ARTICLE Conjugation of Methoxypolyethylene Glycol to the Surface of Bovine Red Blood Cells Sharon I. Gundersen,y Andre F. Palmerz Department of Biomolecular and Chemical Engineering, University of Notre Dame, 182 Fitzpatrick Hall, Notre Dame, Indiana 46656; telephone: 574-631-4776; fax: 574-631-8366; e-mail: [email protected] Received 5 June 2006; accepted 30 August 2006 Published online 28 September 2006 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/bit.21204

ABSTRACT: Methoxypolyethylene glycol (mPEG) covalently bound to the surface of human red blood cells (hRBCs) has been shown to decrease immunological recognition of hRBC surface antigens (Bradley et al., 2002). However, there is an increasing shortage of hRBC donations, thus making hRBCs scarce and expensive (Davey, 2004; Riess, 2001). The goal of this study is to similarly PEGylate the surface of bovine RBCs (bRBCs) with the aim of reducing the demand on human blood donations needed for blood transfusions. This study investigates the feasibility of modifying the surface of bRBCs with the succinimidyl ester of methoxypolyethylene glycol propionic acid (SPA-mPEG) for use as a potential blood substitute. The oxygen binding affinity of PEGylated bRBCs was moderately increased with increasing initial SPA-mPEG concentrations up to 4 mM when reacted with bRBCs at a hematocrit of 12%. Oxygen transport simulations verified that SPA-mPEG conjugated bRBCs could still transport oxygen to pancreatic islet tissues even under extreme conditions. PEGylated bRBCs reconstituted to a hematocrit of 40% exhibited viscosities on the order of !3 cp, similar to hRBCs at the same hematocrit. Taken together, the results of this study demonstrate the success of PEGylating bRBCs to yield modified cells with oxygen binding, transport and flow properties similar to that of hRBCs. Biotechnol. Bioeng. 2007;96: 1199–1210. ! 2006 Wiley Periodicals, Inc. KEYWORDS: methoxypoly(ethylene glycol); stealth erythrocytes; bovine red blood cells; transfusion

Introduction Due to the large variety of blood transfusion needs, blood is in very high demand with an average of 38,000 units (1 unit !450 mL) of human red blood cells (hRBCs) being used every day in the US, supplied solely by volunteer blood

donations (Riess, 2001; Scott et al., 1997b). Threat of disease transmission, complex donor screening process, and costs of storage and handling of blood has resulted in a decreasing US blood supply with a predicted shortage of approximately 4 million units by 2030 (Riess, 2001). Thus, there is a very high demand for a sterile, cost-effective, and universally acceptable alternative to current allogeneic hRBC transfusions. Xenogenic blood from cows could provide a stable blood supply from a controlled and monitored source. However, the major restriction to xenotransfusion would be possible immune recognition of surface antigens on the surface of bovine RBCs (bRBCs). Hence, the surface of bRBCs would have to be modified to overcome this immunological barrier. Research has been ongoing to test the feasibility of modifying the entire surface of hRBCs to produce an immunologically silent RBC by either removing surface antigens or hiding them to prevent antibody binding (Blackall et al., 2001; Scott, 1997a). Unfortunately, removing surface antigens by enzymatic cleavage has been shown to compromise the RBC’s structural and functional integrity (Blackall et al., 2001). Coating the RBC membrane to hide surface antigens is one potential strategy to preserve the structural integrity of the RBC. Poly(ethylene glycol) (PEG) has the potential to mask RBC antigens due to its non-toxic and non-immunogenic characteristics (Jo and Park, 2000; Scott et al., 2000). PEG conjugation neither denatures RBC surface proteins nor influences diffusion and mass transfer of small molecules such as oxygen (Bradley et al., 2002; Scott et al., 2000). PEG has been used extensively to modify surfaces, and it has many unique properties that prevent protein and bacteria adsorption on surfaces such as its hydrophilicity, flexibility, large exclusion volume in water, and is sterically bulky providing a polymer cloud around modified molecules or cells (Holmberg et al., 1993;

y

Graduate Student. Assistant Professor. Correspondence to: A.F. Palmer z

! 2006 Wiley Periodicals, Inc.

Biotechnology and Bioengineering, Vol. 96, No. 6, April 15, 2007

1199

Jo and Park, 2000). PEG has also been shown to prevent adsorption of IgG (a protein involved in activating the complement system), fibrinogen (common plasma protein), and the Streptococcus bacterium onto surfaces (Holmberg et al., 1993). The unique physical properties of PEG have inspired investigations on the effects of covalently binding PEG to the RBC membrane. Numerous studies have conjugated methoxypolyethylene glycol (mPEG) to the hRBC membrane (mPEG-hRBCs), in order to camouflage hRBC surface antigens (Bradley et al., 2002; Murad et al., 1999; Scott et al., 2000). Previous studies demonstrated that conjugating mPEG to the hRBC membrane (PEGylation) had almost no effect on the hRBC’s structure and function. These mPEG-hRBCs were reported to have the same oxygen binding properties as unmodified hRBCs and their deformability was unaltered. Although the chemical cross-linker on the mPEG molecule usually binds to the surface protein involved in anion transport, mPEG modification did not affect cellular ion concentrations (Bradley et al., 2002). Several studies have shown that PEGylation of hRBCs alleviates potential immune responses by preventing cell– cell interactions required for RBC aggregate formation and inhibiting antibody binding (Armstrong et al., 1997; Bradley et al., 2002). Modification of hRBCs with 5 kDa mPEG exhibited a decreased response by anti-sera to the A antigen compared to unmodified hRBCs (Scott, 1997). However, further studies by Bradley et al. (2002) showed that larger mPEG molecules provided longer in vivo RBC survival. Bradley attributed this behavior to the larger mPEG molecule’s ability to immunocamouflage larger surface proteins such as the Kidd protein, involved in urea transport (Bradley et al., 2002; Garratty et al., 2002). Bradley et al. (2002) also found that 20 kDa mPEG was sufficiently large enough to camouflage Kidd blood group antigens. However, conjugating the surface of hRBCs with mPEG is still dependent on the collection of blood from human donors. One possibility of creating a RBC substitute for use in humans would involve conjugating the surface of bRBCs with mPEG (mPEG-bRBCs). Though pigs are considered the best possible source for organ and tissue transplantation, cows provide a more suitable source for blood donations, due to their larger blood volume and easier vein access (Johnstone et al., 2004; Otchet, 2001; Williams, 1996). In addition, bRBCs are more mechanically robust, and have better osmotic stability compared to porcine RBCs. In this study, mPEG was conjugated to the surface of bRBCs, similar to previous conjugations to hRBCs, in an attempt to create a potential non-immunogenic RBC. Thus far we have examined the physical effects of modifying bRBCs with varying concentrations of 20 kDa mPEG to identify optimum reaction conditions that yield a functional RBC. The physical properties of mPEG-bRBCs should be similar to hRBCs, in order to produce effective hRBC substitute with substantial in vivo half-life. Since the bRBC is smaller in size compared to a hRBC (diameter !5.8 mm)

1200

and considered a foreign object, it is not expected to last past the 60–90 day half-life of transfused hRBCs (Biopure, 2004). We predict that the circulation half-life of mPEG-bRBCs will be greater than current LEHb based blood substitutes (>60 h) (Biopure, 2004; Phillips et al., 1999) with clearance mediated through the reticuloendothelial system.

