Broad spectrum immunomodulation using biomimetic blood cell margination for sepsis therapy

(1)

Broad spectrum immunomodulation using

biomimetic blood cell margination for sepsis therapy

The MIT Faculty has made this article openly available.

Please share

how this access benefits you. Your story matters.

Citation

Hou, Han Wei, et al. “Broad Spectrum Immunomodulation Using

Biomimetic Blood Cell Margination for Sepsis Therapy.” Lab on a

Chip 16, 4 (2016): 688–699 © 2016 The Royal Society of Chemistry

As Published

http://dx.doi.org/10.1039/c5lc01110h

Publisher

Royal Society of Chemistry, The

Version

Author's final manuscript

Citable link

http://hdl.handle.net/1721.1/110938

Terms of Use

Creative Commons Attribution-Noncommercial-Share Alike

(2)

Broad spectrum immunomodulation using biomimetic blood cell

margination for sepsis therapy

Han Wei Houa,1,2, Lidan Wub,1, Diana P. Amador-Munozc,1, Miguel Pinilla Verac, Anna

Coronatac_{, Joshua A. Englert}c_{, Bruce D. Levy}c_{, Rebecca M. Baron}c_{, and Jongyoon Han}a,b,# a_{Department of Electrical Engineering & Computer Science, Massachusetts Institute of}

Technology, Cambridge, MA 02139, USA

b_{Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA} 02139, USA

c_{Division of Pulmonary and Critical Care Medicine, Department of Internal Medicine, Brigham and} Women’s Hospital, Harvard Medical School, Boston, MA 02115, USA

Abstract

Sepsis represents a systemic inflammatory response caused by microbial infection in blood. Herein, we present a novel comprehensive approach to mitigate inflammatory responses through broad spectrum removal of pathogens, leukocytes and cytokines based on biomimetic cell

margination. Using a murine model of polymicrobial sepsis induced by cecal ligation and puncture (CLP), we performed extracorporeal blood filtration with the developed microfluidic blood margination (µBM) device. Circulating bacteremia, leukocytes and cytokines in blood decreased post-filtration and significant attenuation of immune cell and cytokine responses were observed 3– 5 days after intervention, indicating successful long-term immunomodulation. A dose-dependent effect on long-term immune cell count was also achieved by varying filtration time. As proof of concept for human therapy, the µBM device was scaled up to achieve ~100-fold higher throughput (~150mLhr−1). With further multiplexing, the µBM technique could be applied in clinical settings as an adjunctive treatment for sepsis and other inflammatory diseases.

Introduction

Several acute and chronic inflammatory conditions display unrestrained inflammation as part of disease pathogenesis. Sepsis, defined as the systemic inflammatory response to infection, is a common example of this problem and is the leading cause of death in intensive care units1. It is well established that key clinical manifestations of sepsis are not caused solely by invading pathogens but instead result from the dysfunction of host defense

#_{Contact: Jongyoon Han (jyhan@mit.edu), Tel: +1-617-253-2290, Room 36-841, Research Laboratory of Electronics, Massachusetts}

Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA. 1These authors contributed equally to this work.

2Current address: Lee Kong Chian School of Medicine, Nanyang Technological University, SINGAPORE

Author contributions

H.W.H, B.D.L, R.M.B and J.H designed research, H.W.H, L.W., D.P.A, M.P.V., A.C. and J.A.E. performed experiments and

HHS Public Access

Author manuscript

Lab Chip. Author manuscript; available in PMC 2017 February 21.

Published in final edited form as:

Lab Chip. 2016 February 21; 16(4): 688–699. doi:10.1039/c5lc01110h.

Author Manuscript

(3)

mechanisms leading to overwhelming cytokine production, widespread activation of vascular endothelium and coagulation cascades, and increased susceptibility to secondary infections2–4. These multi-factorial effects damage the microvasculature which induces tissue hypoxia and leads to shock and sometime death as a result of multiple organ failure. Treatment for sepsis has been limited to source control to remove the nidus of infection, antibiotics, and supportive care, without any existing targeted effective molecular therapy. Current medicinal approaches for modulating the immune system often carry risks of immunosuppression. Limited success has been achieved with pharmacological approaches, most notably with the use of corticosteroids whose effect on sepsis treatment remains controversial5–7. Meanwhile, other immunomodulation modalities have not gained widespread acceptance into clinical practice8, 9, which supports the need to develop new approaches to control the host response during sepsis. Extracorporeal blood purification therapies removing inflammatory mediators using semi-permeable membrane (diffusion/ convection) or sorbent (adsorption) have been shown to improve clinical outcomes in patients10–12. Although modulation of inflammatory signaling has been widely proposed to improve survival in sepsis, there is currently no consensus on which components of the blood should be removed given the complexity of its pathophysiology13. Clinical studies have shown poor efficacy of therapies that block single mediators including tumor necrosis factor-α (TNF-α)14, interleukin-1 (IL-1)15 and recombinant human activated protein C16. Benefits of endotoxin clearance remains controversial with improved survival in some studies17, 18 but not others19. Broad spectrum cytokine removal has been reported to improve survival in animal models20, 21, but its beneficial outcomes are believed to result from modulating mechanisms other than direct cytokine removal22. Moreover, increasing evidence has highlighted the importance of removing activated immune cells in sepsis treatment by hemoadsorption23, 24 or transfusion with healthy granulocytes25, 26. Pathogen removal using magnetic beads coated with human opsonin mannose-binding lectin was also recently reported to improve short-term survival (5 hr) in a rat model of acute endotoxemic shock27. In this regard, it is imperative to develop a new generation of blood purification tools and strategies for sepsis treatment that can remove a broad array of cell targets (pathogens, immune cells, platelets, etc.) along with molecular targets (cytokines, endotoxin etc.) from blood to help modulate the overall inflammatory response.

In this work, we provide evidence for a novel therapeutic approach to ameliorate the effects of an overly exuberant immune response in sepsis through tunable, broad spectrum removal of pathogens, immune cells (including activated neutrophils) and cytokines from peripheral blood using biomimetic cell margination28, 29 (Fig. 1A). We engineered a multiplexed microfluidic blood margination (µBM) device which can be placed in-line with the

circulation in an extracorporeal circuit for continuous blood filtration. To our knowledge, we demonstrate for the first time, non-specific removal of bacteria, cellular components

(platelets and leukocytes) and inflammatory mediators in a murine model of polymicrobial sepsis induced by cecal ligation and puncture (CLP). Remarkably, the µBM intervention resulted in significant long-term immunomodulation in immune cell counts (lower leukocyte and activated neutrophils count) and cytokine responses (lower IL-6, IL-10 and L-selectin) without adverse impact on survival. As cell margination is a passive separation technology and facilitates multiplexing, we scaled up the µBM device massively by channel

Author Manuscript

(4)

parallelization and device stacking to achieve ~100-fold higher throughput (~150mLhr−1) which is relevant for human intervention. With further multiplexing, the margination technique can be tested in clinical settings as an adjunctive treatment for sepsis and other inflammatory diseases.

