UV sensitization with psoralen derivatives

Proc. Nati. Acad. Sci. USA Vol. 91, pp. 5552-5556, June 1994 Medical Sciences

Selective inactivation of viruses in the presence of human platelets: UV sensitization with psoralen derivatives (blood/viral inactivation)

RAYMOND P. GOODRICH*t, NAGENDER R. YERRAM*, BOON H. TAY-GOODRICH*, PATRICIA FORSTER*, MATTHEW S. PLATZt, CHANDRIKA KASTURIt, SANG CHUL PARKS, JEAN NICHOLAS AEBISCHERf, SAROJ RAIt, AND LORETTA KULAGAt *Cryoplhm Corporation, 2585 Nina Street, Pasadena, CA 91107; and tThe Ohio State University, Department of Chemistry,

120 West 18th Avenue,

Columbus, OH 43210

Communicated by John D. Baldeschwieler, January 21, 1994

activated in a manner that triggers a chemical reaction with the virus, which renders the virus inactive. It must perform this function in a manner that does not affect or alter the properties of the cellular or protein elements present in the sample. The desired ideal specificity is not easily achieved. Viruses vary greatly in form or function (5). There are RNA and DNA viruses with single or double strands of the appropriate nucleic acid. There are enveloped and nonenveloped forms of the virus as well as proviral forms of virus and intracellular as well as extracellular forms of viruses. While it is possible to envision a way to target chemical sensitizers to viruses through various measures, including antibodies and membrane selectivity, such methods lack the means to be able to attack the large range of all viruses that may be present in a blood sample. For this work, we have selected a class of sensitizers that are capable of enhanced specific binding of nucleic acids in the presence of lipid membranes and plasma proteins. Since blood platelets, erythrocytes, and plasma proteins do not contain genomic nucleic acid, it is possible, via specific binding of nucleic acids relative to lipid membranes and plasma proteins, to target viruses in the presence of these agents. In this paper, we have attempted to directly address the issues of specificity of the viral targeting, the killing capacity of antiviral agents, and the levels of peripheral damage occurring to the cellular elements present in samples of platelets. We are pleased to report a sensitizer that upon photoactivation can inactivate a model bacteriophage (46) deliberately inoculated into platelet concentrates with acceptable recovery of platelet properties as measured by several assays including a hypotonic shock response (HSR) assay. Furthermore, these results can be obtained in pure plasma under normal oxygen tensions and without the introduction of quenchers.

Inactivation of viruses in blood products reABSTRACT quires that the method employed display selectivity in its action for viral elements while not affecting the biological entity of interest. Several methods have been developed for the treatment of human plasma or products derived from human plasma. An effective technique for the treatment of the cellular components of blood has been lacking, in part due to the inability to develop agents capable of selectively targeting viral agents in the milieu of cellular material. In this paper, we examine the behavior of a group of viral sensitizers designed to be added to cellular samples and be activated upon exposure to UVA light. Upon activation, these agents are capable of disrupting nucleic acids of the virus in a manner that renders them inactive for proliferation. The selectivity observed in this inactivation is determined by the chemical structure of the sensitizer, which can be varied to increase viral killing capacity while diminishing collateral damage to cellular and protein constituents. The use of blood products for therapeutic applications has been and will continue to be an important component of medical science and health care. Because of this need, there will also continue to be a necessity to guard against and be prepared for the possibilities ofdisease transmission via these blood products. The use of screening diagnostics for blood products has greatly reduced the risk of transmission of diseases such as AIDS and hepatitis, but there are still finite risks that are associated with such products (1). Screening methods make the blood supply safer, but they do not make it invulnerable to insult by new and yet undefined agents of disease. It is because of this risk that it is a worthwhile endeavor to develop methods to actively remove or inactivate viral agents that may be present in the blood supply. Several techniques have been examined over the course of the last 10-15 years for the inactivation of viruses in blood products (2). Many of these methods utilize UV and visible light in combination with sensitizers (12). Sensitizers that target viral membranes attempt to exploit quantitative differences in photosensitivity between viral and cellular membranes (3). Often, in application, the lack of significant quantitative differences in photosensitivity of viral and cellular membranes results in unacceptable levels of cellular damage. An ideal chemical agent will be effective as a viral inactivating agent in the presence of cellular or protein products if it is capable of binding to the virus in the presence of the cellular components without significant binding to the other biological elements present. In turn, it must somehow be

MATERIALS AND METHODS Chemical Sensitizers. Chemical agents used for these studies were prepared according to methods described previously or that will be described elsewhere. All chemical agents prepared for this study were characterized by methods of infrared spectroscopy, 1H NMR, 13C NMR, fast atom bombardment mass spectrometry, and UV spectroscopy to confirm the assigned structure and presence offunctional groups. Structures for these agents (sensitizers 1-3) are shown below. Each compound was prepared with a bromine substituent in the C-5 position of psoralen. We have previously reported on the improvement of the level of viral inactivation observed Abbreviations: BSA, bovine serum albumin; HSR, hypotonic shock

The publication costs of this article were defrayed in part by page charge

payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

response.

tTo whom reprint requests should be addressed.

