Antibody-Modified Conduits for Extracorporeal Selective Cytokine Filtration in Sepsis by Morgan K. Moroi Submitted to the Department of Mechanical Engineering in partial fulfillment of the requirements for the degree of Bachelor of Science in Engineering as Recommended by the Department of Mechanical Engineering at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY June 2016 @ Massachusetts Institute of Technology 2016. All rights reserved. Author .... Signature redacted 6' Department of Mechanical Engineering May 6, 2016 Cetfidby.Signature redacted Certified by... ....... ....... Robert S. Langer David H. Koch Institute Professor Thesis Supervisor Accepted bySignature redacted by... - -- - tu r -- e - c e d ............. Anette Hosoi MASSACHUSETTS INSTITUTE Professor of Mechanical Engineering OF TECHNOLOGY Undergraduate Officer JUL 08 2016 LIBRARIES
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Antibody-Modified Conduits for Extracorporeal
Selective Cytokine Filtration in Sepsis
by
Morgan K. Moroi
Submitted to the Department of Mechanical Engineeringin partial fulfillment of the requirements for the degree of
Bachelor of Science in Engineering as Recommended by theDepartment of Mechanical Engineering
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
June 2016
@ Massachusetts Institute of Technology 2016. All rights reserved.
Author ....Signature redacted
6' Department of Mechanical EngineeringMay 6, 2016
Accepted bySignature redactedby... - -- - tu r -- e - c e d .............Anette Hosoi
MASSACHUSETTS INSTITUTE Professor of Mechanical EngineeringOF TECHNOLOGY Undergraduate Officer
JUL 08 2016
LIBRARIES
2
Antibody-Modified Conduits for Extracorporeal Selective
Cytokine Filtration in Sepsis
by
Morgan K. Moroi
Submitted to the Department of Mechanical Engineeringon May 6, 2016, in partial fulfillment of the
requirements for the degree ofBachelor of Science in Engineering as Recommended by the Department of
Mechanical Engineering
Abstract
Sepsis kills millions of people worldwide each year and occurs when microorganismsenter the bloodstream of an infected host. The presence of microorganisms in thebloodstream triggers the body to produce many inflammatory proteins, known ascytokines, that cause damage to blood vessels and vital organs. This leads to cap-illary leak, failing organs, and often death. We have developed a novel approach tomodulate the inflammatory response, using antibody-modified conduits (AMCs) tofilter harmful cytokines selectively from the circulation and in a time-specific manner.Here, we characterize variables that affect AMC performance to determine optimalAMC conditions for later use downstream.
Thesis Supervisor: Robert S. LangerTitle: David H. Koch Institute Professor
3
4
Acknowledgments
The author gratefully acknowledges Dr. J. Brian McAlvin, Dr. Daniel S. Kohane,
and Dr. Robert S. Langer for their guidance throughout the completion of this work.
The author would also like to thank all family and friends for their continual support.
Oxidized PDMS tubes were connected in series and 5% (v/v) APTMS in acetone
was circulated through the tubes at room temperature for 1 hour using a peristaltic
pump (Watson-Marlow, Inc., 120U/DV, Wilmington, MA) set at 40.0 rpm (Figure 2-
1). The resulting aminopropyl silane coated tubes were flushed with ethanol followed
by forced air to remove any residual solution. NHS-PEG,-Maleimide in 1X PBS (pH
8.5) (2 mg/mL) was then circulated at room temperature for 1 hour at 160.0 rpm
to produce Maleimide-PEG, modified tubes (Figure 2-1). Tubes were then flushed
with 10 mM 2-(N-morpholino)ethanesulfonic acid (MES) (pH 5.5) followed by forced
air. To quench residual amines, 200 mM acetic anhydride in methanol (MeOH) was
circulated at room temperature for 2 hours at 160.0 rpm (Figure 2-1). Tubes were then
flushed with 10 mM MES (pH 5.5) followed by forced air. The furan-functionalized
antibody solution was diluted with 10 mM MES (pH 5.5) to an antibody concentration
of 0.5 mg/mL and circulated at 37 'C for 13 hours at 160.0 rpm (Figure 2-1). Before
use, antibody-modified conduits were flushed with sterile IX PBS (pH 7.4).
