Figure 1A schematically illustrates the microfluidic device for immobilization, culturing and in-situ impedance monitoring of single budding yeast cells. The microdevice consists of 5 layers, i.e., a glass substrate, Pt electrodes, a SiN_x passivation layer, SU-8 microfluidic channels and a polydimethylsiloxane (PDMS) cover. The SU-8 fluidic network comprises a main channel (width: 150 μm) and a parallel side channel (width: 300 μm), with 10 bottleneck-like traps (narrow neck width: 4 μm) connecting those two channels. Budding yeast cell suspension and culturing medium were infused into the main channel via their corresponding inlets (i.e., cell inlet and medium inlet, respectively), both at a flow rate of 1.5 μL/min. By precisely controlling the underpressure (about -6,000 Pa) applied to the pressure port, single yeast cells were reliably captured at the trap orifices. After cell immobilization, the pressure was raised to a constant value (about 2,000 Pa) to prevent redundant cells from being captured at the same traps. To integrate electrical impedance biosensor, Pt microelectrode pairs were patterned underneath the SU-8 structures, with the SiN_x passivation layer in between. To implement in-situ impedance measurement, long common electrode in the main channel serves as the stimulus electrode to supply an excitation alternating current (AC) voltage, and tip individual electrodes around trap orifices serve as recording electrodes to receive the weak response current, which is then amplified and converted into a detectable voltage signal. The SiN_x layer was reopened in the sensing region to reduce potential electric crosstalk between adjacent electrodes, and in the contact pads around the device border to connect electrodes to external circuits. Thus, the sensing unit, which features a trap for cell capture and retention and a pair of microelectrodes for in-situ electrical impedance measurement, is the core composition of the EIS-integrated microdevice.

To perform real-time monitoring of yeast dissection events by using in-situ electrical impedance biosensor, single cells are firstly immobilized at the traps and continuously cultured under laminar-perfused medium in the microfluidic device. During cell culturing for an extended time period, electrical impedance measurement is carried out every 2 min to monitor the dynamic cell process, such as budding, cytokinesis and daughter dissection. As soon as the cytokinesis-completed daughter cell is detached from its mother by hydraulic shear forces, it is immediately dragged away by the continuous medium flow, thereby suddenly cutting down the cell impedance between the stimulus and recording electrodes. This drastic decrease in impedance, corresponding to the momentary event of daughter dissection, can be detected by the time-lapse impedance measurement, showing a steep signal jump on the amplitude curve of EIS signal (Figure 1B).

The fabrication process of the microfluidic device (Figure 1A) is shortly described here. First, 200-nm-thick Pt with a 20-nm-thick TiW seed layer were patterned on a 4-inch glass wafer using a lift-off process. Then, a 500-nm-thick SiN_x layer was deposited on the wafer by plasma-enhanced chemical vapor deposition (PECVD), followed by reactive ion etching (RIE) to define the opening in sensing region and contact pads. Subsequently, 30-μm-thick SU-8 3,025 photoresist (MicroChem, United States) was spin-coated on the wafer with patterned microelectrodes, and structured into microfluidic channels by using standard photolithography. After wafer dicing, an unstructured PDMS (Sylgard 184, Dow Corning, United States) layer, punched with holes as fluidic inlets and outlets was used to seal the microfluidic channel network. For an irreversible bond between PDMS and SU-8, both SU-8 surface of each die and oxygen-plasma-activated PDMS cover were modified with 3-aminopropyltriethoxysilane (APTES, Merck KGaA, Germany) in aqueous solution (Zhu et al., 2016).

To achieve single-cell immobilization and real-time impedance measurement in a sensing unit, an experimental setup was designed and its details were presented in our previous work (Zhu et al., 2015). Briefly, a fabricated microfluidic device (Figure 2) was first clamped between a custom-made aluminum (Al) holder and a transparent polymethylmethacrylate (PMMA) cover by using screws. Then, a printed circuit board (PCB) soldered with spring probes was mounted on the PMMA cover to electrically connect microelectrodes with an impedance spectroscope (HF2IS, Zurich Instruments, Switzerland) and a current amplifier (HF2CA, Zurich Instruments, Switzerland). Two glass syringes, prefilled with cell suspension and culturing medium, were respectively connected to cell and medium inlets through polytetrafluoroethylene (PTFE) tubes for fluidic access. Both syringes were affixed on a precision syringe pump (neMESYS, Cetoni, Germany) for sample delivery. A pressure controller (OB1 MK3+, Elveflow, France) was used to adjust the underpressure applied to the pressure port for cell immobilization. For time-lapse imaging, the assembled setup, including the Al holder, microfluidic device, PMMA cover and PCB was placed on the stage of an inverted microscope (FV3000, Olympus, Japan).

