Biomedical Microdevices

, Volume 12, Issue 5, pp 865–874 | Cite as

Size-based separation and collection of mouse pancreatic islets for functional analysis

  • Ki-Hwan Nam
  • Wang Yong
  • Tricia Harvat
  • Adeola Adewola
  • Shesun Wang
  • Jose Oberholzer
  • David T. Eddington


Islet size has recently been demonstrated to be an important factor in determining human islet transplantation outcomes. In this study, a multi-layered microfluidic device was developed and quantified for size-based separation of a heterogeneous population of mouse islets. The device was fabricated using standard soft lithography and polydimethylsiloxane (PDMS). Size-based separation was first demonstrated via injection of a heterogeneous population of glass beads between 50–300 µm in diameter which were separated into five sub-populations based on their diameter. Next, a heterogeneous population of mouse pancreatic islets, between 50–250 µm in diameter was separated into four sub-populations. Throughout this process the islets remained intact without any signs of damage, as indicated by cell viability staining. Islet glucose-stimulated insulin secretion of each sub-population of islets was also evaluated demonstrating that islets smaller than 150 µm have superior stimulation indexes (SI) compared to islets larger than 150 µm. In this study, we found that islets between 100 µm and 150 µm in diameter had the greatest SI value in a heterogeneous population of islets.


Multilayered microfluidic device Particle separation Pancreatic islets 

1 Introduction

Pancreatic islets of Langerhans are a cluster of 1,000–2,000 cells, consisting of multiple cell types that secrete hormones in response to blood glucose levels. Islets are responsible for monitoring and maintaining normal levels of blood glucose through the secretion of insulin and glucagon. Islets have been documented to vary in size between 50 µm and 400 µm in humans and 100–200 µm in rodents. These variations occur due to many physiological and pathological factors such as age, body size, and metabolic requirements. Recently, the understanding of the physiological property of beta-cells to islet size has attracted attention including interest in the interactions between beta-cells as well as the interaction between beta-cells and other cell types. These interactions vary as a function of islet size, as larger islets have more cell-cell contacts (Jo et al. 2005). In the field of islet transplantation, a study demonstrated that islet size is a key factor in determining human islet transplant outcomes (Lehmann et al. 2007). In vitro studies have also demonstrated that small islets have better glucose-dependent insulin secretion and higher oxygen consumption (Papas et al. 1996; MacGregor et al. 2006). Additionally, during both normoxic and hypoxic culture small islets have been shown to be superior to large islets in terms of a higher survival rate (Lehmann et al. 2007). Therefore, the development of a simple tool to separate islets based on size will facilitate in vitro investigations into this phenomena and is the focus of this work.

The application of microfluidic technology for the separation of cells or microparticles has recently gained significant interest (Huang et al. 2004; Takagi et al. 2005; Sai et al. 2006; Yang et al. 2006; Yamada et al. 2007; Choi and Park 2007; Huh et al. 2007; Kulrattanarak et al. 2008; Bhagat et al. 2008; Vahey and Voldman 2008; Lillehoj et al. 2008), including the areas of field flow fraction (FFF) (Wang et al. 1998; Yang et al. 1999; Phelan and Bauer 2009), capillary hydrodynamic fractionation (Yamada et al. 2004), and split-flow thin (SPLITT) fraction (Fuh 2000; Zhang et al. 2005). These microfluidic techniques allow the precise separation of cells and particles through leveraging microscale physics or geometries (Tsutsui and Ho 2009). In addition to the small amount of reagents and analytes required, the portability and low-cost of fabrication are also well known advantages of using microfluidic devices. However, these techniques are time consuming and generally have a restricted sample volume, complex flow control, limited sorting range, and often require external forces that could potentially damage sensitive primary tissue such as islets. In addition, most of these devices have focused on microparticles and single cells, not on multi-cellular structures such as islets. The device shown here has a dynamic sorting range that is able to be tuned to the desired sample by altering device geometries. In addition, the device still provides a fast and easy separation that maintains a high homogeneity of sub-populations without any complex and time consuming flow control.

In this study, a simple gravity-driven microfluidic network that can be used for size-based islet separation is presented and characterized. The device consists of five devices operated in series for continuous separation of a heterogeneous population. Initial tests used glass microbeads for separation to verify the application of this device. Following the initial validation, the device was then used to separate mouse islets into several sub-population groups. Furthermore, we investigated the impact of the separation process on islet cell viability and we measured glucose-stimulated insulin secretion in varying size-based sub-population groups.