Antibody-modified conduits were stored in sterile IX PBS (pH 7.4) at 2 - 8 'C
until use for a maximum of 2 weeks.
19
CH 3 CH 3 CH 3I I I
H3Cnmo' P'CH
Unmodified PDMS
0 NH 0 NH
0 0
O 0
0 0HN HN
0 0
NH NH NHF
S ' Si 'I I |
Antibody Modified Conduit(AMC)
Air Plasma
LINH
F-uran- FuncticAntibody in
pH 5.5
OH OH OHI I I
HO' I 1 0 'OH
Oxidized PDMS
0 N 0 N7 0
0 0
0 0HN HN
0 0
NH NH NHn lzed2)MES. 2
i' ' Si Si
0100
Maleinide-PEG , Modified Surf,(Residual amines blocked by acyl
with acetic anhydride)
H 3COH 3CO-Si NH2
H 3CO
5%' (v/v) APTMSin acetone
NH 2 NH 2 NH 2
HO' I 10 OH
Aminopropyl SilaneCoated Surface
0 0HN
0 ONHS-PEG,-Maleimidein PBS (pH 8.5)
Acetic Anhydride in MeOH0 0
aceation
Figure 2-1: Schema of formation of an antibody-modified conduit. Conjugation ofantibodies onto the inner surface of silicone tubes was accomplished through foursequential reactions. The first step was to activate the silicone surface by oxygenplasma oxidation to create reactive silanol species (Si-OH). The anchoring molecule(APTMS) was then covalently grafted to the Si-OH moieties. Next, a spacer molecule(NHS-PEG,-Maleimiide) was attached to the primary amine of the APTMS anchor.In the final step, the terminal naleimide on the PEG spacer was reacted with thefuran moieties that were added to the antibodies.
20
2.3 In Vitro Demonstration of Selective Cytokine
Filtration
Antibody-modified conduits with a surface area to volume ratio of 0.834 (Table 2.1)
were prepared with NHS-PEG10-Maleimide and a 0.5 mg/mL infiiximab solution and
tested to determine if selective TNF-a filtration could be achieved (PEG10 , 10 kDa
molecular weight PEG) (infiiximab is a monoclonal antibody that binds TNF-a).
Solutions of 53 (w/v) BSA in sterile lX PBS (pH 7.4) were enriched with 2000
pg/mL TNF-a and 2000 pg/mL VEGF-A and circulated through infiiximab-modified
conduits using the experimental setup in Figure 2-2. The cytokine concentrations
were chosen to be an order of magnitude greater than what would be encountered
clinically during an episode of sepsis to sufficiently challenge AMCs.
Unm>dified Pump-Compa111>le -----4"'- ·.ll
Tube
Unm>dified --Tube Antibody-Modified
Conduit
Figure 2-2: Experimental setup for testing selective TNF-a filtration with infiiximabmodified conduits. An infiiximab-modified conduit was connected in series with two other tubes: (1) an unmodified tube (Cole-Parmer Instrument Company, PVC, ID 3/16, OD 7 /16 , Vernon Hills, IL) compatible with the gear pump system (ColeParmer Instrument Company, Vernon Hills, IL) , and (2) an unmodified tube (SigmaAldrich , PVC, ID 3/16, OD 1/8, St. Louis , MO) to return solution to the reservoir. The unmodified pump-compatible tube was clamped within the pump head, and the free ends of the modified conduit and unmodified tube were placed in a reservoir. Reservoirs were made from 50 mL conical tubes, placed on a 37 °C hot plate (IKA@ Works, Inc. , RCT basic , Wilmington, NC), and covered with Parafilm@ M (Brand GMBH, Wertheim, Germany) to avoid evaporation.
