Exploiting osmosis for blood cell sorting
- First Online:
- Cite this article as:
- Parichehreh, V., Estrada, R., Kumar, S.S. et al. Biomed Microdevices (2011) 13: 453. doi:10.1007/s10544-011-9513-y
- 590 Views
Blood is a valuable tissue containing cellular populations rich in information regarding the immediate immune and inflammatory status of the body. Blood leukocytes or white blood cells (WBCs) provide an ideal sample to monitor systemic changes and understand molecular signaling mechanisms in disease processes. Blood samples need to be processed to deplete contaminating erythrocytes or red blood cells (RBCs) and sorted into different WBC sub-populations prior to analysis. This is typically accomplished using immuno-affinity protocols which result in undesirable activation. An alternative is size based sorting which by itself is unsuitable for WBCs sorting due to size overlap between different sub-populations. To overcome this limitation, we investigated the possibility of using controlled osmotic exposure to deplete and/or create a differential size increase between WBC populations. Using a new microfluidic cell docking platform, the response of RBCs and WBCs to deionized (DI) water was evaluated. Time lapse microscopy confirms depletion of RBCs within 15 s and creation of > 3 μm size difference between lymphocytes, monocytes and granulocytes. A flow through microfluidic device was also used to expose different WBCs to DI water for 30, 60 and 90 s to quantify cell loss and activation. Results confirm preservation of ∼ 100% of monocytes, granulocytes and loss of ∼ 30% of lymphocytes (mostly CD3+/CD4+) with minimal activation. These results indicate feasibility of this approach for monocyte, granulocyte and lymphocyte (sub-populations) isolation based on size.
KeywordsMicrofluidicsCell sortingBlood cells
Blood is a living tissue that perfuses the entire body and is the primary mediator of immune homeostasis. Local and systemic changes following injury and disease are almost immediately reflected as changes in molecular and cellular constituents of blood. Cellular populations in blood of particular interest for high-throughput gene and protein expression studies are leukocytes or white blood cells (WBCs) (Calvano et al. 2005; Cobb et al. 2005). Prior to analysis of WBCs, it is often beneficial to deplete contaminating erythrocytes or red blood cells (RBCs) which make up ∼ 99% of the cells in blood (Feezor et al. 2004). Further, separation of WBCs into individual populations can enhance the quality of analysis by minimizing heterogeneity within the sample and reducing noise in molecular expression data (Laudanski et al. 2006). An important yet often overlooked consideration during the isolation process is minimizing cell loss and artificial activation due to the isolation process (Kuijpers et al. 1991; Lundahl et al. 1995; Macey et al. 1995). This process is extremely critical to ensure generation of high fidelity information from subsequent genomics and proteomic analysis and will minimize generation of artifactual data as a consequence of isolation process induced events.
Currently, the most effective techniques for isolation of WBC sub-populations use immuno-specificity, (i.e.) antibodies specific to antigens or cell surface phenotype markers to accomplish separation (Neurauter et al. 2007; Sleasman et al. 1997; Ujam et al. 2003). While such approaches are highly specific and can be integrated with magnetic bead based separation, immuno-affinity columns and density gradient separation, the use of antibodies and the binding event itself is a source of unnecessary signaling and downstream signal transduction. For example, anti- cluster of differentiation (CD) 66b used to phenotype neutrophils and esonophils (sub-populations of granulocytes) is a key mediator in adhesion, activation and downstream signaling (Yoon et al. 2007). Therefore techniques that exploit physical properties of cells rather than chemical or biological targets to accomplish separation hold strong appeal for blood cell sorting.
Size-based sorting is one particular option that has been explored extensively in recent years. Three groups in particular have exploited flow phenomenon that develop in curvilinear micro-channels channels with a rectangular cross-section to separate cells and particles based on size (Kuntaegowdanahalli et al. 2009; Di Carlo et al. 2008; Russom et al. 2009a). This phenomenon is based on trapping cells/particles in equilibrium positions due to inertial lift forces generated in channels with rectangular cross-sections at Reynolds Numbers (Re) between 1–10 and Dean’s vortices that develop perpendicular to the direction of fluid flow when the channel is arranged in a curved or spiral fashion. Other alternatives to inertial sorting like deterministic lateral displacement (Green et al. 2009; Huang et al. 2004), acoustic size-based sorting (Kapishnikov et al. 2006), cross-flow filtration (Chen et al. 2008) and size-based filtration (Ji et al. 2008; Murthy et al. 2006; Sethu et al. 2006a) can also benefit from this technique to enable sorting of blood cells. With inertial sorting techniques, the smallest size difference necessary for reliable discrimination of particles/cells is 3 μm (Russom et al. 2009b) whereas with the deterministic lateral displacement the minimum size difference necessary for sorting is ∼ 2.5 μm (Huang et al. 2004). Other techniques do not report the smallest size difference necessary to accomplish sorting.