Materials and Methods Bovine Red Blood Cells This study used bRBCs purchased from Quad Five (Ryegate, MT). After purchase all bRBC samples were washed three times in PBS (pH ¼ 7.2) by centrifuging at 950 rpm for !30 min, discarding supernatant, and resuspension in PBS three times. Once washed bRBCs were suspended to 40% hematocrit. Quad Five carefully selects and quarantines their donor animals. All animals are maintained under veterinary care and diagnostic protocols, and bled in separate facilities to maintain quarantine conditions. The cattle are routinely screened for tuberculosis, brucellosis, blue-tongue virus, anaplasmosis, leptospirosis, IBR, P13, BVD, and bovine leukosis virus (Micks, 2005).

mPEG Derivative The 20-kDa succinimidyl ester of methoxypolyethylene glycol propionic acid (SPA-mPEG) was purchased from Nektar Therapeutics (Huntsville, AL). The succinimidyl ester moiety mPEG was designed for protein surface modification and was recommended by Nektar for this purpose (Harris and Kozlowski, 1995; Nektar Therapeutics, 2006). These linear monofunctional polymers are capped on one end by a methoxy group to produce products free of cross-linking (Nektar, 2006). The N-hydroxysuccinimide active ester on SPA-mPEG binds to amino acid residues with amines (such as lysines) producing a stable amide link. Since mPEG is covalently linked to surface proteins of bRBCs, and is free of hydrolysis-prone ester linkages, it is unlikely that mPEG will elute or fragment in vivo.

mPEG Derivatization bRBCs were diluted to 12% hematocrit in PBS (pH ¼ 7.2) prior to reaction. Solutions of mPEG were prepared at varying concentrations, 0 mM (buffer control) to 4 mM in 1 mM increments in 50 mM K2HPO4 and 105 mM NaCl (mPEG buffer). An additional unreacted control of bRBCs mixed with PBS was also evaluated to determine if the reaction procedures had any effect on the bRBCs. Bovine erythrocytes and mPEG solutions were mixed in equal volumes, and allowed to react for 1 h at room temperature and pressure on a 3-D rotator as described previously (Bradley et al., 2001, 2002). At the end of the reaction, modified bRBCs were washed three times in PBS and

Biotechnology and Bioengineering, Vol. 96, No. 6, April 15, 2007 DOI 10.1002/bit

resuspended to 3 mL. To determine if any cells lysed during the reaction and subsequent washes, the amount of hemoglobin in the supernatant was assayed with a UV– visible spectrophotometer (Varian, Inc., Palo Alto, CA) as described previously (Bonsen et al., 1977).

Partitioning To determine the extent of bRBC modification with mPEG, mPEG-bRBCs were partitioned in a two-phase system with 8,000 g/mol PEG and dextran (!71 kDa) (Dx) as described previously (Bradley et al., 2002; Gavasane and Gaikar, 2003; Lutwyche et al., 1995; Scott et al., 2000). The PEG stock solution was prepared by mixing 30 g of 8,000 g/mol PEG and 100 g of PEG solution buffer (0.15 mM NaCl, 6.84 mM Na2HPO4, and 3.16 mM NaH2PO4, pH ¼ 7.4). The Dx solution consisted of 30 g of Dx and 100 g of DI water. The two-phase system was prepared as described previously from the stock solutions with 5% (w/w) Dx, 4% (w/w) PEG in 50 mL polypropylene tubes and equilibrated overnight. The upper PEG-rich phase was drawn off, and the bottom Dx-rich phase was collected by puncturing a hole at the bottom of the tube. For the partitioning experiment, 12 mL of mPEG-bRBCs was mixed with 750 mL of top PEG-rich phase, followed by 750 mL of the bottom Dx-rich phase. The mixture was inverted 20 times and allowed to equilibrate for 20 min. After the mixture of PEG, Dx, and cells was allowed to equilibrate, mPEG-bRBCs remained in the PEG-rich (top) phase and most of the unmodified, or slightly modified, bRBCs settled to the PEG–Dx interface (Birkenmeier et al., 1994; Scott et al., 2000). Triplicate cell counts on the top phase with a hematocytomer determined the percentage of bRBCs remaining in the top phase, which corresponded to the percentage of mPEG-bRBCs.

HPLC Analysis The reaction supernatant was assayed on a BioSep-S2000 size exclusion HPLC column (Phenomenex, Torrance, CA) to determine the amount of unreacted mPEG. The mPEG buffer was used as the mobile phase. The mass of reacted mPEG and extent of reaction was calculated from the amount of unreacted PEG. Each sample of the original mPEG solution was diluted to 1 mg mPEG/mL mPEG buffer and 10–100 mL was injected to calibrate the peak area to injected mass. To determine the remaining mPEG mass, the reaction supernatant was diluted to approximately 1 mg mPEG/mL mPEG buffer assuming half the initial mPEG reacted. Twenty to 60 mL of the diluted reaction supernatant was injected and the mass of mPEG was calculated from the peak area. Oxygen Binding Properties As done previously in our group, a Hemox Analyzer (TCS Scientific Corp., New Hope, PA) measured the oxygen

dissociation curves of unmodified bRBCs and mPEGbRBCs (Eike and Palmer, 2004; Gordon et al., 2005). Fifty microliters of the bRBC suspension was mixed with 5 mL of Hemox buffer, 10 mL of anti-foaming agent, and 20 mL of Hemox A additive (TCS Scientific Corp). Air was bubbled through the sample until it reached equilibrium at a pO2 of !150 mmHg. As the sample was deoxygenated by the addition of nitrogen, the Hemox Analyzer records the fraction of oxygen saturated Hb as a function of pO2 until a pO2 of !1.9 mmHg was reached. The Hemox Analyzer software fitted the oxygen binding data to the Hill equation, Equation 1, to yield the P50 and cooperativity of oxygen binding (Hill coefficient) at physiological temperature (378C). The data were also fit to a six-parameter step-wise Adair equation used in later simulations (Eq. 2). Y¼

Y¼

pOn2 n þ P50

pOn2

A1 pO2 þ 2A2 pO2 þ 3A3 pO2 þ 4A4 pO2 4ð1 þ A1 pO2 þ A2 pO2 þ A3 pO2 þ A4 pO2 Þ

(1)

(2)

where Y ¼ fraction of oxygen saturated Hb, pO2 ¼ oxygen partial pressure, P50 ¼ oxygen partial pressure when Hb is half saturated with oxygen (Y ¼ 0.5), n ¼ cooperativity (Hill coefficient), and Ai ¼ Adair parameters.