21
Reservoirs were filled with 20 mL of BSA cytokine solution and a 0.5 mL sample
was taken at time zero. The cytokine solution was circulated at 20 mL/min and addi-
tional 0.5 mL samples were obtained at 0.5, 1, 2, 3, and 4 hours. The reservoir volume
and circulation rate were chosen in anticipation of in vivo experiments (rat model of
sepsis), where the rat circulating blood volume is 20-25 mL and the extracorporeal
circulation rate will be 20 mL/min. Samples were stored at -20 'C until analysis.
VEGF-A and TNF-a concentrations in the samples were determined using VEGF-A
and TNF-a enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, Inc.,
Minneapolis, MN).
This experiment was repeated with bevacizumab-functionalized AMCs with a sur-
face area to volume ratio of 0.834, using a 0.5 mg/mL bevacizumab solution (beva-
cizumab is a monoclonal antibody that binds VEGF-A).
2.4 Influence of Flow Rate on Elimination Kinetics
Infliximab-modified conduits with a surface area to volume ratio of 0.834 (Table 2.1)
were prepared with NHS-PEGiO-Maleimide and tested to characterize how flow rate
affects cytokine filtration. Solutions of 5% (w/v) BSA in sterile 1X PBS (pH 7.4)
were enriched with 2000 pg/mL TNF-a and circulated through infliximab-modified
conduits using the experimental setup in Figure 2-2. Experiments were performed
with flow rates of 10, 20, 30 and 40 mL/min to determine the influence of flow rate
on cytokine elimination kinetics. TNF-a concentration was measured by ELISA at
0, 0.5, 1, 2, 3, and 4 hours.
The rate constant, k (min-1 ), was determined for each flow rate as a measure
of how fast cytokines were cleared from the solution. Assuming a one phase decay
model, each rate constant was determined using a nonlinear curve fit.
22
2.5 Influence of Antibody Surface Packing Density
on Elimination Kinetics
Antibody-modified conduits with a surface area to volume ratio of 0.834 (Table 2.1)
were prepared with NHS-PEGIO-Maleimide, followed by the addition of infliximab us-
ing circulating antibody concentrations of 0.05, 0.5, 1.0, 1.5 and 2.0 mg/mL. Solutions
of 5% (w/v) BSA in sterile 1X PBS (pH 7.4) were enriched with 2000 pg/mL TNF-a
and circulated through infliximab-modified conduits at a flow rate of 20 mL/min, us-
ing the experimental setup in Figure 2-2. Samples were obtained at 0, 0.5, 1, 2, 3 and
4 hours, and TNF-a concentrations were determined by ELISA. The rate constant,
k, was determined for each antibody concentration used.
2.6 Influence of Surface Area to Volume Relation-
ships on Elimination Kinetics
Infliximab-modified conduits with various surface area to volume relationships were
prepared with NHS-PEGIO-Maleimide (Table 2.1). Solutions of 5% (w/v) BSA in
sterile IX PBS (pH 7.4) were enriched with 3000 pg/mL TNF-a and circulated at 20
mL/min through infliximab-modified conduits, using the experimental setup in Figure
2-2. Samples were obtained at 0, 0.5, 1, 2, 3 and 4 hours, and TNF-a concentrations
were measured by ELISA. The rate constant, k, was determined for each surface area
to volume ratio.
2.7 Influence of PEG Spacer Length on Elimina-
tion Kinetics
Infliximab-modified conduits with a surface area to volume ratio of 2.50 (Table 2.1)
were prepared with either NHS-PEGiO-Maleimide or NHS-PEG 3-Maleimide. Solu-
tions of 5% (w/v) BSA in sterile IX PBS (pH 7.4) were enriched with 80,000 pg/mL
23
TNF-a and circulated through infliximab-modified conduits at a flow rate of 20
mL/min, using the experimental setup in Figure 2-2. Samples were obtained at
0, 0.5, 1, 2, 3, and 4 hours, and TNF-a concentrations were measured by ELISA. The
rate constant, k, was determined for each PEG spacer length.
