Characterization of a Centrifugal Microfluidic Orthogonal Flow Platform

Abstract

To bring to bear the power of centrifugal microfluidics on vertical flow immunoassays, control of flow orthogonally through nanoporous membranes is essential. The on-disc approach described here leverages the rapid print-cut-laminate (PCL) disc fabrication and prototyping method to create a permanent seal between disc materials and embedded nanoporous membranes. Rotational forces drive fluid flow, replacing capillary action, and complex pneumatic pumping systems. Adjacent microfluidic features form a flow path that directs fluid orthogonally (vertically) through these embedded membranes during assay execution. This method for membrane incorporation circumvents the need for solvents (e.g., acetone) to create the membrane-disc bond and sidesteps issues related to undesirable bypass flow. In other recently published work, we described an orthogonal flow (OF) platform that exploited embedded membranes for automation of enzyme-linked immunosorbent assays (ELISAs). Here, we more fully characterize flow patterns and cellulosic membrane behavior within the centrifugal orthogonal flow (cOF) format. Specifically, high-speed videography studies demonstrate that sample volume, membrane pore size, and ionic composition of the sample matrix significantly impact membrane behavior, and consequently fluid drainage profiles, especially when cellulosic membranes are used. Finally, prototype discs are used to demonstrate proof-of-principle for sandwich-type antigen capture and immunodetection within the cOF system.

Keywords: centrifugal; decay constant; embedded membrane; exponential decay; membrane deswelling; microfluidic; orthogonal flow.

Figure 1

Schematic overview of a the putative cOF disc and a photograph of a single test domain on a fully assembled cOF disc. Diagrammatic exploded view of a six-layer (5 + 1) microfluidic disc design comprised of one polymethyl methacrylate (PMMA) layer and five polyethylene terephthalate (PET) layers. Layers 2 and 4 serve as the primary fluidic layers. Layer 3 functions a flow through or via layer. Circular cutouts (4- or 5-mm Ø) of porous membranes were placed into cutout recesses in layer 3. Upon lamination, HSA coated layers 2 and 4 bond to the membrane, anchoring it in place.

Figure 2

Proof-of-principle dye studies to demonstrate feasibility of disc-based orthogonal flow through porous membranes. Small aliquots (13 µL) of erioglaucine-spiked artificial blood plasma (ABP) were added to each sample chamber to permit visualization of flow pattern through the membrane. Rotational forces were used to pump the fluid aliquots through membranes comprised of (A) PVDF and (B) nitrocellulose. Fluid intrusion and flow through began at 500 and 4000 rpm, (A) and (B), respectively. Complete flow through was observed in 30–45 s. This figure features representative AFTER images of the (A) PVDF and (B) nitrocellulose membranes.

Figure 3

Exponential decay curves depicting the remaining fluid column height in the sample chamber ($\Delta r$) as a function of elapsed time. Assay buffer (200 µL) was loaded into each sample chamber (Figure 1 inset). Rotational frequency (rpm) was held constant for each fluid drainage trial (n = 4 for each membrane type-rpm pairing). A high-speed, stroboscopic video system was used to visualize incremental changes in the height of the fluid column over time. Plots (A–F) depict mean elapsed time (s) vs. the remaining fluid column height in the sample chamber (mm). Colored points represent mean elapsed time for individual trials at each rotational frequency. Solid colored lines represent exponential best fit curves for mean elapsed time values. The top row, plots (A–C), reflect flow profiles for NC membranes with 0.2 μm pore size. Plots (D,E) (bottom row) reflect drainage profiles for NC membranes with 0.45 μm pore size. Pore size for the Sartorius Unisart membrane was unknown. Note: Rotational frequencies for Sartorius Unisart membranes (panel (F) and (F)-inset) were much lower than other membranes (A–E).

Figure 4

Exponential decay curves depicting the remaining fluid column height in the sample chamber ($\Delta r$) as a function of elapsed time. To assess the impact of differing aqueous sample matrices, 4 mm cutouts of BioRAD 0.2 µm pore size membranes were embedded into each cOF disc. Aliquots of each sample solution (200 µL volumes of artificial urine, assay buffer, and artificial blood plasma) were loaded into the sample chambers (Figure 1 inset). Rotational frequency (rpm) was held constant for each fluid drainage trial (n = 4 for each fluid type-rpm pairing). Note: Y-axis for plot C is larger to capture the full extent of the 750 rpm trials. Plots (A–C) depict mean elapsed time (s) vs. the remaining fluid column height in the sample chamber (mm). Plots (D–F) demonstrate how those exponential decay curves can be linearized via semilog transformation, i.e., x-axis was transformed to Log10 of the remaining fluid column height.

Figure 5

On-disc flowthrough of large volume body fluid samples. (A) The cOF disc was redesigned with 4 mm OF port (same radial distance from CoR) and larger chambers, i.e., thicker 1.5 mm PMMA accessory pieces were affixed to the top and bottom of the disc. At 3000 rpm, this cOF design was able to process 1.2 mL of assay buffer (~3 min). Likewise, at 3000 rpm this design processed 1.5 mL of neat urine (<60 s). The bulk fluid was removed from the recovery chamber and two additional 1.5 mL aliquots of urine were passed through the same cOF membrane (total urine volume = 4.5 mL). (B) The cOF disc was redesigned with a 4 mm OF port (same radial distance from CoR) and a larger waste chamber. At 3000 rpm, this cOF design processed 200 µL of human serum in ~120 s. Three additional aliquots of serum (200 µL ea) were passed through the same membrane (total serum volume = 800 µL).

Figure 6

On-disc immunodetection of Ebola virus-like particles (bottom view). The cOF disc was redesigned with circular 2 mm port in disc layer 5 (directly below the OF fluidic port). The exposed cOF membrane was spiked with 5 µL of a primary capture antibody solution [10 ug/mL]. After drying (30 min@30 °C), each cOF access port was sealed with a 6 mm diameter PeT coverlet. Positive samples consisted of 40 µL of Ebola virus-like particle antigen [10 µg/mL], 5 µL Au tagged mAb, and 5 µL assay buffer. Negative samples consisted of 5 µL Au tagged mAb and 45 µL assay buffer. Spin protocol: 15 s spin cycles at 250 rpm intervals beginning at 1000 and ending at 3000 rpm.