WO2016040850A1 - Microfluidic device for removal of constituents from blood - Google Patents

Abstract

The present disclosure concerns a microfluidic device comprising microchannels that have a surface, and a sorbent molecule attached to that surface. In some embodiments, the sorbent molecule is a peptide. In some embodiments, the microchannels comprise an array of offset microchannels. The sorbent molecule may be attached to the surface through a polymer, such as a triblock polymer. As a fluid, such as blood, flows through the device, the sorbent molecule captures constituents from the fluid, and prevents them from being transported downstream. Typically, the sorbent molecule does not destabilize the constituent so that the constituent does not disintegrate causing fragments of constituent return to the fluid stream.

Description

MICROFLUIDIC DEVICE FOR REMOVAL OF CONSTITUENTS FROM BLOOD

CROSS REFERENCE TO RELATED APPLICATIONS

This application claims the benefit of the earlier filing date of U.S. provisional patent application No. 62/049,296, filed September 11, 2014, which is incorporated herein by reference in its entirety.

FIELD

The present disclosure concerns a microfluidic device comprising at least one

microchannel having a surface coating capable of removing constituents from a blood stream.

ACKNOWLEDGMENT OF GOVERNMENT SUPPORT

This invention was made with government support under Grant No. R01EB011567 awarded by the National Institute of Biomedical Imaging and Bioengineering (NIB IB), and Grant No. 1414400 awarded by the National Science Foundation. The government has certain rights in the invention.

BACKGROUND

There are a wide variety of medical conditions that can be treated by passing blood or plasma through an extracorporeal device to remove specific blood constituents. For example, many autoimmune diseases are mediated by circulating antibodies directed against self-antigens.

Removal of these autoantibodies has proved beneficial for treatment of several diseases, including rheumatoid arthritis, dilated cardiomyopathy, myasthenia gravis, Guillain-Barre syndrome and systemic lupus erythematosus.

Acute rejection following organ transplant can also be mediated by circulating antibodies. Consequently, removal of antibodies is being investigated as a strategy for improving the outcome of ABO incompatible organ transplant procedures.

Numerous other medical conditions can also benefit from devices that provide targeted removal of specific blood constituents. Therefore, improvements in the operation and function of such devices are highly desirable.

SUMMARY

Disclosed herein are embodiments of a device that can remove specific blood

constituents with minimal damage to blood cells and without evoking a host cell response. In some embodiments, the device comprises a first lamina comprising a first plurality of microchannels, with the microchannels having a first microchannel surface, and a first sorbent molecule attached to the first microchannel surface. The microchannels may comprise an array of offset microchannels that are defined by a plurality of offset microchannel walls. In some examples, the offset microchannels have a bifurcation length of from about 50 μιη to about 1 cm, preferably from about 250 μιη to about 5 mm, more preferably from about 500 μιη to about 2 mm. In some embodiments, the offset microchannels have a first bifurcation length of from about 50 μιη to about 1 cm, and a second bifurcation length of from about 50 μιη to about 1 cm, different from the first bifurcation length.

In some embodiments, each microchannel has a microchannel width of from about 10 μιη to about 750 μιη, such as from about 25 μιη to 500 μιη, or from about 50 μιη to about 200 μιη. In other embodiments, each microchannel has a microchannel depth of from about 25 μιη to about 1 mm, such as from about 100 μιη to about 500 μιη.

In some embodiments, the microchannels are defined by microchannel walls that have a substantially rectangular base shape. In some examples, the substantially rectangular base shape has at least one rounded end and/or at least one triangular end.

The sorbent molecule may be a biological molecule, and/or may be a bioactive molecule. In some embodiments, the bioactive molecule is selected from an intact or fragmentary portion of a peptide, protein, enzyme, antibody, antibody fragment, aptamer, single strand DNA, single strand RNA, double-stranded DNA, double-stranded RNA, polysaccharide, glycosaminoglycan, lipid, phospholipid, chelator, antibiotic, synthetic polymer, anticoagulant, anticlotting agent or a combination thereof. The bioactive molecule may be attached to the microchannel surface in such a way as to substantially preserve its solvent accessibility and mobility.

The sorbent molecule may be attached to the surface via a polymer, and in some embodiments, the polymer comprises at least one unsaturated carbon-carbon bond. The polymer may be an amphiphilic diblock or triblock copolymer, and/or may comprise polyethylene oxide, polypropylene oxide, polytetrahydrofuran, polybutadiene, polystyrene, polypentadiene, polyhexadiene, polyacrylonitrile, polyhydroxyethylmethacrylate, polyurethane, polyacrylamide, or a combination thereof. In some embodiments, the polymer comprises polybutadiene, polypentadiene or polyhexadiene, and in particular embodiments, the polymer is selected from poly(ethylene oxide)-polybutadiene-poly(ethylene oxide), poly(tetrahydrofuran)-polybutadiene- poly(tetrahydrofuran), polyhydroxyethylmethacrylate -polybutadiene- polyhydroxyethylmethacrylate, or polyacrylamide -polybutadiene-polyacrylamide. In some embodiments, the sorbent molecule captures a constituent from blood, and in some examples, the constituent is a bacterium, an endotoxin, a cytokine, an antibody or a combination thereof. The sorbent molecule may be a peptide, and in some embodiments, it is an antimicrobial peptide and may substantially retain its antimicrobial activity after attachment to the surface or the polymer. In some examples, the peptide comprises a short hydrophobic sequence at the non-attached or free end of the sequence. The peptide may insert into a membrane of a blood constituent, and in some examples, inserts into the membrane without substantially disrupting the membrane for a length of time of operation of the device. In particular examples, the peptide is polymyxin B (PmB), nisin, WLBU2, WR12, or a mutant or structural variant thereof.

The sorbent molecule may be an antibody, and in some embodiments, the sorbent molecule is one or more antibodies against pro-inflammatory cytokines. In certain examples, the antibody is anti-TNF-a or anti-IL-1.

In some embodiments, the sorbent molecule is an anticoagulant and/or anticlotting agent, and in certain examples, the sorbent molecule is thrombomodulin, heparin, fondaparninux, idraparinux, heparan sulfate, Coumadin, rivaroxaban, apixaban, edoxaban, hirudin, lepirudin, bivalirudin or structural variations or derivatives thereof.

In some embodiments, blood flows through the microchannels during operation of the device, and may flow at a sufficient rate such that about the volume of blood in a human body can pass through the device in 60 minutes or less, preferably in 30 minutes or less. In other embodiments, the blood flows at a rate of from about 25 milliliters per minute to about 250 milliliters per minute, preferably from about 75 milliliters per minute to about 150 milliliters per minute, and in some examples, the rate of blood flow is greater than 100 milliliters per minute.

In some embodiments, the device further comprises a second lamina comprising a second plurality of microchannels, with the microchannels having a second microchannel surface, and a second sorbent molecule attached to the second microchannel surface. The first sorbent molecule may be the same as the second sorbent molecule, or it may be different. In some embodiments, a third sorbent molecule is attached to the first microchannel surface, the second microchannel surface or both. The third sorbent molecule may be different from the first sorbent molecule and the second sorbent molecule.

In some embodiments, the device comprises a first plurality of laminae, with each lamina comprising a first plurality of microchannels, the microchannels having a first microchannel surface, and where a first sorbent molecule is attached to the first microchannel surface. The device further comprises a second plurality of laminae, with each lamina comprising a second plurality of microchannels, the microchannels having a second microchannel surface, and where a second sorbent molecule is attached to the second microchannel surface. The first sorbent molecule and the second sorbent molecule may be different or they may be the same. In some examples, the first plurality of lamina, the second plurality of laminae, or both, comprise from 2 to 1000 laminae, such as from 50 to 500 laminae or from 100 to 250 laminae. In some embodiments, the device further comprises a third sorbent molecule, wherein the third sorbent molecule is attached to the first microchannel surface, the second microchannel surface, or both.

The device may further comprise a valve, where the valve is connected to the first plurality of microchannels and the second plurality of microchannels such that during operation of the device a fluid is directed to the first plurality of microchannels, the second plurality of microchannels, or both. The fluid may be sequentially directed to the first plurality of microchannels and then the second plurality of microchannels, or the valve may direct the fluid to both the first and second pluralities of microchannels substantially simultaneously.

Alternatively, the fluid may be directed to the first plurality of microchannels, the first and second pluralities of microchannels simultaneously, and then the second plurality of

microchannels only.

Also disclosed herein is a method of making the device, comprising providing a lamina comprising a plurality of microchannels, and exposing the microchannels to a sorbent molecule. The method may further comprise contacting the microchannels with a polymer, and in some examples, the polymer may be exposed to radiation, such as gamma radiation. In other embodiments, the sorbent molecule is bound to a polymer.

A method of using the device is also disclosed herein. The method comprises providing an embodiment of the disclosed device, providing blood comprising a constituent, and fluidly coupling the device to the blood such that the blood flows through the device during operation and the constituent is at least partially removed from the blood. In some embodiments, fluidly coupling the device to the blood comprises fluidly coupling the device valve to the blood, such that the blood flows through the first plurality of microchannels, the second plurality of microchannels or both, during operation. Also disclosed is a method of removing a constituent from a blood stream, comprising providing a device comprising a plurality of offset

microchannel walls that define at least one microchannel having a microchannel surface, and a sorbent molecule associated with the microchannel surface; and fluidly coupling the blood stream to flow through a device comprising a lamina comprising a plurality of offset

microchannel walls that define at least one microchannel having a microchannel surface, and a sorbent molecule associated with the microchannel surface, whereby the sorbent molecule removes a constituent from the blood stream. In some embodiments, the constituent is a lipopolysaccharide.

The foregoing and other objects, features, and advantages of the invention will become more apparent from the following detailed description, which proceeds with reference to the accompanying figures.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 provides assembled and exploded views of an exemplary microfluidic device according to the present disclosure, comprising a plurality of laminae.

FIG. 2 is a schematic diagram illustrating a valve directing fluid flow to different groups of laminae.

FIG. 3 is a schematic diagram illustrating shear-induced red blood cell (RBC) migration to the center of the microchannels.

FIG. 4 is a schematic diagram predicting the RBC volume fraction in a 50 μιη x 50 μιη microchannel, with a flow velocity showing the development of a cell-free layer near the channel wall.

FIG. 5 is a schematic diagram predicting the RBC volume fraction in an offset microchannel array with a microchannel cross-section of 100 μιη x 100 μιη and flow from bottom to top, showing disruption and reestablishment of the steady-state RBC profile.

FIG. 6 is a plan view of an exemplary lamina according to the disclosed embodiments, illustrating an array of offset microchannels.

FIG. 7 provides three exemplary geometries suitable for use at the bifurcations at the ends of the microchannel walls.

FIG. 8 is a schematic diagram illustrating one exemplary embodiment where the width in each branch of the bifurcation is different.

FIG. 9 is a schematic diagram illustrating an exemplary polyethylene oxide (PEO) brush surface layer with various tethered sorbents attached.

FIG. 10 is a schematic diagram providing one possible strategy for covalently linking PEO brush layers having various different chain lengths to a polymer surface.

FIG. 11 provides a schematic illustration of WLBU2 at an interface in an entrapped (left) or tethered (right) motif.

FIG. 12 is a graph of frequency versus time indicating the amount of association of lipopolysaccharide (LPS) on various coated or non-coated Au surfaces.

FIG. 13 is a graph of dissipation versus frequency for the surfaces from FIG. 11. FIG. 14 provides circular dichroism (CD) spectra of WLBU2 non- specifically bound to a hydrophobic surface before and after LPS interaction.

FIG. 15 is a graph of frequency (top line) and dissipation (bottom line) versus time for LPS on a surface containing covalently attached PEO polymer only.

