Micro-Nano Technology for Genomics and Proteomics BioMEMs - Ozkan

344 JIM V. ZOVAL AND M.J. MADOU

Dyes	Sample

Optical

Viewing

Window

Valve 1

Valve 2

FIGURE 10.7. Microfluidic pattern for LIVE/DEAD R BacLightTM Bacterial Viability Assay. The dyes and sample are introduced into the reservoir chambers using a pipette. The dyes fill the chamber stopping at a capillary valve (valve 1). Similarly, the sample containing cells is introduced into the sample reservoir. The disc is rotated to a velocity of 800 rpm, the dyes are forced through the capillary valves and they are mixed as they flow through the switchback turns of the microfluidic channels. Simultaneously, the sample passes from the reservoir into a fluid channel where it meets the dye mixture at valve 2. The velocity of the disc is increased to 1600 rpm and the dye mixture and sample combine and mix in the switchback microfluidic path leading to the optical viewing window.

into a fluid channel where it met the dye mixture at valve 2 of Figure 10.7. The velocity of the disc was increased and the dye mixture and sample combine and mix in the switchback microfluidic path leading to the optical viewing window. The dye-sample mixture is allowed to incubate in the dark at room temperature for 5 minutes. The optical viewing chamber was imaged twice, once with optics for the green signal and then with optics for the red signal. A typical fluorescence microscopy image of an overlay of the red and green images of stained E. coli is shown in Figure 10.8.

The instrument for disc rotation and fluorescence imaging (Figure 10.9) used a programmable rotational motor for various velocities and acceleration/deceleration rates. The use of standard microscope objectives enabled the selection of magnification. An automatic focusing system was used. The light source was a mercury lamp, which used standard low-pass excitation filters for fluorescent excitation. A CCD camera was combined with standard emission filter cubes for imaging.

10.5.5. Automated Cell Lysis on a CD

In work done by our group [29], cell lysis was demonstrated on a microfluidic CD (Compact Disc) platform. In this purely mechanical lysis method, spherical particles (beads) embedded in the CD cause shear-induced disruption of mammalian cells, E. coli, and yeast. Interactions between beads and cells were generated in the rimming flow established inside a partially filled annulus chamber in the CD rotating around a horizontal axis of rotation. To maximize bead-cell interactions in the lysis chamber, the CD was spun forward and backwards at high acceleration for up to 5 to 7 minutes. For this novel lysis method we

CENTRIFUGE BASED FLUIDIC PLATFORMS

345

FIGURE 10.8. Fluorescent microscopy over-laid images of red (circled cells) and green (un-circled cells) stained E. coli on CD from LIVE/DEAD R BacLightTM Bacterial Viability Assay.

investigated which of the parameters such as bead density, angular velocity, acceleration rate, and solid volume fraction are the most important for efficient cell lysis. Cell disruption was verified either through direct microscopic viewing or measurement of the DNA concentration extracted upon lysis using the PicoGreen dsDNA Quantitation Kit (Molecular Probes, OR). Lysis efficiency was also compared with a conventional lysis protocol. In the long term, this work is geared towards CD-based sample to answer nucleic acid analysis which will perform cell lysis, DNA purification, DNA amplification, and detection.

Although extensive research on microfluidic devices[6] for biomedical and clinical diagnostic applications has been conducted in the past decade, a majority of the miniaturized devices developed has been limited to single purpose applications due to the complexity and difficulty of integrating various microfluidic components into a single small unit. Compared to other microfluidic technologies [2, 6, 12, 30] for moving small amounts of fluid or

FIGURE 10.9. Left: Optical disc drive/imager with cover removed. Size of unit is made to fit in specific cargo bay of Space-Lab. Right: Zoom of microscope objectives and a disc loaded in the drive.

