epH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e% release\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6.9\u0026plusmn;4.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e2.0\u0026plusmn;1.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e9.2\u0026plusmn;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1.5\u0026plusmn;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e3.4\u0026plusmn;1.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1.9\u0026plusmn;1.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e2.5\u0026plusmn;1.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e1.2\u0026plusmn;1.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e3.0\u0026plusmn;1.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e21.1\u0026plusmn;1.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e2.7\u0026plusmn;1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e24.0\u0026plusmn;3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e54.3\u0026plusmn;4.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e12.6\u0026plusmn;6.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e3.5\u0026plusmn;2.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e12.3\u0026plusmn;0.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e2.4\u0026plusmn;1.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; OACD data further generated a quadratic response surface model (R\u003csup\u003e2\u003c/sup\u003e=0.89, Fig. 5), revealing strong interactions among factors. The global optimum predicted by the model (46% release) was achieved at 10.5 min and 90\u0026deg;C (pH 9). Fig. 5a instead illustrates a local optimum at intermediate temperatures and incubation durations, with reduced release at very short times and at the highest temperatures. Fig. 5b shows that, when time is fixed, elevated pH (\u0026ge;8) consistently enhances release efficiency across tested temperatures. Fig. 5c instead predicts that maximum efficiencies are obtained at moderate incubation times and high pH values. The linear temperature term (f\u003csub\u003e2\u003c/sub\u003e) accounted for the highest relative contribution (50%), followed by the quadratic temperature term (f\u003csub\u003e2\u003c/sub\u003e\u003csup\u003e2\u003c/sup\u003e, 25%) and the intercept (25%) in the sensitivity analysis (Fig. 5d). In contrast, time (f\u003csub\u003e1\u003c/sub\u003e), pH (f\u003csub\u003e3\u003c/sub\u003e), and all interaction terms (f\u003csub\u003e1\u003c/sub\u003ef\u003csub\u003e2\u003c/sub\u003e, f\u003csub\u003e1\u003c/sub\u003ef\u003csub\u003e3\u003c/sub\u003e, f\u003csub\u003e2\u003c/sub\u003ef\u003csub\u003e3\u003c/sub\u003e) were each \u0026lt;1%, with commensurately low coefficients in Eq. 1; their interactions with temperature were also weak. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEq. 1:\u003c/strong\u003e Release efficiency=-6550.4+(14.84)f1+(-0.86)f1^2+(146.56)f2+(-0.82)f2^2+(-18.8)f3+(0.42)f3^2+(0.11)f2f3+(0.35)f1f3+(-0.01)f1f2 \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThese results indicate that temperature is the dominant determinant of mRNA release within the design space. Its negative quadratic coefficient in Eq. 1 suggests that excessive heating leads to reduced efficiency, likely due to destabilization of mRNA structures or probe-bead interactions. The negative intercept reflects extrapolation of the model outside the parameter space. The system did not reach the \u0026gt;70% threshold required for robust puromycin-linker annealing and downstream TRAP display screening, warranting further optimization.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4.2 Second CSR round\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eInformed by first-round trends, we fixed Tris-HCl at pH 9 and expanded the temperature (60-95\u0026deg;C) and time (5-30 min) ranges, generating a new 9-run matrix (Table 2). Lowest release (5.9\u0026plusmn;2.8%) occurred at 77.5\u0026deg;C for 5 min, likely reflecting insufficient thermal energy to disrupt mRNA-probe-bead hybridization. The highest release (74\u0026plusmn;12%) was achieved at 60\u0026deg;C for 30 min, suggesting that prolonged gentle heating permits better kinetic release without mRNA damage. Intermediate conditions (17.5 min at 77.5-95\u0026deg;C) showed