This manuscript describes efforts taken to determine if controlled exposure to hypotonic solution like deionized (DI) water can be used to deplete certain blood cell populations (particularly RBCs) and create a suitable size difference between other WBC populations to facilitate separation into unique sub-populations. To accomplish this, a cell docking module was designed and fabricated to allow extra-cellular conditions to be changed almost instantaneously while the response of cells can be monitored and characterized using an inverted microscope. Computational fluid dynamics (CFD) modeling was initially performed to determine if trapped cells remained within the grooves and mass transfer was almost instantaneous. The model was then experimentally validated and used to study the response of different blood cells (RBCs and WBCs) to DI water to characterize the size increase and eventual lysis as a function of time. Results indicate that specific exposure times allow complete depletion of RBCs and sorting of WBCs. Cell lysis and activation were also evaluated using a flow-through device to subject isolated WBCs to 30, 60 and 90 s exposure to DI water. This device essentially performs the same function as the docking device but does not allow for visualization of the cells. Since sample continuously flows through the channels, larger sample volumes can be processed and sufficient number of cells can be obtained for flow cytometry. Preservation of ∼ 100% of granulocytes and monocytes was observed along with ∼ 70% recovery of lymphocytes. Activation studies using early activation markers reveal that lymphocytes and monocytes are not activated even after 90 s of exposure to DI water whereas there was minimal (statistically insignificant) activation of granulocytes.
2 Materials and methods
2.1 Blood samples
4 mL of blood was withdrawn from healthy volunteers by venipuncture as per protocols approved by the University of Louisville Institutional Regulatory Board (IRB) and collected in green top vacctuainers (Fisher Scientific, Florence, KY) with heparin as anticoagulant. Samples were maintained on ice until use. All data is represented as mean values for a sample size of n = 5 ± SEM where SEM is the standard error of the mean and can be calculated by dividing the standard deviation by the square root of the sample size.
2.2 Cell docking device fabrication
2.3 Controlled exposure device
2.4 Fluid flow modeling
2.5 Experimental procedure for immobilization and analysis of cells
The device was primed with 1 × phosphate buffered saline (PBS) to eliminate air and trapped bubbles within the device prior to use. Cells were suspended in 1 × PBS with 0.4% trypan blue solution to aid visualization and introduced into the device. Cell concentration was adjusted to ∼ 1 × 105 cells/ml to provide sufficient cellular populations for analysis. Cells were allowed to sediment within the device for ∼ 2 min and then the cells not immobilized within the grooves were removed using 1 × PBS with 0.4% trypan blue. A 10 mL syringe was filled with DI water and loaded onto a syringe pump (Harvard Apparatus, Holliston, MA) which was set to deliver fluids at ∼ 100 μL/min. The flow of DI water was initiated and a stop clock was started as soon as a visible color change was observed within the grooves (DI water replacing the 1 × PBS with trypan blue). Images were taken every 3 s and analyzed using Metamorph software (Molecular Devices, Sunnyvale, CA). Briefly, the measurements of cell size were accomplished manually. Once the image was captured and the diameter of each individual cell was estimated by measuring the length of the cell from end to end. This was accomplished 10 different times from different directions and the average values were used to determine the cell diameter.
2.6 Separation of peripheral blood mononuclear cells (PBMCs) and polymorphonuclear cells (PNMs)
PBMC and PNM separation from whole blood was accomplished using standard density gradient separation. Briefly, 4 mL of blood was mixed with 4 mL of 1 × phosphate buffered saline (PBS) and carefully layered on top of 10 mL of Ficol Paque (GE Healthcare, Waukesha, WI) in a 50 mL tube. Samples were then balanced and centrifuged for 30 min at 450 g. Following this procedure, PBMCs were fractionated as a narrow band below the plasma and above the Ficol whereas the PNMs and RBCs pellet to the bottom below the Ficol. Each fraction was isolated and the PBMCs were washed once with 1 × PBS whereas the PNM and RBC mixture was mixed with 30 mL of ammonium chloride lysis buffer for 5 min to eliminate RBCs. The samples were then again centrifuged at 250 g for 5 min and resuspended in 1 × PBS and maintained at room temperature until needed.