Rheology A Carri-Med controlled stress rheometer with a 4 cm parallel plate (Carri-Med Ltd., Dorking, Surrey, England, no longer in manufacture) was used to measure the shear rate versus applied shear stress (0 to !64 dyne/cm2) of mPEG-bRBCs at 40% hematocrit at 378C. Since blood is a non-Newtonian fluid, the shear stress–shear rate curve was fit to the Casson equation to regress the high shear rate viscosity and the yield stress. Due to daily variances in setting the rheometer’s gap width, the high shear rate viscosity was normalized such that a control of plain washed bRBCs at 40% hematocrit in PBS had a viscosity of 3 cp.

Capillary Wetting The capillary wetting experiment examining the deformability of mPEG-bRBCs was described previously by Zhou and Chang (2005) and Zhou et al. (2006). Briefly, mPEGbRBCs were diluted in PBS to 1% hematocrit. A 10-mL drop was placed at the end of a glass capillary tube with a 21 mm inner diameter (Polymicro Technologies, Phoenix, AZ). The blood suspension enters the tube by capillary wetting penetration. The meniscus flow was recorded using a highspeed video camera at 500 frames/s, and the meniscus

Gundersen and Palmer: PEGylated Bovine Erythrocytes Biotechnology and Bioengineering. DOI 10.1002/bit

1201

velocity was calculated at five different lengths down the capillary.

Oxygen Transport Simulations To insure modified bRBCs were still capable of transporting oxygen to tissues under a variety of simulation conditions, oxygen transport simulations were conducted to model oxygen transport from the mPEG-bRBCs in a capillary to the surrounding tissue. Although the changes in oxygen binding properties (P50 and n) were not very considerable, simulations were conducted on all mPEG-bRBC solutions and controls consisting of bRBCs and hRBCs. The oxygen transport model described herein is a modification of the Krogh tissue cylinder model, shown in Figure 1. In this model, blood is assumed to behave as a homogenous fluid with oxygen in equilibrium with oxyhemoglobin as described mathematically by the specified oxygen binding characteristics of the Adair equation. The Hemox-derived Adair parameters were used for mPEGbRBC solutions and controls. The Adair parameters for human RBCs were taken from McCarthy et al. (2001). Oxygen diffuses through and out of the plasma and through the capillary wall where some of it is consumed by the capillary wall’s endothelial cells. Next, it diffuses through the interstitial space to the tissue space where it is primarily

Figure 1. Modified Krogh tissue cylinder model including the capillary wall and interstitial space between the capillary wall and the first tissue cells. [Color figure can be seen in the online version of this article, available at www.interscience.wiley.com.] consumed by the tissue cells. Tables I and II summarize the dimensions and parameters used in the model. Capillary Space Differential mass balances on dissolved oxygen (O2) and oxygenated hemoglobin (HbO2) were derived for the capillary lumen. O2 and HbO2 were assumed to be in chemical equilibrium, thus RO2 ;formation ¼ &RHbO2 ;formation .

Table I. Oxygen transport simulation parameters. References Capillary space pO2,in rc L Vz [Hb]total DO2;plasma DHb,plasma HO2;plasma Adair Constants listed in Table II Capillary wall rw rw- rc DO2;wall HO2;wall Vmax,wall Interstitial space ri ri- rw DO2; interstitium HO2;interstitium Tissue space rT rT – ri DO2;tissue HO2;tissue Km Vmax

1202

20, 40, 60, 80, 95 mmHg 5 mm 0.1 cm 0.5 cm/s 8,800 mM 1.85E-05 cm2/s

Model parameters Fournier (1999) Fournier (1999) Fournier (1999) Fournier (1999) Vadapalli et al. (2002)

5.30E-07 cm2/s 0.74 mmHg/mM

Vadapalli et al. (2002) Fournier (1999) Experimental and McCarthy et al. (2001)

5.3 mm 0.3 mm 8.73E-06 cm2/s 0.576 mmHg/mM 4.4615 mM/s

Vadapalli et al. (2002) Vadapalli et al. (2002) Vadapalli et al. (2002)

5.64 mm 0.34 mm 2.81E-05 cm2/s 0.798 mmHg/mM

Vadapalli Vadapalli Vadapalli Vadapalli

13.7 mm 8.06 mm 6.30E-06 cm2/s 0.678 mmHg/mM 0.44 mmHg 26 (low glucose) mM/s 46 (high glucose) mM/s

Fournier (1999)

Vadapalli et al. (2002)

et et et et

al. al. al. al.

(2002) (2002) (2002) (2002)

Vadapalli et al. (2002) Vadapalli et al. (2002) Fournier (1999) Fournier (1999) Fournier (1999)

Biotechnology and Bioengineering, Vol. 96, No. 6, April 15, 2007 DOI 10.1002/bit

Table II. Adair constants for mPEG-bRBCs and hRBCs.

4 mM 3 mM 2 mM 1 mM 0 mM PBS hRBCs

A1

A2

A3

A4

0.0441 0.0715 0.0349 0.0403 0.0585 0.0285 0.0153

3.917E-04 3.992E-04 6.550E-04 5.750E-04 3.768E-04 3.805E-04 1.100E-03

8.545E-06 3.851E-06 5.875E-13 5.601E-11 4.163E-13 3.868E-10 1.240E-07

4.331E-06 5.265E-06 3.388E-06 3.891E-06 2.998E-06 2.298E-06 1.810E-06

The two differential mass balances were combined to eliminate the rate term yielding one equation in terms of oxygen partial pressure (pO2) (Eq. 3). The equilibrium relationship between HbO2 and O2 is described mathematically by the Adair equation. The simulated inlet pO2s (z ¼ 0) (Eq. 4.a) are listed in Table I. The remaining boundary conditions consisted of convective flux of oxygen at the outlet of the capillary (z ¼ L), axial symmetry at the center of the capillary (r ¼ 0), and continuity of oxygen at the capillary wall (r ¼ rc) (Eqs. 4.b–d). The convective flux boundary condition was used at z ¼ L in the capillary (and all other regions), since it assumes that all endothelial and tissue cells are absent and no more oxygen can be consumed at that point. The length and radius of the capillary space and plasma properties of blood are average representative values for human capillaries as per Fournier (1999). Vz ð1 þ mÞ