moderate release (47-72%), with a local maximum at 77.5\u0026deg;C for 17.5 min (73\u0026plusmn;11%). A 3D response surface illustrating effects of time and temperature (Fig. 6a) on mRNA thermal release efficiency revealed a broad, high-efficiency region at 60-77.5\u0026deg;C at 20-30 min incubation. There was a sharp decline at higher temperatures, demonstrating that gentler but longer heating is superior. The optimal conditions, which led to a hypothetical 74% release, were 23.5 min at 60\u0026deg;C at pH 9 (vs. 90\u0026deg;C for 10 min for the standard protocol). The pie chart of Fig. 6b depicts that both temperature and time significantly affected release, though temperature slightly more. Under these conditions, a release of 71\u0026plusmn;7% was obtained, ~3% below the hypothetical max.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEq. 2:\u003c/strong\u003e Release efficiency=(182.14)+(8.14) f1+(-0.10)f1^2 (-4.80)f2+(0.03)f2^2+(-0.06)f1f2 \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eEq. 2 confirms the positive effect of time (f\u003csub\u003e1\u003c/sub\u003e), though its negative quadratic term (-0.10) suggests a diminishing return at longer durations; this could reflect a saturation effect or possible re-adsorption or degradation. In contrast, temperature (f\u003csub\u003e2\u003c/sub\u003e) showed a negative linear coefficient (-4.80), coupled with a positive quadratic term (0.03), suggesting that while moderate increases may initially reduce release efficiency, further increases can partially compensate. Such behavior may arise from competing mechanisms, including enhanced thermal dissociation at higher temperatures and potential structural destabilization at intermediate ones. The interaction term (-0.06) indicates weak temperature-time coupling.\u003c/p\u003e\n\u003cp\u003eThe observation that 60\u0026deg;C for 23.5 min outperformed the standard protocol is worth discussing. First, 60\u0026deg;C only marginally exceeds the melting temperature (~58.6\u0026deg;C) of our 20-mer probe, permitting gradual strand separation via a kinetic route without thermal damage. The presence of 10 mM Tris-HCl at pH 9 helps maintain a stable alkaline environment that reduces acid-driven RNA degradation and permits longer incubations. Furthermore, our moderate optimized temperature of 60\u0026deg;C further preserves target-specific hybridization stringency while minimizing mRNA degradation. By restricting initial exploration to narrow ranges and then iteratively broadening high-value parameters, CSR mitigates overfitting and identifies potentially unexpected global optima.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2:\u003c/strong\u003e CSR-predicted optimal parameter combinations used to guide experiments, along with corresponding measured mRNA release efficiencies. Error terms represent std. dev. (n=3).\u003c/p\u003e\n\u003cdiv align=\"center\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"438\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eRun no.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTime (min)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTemperature (\u0026deg;C)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e% release\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e60.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e62.0\u0026plusmn;5.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e77.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5.8\u0026plusmn;2.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e5.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e55.4\u0026plusmn;7.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e17.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e60.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e44.0\u0026plusmn;5.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e17.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e77.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e72.5\u0026plusmn;10.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e17.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e62.0\u0026plusmn;9.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e30.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e60.