346	JIM V. ZOVAL AND M.J. MADOU

suspended particles from site to site, a centrifuge-based system is well suited for various crucial microfluidic functions such as flow sequencing, cascade mixing, capillary metering and flow switching.[11, 31, 32] Those functions can be implemented through the exploitation of (1) centrifugal and coriolis forces induced by spinning the CD for sample propulsion, (2) capillary force due to interfacial tension for stopping the flow and (3) microfluidic designs. In this regard, a polymer based microfluidic CD platform is a highly promising approach, not only for the integration of multi-microfluidic functions for diagnostic applications but also as a platform enabling the automated, multiple parallel processing needs in high-throughput screening (HTS). In this paper, we investigate the achievability of mechanical cell lysis on a CD platform and compare its performance with a traditional lysis protocol.

There are many types of cell lysis methods used today that are based on mechanical [33], physico-chemical [34], chemical [35], and enzymatic [35] principles. The most commonly used methods in biology research labs rely on chemical and enzymatic principles. The main drawbacks of those procedures include intensive labor, adulteration of cell lysate, and the need for additional purification steps. In order to minimize the required steps for cell lysis, a rapid and reagentless cell lysis method [33] would be greatly appreciated especially in sample preparation on a microfluidic CD. This would also help avoid the necessity of storing chemicals in chambers and of intensive mixing which is generally considered difficult in the micro domain. In this work, a purely mechanical lysis method was investigated based on rapid granular shear flow [36, 37] in an annulus chamber which spins about a horizontal axis of rotation alternating between the forward and reverse directions. When the CD is at rest, spherical particles, such as glass beads, suspended in an aqueous medium containing the cells, lie in a pool at the bottom of the chamber. However, during spinning all particles are dragged up by the shear of the surrounding medium and uniformly coat the outer wall of the chamber in response to both shear and centrifugal forces as shown in Figure 10.10. This flow is often referred to as “rimming flow” [38, 39]. In order for a cell to be disrupted in a rapid granular flow, the cell should be brought into physical contact with the colliding particles. The bead interactions with the cells consist of two main types: impulsive contact (collision)

(a)

(b)

FIGURE 10.10. Flow patterns for two rotational states (a) At rest: fluidized beads are sedimented at the bottom of chamber (b) At spinning: two circumferential bands of beads and liquid are observed for a constant velocity.

CENTRIFUGE BASED FLUIDIC PLATFORMS

347

ω(t)

uT(t)

us(t) u(t) ro

u1

u2

UT(t)=roω(t)

FIGURE 10.11. Illustrative views of particle interactions (collision and friction) that are generated in colliding particles in acceleration of a CD. A tangential velocity profile of fluid in the chamber is depicted on the right and varies with radial locations and time. uT(t) represents tangential velocity of a particle and follows with the flow passing around it. us(t) indicates settling velocity under a centrifugal field.

due to the beads responding to the centrifugal field and sustained contact (friction) due to shearing.

We first discuss inter-particle forces involved in the rimming flow and identify those parameters that may have the largest contribution to the inter-particle forces and the particle interaction frequency. The influence of these parameters on cell lysis is then verified experimentally by microscopy or by measuring the double stranded DNA concentration extracted from the cells upon cell lysis.

Rapid granular rimming flow in a horizontally rotating annulus chamber features two types of particle interactions: collision and friction (grinding). In Fig. 10.11 we schematically depict these particle interactions generated during an acceleration period in a partiallyfilled lysis chamber. Collisions occur when suspended particles are forced to move outwards radially (vector us(t)) responding to the centrifugal force and impact upon particles closer to the outer wall. Friction between particles is generated in the process of rearranging the impacting particles into a homogeneous granule band along the outer wall. The impacting particles (vector u1) travel slower than the particles close to the chamber outer wall (vector u2) due to differential shear rates imposed on them. Particles accumulated against the chamber outer wall move almost at the same tangential velocity (see UT(t)) as the outer chamber wall as they more or less stick to it. It is actually the velocity differential between colliding particles moving in the same tangential direction that leads to the friction (grinding) between those particles.