FIG. 16 is a graph of frequency and dissipation versus time for LPS on a surface containing WLBU2 entrapped in a PEO layer.

FIG. 17 provides CD spectra of entrapped WLBU2 on hydrophobic nanoparticles mixed with 0.1 mg/mL LPS.

FIG. 18 is a schematic diagram illustrating covalent association of Cys-WLBU2 with EGAP-PDS to create EGAP-WLBU2.

FIG. 19 is a schematic diagram of WLBU2 interaction with LSP vesicles.

FIG. 20 provides CD spectra of EGAP-WLBU2 mixed with 0.1 mg/mL LPS.

FIG. 21 provides CD spectra of EGAP-WLBU2 on hydrophobic nanoparticles mixed with 0.1 mg/mL LPS.

FIG. 22 provides NMR spectra of non-irradiated WLBU2 and 0.3 Mrad γ-irradiated

WLBU2.

FIG. 23 provides UV/Vis spectra of non-irradiated WLBU2 and 0.3 Mrad γ-irradiated WLBU2.

FIG. 24 provides CD spectra of non-irradiated WLBU2 and 0.3 Mrad γ-irradiated WLBU2.

FIG. 25 is a graph of frequency and dissipation versus time for LPS on a surface containing covalently attached EGAP-WLBU2.

FIG. 26 is a schematic diagram illustrating one possible explanation of how entrapped WLBU2 is able to "capture" LPS.

FIG. 27 is a graph of frequency (top) and dissipation (bottom) for fibrinogen on a surface containing covalently bound PEO polymer only.

FIG. 28 is a graph of frequency and dissipation for a mixture of fibrinogen and LPS on a surface containing entrapped WLBU2.

FIG. 29 is a graph of frequency and dissipation for a mixture of fibrinogen and LPS on a surface containing covalently attached EGAP-WLBU2.

FIG. 30 is a graph of frequency versus time for entrapped WLBU2 and tethered WLBU2 challenged with fibrinogen.

FIG. 31 is a graph of dissipation versus frequency for entrapped WLBU2 and tethered WLBU2 challenged with fibrinogen. FIG. 32 is a graph of frequency/dissipation versus time of LPS, fibrinogen and a fibrinogen/LPS mixture on surfaces with entrapped WLBU2.

FIG. 33 is a graph of frequency/dissipation versus time of LPS, fibrinogen and a fibrinogen/LPS mixture on surfaces with tethered WLBU2.

FIG. 34 is a graph of surface tension versus time illustrating the air-water tensiometry of suspensions of various concentrations of PmB and LPS in PBS as individual species, with the standard deviation (n=5) shown for LPS.

FIG. 35 is a graph of surface tension versus time illustrating the air-water tensiometry of suspensions of various concentrations of PmB and LPS in PBS as mixtures of peptide and LPS, with the standard deviation (n=5) shown for LPS.

FIG. 36 is a graph of surface tension versus time illustrating the air-water tensiometry of suspensions of various concentrations of WLBU2 and LPS in PBS as individual species, with the standard deviation (n=5) shown for LPS.

FIG. 37 is a graph of surface tension versus time, illustrating the air- water tensiometry of suspensions of various concentrations of WLBU2 and LPS in PBS as mixtures of peptide and LPS, with the standard deviation (n=5) shown for LPS.

FIG. 38 provides the molecular structure and approximate dimensions of PmB and helical form of WLBU2 peptides.

FIG. 39 is a graph of adsorbed mass versus time illustrating optical waveguide lightmode spectroscopy (OWLS) kinetic data for competitive adsorption from mixtures of LPS and peptide concentrations.

FIG. 40 provides CD spectra of 50 μιη WLBU2 in PBS, with helix-inducing perchlorate ions or in the presence of LPS vesicles.

FIG. 41 provides CD spectra of 50 μιη PmB in PBS, with helix-inducing perchlorate ions or in the presence of LPS vesicles.

FIG. 42 is a graph of probability of diameter size at or below X-value versus diameter, illustrating the cumulative oversize distribution of particle diameter in peptide-LBS suspensions from dynamic light scattering (DLS).

FIG. 43 provides photographs illustrating visible aggregation rapidly occurring in concentrated mixtures of WLBU2 and LPS (top) but not in PmB-LPS (middle) or peptide-free LPS suspensions (bottom).

FIGS. 44A and 44B provides the synthetic pathway used to make an exemplary "trident" construct PEO polymer. FIG. 45 is a graph of endotoxin capture versus time, illustrating the capture and elution of endotoxins from immobilized polymers.

DETAILED DESCRIPTION

The following explanations of terms and abbreviations are provided to better describe the present disclosure and to guide those of ordinary skill in the art in the practice of the present disclosure. As used herein, "comprising" means "including" and the singular forms "a" or "an" or "the" include plural references unless the context clearly dictates otherwise. The term "or" refers to a single element of stated alternative elements or a combination of two or more elements, unless the context clearly indicates otherwise.

Unless explained otherwise, all technical and scientific terms used herein have the same meaning as commonly understood to one of ordinary skill in the art to which this disclosure belongs. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present disclosure, suitable methods and materials are described below. The materials, methods, and examples are illustrative only and not intended to be limiting. Other features of the disclosure are apparent from the following detailed description and the claims.

"Providing a device" or "providing the device" refers to a manufacturer who makes the device and provides instructions for its use, establishing the manner and timing of using the device; a supplier who supplier the device and provides instructions for its use, establishing the manner and timing of using the device; a facility, such as a hospital or clinic, that uses the device to treat a subject; and/or a subject who uses the device themselves. The manufacturer, supplier, facility and/or subject may act jointly or as a joint enterprise by agreement, by a common purpose, a community of pecuniary interest, and/or equal say in direction of using the device. Alternatively, or additionally, the manufacturer, supplier, facility and/or subject may condition participation in an activity or receipt of a benefit upon performance of a step or steps of the method of using the device disclosed herein, and establish the manner or timing of that performance. I. Overview

There are a wide variety of medical conditions that can be treated by passing blood or plasma through an extracorporeal device containing a sorbent to remove specific blood constituents. For instance, many autoimmune diseases are mediated by circulating antibodies directed against self-antigens. Removal of these autoantibodies has proved beneficial for treatment of several diseases, including rheumatoid arthritis, dilated cardiomyopathy, myasthenia gravis, Guillain-Barre syndrome and systemic lupus erythematosus. Acute rejection following organ transplant is also mediated by circulating antibodies and consequently removal of antibodies is being investigated as a strategy for improving the outcome of ABO incompatible organ transplant procedures. Sorbent-based approaches can also be relevant to the treatment of high cholesterol, poisoning, kidney failure and sepsis.

As discussed in more detail below, successful hemoperfusion for sepsis treatment requires surface modification that will ensure highly selective capture of bacteria and endotoxin that reach the interface. In addition, surface coatings must provide constituent binding functionality without evoking a host cell response, without nonspecific adsorption of protein, and without platelet activation and blood cell damage caused by cell- surface interactions.

II. Microfluidic Device

Disclosed herein are embodiments of a microfluidic device that is tailored to minimize damage to blood cells and enhance adsorption at the device surface. While microfluidic devices typically operate at low flow rates, clinically relevant flow rates can be achieved using a "number up" approach. To increase the volumetric throughput, a single device typically comprises a stack of laminae operated in parallel, as illustrated in FIG. 1. Laminae 10 can be stacked upon each other and bonded together, and a first end member 12 and a second end member 14 can be provided to allow passage of the fluid into and out of the array of

microchannels for treatment of the fluid. In this manner, microfluidic device 16 can

significantly increase volumetric throughput. This "number up" strategy for device scale up is advantageous in that the full scale devices retain the same microscale control over fluid flow and mass transfer as the small scale devices. In some embodiments, the laminae are stacked together without a membrane between them. If desired, a plurality of microfluidic devices 16 can be combined to form a bank of devices 16 to allow even greater fluid flow through the system.

In some embodiments, the device also comprises at least one valve, fluidly coupled to the laminae stack (FIG. 2). In some embodiments, the valve or valves 20 direct the flow of the fluid to one or more of a plurality groups of laminae 22. In FIG. 2, two groups 24 and 26 are shown for illustrative purposes, and the arrows 28 and 30 illustrate the flow of the fluid to these groups. This enables a first portion of a fluid stream to be directed to a first group of laminae

24, and a second portion of the fluid stream to be directed to an unused second group of laminae

26, thus minimizing the exposure of the second portion of the fluid stream to constituents of the fluid stream that have been removed from the first portion. For example, if the fluid is blood, and the constituent being removed is a pathogen, the second portion of the blood stream is not exposed to the pathogen already removed from the first portion of the blood stream. A person of ordinary skill in the art will appreciate that a device may have more than two groups of laminae, and that the valve or valves can direct the flow of the fluid to any one of the groups, any combination of the groups, or all the groups at the same time.

The microfluidic device architecture leverages the natural tendency of red blood cells (RBCs) to migrate to the center of microchannels in order to enhance adsorption of other blood constituents (e.g., endotoxin or antibodies) at the device wall. The importance of RBC migration is well recognized in human physiology. In the microvasculature, RBCs migrate toward the center of the vessel, which in turn causes the other constituents of blood to become enriched in the margins of the flow near the vessel wall. This process is known as

"margination." Enrichment of the non-RBC components of blood near the vessel wall is critical for proper function of platelets and white blood cells. To form a clot, platelets must bind to the vessel surface via specific receptor-ligand interactions. Similarly, extravasation of white blood cells at the site of infection involves binding to adhesion ligands at the vessel surface. The margination process facilitates adsorption at the vessel wall.

To the inventors' knowledge, the present disclosure is the first to take advantage of RBC migration to enhance adsorption at the wall of a microfluidic device, such as for removal of blood constituents and/or toxins. RBC migration is leveraged to enhance mass transfer and create an efficient adsorption device that is safe to use with whole blood. In the absence of RBC migration and the associated margination effects, mass transfer to the surface of a sorbent is governed by diffusion. Consequently short diffusive lengths are necessary and the gaps between beads in a typical packed column are small. This leads to relatively high shear stresses and the potential for damage to blood cells. By taking advantage of margination, similar mass transfer kinetics can be achieved using a microfluidic device with relatively large gaps between the solid surfaces. The resulting shear stresses are expected to be relatively small and hence less damaging to blood cells.

In some embodiments, the strategy involves periodic disruption of the steady-state RBC distribution using an offset array of microchannels 110 defined by microchannel walls 120, as illustrated in FIG. 3. Within each segment of the array, RBCs migrate to the center of the channel, causing enrichment of other blood constituents including pathogens near the channel wall, as indicated by the arrows. Mass transfer is further enhanced by disrupting the RBC distribution at each lateral offset in the microchannel array. To enable clinically relevant flow rates and selective adsorption at the device surface, multiple parallel microchannel arrays can be used (see FIG. 1).

To facilitate identification of promising device designs, a computational model was developed capable of capturing, through a statistical approach, the net effects of RBC migration and the formation of the cell-free plasma layer within microfluidic device architectures. The modeling approach was first devised by Gidaspow (Gidaspow, D. and J. Huang, Kinetic theory based model for blood flow and its viscosity Ann Biomed Eng, 2009. 37(8): p. 1534-45, incorporated herein by reference) and used to study blood flow in cylindrical tubes to predict potential sites of atherosclerotic lesions in vivo. This model applies granular flow theory and uses statistical methods analogous to kinetic theory of dense gases in order to represent RBCs as a continuous phase. Gidaspow demonstrated that the model can accurately predict some commonly-observed trends, including the dependence of apparent blood viscosity on hematocrit and tube diameter. This model was adapted to provide the increased resolution necessary for capturing the cell free layer and for use in rectangular microchannels, the most common geometry in microfluidic devices.