2.7 Cell counts
Sample cell concentration was estimated prior to introduction into the controlled exposure device and again counted after exposure to DI water and return to isotonic conditions at the outlet. Cells were counted using a hemocytometer. The difference between the inlet and outlet counts represents the number of cells lost due to DI water exposure. Cell Counts of different sub-populations was estimated using flow cytometry using a combination of light scatter and phenotype markers.
2.8 Activation studies
Activation studies were performed using flow cytometry. Samples were first suspended in a flow cytometry buffer consisting of 1 × PBS and 1% bovine serum albumin (BSA). Antibodies used were conjugated to fluorescently labeled tags such as Fluorescein isothiocyanate (FITC), phycoerythrin (PE), Peridinin Chlorophyll Protein Complex (PerCP) and Allophycocyanin (APC) and evaluated using a four color flow cytometer (BD FACS Calibur, BD Biosciences, Nashville, TN). Antibodies specific to cell surface markers CD3 and CD4 for lymphocytes, CD14 for monocytes and CD66b for granulocytes were used to phenotype white blood cells into major sub-populations. Once cells were phenotyped, they were evaluated for activation using early activation markers identified previously (Sethu et al. 2006b). Expression of CD25 and CD69 were evaluated for lymphocytes and expression of CD18 and CD29 were evaluated for monocytes and granulocytes.
3.1 Device design and modeling
To enable characterization of changes in cell size in response to changing extra-cellular conditions a device was designed to first immobilize cells within grooves (50 μm × 50 μm) in a microfluidic channel (Fig. 1). This device is compatible with an inverted microscope where images of cells at different time points can be acquired using a charge coupled device (CCD) camera and analyzed using image processing software. The aspect ratio of the grooves within the microchannel was designed to be ∼ 1:1. CFD modeling was used to determine the effect of various inlet flow rates on cells trapped at the bottom of the grooves. CFD results indicate that the fluid flowing in the main channel at a flow rate of 100 μL/min, corresponding to a main channel velocity of 8 mm/s does not directly affect cells trapped at the bottom of the channel. Rather, the fluid flow in the main channel causes recirculation at lower velocities (~ 0.014 mm/s) within the bottom half of each groove (Fig. 2). These results confirm theoretically, that cells trapped at the bottom of the grooves do not get carried away by the fluid in the main channel and mass transfer is accomplished instantaneously and effectively due to a combination of diffusion mixing and recirculation effects at the bottom of the channel. These modeling results were confirmed experimentally using cells suspended in trypan blue and introduced into the device and flushing with 1 × PBS. Cells remained immobilized within the grooves and mass transfer was accomplished in ∼ 2 s following entry into the device as visualized by replacement of the blue color of trypan blue in the grooves. This ∼ 2 s equilibration time was taken into account for all subsequent experiments.
3.2 RBC characterization
3.3 PBMC characterization
PBMCs or the mononuclear cell fraction is a mixture of primarily monocytes and lymphocytes. In isotonic conditions monocytes are typically ∼ 10.7 ± 0.291 μm in diameter and lymphocytes are ∼ 7.6 ± 0.156 μm in diameter allowing visual discrimination between the two cell types. PBMC samples with trypan blue were introduced into the device and evaluated in a similar fashion to the RBCs. Image analysis was performed to track cells identified as lymphocytes and monocytes based on initial size (Fig. 4(b) and (c)). Monocytes, initially ∼10.7 ± 0.291 μm in diameter swell and reach a size of 17 ± 0.210 μm following 100 s of exposure. Lymphocytes on the other hand increase from their initial size of ∼ 7.6 ± 0.156 μm to 11 ± 0.071 μm within 30 s. At this time lysis of lymphocytes begins and a ∼ 30% of the lymphocytes begin to lyse and are completely lysed by 40 s. The remaining 70% of the lymphocytes remain at a size of 11 ± 0.071 μm until 100 s without any change in size.
3.4 PNM characterization
PNMs collectively refer to granulocytes including neutrophils, basophils, eosinophils and other cells including mast cells. These cells are characterized by multiple nuclei and highly granular cytoplasm which is a source of high side scatter in flow cytometry. PNMs obtained following lysis of contaminating RBCs was mixed with trypan blue, introduced into the device and characterized in a similar fashion to RBCs and PBMCs. Image analysis of granulocytes in isotonic conditions measure ∼ 8.03 ± 0.262 μm in diameter and following exposure to DI water gradually increase in size to ∼ 14.8 ± 0.176 μm in diameter and remain intact even 100 s in DI water (Fig. 4(d)).
