Abstract
Microfluidic Platforms
Microfluidic Platform Combining Droplets and Magnetic Tweezers: Application to HER2 Expression in Cancer Diagnosis
With the progression of oncology biomedicine, greater numbers of assays are needed with a trend toward the reduction in clinical sample size. To obtain the greatest amount of information possible from biological samples, innovative technologies are being developed for rapid, multigene analysis from minute clinical sample material. In breast cancer, HER2 (human epidermal growth factor receptor 2) is a critical biomarker because its overexpression is typically associated with more aggressive clinical behavior. Consequently, numerous drugs have been developed precisely to target HER2 overexpressed cancer. Whereas quantitative PCR (qPCR) has been used to quantify HER2 expression levels for cancer classification, methods to increase sample throughput and multiplexing are still highly in demand.
Using microfluidics for gene detection is a promising approach with significant advantages, such as volume reduction, automation, and confinement. However, certain microfluidic designs are limited with respect to device and fabrication complexity, cost-effectiveness, and inefficient sampling procedures.
The authors describe a platform that utilizes droplet microfluidics, yet is simple and robust and maintains compatibility and interoperability with established qPCR protocols and workflows. A motorized pipetting arm deposits equally spaced sample droplets into a confined stream of oil (plug microfluidics). Such designs reduce contamination risk, and their sequential nature means that droplet identity can be established without additional identity tagging material. In addition, magnetic tweezers allow the transfer of functionalized magnetic particles between droplets to specifically capture and purify mRNA. This platform comprises a thermocycler in which the reverse transcription (RT) and subsequent PCR cycling can be performed. The entire protocol occurs in a train of confined droplets containing the required reagents and samples. These trains can be loaded sequentially and processed fully automated, reducing labor and the risks of manipulation and contamination errors with minimal samples and reagents. To account for potential sensitivity and stability differences, the final analysis step is performed using conventional qPCR equipment.
Calibration curves are obtained for TMP (an internal control gene with minimal expression variability) and HER2 at a 6 ng/µL to 60 ng/µL range. This is equivalent to the total mRNA of a single cell or 10–1000 copies of target mRNA per droplet (using typical breast cancer cell lines). Calibration curves are then obtained for MCF7 (HER2+ overexpressing) and SKBR3 (HER2 nonoverexpressing) cell lines. Calibration curves are observed to be linear in the investigated range (R2 > 0.96), which demonstrates that mRNA capture efficiency and RT reaction efficiency are not influenced by mRNA seeding quantity. Ct (threshold cycle) values are similar for TBP expression in both cell types; however, HER2 is 5 units smaller in MCF7, confirming that it overexpresses HER2. Further characterization then demonstrates that the novel droplet PCR technique experiences a performance similar to that of conventional PCR.
Finally, clinical validation is performed in a blind protocol, using 21 samples of RNA extracted from biopsies of breast cancer patients. Data from both droplet and conventional PCR methods are compared with predefined reference samples to obtain fold difference values. This provides a relative value to identify HER2 expression in relation to reference genes. The linear correlation between both methods yields a Pearson correlation coefficient of 0.84. Crucially, positive and negative HER2 samples are identically segregated. Thus, as a front-end device for preparing direct or preamplified cDNA or as a fully integrated device, including detection, this droplet microfluidic PCR device provides gains in cost, robustness, and simplicity of operation for a number of nucleic acid analysis applications, especially for diagnosis (Ferraro, D.; et al. Sci. Rep.
Enhanced and Differential Capture of Circulating Tumor Cells from Lung Cancer Patients by Microfluidic Assays Using Aptamer Cocktail
Liquid biopsy methods, such as collecting circulating tumor cells (CTCs), have risen in popularity due to their noninvasive nature, accessibility, and ability to repeatedly perform sampling on patients. As a means to characterize tumor heterogeneity, they have limitations because CTC abundance is very low (1 CTC per billion hematopoietic cells in peripheral blood at advanced stages), and current CTC capture designs give little consideration to cancer cells with different phenotypes. Anti-EpCAM (epithelial cell adhesion molecule)-based CTC enrichment, while being the most popular method, is unable to capture and diagnose CTCs with low EpCAM expression levels. Using additional antibodies may mitigate this to some extent, but the available antibodies for tumor-specific surface markers are considered the limiting factor. Aptamers, however, may expand CTC enrichment potential. Aptamer binding has recently been verified with cancer cell line samples and cultivated mouse tumors. They also have greater long-term stability, synthetic reproducibility, and convenience in modification. Crucially, aptamers can be identified using cell-SELEX without a priori knowledge of molecular targets.