@pO2 @z

¼ ðDO2 ;plasma þ mDHbO2 ;plasma Þ # 2 ! "$ @ pO2 1 @ @pO2 r ' þ r @r @z 2 @r

(3)

@Y where m ¼ HO2 ½Hb)total @pO Y ¼ fraction of oxygen saturated 2 Hb

z¼0

r ¼ rc

pO2 ¼ pO2;in

(4.a)

z¼L

@pO2 ¼0 @z

(4.b)

r¼0

@pO2 ¼0 @r

(4.c)

DO2 ;plasma

@pO2 @pO2 ¼ DO2 ;wall @r @r

cells. Equations 6.a–c represent the continuity boundary conditions for the sides of the capillary wall, r ¼ rc and r ¼ rw, and the isolated symmetry boundary condition for both ends of the capillary at x ¼ 0 and x ¼ L. The oxygen consumption kinetics of endothelial cells, diffusional properties, and capillary wall dimensions were taken from Vadapalli et al. (2002). ! "# ! "$ DO2 ;plasma @2 pO2 1 @ @pO2 r 0¼ þ r @r HO2 ;wall @z 2 @r þ Vmax;wall

(5)

z ¼ 0 and z ¼ L r ¼ rc r ¼ rw

(6.a)

@pO2 @pO2 ¼ DO2 ;wall @r @r

(6.b)

@pO2 @pO2 ¼ DO2 ;interstitium @r @r

(6.c)

DO2 ;plasma

DO2 ;wall

@pO2 ¼0 @r

Interstitial Space The interstitial space was modeled with a differential mass balance on O2 with model dimensions as per Vadapalli et al. (2002) (Eq. 7). It is assumed that the interstitial space has no tissue cells, thus oxygen is not produced or consumed in this region. The boundary conditions consisted of continuity for both sides of the region, r ¼ rw and r ¼ ri, and isolated symmetry for both ends of the interstitial space at x ¼ 0 and x ¼ L (Eqs. 8.a–c). 0¼

!

DO2 ;interstitium HO2 ;interstitium

"#

! "$ @2 pO2 1 @ @pO2 r þ r @r @z 2 @r

z ¼ 0 and z ¼ L

@pO2 ¼0 @r

(7)

(8.a)

@pO2 @pO2 ¼ DO2 ;interstitium @r @r

(8.b)

@pO2 @pO2 ¼ DO2 ;tissue @r @r

(8.c)

r ¼ rw

DO2 ;wall

r ¼ ri

DO2 ;interstitium

(4.d) Tissue Space

Capillary Wall A differential mass balance on O2 was derived for the capillary wall, since it is assumed that Hb and HbO2 cannot diffuse through the wall (Eq. 5). Zeroth order kinetics describe O2 consumption by the capillary wall endothelial

Equation 9 shows the differential O2 mass balance in the tissue space with first-order Michaelis–Menten kinetics describing oxygen consumption of pancreas islet tissues at high and low glucose levels. The two glucose levels correspond to high and low Vmax, respectively, to provide a range of oxygen consumption rates. The boundary

Gundersen and Palmer: PEGylated Bovine Erythrocytes Biotechnology and Bioengineering. DOI 10.1002/bit

1203

conditions consisted of isolated symmetry for the three sides of the modeled tissue space at x ¼ 0, x ¼ L, and r ¼ RT, and a continuity boundary condition between the interstitium and tissue space (Eqs. 10.a–b). The oxygen consumption kinetics and tissue dimensions were modeled as per Fournier (1999). 0¼

!

DO2 ;tissue HO2 ;tissue

þ

"#

Vmax pO2 Km þ pO2

! "$ @2 pO2 1 @ @pO2 r þ r @r @z 2 @r (9)

z ¼ 0; z ¼ L; and r ¼ RT r ¼ ri

DO2 ;interstitium

@pO2 ¼0 @r

@pO2 @pO2 ¼ DO2 ;tissue @r @r

(10.a)

(10.b)

Results Partitioning experiments suggest that bRBCs are being effectively surface conjugated with mPEG. bRBCs, initially at 12% hematocrit in PBS reacted with 0–4 mM SPA-mPEG. Figure 2 shows the partitioning results as a ratio of the percentage of mPEG-bRBCs in the top PEG-rich phase to the percentage of unreacted (PBS control) bRBCs in the PEG-rich phase. The 0-mM control corresponding to that day’s experiment overlaps the experimental sample. Since there was some variability in the amount of partitioning between experiments done on different days, all data were

normalized to the PBS control to be able to compare the data. bRBCs modified with 2, 3, and 4 mM SPA-mPEG noticeably partitioned into the top (PEG-rich) phase to a greater extent versus the controls. Reactions with 4 mM mPEG partitioned up to 10 times more compared to the PBS control. bRBCs reacted with 2 and 3 mM mPEG partitioned about two and six times more compared to the control, respectively. The reaction with 1 mM mPEG had only slightly increased partitioning. As would be expected, greater concentrations of mPEG resulted in the formation of more mPEG-bRBCs favoring the PEG phase, thus indicating a higher amount of mPEG surface coverage. HPLC analysis (Fig. 3) shows the amount of unreacted mPEG remaining in solution after the 1-h reaction period. The percentage of reacted SPA-mPEG was greatest with 1 mM SPA-mPEG at 50.5% with other mPEG concentrations (2, 3, and 4 mM) resulted in approximately 38% of reacted mPEG. At 4 mM mPEG, 90.3 mg mPEG reacted with bRBCs. At lower mPEG concentrations, 29.8, 44.2, and 66.4 mg of mPEG reacted at 1, 2, and 3 mM of mPEG, respectively. Since bRBCs reacted at the same hematocrit, this corresponds to more mPEG molecules being conjugated per cell at higher initial mPEG concentrations. Assuming bRBCs have a diameter of about 6 mm, these results correspond to 1.4, 2.1, 3.2, and 4.3 ' 108 mPEG molecules/ bRBC for initial concentrations of 1, 2, 3, and 4 mM mPEG, respectively. Oxygen binding measurements show a slight drop in P50 with increased bRBC surface coverage with mPEG. The oxygen dissociation curves were fitted to the Hill Equation to regress the P50 and cooperativity (Hill coefficient), shown in Figure 4A and B (Voet and Voet, 1995). The mPEG layer conjugated to the surface of bRBCs slightly increases the overall O2 affinity as shown by the left shifted curves

Figure 2.