FIG. 4 provides preliminary computational results for a 50 μιη x 50 μιη square microchannel, with the flow velocity set at 2 mm/second and hematocrit at 40%, illustrating the local RBC concentrations in a cross sectional area of the channel, as well the development of the cell-free layer downstream of a uniform inlet. Preliminary simulations of blood flow through a representative region within an array of offset microchannels with a cross section of 100 μιη x 100 μιη, a flow velocity of 2 mm/second and hematocrit at 40%, have also been performed. The results, shown in FIG. 5, illustrate that the RBC distribution was disrupted at the bifurcation, leading to RBC migration and establishment of a new steady- state downstream.

Based on the results from the computational models, laminae were designed comprising an array of offset microchannels. FIG. 6 provides a close-up view of an exemplary lamina surface 100, illustrating the array of offset microchannels 110. The microchannels 110 are defined by microchannel walls 120 that are substantially rectangular in shape. In the illustrated embodiment, the microchannel walls 120 have an end geometry 130 that is triangular, but a person of ordinary skill in the art will appreciate that other geometries are possible. For example, FIG. 7 illustrates exemplary end geometries, including a square end (130a), a rounded end (130c) and a triangular end (130b) like that shown in FIG. 6. In some embodiments the two ends of the microchannel wall have the same geometry, as illustrated in FIG. 6. However, in alternative embodiments, the upstream wall end has a different geometry from the downstream wall end. In addition, adjacent end geometries 130 can have at least partially overlapping end geometries 130 such as that shown in the triangular ends (130b) of FIG. 6 or adjacent offset end geometries can be substantially even with each other or otherwise spaced apart as shown in the square end (130a) embodiment of FIG. 5. As illustrated in FIG. 8, the width in each branch of the bifurcation need not be identical. Referring to FIG. 8, each width A and B independently can be the same (as in FIG. 6) or different (as in FIG. 8). If the widths are different, overall symmetry can still be maintained by periodically alternating which branch of the bifurcation (top or bottom) has the larger width, as shown in the exemplary embodiment in FIG. 8.

Alternatively, the width of each branch of the bifurcation may vary independently from one another, and/or may vary in an unsymmetrical, and/or random pattern.

Referring again to FIG. 6, the microchannels have a width 140. Width 140 can be any width suitable to allow fluid flow. In embodiments, the microchannels have a width of from greater than zero to about 1 mm, such as from about 10 μιη to about 750 μιη, from about 25 μιη to about 500 μιη, or from about 50 μιη to about 200 μιη. In particular embodiments, the microchannel width 140 is about 50 μιη and in other embodiments it is about 200 μιη. The microchannels also have a depth, not shown in FIG. 6. The microchannel depth can be any depth suitable to allow fluid flow. In some embodiments, the microchannels have a depth of from greater than zero to about 1.5 mm, such as from about 25 μιη to about 1 mm or from about 100 μιη to about 500 μιη. In some embodiments the microchannel width is substantially the same as the microchannel depth, and in other embodiments the microchannel width is greater than or less than the depth.

The microchannel bifurcation length 150 can be any length suitable to achieve suitable mass transfer kinetics while minimizing shear stresses and potential damage to blood cells. In some embodiments, the bifurcation length 150 is from about 25 μιη to about 1.5 cm, such as from about 50 μιη to about 1 cm or from about 250 μιη to about 5 mm. In particular

embodiments, the bifurcation length 150 is from about 500 μιη to about 2 mm.

Microchannel arrays can be created with the offset patterning discussed above by machining (e.g., laser machining), forming (e.g., stamping), micromolding, and/or any other suitable manufacturing technique. Lamina can be bonded together to form the array by diffusion bonding, solder past, laser welding, and/or other suitable coupling methods. In some embodiments, the device, when fluidly coupled to a blood stream, can operate with a sufficient rate of blood flow through the device, such that a body volume of blood can pass through the device in a clinically relevant length of time, such as in two hours or less, one hour or less, or 30 minutes or less. In other embodiments, the blood flows at a rate of from about 25

milliliters/minute to about 250 milliliters/minute, such as from about 75 milliliters/minute to about 150 milliliter s/minute. In particular embodiments, the blood flows at a rate from greater than about 100 milliliters/minute to about 250 milliliter s/minute.

Offset microchannels (as defined by offset microchannel walls) can increase the efficiency of operation of the device in a manner consistent with the enhancement of adsorption due to RBC migration. In one example, removal of biotin-coated microspheres was measured after flowing whole blood or blood plasma through an exemplary device coated with

streptavidin. These experiments revealed two clear trends that strongly supported the concept of margination-enhanced adsorption: (1) the presence of bifurcations significantly increased microsphere removal; and (2) the presence of RBCs significantly increased microsphere removal. In experiments with whole blood, 24% of the microspheres were removed in an exemplary straight microchannel array, while 41% of the microspheres were removed in an exemplary offset microchannel array. This corresponded to more than a 70% improvement in adsorption efficiency for the offset microchannel array. When plasma was used, the extent of microsphere removal significantly decreased; in the device with straight microchannels, only 12% of the microspheres were removed, whereas 18% were removed in the bifurcated microchannel array. These results demonstrated that the presence of RBCs significantly enhanced adsorption at the device surface. Together, the results also indicated a synergistic effect between bifurcations and the presence of RBCs, strongly supporting the idea that bifurcations can be used to enhance adsorption by disrupting and then re-establishing the RBC distribution.

III. Surface Coating

A. Molecules for functional surface coatings

A wide range of biological functions can be imparted to the coated surface by inclusion of one or more molecules, such as biological molecules, some of which are bioactive, of similar or dissimilar function, into or onto the coating. The biological molecule can be isolated or synthesized. Such molecules include, but are not limited to, intact or fragmentary portions of peptides, proteins, anticoagulants, anticlotting agents, enzymes, antibodies or antibody fragments, aptamers, single or double- stranded DNA or RNA, polysaccharides or

glycosaminoglycans, lipids or phospholipids, chelators, antibiotics, quaternary amines, or synthetic polymers, as well as specific ligands recognized by antibodies, cells, or other blood components, or combinations thereof. In some disclosed embodiments, a surface coating based on an immobilized molecule is used to remove constituents from the fluid flowing through the microfluidic device. In some embodiments, the microfluidic device enhances the kinetics of enzyme catalyzed reactions by bringing reactant molecules to the device surface. In some embodiments, one type of bioactive molecule is attached to a surface, but in other embodiments, there are multiple types of bioactive molecules on a single surface, such as two, three, four or more types of bioactive molecule. For example, an individual surface may have anticoagulant and antibiotic bioactive molecules attached.

In some embodiments, the bioactive molecule is a peptide or a useful portion thereof. Suitable peptides are any peptide that can bind to a target molecule or cell and prevent that target from flowing with the fluid stream. Suitable peptides include, but are not limited to, cationic amphiphilic peptides, lantibiotics and other peptide antibiotics. Exemplary peptides include, but are not limited to, polymyxin B (PmB), nisin, WLBU2, WR12, and mutants or structural variants or derivatives thereof. In addition, a short hydrophobic "leader sequence" may be added to any peptide to improve selectivity for bacterial membranes versus blood/host cells. In some embodiments, the short hydrophobic sequence is at the 'non-attached' or free end of the peptide - the end that is not attached to the surface or the polymer. In some embodiments, the short hydrophobic sequence has up to at least 100 amino acids, such as from 2 to 100 amino acids long, or from 2 to 50 amino acids long, or from 2 to 25 amino acids long.

Cationic amphiphilic peptides (CAPs) constitute a major class of antimicrobials that allow neutrophils and epithelial surfaces to rapidly inactivate invading pathogens. A number of CAPs have been shown to bind LPS with affinities comparable to PmB. For example, the CAP human cathelicidin peptide LL-37 has been shown to neutralize the biological activity of LPS and to protect rats from lethal endotoxin shock, revealing no statistically significant differences in antimicrobial or anti-endotoxin activities between LL-37 and PmB. Despite the broad activity of LL-37 and other natural CAPs, their potency is inhibited in the presence of physiological concentrations of sodium chloride (NaCl) and divalent cations. However the 24-residue, de novo engineered peptide WLBU2, a synthetic analogue of LL-37, shows highly selective, potent activity against a broad spectrum of Gram-positive and Gram-negative bacteria at physiologic NaCl and serum concentrations of Mg²⁺ and Ca²⁺. Moreover WLBU2 shows greater

antimicrobial activity than either LL-37 or PmB, and is active against a much broader spectrum of bacteria.

A major distinguishing feature of CAPs is their capacity to adopt an amphiphilic secondary structure in bacterial membranes, typically involving segregation of their positively- charged and hydrophobic groups onto opposing faces of an a-helix. The propensity for a-helix formation in cell membranes correlates positively with CAP activity and selectivity of bacterial over human cells, and WLBU2 has been optimized specifically for formation of an amphipathic α-helix conformation in cell membranes. Finally, in addition to its broad-spectrum antimicrobial activity in blood, WLBU2 retains potency while bound to solid surfaces and importantly, shows high affinity for adhesion of susceptible bacteria. The related peptide WR12 also exhibits substantial antimicrobial and binding activity, while being smaller (12 amino acids) and chemically simpler than WLBU2, and may also be used in some embodiments.

The bioactive molecule may be an antibody or aptamer. In some embodiments, the antibody or aptamer specifically binds to pro-inflammatory cytokines. Without being bound to a particular theory, these cytokines are implicated in the initial dysregulation and inflammatory response of the immune system, and their removal from the bloodstream during treatment of early sepsis may improve patient outcomes by helping to quell the inflammatory response. In some embodiments, antibodies specific for the pro-inflammatory cytokines TNF- and/or IL-1 can be incorporated to remove these molecules from blood. A person of ordinary skill in the art will understand that these and other cytokines may also be implicated in cancer, heart disease, Alzheimer's disease, autoimmune disorders, and other conditions. Additional information can be found in Schulte, W., J. Bernhagen and R. Bucala, Cytokines in sepsis: Potent

immunoregulators and potential therapeutic targets - an updated view. Mediat Inflamm, 2013, article ID 165974, incorporated herein by reference. Selective and efficient removal of these molecules from blood in a microfluidic device may thus provide an important additional therapeutic treatment method for many common diseases.

Other therapeutically important bioactive molecules include anticoagulants, which either prevent the formation of fibrin clots, which may occur when blood is in contact with a synthetic material surface, or induce the breakdown and elimination of small fibrin mats before they become clinically important. In some embodiments, the bioactive molecules may thus include heparin and/or its derivatives including fondaparninux and idraparinux, heparin sulfate,

Coumadin and its derivatives, rivaroxaban, apixaban, edoxaban, hirudin, lepirudin, bivalirudin, or other anticoagulants or combinations thereof. In addition, enzymes such as thrombomodulin may be incorporated into a device, to catalyze the destruction of small circulating fibrin clots.

Other exemplary enzymes that may be used in the device include, but are not limited to, heparinase, bilirubin oxidase, and phospholipase A2. Heparinase may be useful for enabling local anticoagulation within an extracorporeal circuit, where heparin is broken down prior to returning the blood to the body. Bilirubin oxidase may be useful for converting bilirubin to nontoxic products for treatment of jaundice. Phospholipase A2 may be useful for converting low density lipoprotein to more rapidly cleared products. B. Linkers and activated end-groups

In some embodiments, the sorbent molecule is associated with and/or attached to the surface of the microchannel with an intervening linker. In some embodiments, the linker is a polymer. Suitable polymers include any polymer that can be associated with the surface of the microchannel, and also associated with the peptide. The polymer may be associated with the surface of the microchannel through any suitable technique, such as, but not limited to, grafting to or from the surface, covalent bonding, ionic bonding, electrostatic attraction, adsorption or combinations thereof. In particular embodiments, the polymer is adsorbed onto the surface of the microchannel, and in certain examples, the adsorbed polymer is then exposed to radiation, such as gamma radiation, to covalently attach the polymer to the surface.