2 Methods and materials

2.1 Device fabrication

PDMS-based microfluidic devices with multiple heights were fabricated with one molding step, as described previously (Unger et al. 2000; Leclere et al. 2003; Nam and Eddington 2010). A schematic illustration to describe the fabrication process and design of a multi-layered microfluidic device is shown in Fig. 1. Briefly, the process consisted of fabricating a SU-8 master by using the methods of soft-lithography (Xia and Whitesides 1998) and molding micro-patterned PDMS plates. A SU-8 negative photoresist (MicroChem, MA) was spun on a clean silicon wafer (Laurell Technologies Corporation, Model WS-400B-6NPP/NITE, North Wales, PA), with defined spin speed at a specific thickness, and exposed to UV light (OmniCure, S1000, Mississauga, Ontario, Canada) through a photomask (Fineline Imaging, Colorado Springs, CO) including each constriction channel structure. The unpolymerized resist was washed away, creating the micro-patterned channel networks. The process to make the monolayer structure was repeated with a second photomask containing the distribution and extraction channel structures. The PDMS prepolymer (Sylgard 184, DOW) was cast on the micro-patterned SU-8 master and a PDMS microfluidic device was released. Next, holes were bored into the microfluidic device to make the inlet and outlet ports. The PDMS mold was bonded using O2 plasma treatment (Plasmatic Systems, Inc. Plasma-Preen II-862, North Brunswick, NJ) with a 2 mm thick flat PDMS plate to form the microchannel structure. The height of each constriction channel was controlled by the type of SU-8 (2050, 2100, and 2150) and spin speed following standard processing procedures. Each device was connected in series via plasma bonding. The outlet of one device was bonded to the inlet of the following device which had a constriction channel with a lower height as shown in Fig. 1(a).
Fig. 1

(a) Multi-connected microfluidic device depicting the principle of size-based separation in a multi-connected microfluidic device. (b) Schematic illustration of fabrication process. (c) Design of a multilayered microfluidic device

2.2 Device design

A multi-layered microfluidic device composed of a distribution channel that is 20 × 33 mm with a single height constriction channel of differing sizes was capable of size-based separation as shown in Fig. 1(c). The constriction channel heights and the distribution channel at each device were: 290 ± 10 and 570 ± 20, 225 ± 5 and 520 ± 15, 180 ± 5 and 410 ± 15, 135 ± 5 and 365 ± 10, 95 ± 5 and 280 ± 10 µm respectively.

To visualize the architecture, the microfluidic channel was filled with a blue dye. Small white circular dots (D = 500 µm) indicate microposts utilized to support a wide and thin microfluidic distribution channel. The dark regions indicate the dye-filled microfluidic channels and the bright regions indicate the microposts attached to the bottom plate. Square microposts were also constructed to prevent the constriction channels from clinging to the bottom plate at each constriction channel. The ratio of the height of the post to the length between the two posts was optimized based on previous work, in which it was demonstrated that any ratio lower then 1:25 caused the channels to collapse (Nam and Eddington 2010). Therefore, we chose the ratio of 1:5. Five devices composed of a single height constriction channel of differing heights were connected with each other. An outlet of one device was connected to an inlet of the following device composed of a lower constriction channel height as shown in Fig. 1(a). In addition, extraction channels were created to collect trapped particles and cells from the device. The inlet channel of the extraction channels was linked at the downside of one end of the constriction channel, and an outlet was linked at the topside of the opposite end of the constriction channel as shown in Fig. 1(c). This device was designed to make a simultaneous downward stream from the top to the bottom of an extraction channel and a backward stream from the outlet to the inlet of a distribution channel which is perpendicular to the direction of particle injection. The flow was analyzed through the tracking of fluorescent particles.