Percentage of mPEG-bRBCs remaining in the top, PEG-rich phase after partitioning, normalized to the corresponding unreacted bRBCs (PBS control) in the PEG-rich phase. bRBCs initially at 12% hematocrit were reacted with the specified concentrations of SPA-mPEG. After reaction, bRBCs were reconstituted in a PEG–Dx solution. bRBCs modified with mPEG should remain in the top PEG-rich phase. Results, in solid bars, show the percentage of mPEG-bRBCs remaining in the PEG-rich phase/ percentage of bRBCs that reacted with PBS in the PEG phase after the partitioning experiment. Overlapping, in dotted bars, is the additional control of 0 mM mPEG corresponding to that sample. Error bars represent the standard error in cell counts from three samples propagated through to the normalized percentage. [Color figure can be seen in the online version of this article, available at www.interscience. wiley.com.]

1204

Figure 3. Mass of SPA-mPEG that reacted with bRBCs (solid bars) and the percentage of reacted mPEG (dotted bars) derived from HPLC measurements. Left axis: mass of mPEG reacted; right axis: percent of initial mPEG that reacted with bRBCs. Error bars represent the standard error in HPLC peak area averaged from three experiments propagated through to the percentage of reacted mPEG. [Color figure can be seen in the online version of this article, available at www.interscience.wiley.com.]

Biotechnology and Bioengineering, Vol. 96, No. 6, April 15, 2007 DOI 10.1002/bit

Figure 5. High shear rate viscosities of mPEG-bRBCs and controls normalized to 3 cp for that day’s 40% hematocrit control. bRBCs were reacted with 4–0 mM mPEG or unreacted PBS as indicated and resuspended to 40% hematocrit (*1.5% hematocrit). The dotted line is at 3 cp, which is the average high shear rates viscosity for human blood at 40% hematocrit with or without fibrinogen. Error bars represent the standard error of the average fitted high shear rate viscosity of three experiments.

Figure 4. A: P50 and B: cooperativity (Hill coefficient) of mPEG-bRBCs as determined from curve fitting experimental oxygen dissociation curves to the Hill equation. The dotted line in (A) shows the average P50 for human RBCs !27 mmHg. All samples were mixed with equal volumes of bRBCs at 12% hematocrit and the corresponding SPA-mPEG concentration. Error bars represent the standard error of the average of three experiments.

corresponding to a lower P50. The average P50 for bRBCs is 27.92 mmHg, which drops to 26.04 mmHg for bRBCs reacted with 2 mM mPEG, and finally to 24.38 mmHg for a reaction with 4 mM mPEG. The cooperativity of Hb’s O2 binding (Fig. 4B) is maintained with PEGylation of the bRBC membrane yielding Hill coefficients >2. The viscosity results in Figure 5 show similar high shear rate viscosities for modified bRBCs and the control group of !3 cp. All samples were resuspended to a hematocrit of 40 * 1.5%. Some of the variability in the measured viscosity is due to the slightly varying hematocrits. bRBCs reacted at the highest concentration, 4 mM mPEG, exhibited a measured viscosity of 2.8 cp. The remaining samples reacted at 0, 1, 2, 3 mM mPEG, and PBS exhibited apparent viscosities of 2.9, 2.6, 2.7, 2.3, 3.1 cp, respectively. Although the rheometer software fitted the experimentally measured shear stress versus shear rate to the Casson equation for blood, the yield stress values were all very low, approximately zero, and out of the measurement range of the instrument and will not be included.

After reacting bRBCs with mPEG, mPEG-bRBCs were washed six times in PBS. After each centrifugation step, the supernatant was collected and tested for free Hb to determine if the conjugation reaction with mPEG or the subsequent PBS washes elicited cell lysis (data not shown). On average, reactions with up to 4 mM mPEG had less than 1% of acellular Hb in the supernatant for each wash. Therefore, the reaction alone does not immediately compromise the bRBC’s membrane integrity. Capillary wetting experiments showed some unexpected results. The meniscus velocity remained the same for all mPEG-bRBCs, as shown in Figure 6. Modified bRBCs displayed similar meniscus packing behavior compared to normal bRBCs. However, mPEG-bRBCs would pack into periodic plugs. Figure 7 shows sequential screen shots of the

Figure 6. Meniscus velocity versus the wetted capillary length of several mPEG modifications of bRBCs. [Color figure can be seen in the online version of this article, available at www.interscience.wiley.com.]

Gundersen and Palmer: PEGylated Bovine Erythrocytes Biotechnology and Bioengineering. DOI 10.1002/bit

1205

Figure 7. Capillary wetting snapshots. A: Unmodified bRBCs in PBS pack at the meniscus (left pane 1) and will migrate to the axial centerline further downstream in panes 2 and 3. B: Modified cells (at 4 mM mPEG) will still pack at the meniscus (left pane 1) and migrate to the axial centerline (pane 2). However, the amount of cells at the axial centerline will slowly decline until the tube is empty (pane 3) and then be followed by a second slug of meniscus-like packed cells (pane 4). [Color figure can be seen in the online version of this article, available at www.interscience.wiley.com.]

video of (A) unreacted bRBCs and (B) 4 mM mPEG modified mPEG-bRBCs as they travel down the length of the glass capillary. The left hand frame (Frame 1) is the first snapshot of the meniscus. Frames 2–4 show subsequent snapshots in the order they were taken. As can be seen in Figure 7A, normal bRBCs will pack at the meniscus, Frame 1, and further upstream are migrating to the capillary centerline, Frames 2 and 3. PEGylated bRBCs also pack at the meniscus, as seen in Figure 7B Frame 1, and upstream of

the meniscus will migrate to the center, Frame 2. However, no to very little bRBCs are present further upstream, Frame 3, until a second plug of cells appears in Frame 4. This second plug of cells follows a similar trend of packing at the beginning of the plug (as opposed to the meniscus) and migrating to the centerline after the plug. For oxygen transport simulations, examination of the axial centerline (r ¼ 0) pO2 profile for all mPEG-bRBCs (Fig. 8A) shows that the oxygen concentration decreases

Figure 8. Simulated oxygen concentration profiles at the axial centerline (r ¼ 0) at (A) pO2,in ¼ 95 mmHg and (B) pO2,in ¼ 20 mmHg. Simulated oxygen concentration profiles at the tissue radius (r ¼ RT) at (C) pO2,in ¼ 95 mmHg and (D) pO2,in ¼ 20 mmHg. The solid lines represent simulations at low glucose levels, low Vmax, and the dotted lines represent simulations at high glucose levels, high Vmax. [Color figure can be seen in the online version of this article, available at www.interscience.wiley.com.]