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e73.9\u0026plusmn;12.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e30.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e77.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e47.4\u0026plusmn;1.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e30.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e95.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e17.7\u0026plusmn;3.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eOur finding outperforms aggressive high-temperature/short-time elution. The 60\u0026deg;C marginally exceeds the melting temperature (Tm ~58.55\u0026deg;C) of our 20-mer complementary sequences, permitting gradual strand separation via a kinetic route without thermal damage. The presence of 10 mM Tris-HCl, pH 9, helps maintain a stable alkaline environment, which reduces acid-driven RNA degradation and allows longer incubation without significant loss of mRNA integrity. This dual-round CSR approach shows design-of-experiments efficiency in microfluidics. Comparable CSR/OACD methods in microfluidic drug screening achieved 50-fold precision improvements in dose-response curves via ~10,000-point droplet microfluidic assays, and AI-CSR optimization of 155 cocktail drug combinations on hydrogel chips in 2.5 hr demonstrated analogous rapid exploration of combinatorial landscapes (Kaladharan et al. 2024). By restricting initial exploration to narrow ranges and then iteratively broadening high-value parameters, CSR mitigates overfitting and identifies surprising global optima, here, the unexpected superiority of mild, prolonged elution.\u0026nbsp;\u003c/p\u003e"},{"header":"4. Conclusion","content":"\u003cp\u003eWe showcased CSR-based optimization as a powerful methodology for microfluidic bioassay development, delivering an mRNA thermal release efficiency of 71% at 60\u0026deg;C for 23.45 min (pH 9); gentle, prolonged heating outperformed conventional high-temperature elution. This release efficiency is high enough for non-poly(A) mRNA processing for synthetic biology and therapeutic peptide discovery. Integration with on-chip puromycin annealing and automated TRAP library screening will establish a complete sample-to-peptide pipeline for high-throughput functional screening.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eCRediT authorship contribution statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Swati T. Gurme designed and conducted the experiments, analysed the results, and wrote the manuscript; Yi-Cheng Tsai performed the CSR analysis; Bhushan Koparde performed microfluidic analyses and chip operations; Da-Jeng Yao provided CSR and OACD algorithms; Lily Hui-Ching Wang validated the experimental data and provided the mRNA extraction protocol. Gwo-Bin Lee supervised, administered the work, and proofread the manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of competing interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp;Data will be made available upon request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank the financial support from the National Science and Technology Council (NSTC) of Taiwan (NSTC 114-2221-E-007-009-MY3, NSTC 112-2221-E-007-077-MY3, MOST 111-2221-E-007-062-MY3, and NSTC 113-2221-E-007-144-MY3).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; The authors thank the financial support from the National Science and Technology Council (NSTC) of Taiwan (NSTC 114-2221-E-007-009-MY3, NSTC 112-2221-E-007-077-MY3, MOST 111-2221-E-007-062-MY3, and NSTC 113-2221-E-007-144-MY3). We sincerely acknowledge Dr. Chih-Ming Ho from NTHU for providing valuable discussion on CSR and the OACD algorithm.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAdams NM, Bordelon H, Wang KK, Albert LE, Wright DW and Haselton FR (2015) Comparison of three magnetic bead surface functionalities for RNA extraction and detection. ACS Applied Materials \u0026amp; Interfaces 7(11):6062-6069. DOI: 10.1021/am506374t.\u003c/li\u003e\n \u003cli\u003eBustin SA (2002) Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): trends and problems. J Mol Endocrinol 29(1): 23-39. DOI: 10.1677/jme.0.0290023.