2.3 Separation of microbeads

The size-dependent separation of the microbeads was verified in a two-layered microfluidic device composed of a single constriction channel with a specified height. Five types of devices were prepared to measure the distribution of particles trapped at each device according to the size of particles (DP) for the height of the constriction channels. Moreover, to create the heterogeneous population of microparticles, five types of glass beads were used with diameter ranges of 50∼100 µm, 105∼150 µm, 150∼210 µm, 210∼250 µm, and 250∼300 µm (Polysciences, Inc., PA) respectively. The concentration of the heterogeneous population of glass beads was 500 µg/mL composed of 100 µg of each population. In addition, different volumes of the heterogeneous population were injected into the device to verify the ability and resolution of the device for the size-based separation. The injected amount of the population was 20 µL (350 ± 30 particles), 30 µL (570 ± 45 particles), and 40 µL (720 ± 40 particles), respectively. During microbead loading, the inlet of the microfluidic device was tilted up, allowing gravity to force the flow from the inlet to the outlet of the microfluidic channel. The angle that the device made with the horizontal plane was approximately 15°. The inlet and outlet chambers of the extraction channel at each device were closed during injection of the microparticles. Before injection of the population, the devices were carefully filled with water to minimize bubble introduction and formation. Each volume of the population was aliquoted into the inlet chamber using a micropipette.

2.4 Separation of mouse islets

Prior to the injection of 100–200 mouse islets, 5 mL of 70% ethanol was injected into the device, followed by HBSS (Mediatech, CellGRO), and RPMI1640 culture media (Invitrogen). These treatments improved the separation by preventing islet adhesion to the device surface. The device was then tilted, as previously described, approximately 15° to introduce a gravity-driven flow. The cells were then introduced into the inlet chamber of the media-filled device using a micropipette. Care was taken while introducing the islets into the chamber to minimizing mechanical stress to the islets. After introducing the islets into the distribution channel of the device, culture media fell in successive drops down to an inlet chamber of the microchannel. A p20 micropipette was initially used to push out the distributed islets into the wide distribution channel. The islets then rolled slowly along the distribution channel in the tilted device and were trapped at the constriction channels based on their size. After all of the introduced islets were trapped at the five constriction channels, inlet and outlet ports of the distribution channel were closed by small valves. During this separation process, all inlets and outlets of the extraction channels in each device were closed.

2.5 Microscopy and analysis

Once all particles or islets were trapped at the constriction channels, the particles or islets were imaged using a 4× magnification (Olympus IX71) that was attached to a digital camera (QImaging, RETIGA-SRV, CA, USA). The number and size of trapped particles and islets were then measured and analyzed with Image J to verify the distribution of the particle and islets. For each data point, averages of three images were compiled to ensure an adequate sample size.

2.6 Collection of islets from the device

After the size-based separation of the microparticles or mouse islets in the microfluidic device, inlet and outlet ports of the distribution channel were closed and the device was rotated 90° so that the openings in the extraction channel became the new inlet and outlet ports of the device. When outlet ports of the extraction channel were opened, the trapped particles at the constriction channel moved down towards the outlet of the extraction channel. Islets were collected from the outlet chamber using a micropipette.

2.7 Isolation and culturing of islets of Langerhans

The pancreatic islets of 8–12 weeks old C57/B six mice (Jackson Laboratory, Main) were isolated and cultured as previously described (Lacy and Kostianovky 1967). Briefly, the pancreas was injected in a retrograde manner through the pancreatic duct with 0.375 mg/dl Collagenase P. The distended pancreas was incubated at 37°C for 15 min and shaken vigorously for 5 s to dissociate the pancreatic acinar tissue and then washed twice with HBSS. Discontinuous Ficoll (Mediatech, VA) gradients were used for islet purification. The islets were then cultured in RPMI-1640 medium (Mediatech, VA) supplemented with 10% fetal bovine serum (Hyclone Inc, MA) and antibiotics at 37°C.

2.8 Islet cell viability

The separated rodent islets were evaluated for viability using Syto-green and Ethidium Bromide (fluorescence inclusion and exclusion dyes) as previously described (Salehi et al. 2006). These double fluorescence dyes were used to assess the amount of live (green) versus dead (red) cells.

2.9 Glucose-stimulated insulin secretion and total insulin measurement

Glucose-stimulated insulin secretion (GSIS) in mouse islets was performed by static glucose incubation and expressed in terms of stimulation index (SI) which is a measure of how the islets respond to glucose challenge. The SI was calculated by dividing insulin levels stimulated by 16.7 mM glucose by the insulin levels stimulated by 1.67 mM glucose. In brief, five of the separated islets from the microfluidic device were handpicked and incubated in Krebs-Ringer bicarbonate buffer (KRBB, pH 7.4) containing 1.67 mM glucose, 10 mM HEPES and 0.3% BSA for 30 min to stabilize islet metabolism and insulin secretion. These supernatants were removed and discarded. New 1.67 mM (low) glucose solution was then added to the islets for 1 h at 37°C and 5% CO2, and supernatants were collected at end of the incubation. The islets were then transferred to a 16.7 mM (high) glucose solution for an additional hour at which time the supernatants were collected. These islets were then incubated with cold acid ethanol for 15 min, sonicated for 10 s, and incubated overnight at 4°C. The supernants were collected in order to measure the total insulin content, after spinning at 12,000 rpm for 15 min. All supernatants were frozen at −20°C until the analysis of mouse insulin production by ELISA (Mercodia, Uppsala, Sweden).