Methods

Device fabrication

Microfluidic devices were fabricated in polydimethylsiloxane (PDMS) using standard microfabrication soft-lithographic techniques. Briefly, patterned silicon wafers were silanized with trichloro (1H,1H,2H,2H-perfluorooctyl) silane (Sigma Aldrich, St Louis, MO) and PDMS prepolymer (Sylgard 184, Dow Corning, Midland, MI) mixed in 10:1 (w/w) ratio with curing agent was poured onto the silanized wafer and cured at 80 °C for 1– 2 hrs. The cured PDMS mold was silanized and acted as a template (negative replica) for subsequent PDMS casting which gave the final PDMS microchannels. Holes (1.5 mm) for fluidic inlets and outlets were punched and the PDMS devices were irreversibly bonded to microscopic glass slides using an air plasma machine (Harrick Plasma Cleaner, Ithaca, NY).

In vitro characterization

Human whole blood (Research Blood Components, Brighton, MA) was washed and adjusted to physiological hematocrit (~45%) in sample buffer containing 1× PBS, 0.5% v/v bovine serum albumin (BSA) (Miltenyi Biotec, Auburn, CA) and 3% w/v Dextran 40 (Sigma Aldrich, St Louis, MO). For bacteria visualization, Alexa Fluor® 488 E. coli (K-12 strain) BioParticles® conjugates (Invitrogen, Carlsbad, CA) were added to blood samples (106–7_{/mL) and pumped through the µBM device using a syringe pump (PHD 2000, Harvard}

Apparatus, Holliston, MA) at 30 µLmin−1_{. A separate syringe pump was used for saline}

infusion at 10 µLmin−1_{. The device was mounted on an inverted phase contrast microscope}

(Olympus IX71) equipped with a Hamamatsu Model C4742-80-12AG CCD camera (Hamamatsu Photonics, Japan) for imaging of fluorescent bacteria. A high speed CCD camera (Phantom v9, Vision Research Inc., Wayne, NJ) was also used to capture blood cells flowing through the device and analyzed offline using ImageJ®_{software. To determine}

bacteria separation efficiency, the filtered centre outlet of the µBM device was collected for FACS analysis using Accuri™ C6 flow cytometer (BD Biosciences, San Jose, CA) and normalized with inlet sample. For whole blood analysis, whole blood was incubated at 4 °C for 30 minutes with FITC conjugated CD41a antibodies (1:50, BD Biosciences, San Jose, CA) and allophycocyanin (APC) conjugated CD45 marker (1:100, BD Biosciences, San Jose, CA) to identify platelets and leukocytes respectively. Separation efficiency was determined using flow cytometry with eluents from the filtered outlet and inlet sample. RBC enumeration was done with diluted blood samples from inlet and filtered outlet using hemacytometer.

Cecal ligation and puncture

6–8 week C57BL/6 male mice (~20 grams) were anesthetized with ketamine/Xylazine intraperitoneally (i.p.). With aseptic technique, a 2 cm laparotomy was performed. The cecum was exposed, its length measured from the ileocecal valve, and half of it was ligated

Author Manuscript

(5)

with two 6-0 sutures. At two millimeters distally to the ligation, the cecum was punctured once on the antimesenteric edge with a 19G needle (BD Biosciences, San Jose, CA). Approximately 2 mm of cecal content was gently squeezed through the puncture, and the cecum was placed back into the peritoneal cavity. The abdomen was closed in layers using 6-0 silk sutures (Ethicon, San Angelo, TX). The mice were resuscitated with 1 mL of PBS (Corning-Cellgro, Manasas, VA) i.p., put on a plate heater for recovering, and then returned to their cages with unrestricted access to water and chow. For sham animals, the same procedure was performed, with exception of the ligation and puncture of the cecum. All mice were maintained in a barrier animal facility, and all animal protocols were approved by the Harvard Medical Area Standing Committee on Animals, Harvard Medical School as well as Space and Naval Warfare Systems Center Pacific (SSC Pacific) Institutional Animal Care and Use Committee (IACUC).

Ex vivo characterization

At 24 hours after CLP, mice were anesthetized and exsanguinated through right ventricular puncture. Blood collected was stored in BD Microtainer® tube (BD, Franklin Lakes, NJ) and pumped through the device at 30 µLmin−1 using a peristaltic pump (P720, Instech Laboratories, Inc., Plymouth Meeting, PA). A separate syringe pump was used for saline infusion at 10 µLmin−1. Bacteria clearance was determined by plating diluted blood samples from baseline (pre-filtration), filtered (post-filtration) and side outlets on LB agar. Colony-forming units (CFUs) were counted after 24 hour incubation at 37 °C and normalized to baseline. For blood analysis, diluted mouse blood (1:5) was incubated at 4 °C for 30 minutes with CD41a-FITC (1:50, eBioscience, San Diego, CA) and CD45-APC (1:100, eBioscience, San Diego, CA) for quantification of platelets and leukocytes respectively. To determine neutrophil and lymphocyte concentrations, whole blood was lysed with 1× RBC lysis buffer (1:10, eBioscience, San Diego, CA) and washed twice with 1× PBS, followed by staining with CD45-APC (1:100, eBioscience, San Diego, CA). Leukocytes were then differentiated based on fluorescence signal and side scatters. For neutrophil activation assay, lysed blood was incubated with Ly-6G (Gr-1)-APC (1:100, eBioscience, San Diego, CA) and CD18-FITC (1:100, eBioscience, San Diego, CA) at 4 °C for 30 minutes. Activated neutrophils were gated based on up-regulation of both fluorescence signals which was pre-determined using healthy and PMA-stimulated neutrophils (data not shown).

