Structural and Functional Consequences of Antigenic Modulation of Red Blood Cells With Methoxypoly(Ethylene Glycol)

After informed consent was obtained, whole blood was collected in heparin, acid citrate dextrose (ACD), or EDTA from laboratory volunteers and immediately processed. Volunteers were selected to insure adequate representation, and no individuals were excluded based on race or gender. Statistical analyses were performed by the Student’s t-test or analysis of variance (ANOVA).8 Commercial anti-A and anti-B antisera were obtained from Carolina Biological Supply (Burlington, NC). All other biochemicals, unless otherwise noted, were obtained from Sigma Chemical Co (St Louis, MO).

Unless specifically stated, derivatization of erythrocytes was performed as previously described.4,5 Briefly, washed RBC were suspended to a hematocrit of approximately 12% (≈1.5 × 10⁹ cells/mL) in isotonic alkaline phosphate buffer (phosphate-buffered saline [PBS]: 50 mmol/L K₂HPO₄ and 105 mmol/L NaCl) and incubated with varying concentrations of cyanuric chloride-activated mPEG (MW 5 kD). As indicated, the effects of pH (9.2 and 8.0), temperature (4°C and 25°C), and time (30 minutes and 60 minutes) on the efficacy of derivatization were assessed. However, unless otherwise noted, all functional and structural analyses used RBC that were reacted with mPEG at pH 9.2, at 4°C for 30 minutes. Effects of storage on the efficacy of RBC derivatization were assessed by storing blood at 4°C in ACD vacutainer tubes for up to 45 days.

Antigenic camouflage was assessed by attenuation of anti-A and anti-B human antisera-mediated RBC aggregation using a platelet aggregometer (Chrono-Log, Havertown, PA), as previously described.4,5Briefly, 450 μL of an RBC suspension (6% hematocrit in isotonic saline) was placed in an aggregometer cuvette at 37°C, with stirring, and 20 μL of anti-A and/or anti-B typing serum (or autologous/heterologous serum) was added. RBC aggregation was then observed over time. Our previous studies show a direct relationship between effects on aggregation and measured binding of appropriate antisera to intact RBC.4-7 

To test the integrity of mPEG-derivatized RBC, changes in gross morphology, membrane stability, deformability, ion exchange, and RBC membrane proteins were examined. RBC indices of control and mPEG-modified RBC were determined using a Swelab 920EO hematology analyzer (DiaPharma, Franklin, OH). The morphology of both control and mPEG-derivatized cells was assessed by both light and scanning electron microscopy (SEM), as previously described.4 Spontaneous lysis was measured by determining total hemoglobin concentration spectrophotometrically (540 nm) for both the total cell suspension and the cell supernatant using Drabkin’s reagent.9 

Gross changes in the membrane protein pattern were analyzed by one-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of membrane ghosts, as previously described.4,10-12 Functional integrity of the RBC membrane, as well as functioning of membrane pumps, was assessed by ion homeostasis. Cation (Na⁺, K⁺) and anion (SO₄²⁻) fluxes were examined by flame photometry in accordance with standard procedures.13,14 

Cellular deformability was determined using a Technicon ektacytometer (Bayer Diagnostics, Tarrytown, NY). Ektacytometry is a sensitive method for detecting population changes in deformability based on changes in cell geometry, surface area, cytoplasmic viscosity, and cellular hydration.15-17 In brief, control and mPEG-derivatized RBC were suspended in 4% polyvinylpyrrolidone. A constant shear stress of 125 dyne/cm² was applied to the cells, and the osmolality of the suspending medium was gradually changed over a range of 100 to 500 mOsm/kg.

In vivo studies provide the ultimate indication of the normality (ie, viable structure and function) of PEG-derivatized RBC. Consequently, mice (Balb/c) were transfused as previously described.4,18The concentrations of mPEG used to treat murine RBC in these experiments ranged from 0 to 5 mmol/L. Murine RBC were labeled using a fluorescent, membrane-anchoring marker, PKH-26 (Sigma Chemical Co). Equal numbers of control and mPEG-derivatized RBC (40% hematocrit) were injected intraperitoneally (IP; a common mode of transfusion in veterinary medicine) into recipient Balb/c mice. Blood samples from the recipient mice were observed until the labeled RBC were cleared from circulation (approximately 40 to 50 days for allotransfusions). Survival of fluorescently labeled mPEG-treated and control RBC was monitored by analyzing the percentage of fluorescently labeled RBC by flow cytometry.18 In some studies, mice were “hypertransfused” using the above protocol to determine the effects of a high hematocrit of mPEG-modified RBC. Because hypertransfusion results in suppression of endogenous RBC production, this procedure allowed for an almost complete blood exchange. Mice were hypertransfused by injecting (IP) RBC at 3- to 5-day intervals for 62 days. The percentage of PKH-26–labeled RBC in mice receiving either control or mPEG-modified (0.4 mmol/L) RBC was determined by flow cytometry at 24 hours posttransfusion.