3.5 Combined analysis
3.6 Cell recovery
Cell recovery following exposure to DI water (sample size n = 5)
% of cells recovered
3.7 Cell activation
Blood cell sorting without the use of cell specific antibodies is important to minimize unnecessary signaling in cell samples prior to molecular expression analysis. Inertial microfluidic sorters that exploit flow phenomenon in curvilinear channels with rectangular cross-section hold great promise to accomplish antibody-free sorting based on size. Without appropriate sample pre-processing, blood cells cannot be sorted into WBC subpopulations using inertial microfluidic sorters due to abundance of unwanted RBCs and size overlap between different WBC populations. Other size based sorting techniques like lateral deterministic displacement, size-based filtration, size-based acoustic sorting and cross-flow filtration can also benefit from cell size amplification. Subjecting cells to hypotonic solutions like deionized water results in fluid transport across the cell membrane and a buildup in pressure within the cells. The process of cell swelling and ultimately lysis depends on the cells membrane properties including presence or absence of water channels (aquaporins), cytoplasm to nucleus ratio and intracellular ratios of F-actin to G-actin. This process can be potentially exploited to selectively deplete certain cells while also creating a sufficient size difference between remaining cell populations to provide a suitable starting sample for size-based sorting.
To determine the feasibility of using an osmosis based approach for sample pre-processing, the response of blood cells to hypotonic solutions like DI water needs to be characterized. A new microfluidic cell docking device was designed and constructed for this purpose. Fluid flow in channels with grooves that have aspect ratio ∼ 1:1, 50 μm (width): 50 μm (height) does not affect the bottom of the grooves directly but creates recirculation patterns at the bottom of the grooves which is extremely important in ensuring mass transfer is rapid and the conditions within the wells match conditions in the main channel. This provides a unique platform for visually characterizing the immediate and time dependent response of cells to changing extracellular stimulus.
Experiments performed with the cell docking device indicate that RBCs can be eliminated within 15 s of hypotonic exposure without loss of any WBC populations. At the 30 s time point, the lymphocytes begin to lyse and within 40 s of exposure, ~ 30% of the lymphocytes are lysed. The remaining 70% of the lymphocytes remain at a size of 11 μm in diameter till the conclusion of the experiment at 100 s. Lymphocytes consist of four unique sub-populations: T helper cells, cytolytic T cells, B cells and natural killer (NK) cells. We speculate that the remaining cells may represent a unique sub-population of lymphocytes. At the 40 s time point, the size difference is still unsuitable for separation of the three sub-populations. However, with increase in exposure time the granulocytes and monocytes differentially increase in size whereas the lymphocytes remain at 11 μm. At 60 s the size difference between the remaining lymphocytes and the larger cells is > 3 μm and which provides an opportunity to sort the remaining lymphocytes. A ∼ 3 μm size difference between granulocytes and monocytes appears only following 90–100 s of exposure. Therefore, a 90 s exposure provides the best opportunity to eliminate RBCs and sort monocytes, granulocytes and a sub-population of lymphocytes based on size.
To confirm this technique does not induce WBC activation, activation studies were performed using commonly used early activation markers. These studies confirm that that most cells were not activated and the granulocytes showed minimal activation which was found to be statistically insignificant. However, there was loss of lymphocytes following > 30 s of exposure to DI water. These cells were predominantly CD3+/CD4+ lymphocytes. Despite loss of this sub-population of lymphocytes, this technique can be used to isolate other lymphocyte subpopulations as well as granulocytes and monocytes without loss and minimal activation.
Size-based sorting is an attractive alternative to immuno-affinity based approaches. However, prior to application for blood cell sorting RBCs need to be eliminated and sufficient size difference need to created between different WBC populations. Using the microfluidic docking device, we have demonstrated that osmosis that ensues following DI water exposure can be exploited to first deplete RBCs and create sufficient size differences between different WBC populations for subsequent sorting using size-based sorting techniques.
The authors would like to thank Tim Andrews, phlebotomist, Division of Pediatric Hematology/Oncology for help with obtaining blood samples and helpful discussions. This work was supported through a proof-of-concept grant (POCG) through the Office of Technology Transfer and the Vice President of Research at the University of Louisville and by the National Science Foundation under Grant No. 0814194.