In vivo characterization

Following approval of our institution animal care protocols and approval from all DARPA-affiliated agencies, 6–8 week C57BL/6 male mice (~20 grams) bearing right carotid and left internal jugular catheters (Charles River laboratories, Wilmington, MA) were subjected to the CLP procedure described above. 22 hours after CLP (2 hours prior to filtration), mice received 20 µL/g of body weight of saline subcutaneously and were randomly assigned to receive either µBM filtration (µBM, n = 14), pump filtration (extracorporeal filtration on the peristaltic pump without µBM device) (pump, n = 8) or no treatment (control, n = 8). All µBM devices were sterilized with ethanol, followed by heparin solution (50 U/mL) before use. 24 hours after CLP, mice were anesthetized with 2% isoflurane (Drager Vapor 19.1 vaporizer, Telford, PA) and saline (10 µL/g of body weight) was administered through the jugular catheter. Baseline blood sample (~80 µL) was obtained from mice through the

Author Manuscript

(6)

arterial catheter 2 minutes later and collected in Microtainer® tube. For µBM intervention, a µBM device was connected to the mouse via arterial and venous catheters under isoflurane anesthesia (maintained at 1%) and subjected to continuous filtration (30 µLmin−1, ~90 mL/kg/hr, ~25mins) using a peristaltic pump inside a fume hood. A separate syringe pump was used for saline infusion at 10 µLmin−1 and the device was continuously monitored on a portable microscope (SVM340, LabSmith, Livermore, CA) during the filtration. Mice in the pump-filtered group underwent similar intervention in an extracorporeal circuit without µBM devices. Tygon tubing (0.02” ID) was used in the extracorporeal circuit (dead volume less than 100 µL) and was connected to the implanted polyurethane catheter (3 Fr) using a stainless steel coupler (gauge 23). After filtration, a post-filtered blood sample (~80 µL) was immediately collected from the arterial catheter of µBM-filtered and pump-filtered mice. All mice (including control mice) were then given additional saline (~15 µL/g of body weight) intravenously and allowed to recover from anesthesia on a warming plate (37°C).

Bacteremia and blood analyses were performed similar to ex vivo characterization described above. For harvesting, either post filtration or at day 6, anesthetized mice were

exsanguinated through right ventricular puncture, and the blood was stored in EDTA microvacutainer. Plasma was obtained by centrifuge and frozen at −80°C for cytokine analysis (Aushon Biosystems Inc, Billerica, MA). The pulmonary circulation was perfused with 10 mL of PBS via the right ventricle, and lungs, liver, spleen, heart, and kidneys were collected for future analysis. For half-dose experiments, CLP mice were subjected to same intervention with shorter filtration time (~13 mins).

Statistical analysis

All numerical data were expressed as mean ± standard error (s.e.m.) unless specified otherwise. We assessed the statistical significance of the difference between two sets of data using Mann-Whitney test (unless specified otherwise) with P < 0.05 to be considered of significant difference. All analysis was performed using GraphPad Prism V5.0 (GraphPad Software, San Diego, CA).

Results

Microfluidic blood margination (µBM) device design

The multiplexed µBM device (Figure 1) is made of polydimethylsiloxane (PDMS) and consists of 3 channels; each 6 mm long, 20 µm × 40 µm (W×H) with 3 stages of bifurcation at 2 mm channel interval (Fig. 1B, Supplementary Fig. S1). A filter region with an array of square pillars (100 µm gap) is added upstream of the margination channels to trap large thrombi or cell clusters to ensure smooth flow within the device. As blood flows through the margination channel, deformable RBCs migrate axially towards the channel centre and mechanical collisions between the migrating RBCs and other cell types (bacteria, platelets and leukocytes) result in the lateral margination of larger and/or stiffer cells towards the side walls which are removed via the smaller side channels (Fig. 1C). To minimize RBC loss, the side channel skimming volume is progressively lower (15%, 10%, 5%) at each bifurcation stage with a total volume of ~30%. This is important since the filtered blood is returned to the animal directly and having more bifurcations may decrease the hematocrit significantly. Therefore, the µBM system is capable of extracting up to ~30% of plasma (hence reducing

Author Manuscript

(7)

plasma mediators such as cytokines) from the incoming blood in addition to cell separation by margination. During the extracorporeal blood filtration, the catheterized mouse is under isoflurane anaesthesia while blood is drawn from the mouse arterial (carotid) catheter using a peristaltic pump into the µBM device which separates the blood components and returns the filtered blood to the animal through the venous (jugular) catheter (Fig. 1D,E). A saline infusion port is also added to the filtered centre outlet to account for volume loss into the side channels. The simplicity in this extracorporeal circuit greatly minimizes tubing length which is a key feature in reducing dead volume (total blood volume of mouse is ~1 mL, dead volume ~100 µL) as well as coagulation issues due to blood contact with foreign surface.

Non-specific pathogen and inflammatory cell removal using µBM

As the µBM device was developed specifically for a murine experimental model, we characterized the platform using human whole blood at a flow rate of 30 µLmin−1 (90 mL/kg/hr) which is comparable to clinical human hemofiltration (45–60 mL/kg/hr). The addition of saline infusion (10 µLmin−1) to the filtered blood resulted in modest hematocrit decrease (~10%) since Fahreaus effect30 increased the initial hematocrit by >15% post margination (Supplementary Fig. S2). We next studied the cell margination effect in channels of different heights. As expected, the smaller but stiffer platelets and bacteria undergo similar level of margination in the µBM device of different aspect ratios (changing height), achieving ~70% removal efficiency in a 40 µm height system based on flow cytometry analysis (Supplementary Fig. S3). On the contrary, the efficiency of leukocyte margination decreased with increasing channel height which is consistent with previous work31, 32 and facilitates the adjustment of leukocyte removal efficiency based on particular therapeutic needs. While further studies are warranted to better understand the mechanisms governing margination for different target cells, these findings validated broad spectrum removal of pathogen and inflammatory cells (platelets and leukocytes) from human blood using the µBM device.

Ex vivo µBM characterization using CLP mice

To assess the immunological responses to sepsis, we used the CLP mouse model of polymicrobial sepsis that mirrors the course of sepsis in humans33, 34. In healthy, CLP and sham (laparotomy without cecal ligation or perforation) mice, immune cell counts were determined at multiple early timepoints (first 24 hours after CLP surgery) to track disease progression (Supplementary Fig. S4A). Neutrophil activation (upregulated Ly-6G (Gr-1) and CD18 expression) increased as early as 6 hr after CLP and continued to increase during the first 24hr. This surge in neutrophil activation in CLP-mice was accompanied by circulating leukocyte reduction (leukopenia), which suggests an on-going systemic

inflammatory response during this period (Supplementary Fig. S4B). In contrast, sham mice recovered from surgery, and their immune cell count and neutrophil activation returned to normal range after 12 hours. These results indicate that the host immune response to sepsis increased during the first 24 hours after CLP surgery (CLP-24hr) in this mouse model. Based on these data, we chose 24 hours as the timepoint for µBM intervention, as there was a sufficient inflammatory response and this was also a time that the mouse had sufficiently recovered from the CLP surgery so as to tolerate the µBM intervention35.