1206

Biotechnology and Bioengineering, Vol. 96, No. 6, April 15, 2007 DOI 10.1002/bit

down the capillary centerline as the pancreatic tissue and endothelial cells consume oxygen. The pO2 profile at the tissue edge (Fig. 8B) also exhibits the same decreasing trend for various pO2,in and Vmax values, but shifted to lower pO2s due to its increased distance from the oxygen source in the capillary. In the extreme case of a high Vmax value and hypoxia (pO2,in ¼ 20 mmHg), the tissue pO2 has a lower limit of 15.5 mmHg. The percentage of total incoming oxygen transferred to the tissues was determined from an overall oxygen mass balance on the capillary space. Figure 9A displays the percent decrease of total dissolved oxygen in the capillary for

modified bRBCs at various inlet pO2s at low Vmax. Since the oxygen consumption rate is dependent on the pO2 via Michaelis–Menten kinetics, it is expected that more O2 is consumed at higher pO2s. However, since more oxygen is available at higher inlet pO2s, the percentage of O2 consumed is lower at higher initial pO2s. Equivalently, there is a larger percent decrease in O2 at low inlet pO2s, even though the tissues consume less oxygen. Furthermore, as would be expected, the larger percentage of oxygen lost to tissues at a high Vmax value at a given entering pO2 is due to the higher tissue oxygen consumption rate as the cells metabolize more oxygen (Fig. 9B). Although, the P50 generally decreased with increasing initial mPEG concentration, the Hill coefficient did not follow this trend. This would correspond to varying Adair constants, causing a lack of an apparent trend between mPEG concentration and the percentage of oxygen transported to the tissue space.

Discussion

Figure 9. Percentage of total incoming oxygen lost from the capillary to the tissues at (A) low glucose levels, low Vmax, and (B) high glucose levels, high Vmax. C: Representative %O2 transfer to tissue values for both high Vmax (dotted lines) and low Vmax (solid lines). [Color figure can be seen in the online version of this article, available at www.interscience.wiley.com.]

As the primary hRBC donor population shrinks in size, a viable substitute for hRBCs is in high demand. Although modified Hbs show good potential as oxygen carriers, they still have many problems to overcome. Also, more importantly the production of modified Hbs and mPEGhRBCs are highly dependent on the dwindling supply of human blood. Hence, modified bRBCs constitute a possible solution to ease the demand on human blood donations and the cost of donated blood. Although the amount of mPEG required to immunologically silent hRBCs should be approximately the same as for bRBCs, the main cost savings would derive from the blood source (!$100 less for bRBCs). The evaluation and control of select physical properties of mPEG-bRBCs is essential for the successful application of this novel xenotransfusion technology. For example, if the mPEG-bRBCs cannot support normal oxygen delivery or are too rigid, they could potentially clog the capillaries or severely impair oxygen transport to surrounding tissues. Hence, this study focused on elucidating the physical properties of mPEG-bRBCs, that is, oxygen binding, rheology, and deformability. Since the primary reason for mPEG modification is to effectively cover antigenic sites on the bRBC membrane, it is important to determine how well bRBCs are surface conjugated with mPEG. Both partitioning and HPLC analysis (Figs. 2 and 3) showed that bRBCs were successfully modified with SPA-mPEG in a dose-dependent manner. Although the percentage of initial mPEG reacting with bRBCs levels off, the mass of SPA-mPEG reacting with bRBCs increases with increasing mPEG concentration up to 4 mM. The results in Figures 2 and 3 agree, showing that more mPEG molecules are conjugated to the bRBC surface when reacted at higher initial mPEG concentrations. These results are in agreement with Bradley et al. (2001, 2002), who showed that surface coverage of mPEGs on hRBCs increased with increasing initial mPEG concentrations. The surface coverage of mPEG on the bRBC membrane should increase

Gundersen and Palmer: PEGylated Bovine Erythrocytes Biotechnology and Bioengineering. DOI 10.1002/bit

1207

until it reaches a maximum saturation value such that increasing the initial mPEG concentration will not affect the amount of surface conjugated mPEG molecules. This will be influenced by the number of possible cross-linking sites available on the bRBC surface and steric exclusion between neighboring mPEG molecules. Since the reactive ester group targets positively charged amine groups, it should not react with itself at high concentrations (Nektar, 2006). Even though the average number of mPEG molecules per cell was quite high, some bRBCs remained unmodified or slightly modified due to possible poor mixing or were trapped near the glass wall of the reaction vessel during the reaction. In this study, partitioning was used only as a qualitative test to determine the relative degree of mPEG modification. By changing the molecular weight and concentrations of the PEG and Dx partitioning solutions, this technique can be ‘‘fine-tuned’’ so all unmodified and lightly modified cells will settle at the PEG–Dx interface, while successfully modified bRBCs will remain in the PEG-rich layer. This will be especially important for potential clinical application as the mPEG-bRBC solution will have to be free of unmodified and poorly modified bRBCs to not elicit an immune response. There was minimal acellular Hb present in the reaction mixture after the 1-h reaction period and in the supernatant during subsequent PBS washes. Less than 5% of cellular Hb was found in the supernatant after the five washes. This small amount could be enough to initiate an immune response if left in the solution. However, with washes and resuspension in the appropriate media for storage, such as citrate phosphate dextrose for hRBCs, this should be able to be removed before transfusion. Modified bRBC solutions must be able to be resuspended to a hematocrit exhibiting physiological and rheological properties, while maintaining efficient oxygen delivery. In normal human blood flow, the shear rate is !100 s&1 where blood behaves like a Newtonian fluid with a viscosity of 3 cp (Fournier, 1999). Our rheometry studies showed that mPEG-bRBCs at 40 * 1.5% hematocrit exhibited a high shear rate viscosity of !3cp. The appearance of a yield stress and non-Newtonian behavior of normal whole blood at low shear rates is due to RBC aggregate formation and other plasma proteins, particularly fibrinogen (Armstrong et al., 1997; Fournier, 1999). Thus, the low yield stress values observed in our experiments are expected for RBCs resuspended in PBS due to the absence of fibrinogen and other plasma contents (Fournier, 1999). These results are in agreement with the general behavior of blood solutions and with experiments by Armstrong et al. (1997) who showed that PEGylation of hRBCs resulted in lower viscosities at low shear rates (lower yield stress values) due to conjugated mPEG hindering RBC aggregate formation. There is a slight decrease in viscosity for bRBCs reacted at 2 and 3 mM mPEG. However, this is most likely due to the slight variances in the hematocrit than a function of PEGylation. It is also important for mPEG-bRBCs to be flexible, as RBCs have to deform to fit through smaller sized capillaries.