In some embodiments, the polymer is an amphiphilic block copolymer, comprising hydrophobic and hydrophilic polymers, and in certain embodiments the polymer is a triblock copolymer. Amphiphilic block copolymers will self-assemble from aqueous or organic solutions to form a pendant 'brush layer' attached to the surface through the middle block by one or more of the methods described above (FIG. 9). In some embodiments, the block copolymer comprises polyethylene oxide (PEO), polypropylene oxide (PPO),

polytetrahydrofuran (PTHF), polybutadiene (PB), polystyrene, polypentadiene, polyhexadiene, polyacrylonitrile, polyhydroxyethylmethacrylate (PHEMA), polyurethane (PU), polyacrylamide (PA), or combinations thereof. In particular embodiments, the polymer is PEO-PPO-PEO, PEO- PB-PEO, PTHF-PB-PTHF, PHEMA-PB -PHEMA, or PA-PB-PA.

The polymer may be a straight chain polymer, or it may be a branched chain polymer, with one or more branches. In some embodiments, the branched polymer has 1, 2, 3, 4, 5 or six branches. Formula I illustrates one example of a branched polymer:

I

where each n independently is from 0 to 250 or more, such as from 10 to 250 or 25 to 200, or about 25, 50, 75, 100, 150 or 200. In certain embodiments, all the n values are substantially the same. Additionally, or alternatively, the branched polymer may have an average molecular weight of from 1 kDa to 25 kDa or more, such as from 5 kDa to 20 kDa, or from 7 kDa to 15 kDa. In certain embodiments, the branched polymer has an average molecular weight of about 5 kDa, about 10 kDa about 15 kDa or about 20 kDa. As used herein the average molecular weight is a number averaged molecular weight, as determined by any suitable technique known to persons of ordinary skill in the art, such as NMR, size exclusion chromatography and/or mass spectrometry.

In some embodiments, the branched polymer is a 'star' polymer, such as the example shown in Formula I. A 'star' polymer is a branched polymer where the branching chains are attached at the same carbon, and the lengths of the branches are such that the arms of the star are of equal length. In other embodiments, at least one, or more than one or all of the branching chains are of different lengths to each other, and/or the attachment positions of each branching chain are different.

In still other embodiments, the branched polymer may be a dendrimer (e.g. PAMAM), or other hyperbranched polymer. The dendrimer may be a 1^st, 2^nd, 3^rd, 4^th, 5^th, 6^th or more generation dendrimer. In some embodiments, an η^ώ generation dendrimer has 2ⁿ branches. For example, a 5^th generation dendrimer may have 2⁵ or 32 branches. In other embodiments, an η^ώ generation dendrimer has 2ⁿ⁺² branches, such that a 5^th generation dendrimer may have 2⁷ or 128 branches.

The polymer may further comprise at least one functional group that can attach to the bioactive molecule. The functional group may be any functional group suitable for reaction with the bioactive molecule. Polymers comprising a functional group may be referred to as end group activated polymers (EGAPs). Exemplary functional groups include, but are not limited to p-nitrophenol, [sulfo-]N-hydroysuccinimide, imidoester, hydrazine, maleimide, thiopyridyl, iodoacetyl, tyrosyl residue, vinylsulfone, iodoacetimide, epoxide, thiol, amine, iso[thio]cyanate, aldehyde, azide, alkene, alkyne, phosphine, or disulfide. In certain embodiments, the reactive group is selected from functional groups known to be stable in an aqueous environment, including, but not limited to, hydrazine, imidoester, maleimide, thiopyridyl, iodoacetyl, tyrosyl residue, vinylsulfone, iodoacetimide, aldehyde, azide, alkene, alkyne, or disulfide.

In a preferred embodiment, the functional group comprises an R'— S— S group where R' is to be displaced to facilitate the immobilization of an antimicrobial peptide. In some embodiments, the substituent R' is selected from 2-benzothiazolyl, 5-nitro-2-pyridyl, 2-pyridyl, 4-pyridyl, 5-carboxy-2-pyridyl, or N-oxides of any the pyridyl-containing groups. A preferred end group includes 2-pyridyl disulfide or 4-pyridyl disulfide (PDS). The reactivity of these groups with proteins and polypeptides is discussed in U.S. Patent No. 4,149,003 to Carlsson et al. and U.S. Patent No. 4,711,951 to Axen et al, all of which are hereby incorporated by reference.

C. Coupling the sorbent molecule to the polymer

The coupling of a sorbent molecule to the end of the polymer chain[s] may be achieved using a variety of methods known to a person of ordinary skill in the art. Additional information on coupling sorbent molecules to polymers is provided by Greg T. Hermanson, Bioconjugate Techniques; Academic Press, 1996, which is incorporated herein by reference. In particular embodiments, a sorbent molecule, such as an antimicrobial peptide or any of the other molecules described above, is attached to the end group activated polymers through thioether or disulfide linkages. Accordingly, the sorbent molecule may be modified to include a thiol group, so that this thioether or disulfide linkage may be made. In some embodiments, this modification is made to the C- or N-terminal of a peptide. A person of ordinary skill in the art will be able to recognize where on the sorbent molecule, such as which peptide terminal, is suitable for the introduction of such a thiol group. Another method involves using recombinant or synthetic protein engineering techniques to introduce a moiety that can bond an antimicrobial peptide to EGAP. For example, Cys-WLBU2 or WLBU2-Cys may be produced that contains an N- terminal or C-terminal cysteine residue respectively. In some cases, the cysteine residue may be separated from the native sequence of the peptide by one or more amino acid (e.g. glycine) spacers or a polyethylene oxide (PEO) spacer. Accordingly, the thiol-terminated or cysteine- terminated peptide may then be linked to a pyridyl disulfide activated polymer through a disulfide exchange, or to a polymer activated with maleimide or iodoacetamide to form a thioether bond. Another method involves selecting a form of the end group activated polymer that is capable of reacting directly with the native form of the peptide. This method can be achieved by reacting a p-nitrophenol activated polymer or an [sulfo-]N-hydroxysuccinimide activated polymer with the N-terminal or side-chain amines of the peptide. Yet another method involves producing a peptide that is recombinantly or synthetically engineered to produce a peptide having a terminal polyhistidine tag. A polyhistidine-tagged antimicrobial peptide is then bound through very strong ionic interactions, in the presence of divalent metal ions, to a polymer that has an activated end group in which one or more of the hydroxyl end groups of the PEO groups have been replaced with a nitrilotriacetic acid group.

Other ways and/or methods for introducing the thiol group or other reactive groups may be used and will depend (in part) on the particular sorbent molecule is being used. Further, with respect to other types of activated end groups that may be introduced to polymers, a variety of other modifications/chemical groups may be added to the sorbent molecule in order to ensure coupling between the polymer and the sorbent molecule, as will be recognized by those of ordinary skill in the art.

A person of ordinary skill in the art will appreciate that the order of attachment of the polymer and sorbent molecule may also vary according to the properties of different molecules. In some embodiments, the polymer is first associated with the surface of the microchannel, then activated if necessary, and finally exposed to the sorbent molecule, allowing it to attach to the polymer. In other embodiments the polymer is first activated and associated with the surface, and then exposed to the sorbent molecule (FIG. 10). The microchannel is then rinsed or flushed with water, buffer or solvent to remove excess or unreacted molecules.

Alternatively, the polymer may be conjugated with the sorbent molecule prior to association with the surface. The resulting polymer- sorbent molecule construct is purified by any suitable method, such as dialysis, chromatography, or precipitation. The microchannel surface is then exposed to the purified construct. A person of ordinary skill in the art will be able to select the order of attachment based on the particular sorbent molecule. For example, a sorbent molecule that is not resistant to gamma or ultraviolet radiation will be attached to the polymer after the polymer is exposed to the radiation to immobilize it on the surface.

IV. Applications

There are a wide variety of medical conditions that can be treated by passing blood or plasma through an extracorporeal device containing a sorbent to remove specific blood constituents. For instance, many autoimmune diseases are mediated by circulating antibodies directed against self-antigens. Removal of these autoantibodies has proved beneficial for treatment of several diseases, including rheumatoid arthritis, dilated cardiomyopathy, myasthenia gravis, Guillain-Barre syndrome and systemic lupus erythematosus. Acute rejection following organ transplant is also mediated by circulating antibodies and consequently removal of antibodies is being investigated as a strategy for improving the outcome of ABO incompatible organ transplant procedures. Sorbent-based approaches are also useful in the treatment of high cholesterol, poisoning, kidney failure and sepsis.

Severe sepsis is a blood infection that in the US alone affects about 750,000 people each year, killing 28-50% of them, and accounting for nearly $17 billion in treatment costs. The number of sepsis-related deaths continues to increase, and is already greater than the annual number of deaths in the US from prostate cancer, breast cancer and AIDS combined. During bacterial growth, or as a result of the action of antibacterial host factors, lipopolysaccharide (LPS, endotoxin) is released from the cell wall of Gram- negative bacteria. The high immuno stimulatory potency of endotoxin causes dysregulation of the inflammatory response with elevated production and release of proinflammatory cytokines, leading to blood vessel damage and organ failure. For sepsis, it is the dysregulated inflammatory response caused by the pathogen and not the pathogen itself that leads to the complications associated with sepsis. The current standard of practice is systemic administration of antibiotics, which helps to reduce bacterial load but may release lipopolysaccharide-rich bacterial fragments (endotoxin) into the blood. These bacterial fragments are potent triggers of the systemic inflammatory response. Embodiments of the disclosed device treat sepsis by targeting removal of endotoxin from the blood.

A. Interaction between LPS and surface bound peptide.

For the microchannel device to operate once the peptide is attached, the peptide must be tethered to the microchannel surface in such a way as to preserve its solvent accessibility and mobility. To assess this for WLBU2, the interactions between the peptide and LPS were investigated for surface immobilized, PEO-chain-entrapped and PEO-chain-tethered WLBU2. FIG. 11 provides an illustration comparing entrapped and tethered peptides.

The interactions between LPS (0.1 mg/mL) and surface bound WLBU2, and WLBU2 variants recorded by QCM-D are presented in FIG. 12 and FIG. 13. Adsorbed mass of WLBU2, CysWLBU2, and WLBU2Cys prior to LPS adsorption was about 300 ng/cm², about 425 ng/cm², and about 400 ng/cm², respectively, calculated by the Sauerbrey equation. FIG. 12 indicates that adsorption to a peptide coated surface was similar for all WLBU2 variants and was greater than adsorption of LPS to a bare gold surface. Without being bound to a particular theory, the enhanced adsorption of LPS vesicles at the peptide-coated surface was likely caused primarily by electrostatic interaction between the negatively charge LPS vesicle with WLBU2, which carried an out-of-balance charge of +13 at physiologic pH. WLBU2 with a cysteine added to either the amine- (CysWLBU2) or carboxy-terminated (WLBU2Cys) end was expected to adsorb "end-on" to the gold surface mediated by the high-avidity, gold-thiol association.

Chemical bonding energies can vary greatly, with hydrophobic association at about 0.8 kcal/mol, common hydrogen bonds ranging from 2 to 7 kcal/mol, C-C bonds at 83 kcal/mol, and gold-thiol bonds at 45 kcal/mol. As thiol-gold interactions approximate covalent attachment (45 v 83 kcal/mol, compared to 7 for H-bonding), it was expected that WLBU2 associated in this manner would not be replaced by LPS. On the other hand, WLBU2 randomly associated to a surface may have been removable. As seen in the QCM-D data (FIG. 12), adsorption of LPS to each of the three WLBU2-variant coated surfaces was similar. LPS association to end-on oriented WLBU2 molecules, presenting either the amine or carboxy end, behaved substantially the same as to randomly adsorbed WLBU2, suggesting that LPS association to each of these layers was likely electrostatic, and not indicative of higher order interaction. Since the orientation of WLBU2 did not seem to change the adsorption of LPS, and modifying peptides at the N-terminus is more straightforward than the C-terminus, only the CysWLBU2 variant was used for further experimentation.