As we have previously shown, mPEG-derivatization decreased antisera-induced aggregation in a dose-dependent manner.4-7These previous studies derivatized the RBC at a pH of 9.2 for 30 minutes at 4°C. This high pH was used because it was previously reported to be the optimal pH of the chemical reaction between cyanuric chloride–activated mPEG.19 To determine whether human RBC could be efficiently modified at a more physiological pH and under less restrictive conditions, we examined the effects of pH, derivatization time, and temperature on antisera-mediated aggregation. As shown in Fig1A, the derivatization process is quite malleable. Indeed, derivatization at pH 8.0 for 60 minutes was found to consistently result in a more efficient antigen masking, as assessed by antisera-mediated aggregation. In addition, as shown in Fig 1B, mPEG-modified RBC effectively inhibit antisera-mediated aggregation of unmodified cells in a dose-dependent manner.

Stored RBC were found to be readily derivatized even after storage at 4°C for 0, 7, and 45 days (Fig 2). Importantly, the derivatization process had no significant effects on RBC lysis (<0.5% lysis even in the presence of 10 mmol/L activated mPEG) or hemoglobin oxidation (>98% oxyhemoglobin after derivatization). Indeed, morphologically, the derivatized RBC appeared normal by both light microscopy and SEM (Fig3). Analysis of RBC indices also showed no overall alterations arising from mPEG-modification (Table1). Although the mPEG-modified cells showed a tendency towards slightly (but not significantly) decreased intracellular K⁺ concentrations relative to the pH control, only at the highest levels of derivatization (>5 mmol/L mPEG) did one see a significant increase in intracellular Na⁺ (Table 1).

The above findings are consistent with our earlier reports indicating normal lysis rates and osmotic fragility after in vitro incubations of up to 48 hours.4,5 Similarly, oxygen binding was unaffected by mPEG-derivatization of the membrane. In agreement with recently published findings from our Korean colleagues,20 the P₅₀ of the mPEG-modified RBC was unchanged from that of the control cells previously subjected to pH 9.2 conditions (P₅₀ = 22.3 and 23.0, respectively).

Closely related to cell shape and membrane stability is cation and anion homeostasis. Ion homeostasis is governed in large part by integral membrane proteins that are subject to mPEG-derivatization. As we previously showed, the anion transporter Band 3 is a major site of mPEG-binding.4 Consequently, to determine if derivatization adversely affected Band 3 function,³⁵SO₄⁻ influx in the absence and presence of the Band 3 inhibitor di-isothiocyano-disulfonyl stilbene (DIDS) was examined. As shown in Fig 4A, RBC derivatization with up to 5 mmol/L mPEG did not alter³⁵SO₄⁻ influx. Furthermore, Band 3 dependent anion transport—even in the pegylated cells—was readily inhibited by DIDS.

Similarly, active cation transport was unaffected by mPEG-derivatization, as shown by the ouabain-sensitive Na⁺efflux, which measures the maximal functional capacity of the Na⁺-K⁺ adenosine triphosphatase (ATPase) (Fig4B). There were also no changes induced by mPEG-derivatization on the maximal activity of the Na⁺-K⁺-Cl⁻ cotransport system (expressed as bumetanide-sensitive Na⁺ or K⁺effluxes, data not shown). No effects of mPEG-derivatization could be shown on the passive permeability of the erythrocyte. In particular, the efflux of K⁺ from fresh cells in the presence of 0.1 mmol/L ouabain and 0.01 mmol/L bumetanide was unaffected by mPEG-derivatization (data not shown). The finding that transport inhibitors like DIDS, ouabain, and bumetanide could readily inhibit their targets further attests to the functional integrity of integral membrane proteins of pegylated RBCs. Furthermore, the results obtained with these inhibitors suggest that other small molecules (eg, oxygen, glucose) can access and interact with the membrane of pegylated RBC, whereas large molecules (eg, IgG) and cells cannot.