1208

Even though bRBCs are slightly smaller than hRBCs, human capillaries can possess a diameter less than bRBCs as well. In capillary flow, shear-induced migration will cause deformable particles to migrate to the axial centerline, while perfectly rigid particles will remain uniformly suspended (King and Leighton, 2001). Since the deformable particles at the centerline travel faster than the average velocity, they will begin to pack at the meniscus of the flow. This will form a plug at the meniscus and will subsequently cause the meniscus to slow down. Thus, more deformable particles will lead to slower capillary penetration. The capillary wetting experiments showed mPEG-bRBCs to be deformable, even though they exhibited slightly different entrance phenomena in a glass capillary compared to normal bRBCs. This should not affect in vivo properties, since this particular glass capillary entrance geometry is not observed in vivo. Since capillary flow was the only concern, the most important outcome of this experiment is that the mPEGbRBCs still packed at the meniscus and migrated to the axial centerline. In accordance with results from Zhou and Chang (2005), mPEG-bRBCs can be considered deformable particles. The reason for the different entrance behaviors between mPEG-bRBCs and normal bRBCs could be due to mPEG molecules preventing cell–cell interactions or differences in surface charge density. However, future zeta potential measurements will be needed to confirm this hypothesis. Naturally it is essential for mPEG-bRBCs to perform RBC’s most important function of delivering oxygen to the tissues. Hence, the oxygen binding properties of mPEGbRBCs should be similar to hRBCs. Although the P50 dropped slightly, mPEG-bRBCs maintained a high P50 for bRBCs which reacted at lower mPEG concentrations close to the hRBC physiological value of 27 mmHg. Modified bRBCs maintained cooperative oxygen binding with cooperativities >2 as would be expected, since mPEG modification of the cell’s surface should not have an affect on the cooperativity of Hb. Scott and Chen (2004) reported similar drops in P50 with mPEG modifications up to 5 mM cyanuric chlorideactivated mPEG on human RBCs with no change in cooperativity. The reason for the slight drop in P50 is still under investigation and poorly understood. However, it is assumed this change is not due to the mPEG diffusing inside the cell and reacting with Hb directly. Methoxy PEG should not be able to enter the cell due to its hydrophilic nature. Thus, it is highly unlikely that SPA-mPEG reacts directly with Hb to influence Hb-O2 binding. Next, the experimentally measured oxygen binding parameters of mPEG-bRBCs were used as input variables in oxygen transport simulations to confirm that the slight decrease in P50 of mPEG-bRBCs can still result in the delivery of sufficient oxygen to tissues similarly to hRBCs even under extreme oxygenation. The most important outcome from these simulations shows mPEG-bRBCs ability to maintain the tissue pO2. Since normal, nontumorgenic tissues require a minimum pO2 of 7–10 mmHg to avoid hypoxia, the mPEG-bRBCs must be able to

Biotechnology and Bioengineering, Vol. 96, No. 6, April 15, 2007 DOI 10.1002/bit

transport enough oxygen to sustain tissue pO2s above this limit (Harrison and Blackwell, 2004; Vaupel, 2004). As seen from the oxygen transport simulations, the tissue pO2 level remained above 10 mmHg for all surface coverages of mPEG on the bRBC surface even under extreme oxygenation scenarios characterized by low entering pO2s and high oxygen consumption rates in the tissues. These simulations demonstrate that mPEG-bRBCs should be able to transport the necessary amount of oxygen to tissues under normal physiological conditions. Figure 8 does not appear to agree with the convective flux boundary condition imposed at z ¼ L. The O2 profile is expected to decrease to a slope of zero at the end of the capillary due to the exiting boundary condition. This was not observed since there are pancreatic and endothelial cells consuming oxygen until z ¼ L. Therefore, the pO2 will continue to decrease until either there is no more O2 consuming tissue, which is located at z ¼ L in this simulation, or the oxygen is completely consumed in the capillary and pO2 ¼ 0. Should bRBCs prove to be a feasible hRBC substitute, they could help reduce the demand on human blood transfusions. Also, mPEG-bRBCs could be cross-linked with gluteraldehyde to further alter the P50 for oxygenation in bio-artificial devices (Gordon et al., 2005; Sullivan et al., 2006). Before application of this novel blood substitute, the most important concern will be the safety of the bovine blood supply. In addition to Quad Five’s quarantine and screening procedures, the presence of TSE in cow’s blood can be excluded by testing two generations of founders and quarantining the offspring (Johnstone et al., 2004). Furthermore, it will be necessary to obtain a continuous fresh bovine blood source for further tests, especially for clinical trials and Food and Drug Administration approval, however, this is beyond the scope of this current study.

Conclusions The results of this study show that bRBCs can be effectively surface conjugated with mPEG, just as hRBCs can, without compromising the cell’s structure and oxygen binding functions. There is significantly more partitioning of mPEGbRBCs in the PEG-rich phase compared to control bRBCs. HPLC analysis also shows that bRBCs are being conjugated with SPA-mPEG due to the significant decrease in unreacted mPEG molecules. Also, mPEG-bRBCs demonstrated similar viscosities compared to unmodified bRBCs and hRBCs. Modified bRBCs retain their ability to bind and release oxygen, as shown by only moderate changes in oxygen binding properties, such as the P50 and cooperativity coefficient. This was further confirmed in oxygen transport simulations, which demonstrated that PEGylation does not affect the ability of mPEG-bRBCs to deliver oxygen to tissues under a variety of oxygenation conditions. Even under the most extreme circumstance of pO2,in ¼ 20 mmHg and high Vmax value, tissues still received enough oxygen to remain

viable (>10 mmHg). Taken together, these results demonstrate that mPEG-bRBCs at any level of PEGylation can function as a suitable human RBC substitute from a physical standpoint. Of course, future immunological studies are needed to prove the feasibility of using mPEG-bRBCs as a hRBC substitute. This research was supported by United States Public Health Service grants HL078840 and DK070862 to A.F.P.