FIG. 13 shows dissipation versus frequency curves for each experiment. Data in this format allows visual inspection of the quality of the adsorbed layer or adlayer. In particular, if the data shows hysteresis effects, adsorption is likely changing the structure of the adsorbed layer. More generally, data displayed in this manner describes the comparable rigidity of a layer changing with adsorption and elution; as the slope decreases, the rigidity increases. The data shown in FIG. 13 nearly overlap for each of the LPS on peptide experiments, with hysteresis ranging from 0.4 - 1.9%. This strongly suggested the structure of LPS did not change upon adsorption to a WLBU2 coated surface, whether adsorbed randomly or end on. For LPS adsorbed to Au, the slope of the dissipation versus frequency was greater than any of the LPS on peptide experiments, suggesting a much more rigid layer, and this curve showed hysteresis upon elution, with a change of nearly 18%. Taken together, this strongly suggested that LPS vesicles unfolded at a gold surface, but remained largely intact and electrostatically adsorbed to a peptide coated surface.

FIG. 14 provides CD spectra for WLBU2 in suspension with hydrophobic nanoparticles with and without LPS. The a-helicity changed from 10% on the bare particle without LPS to 23% with LPS included. This data supported the hypothesis drawn from the QCM-D data that surface bound WLBU2 did not substantially interact with LPS in a meaningful way.

B. Interaction between LPS and PEO layers

PEO layers are commonly considered to be nonfouling. As such, no irreversible location of LPS at pendent PEO layers was expected. FIG. 15 shows that this was indeed the case as changes in both the frequency and dissipation returned to the original baseline upon elution.

C. Interaction between LPS and peptide at PEO layers

I. LPS at peptide entrapped layers

It has been demonstrated that WLBU2, like some other peptides, is able to penetrate

PEO brush layers. This tendency toward small peptide entrapment requires that, for the purposes of peptide tethering, EGAP-WLBU2 constructs must be prepared in advance of adsorption to a surface. If attachment were to be conducted in situ, location of WLBU2, either entrapped or tethered would not be discernable. FIG. 16 shows QCM-D data for LPS association at peptide entrapped PEO layers. For the concentration used, a maximum loading of WLBU2 was expected to be around 0.2 molecules/nm² (about 120 ng/cm²) or less for entrapped peptide in membrane mimetic solvents, which would encourage a-helicity. In fact, when calculating the concentration of WLBU2 in solution by UV/Vis spectroscopy for entrapped WLBU2 on nanoparticles for CD analysis, the apparent concentration of WLBU2 was below the detectable limit. FIG. 17 shows that for entrapped peptide, there was very little initial peptide present, but upon introduction of LPS, the a-helicity increased from 3% to 8%.

//. LPS at peptide tethered PEO layers

In the context of LPS capture in a hemoperfusion device, it is preferable to retain the mobility and solvent accessibility of the active capture agent, in this case WLBU2. Further, management of peptide density and distance from the primary interface typically requires more control than peptide entrapment allows. Thus, preferentially the peptide should be tethered to the surface, and to avoid convolution with potential entrapment of the peptide, it is important to build the tethered- WLBU2 construct prior to surface immobilization. Typically, γ-irradiation is used to covalently attach triblocks to surfaces. It was therefore important to first investigate the effect of both a tether and γ-irradiation on the structure and function of WLBU2.

Covalent attachment of WLBU2 with EGAP-PDS occured spontaneously at room temperature according to the reaction scheme presented in FIG. 18. The release of pyridine-2- thione (P2T) allowed for the direct calculation of the total amount of construct produced.

Constructs were prepared by mixing equimolar quantities of EGAP-PDS and CysWLBU2, and average total conversion was greater than 50% as evaluated by the evolution of P2T absorption at 343 nm (ε = 8080 M^"1 cm^"1). Unmodified WLBU2 showed a substantial increase in a-helicity upon exposure to LPS, increasing in helicity from about 0 to 78%, illustrated in FIG. 19. This is owing to the WLBU2 infiltrating the lipid A region of the LPS vesicle, a prerequisite for vesicle capture. EGAP-WLBU2 was evaluated in a similar manner and the results are shown in FIG. 20. The data showed an increase in helicity from 2% in HPLC H₂0 to 16% upon addition of

LPS to 0.1 mg/mL. FIG. 21 provides CD spectra of the interaction of LPS with a tethered peptide associated with an interface. The data showed convincingly that WLBU2, when tethered to a surface, retained its ability to adopt an a-helix upon introduction of LPS, changing from 2% helix without LPS to 17% including LPS. These data clearly indicated that the inclusion of a covalent tether, on the order of 12.5 kDa, did not prevent WLBU2 from interacting with LPS in a manner keeping with that of unmodified WLBU2, and further suggested that WLBU2 covalently tethered to a surface would retain the ability to interact with LPS, and therefore capture vesicles from solution.

While the data shown in FIGS. 20 and 21 provided compelling evidence and support for the potential of tethered WLBU2 to capture LPS from solution, these systems had not been γ- irradiated. FIGS. 22-24 show the effects of comparable doses of γ-irradiation as used for covalent attachment of triblocks as evaluated by NMR (FIG. 22, about 1000 μΜ WLBU2), and UV/Vis and CD (FIGS. 23 and 24 respectively, about 35 μΜ WLBU2).

The NMR spectra shown in FIG. 22 show that the structure of WLBU2 remained intact upon irradiation, showing no significant difference in structure. Spectra recorded using UV/Vis spectroscopy (FIG. 23) showed more substantial change in the characteristic curve, as the curve broadens and its peak at 280 nm was reduced by 13%. Finally, in a helix inducing solvent (perchloric acid), WLBU2 was shown to decrease in a-helicity from 65% to 36% upon irradiation as indicated by the CD spectra shown in FIG. 24.

With the effects of tethering and γ-irradiation understood, the capture of LPS by tethered WLBU2 as witnessed in QCM-D and shown in FIG. 25 was appropriately evaluated. The QCM-D data presented showed clear evidence of LPS capture by tethered WLBU2 indicating that this system worked.

When comparing the capture of LPS between entrapped peptide PEO layers (FIG. 16) and tethered peptide PEO layers (FIG. 25), it seemed that the entrapped peptide captured a greater amount of LPS than the tethered peptide did, as AFentrapped/AFtethered = 2.3 at the end of the elution step, despite the lower surface concentration in the entrapped case. This result, however, was consistent with the removal of WLBU2 from an entrapped motif and the creation of peptide bridges and LPS aggregation, as has been shown in previous work. Without being bound to a particular theory, the resultant LPS-WLBU2 association may not indicate capture, but merely that the aggregates resist the flow parameters and do not leave the interface, illustrated in FIG. 26. This is further consistent with the very slow elution kinetics seen in FIG. 16 and the low elutability of only 24%. Because WLBU2 cannot participate in LPS bridging in the tethered motif, what remained at the surface upon elution (76% elutability) is likely due only to capture, and not convoluting complexes. D. Effect of fibrinogen on LPS capture

A clinically relevant device must be able to capture LPS from whole blood in a hemoperfusive device. To that end, it must be demonstrated that LPS capture is possible from a complex milieu containing blood proteins. FIG. 27 shows QCM-D evidence that fibrinogen does not substantially adsorb or remain on a surface containing only F127 polymer. Thus, any interaction described upon the inclusion of LPS and/or WLBU2 would suggest that location of fibrinogen was modulated by those excipients, and not by the PEO brush layer itself. FIGS. 28 and 29 show the adsorption and elution profiles, by f md AD, of a mixture of fibrinogen and LPS on a surface containing entrapped WLBU2 (FIG. 28) and one with covalently attached EGAP-WLBU2 (FIG. 29). The concentration of fibrinogen was physiologically relevant, at 2 mg/mL, and LPS was at the same concentration for all other experiments, 0.1 mg/mL.

It was clear from the shape of these curves that the adsorption and elution of fibrinogen/LPS mixtures was complex. For both sets of data, the total adsorbed amount was higher (-Af <x Am) at the end of both the adsorption and elution steps in the experiment than for LPS adsorption on respective surfaces alone. In the case of entrapped peptide contacted by the mixture (FIG. 28), 82% of the adsorbed mass was removed upon elution. For the EGAP- WLBU2 construct challenged by the mixture (FIG. 29), 51% of the mass was removed upon elution. Because the total adsorbed mass of both fibrinogen/LPS mixtures was greater than that for LPS alone, it was clear that fibrinogen itself interacted with WLBU2, interacted with LPS in a manner further encouraging location or capture at the interface, or some combination of the two.

FIG. 30 shows the comparison of Afvs time for fibrinogen on an entrapped or tethered surface motif, and FIG. 31 provides the Adissipation vs Afior these same surfaces. It was clear from the data shown in these figures that fibrinogen did in fact interact with surfaces that contained WLBU2. The nature of this interaction appeared not to depend on whether the surface contained entrapped or tethered WLBU2, as shown in FIG. 31, despite overall mass loading being dissimilar (FIG. 30). Further evidence of this was that the percent mass eluted was 14% and 15% for fibrinogen on the entrapped or tethered peptide, respectively. This interaction was directly related to the inclusion of WLBU2, as fibrinogen did not substantially adsorb or remain on a surface containing only F127 polymer, shown in FIG. 27.

Without being bound to a particular theory, the interaction between fibrinogen and

WLBU2 associated surfaces was likely not due to a higher order interaction between the two proteins, but likely only suggestive of an electrostatic interaction between the two. As stated previously, WLBU2 contains an out-of-balance net positive charge of 13, while the outer regions of fibrinogen carry a net negative charge. The adsorption and elution profile of fibrinogen as compared to fibrinogen/LPS mixtures adsorbed and retained more mass for both entrapped and tethered peptide, suggesting that fibrinogen/LPS mixtures existed as more than a binary mixture of discrete molecules, but rather as a fibrinogen-LPS complex. Further, although the experiments included a physiologically relevant concentration of fibrinogen, the

concentration of LPS was well beyond what may be expected in a clinical setting; the 0.1 mg/mL used in this work corresponded to about 500 grams of LPS circulating in the human body. For this experimentation, using smaller concentrations of LPS was not efficacious because the capture seen already by this non-optimized device analogue was rather low (FIG. 25). More direct investigation of the interactions of fibrinogen and LPS may be required to elucidate nuances shown in the data presented.

Despite evidence that fibrinogen interacted with surfaces containing WLBU2, and that LPS and fibrinogen may have created a complex structure, FIGS. 32 and 33 suggest that WLBU2, whether entrapped or tethered, preferentially captured LPS over fibrinogen. The data shown is of the ratio of / to ΔΌ vs time. Data shown in this manner allowed for more direct comparison of surface characteristics with respect to adsorption and elution of various species, and time. Further, viewing the data in this manner revealed intricacies not captured by other graphical methods. For instance, as the ratio -Af/ΔΌ increased, the adsorbed mass was changing more rapidly than is the dissipation, suggestive of increasing rigidity. Conversely, as this ratio decreased, the dissipation was increasing more rapidly than the frequency (or was decreasing less rapidly), indicating that the overlayer became less rigid. In both FIGS. 32 and 33, the overall shape of the curves was similar for each surface. For the surfaces challenged with LPS, the ratio, - f/ D, did not appear to change much upon elution. This suggested that the overall structure of the LPS did not change upon elution, or more specifically, that there was unlikely to be an under layer of spread LPS. The slight decrease, in fact, indicated that the layer became less rigid, as mass was seen to decrease (frequency increases) upon elution in all figures showing AF vs time. For fibrinogen, the situation was quite the opposite, because the ratio of -AFrequency/ADissipation increased upon elution, and the mass decreased, the layer must (i) become rigid upon elution, or (ii) loosely bound fibrinogen was removed, revealing a rigid underlayer of associated protein. Finally, for the fibrinogen/LPS mixture on the entrapped

WLBU2 surface, ratio of frequency to dissipation was nearly identical to that for fibrinogen at the end of the adsorption cycle, but upon elution came almost to the same point as that for LPS.