The particles make a complete circle on a typical time scale of 0.1 sec at angular velocities ranging from 10 to 20 rev/sec. This velocity (vector uT(t)) is approximately three orders of magnitude larger than the rate of radial motion of particles (vector us(t)) i.e.,

348	JIM V. ZOVAL AND M.J. MADOU

particle settling velocity (see also below). On the basis of this consideration, one can expect that the suspended particles, upon spinning the CD, uniformly coat the chamber wall while undergoing both collision and friction.

We discuss the grinding force here as a compressive pressure Pg exerted between the particles in a shear flow. It was empirically found [40] that the compressive pressure is proportional to a shear stress τ which has the following relation

τ ρp · D2p · γ 2 · f (ν)

(10.12)

where ρp is particle density, Dp is particle diameter, γ is shear rate, and f (ν) is a function of the volumetric concentration of particles.

Frictional contact of particles depends on the shear stress τ which is proportional to the density of particles ρp and the square of local shear rate γ. This is similar to a plane Couette flow where the shear rate (dUT/dy) in steady-state is proportional to the angular velocity of the CD. During the steady-state many particles will, however, have already reached the mound of particles at the outer wall of the chamber, and therefore make no relative motions contributing to lysing. Hence, an acceleration period is more contributive to the shear rate than a steady state mode. Accordingly, higher acceleration rate will lead to larger magnitude of frictional contact of particles.

Most of the theories on particle collisions [41, 42] in granular flow are based on models describing the individual collisions of spheres such that the impact properties employed in the theories should adequately describe the average properties of the flow. However, the measurement of impact coefficients to test these theories is extremely challenging in the current application and is beyond the scope of this treatise. Therefore, we have attempted to evaluate the parameters associated with the particle collisions experimentally to predict the magnitude of collisions qualitatively.

In the work by Batchelor [43] and Zenit [44], a collisional particle pressure in a solidliquid mixture is represented as

Pc = ν · ρp · u¯2 · F(ν)

(10.13)

where ν is solid volume fraction (SVF), ρp is particle density, and u¯ is local mean particle velocity. Batchelor suggests the following relation for a function of solid volume fraction ν

F(ν) ≈	ν	1 −	ν		(10.14)
	νcp		νcp

where νcp is the closed-packed solid fraction (νcp ≈ 0.62) for a random packed bed of uniform sized particles. Here, considering a solid-liquid suspension in a centrifugal field, it is appropriate that we substitute terminal settling velocity ut with the local mean particle velocity u in Eqn. (10.13). From a balance of the centrifugal, buoyancy and drag force, the motion of a spherical particle is expressed as

du				ρ		CD
m		= mr ω2	− m		r ω2 −		u2 A p ρ	(10.15)
	dt			ρp		2

CENTRIFUGE BASED FLUIDIC PLATFORMS								349
which can be rewritten as
du		=	r ω2	ρp − ρ	−	CD A p ρ	u2	(10.16)
	dt
				ρp		2m

where ω is the angular frequency, r is distance of particle from the center of the disc, CD is drag coefficient, and Ap is projected area of particle. In Eqn. (10.16), u is the velocity of the particle relative to the fluid and is directed outwardly along a radius. As a particle travels, the drag quadratically increases with the velocity. Therefore, the particle acceleration quickly decreases and approaches zero. The particle then reaches a maximum constant velocity, terminal settling velocity, which is represented as

ut =		·	3·		CD·		ρ	−		(10.17)
	4		r ω2	Dp		(ρp			ρ)
				·		·

Since the particle Reynolds number, Rep = Dp ut ρ f η, is calculated to be slightly larger than 1 for the current experiments, falling into a transition region very close to Stokes regime, the drag coefficient [45] is found as

24

3

+ 0.34

(10.18)

CD =

+

Rep

Rep

Thus, the collisional pressure in Eqn. (10.13) can now be rewritten as

Pc =

3

·

CDp · ρp ρp − 1 · ω2

· ν F(ν)