Functionally related to ion homeostasis (via cell size and intracellular viscosity) and membrane integrity is the ability of RBC to deform. The ability of the erythrocyte to undergo deformation is crucial to not only its physiological function of delivering oxygen, but also to its survival within the circulatory system. Indeed, genetic conditions (eg, sickle hemoglobin) or pharmacological agents (eg, membrane oxidants) that affect cellular deformability have been shown to have profound effects on erythrocyte function and survival.16,17,21 Hence, because the membrane of the mPEG-treated cell undergoes significant surface modification, we examined what, if any, effects variable levels of mPEG modification would have on RBC deformability using ektacytometry. As shown in Fig5, only at very high levels of derivatization (≥5 mmol/L mPEG) did the deformability profile of the modified cells show any significant differences from the control RBC. Of particular interest in analyzing the ektacytometric curve is the maximum deformability index (DI_max) at isotonicity (∼290 mOsm). As shown, all samples with the exception of the 5 mmol/L-modified sample fall within the normal range (0.45 ± 0.08). However, even in the 5 mmol/L-treated samples, the deformability is still significantly better than that seen in sickle, β thalassemic, or slightly oxidant stressed glucose 6 phosphate dehydrogenase (G6PD)-deficient RBC.16,21 All other parameters (0_min = Minimum DI and is where 50% hemolysis is observed in a classical osmotic fragility test; 0_hyp = 1/2DI_max and arises from the hypertonic osmolality and resultant cell shrinkage) are within the normal range.

In aggregate, all in vitro studies suggested that the mPEG-modified RBC show normal structure and function. However, the ultimate test as to the normality of modified RBC is in vivo survival. To examine this question, we used a murine (Balb/c) model using the transfusion of syngeneic control and mPEG-modified cells. With this model, we examined the effects of the degree of mPEG-derivatization and whether mPEG-modified cells could efficaciously replace normal, unmodified RBC. Despite the inherent fragility of mouse erythrocytes, a significant degree of derivatization was possible; though as shown in Fig6, it was possible to overmodify the RBC, such that a more rapid clearance of the transfused RBC occurred. How this finding relates to human RBC is somewhat unclear at this point. However, it is important to note that significant in vitro lysis occurred in the murine cells at mPEG concentrations ≥0.6 mmol/L. In contrast, under the same conditions, significant lysis of human RBC only occurred at mPEG concentrations ≥12 mmol/L. Furthermore, as we previously reported, repeated transfusion of mice with mPEG-modified (0.4 mmol/L) RBC did not sensitize the animals to the modified RBC, and these cells showed normal survival even after five transfusions each separated by a minimum of 50 days.

Additional studies further showed that mPEG-derivatized murine RBC, at levels that yielded normal in vivo RBC survival (shown is 0.4 mmol/L mPEG), were readily and safely tolerated in hypertransfused mice (Fig 6insert). In this study, the total time from the initial transfusion to the final clearance of the transfused RBC was approximately 140 days, with a maximum of approximately 80% of the mouse RBC volume consisting of mPEG or labeled control RBC. During this time, the PEG-transfused mice received a total of 6.7 g of mPEG per kg body weight and showed no evident toxicity relative to the mice hypertransfused with control RBC. However, both populations of mice showed secondary iron overload to the repeated transfusions. On necropsy, gross examination of the organs (with one exception discussed below) showed no significant differences in appearance between the three groups (nontransfused, control-transfused, mPEG-RBC transfused). Furthermore, no differences in body weight or the wet weight of the liver, kidney, heart, or brain were noted between the mice hypertransfused with control cells (Table2). Interestingly, however, the mice hypertransfused with control-RBC showed a significant increase in spleen size, whereas the mPEG-transfused animals showed no increase in the spleen relative to the nontransfused animals.

Despite medical advances in transfusion medicine, transfusion reactions are still a clinically significant problem. Although the NIH estimates that 1:100,000 transfusions result in a fatal rejection reaction,22 it is estimated that 1 of every 4,000 units transfused results in a nonfatal transfusion reaction (eg, accelerated clearance of transfused RBC).23 The majority of these clinically significant events arise because of the presence of antibodies to non-ABO/RhD RBC antigens that are not typically screened for by blood banks. Even in those patients at greatest risk of allosensitization, ie, the chronically transfused, debate continues as to the efficacy of pretransfusion matching of donor blood and prevention of clinically significant sequela, as well as subsequent allosensitization. In part, this debate arises from the expense of laboratory testing and the sheer abundance (>300) of potentially immunogenic non-ABO/RhD blood group antigens.24,25Consequently, a need continues to exist to diminish adverse immune reactions while providing the necessary oxygen carrying capacity in the chronically transfused patient.