2.10 Statistics

Statistical analysis was carried out by Student’s t-test and Pearson chi-square test. P values <0.05 were regarded as statistically significant.

3 Results and discussion

3.1 Multi-connected microfluidic device

A heterogeneous population of microbeads and mouse pancreatic islets were used to validate this device. The heterogeneous population of the particles injected into the tilted microfluidic device was evenly trapped at the constriction channels according to the size of particles in serial order as shown in Figs. 2(a) and 3(a). Microposts and a bell-shaped distribution channel contributed to the good distribution of the injected particles in the device. The particles injected into the device using a pipette slowly rolled down along the distribution channel. Additional media was added, pushing the islets or particles along until they were trapped at the constriction channels based on their size. Finally, the trapped particles were collected through the extraction channel of the device. The whole process from the separation to the collection was finished within 30 min.
Fig. 2

Size-based separation of a heterogeneous population of microparticles. (a) Particles between 50 µm and 300 μm in diameter were trapped at each constriction channel of differing heights capable of size-based separation and they were separated into five sub-populations based on size. The height of each constriction channel was 225 µm, 180 µm, 135 µm, and 95 µm respectively. (b) Size-based distribution of microparticles in the device is shown. The size of the particles indicates the diameter of each trapped particle

Fig. 3

Size-based separation of a heterogeneous population of islets. (a) The islets introduced into the device were trapped at each constriction channel of differing height respectively capable of size-based separation. (b) Size-based distribution of islets in the device. (c) Total insulin contents of each collected sub-population of mouse islets (n = 6 for each group, ng/islet±SE). Each sub-population of 1, 2, 3, and 4 was composed of 74–112 µm, 106–154 µm, 145–205 µm, and 200–245 µm in islet diameter

Inlet and outlet ports of an extraction channel at each device were closed during the separation process while inlet and outlet ports of a distribution channel were opened to hinder injected particles from going into the extraction channels. Conversely, the inlet and outlet ports of the distribution channel at each device were closed during the collection process while the inlet and outlet ports of an extraction channel were opened to interrupt the trapped particles from spreading out to the distribution channel. As a result, the particles did not go out of the extraction channel during separation nor did they spread out to the distribution channel during collection.

The multilayered microfluidic device was made of PDMS, which is an elastomeric material and thus can be deformed under the pressure used to drive the flow (Hardy et al. 2008; Nam and Eddington 2010). In this study, no external force such as a pumping system was used to inject particles or cells into the device. During the injection and collection of particles using a micropipette, there were no observable changes in the height of the constriction channels. Consequently, deformation of the device under pressurized flow is not an issue. Handling the micro-level flow through pumps makes the system complicated and, therefore, can be an obstacle for routine and commercial use of a microfluidic device. However, through the use of gravity flow and simple tilting of the device, no external pumping equipment is required for operation, which facilitates the use of this device by non-specialized users.

3.2 Separation of a heterogeneous population of microbeads

Figure 2 (a) shows size-based separation of a heterogeneous population of glass beads in five two-layered microfluidic devices each composed of a single constriction channel of differing heights in each device. The size of particles included in the heterogeneous population was between 50 µm and 300 µm diameter particles. The particles were injected into the device using a micropipette and the particles were homogeneously trapped by size along the 2 mm wide constriction channels.