(10.19)

4

r D

ρ

From equations 10.12 and 10.19, the important parameters for lysis appear to be acceleration rate, particle density, angular velocity, and solid volume fraction. The particle size is excluded from the important influencing parameters here because increasing particle size conversely creates larger interparticle voids, which will reduce the number of particle collisions with cells squeezed in-between (i.e., the effective impact frequency). The particle size is expected to be more of an important parameter for the effective impact frequency. Concerning the volume fraction of particles, it plays a binary role in both the magnitude of particle interactions and the effective impact frequency. If only the collisional pressure in Eqn. (10.13) is considered for the inter-particle force, the highest collisional pressure will lie at the SVF of 0.4 as shown in Fig. 10.12.

The CDs were fabricated using the Polydimethylsiloxane (PDMS) replication techniques described earlier [26]. For the purpose of mechanical cell disruption, an ultra-thick SU-8 process was developed to fabricate a mold featuring extra high structures (thickness of a 1mm) so that sufficiently high spaces could be formed in the PDMS for the interacting beads. The rotating CD platform was completed by sandwiching the micromachined PDMS sheet between two polycarbonate discs. Fig 10.13 shows a CD with an annulus chamber designed for cell lysis.

As shown in Fig. 10.14, a custom CD spin-stand is equipped with a motor and an amplifier/controller to allow for various rotational profiles such as RPM-specified rotation

350

Collisional Pressure (P)

JIM V. ZOVAL AND M.J. MADOU

Collisional Pressure (P = v*F(v))

0.1

0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
00	0.1	0.2	0.3	0.4	0.5	0.6	0.7
			Solid Volume Fraction (v)

FIGURE 10.12. Plot of collisional pressure with respect to solid volume fraction.

and precise positioning. A centrifuge vision system was used to view a sequence of color images of the area of interest on the CD in real time while it is running and store the captured frames on a computer.

Periodic granular rimming flow is induced in a partially filled circular chamber by spinning a CD around a horizontal axis of symmetry with high acceleration (e.g., 200 rev/sec2) and changing rotational directions with a tiny rest period in between (e.g., 150 msec) to

1 mm

FIGURE 10.13. Microfluidic view of a CD (a) CD with a circular chamber (1mL) (b) SEM photo showing the side view of inner wall of the chamber.

CENTRIFUGE BASED FLUIDIC PLATFORMS

351

FIGURE 10.14. Spin-stand and centrifuge vision system. the CD spin-stand is equipped with a motor (Pacific Scientific Servo Motor, 9000rpm) and an amplifier/controller (PAC SCI Programmable Servo Drive) to allow for various rotational profiles such as RPM-specified rotation and precise positioning. The servo drive uses a graphical user interface program, ToolPAC, to easily configure and program the motor to specific applications. A microfluidic CD is placed on the aluminum saddle coupled with the motor shaft. The digital video recording system is composed of: a camera (Basler A301bc, 640 × 480 pixels, 80 fps max., 10× zoom lens mounted), a strobe light (PerkinElmer MVS-4200, 6µs duration), and a retro-reflective sensor (Banner D10 Expert Fiber-Optic Sensor). The strobe light with a 100 Hz maximum repetition frequency is employed to reduce blurry images of a fast moving object.

accumulate a solid-liquid mixture at the bottom of the chamber. The CD makes 3 or 4 revolutions depending on the chosen angular velocity (e.g., 10–20 rev/sec) and returns to the original position with the same angular velocity and the same number of revolutions—a typical cycle profile is shown in Fig. 10.15. The CD then runs for hundreds of such cycles to ensure that a large number of effective particle collisions occur. The angular acceleration rate is so high that a circumferential particle band begins to be observed in one revolution corresponding to about 0.1 sec. Fig. 10.16 shows a sequence of flow patterns to illustrate how a solid-liquid mixture forms into rimming flow.