Because transfusion rejections are immune-mediated responses arising from the presence of antigenic proteins, glycoproteins, and glycolipids located on the RBC membrane, significant research over the last 20 years has been devoted to the development of acellular blood substitutes lacking these surface antigens.26,27 Though commendable in theory, the safety and efficacy of acellular/liposome-encapsulated blood substitutes is still in question.28 Indeed, the short circulating life as well as secondary iron toxicity after repeated transfusions with these agents is of major concern. In contrast, our work on PEG-modified RBC is aimed at maintaining cellular integrity, longevity, and the myriad of other functions of the RBC while reducing its inherent antigenicity and immunogenicity by physically masking membrane antigenic sites.

Hence, establishing a balance between decreased cellular antigenicity and normal RBC structure and function is a major concern in PEG-modification of cells. Indeed, the need for balance is shown by the fact that at high levels of derivatization, human and murine RBC show some loss of function. Importantly though, these adverse structural and functional effects in human RBC occur at derivatization dosages (>5 mmol/L) well in excess of that needed to effectively attenuate antigenic recognition of RBC antigens as shown in Figs 3 through 5.

Indeed, the ability to globally camouflage the RBC surface should be particularly desirable in transfusion medicine because of the sheer abundance of RBC antigens, their ability to mediate a potent immune response, and the difficulty/expense of adequately screening donors and recipients. Effective PEG coverage of the cell surface also diminishes cell-cell interactions. In mixed cell populations (Fig 1B), mPEG-modified cells significantly diminished aggregation of the unmodified RBC in response to antisera. Others have shown that PEG-derivatization of human RBC greatly reduces the low shear viscosity of these cells when suspended in autologous plasma.29 These findings may be of particular interest in the treatment of diseases such as sickle cell anemia, in which vasocclusive events—mediated by vascular wall adherence and RBC aggregate formation—play an important role in the disease pathology. Use of the less “adhesive” PEG-RBC may make it possible to transfuse sickle patients at a lower level than the new NIH guidelines recommend because of the antiaggregation effect of PEG-RBC.

Of potential concern is whether the chronic administration of mPEG-modified RBC is safe. As shown in this study, mice hypertransfused with mPEG-RBC showed no obvious toxicity. Furthermore, based on projections of human dosages, annual PEG administration (eg, to chronically transfused adult sickle or thalassemic patients) will range from 230 to 760 g of PEG (based on 18 and 30 U of blood per year, respectively). Using a mean human body weight of 68.2 kg (150 lb), this results in an annual PEG dosage of 3.4 to 11.1 g/kg. This is substantially less than the amount of PEG administered to the hypertransfused mice when adjusted to an annual rate (ie, 28 g/kg body weight). Hence, the lack of observed toxicity in our repetitively transfused mice suggests—but does not prove—that the acute and short-term transfusion of PEG-RBC should be safe and well tolerated in humans. Unresolved is the question as to whether the long-term (eg, 10 years) administration of PEG-RBC to the chronically transfused patient will be safe. However, in support of its potential safety, animal models have shown that PEG (up to approximately 10 kD) and PEG-conjugated proteins (eg, PEG-Hb) are effectively excreted via the kidneys, and thus, would not likely accumulate within the transfused patient.30 

Finally, as shown in Fig 6, extensive derivatization of murine cells did result in foreshortened in vivo survival. While human, canine, and sheep RBC show much greater in vitro “tolerance” to mPEG-modification than the inherently fragile murine cell, it may be that to effectively camouflage all RBC antigens (eg, the A and B determinants to yield a “universal” RBC), some decrement in circulation time might be observed. This, however, might be tolerable because in profoundly anemic patients already allosensitized, transfusion of lysis-resistant RBC that effectively oxygenate the tissues would still be of significant benefit. Furthermore, if these cells prove efficacious in diminishing the risk of allosensitization, a moderate decrease in the in vivo survival would also be acceptable, because the consequences of sensitization are so much greater.

In summary, our immediate goal in developing PEG-derivatized RBC is not to necessarily create a universal RBC, but rather to produce antigenically and immunogenically attenuated cells with normal or near normal in vivo survival for use in the chronically transfused or allosensitized patient. To this end, we have shown that mPEG derivatization has no detrimental effect on RBC structure or function (morphology, membrane stability, P₅₀, ion homeostasis, and cellular deformability) at concentrations that significantly reduce RBC antigenicity and immunogenicity.4-7 Thus, use of these “stealth erythrocytes”4 in the chronically transfused or in patients with rare blood types may be effective at both delivering oxygen as well as attenuating the risk of allosensitization.

The authors thank Simone Petrocine, Todd P. Christian, Dave L. Devernoe, Scott G. Menzie, and Stanley P. Mudzinski for technical assistance and advice.