The authors combine an aptamer cocktail with silicon nanowire substrates embedded on a microfluidic chip. This platform allows enhanced and differential capture of CTCs from non-small-cell lung cancer (NSCLC) patients. Starting from a panel of 10 aptamers generated through cell-SELEX from A549 cells, these are reduced to four sequences following evaluation on another five different NSCLC cell lines. Thereafter, the usage of multiple aptamers demonstrates enhanced capture efficiency, with dual- and triple-aptamer combinations each outperforming single-aptamer combinations. Following the optimization process, three cocktail combinations exhibit >50% capture performance across the five cell lines—cocktail A (Ap1 + Ap2), cocktail B (Ap1 + Ap2 + Ap4), and cocktail C (Ap1 + Ap3 + Ap4), along with single aptamer Ap1 (the most effective signal capture agent), are trialed on blood samples obtained from NSCLC patients.
Between NSCLC patients and healthy donors, statistically significant differences are observed. Among 36 healthy donors, a single false-positive cell is detected. The standard deviation and coefficient of variation for all 12 patients are considered low, which suggests the technique is highly reproducible. This aptamer method, compared with anti-EpCAM capture, shows greater CTC capture ability of the aptamer strategy. This aptamer capture strategy is also able to monitor changes in CTC numbers following various clinical treatment procedures that are validated using computed tomography imaging.
Thus, this is the first demonstration of enrichment and capture of CTCs from NSCLC patients using an aptamer-based microfluidic assay. Using different aptamer cocktails facilitates selective enrichment of CTC subpopulations and real-time characterization of CTC heterogeneity (Zhao, L.; et al. Small
Fully Automated and Colorimetric Foodborne Pathogen Detection on an Integrated Centrifugal Microfluidic Device
Foodborne pathogens are responsible for more than 600 million foodborne illnesses and 420,000 deaths globally per year. To detect contaminated food, precise, rapid, and user-friendly point-of-care diagnostics have been developed. Isothermal DNA amplification methods, such as loop-mediated isothermal amplification (LAMP), amplify target genes at fixed temperatures with high sensitivity and specificity using specially designed primers and strand-displacing DNA polymerases. To date, LAMP has been explored for analyzing pathogenic bacteria, infectious viruses, and malaria-carrying mosquitoes.
Recently, centrifugal microfluidics has been combined with molecular diagnostics for sample-in, answer-out molecular analysis of foodborne pathogens. Although some success has been attained, current detection methods have been found to be time-consuming, and multiplexing is not considered satisfactory.
In this study, an integrated centrifugal microfluidic device that multiplexes LAMP to identify food pathogens is described. This device uses features such as microbead-assisted DNA purification, isothermal DNA amplification by LAMP, and colorimetric detection (on a single lab-on-a-disc) for automated and rapid food pathogen identification.
Highly specific primers are used to detect Escherichia coli and its specificity compared with three other nontarget bacteria and blanks. Promisingly, the initial purple color changes to navy blue before turning to sky blue to yield positive identification. This coincides with the absorbance peak shifting from 570 to 640 nm. Positively identified samples also have a 10-fold higher A640 nm/A570 nm ratio, and a minimum of 2.7 × 104 cells/mL can be detected with this assay. Thereafter, it is extended to identify other foodborne pathogens. In addition, its advantages include handheld dimensions, no requirement for tubing and valves, and naked eye detection. These underscore the potential for point-of-care pathogen diagnosis of pathogens (Oh, S.; et al. Lab Chip
Diagnostics
Colorimetric Detection of Staphylococcus aureus–Contaminated Solutions without Purification
Detection of the dangerous pathogen Staphylococcus aureus relies on agar culture that can take as long as 72 h. This delay in sample processing presents a potential public hazard because it means a delay in administering appropriate therapeutics. Colorimetric assays such as enzyme-linked immunosorbent assay, oligonucleotide embedded agar, PCR, and bacteriophages, have been used as alternatives. These methods, which require culture, can be time-consuming, laborious, and dependent on sophisticated laboratory instruments. This makes portable, point-of-care diagnostics a highly attractive alternative.