Author Manuscript

(8)

Using ex vivo blood samples obtained from healthy mice spiked with fluorescent E. Coli, we tested the µBM device and showed successive removal of marginated E. Coli into the side channels at each bifurcation stage (Supplementary Movie 1). The average fluorescence intensity of the central channel for filtered blood decreased by ~50% after 3 stages of cascaded separation (Fig. 2A). The ex vivo filtration performance was also corroborated using septic mice blood which achieved ~60% removal for all cell targets (Fig. 2B, C). Importantly, as cell margination removes all leukocytes based on physical differences (cell size and stiffness), activated neutrophils were significantly reduced (P < 0.05) which is an appealing target to mitigate tissue damage from release of oxygen radicals36 (Fig. 2D). We next measured cytokines and other pro-inflammatory molecules which are commonly upregulated during sepsis, including interleukin (IL)-6, IL-10, tumor necrosis factor-α (TNF-α) and L-selectin. Overall, an average of ~30% reduction in circulating plasma biomolecule concentrations was achieved in the post-filtered blood and was consistent with the plasma skimming volume (Fig. 2E). Interestingly, pro-inflammatory cytokines (IL-6 and TNF-α) increased for 1 mouse after µBM filtration. This is likely due to secretion from remaining immune cells at the filtered blood that became activated as they experienced high stress when passing through the device37. Nevertheless, the consistent presence of cytokines in the side outlets indicates that our µBM technique was able to deplete circulating cytokines by plasma skimming. Therefore, the ex vivo characterization using blood from CLP-24hr mice further validated the utility of this µBM technique in partial, non-specific removal of relevant cells and biomolecule targets that are potentially harmful in sepsis.

Continuous blood filtration in vivo using µBM in CLP mice

To assess the impact of filtration on host responses during the development of sepsis, µBM intervention was initiated at 24 hours after CLP surgery. The intervention time was ~25 minutes at 30 µLmin−1 which was chosen to be equivalent to the time required to process the circulating blood volume once (~750 µL). The mouse was maintained in a plane of light anesthesia using inhaled isoflurane to allow connection of the inserted arterial and venous catheters to the device and for blood circulation using the peristaltic pump. Extracorporeal blood filtration was well tolerated by septic mice with no observable complications and all mice recovered quickly (~minutes) from removal of anesthesia after the intervention was completed. To assess the filtration efficacy of µBM device, we sampled blood at baseline (pre-filtration) and immediately after the intervention (post-filtration) to quantify immune cell counts. As shown in Fig. 3A, µBM-filtered mice had higher removal of platelets and leukocytes (P < 0.05) as compared to pump-filtered mice (CLP-24hr mice on the peristaltic pump without µBM device), with similar reduction of activated neutrophils and leukocytes in individual µBM-filtered mice (Fig. 3B). The platelet count decreased slightly in pump-filtered mice which is consistent with other reports due to platelet adherence to the tubing24. The impact of µBM intervention on bacteria clearance was also determined, and animals were not administered antibiotics in these experiments specifically so that we could assess pathogen clearance by the device. Although not all CLP mice developed measurable bacteremia at 24 hours post CLP, when present, the bacteremia was successfully reduced in µBM-filtered mice (n=4) after intervention, and this was confirmed by side outlet bacterial count which was approximately twice baseline (P < 0.05) (Fig. 3C, Supplementary Fig. S5). Notably, these in vivo results were consistent with in vitro results (continuous filtration of 1

Author Manuscript

(9)

mL blood sample spiked with fluorescent bacteria) and standard Monod kinetics that predicted 30~40% decrease in bacteremia after 25 minutes of closed-loop filtration at 70% removal efficiency (Fig. 3D). We next measured IL-6, IL-10, TNF-α, and L-selectin at baseline (pre-filtration), immediately post-filtration, and in side outlets to assess cytokine removal efficiency in vivo using the µBM device. A slight decrease in cytokine levels (10– 15%) was observed in µBM-filtered mice after 25 minutes of filtration, and similar concentrations of cytokines were also detected in the side outlets, thus confirming cytokine removal from the animal (Fig. 3E). In contrast, pump-filtered mice had elevated cytokine levels after 25 minutes of intervention which suggests active production of cytokines in vivo during the intervention (Fig. 3F). Similar observations were also reported by Peng and colleagues22 and a possible explanation for the modest cytokine depletion in µBM-filtered mice could be the circulating cytokines in the blood stream were continuously “replenished” from injury sites or by immune cells sequestered in organs.

Long-term immunomodulation from µBM intervention

To evaluate the long-term effects of our cell-based therapeutic approach, we measured immune cell counts in µBM-filtered (CLP-24hr mice with µBM filtration), pump-filtered (CLP-24hr mice on the peristaltic pump without µBM device) and control mice (CLP-24hr mice that did not undergo any intervention) at 6 days post CLP (5 days after µBM

intervention) (Fig. 4A). A saline bolus (~ 300 µL) was given intravenously to all mice after intervention to assist their recovery and minimize the effect of fluid resuscitation in the long-term study of µBM intervention. No significant difference in long-term survival (6 days post CLP) was observed between groups, which suggest the safe usage of the µBM

intervention (Supplementary Fig. S6). Platelet counts were comparable between each group with µBM-filtered mice having slightly larger rebound in platelet count at day 6 compared with baseline than pump-filtered and control mice (Supplementary Fig. S7A). Interestingly, the leukocyte count remained significantly lower in µBM-filtered mice at day 6 (~2-fold lower) as compared to pump-filtered (P < 0.01) and control mice (P < 0.005) in which both leukocyte counts were higher than healthy mice and consistent with previous work38 (Fig. 4B). This trend was similar for both lymphocytes and neutrophils (Supplementary Fig. S7B), and the activated neutrophil count remained lower in µBM-filtered mice (P < 0.05),

suggesting significant attenuation of the immune response due to µBM intervention (Fig. 4C). In a subset of mice, we measured plasma cytokine levels at day 3 post CLP and observed a significant increase in pro-inflammatory IL-6 but not TNF-α in control CLP mice as compared to µBM-filtered and pump-filtered mice (P < 0.05). Anti-inflammatory IL-10 was higher in control group (44.6 ± 25.7 pg/mL) as compared to pump-filtered (12.2 ± 4.00 pg/mL) and µBM-filtered mice (8.47 ± 1.89 pg/mL) although not statistically

significant. L-selectin, a marker for inflammation, was also slightly higher in pump-filtered (6.84 ± 1.07 × 105 pg/mL) and control group (6.95 ± 2.35 × 105 pg/mL) than in µBM-filtered mice (5.55 ± 2.1 × 105 pg/mL) (Fig. 4D). When compared to fold-change in

cytokine level over baseline (before intervention), both µBM-filtered and pump-filtered mice had lower IL-6 levels and only control mice increased. TNF-α and IL-10 levels decreased significantly in all groups, suggesting predominant immunosuppression state by day 3 (Fig. 4E). For L-selectin, control mice had slightly higher fold change (1.56 ± 0.14) as compared to µBM-filtered (1.38 ± 0.06) and pump-filtered mice (1.33 ± 0.15). Lastly, we harvested