\u003c/li\u003e\n \u003cli\u003eEastburn DJ, Sciambi A and Abate AR (2013) Ultrahigh-throughput mammalian single-cell reverse-transcriptase polymerase chain reaction in microfluidic drops. Analytical Chemistry 85(16): 8016-8021. DOI: 10.1021/ac402057q.\u003c/li\u003e\n \u003cli\u003eFiliatrault MJ (2011) Progress in prokaryotic transcriptomics. Current Opinion in Microbiology 14(5):579-586. DOI:10.1016/j.mib.2011.07.023.\u003c/li\u003e\n \u003cli\u003eGurme ST, Su YT, Koparde B, Wang LHC and Lee GB (2026) An integrated microfluidic system for automated extraction of \u003cem\u003ein vitro\u003c/em\u003e transcribed mRNAs via probe-coated magnetic beads. Lab on a Chip 26:471-480. DOI: 10.1039/D5LC00938C.\u003c/li\u003e\n \u003cli\u003eHan N, Shin JH and Han KH (2014) An on-chip RT-PCR microfluidic device, that integrates mRNA extraction, cDNA synthesis, and gene amplification. RSC Advances 4(18): 9160-9165. DOI: 10.1039/C3RA47980C.\u003c/li\u003e\n \u003cli\u003eHe X, Zhang C, Shi C and Lu Q (2018) Meta-analysis of mRNA expression profiles to identify differentially expressed genes in lung adenocarcinoma tissue from smokers and non-smokers. Oncology Reports 39(3): 929-938. DOI: 10.3892/or.2018.6197.\u003c/li\u003e\n \u003cli\u003eIshizawa T, Kawakami T, Reid PC and Murakami H (2013) TRAP Display: A High-Speed Selection Method for the Generation of Functional Polypeptides. Journal of the American Chemical Society 135(14): 5433-5440. DOI: 10.1021/ja312579u.\u003c/li\u003e\n \u003cli\u003eJi J, Hu C, Pang X, Liang J, Huang Q, Hu S, Mei Q and Ma H (2024) Open thermal control system for stable polymerase chain reaction on a digital microfluidic chip. ACS Omega 9(9): 10937-10944. DOI: 10.1021/acsomega.3c10312.\u003c/li\u003e\n \u003cli\u003eJiang G and Harrison DJ (2000) mRNA isolation in a microfluidic device for eventual integration of cDNA library construction. Analyst 125(12):2176-2179. DOI: 10.1039/B005999O.\u003c/li\u003e\n \u003cli\u003eKaladharan K, Ouyang CH, Yang HY and Tseng FG (2024) Selectively cross-linked hydrogel-based cocktail drug delivery micro-chip for colon cancer combinatorial drug screening using AI-CSR platform for precision medicine. Lab on a Chip 24(20):4766-4777. DOI: 10.1039/D4LC00520A.\u003c/li\u003e\n \u003cli\u003eKawakami T, Ogawa K, Goshima N and Natsume T (2015) DIVERSE system: de novo creation of peptide tags for non-enzymatic covalent labeling by \u003cem\u003ein vitro\u003c/em\u003e evolution for protein imaging inside living cells. Chemistry \u0026amp; Biology 22(12):1671-1679. DOI: 10.1016/j.chembiol.2015.10.016.\u003c/li\u003e\n \u003cli\u003eKhong J, Lee M, Warren C, Kim UB, Duarte S, Andreoni KA, Shrestha S, Johnson MW, Battula NR and McKimmy DM (2025) Tacrolimus dosing in liver transplant recipients using phenotypic personalized medicine: A phase 2 randomized clinical trial. Nature Communications 16(1):4558.DOI: 10.1038/s41467-025-59739-6.\u003c/li\u003e\n \u003cli\u003eLiu CJ, Lien KY, Weng CY, Shin JW, Chang TY and Lee GB (2009) Magnetic-bead-based microfluidic system for ribonucleic acid extraction and reverse transcription processes. Biomedical Microdevices 11:339-350. DOI: 10.1007/s10544-008-9240-1.\u003c/li\u003e\n \u003cli\u003eNestorova GG, Hasenstein K, Nguyen N, DeCoster MA and Crews ND (2017) Lab-on-a-chip mRNA purification and reverse transcription via a solid-phase gene extraction technique. Lab on a Chip 17(6):1128-1136. DOI: 10.1039/C6LC01421F.\u003c/li\u003e\n \u003cli\u003eNewton MS, Cabezas-Perusse Y, Tong CL and Seelig B (2020) \u003cem\u003ein vitro\u003c/em\u003e selection of peptides and proteins-advantages of mRNA display. ACS Synth. Biol 9:181-190. DOI: 10.1021/acssynbio.9b00419.\u003c/li\u003e\n \u003cli\u003eNi J, Huang F, Wang B, Li B and Lin Q (2010) A planar PDMS micropump using in-contact minimized-leakage check valves. J Micromech Microeng 20(9):095033. DOI: 10.1088/0960-1317/20/9/095033.\u003c/li\u003e\n \u003cli\u003eNi J, Li B and Yang JG (2012) A pneumatic PDMS micropump with in-plane check valves for disposable microfluidic systems. Microelectronic Engineering 99:28-32. DOI: 10.1016/j.mee.2012.04.002.\u003c/li\u003e\n \u003cli\u003ePark D, Lee S and Min KT (2020) Techniques for investigating mitochondrial gene expression. BMB Reports 53(1):3-9. DOI: 10.5483/BMBRep.2020.53.1.272.