References Armstrong JK, Meiselman HJ, Fisher TC. 1997. Covalent binding of poly(ethylene glycol) (PEG) to the surface of red blood cells inhibits aggregation and reduces low shear blood viscosity. Am J Hematol 56:26–28. Biopure. 2004. Biopure annual report 2004. http://www.biopure.com/corporate/reports/ar04/page5.htm (accessed on May 26, 2005). Birkenmeier G, Walter H, Widen KE. 1994. Factors in the affinity extraction of red blood cells using poly(ethylene glycol)-metal chelate. Methods Enzymol 228:368–390. Blackall DP, Armstrong JK, Meiselman HJ, Fisher TC. 2001. Polyethylene glycol-coated red blood cells fail to bind glycophorin A-specific antibodies and are impervious to invasion by the Plasmodium falciparum malaria parasite. Blood 97:551–556. Bonsen P, Laver MB, Morris KC, Alza Corporation, assignee. 1977. Novel polymerized, cross-linked, stromal-free hemoglobin. United States patent 4,001,200. Bradley AJ, Test ST, Murad KL, Mitsuyoshi J, Scott MD. 2001. Interactions of IgMABO antibodies and complement with methoxy-PEG-modified human RBCs. Transfusion 41:1225–1233. Bradley AJ, Murad KL, Regan KL, Scott MD. 2002. Biophysical consequences of link chemistry and polymer size on stealth erythrocytes: Size does matter. Biochim Biophys Acta 1561:147–158. Davey RJ. 2004. Overcoming blood shortages: Red blood cell and platelet substitutes and membrane modifications. Curr Hematol Rep 3(2):92– 96. Eike JH, Palmer AF. 2004. Effect of NaBH4 concentration and reaction time on physical properties of glutaraldehyde-polymerized hemoglobin. Biotechnol Prog 20(3):946–952. Fournier RL. 1999. Basic transport phenomena in biomedical engineering. Philadelphia: Taylor & Francis. Garratty G, Telen M, Petz LD. 2002. Red cell antigens as functional molecules and obstacles to transfusion. Hematology 2002:445–462. Gavasane MR, Gaikar VG. 2003. Aqueous two-phase affinity partitioning of penicillin acylase from E. coli in presence of PEG-derivatives. Enzyme Microb Technol 32:665–675. Gordon JE, Dare MR, Palmer AF. 2005. Engineering select physical properties of cross-linked red blood cells and a simple a priori estimation of their efficacy as an oxygen delivery vehicle within the context of a hepatic hollow fiber bioreactor. Biotechnol Prog 21(6):1700–1707. Harris JM, Kozlowski A, Shearwater Polymers, Inc., assignee. 1995. Poly(ethylene glycol) and related polymers monosubtituted with propionic or butanoic acids and functional derivatives thereof for Biotechnical Applications. Patent 5672662 (9/30/1997). Harrison L, Blackwell K. 2004. Hypoxia and anemia: Factors in decreased sensitivity to radiation therapy and chemotherapy? Oncologist 9:31–40. ¨ sterberg E, Tiberg F, Harris JM. 1993. Holmberg K, Bergstro¨m K, Brink C, O Effects on protein adsorption, bacterial adhesion and contact-angle of grafting PEG chains to polystyrene. J Adhes Sci Technol 7(6):503–517. Jo S, Park K. 2000. Surface modification using silanated poly(ethylene glycol)s. Biomaterials 21:605–616. Johnstone JE, MacLaren LA, Doucet J, McAlister VC. 2004. In vitro studies regarding the feasibility of bovine erythrocyte xenotransfusion. Xenotransplantation 11(1):11–17.

Gundersen and Palmer: PEGylated Bovine Erythrocytes Biotechnology and Bioengineering. DOI 10.1002/bit

1209

King MR, Leighton DT. 2001. Measurement of shear-induced dispersion in a dilute emulsion. Phys Fluids 13(2):397–406. Lutwyche P, Norris-Jones R, Brooks DE. 1995. Aqueous two-phase polymer systems as tools for the study of recombinant surface-expressed Escherichia Coli hemagglutinin. Appl Environ Microbiol 61(9):3251–3255. McCarthy MR, Vandegriff KD, Winslow RM. 2001. The role of facilitated diffusion in oxygen transport by cell-free hemoglobins: implications for the design of hemoglobin-based oxygen carriers. Biophysical Chemistry 92:103–117. Micks W. 2005. Quad Five’s collection and virus screening. Ryegate, MT: Fax transmittal received by S. Gundersen on 5/16/2005. Murad KL, Mahany KL, Brugnara C, Kuypers FA, Eaton JW, Scott MD. 1999. Structural and functional consequeces of antigenic modulation of red blood cells with methoxypoly(ethylene glycol). Blood 93(6):2121– 2127. Nektar Therapeutics. 2006. Polyethylene glycol and derivatives for Advanced PEGylation Catalog 2005–2006. http://www.nektar.com/ (accessed on 06/14/2006). Otchet A. 2001. Animal transplants: Safe or sorry? UNESCO Courier. http:// www.unesco.org/courier/2000_3/uk/ethique/txt1.htm (accessed on 6/7/2004). Phillips WT, Klipper RW, Awasthi VD, Rudolph AS, Cliff R, Kwasiborski V, Goins BA. 1999. Polyethylene glycol-modified liposome-encapsulated hemoglobin: A long circulating red cell substitute. J Pharmacol Exp Ther 288(2):665–670. Quad 5. Quad 5 Animal diagnostic blood products. http://www.quadfive.com/ (accessed on May 15. 2002). Circular of Information for the Use of Human Blood and Blood Components. South Bend Medical Foundation, Inc., editor: American Association of Blood Banks, America’s Blood Centers, & American Red Cross. p 1–39.

1210

Riess JG. 2001. Oxygen carriers (‘‘blood substitutes’’)—raison d’etre, chemistry, and some physiology. Chem Rev 101(9):2797–2919. Scott MD, Murad KL, Koumpouras F, Talbot M, Eaton JW. 1997a. Chemical camouflage of antigenic determinants: Stealth erythrocytes. Proc Natl Acad Sci 94:7566–7571. Scott MD, Kucik DF, Goodnough LT, Monk TG. 1997b. Blood substitutes: Evolution and future applications. Clin Chem 43(9):1724– 1731. Scott MD, Bradley AJ, Murad KL. 2000. Camouflaged blood cells: Lowtechnology bioengineering for transfusion medicine. Transfus Med Rev 14(1):53–63. Scott MD, Chen AM. 2004. Beyond the red cell: pegylation of other blood cells and tissues. Transfusion Clinique et Biologique 11:40–46. Sullivan JP, Gordon JE, Palmer AF. 2006. Simulation of oxygen carrier mediated oxygen transport to C3A hepatoma cells housed within a hollow fiber bioreactor. Biotechnol Bioeng 93(2):306–317. Vadapalli A, Goldman D, Popel AS. 2002. Calculation of oxygen transport by red blood cells and hemoglobin solutions in capillaries. Artif Cells Blood Substit Immobil Biotechnol 30(3):157–188. Vaupel P. 2004. The role of hypoxia-induced factors in tumor progression. Oncologist 9(Suppl 5):10–17. Voet D, Voet JG. 1995. Biochemistry. New York: John Wiley & Sons, Inc. Williams RD. 1996. Organ transplants from animals: Examining the possibilities. FDA Consumer Magazine 30(5). http://www.fda.gov/ fdac/features/596_xeno.html (accessed on 6/7/2004). Zhou R, Chang H-C. 2005. Wetting penetration failure of blood suspensions. J Colloid Interface Sci 287:647–656. Zhou RH, Gordon J, Palmer AF, Chang HC. 2006. Role of erythrocyte deformability during capillary wetting. Biotechnol Bioeng 93(2):201– 211.

Biotechnology and Bioengineering, Vol. 96, No. 6, April 15, 2007 DOI 10.1002/bit