This pattern was consistent with a system wherein a smaller molecule with slight affinity for the presented surface approached that surface more quickly than its larger counterpart that has a higher affinity. Over time the larger, higher affinity molecule would replace the smaller one, resulting in a surface that initially behaved like one containing only the smaller molecule, but ended similarly to a system containing only the larger. As this pattern was seen in FIG. 32 and the pattern was similar, albeit not to the same extent in FIG. 33, LPS capture may be possible with both entrapped and tethered WLBU2 even in a complex milieu containing physiological quantities of fibrinogen.

Analysis of the interaction between surface-immobilized, PEG chain entrapped, and PEG chain tethered WLBU2 with LPS using QCM-D, and CD, as well as the effects of γ-irradiation on PEGylated WLBU2 using UV/Vis spectroscopy and NMR, all revealed that WLBU2 could interact with LPS in a manner keeping with its purpose whether irradiated, PEGylated, or tethered. In this way, WLBU2 is suitable for use in a hemoperfusive device for the capture of sepsis causing LPS. QCM-D data suggested that LPS capture by tethered WLBU2 likely retains its overall structure. This indicates that the LPS membranes did not rupture, which would result in LPS fragments passing back into the body after treatment, potentially causing complications.

V. Examples

Example I

Comparison of Straight Microchannels versus Offset Microchannels In order to improve the removal of endotoxin through adsorption onto the WLBU2 peptide, the microchannel geometry was designed to make use of the collisional and volume displacement effects of red blood cell migration to transport endotoxin aggregates and bacterial cells towards the channel walls, where contact with the modified surface permited selective adsorption. This required a sufficient channel length to permit a steady-state flow profile to develop, for margination to occur, and for adsorption of the target species from the cell-free peripheral layer. Adsorption at the device walls depleted this peripheral layer of endotoxin at which point the channels will bifurcate, disturbing the flow profile and creating a homogenous mixture. This then allowed the margination process to re-occur and the peripheral layer to be replenished with endotoxin.

Experiments were performed on a system using the well-known binding affinity between streptavidin and biotin. The two principal concepts of margination-enhanced absorption were demonstrated through an increase in capture efficiency with the presence of red blood cells over plasma alone and with the use of a periodically bifurcated array of microchannels over a simple straight channel array. For this system, the three inch square polycarbonate laminae were coated with 100 μΐ_^ of fluorescently labeled streptavidin at a concentration of 1 mg/mL. This coating was allowed to adsorb overnight, after which the lamina was washed three times in a PBS solution to remove all but the adsorbed monolayer. To confirm adsorption of a streptavidin layer, the fluorescence response of the device surface was measured by microscopy. Either whole blood or isolated blood plasma was spiked with 2.0 μιη biotin-conjugated microspheres and flowed through the device at 0.5 mL/hour for six hours. The concentration of microspheres was measured at the device inlet and outlet to determine the overall capture efficiency of the device under varying operating conditions.

Tests comparing streptavidin and Pluronic F-108 coated laminae were used to rule out non-specific adsorption as the cause of microsphere removal. These tests indicated that the those laminae coated with the F-108 surfactant removed less than 5% of the microspheres, while the streptavidin coated laminae removed approximately between 10 and 40% over the range of conditions tested. Once these evaluations ruled out non-specific absorption as the means of removal, additional comparative tests were run to evaluate the effects of the channel bifurcations as well as the influence of red blood cells on microsphere capture efficiency. These experiments evaluated the microsphere capture efficiency using both whole human blood as well as human plasma through 40 mm long, 200 μιη by 100 μιη channel arrays with and without periodic bifurcations (Table 1).

Table 1.

Comparison of the efficiency of offset and straight microchannels.

The data suggested two distinct trends which strongly supported the proposed theory of margination-enhanced adsorption. In both bifurcated and straight channel arrays, a marked improvement of biotin removal was observed through the presence of red blood cells compared to the experiments using only biotin- spiked human plasma. The extent of biotin removal increased from approximately 10-15% up to 20-30% in straight channels, and from

approximately 15-25% up to 40-50% in channel arrays featuring bifurcations at 2 mm intervals, indicating that the presence of red blood cells had a clear, positive impact on the adsorption of biotin microspheres within the device. Without being bound to a particular theory, this was attributed to the preferential movement of red blood cells to the center of the channels and the subsequent margination of biotin spheres closer to the channels walls where adsorption occurred. Furthermore, it was also apparent that the presence of periodic bifurcations provided an added improvement in capture efficiency from both whole blood and blood plasma. The use of bifurcations yielded an increase of from around 20% to above 40% when processing whole blood, and an increase of from around 10% to greater than 20% when processing plasma only. The increase in removal observed with the incorporation of channel bifurcations but without the presence of red blood cells was likely a result of fluid mixing that was occurring as the steady- state flow profile was disturbed. This mixing provided a similar, but less substantial, improvement in capture efficiency compared with the RBC-induced margination process that occurred in whole blood.

Peptides and Lipopoly 'saccharides

Unless otherwise specified, all reagents were purchased from commercial vendors and were of analytical reagent or higher grade. WLBU2

(RRWVRRVRRWVRRVVRVVRRWVRR, 3.4 Da), CysWLBU2 and WLBU2Cys were obtained from GenScript (Piscataway, NJ). The latter peptides are structurally identical to the original WLBU2 except for an additional cysteine at the N-terminus and C-terminus, respectively. Polymyxin B sulfate (PmB, 1385.6 Da), fibrinogen from bovine plasma (Fib, 340 kDa) and purified Pseudomonas aeruginosa lipopolysaccharide (LPS) were purchased from

Sigma- Aldrich (St Louis, MO). All solutions were prepared using HPLC-grade water, and all peptides and LPS were used as received, without further purification. Fibrinogen was prepared in HPLC, incubated at 37 °C for 2 hours, and 0.45 μιη filtered prior to use. Fibrinogen was used at 2 mg/mL in all cases.

Stock solutions of WLBU2 were made in phosphate buffered saline (PBS, 10 mM sodium phosphate with 150 mM NaCl at pH 7.4), or in 0.5 M HCIO4 for circular dichroism. Working solutions at 50 μΜ or 5 μΜ concentrations were prepared in degassed PBS, using the calculated molar extinction coefficient at 280 nm (16,500 M^"1 cm^"1) of WLBU2. Similarly, 10 mg/mL stock solutions of PmB in degassed PBS were diluted to 50 μΜ or 5 μΜ. LPS was dissolved in PBS to 10 mg/mL, and diluted to 0.1 mg/mL in degassed PBS. All dilute peptide/LPS solutions were prepared and degassed under vacuum with sonication immediately before use.

Surface Modification of OWLS Sensors

Si0₂-coated OW2400c optical waveguide lightmode spectroscopy (OWLS) waveguides (Micro Vacuum, Budapest, Hungary) were cleaned by submersion in 5% (w/v) sodium dodecyl sulfate (SDS) for thirty minutes, followed by 10 minutes at 80 °C in 5: 1: 1 H₂0:27 HC1:30% H₂0₂, then rinsed with HPLC H₂0 and dried under a stream of nitrogen. Cleaned waveguide surfaces were modified with trichlorovinylsilane (TCVS, TCI America, Portland, OR) by a variation of the method of Popat (K.C. Popat, R.W. Johnson, T.A. Desai, Characterization of vapor deposited thin silane films on silicon substrates for biomedical microdevices, Surf. Coat. Tech-nol. 154 (2002) 253-261, incorporated herein by reference). Briefly, clean OWLS sensors were exposed to flowing dry nitrogen in a sealed vessel for 1 hour to remove any residual surface moisture, after which 200 μΐ_^ of TCVS was added and allowed to vaporize at 25 °C, while flowing nitrogen transported the TCVS vapor across the waveguide surfaces. The nitrogen flow was maintained for three hours, after which the sensors were cured at 120 °C for 30 minutes to stabilize the vinylsilane layer. Cleaned and modified sensors were stored in 1.5 mL centrifuge vials under nitrogen in the dark to prevent oxidation of the vinyl moieties.

Optical Waveguide Lightmode Spectroscopy

Silanized waveguides were equilibrated prior to use by incubation overnight in PBS, then rinsed with HPLC-grade H₂0, dried with nitrogen, and immediately installed in the flow cell (4.8 μΐ_^ total volume) of a Micro Vacuum OWLS 210 instrument (Budapest, Hungary) equipped with a 4 mL narrow-bore Tygon® flow loop in line with the flow cell. Incoupling peak angles (+TE and TM) were recorded about four times per minute at 20 °C, and a stable baseline was achieved with PBS prior to the injection of peptide or LPS. Unless otherwise indicated, flow rates were maintained at 50 μί/ι ίηυίε during adsorption and elution steps. Peptides and LPS were introduced as mixtures (competitive adsorption) for 40 minutes, followed by a 40-minute rinse with flowing PBS. Interfacial Tensiometry

o

A FTA model T10 (First Ten Angstroms, Portsmouth, VA) equipped with a Du Nuoy ring (CSC Scientific Co, Fairfax, VA) was used to measure the baseline surface tension of 6.5 mL of PBS, after which 500 μΐ_^ of peptide or LPS stock solution was injected to reach final concentrations of 5 or 50 μΜ WLBU2 or PmB in PBS, with or without 0.1 mg/mL LPS. Data was collected for at least 20 minutes to determine the steady state surface tension of the resulting peptide and/or LPS solutions. The platinum ring was flamed to remove contaminants between experiments.

Dynamic Light Scattering

Apparent particle sizes of peptide and LPS solutions and mixtures were measured at 20 °C by dynamic light scattering (DLS) at 635 nm, using a Brookhaven Instruments 90 Plus Particle Size Analyzer (Holtsville, NY). Ten 1 -minute scans were averaged for each sample, and cumulative size distributions extracted from the multimodal size distribution data.

Surfactant Preparation

Self-assembled PEO brush layers were formed by suspension of hydrophobic silica nanoparticles (R816, Degussa, 190 m²/g, 10-12 nm) in Pluronic^® F127 (PEO101-PPO56-PEO101, -12.6 kDa, BASF) in HPLC water (1 % wt/v) overnight on a rotator. F127 was also used in conjunction with QCM-D sensors, described later. End Group Activated F127 Pluronic, activated with pyridyl disulfide (EGAP-PDS) was obtained from Allvivo Vascular, Inc. EGAP- PDS was incubated with CysWLBU2 in equimolar concentrations for 8 hours at room temperature (EGAP-WLBU2) before use similar to F 127. Surface Modification of QCM-D sensors

QSX303 silicon dioxide QCM-D sensors were purchased from Biolin (Linthicum, MD) and were cleaned by 15 minutes in UV/O3 clean followed by 1 hour in 5 % w/v sodium dodecyl sulfate (SDS) and then another 15 minutes in UV/O3. Cleaned sensor surfaces were modified with trichlorovinylsilane (TCVS, TCI America, Portland, OR) by a variation of the method of Popat. Briefly, clean QCM-D sensors were exposed to flowing dry nitrogen in a sealed vessel for 1 hour to remove any residual surface moisture, after which 200 μΐ_^ of TCVS was added and allowed to vaporize at 25 °C, while flowing nitrogen transported the TCVS vapor across the sensor surfaces. The nitrogen flow was maintained for three hours, after which the sensors were cured at 120 °C for 30 minutes to stabilize the vinylsilane layer. Cleaned and modified sensors were stored in 1.5 mL centrifuge vials under nitrogen in the dark until further use. All silanized sensors were submerged in 1% w/v F127 or EGAP-WLBU2 and exposed to γ-radiation from a ⁶⁰Co source (Oregon State University Radiation Center) for a total dose of 0.3 Mrad to achieve polymer grafting. Sensors were used immediately after surface preparation.