2.8 Determination of Cytokine Binding Capacity
Infliximab-modified conduits were prepared with the highest performing characteris-
tics (2.50 surface area to volume ratio, NHS-PEGiO-Maleimide, 1 mg/mL infliximab
solution) to determine the maximum cytokine binding capacity for AMCs. Solutions
of 5% (w/v) BSA in sterile 1X PBS (pH 7.4) were enriched with 80,000 pg/mL TNF-
a to saturate AMCs and circulated through infliximab-modified conduits, using the
experimental setup in Figure 2-2. Samples were obtained at 0, 0.5, 1, 2, 3, 4, 5, 6, 7,
and 8 hours, and TNF-a concentrations were measured by ELISA.
Cytokine binding capacity was calculated using the following equation:
Capacity = ([Cytokine]initiai - [Cytokine]saturation) x Volumecirculating fluid (2.1)
2.9 Statistics
Statistical analyses were performed with Prism (Graph-Pad Software Inc., San Diego,
CA). All data were presented as mean SD. In Sections 3.2 - 3.4, one-way analysis
of variance (ANOVA) was performed to evaluate the significance of the data and
calculate the P value. In Section 3.5, unpaired Student t-tests were performed to
evaluate the significance of the data and calculate the P value.
24
Chapter 3
Results
3.1 In Vitro Demonstration of Selective Cytokine
Filtration
AMCs performed selective cytokine filtration by only clearing the cytokine they were
designed to filter. AMCs modified with infliximab removed 82% of the TNF-a present
at the starting concentration after 4 hours, while VEGF-A levels remained unaffected
(Figure 3-1). AMCs modified with bevacizumab removed 78% of the VEGF-A present
at the starting concentration after 4 hours, while TNF-a levels remained unaffected
(Figure 3-1).
25
E 3000
CL 2500
C
2000 '.
............m 1000e'-h..s
S 0 1 2 34Time (hours)
TNF-a (infliximab conduit)
TNF-a (bevacizumab conduit)
VEGF-A (infliximab conduit)
VEGF-A (bevacizumab conduit)
Figure 3-1: Type of cytokine filtered was determined by the selection of AMC surfaceantibody. Conduits modified with infliximab (monoclonal antibody against TNF-o)removed TNF-o from the circulation. Conduits modified with bevacizumab (mono-clonal antibody against VEGF-A) removed VEGF-A. Data were presented as mean
SD.
9
3.2 Influence of Flow Rate on Elimination Kinetics
Rate of TNF-a elimination was directly proportional to flow rate (Figure 3-2). Curve
fitting analysis revealed that, TNF-a elimination followed a single phase, exponential
decay with strong correlation (R 2 > 0.90 for all groups) (Figure 3-2A). The differ-
ences in k values were determined to be statistically significant (P = 0.0004, one-way
ANOVA) (Figure 3-2B). Higher flow rates were correlated with higher k values, sug-
gesting that faster and more favorable kinetics were achieved at higher flow rates
Figure 3-2: TNF-n filtration by infliximab-modified conduits, where the BSA TNF-a solution was circulated through AMCs at varying flow rates. All conduits had asurface area to volume ratio of 0.834 (Table 2.1) and were prepared with NHS-PEGio-Maleimide and a 0.5 ing/mL infliximab solution. (A) TNF-a elimination profiles foreach group. (B) Exponential decay rate constants (k) for each group. Data were
presented as mean SD.
3.3 Influence of Antibody Surface Packing Density
on Elimination Kinetics
Rate of TNF-a elimination was affected by antibody surface packing density (Figure
3-3). Curve fitting analysis revealed that TNF-a elimination followed a single phase,
Figure 3-3: TNF-o filtration by infliximab-modified conduits with a surface area tovolume ratio of 0.834 (Table 2.1). AMCs were prepared with NHS-PEGio-Maleimnide.Surface antibody addition was performed with solutions of varying infliximab con-centration. (A) TNF-o elimination profiles for each group. (B) Exponential decayrate constants (k) for each group. Data were presented as mean SD.