Author Manuscript

(10)

lungs from surviving mice at day 6 post CLP for immunohistochemistry and found reduced leukocyte (CD45) and neutrophil (GR-1) sequestration in the lung tissues of µBM-filtered compared with pump-filtered and control mice, indicating decreased lung inflammation with µBM intervention (Supplementary Fig. S8).

Dose-dependence of µBM intervention

For therapeutic applications, it is important to assess the “tunability” of the µBM

intervention in attenuating long-term inflammatory responses. Therefore, we sought to study the effects on immune cell response by halving the filtration time (‘dose’) to ~13 minutes for both µBM-filtered (µBM-0.5) and pump-filtered (pump-0.5) mice and measuring their cell counts at day 6 post CLP. As shown in Figure 5A and 5B, both leukocyte and activated neutrophil counts increased for µBM-0.5 mice as compared to µBM group mice (full dose, i.e. ~25 minutes of filtration) while pump-0.5 group unexpectedly had lower cell counts than pump group mice. Because CLP severity is highly heterogeneous among different mice even with the same surgical procedures, we normalized the cell count at day 6 to their individual baseline (day 1 before intervention) for more accurate assessment of the immune response. Our results showed that µBM-0.5 mice had higher rebound in platelet count (Supplementary Fig. S9), leukocyte count (2.22 ± 0.35) and activated neutrophil count (1.17 ± 0.39) as compared to µBM group mice (1.29 ± 0.19 for leukocyte and 0.48 ± 0.05 for activated neutrophils), suggesting that longer uBM intervention resulted in a more durable response. Importantly, fold-change in immune cell counts remained comparable between pump-0.5, pump and control group mice (~3.3–4.1 for leukocyte and ~1.90–2.3 for activated neutrophils) (Fig. 5C,D).

High throughput µBM intervention

As proof of concept for future human intervention, we scaled up our µBM platform through channel parallelization and device stacking to achieve ~100-fold higher throughput (150 mLhr−1_{) in a 32-channel margination platform (Fig. 6A). Unique design features include the}

radial arrangement of channels to allow a common single blood inlet and easy collection of filtered blood from all channels using drilled holes through the bottom glass slide (Fig. 6B). Each channel also functions independently with a wide range of working flow rates (40–80 µLmin−1_{) thus giving a more robust filtration performance. We validated the multiplexing}

principle with human whole blood and showed similar separation performance of platelets, leukocytes and bacteria between the 16-channel and single channel device (Supplementary Fig. S10). Using 50 mL of reconstituted human blood spiked with fluorescent bacteria, we further demonstrated continuous bacteria clearance at 90 mLhr−1_{using a 16-channel system}

and depleted bacteria concentration >50% in an hour which is in good agreement with the Monod kinetics (Fig. 6C). By stacking more layers of channel onto a single device, we envision further multiplexing to increase throughput by 5–10 fold (~1–2 Lhr−1_).

Discussion

Cell margination occurs naturally in microvasculature where deformable RBCs migrate to axial center of the vessel due to Poiseuille flow profile (Fahreaus effect), resulting in margination of stiffer cells, including platelets39, 40_{and leukocytes}41, 42_{towards the vessel}

Author Manuscript

(11)

wall. This margination process is crucial for the physiological functions of platelets and leukocytes: repairing damaged vessels and initiating extravasation through the vascular endothelium during inflammation, respectively. Inspired by this in vivo phenomenon, we previously developed a label-free separation approach for enrichment of malaria-infected red blood cells28_{and pathogens}29_{from human whole blood. To determine if cell}

margination can be used to mitigate overwhelming host immune response during sepsis, we herein developed a novel microfluidic blood margination (µBM) device for extracorporeal blood filtration and applied the µBM-based blood filtration therapy in a CLP mouse model that mirrors the bacteremia and cytokine release of human sepsis. The capacity of the µBM device was systemically validated ex vivo and in vivo with the mouse model of sepsis to achieve broad spectrum removal of pathogens, inflammatory cellular components and cytokines. Interestingly, significant long-term immunomodulation of immune cells, cytokine responses and end-organ leukocyte recruitment were observed in µBM-filtered mice several days after intervention, and a dose-dependent effect on immune cell count was also achieved by tuning µBM filtration time.

Given the complexity of sepsis pathophysiology and host response, broad spectrum immunomodulation is becoming increasingly important since there is currently no general consensus on which components of the blood should be removed for sepsis therapy43. During the early onset of sepsis typically characterized by pro-inflammatory phase (“cytokine storm”), extracorporeal blood purification processes targeting inflammatory mediators via membrane dialysis (hemofiltration) or adsorption (CytoSorb®) reduce blood-borne cytokines20, 21 and affect leukocyte trafficking by the modulation of chemokine gradient44. As the widespread activation of circulating leukocytes45, 46, inappropriate neutrophil apoptosis47 and platelet coagulation48 contribute to the pathogenesis of sepsis by impaired perfusion or hypoxia induced inflammation49, partial removal of circulating activated immune cells can also help to reduce organ inflammation50 and vascular

endothelial damage51, which is a rather under-studied therapeutic strategy. The importance of modulating immune cells in sepsis treatment is further evident in a mouse model of sepsis where administration of mesenchymal stem cells led to reprogramming of immune cells towards a less inflammatory phenotype, thereby reducing mortality and improving organ functions52. Hence, in addition to pathogens clearance and antibiotics treatment, it might be beneficial to remove immune cells and cytokines in a targeted and titratable manner from blood, as the removal of those inflammatory components could lead to better attenuation of host inflammatory response during sepsis than direct cytokine removal.