\u003c/li\u003e\n \u003cli\u003eRoberts RW and Szostak JW (1997) RNA-peptide fusions for the \u003cem\u003ein vitro\u003c/em\u003e selection of peptides and proteins. Proceedings of the National Academy of Sciences USA 94(23):12297-12302. DOI: 10.1073/pnas.94.23.12297.\u003c/li\u003e\n \u003cli\u003eSarkar TR and Irudayaraj J (2008) Carboxyl-coated magnetic nanoparticles for mRNA isolation and extraction of supercoiled plasmid DNA. Analytical Biochemistry 379(1): 130-132. DOI: 10.1016/j.ab.2008.04.016.\u003c/li\u003e\n \u003cli\u003eSatterfield BC, Stern S, Caplan MR, Hukari KW and West JAA (2007) Microfluidic purification and preconcentration of mRNA by flow-through polymeric monolith. Analytical Chemistry 79(16): 6230-6235. DOI: 10.1021/ac0709201.\u003c/li\u003e\n \u003cli\u003eThompson AM, Gansen A, Paguirigan AL, Kreutz JE, Radich JP and Chiu DT (2014) Self-digitization microfluidic chip for absolute quantification of mRNA in single cells. Analytical Chemistry 86(24):12308-12314. DOI: 10.1021/ac5035924.\u003c/li\u003e\n \u003cli\u003eToriello NM, Douglas ES, Thaitrong N, Hsiao SC, Francis MB, Bertozzi CR and Mathies RA (2008) Integrated microfluidic bioprocessor for single-cell gene expression analysis. Proc Natl Acad Sci USA 105(51):20173-20178. DOI: 10.1073/pnas.0806355106.\u003c/li\u003e\n \u003cli\u003eTsai YC, Patil PJ, Lin CN, Chen WJ, Hsu KF, Shan YS, Yao DJ, Ho CM and Lee GB (2025) Optimization of drug-dosage combination of cancer therapy for ovarian cancer and cholangiocarcinoma by using complex system response technology. 2025 IEEE 4th International Conference on Micro/Nano Sensors for AI, Healthcare, and Robotics (NSENS).\u003c/li\u003e\n \u003cli\u003eWang CH, Shao YS, Hsu KF and Lee GB (2024) Detection of dual-methylated BRCA1/BRCA2 cell-free DNA for ovarian cancer on an aptamer-based integrated microfluidic system. Chemical Engineering Journal 495:153478. DOI: 10.1016/j.cej.2024.153478.\u003c/li\u003e\n \u003cli\u003eWang W, Jiang Q, Argentini M, Cornu D, Gigant B, Knossow M and Wang C (2012) Kif2C minimal functional domain has unusual nucleotide binding properties that are adapted to microtubule depolymerization. Journal of Biological Chemistry 287(18):15143-15153. DOI: 10.1074/jbc.M111.317859.\u003c/li\u003e\n \u003cli\u003eWeiss A, Ding X, van Beijnum JR, Wong I, Wong TJ, Berndsen RH, et al. (2015) Rapid optimization of drug combinations for the optimal angiostatic treatment of cancer. Angiogenesis 18(3):233-244. DOI: 10.1007/s10456-015-9462-9.\u003c/li\u003e\n \u003cli\u003eWhite AK, VanInsberghe M, Petriv OI, Hamidi M, Sikorski D, Marra MA, Piret J, Aparicio S and Hansen CL (2011) High-throughput microfluidic single-cell RT-qPCR. Proceedings of the National Academy of Sciences 108(34):13999-14004. DOI: 10.1073/pnas.1019446108.\u003c/li\u003e\n \u003cli\u003eWong PK, Yu F, Shahangian A, Cheng G, Sun R and Ho CM (2008) Closed-loop control of cellular functions using combinatory drugs guided by a stochastic search algorithm. Proc Natl Acad Sci USA 105(13):5105-5110. DOI: 10.1073/pnas.0800823105.\u003c/li\u003e\n \u003cli\u003eYu T, Tang C, Zhang Y, Zhang R and Yan W (2017) Microfluidics-based digital quantitative PCR for single-cell small RNA quantification. Biol Reprod 97(3): 490-496. DOI: 10.1093/biolre/iox102.\u003c/li\u003e\n \u003cli\u003eZeng Y, Woolley M, Chockalingam K, Thomas B, Arora S, Hook M and Chen Z (2023) Click display: a rapid and efficient in vitro protein display method for directed evolution. Nucleic Acids Research 51(16): e89-e89. DOI: 10.1093/nar/gkad643.\u003c/li\u003e\n \u003cli\u003eZuo X, Meng P, Bao Y, Tao C, Wang Y, Liu X, Bu Y and Zhu J (2023) Cell cycle dysregulation with overexpression of KIF2C/MCAK is a critical event in nasopharyngeal carcinoma. Genes \u0026amp; Diseases 10(1):212-227. DOI: 10.1016/j.gendis.2021.05.003.