Building on a previous in vitro assay containing oligonucleotide probes, the authors developed a colorimetric assay that does not require instrumentation and is amenable for remote location use. The authors hypothesize that released micrococcal nuclease (MN) from S. aureus would cleave the previously described oligonucleotides functionalized on gold nanoparticles (AuNPs) and cause them to aggregate. AuNP aggregation would cause a color change from red to purple and become a sensitive assay detectable using the naked eye. This assay can also be stabilized with trehalose, stored as a lyophilized powder and reconstituted in solution.
MN cleavage of the oligonucleotide is first demonstrated (red to purple color change, shift in absorbance from 530 to 570 nm) and optimized. Its specificity is demonstrated when RNase addition is shown to inhibit its activity. The assay limit of detection is next characterized and measured using the ratio of ultraviolet signal at 570 to 530 nm. As a criterion for significant peak shift, the absorbance peak would need to shift to at least 550 nm, and the 570/530 nm ratio would need to be more than 0.5 and produce a visible color change. Promisingly, the limit of detection is found to be between 0.04 and 0.05 µM of MN. Remarkably, this is 10-fold lower than the (previous) fluorescent probe version (0.496 µM). The authors surmise that this is due to a small number of oligonucleotide strands needing to be cleaved for this aggregation event to occur. Between 0.04 and 0.496 µM, the solution changes from pink to light purple, but higher concentrations (9.917 µM) cause the AuNPs to aggregate rapidly, precipitate out, and coat the tube surface. Intermediate concentrations of MN (0.05 µM) lead to slower aggregation and develop a deep purple color.
Further tests suggest that nonexpert users could identify S. aureus concentrations below the limit of detection (LOD) from those above it through naked-eye detection, and that the assay works within reasonable probe concentrations of 0.25–1 nM at a reasonable pH range of 4–10. To replicate real-world conditions, the MN probe is first lyophilized, stored as dry powder, and tested in creek and ocean water spiked with S. aureus. In ocean water, the LOD decreases by almost 100-fold, which is likely due to salinity and other contaminants obscuring detection ability. Finally, probe performance is validated against non-MN-bearing bacteria and knockout variants of S. aureus. Thus, the authors show that an oligonucleotide-AuNP can serve as a rapid, convenient colorimetric assay for S. aureus detection based on particle aggregation. This has the potential to be used in remote, low-resource location water quality analysis and is sufficiently versatile to detect other bacteria of interest (Tiet, P.; et al. Bioconjug. Chem.
Rapid and Selective Detection of Pathogenic Bacteria in Bloodstream Infections with Aptamer-Based Recognition
Bacterial bloodstream infections (BSIs) can be potentially life threatening and are associated with one-third of global mortality. Low infectious doses (~100 cells/mL) make it a challenge to detect such pathogens from unprocessed blood samples. The current gold standard of bloodstream infection diagnosis involves incubation of patient blood for 3–5 days and monitoring of bacterial growth through metabolite production and a further 12 h culture on solid media for definitive identification. Delay in its diagnosis can be fatal, as every hour of delay in administering effective antibiotics leads to a 7.6% decrease in survival rate. Thus, there is an urgent need for selective and rapid pathogen diagnosis from blood samples.
Efforts to develop such diagnostic means have a number of limitations, such as the requirement for tedious work (cytolysis, nucleic acid extraction, etc.) and expensive equipment that restricts widespread usage. The authors propose a simple capture platform comprising mesoporous TiO2-coated magnetic nanoparticles functionalized with aptamers to isolate bacteria from blood to rapidly detect pathogens. This platform integrates bacteria recognition, capture, and enrichment. The TiO2 nanoparticles are incubated with fresh blood from patients, which allows the aptamer to bind with target bacteria. The captured bacteria are concentrated using a magnetic field before confirmation and identification. The chosen aptamer demonstrates good specificity by binding well with Staphylococcus aureus, whereas it does not bind well with Escherichia coli. The aptamer captured >89% of the microbial population.