Microfluidic blood separation technologies have been developed for removal of single cell types including pathogens53–55 or leukocytes56, 57. Since cell margination enables non-specific separation of any cell type from whole blood based on physical properties, we implemented the technique in a novel blood filtration platform to simultaneously remove the stiffer bacteria/platelets and larger immune cells from the bulk RBCs. The developed µBM device offers significant advantages, including whole blood processing without any sample preparation, return of filtered blood directly to the animal, and non-labeled, non-specific removal of a broad range of blood-borne pathogens. Notably, the label-free separation is highly relevant in clinical settings as it enables non-specific removal of blood-borne

Author Manuscript

(12)

pathogens and is useful for early intervention and separation of unknown infecting species. Currently, the lack of understanding of sepsis pathophysiology makes it unclear to what degree immune cells should be removed to improve outcomes. As the CLP model results in varying degree of inflammation between mice, it is important to continuously monitor the inflammatory response in individual CLP mouse to identify a timepoint where the proinflammatory response predominates and our intervention would be most effective. In this study, the µBM device was operated at a flow rate of 30 µLmin−1 (~90 mL/kg/hr) which depleted pathogen and immune cells by ~20–40% in mice immediately after ~25–30 minute intervention. Even with this moderate filtration dose used, we observed the long-term potency of µBM technique in mitigation of circulating immune cells, cytokine level and end-organ leukocyte recruitment in septic mice. Furthermore, our dose-dependent results suggest the possibility of adjusting filtration time and frequency to achieve the desired amount of immunomodulation in mice. Such tunability would be critically important in sepsis

treatment where the immunological status and host response to infection varies significantly between individuals58.

In summary, we introduced a novel extracorporeal blood filtration approach (µBM) based on the biomimetic microcirculatory phenomenon of cell margination and validated the principle using microfluidics and a mouse model of sepsis. As proof of concept for human

intervention, the margination technique was successfully scaled up by ~100-fold, and we envision further multiplexing strategies to achieve higher throughput, comparable to convectional blood dialysis (~1–2 Lhr−1) used in clinical settings. Potential hazards include excessive removal of immune cells or cytokines which can be adjusted by varying the resistance of the side channels to collect lower plasma volume and marginated components. The µBM intervention could serve as an adjunctive immunomodulation therapy in addition to antibiotic treatment and current clinical care in managing patients with sepsis. While further studies will be needed to refine this technique and validate its effectiveness in larger animal models, our data supports the feasibility of an exciting and novel filtration approach in sepsis.

Supplementary Material

Refer to Web version on PubMed Central for supplementary material.

Acknowledgments

Financial support by DARPA’s Dialysis-like therapy (DLT) program under SSC Pacific grant N66001-11-1-4182 is gratefully acknowledged. This work is also supported by the use of MIT’s Microsystems Technology Laboratories. H.W.H is supported by Lee Kong Chian School of Medicine Postdoctoral Fellowship. J.A.E is supported by National Institutes of Health grant K08 GM102695.

References

1. Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, Pinsky MR. Crit. Care Med. 2001; 29:1303–1310. [PubMed: 11445675]

2. Schouten M, Wiersinga WJ, Levi M, van der Poll T. Journal of Leukocyte Biology. 2008; 83:536– 545. [PubMed: 18032692]

3. Cohen J. Nature. 2002; 420:885–891. [PubMed: 12490963]

Author Manuscript

(13)

4. Legrand M, Klijn E, Payen D, Ince C. Journal of Molecular Medicine. 2010; 88:127–133. [PubMed: 20119709]

5. Annane D, Bellissant E, Bollaert P, et al. JAMA. 2009; 301:2362–2375. [PubMed: 19509383] 6. Lefering R, Neugebauer EAM. Critical Care Medicine. 1995; 23:1294–1303. [PubMed: 7600840] 7. Annane D, Sébille V, Charpentier C, et al. JAMA. 2002; 288:862–871. [PubMed: 12186604] 8. Christaki E, Anyfanti P, Opal SM. Expert Review of Anti-infective Therapy. 2011; 9:1013–1033.

[PubMed: 22029521]

9. Monneret G, Venet F. Expert Review of Anti-infective Therapy. 2010; 8:1109–1112. [PubMed: 20954877]

10. Rimmelé T, Kellum J. Critical Care. 2011; 15:1–10.

11. Hoffmann JN, Hartl WH, Deppisch R, Faist E, Jochum M, Inthorn D. Kidney Int. 1995; 48:1563– 1570. [PubMed: 8544415]

12. Cole, Bellomo, Journois, Davenport, Baldwin, Tipping. Intensive Care Medicine. 2001; 27:978– 986. [PubMed: 11497156]

13. Hotchkiss RS, Karl IE. New England Journal of Medicine. 2003; 348:138–150. [PubMed: 12519925]

14. Abraham E, Wunderink R, Silverman H, et al. JAMA. 1995; 273:934–941. [PubMed: 7884952] 15. Fisher CJ Jr, Dhainaut JA, Opal SM, et al. JAMA. 1994; 271:1836–1843. [PubMed: 8196140] 16. Ranieri VM, Thompson BT, Barie PS, Dhainaut J-Fo, Douglas IS, Finfer S, Gårdlund B, Marshall

JC, Rhodes A, Artigas A, Payen D, Tenhunen J, Al-Khalidi HR, Thompson V, Janes J, Macias WL, Vangerow B, Williams MD. New England Journal of Medicine. 2012; 366:2055–2064. [PubMed: 22616830]

17. Cruz DN, Antonelli M, Fumagalli R, et al. JAMA. 2009; 301:2445–2452. [PubMed: 19531784] 18. Bengsch S, Boos K-S, Nagel D, Seidel D, Inthorn D. Shock. 2005; 23:494–500. [PubMed:

15897800]

19. Ziegler EJ, Fisher CJ, Sprung CL, Straube RC, Sadoff JC, Foulke GE, Wortel CH, Fink MP, Dellinger RP, Teng NNH, Allen IE, Berger HJ, Knatterud GL, LoBuglio AF, Smith CR. New England Journal of Medicine. 1991; 324:429–436. [PubMed: 1988827]

20. Kellum JA, Song M, Venkataraman R. Crit. Care Med. 2004; 32:801–805. [PubMed: 15090965] 21. Peng Z-Y, Carter MJ, Kellum JA. Crit. Care Med. 2008; 36:1573–1577. 1510.1097/CCM.