Au coated QCM-D sensors (Biolin, Linthicum, MD) were cleaned by a 15 minute UV/O3 clean followed by 10 minutes in 5: 1: 1 H₂O:30 H₂0₂:27 NH₄OH solution at 80 °C, followed by another 15 minute UV/O3 clean. These sensors were used immediately after cleaning.

Quartz Crystal Microbalance with Dissipation

All modified sensors were submerged in HPLC water for 1 hour prior to instrument use to remove residual F127 or EGAP. The adsorption and elution of peptides, LPS, and fibrinogen were measured with a Q-Sense E4 QCM-D (Q-Sense, Linthicum, MD). QCM-D allows simultaneous measurement of changes in resonance frequency (Af) and energy dissipation (AD) caused by adsorbed mass on QCM-D sensors. For rigid layers, changes in mass can be directly calculated by the Sauerbrey equation:

i

Am = — C—Af

n

Where Am is the change in adsorbed mass, Af is the change in frequency, n is the frequency overtone, and C is a constant parameter characteristic to the quartz crystal, very commonly 17.7 ng/cm²- s.

A high precision peristaltic pump was used to flow sample solutions over QCM-D sensors. Flow rates were maintained at 50 μί/ηιίηυίε, and solution temperature was maintained at 20 °C. QCM-D experiments began by collecting baseline data of a peptide free phosphate buffered saline solution (10 mM PBS, 150 mM NaCl) followed by introduction of WLBU2, or variant, followed by rinse with PBS, a subsequent challenge with LPS, fibrinogen, or a mixture of both, and a final rinse with PBS. All adsorption and elution steps proceeded for 40 minutes. In all QCM-D data presented, the Af is from the 5th overtone. All QCM-D data on surfaces containing pre-adsorbed triblock was baseline adjusted using MatLab prior to use. In brief, this was accomplished by modeling the baseline assuming a simple kinetic model for the removal of excess triblock, suggested by an initial increase in frequency. The model was fit to the baseline by minimizing the residual between the model and the data, and the model was subtracted from the whole subset of data. Circular Dichroism

Peptide secondary structure in the presence or absence of LPS was evaluated in triplicate by circular dichroism (CD) using a Jasco J-815 spectropolarimeter (Easton, MD) at 25 °C.

Spectra were recorded in a cylindrical cuvette (0.1 cm pathlength) from 185 to 260 nm in 0.5 nm increments after calibration with 0.6 mg/mL D(+)-camphorsulfonic acid, and 10 scans/sample were averaged to increase the signal-to-noise ratio. All concentrations of peptides and LPS were the same as for tensiometry and OWLS. The spectra from each of the three replicates for each sample differed only slightly (-5%) in signal intensity. Peptide a-helix content was estimated from CD spectra using DichroWeb.

UV/Vis Spectroscopy

Peptide concentration, as well as the extent of WLBU2 attachment to EGAP-PDS was assessed using a Thermo -Electron Genesys 6 UV-Vis spectrophotometer (Madison, WI).

Concentration of WLBU2 and variants was assessed using the calculated molar extinction coefficient at 280 nm (16,500 M^"1 cm^"1) of WLBU2. The extent of covalent attachment was assessed at 343 nm by the increase in pyridine-2-thione (P2T) concentration (8080 M^"1 cm^"1).

Nuclear Magnetic Resonance

Proton nuclear magnetic resonance (1H-NMR) spectra were taken using a Bruker (Billerica, MA) Robinson 400 MHz NMR spectrometer with TopSpin 2.1 software at room temperature (25 °C) using about 1000 μΜ WLBU2 and y-WLBU2 in D₂0. Each sample was measured using 128 scans. The spectra were post processed by setting the line broadening factor to 0.8 Hz. Results and Discussion

Surface tension depression was recorded for mixtures of LPS (0.1 mg/mL) and peptide at high (50 μΜ) or low (5 μΜ) peptide concentrations in buffer (FIGS. 34-36). In the absence of peptide, LPS vesicles decreased surface tension to a steady value of about 40 mN/m. In contrast, while 50 μΜ PmB slightly reduced surface tension, PMB had almost no effect on surface tension at 5 μΜ (FIG. 34). However, when PmB is mixed with LPS, a faster rate of surface tension decrease is observed at each concentration, and, in the case of 50 μΜ PmB, the surface tension is reduced to a greater extent than observed with LPS alone (FIG. 35).

As with PmB, WLBU2 in the absence of LPS did not substantially decrease surface tension at either concentration (FIGS. 36 and 37). However, unlike PmB, the similarity in the rate and extent of surface tension depression at each WLBU2 concentration suggests that monolayer coverage of the interface is achieved in each case.

The dimensions of the peptides were determined using the open-source viewer Jmol™ from structures of PmB from the NCBI PubChem repository (CID 49800003), and a helical structure of WLBU2 predicted using PEP-Fold (FIG. 38). From those dimensions, the expected surface concentrations of PmB and WLBU2 peptides adsorbed in a monolayer in a "side-on" or "end-on" conformation were estimated, assuming a footprint of the solution dimensions and close-packed rectangular (side-on) or hex-packed circular (end-on) configurations (Table 2). The ratio of the surface tension depression for WLBU2 relative to PmB (FIG. 34 and 35) was about 3.23 at 5 μΜ peptide, and about 1.55 at 50 μΜ peptide. These values fell within limits based on expectations for monolayer coverage.

Table 2.

Size and estimated packing density of PmB and WLBU2 adsorbed "side-on" and "end-on" at an interface.

Mixtures of LPS and WLBU2 behaved quite differently from the mixtures of PmB and LPS (FIGS. 35 and 37). In particular, the presence of WLBU2 with LPS resulted in appreciably reduced surface tension depression when compared to LPS alone (FIG. 37). At the low (5 μΜ) concentration of WLBU2, the surface tension depression was nearly negligible compared to that associated with either WLBU2 or LPS alone. At higher (50 μΜ) WLBU2 concentrations in an LPS-WLBU2 mixture, the surface tension was depressed substantially, but did not reach that of LPS alone.

These results suggested that suspensions of LPS with WLBU2 were more stable than similar suspensions of LPS with PmB. In particular, suspensions of LPS with WLBU2 showed substantially less surface activity (e.g., vesicle adsorption and spreading at the interface) than was exhibited by LPS alone (FIG. 37). In contrast, suspensions of LPS with polymyxin B showed greater surface activity than was observed for LPS alone (FIG. 35). Without being bound to a particular theory, these findings were consistent with the notion that peptide insertion (into the vesicle membrane) and stabilization of intact LPS vesicles occurs in the case of WLBU2, while peptide-induced destabilization of LPS vesicles occurs in the case of PmB. This was further tested by evaluating the adsorption behavior of peptide-LPS mixtures at a hydrophobic solid surface, and observation of peptide secondary structure and particle size distributions in such mixtures.

Competitive adsorption of peptides and LPS at a hydrophobic solid surface.

FIG. 39 shows the adsorption and elution kinetics recorded with mixtures of LPS (0.1 mg/mL) and peptide at high (50 μΜ) or low (5 μΜ) peptide concentrations. The total mass remaining after elution was similar for both mixtures containing PmB, with final adsorbed masses of 74 or 55 ng/cm², respectively. The adsorption kinetics of LPS in the presence of PmB (FIG. 39) were also consistent with the tensiometry results of FIG. 34, and suggested that destabilized LPS vesicles adsorb and spread at the interface.

In contrast, the final adsorbed masses after elution for mixtures containing 0.1 mg/mL LPS and 5 or 50 μΜ WLBU2 were substantially different. The final adsorbed mass was nearly zero at the low peptide concentration, but reached 590 ng/cm² with 50 μΜ WLBU2. The observation of extremely low surface activity (i.e. adsorbed amounts) in WLBU2-LPS mixtures at low peptide concentration was consistent with the tensiometry results (FIG. 37). It also suggested that LPS vesicles formed under these conditions were less able to spread at the hydrophobic surface, presumably due to their association with the membrane- active peptide WLBU2. The reason for the high value of adsorbed mass remaining after elution in the case of the 50 μΜ WLBU2-LPS mixture was not obvious. With reference to FIG. 37, however, the high adsorption would not be consistent with any enhancement of LPS vesicle spreading at the interface. Rather, it suggested that intact WLBU2-associated LPS vesicles were adsorbed at the solid-liquid interface, without spreading.

Peptide structure in peptide-LPS mixtures

WLBU2 structure is substantially disordered in aqueous solution, but becomes increasingly helical in the presence of certain anions (e.g. C104^"), membrane-mimetic solvents, or bacterial membranes. For example, it has been shown that WLBU2 has no appreciable stable structure in water, but reaches 81% a-helix content in an ideal membrane mimetic solvent (30% trifluoroethanol in phosphate buffer). Circular dichroism showed that WLBU2 gained substantial helicity when mixed with LPS (FIG. 40), reaching 78% a-helix content. This suggested that the peptide was located almost exclusively within the membranes of the LPS vesicles. Due to its rigid cyclic structure, PmB would not be expected to become substantially a-helical, and in fact showed no appreciable helical structure under any conditions (FIG. 41). The CD spectrum from the PmB-LPS mixture appeared to be primarily the sum of the CD signal from PmB and LPS alone.

Vesicle size distribution in peptide-LPS mixtures

Dynamic light scattering analysis of peptide-LPS mixtures and peptide-free LPS suspensions are presented in FIG. 42 as the cumulative oversize distribution of particle diameter. The particle size distribution was bimodal in all cases. At the lower mode, the presence of WLBU2 increased the apparent particle diameter of LPS, from 95 ± 11 nm to 195 + 13 nm (mean + standard deviation, n = 3), while addition of PmB had very little effect on particle size in the lower mode (89+ 19 nm). At the upper mode, however, the presence of PmB decreased the mean particle diameter from 408 + 56 nm to 262 + 26 nm, consistent with disruption of the LPS vesicles. In contrast, the presence of WLBU2 greatly increased both the mean and the range of particle sizes, with a mean diameter of 909 + 204 nm. This increase in particle size and polydispersity suggested that WLBU2 induced aggregation of LPS vesicles, rather than disruption.

Visual evidence recorded during the preparation of peptide-LPS suspensions at high concentrations (700 μΜ peptide and 1.4 mg/mL LPS) suggested that the increase in LPS particle diameter in the presence of WLBU2 was not caused by an increase in the individual vesicle size, but rather was due to large-scale aggregation of vesicles (FIG. 43). While there was a slight increase in opalescence of LPS suspensions when PmB was added, the large-scale aggregation observed with WLBU2-LPS did not occur in either the PmB-LPS or peptide-free LPS suspensions. No aggregation was observed in the LPS-free solutions of peptide, even at concentrations as high as 1.4 mM.