In human blood samples containing 105 CFU/mL, the nanoparticles capture and separate S. aureus in the presence of a magnetic field (2 min exposure using a bar magnet). Agar culture shows a significant reduction in the colony-forming potential of the purified blood, which is quantified to approximately 83%. On the other hand, using E. coli targeting aptamers has negligible capture efficiency.
Thereafter, the capture efficiency of this platform is evaluated over multiple concentrations, ranging from low (10 CFU/mL) to high (2000 CFU/mL). Effective capture is further demonstrated for E. coli by using different aptamers, which achieve a high enrichment efficiency of ~80% even for low microbial concentrations. Finally, time-to-detection (TTD) studies show that 2 h is sufficient to obtain a final capture efficiency of 80%. In contrast, it takes traditional bacteria culture 8 h to achieve the same colony numbers from plate counting. This highlights the ability to shorten the diagnosis process required to reach definitive bacterial identification.
In summary, the authors demonstrate a highly versatile bacteria capture and detection platform using aptamers combined with magnetic nanoparticles. It achieves high selectivity and binding affinity without compromising bacteria viability and maintains good enrichment efficiency across a range of bacteria concentrations with rapid enrichment and separation (2 h). Efficient capture and detection can facilitate the early diagnosis and administration of effective therapeutics for BSI (Shen, H.; et al. ACS Appl. Mater. Interfaces
IRIDICA BAC BSI Assay: Rapid, Sensitive, and Culture-Independent Identification of Bacteria and Candida in Blood
The problem of sepsis (bloodstream infections [BSIs]) is increasing with the number of elderly and immunocompromised, and the increased use of implanted medical devices. The main factor that determines clinical outcome and financial burden of BSI is prompt treatment using appropriate antibiotics. Mortality risk doubles with a 24 h delay for bacteremia and every 12 h for candidemia. However, current culture-based methods require incubation times of up to 5 days to identify the bacteria or fungi species causing BSI. To circumvent these issues, broad-spectrum antibiotics are often inappropriately administered, or the causal agents misidentified using biochemical analysis.
To fulfill this unmet need for rapid, sensitive molecular detection methods of BSI agents from whole blood, the authors have developed a method involving lysis and DNA extraction from a 5 mL blood sample. This is paired with conserved-site PCR primers to amplify targets from >95% of eubacteria and Candida species that cause BSI. This method is also sufficiently compatible with the high concentrations of background human DNA and capable of providing identification within 8 h of sample collection. The generated amplicons are then prepared for mass spectrometry to discriminate sequence variants from one or more different species present in the sample. Limit of detection (LOD) studies are first performed for four core organisms: Candida albicans, Staphylococcus aureus, Enterococcus faecium, and Klebsiella pneumoniae. A LOD range of 0.25–128 CFU/mL suggests that bacteria can be detected directly from blood. In total, the assay successfully detects 47 additional bacteria and Candida species.
The lack of assay cross-reactivity is reported when samples containing 105 CFU/mL of bacteria, fungi, and viruses not designed for the assay to report are assessed. From a list of 20 potentially interfering substances, bilirubin is found to interfere with the assay. From a host of different microbial infections, at least one of two organisms could be identified in 18/18 instances, and positive identification for both organisms is achieved in 15/18 instances. The assay is also found to be reproducible from three operators using three lots of assay strips using three different instruments over 5 days.
Thereafter, the IRIDICA BAC BSI assay is compared with blood culture. While culture identifies 45 infectious organisms, the assay identifies 85 organisms from 285 clinical blood samples. Because the culture and BSI assays use single reagents for detection, the presence of a detected organism masks the other organisms through competitive interference. In culture, faster organisms outcompete slower ones. In PCR-based detection (such as this BSI assay), targets with better matches to primer pairs may outcompete other species during amplification, which obscures samples with at least one positive analyte. Among the samples, 207 yielded negative results by both methods. The IRIDICA BAC BSI assay matches 32 of 40 detections reported by culture (80% positive agreement) and detects an additional 34 in culture-negative samples. The majority of the BSI assay-positive, culture-negative identified species are organisms commonly associated with BSI. Some of these also include clinically relevant pathogens that are difficult to grow and common contaminants that demonstrate their comparative advantage over culture detection methods.