To validate separation efficiency and resolution of the microfluidic device as a function of the number of particles, various amounts of particles, including 350, 570 and 720 particles, were added to the device in different volumes (20 µL, 30 µL, and 40 µL) to achieve an overall concentration of 500 µg/mL. The particles were successfully separated based on their size which is limited by the height of constriction channels as shown in Fig. 2(b). The size of the particles indicates the diameter of each trapped particle measured through images taken by a microscope. The results of the three different amounts of a heterogeneous population of particles described above were included in the same plot as there was no statistical difference according to the amount of particles injected. Microscopic observation showed that very few particles stuck to the PDMS surfaces in the distribution channels, but the number was insignificant compared with the number of trapped particles at the constriction channel. The size range of the particles trapped at each constriction channel were as follows; 224–288 µm particles at the 225 µm tall constriction channel, 179–215 µm at the 180 µm, 135–173 µm at the 135 µm, 99–138 µm at the 95 µm, and 51–83 µm particles in the outlet of the device. There were no particles trapped at the 288 µm tall constriction channel. The size of approximately 97% of trapped particles at each constriction channel was larger than the height of the constriction channel where they were trapped, and smaller than the height of an upper constriction channel.

After trapping particles and closing all inlets and outlets, the device was gently shaken from side to side. The small particles that were blocked by the larger particles were able to pass through the constriction channel and moved to the next device. Therefore, by avoiding the larger particles from blocking the smaller particles, the efficiency of the device was improved in terms of size-based separation and enhanced homogeneity of each separated sub-population of particles.

3.3 Size-based separation of mouse islets

Size-based separation of mouse pancreatic islets was accomplished in our multi-connected microfluidic device and the glucose-stimulated insulin secretion of each sub-population of islets was evaluated. Figure 3(a) shows evenly trapped islets based on size at the 24 mm wide constriction channels of varying heights. Moreover, a heterogeneous population of islets introduced into the device was successfully separated based on the size which was limited by the height of constriction channels as shown in Fig. 3(b). The size of islets obtained from the top view images indicates an average of two lines, the longest and shortest diameter of each trapped islet. No islets were trapped at the 288 µm and 226 µm tall constriction channels. The size range of the trapped islet at each constriction channel was as follows: 200–245 µm at the 180 µm tall constriction channel, 145–205 µm at the 135 µm, 106–154 µm at the 95 µm, and 75–112 µm in the outlet of the device. Consequently, a heterogeneous population of islets was separated into four sub-populations based on size. In order to further confirm the separation accuracy beyond microscopic observation, total insulin content of each islet sub-population was measured as shown in Fig. 3(c) These results demonstrated that the total insulin contents increased when the size of the islets increased as expected.

As a result of this study, the size-based separation of a heterogeneous population of islet was successfully achieved in the multi-connected microfluidic device composed of constriction channels of differing heights. There were a small percentage of islets that overlapped with the larger and smaller population at each constriction height. The results showed that each sub-population contains at least 9% (±3) larger islets than the height of the constriction channel where they were trapped, and at most 9% (±4) larger islets than the height of an upper constriction channel. Studies using islets demonstrated a significantly higher statistical variation compared to the studies using glass beads, however, glass beads are firm and uniform in diameter as compared with islets which are soft and asymmetric. In addition, unlike microbeads, the size of islets could also be influenced by several other factors including flow pressure and osmolarity of the solutions. Yet, similar to microbeads separation, blocking smaller islets by previously trapped larger islets was not observed in this microfluidic device. Future design of the device will take these observations into consideration, by making constriction channels 10% lower than expected islet size.

3.4 Collection of particles from the device

After the separation was accomplished, the inlet and outlet ports of a distribution channel were closed and the device was rotated 90°. Most of the trapped particles rolled to the outlet chamber of the extraction channel and the remainder were dislodged by adding several drops of water or media into the inlet of the extraction channel with a micropipette. Consequently, 99% of trapped particles at each constriction channel were collected from the device through the extraction channels.

3.5 Post-separation islet viability and insulin secretion capabilities of islet sub-populations

Pancreatic islets are prone to mechanical damage. Islet viability was assessed after separation to determine if there was any mechanical stress and damage from the microfluidic device, the loading process or the separation maneuver. Our results shown in Fig. 4(a), indicated that the islets from each sub-population group are viable (green channel) with very little ethidium bromide staining (red channel), suggesting that the islets remained intact without damage.
Fig. 4

(a) Viability test of each sub-population. Live cells (green channel) and dead cells (red channel). (b) Stimulation index (SI). SI demonstrates insulin secretion of islets in response to glucose stimulation and calculated by [insulin]16.7 mM glucose/[insulin] 1.67 mM glucose±SE (*P < 0.01, n = 6 for each group)