CHO-K1 (Clontech, CA) was used as a model cell line for mammalian cell lysis. The cell density was 100,000 cells/mL. Saccharomyces cerevisiae strain TMy16 (MATa trp1- 901 ura3-52 his3- 201 ade2-101 lys2-1 leu1-12 can1-100 GAL+) [46] and Escherichia coli

XL1-Blue (Stratagene, CA) were used for yeast and bacteria lysis. The cell concentration was measured at O.D600with a spectrophotometer.

352	JIM V. ZOVAL AND M.J. MADOU

V (rev/sec)

20

t (sec)

0.1	0.3	0.45
4 revolutions		pause

FIGURE 10.15. Schematic rotational profiles for one cycle: A CD ramps up to 20 rev/sec with 200 rev/sec2 making 4 revolutions and spins back to the original position.

(a)

(b)

(c)

(d)

FIGURE 10.16. Sequential images of flow of a mixture of Zirconia-Silica beads and blue-dyed liquid forming into circumferential bands (rimming flow): solid volume fraction of 1/3 and angular velocity of 20 rev/sec. The frames are captured with a continuous pulses generation at a rate of 60 Hz without using the fiber optical sensor.

CENTRIFUGE BASED FLUIDIC PLATFORMS

353

The protocol we compared our CD based lysis with will now be described. A volume of 0.3 mL of Yeast cells (O.D.600 = 2.0) was transferred to a microcentrifuge tube and centrifuged for 1 minute to pellet the cells. The supernatant was discarded and 200µL of lysis buffer (1%SDS, 1mM EDTA, 100mM NaCl, 10mM Tris (pH 8), 2% Triton X-100), phenol-chloroform:isoamyl alcohol and 0.3g of glass beads were added. Using a vortex placed in the refrigerator, we vortexed the microtube for 5 min. After vortexing we added 200µL of water, briefly mixed manually, spun the tube for 10 more minutes and finally transferred the top aqueous layer to a new microcentrifuge tube to be analyzed.

CHO-K1 cells were observed by a microscope and photographed before and after CD spinning to verify the cell disruption. For E. coli and yeast, the amount of DNA extracted from lysis was measured by PicoGreen Quantitation Kit (Molecular Probes, Eugene, OR) that exploits an affinity property of PicoGreen reagent to double stranded DNA and its fluorescence intensity dependence on the dsDNA concentration. The samples were excited at 480 nm and the fluorescence emission intensity measured at 520 nm. The DNA concentration of each sample was extrapolated from a standard curve of fluorescence emission intensity versus DNA concentration.

A volume of 0.3 mL Chinese Hamster Ovary (CHO-K1) cells were loaded into a CD lysis chamber along with 0.25 g of acid-washed glass beads (density of 2.5 g/cc). Two sizes of glass beads were used, the first is around 106 µm in diameter and the second ranges from 150 to 212 µm. In the test of the mammalian cells, the two types of beads were mixed in a 7:3 ratio. A mixture of medium and beads approximately filled the lysis chamber halfway with a solid volume fraction of 0.33. The CD was then spun up to 10 rev/sec for a period of 300 cycles. Mammalian cells are, in general, easy to disrupt and thus may not be the optimal choice to test the important parameters for lysis discussed before, so we only checked mammalian cell lysing using simple microscopy. A microscope view before and after spinning the CD in Fig. 10.17 shows that almost all the cells are lysed leaving large amounts of cell debris.

For E. coli lysis tests, the same glass beads used in the mammalian test were exploited to study the effect of parameters such as solid volume fractions (SVF) and angular velocity. The two sizes of glass beads were also mixed at the same ratio (i.e., 7:3). The total volume

0 200 m

FIGURE 10.17. Comparison of before (left) and after (right) spinning for mammalian cells. The large spheres are glass beads and the small semicircular spot ( 10 µm) on the left are CHO-K1. The image on the right shows cell debris from the lysis of the CHO-K1 cells.