1570b1013e318170b318179a318177. [PubMed: 18434884]

22. Peng Z-Y, Wang H-Z, Carter MJ, Dileo MV, Bishop JV, Zhou F-H, Wen X-Y, Rimmele T, Singbartl K, Federspiel WJ, Clermont G, Kellum JA. Kidney Int. 2012; 81:363–369. [PubMed: 21918497]

23. Song SO, Hogg J, Peng Z-Y, Parker R, Kellum JA, Clermont G. PLoS Comput Biol. 2012; 8:e1002422. [PubMed: 22412365]

24. Rimmele T, Kaynar A, McLaughlin J, Bishop J, Fedorchak M, Chuasuwan A, Peng Z, Singbartl K, Frederick D, Zhu L, Carter M, Federspiel W, Zeevi A, Kellum J. Critical Care. 2013; 17:R59. [PubMed: 23531333]

25. Sauer M, Altrichter J, Kreutzer HJ, Lögters T, Scholz M, Nöldge-Schomburg G, Schmidt R, Mitzner SR. Crit. Care Med. 2009; 37:606–613. [PubMed: 19114920]

26. Altrichter J, Sauer M, Kaftan K, Birken T, Gloger D, Gloger M, Henschel J, Hickstein H, Klar E, Koball S, Pertschy A, Noldge-Schomburg G, Vagts D, Mitzner S. Critical Care. 2011; 15:R82. [PubMed: 21371308]

27. Kang JH, Super M, Yung CW, Cooper RM, Domansky K, Graveline AR, Mammoto T, Berthet JB, Tobin H, Cartwright MJ, Watters AL, Rottman M, Waterhouse A, Mammoto A, Gamini N, Rodas MJ, Kole A, Jiang A, Valentin TM, Diaz A, Takahashi K, Ingber DE. Nat Med. 2014; 20:1211– 1216. [PubMed: 25216635]

28. Hou HW, Bhagat AAS, Chong AGL, Mao P, Tan KSW, Han J, Lim CT. Lab on a Chip. 2010; 10:2605–2613. [PubMed: 20689864]

29. Hou HW, Gan HY, Bhagat AAS, Li LD, Lim CT, Han J. Biomicrofluidics. 2012; 6:024115– 024113.

Author Manuscript

(14)

30. Goldsmith HL, Cokelet GR, Gaehtgens P. American Journal of Physiology - Heart and Circulatory Physiology. 1989; 257:H1005–H1015.

31. Jain A, Munn LL. PLoS ONE. 2009; 4:e7104. [PubMed: 19768109]

32. Shevkoplyas SS, Yoshida T, Munn LL, Bitensky MW. Analytical Chemistry. 2005; 77:933–937. [PubMed: 15679363]

33. Wichterman KA, Baue AE, Chaudry IH. Journal of Surgical Research. 1980; 29:189–201. [PubMed: 6997619]

34. Buras JA, Holzmann B, Sitkovsky M. Nat Rev Drug Discov. 2005; 4:854–865. [PubMed: 16224456]

35. Villa P, Sartor G, Angelini M, Sironi M, Conni M, Gnocchi P, Isetta AM, Grau G, Buurman W, van Tits LJ. Clinical and Diagnostic Laboratory Immunology. 1995; 2:549–553. [PubMed: 8548533]

36. Hoesel LM, Neff TA, Neff SB, Younger JG, Olle EW, Gao H, Pianko MJ, Bernacki KD, Sarma JV, Ward PA. Shock. 2005; 24:40–47. [PubMed: 15988319]

37. Mantovani A, Cassatella MA, Costantini C, Jaillon Sb. Nat Rev Immunol. 2011; 11:519–531. [PubMed: 21785456]

38. Xiao H, Siddiqui J, Remick DG. Infection and Immunity. 2006; 74:5227–5235. [PubMed: 16926416]

39. Aarts P, van den Broek S, Prins G, Kuiken G, Sixma J, Heethaar R. Arteriosclerosis, Thrombosis, and Vascular Biology. 1988; 8:819–824.

40. Zhao R, Kameneva MV, Antaki JF. Biorheology. 2007; 44:161–177. [PubMed: 17851165] 41. Fiebig E, Ley K, Arfors KE. Int. J. Microcirc.-Clin. Exp. 1991; 10:127–144. [PubMed: 2060996] 42. Goldsmith HL, Spain S. Microvasc. Res. 1984; 27:204–222. [PubMed: 6708830]

43. Cohen J, Opal S, Calandra T. The Lancet Infectious Diseases. 2012; 12:503–505. [PubMed: 22742624]

44. Peng Z-Y, Bishop J, Wen X-Y, Elder M, Zhou F, Chuasuwan A, Carter M, Devlin J, Kaynar A, Singbartl K, Pike F, Parker R, Clermont G, Federspiel W, Kellum J. Critical Care. 2014; 18:R141. [PubMed: 24992991]

45. Brown KA, Brain SD, Pearson JD, Edgeworth JD, Lewis SM, Treacher DF. The Lancet. 2006; 368:157–169.

46. Muller Kobold AC, Tulleken JE, Zijlstra JG, Sluiter W, Hermans J, Kallenberg CGM, Cohen Tervaert JW. Intensive Care Medicine. 2000; 26:883–892. [PubMed: 10990102]

47. El Kebir D, Gjorstrup P, Filep JG. Proceedings of the National Academy of Sciences. 2012; 109:14983–14988.

48. Mavrommatis AC, Theodoridis T, Orfanidou A, Roussos C, Christopoulou-Kokkinou V, Zakynthinos S. Critical Care Medicine. 2000; 28:451–457. [PubMed: 10708182]

49. Eltzschig HK, Carmeliet P. New England Journal of Medicine. 2011; 364:656–665. [PubMed: 21323543]

50. Ding F, Song JH, Jung JY, Lou L, Wang M, Charles L, Westover A, Smith PL, Pino CJ, Buffington DA, Humes HD. PLoS ONE. 2011; 6:e18584. [PubMed: 21533222]

51. Kumagai T, Takeyama N, Yabuki T, Harada M, Miki Y, Kanou H, Inoue S, Nakagawa T, Noguchi H. Shock. 2010; 34:461–466. [PubMed: 20386499]

52. Nemeth K, Leelahavanichkul A, Yuen PST, Mayer B, Parmelee A, Doi K, Robey PG,

Leelahavanichkul K, Koller BH, Brown JM, Hu X, Jelinek I, Star RA, Mezey E. Nat Med. 2009; 15:42–49. [PubMed: 19098906]

53. Xia N, Hunt T, Mayers B, Alsberg E, Whitesides G, Westervelt R, Ingber D. Biomedical Microdevices. 2006; 8:299–308. [PubMed: 17003962]