3.4 Influence of Surface Area to Volume Relation-
ships on Elimination Kinetics
Rate of TNF-o elimination was directly proportional to the ratio of AMC surface
area to volume (Figure 3-4). Curve fitting analysis revealed that TNF-a elimination
followed a single phase, exponential decay with strong correlation (R' > 0.93 for all
groups) (Figure 3-4A). The differences in k values were determined to be statistically
significant (P < 0.0001, one-way ANOVA) (Figure 3-4B). Higher surface area to
volume ratios were correlated with higher k values., suggesting that faster and more
favorable kinetics were achieved by AMCs with higher surface area to volume ratios
(Figure 3-4B).
28
-- +*-- 2.50 mm- 1. -~1
- 0.4-..
---- 0.--- + -- I-7.' --- I.
A -4000
.
S30000
r 2000-
0 1000-
ILL0 0
B25 mm834 mm-1625 mm-1416 mm-1
4.
3-
1. *
0.0 0.5 1.0 1.5 2.0 2.5 3.0Surface Area to Volume Ratio (1/mm)
Figure 3-4: TNF-i filtration by infliximab-modified conduits with varying surfacearea to volume ratios. All conduits were prepared with NHS-PEG1 0 -Maleimide and a0.5 mig/niL infliximab solution. (A) TNF-o elimination profiles for each group. (B)Exponential decay rate constants (k) for each group. Data were presented as mean
SD.
3.5 Influence of PEG Spacer Length on Elimina-
tion Kinetics
AMC perfornanice was enhanced by increasing PEG spacer length (Figure 3-5). Curve
fitting analysis revealed that TNF-a elimination followed a single phase. exponential
decay with strong correlation (B2 = 0.73, 0.99 for 3 kDa PEG. 10 kDa PEG. respec-
tively) (Figure 3-5A). The difference in k values was determined to be statistically
Figure 3-5: TNF-o filtration by inflixinab-iodified conduits. Conduits with a surfacearea to volune ratio of 2.50 were prepared with varying PEG spacer lengths and a0.5 nig/mL inflixiinab solution. (A) TNF-i elimination profiles for each group. (B)Exponential decay rate constants (k) for each group. Data were presented as mean
2.50 surface area to voluie ratio) performed selective cytokine filtration until satu-
ration (Figure 3-6). Using Equation 2.1, the binding capacity for an ideal AMC was
found to be 394 47 pg/nun 2.
2'1000000.
C 800000 .1
0* 60000C
40000,0
20000-
0'*0 1 2 3 4 5 6 7 8
Time (hours)
Figure 3-6: TNF-o, filtration by ideal infliximab-modified conduits (2.50 surface areato volume ratio, NHS-PEGIO-Maleinide, 1 ng/niL infliximab solution). Cytokinebinding capacity for this ideal type of AMC was determined to be 394 47 pg/1i11112.Data were presented as mean SD.
30
Chapter 4
Discussion
The purpose of this thesis was to demonstrate the concept of selective blood filtra-
tion, using cytokines as the model target. Ultimately, these data will be used to
develop a novel therapeutic approach to the treatment of sepsis, where circulating
blood is purified by selective removal of the signaling molecules that drive systemic
inflammation. Here, we successfully characterized the variables that influence AMC
performance.
Ideally, AMCs must fulfill the following criteria: they must (1) only bind the
targeted cytokine, and (2) be able to reduce clinically relevant cytokine levels in a
timely manner and maintain them at zero. Objective (1) was fulfilled by circulating
two types of cytokines through an AMC and observing only the removal of the type
of cytokine targeted by the antibody (Section 3.1). Objective (2) was fulfilled by
characterizing AMC variables and designing an ideal AMC to have quick elimination
kinetics and a large cytokine binding capacity (Sections 3.2 - 3.6).
AMCs were determined to be capable of selective cytokine filtration. Conduits
cleared targeted cytokines, whereas non-targeted cytokines remained unaffected and
in the circulating fluid. This proof of concept provides confidence that conjugating
any type of antibody to the inner surface of silicone circuits will allow for the rapid
removal of specific cytokines. By applying this concept and expanding this technology
to target other key cytokines involved in sepsis, we hope to be able to successfully
create a platform for immune modulation via selective cytokine filtration.