Since the IRIDICA BAC BSI assay cannot determine antibiotic resistance phenotypes, it should be considered an additional tool in a diagnostic regime, as opposed to a complete replacement for culture-based methods. The numbers of BSI assay–positive, culture-negative identified pathogens lend further weight to the lack of detection sensitivity of culture-based methods. Thus, this BSI assay demonstrates sensitive and specific detection of one or more species responsible for BSI, even with high background levels of human DNA, over a relatively short period of time (Metzgar, D.; et al. PLoS One
Culture Systems, Disease Modeling, and Drug Screening
Interstitial Flow Regulates the Angiogenic Response and Phenotype of Endothelial Cells in a 3D Culture Model
Fluid flow and mechanical stress are intimately linked with physiological and pathological phenomena in blood vessels. Recent microfabricated three-dimensional (3D) systems have shown that mechanical stimuli regulate vascular morphogenesis. Using such culture systems, mechanical stimuli, and vascular morphogens, blood vessel development can be studied under scenarios that better mimic native behavior. In this report, a platform is described that allows delineation of the role interstitial flow (IF) plays during blood vessel initiation (angiogenesis).
The authors demonstrate that IF directed against the vessel sprouting direction synergizes with pro-angiogenic factors (vascular endothelial growth factor [VEGF] and sphingosine-1-phosphate [S1P]) to facilitate vascular patterning in a multichannel system containing fibrin matrix, endothelial cells, and fibroblasts. This allows more reliable mimicry of endothelial mechanosensing during pathological neovascularization.
On the other hand, reversing the IF flow direction drastically attenuates vessel sprout formation, and clearly demonstrates how different flow regimes affect sprouting behavior. Adding various vascular morphogens (VEGF and S1P) also reacts synergistically with IF flow in 3D endothelial cultures to influence vessel sprouting behavior. Having demonstrated efficient modeling of the vascular morphogenic process, the system is further used to screen antiangiogenic drugs.
Using such systems that combine mechanical forces, morphogens and 3D coculture allow greater control over blood vessel growth to be achieved in vitro to better mimic vascularization processes in the tumor microenvironment. This facilitates identification of novel drugs and molecular mechanisms relating to angiogenic sprouting (Kim, S.; et al. Lab Chip
Drug Testing and Flow Cytometry Analysis on a Large Number of Uniform-Sized Tumor Spheroids Using a Microfluidic Device
Three-dimensional (3D) cell cultures models have gained interest in recent years for cancer modeling due to their greater similarity to native tissue compared with two-dimensional (2D) cell models. These cell colony clusters develop gradients of nutrients, metabolites, catabolites, biological signals, and oxygen that mimic native solid tumors. Unsurprisingly, this gives rise to heterogeneous cell populations. Various methods to model tumor behavior have utilized cell spheroids in microfluidics and applied various fluorescent dyes to study cancer cell characteristics. To date, each method has its own limitations.
These authors report an integrated microfluidic and flow cytometry approach for high-throughput drug testing and analysis on cancer cell spheroids. A two-layered microfluidic device is used to generate uniform-sized 3D spheroids with various channels for medium delivery and exchange. Following treatment, the spheroids can be harvested readily and dissociated for flow cytometry analysis. HepG2 cell spheroids (3%–6% variation) spontaneously aggregate inside the microfluidic device within 24 h of culture. Flow cytometry analysis suggests that >92% of the cells are alive after 72 h of microfluidic culture.
In control 2D petri dish culture formats, at least 30% of cells are observed to be nonviable following treatment using anticancer chemicals (cisplatin, resveratrol, and tirapazamine [TPZ]). In contrast, smaller 3D tumor spheroid cultures lead to a greater proportion of early apoptotic cells following cisplatin treatment. Cytotoxicity is less pronounced with resveratrol and TPZ treatment. The 3D spheroids are also shown to have 50% more viable cells than 2D cultures. In contrast, larger cancer cell spheroids exhibit greater drug resistance where more cells appear viable. A combination of the three drugs leads to higher cell death in 2D culture. This drug combination, however, does not significantly increase cytotoxicity in the 3D spheroid cultures, with most cells expressing early rather than late apoptosis features.
Using an integrated drug testing and analysis platform, the authors demonstrate the importance of drug testing on 3D spheroid cancer cell models. Compared with 2D cultures, the 3D models exhibit different proportions of early and late apoptosis and cell viability. In particular, cells in the 3D configuration are more resistant to drugs than their 2D counterparts. Different drug combinations and spheroid sizes are also critical to cancer cell response (Patra, B.; et al. Sci. Rep.