Insulin secretion capacities of each separated sub-population in response to glucose stimulation were evaluated, as shown in Fig. 4(b), demonstrating that islets smaller than 150 µm have superior secretion compared to islets larger than 150 µm. Stimulation indexes of small islets was double that of larger islets (SI: 3.207 ± 0.643 for 50–100 µm; 3.919 ± 0.441 for 100–150 µm; 1.029 ± 0.118 for 150–200 µm; 1.567 ± 0.198 for 200–250 µm, P < 0.01). The data are in line with recent findings from a rodent study demonstrating that small islet are superior to large islets in vitro and in vivo based on their function when transplanted under the kidney capsule of diabetic rats (MacGregor et al. 2006). In this study, islets between 100 µm and 150 µm in diameter had the greatest SI value in a heterogeneous population of islets.

Several mechanisms may explain why small islets have better glucose-stimulated insulin secretion than large islets. Without vascular support and disruption of innervations after isolation, the isolated islets lose vascular-dependent and neural regulated signals required for synchronization between beta-cells, leading to irregularities of insulin secretion. The glucose induced insulin secretion in the isolated islets depends on passive glucose diffusion and direct interaction between beta-cells through gap junctions. Hence, smaller islets have a smaller diffusion barrier and are able to respond more efficiently to changes in glucose concentrations than larger islets.

Currently, human islet transplantations are limited by the scarcity of donor pancreases. In addition, more than one donor is often required to reach normal blood glucose levels after transplantation. Current protocols for transplantation do not take into account islet size prior to transplantation, only the total amount to islets needed based on the body weight of the recipient. The device presented in this work passively separates the islets into four sub-populations, therefore, this device could allow for accurate size estimations of a given isolation preparation to be easily generated without time consuming manual measurements through simply counting the number of islets at each constriction region. The current practice of size estimation involves manual measurement of islet diameters which is often difficult due to their asymmetric nature. The device presented here could be used to rapidly assess the islet size fractions without susceptibility of user bias as found with manual measurements. More than 1,000,000 islets are required for each human islet transplantation, therefore, scaling up the method presented here would be a significant, but not insurmountable engineering challenge. However, it would not be advantageous to sort the entire isolation preparation as all the isolated tissue would most likely be transplanted. Instead, the device presented in this work could be used to provide a quick estimation of the size range of the isolated tissue to provide another set of data to help predict the success of an islet transplantation. As shown by our SI data, 1,000,000 large islets would behave much differently than 1,000,000 smaller islets.

4 Conclusions

A novel microfluidic device using gravity force for size-dependent separation of microbeads and pancreatic islets has been presented. It is a big challenge to separate an islet population into multiple sub-populations based on islet size, especially without external force. Our device is a simple, yet versatile tool for islet separation that does not sacrifice separation resolution, sample retrieval efficacy, or analysis time. It has been demonstrated that islet transplantation outcomes are size-dependent in rodent and human models (Lehmann et al. 2007). These studies were done by handpicking the islets and generally only two groups of islets (small and large) were defined. Our microfluidic network has fulfilled the needs for grouping multiple islet sub-populations. Our results have also demonstrated that islets less than 150 µm have superior insulin secretion in response to glucose stimulation, when compared to islets larger than 150 µm.



This work was supported by the NIH (JO) and the Alfred P. Sloan Foundation (DTE).