For time-lapse EIS measurement, a swept-frequency AC voltage V s t i ∗ (10 kHz–10 MHz, V_p: 1 V) from the impedance spectroscope was applied to the stimulus electrode at an interval of 2 min, and the response current I r e s ∗ obtained from the corresponding recording electrode was amplified and converted into a voltage signal V r e c ∗ using the current amplifier, and ultimately recorded by the impedance spectroscope. The relationship among the measured impedance Z ∗ , the response current I r e s ∗ and the recorded voltage V r e c ∗ can be expressed by following equations: Z ∗ = V s t i ∗ / I r e s ∗ , V r e c ∗ = G · I r e s ∗ , and Z ∗ = G · V s t i ∗ / V r e c ∗ . Accordingly, the measured impedance Z ∗ is inversely proportional to the response current I r e s ∗ ( Z ∗ ∝ 1 / I r e s ∗ ), and also inversely proportional to the recorded voltage V r e c ∗ ( Z ∗ ∝ 1 / I r e c ∗ and | Z ∗ | = G / A , where A is the amplitude of V r e c ∗ ). Thus, the EIS amplitude A can be directly used to characterize the impedance variation induced by the immobilized budding yeast cells.

A budding yeast (S. cerevisiae) cell strain, derived from BY4743, was used in the experimental verification of the EIS-biosensor-integrated microfluidic device. Yeast cells were cultured using standard methods as described in literature (Sherman, 2002). Liquid cultures of yeast were grown at 30°C in a synthetic complete (SC) medium, which contained 0.17% yeast nitrogen base without amino acids or ammonium sulfate, 2% glucose, and a complement of amino acids and nucleotides. 0.05% w/v Pluronic^® F127 solution was added in the SC medium to prevent hydrophobic components from sticking together or accumulating on the channel surface. Prior to cell filling in the glass syringe, the suspension of yeast cells was diluted to reach a concentration of approximately 1 × 10⁶ cells/mL in SC medium. All chemicals were purchased from Merck, Germany.

Figure 3 shows the experimental results of single-cell trapping and continuous culturing. Single budding yeast cells were initially captured at the traps, and were reliably retained for bud growth and daughter dissection under the continuous perfusion of cell-culturing medium. Due to the size difference between yeast mother cells and buds, buds usually grew inside the trap or rotated into the trap for subsequent growth by hydrodynamic forces. After the completion of cell division and cytokinesis, daughter cells were ultimately dissected by fluid shear forces and washed away with the flowing medium. The automatic detachment of a yeast daughter cell under the hydraulic shear forces was known as a daughter-dissection event. For instance, the mother cell in Figure 3A sprouted at 10 min, its bud gradually grew to be a mature daughter, and finally the daughter cell was successfully removed at 90 min.

New buds of yeast cells may start to sprout when the previous daughter cells are not completely detached from mother cells in liquid culture. As shown in Figure 3B, both bud 1 and bud 2 started to grow when the former daughters were not dissected. After the mature progenies were automatically removed by the fluid flow (at 40 and 150 min), newborn buds were successfully rotated into the narrow orifice of the trap. In this case, cell rotation could be attributed to two reasons: i) the high flow velocity distributed in the narrow orifice exerting sufficient drag forces on buds from the main channel towards the side channel and ii) the high channel height providing adequate space in vertical direction for cell rotation. Therefore, the bud rotation and reorientation towards the side channel is of benefit to coincident and efficient daughter-cell dissections.

During the continuous cell culturing, in-situ EIS measurement was performed every 2 min to monitor yeast budding process in a label-free and real-time manner. For the wide-band EIS, electric current at low frequencies (typically below 1 MHz) flows mostly around cells and very few penetrates cell membrane. Thus, EIS signal at 1 MHz can be directly used to characterize cell size or morphology, which has been validated previously (Zhu et al., 2015). In Figure 4A, raw signal amplitude at 1 MHz shows 4 remarkable jumps when 4 daughters were dissected and washed away from the traps in sequence. Such activities of daughter dissection cause abrupt reduction of cell size or volume, which can be sensitively recorded by the real-time EIS measurement.

The recorded raw EIS signal, referring to the instantaneous impedance of yeast cell and its surrounding medium, may exhibit drifts along the recording time period. They could be attributed to bud swing, cell reorientation, or dirt attachment in the sensing unit during cell culturing. To compensate signal drifts and accurately identify the momentary events of daughters separating from their mother, EIS amplitude at time point t , A t , was divided by the amplitude value at the previous time point, A t − 1 , thereby yielding a “Dissection Index (DI)” ( D I = A t / A t − 1 ) curve (Figure 4B). During yeast budding, EIS signal varied slowly and slightly, resulting in the ratio DI approaching the baseline. Since the signal drifts were relatively low in amplitude, they were also normalized around the baseline. This simple normalization method could thus compensate potential drifts of EIS signal. In comparison, the sudden removal of a daughter from its immobilized mother cell at time point t caused the recorded signal amplitude A t to abruptly jump to a value much higher than that of A t − 1 , thus leading to a conspicuous peak on the DI curve. Moreover, by using K-means cluster analysis, data points on the DI curve were adaptively classified into two categories, i.e., the baseline cluster centered around 0.9998 with minor signal fluctuations, and the peak cluster centered around 1.0167 indicating the occurrence of daughter dissections. Therefore, the moments, when daughters were detached from the mother cell in sequence, were accurately extracted from the recorded EIS signals, and 4 daughter-dissection events were highlighted as 4 sharp spikes on the DI curve.