54. Yung CW, Fiering J, Mueller AJ, Ingber DE. Lab on a Chip. 2009; 9:1171–1177. [PubMed: 19370233]

55. Lee J-J, Jeong KJ, Hashimoto M, Kwon AH, Rwei A, Shankarappa SA, Tsui JH, Kohane DS. Nano Letters. 2013

56. Shin JH, Lee MG, Choi S, Park J-K. RSC Advances. 2014; 4:39140–39144.

Author Manuscript

(15)

57. Kotz KT, Xiao W, Miller-Graziano C, Qian W-J, Russom A, Warner EA, Moldawer LL, De A, Bankey PE, Petritis BO, Camp DG, Rosenbach AE, Goverman J, Fagan SP, Brownstein BH, Irimia D, Xu W, Wilhelmy J, Mindrinos MN, Smith RD, Davis RW, Tompkins RG, Toner M. Nat Med. 2010; 16:1042–1047. [PubMed: 20802500]

58. Hotchkiss RS, Sherwood ER. Science. 2015; 347:1201–1202. [PubMed: 25766219]

Author Manuscript

(16)

Figure 1.

Broad spectrum immunomodulation using biomimetic cell margination. (A) Overview of blood cell margination in microfluidics. (Cross-sectional view) Deformable RBCs migrate to axial center of microchannel during flow which results in size and deformability-based margination of pathogen and inflammatory cellular targets (platelets and leukocytes) towards the channel wall. (B) Schematic illustration and image (filled with red dye) of the microfluidic blood margination (µBM) device developed for blood filtration in mice. (C) Separation principle at each bifurcation stage (yellow box in B). Plasma and marginated cell components are removed via the side outlets while filtered blood at the channel center (RBCs enriched) is returned to the animal. (D) Schematic representation of the

extracorporeal circuit for continuous blood filtration in a catheterized mouse. A peristaltic pump is used to pump blood from the carotid artery into µBM device and filtered blood is returned to the animal via jugular vein. Saline infusion is added at the filtered outlet to account for volume loss (~30% into side outlets) during filtration. (E) Extracorporeal µBM setup for a catheterized mouse under isoflurane anaesthesia.

Author Manuscript

(17)

Figure 2.

Ex vivo characterization of µBM device using blood from mice which underwent cecal ligation and puncture (CLP) after 24 hours. (A) Average composite images and normalized fluorescence intensity plot (channel central region) illustrating margination and successive removal of FITC-labeled E. coli spiked in whole blood at each bifurcation stage. (B) High speed images illustrate successive plasma skimming into the smaller side channels. (C) Removal efficiency of cellular targets using the developed µBM device (40 µm channel height). Mean ± s.e.m. from n = 4 mice. (D) Significant reduction of activated neutrophils in filtered blood collected from the centre outlet. Mean ± s.e.m. from n = 6 mice. *P < 0.05 by paired t-test. (E) Overall decrease (~30%) in plasma-borne cytokines post filtration. *P < 0.05 vs. pre-filtration. Mean ± s.e.m. from n = 4 mice.

Author Manuscript

(18)

Figure 3.

Continuous blood filtration of septic mice (24 hours after CLP) in an extracorporeal circuit (~25 mins at 30 µLmin−1). (A) Higher removal of platelets and leukocytes in µBM-filtered mice as compared to pump-filtered mice (CLP-24hr mice on the peristaltic pump without µBM device). Each cell count post filtration is normalized to baseline (before filtration) and ratio of 1 (red dotted line) indicates no change. n = 8 and 6 for µBM-filtered and pump-filtered mice, respectively. *P < 0.05 (B) Similar removal of leukocytes and activated neutrophils in each mouse and overall removal efficiency (right inset) after µBM

Author Manuscript

(19)

intervention. (C) Bacteremia at baseline (pre-filtration), after µBM intervention (post filtration) and side outlets of device. Mean ± s.e.m. from n = 4 mice. *P < 0.05 versus pre- and post-filtration. (D) In vivo and in vitro (continuous filtration of 1mL of blood at 30 µLmin−1) experimental results of bacteria removal were similar to standard Monod kinetics model (70% removal efficiency). In vitro data represent mean ± s.d. from n ≥ 3 separate experiments. (E) Cytokine concentrations at baseline (pre-filtration), after µBM intervention (post filtration) and side outlets of device. Mean from n = 4 mice. (F) Normalized

concentration of cytokine removal for µBM-filtered and pump-filtered mice. Mean ± s.e.m. from n = 4 mice for each group. Normalized value of 1 (red dotted line) indicates no change after filtration.

Author Manuscript

(20)

Figure 4.

Effect of µBM intervention on long term immune cell and cytokine responses. (A)

Experimental timeline (6 days post CLP) involving 3 groups of mice: 1) µBM-filtered mice, 2) pump-filtered mice (CLP-24hr mice on the peristaltic pump without µBM device) and 3) control mice (CLP-24hr mice which did not undergo any intervention). Significant

attenuation of (B) leukocyte and (C) activated neutrophil count in µBM-filtered mice as compared to pump-filtered and control mice at day 6. Purple dotted line and shaded region indicate mean ± s.e.m for healthy mice. Mean ± s.e.m from n = 8–14 mice in each group. *P < 0.05, **P < 0.01 and ***P < 0.005. (D) Absolute cytokine concentration and (E) fold change in cytokine level at day 3 when normalized to baseline (before intervention). Mean ± s.e.m from n = 5–6 mice in each group. *P < 0.05.

Author Manuscript

(21)

Figure 5.

Dose-dependent long term effect of µBM intervention on immune cell response. (A) Leukocyte and (B) activated neutrophil count at day 6. µBM-0.5 and Pump-0.5 groups correspond to mice on shorter intervention time (~13 mins). Purple dotted line and shaded region indicate mean ± s.e.m for healthy mice. Mean ± s.e.m from n = 8–14 mice in each group. Fold change in (C) leukocytes and (D) activated neutrophils when normalized to baseline (day 1 before intervention). Normalized value of 1(purple dotted line) indicates no change.

Author Manuscript

(22)

Figure 6.

Scaling up µBM technology for (ultra)high throughput. (A) Multiplexed margination platform using channel parallelization and device stacking. (B) Optical image of a dual-stacked margination device capable of processing whole blood at 150 mLhr−1. (C) Decrease in bacteria concentration in 50 mL sample at 90 mLhr−1 using a 16-channel device. Mean ± s.d. from n ≥ 3 separate experiments.