31
Before moving forward with expansion and further testing, it was important to
characterize variables that affected AMC performance so that the highest performing
AMCs could be used for future experimentation. These variables included: fluid flow
rate, antibody loading concentration, tubing dimension, and PEG spacer length.
In varying flow rates, we observed that higher flow rates were correlated with
faster elimination kinetics. By analogy, solute clearance with traditional hemodialysis
is enhanced by increasing the flow rate of blood [16]. Given that the highest flow rate
tested (40 mL/min) achieved the best kinetics, a flow rate of 40 mL/min will be
used in future tests. 40 mL/min is equivalent to 100 mL/kg/min for a 400 gram rat,
and this normalized flow rate is in range with those achieved during extracorporeal
membrane oxygenation (ECMO) in children and infants (75 - 150 mL/kg/min) [17].
In varying antibody loading concentration, we discovered that AMC performance
was enhanced up to an antibody loading concentration of 1.0 mg/mL. Above 1.0
mg/mL, the kinetics remained the same. We hypothesize that this occurred because
either: (1) the AMCs became saturated with antibodies when circulated with con-
centrations of 1.0 mg/mL or higher, or (2) the AMCs became oversaturated with an-
tibodies when circulated with concentrations greater than 1.0 mg/mL, increasing the
extent of steric hindrance and preventing antibodies from favorably binding cytokines
[18]. In the future, AMCs will be prepared with an antibody loading concentration
of 1.0 mg/mL.
In varying AMC dimensions, we observed that AMCs with higher surface area to
volume ratios had faster elimination kinetics. This is most likely because AMCs with
a large surface area to volume ratio possess a large blood-circuit interface, allowing
their surface antibodies to be exposed to a greater proportion of the circulating fluid.
Given this information, future AMCs will be prepared with silicone tubing with a
surface area to volume ratio of 2.50. It is important to note, however, that changing
the surface area to volume ratio will also change the fluid dynamics of the system.
The resistance to blood flow is directly proportional to the length and inversely pro-
portional to the radius to the fourth power. This could possibly cause hemolysis due
to increased shear forces, as increased resistance has been shown to cause hemolysis in
32
extracorporeal circuits [19]. Future in vitro studies will need to be conducted, using
blood as the circulating fluid, to determine whether AMCs with an aspect ratio of
2.50 cause hemolysis.
In varying PEG spacer length, we found that AMCs prepared with a 10 kDa PEG
spacer outperformed those prepared with a 3 kDa PEG spacer. The use of spacers to
distance immobilized antibodies from a surface has been shown to introduce greater
freedom of antibody motion [20], reduce steric interference [20,21], and improve sur-
face antibody orientation [21]. Additionally, it is known that the use of PEG spacers
increases antibody-antigen interactions by up to 90% and is dependent upon spacer
length [21]. Thus, it is not surprising that the longer spacer displayed faster elimina-
tion kinetics. Moving forward, AMCs will be prepared with NHS-PEGiO-Maleimide
(PEG10 , 10 kDa PEG).
Given these findings, ideal tubes (NHS-PEGio-Maleimide, 1.0 mg/mL antibody
loading concentration, 2.50 surface area to volume ratio, 40 mL/min flow rate) were
tested, and the total cytokine binding capacity was calculated. The total binding
capacity was determined to be 394 47 pg/mm 2 . This capacity will be helpful in
estimating the number of AMCs needed for use in various in vivo models.
We now know that selective cytokine filtration with AMCs yields rapid elimination
of a specific cytokine with the capacity to sustain elimination for prolonged periods.
Moving forward, we plan to modulate cytokine concentrations in vivo. While we
have demonstrated proof of concept, a crucial test of this technology will be to use
AMCs, singly or in combination, to regulate cytokine levels in rats rendered septic.
In testing our device in vivo, we hope to validate that customized AMC combinations
can provide precise temporal control over circulating cytokine concentrations, thereby
preventing organ failures and death.
Also, it is important to note that while we chose sepsis as the model disease for this
technology, the underlying mechanism being addressed is the systemic inflammatory
response syndrome (SIRS), which is a common mechanism for a variety of critical