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"microfluidics-and-nanofluidics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mano","sideBox":"Learn more about [Microfluidics and Nanofluidics](http://link.springer.com/journal/10404)","snPcode":"10404","submissionUrl":"https://submission.nature.com/new-submission/10404/3","title":"Microfluidics and Nanofluidics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"complex system response, microfluidics, non-poly(A) mRNA extraction, qRT-PCR, thermal release, optimization","lastPublishedDoi":"10.21203/rs.3.rs-9465529/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9465529/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eNon-poly(A) messenger ribonucleic acid (mRNA) purification remains a critical bottleneck for cell-free translation platforms like \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003etr\u003c/span\u003eanscription-translation coupled with the \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ea\u003c/span\u003essociation of \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003ep\u003c/span\u003euromycin linker (TRAP) display for peptide screening, where sub-optimal thermal release can compromise results. Herein, we demonstrate a \u0026ldquo;complex system response\u0026rdquo; approach for optimizing thermal elution from probe-coated magnetic beads. Using an orthogonal array composite design with two optimization rounds (17 three-factor\u0026thinsp;+\u0026thinsp;9 two-factor experiments), we systematically explored temperature (60\u0026ndash;95\u0026deg;C), time (5\u0026ndash;30 min), pH (7\u0026ndash;9), and their interactions, revealing an unanticipated optimum of 60\u0026deg;C for 23.5 min at pH 9 (vs. conventional protocols of 90\u0026deg;C for 10 min). These conditions were associated with a maximum release efficiency of 71\u0026thinsp;\u0026plusmn;\u0026thinsp;7% with 10 pM RNA, sufficient for downstream TRAP library construction.\u003c/p\u003e","manuscriptTitle":"Optimization of mRNA Thermal Release on an Integrated Microfluidic Platform by a Complex System Response Method","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-18 13:48:50","doi":"10.21203/rs.3.rs-9465529/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-17T01:19:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"52489060888903101122874953834197646139","date":"2026-05-12T17:01:10+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"109373537998923715862970891846324452146","date":"2026-05-11T16:33:06+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"182705106034822535432241304219880863091","date":"2026-05-11T09:46:24+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"176480183062610555411468379225249095997","date":"2026-05-07T15:54:12+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-05-07T09:39:31+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-05-07T09:20:46+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-05-05T14:03:17+00:00","index":"","fulltext":""},{"type":"submitted","content":"Microfluidics and Nanofluidics","date":"2026-04-20T01:46:48+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"microfluidics-and-nanofluidics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"mano","sideBox":"Learn more about [Microfluidics and Nanofluidics](http://link.springer.com/journal/10404)","snPcode":"10404","submissionUrl":"https://submission.nature.com/new-submission/10404/3","title":"Microfluidics and Nanofluidics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"ffaa32c1-e914-44cb-99c2-d539cd020968","owner":[],"postedDate":"May 18th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-17T01:19:55+00:00","index":29,"fulltext":""},{"type":"reviewerAgreed","content":"52489060888903101122874953834197646139","date":"2026-05-12T17:01:10+00:00","index":28,"fulltext":""},{"type":"reviewerAgreed","content":"109373537998923715862970891846324452146","date":"2026-05-11T16:33:06+00:00","index":25,"fulltext":""},{"type":"reviewerAgreed","content":"182705106034822535432241304219880863091","date":"2026-05-11T09:46:24+00:00","index":24,"fulltext":""},{"type":"reviewerAgreed","content":"176480183062610555411468379225249095997","date":"2026-05-07T15:54:12+00:00","index":22,"fulltext":""},{"type":"reviewersInvited","content":"15","date":"2026-05-07T09:39:31+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-05-07T09:20:46+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-05-05T14:03:17+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-18T13:48:50+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-18 13:48:50","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9465529","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9465529","identity":"rs-9465529","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}