  1. A.A.S. Bhagat, S.S. Kuntaegowdanahalli, I. Papautsky, Lab Chip 8, 1906 (2008)CrossRefGoogle Scholar
  2. S. Choi, J.K. Park, Lab Chip 7, 890 (2007)CrossRefGoogle Scholar
  3. C.B. Fuh, Anal. Chem. 72, 266A (2000)Google Scholar
  4. B.S. Hardy, K. Uechi, J. Zhen, H. Pirouz Kavehpour, Lab Chip 9, 935 (2008)CrossRefGoogle Scholar
  5. L.R. Huang, E.C. Cox, R.H. Austin, J.C. Sturm, Science 304, 987 (2004)CrossRefGoogle Scholar
  6. D. Huh, J.H. Bahng, Y. Ling, H.-H. Wei, O.D. Kripfgans, J.B. Fowlkes, J.B. Grotberg, S. Takayama, Anal. Chem. 79, 1369 (2007)CrossRefGoogle Scholar
  7. J. Jo, H. Kang, M.Y. Choi, D.S. Koh, Biophys. J. 89, 1534 (2005)CrossRefGoogle Scholar
  8. T. Kulrattanarak, R.G.M. van der Sman, C.G.P.H. Schroën, R.M. Boom, Adv. Colloid Interface Sci. 142, 53 (2008)CrossRefGoogle Scholar
  9. E.P. Lacy, M. Kostianovky, Diabetes 16, 35 (1967)Google Scholar
  10. E. Leclere, Y. Sakai, T. Fujii, 2003 IEEE, 415 (2003)Google Scholar
  11. R. Lehmann, R.A. Zuellig, P. Kugelmeier, P.B. Baenninger, W. Moritz, A. Perren, P.A. Clavien, M. Weber, G.A. Spinas, Diabetes 56, 594 (2007)CrossRefGoogle Scholar
  12. P. Lillehoj, N. Li, H. Tsutsui, and C-M. Ho, 2008 IEEE, 292 (2008)Google Scholar
  13. R.R. MacGregor, S.J. Williams, P.Y. Tong, K. Kover, W.V. Moore, L. Stehno-Bittel, Am J Physiol Endocrinol Metab. 290, 771 (2006)CrossRefGoogle Scholar
  14. K. Nam, D.T. Eddington, Size based separation of microparticles in a multilayered microfluidic device, J. Microelectromechanical Systems 19(2), 375–383 (2010)Google Scholar
  15. K.K. Papas, R.C. Long Jr., I. Constantinidis, A. Sambanis, Biochim. Biophys. Acta. 1291, 163 (1996)Google Scholar
  16. F.R. Phelan Jr., B.J. Bauer, Chem Eng Sci 64, 1747 (2009)CrossRefGoogle Scholar
  17. Y. Sai, M. Yamada, M. Yasuda, M. Seki, J. Chrom. A 1127, 214 (2006)CrossRefGoogle Scholar
  18. P. Salehi, M.A. Hansen, J.G. Avila, B. Barbaro, A. Gangemi, T. Romagnoli, Y. Wang, M. Qi, P. Murdock, E. Benedetti, J. Oberholzer, Transplantation 82, 983 (2006)CrossRefGoogle Scholar
  19. J. Takagi, M. Yamada, M. Yasuda, M. Seki, Lab Chip 5, 778 (2005)CrossRefGoogle Scholar
  20. H. Tsutsui, C.-M. Ho, Mech Res Comm 36, 92 (2009)CrossRefGoogle Scholar
  21. M.A. Unger, H.-P. Chou, T. Thorsen, A. Scherer, S.R. Quake, Science 288, 113 (2000)CrossRefGoogle Scholar
  22. M.D. Vahey, J. Voldman, Anal. Chem. 80, 3135 (2008)CrossRefGoogle Scholar
  23. X.B. Wang, J. Vykoukal, F.F. Becker, P.R. Gascoyne, Biophys. J. 74, 2689 (1998)CrossRefGoogle Scholar
  24. Y. Xia, G.M. Whitesides, Annu. Rev. Mater. Sci. 28, 153 (1998)CrossRefGoogle Scholar
  25. M. Yamada, M. Nakashima, M. Seki, Anal. Chem. 76, 5465 (2004)CrossRefGoogle Scholar
  26. M. Yamada, K. Kano, Y. Tsuda, J. Kobayashi, M. Yamato, M. Seki, T. Okano, Biomed. Microdevices 9, 637 (2007)CrossRefGoogle Scholar
  27. J. Yang, Y. Huang, X.-B. Wang, F.F. Becker, P.R.C. Gascoyne, Anal. Chem. 71, 911 (1999)CrossRefGoogle Scholar
  28. S. Yang, A. Undar, J.D. Zahn, Lab Chip 6, 871 (2006)CrossRefGoogle Scholar
  29. Y. Zhang, R.W. Barber, D.R. Emerson, Curr. Anal. Chem. 1, 345 (2005)MATHCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2010

Authors and Affiliations

  • Ki-Hwan Nam
    • 1
  • Wang Yong
    • 2
  • Tricia Harvat
    • 2
  • Adeola Adewola
    • 2
  • Shesun Wang
    • 2
  • Jose Oberholzer
    • 1
    • 2
  • David T. Eddington
    • 1
  1. 1.Department of BioengineeringUniversity of Illinois at ChicagoChicagoUSA
  2. 2.Department of SurgeryUniversity of IllinoisChicagoUSA

Personalised recommendations