Biomechanically Primed Liver Microtumor Array as a High-Throughput Mechanopharmacological Screening Platform for Stroma-Reprogrammed Combinatorial Therapy
Tumors are highly complex and heterogeneous tissue. Many solid tumors (breast, pancreas, liver, lung, etc.) have increased stromal stiffness and abundant extracellular matrix (ECM) constituents that amplify pro-tumorigenic responses. Because this stromal barrier can obstruct drug infiltration, this has led to the development of therapeutics to deactivate tumor stroma. While vitamin D receptor (VDR) ligand therapy has been considered a stroma deactivation adjuvant, its effects have been limited.
In this report, different combinations of VDR ligands and chemotherapy drugs are tested in vitro using bioluminescence cancer cells to assay viability. To create stromal tumor models, HepG2 cells are cocultured with the hepatic stellate cell (HSC) line LX-2. HSCs are seeded on these hydrogel substrates with a stiffness ranging from 130 to 1200 Pa. Hydrogels with greater stiffness produce a spread phenotype with intense actin fibers and higher proliferation rates, while softer gels yield more rounded cells. Thermoresponsive hydrogels also allow ease of cell harvesting when the temperature is changed from 37 to 4 °C.
Rigid substrates promote the expression of α-smooth muscle actin (SMA) and other ECM proteins that signify fibroblast activation. Another indicator is that the YES-associated protein (YAP) in rigid substrates accumulates in the nucleus, whereas softer substrates generate cells with cytoplasmic YAP expression. Other target genes, such as CYP24a1, hepatic growth factor (HGF), and epidermal growth factor (EGF), are expressed at higher levels. When seeded in collagen substrates, greater contraction is also observed from the HSCs grown in stiffer substrates. This confirms greater activation in rigid substrates than in softer ones.
Using the thermoresponsive hydrogels, the authors harvest activated stromal fibroblasts and coculture them with HepG2 cells in gelatin microscaffolds to generate liver tumor models. Doxorubicin (Dox, red autofluorescence) is first utilized as a model drug, and its diffusion is examined in three different types of stroma tumor models—nonactivated microtumors (NµT), less activated microtumors (LµT), and activated microtumors (AµT)—with increasing degrees of cell activation.
Confocal microscopy allows assessment of the microtumors, which show a sevenfold difference in the number of Dox+ cells in the periphery compared with the core of AµT tumor models. In contrast, the number of Dox+ cells on the periphery versus the core does not differ as much for NµT. This is a clear demonstration of how activated stroma can retard cancer drug infiltration into tumors, whereas nonactivated stroma do not present the same barrier. In addition, HepG2 cells from AµT show higher proliferation and viability characteristics than LµT and NµT. The drug response of AµT is also markedly different, with a higher half maximal inhibitory concentration (IC50), indicating that drug resistance correlates with stromal activation. Increasing the seeding number of activated stromal cells further increases Dox resistance.
Thereafter, stroma-reprogramming treatment is performed to examine the influence of the VDR ligand, calcipotriol. Its addition leads to inhibited α-SMA, ECM proteins, and cytokines. Drug infiltration into AµT also markedly improves, achieving whole tumor penetration. However, addition of calcipotriol does not improve the drug response for LµT, NµT, and two-dimensional cultures, which further demonstrates the role activated stroma plays in cancer drug resistance.
To screen larger numbers of drugs, the authors looked at 15 generic chemotherapy drugs and found that calcipotriol enhances drug sensitivity in AµT models. Carboplatin and other platin class drugs are found to be particularly effective with calcipotriol coadministration, whereas no significant changes are observed when NµT models are used. This stroma coculture model is then transferred to mouse models that similarly demonstrate that AµT models have a growth advantage over NµT after 10 days of culture. Finally, the authors were able to demonstrate the accuracy of the earlier screening studies, identifying carboplatin and calcipotriol as the most effective drug combination for activated stroma tumors.
In closing, this screening platform imparts greater realism to cancer studies, as it recreates tumors with activated stroma that increase drug resistance. This can aid in the translation of novel mechanopharmacology therapeutics (Zhu, L.; et al. Biomaterials