Without being bound to a particular theory, these results suggested that peptide insertion and stabilization of intact LPS vesicles occurred in the case of WLBU2, while PmB causes peptide-induced destabilization and disruption of LPS vesicles. Additionally, the results appeared to suggest that PmB caused peptide-induced destabilization and disruption of LPS vesicles. In contrast, WLBU2 did not appear to destabilize LPS vesicles, and may have induced aggregation of the vesicles. Moreover, the results suggested that the high value of adsorbed mass for WLBU2-LPS mixtures at high peptide concentration (FIG. 39) could be attributed to location of intact WLBU2-LPS vesicles or vesicle aggregates at the interface. Conclusion

Analysis of the interfacial behavior of mixtures of LPS and peptides using interfacial tensiometry and OWLS, evaluation of peptide structure using CD, and determination of the particle size distributions using DLS all suggested that peptide insertion into intact LPS vesicles occurred without destabilization in the case of WLBU2, while PmB appears to cause peptide- induced destabilization and disruption of LPS vesicles. In the context of blood purification with hemoperfusion, the most desirable outcome is insertion and tight binding of the peptide in the bacterial membrane or LPS vesicle, without destabilizing the membrane. Disruption and concomitant lysis of the membrane could cause the return of LPS or cellular degradation products to the circulating blood, and is not desirable. Thus, the presentation of WLBU2 at an interface, tethered in a fashion preserving its solvent accessibility and mobility, could promote the capture of pathogens or endotoxin that reach the surface without destabilizing or disrupting the captured vesicle or pathogen. Based on the results provided here, there is no reason to expect a similar outcome with PmB.

Example II

Synthetic Pathway used to make the "Trident" Construct Briefly, a 4-arm "star" PEO polymer (about 10 kDa; Laysan Bio) was first modified by reaction with a single equivalent of succinimidyl 3-(2-pyridyldithio)propionate (SPDP), to introduce a pyridyldisulfide (PDS) group. The remaining three -NH₂ groups were then converted to carboxylic acids by reaction with glutaric anhydride and pyridine (FIG. 44A). The carboxyl/PDS-modified polymer was purified by precipitation in cold methanol and extraction from aqueous bicarbonate with CHCI3. The organic extract was dried with anhydrous MgS0₄, the solvent evaporated, and the polymer dried under vacuum overnight. The terminal -COOH's were converted to amine-reactive NHS-esters by Steglich esterification with N- hydroxysuccinimide (NHS). Needles of dicyclohexylurea were removed by filtration, and the polymer again precipitated in cold methanol. The NHS-activated polymer was then conjugated with three equivalents of H₂N-WLBU2-COOH peptide via NHS-amine reaction in dry DMSO, and precipitated again in cold methanol. Finally, the construct was dried and dissolved in phosphate buffered saline (PBS, pH 7.4), then passed over a column of immobilized

triscarboxyethylphosphine (TCEP) to reduce the PDS terminal group to produce a free thiol (FIG. 44B). Endotoxin-Binding Activity of WLBU2 Immobilized on a Solid Surface FIG. 45 illustrates the ability of the "trident" construct (FIG. 45, top right) to capture endotoxins. The "trident" construct consisted of three WLBU2 peptides covalently linked to three of the four arms of a polyethylene oxide "star" polymer (about 2.5 kDa PEO per arm). The fourth arm of the star was terminated with a free thiol (-SH) group, which formed stable Au-thiol bonds that immobilized the constructs on the gold surface of a quartz-crystal microbalance (QCM) sensor. For comparison, single WLBU2 peptides were also tethered on 5- kDa thiol-terminated PEO tethers (FIG. 45, lower left). Both coatings provided substantial (about 200 ng/cm²) binding of purified Pseudomonas aeruginosa PA- 10 endotoxin in buffer. The WLBU2 "trident" constructs appeared to bind the endotoxin with similar kinetics (solid black line) and final captured amount as the single peptides (dotted). However, elution of the endotoxin from "tridents" was considerably slower than from the single peptides, as evidenced by the shallower slope of the solid black line during the rinse step (t approximately 40 minutes).

In view of the many possible embodiments to which the principles of the disclosed invention may be applied, it should be recognized that the illustrated embodiments are only preferred examples of the invention and should not be taken as limiting the scope of the invention. Rather, the scope of the invention is defined by the following claims. We therefore claim as our invention all that comes within the scope and spirit of these claims.

Claims

We claim:

1. A device, comprising:

a lamina comprising a plurality of offset microchannel walls that define at least one microchannel having a microchannel surface; and

a sorbent molecule associated with the microchannel surface.

2. The device of claim 1, wherein the sorbent molecule is a peptide, protein, enzyme, antibody, antibody fragment, aptamer, single strand DNA, single strand RNA, double- stranded DNA, double-stranded RNA, polysaccharide, glycosaminoglycan, lipid, phospholipid, chelator, antibiotic, synthetic polymer, anticoagulant, anticlotting agent or a combination thereof.

3. The device of claim 1, wherein the sorbent molecule is attached to the surface via a linker.

4. The device of claim 3, wherein the linker is a polymer that, prior to attachment to the surface, comprises at least one unsaturated carbon-carbon bond.

5. The device of claim 3, wherein the linker is an amphiphilic diblock or triblock copolymer.

6. The device of claim 3, wherein the linker comprises polyethylene oxide, polypropylene oxide, polytetrahydrofuran, polybutadiene, polypentadiene, polyhexadiene, polystyrene, polyacrylonitrile, polyhydroxyethylmethacrylate, polyurethane, polyacrylamide or a combination thereof.

7. The device of claim 3, wherein the linker is poly(ethylene oxide)-polybutadiene- poly(ethylene oxide), poly(tetrahydrofuran)-polybutadiene-poly(tetrahydrofuran),

polyhydroxyethylmethacrylate-polybutadiene-polyhydroxyethylmethacrylate, or

polyacrylamide-polybutadiene-polyacrylamide.

8. The device of claim 3, wherein the linker is a branched polymer.

9. The device of claim 8, wherein the branched polymer has a formula

; and n is from 0 to 250.

10. The device of claim 1, wherein the sorbent molecule is a peptide.

11. The device of claim 10, wherein the peptide is an antimicrobial peptide.

12. The device of claim 11, wherein the antimicrobial peptide is an antimicrobial peptide that substantially retains antimicrobial activity after attachment to the surface or the polymer.

13. The device of claim 12, wherein the antimicrobial peptide is an antimicrobial peptide that inserts into a membrane of a blood constituent without substantially disrupting the membrane for a length of time of operation of the device.

14. The device of claim 10, wherein the peptide is polymyxin B (PmB), nisin, WLBU2, WR12, or a mutant or structural variant thereof.

The device of claim 1, wherein the sorbent molecule is an antibody.

16. The device of claim 15, wherein the antibody is an antibody against inflammatory cytokines.

17. The device of claim 15, wherein the antibody is anti-TNF-a or anti-IL-1.

18. The device of claim 1, wherein the sorbent molecule is an anticoagulant or anticlotting agent.

19. The device of claim 18, wherein the sorbent molecule is thrombomodulin, heparin, fondaparninux, idraparinux, heparin sulfate, Coumadin, rivaroxaban, apixaban, edoxaban, hirudin, lepirudin, bivalirudin or structural variations thereof.

20. The device of claim 1, wherein the sorbent molecule is an enzyme.

21. The device of claim 20, wherein the enzyme is heparinase, bilirubin oxidase, phospholipase A2, or combinations thereof.

22. The device of claim 1, wherein the at least one microchannel has a microchannel width of from about 10 μιη to about 750 μιη.

23. The device of claim 1, wherein the at least one microchannel has a microchannel depth of from about 25 μιη to about 1 mm.

24. The device of claim 1, wherein the offset microchannel walls are selected such that red blood cells substantially migrate to the center of each channel during operation.

25. The device of claim 1, wherein the offset microchannels have a bifurcation length of from about 50 μιη to about 1 cm.

26. The device of claim 1, wherein the microchannel walls have a substantially rectangular base shape.

27. The device of claim 26, wherein the substantially rectangular base shape has at least one rounded end.

28. The device of claim 26, wherein the substantially rectangular base shape has at least one triangular end.

29. The device of claim 1, further comprising:

a second lamina comprising a second plurality of offset microchannel walls that define at least one microchannel having a second microchannel surface; and a second sorbent molecule associated with the second microchannel surface.

30. The device of claim 29, wherein the sorbent molecule and the second sorbent molecule are the same.

31. The device of claim 29, wherein the sorbent molecule and the second sorbent molecule are different.

32. The device of claim 29, further comprising a third sorbent molecule, wherein the third sorbent molecule is attached to the microchannel surface, the second microchannel surface, or both.

33. The device of claim 32, wherein the third sorbent molecule is different from the sorbent molecule and the second sorbent molecule.

34. A device, comprising:

a first plurality of laminae, each lamina comprising a first plurality of offset

microchannel walls that define at least one first microchannel having a first microchannel surface;

a first sorbent molecule attached to the first microchannel surface;

a second plurality of laminae, each lamina comprising a second plurality of offset microchannel walls that define at least one second microchannel having a second microchannel surface; and

a second sorbent molecule attached to the second microchannel surface.

35. The device of claim 34, wherein the first sorbent molecule and the second sorbent molecule are different.

36. The device of claim 34, wherein the first sorbent molecule and the second sorbent molecule are the same.

37. The device of any one of claims 34-36, wherein the first plurality of lamina, the second plurality of laminae, or both, comprises from 2 to 1000 laminae.

38. The device of claim 34, further comprising a third sorbent molecule, wherein the third sorbent molecule is attached to the first microchannel surface, the second microchannel surface, or both.

39. The device of claim 34, further comprising a valve, the valve connected to the first plurality of offset microchannel walls and the second plurality of offset microchannel walls, such that during operation of the device a fluid is sequentially or substantially simultaneously directed to the at least one first microchannel and the at least one second microchannel.

40. A method of making a device, comprising:

providing a lamina comprising a plurality of offset microchannel walls that define at least one microchannel having a microchannel surface; and

exposing the at least one microchannel to a sorbent molecule.

41. The method of claim 40, further comprising contacting the microchannel surface with a polymer.

42. The method of claim 41, further comprising exposing the polymer to gamma radiation.

43. The method of claim 40, wherein the sorbent molecule is bound to a polymer.

44. The method of claim 43, wherein the polymer is a branched polymer.

45. A method, comprising:

associating a device with a subject, the device comprising a lamina comprising a plurality of offset microchannel walls that define at least one microchannel having a

microchannel surface, and a sorbent molecule associated with the microchannel surface, to flow a blood stream through the device.

46. A method of removing a constituent from a blood stream, comprising:

providing a device comprising a plurality of offset microchannel walls that define at least one microchannel having a microchannel surface, and a sorbent molecule associated with the microchannel surface; and fluidly coupling the blood stream to flow through a device comprising a lamina comprising a plurality of offset microchannel walls that define at least one microchannel having a microchannel surface, and a sorbent molecule associated with the microchannel surface, whereby the sorbent molecule removes a constituent from the blood stream.

47. The method of claim 46, wherein fluidly coupling the blood stream comprises fluidly coupling a device valve to the blood stream, such that the blood stream flows through the first plurality of microchannels, the second plurality of microchannels or both, during operation.

48. The method of claim 46, wherein the blood stream flows at a sufficient rate such that about a body volume of blood can pass through the device in 60 minutes or less.

49. The method of claim 46, wherein the constituent is a bacterium, endotoxin, cytokine, antibody or a combination thereof.

50. A method of removing a lipopolysaccharide from a blood stream, the method comprising fluidly coupling the blood stream to a device comprising a first plurality of laminae, each lamina comprising a first plurality of offset microchannel walls that define at least one first microchannel having a first microchannel surface, and a first sorbent molecule attached to the first microchannel surface, the first sorbent molecule removing a lipopolysaccharide from the blood stream during operation.

51. The method of claim 50, wherein the device further comprises:

a second plurality of laminae, each lamina comprising a second plurality of offset microchannel walls that define at least one second microchannel having a second microchannel surface; and

a second sorbent molecule attached to the second microchannel surface.

52. The method of claim 50, wherein the first sorbent molecule is WLBU2.