5552

Medical Sciences: Goodrich et al. Br

0

1 OCH2CH2CH2R

1

R=NH3+

2

R = NHEt2

3

R = NEt3

Sensitizers 1-3

when this derivation of the psoralen molecule is made and provided an explanation for this behavior (6). The sensitizers are derivatives of the parent 8-methoxypsoralen and contain a positively charged side chain group to increase the water solubility of the molecules. Each derivative was prepared with a positively charged group but with varying amounts of acidic hydrogens to permit the separation of charge effects from hydrogen bonding effects when examining the overall behavior of the molecules in solution. Sensitizer 1 possesses three acidic hydrogen atoms on the ammonium group, whereas sensitizer 2 possesses one and sensitizer 3 possesses no acidic hydrogens, respectively. Dialysis Binding Experiments. To determine the binding preferences of the sensitizers to various agents, a set of dialysis experiments was performed using a set of custommade polystyrene dialysis chambers. The unit consists of three chambers capable of holding a volume of 10 ml of solution. Each chamber was separated from the adjoining chamber by a dialysis membrane (Mr cutoff of 5000; Fischer). The center chamber was loaded with solution containing sensitizer at 0.5 mg/ml in phosphate-buffered saline (PBS). The other two adjoining chambers were loaded with solution containing the agents for which binding was to be tested. For example, to test preferential binding of sensitizer to membranes versus nucleic acid, one chamber was loaded with a suspension of multilamellar vesicles composed of dioleoylphosphatidylserine in PBS. These vesicles were prepared by vortexing the lipid (4.17 mg/ml; Avanti Polar Lipids) for 30 sec to obtain a uniform suspension of vesicles in PBS. The other chamber was loaded with a suspension of calf thymus double-helical DNA (Sigma) in PBS. Samples were allowed to equilibrate with constant agitation for a period of 24 hr. At the end of the 24 hr, the solutions were removed from the individual chambers. The optical density of each compartment was then read on a spectrophotometer at 350 nm to determine the amount of sensitizer present in each chamber. For experiments involving vesicles, 5% Triton X-100 (Sigma) was used to dissolve vesicles and clarify the solution. The same volume of Triton X-100 was added to the other chambers to compensate for dilution and background effects. Values expressed in Table 1 represent the percent increase of the sensitizer concentration relative to the value observed at equilibrium for the control sample-i.e., [(absorbance in sample chamber/total absorbance in all chambers) x 100%] - 33.3%. Similar experiments were performed testing the preferential binding of sensitizer to bovine serum albumin (BSA; Sigma) versus DNA as well as DNA versus poly(adenylic acid) [poly(A); Sigma]. Cell Samples. All platelet samples used in this experiment were obtained from normal healthy volunteers with consent. Samples were obtained as platelet concentrates from the San Diego Blood Bank. Samples had been drawn into citrate/ phosphate/dextrose/adenine and separated into platelet concentrates using normal blood banking procedures. All units were 18-24 hr old at the point at which they were treated.

Proc. Nati. Acad. Sci. USA 91 (1994)

5553

Samples of two platelet concentrates were pooled and then subsequently split into two units each for paired controls. This was repeated for a total of n = 5 paired sets of samples. Untreated controls were not exposed to virus, sensitizer, or UV light. For viral inactivation studies, samples were spiked with 06 virus to provide a starting viral titer of at least 6.0 loglo units. This virus is a bacteriophage that possesses a lipid envelope and double-stranded RNA. It was selected due to ease of handling. Sensitizer was added to the samples from a stock solution of 5.0 mg/ml in PBS to provide a final concentration as specified. For irradiation,.samples were transferred to 500-ml Cryocyte bags (PL-269; Fenwal Laboratories, Deerfield, IL). The material making up these containers permits transmittance of 65% ofthe incident light at 350 nm. All samples were irradiated using a homemade UV reactor system. This reactor consists of two banks of UV lights capable of generating UV light in the 300- to 400-nm region (UVA light with a A. = 350 nm). Maximal UV intensity at the sample position was measured at 3.6 mW/cm2 (top) and 3.4 mW/cm2 (bottom) using a light meter (Graseby Optronics, Orlando, FL). Samples were mixed on an orbital shaker (Lab-Line Instruments model 351, Melrose Park, IL) while being irradiated for the times specified. After treatment, samples were stored on a platelet shaker (Helmer Laboratories, Chicago) in a 220C incubator (Forma Scientific, Marietta, OH). All samples were stored in the standard PL-732 (Fenwal Laboratories) platelet storage container in which they were originally provided. All control and treated samples were assayed daily for functional properties over the course of a 4-day storage period as follows. Agregation with ADP and Collan. Response of platelets to ADP and collagen was tested using a ChronoLog LumiAggregometer. Each sample was tested for aggregation against a 20 pM concentration of ADP or a collagen (Sigma) concentration of 20 pg/ml. The degree of aggregation was calculated as the percent change in light transmittance occurring after addition of ADP or collagen to the samples. Samples tested for ADP aggregation were also supplemented with 2.5 mM calcium chloride (Sigma). During the testing, samples were stirred at 1000 rpm at a temperature of 3rC. For each test, the platelet number was adjusted to 300-500 x 103/jud. In each case, autologous platelet-poor plasma was used as the blank. HSR. HSR was determined as described by Armitage et al. (7) with minor modifications. Cell concentrations were adjusted to 300-500 x 103/pul. Samples were run on a Perkin-Elmer Lambda 4A spectrophotometer operating at 610 nm. An 800-IlI sample was used per test. The baseline was recorded by adding 400 td of 0.9%o saline to the cuvette and allowing the sample to run for 10 min. The percent transmittance was recorded after 10 min (Tb). The same steps were repeated for a second aliquot of the sample. In this case, 400 pl of water was used for the dilution step. The maximum transmittance observed (Tmx) was recorded. At 10 min post-mixing with water, a final transmittance (T10) was recorded. The percent reversal was determined according to the following formula: % reversal = [(Tm,, - Tlo)/(Tm - Tb)] x 100.

Morphology Score. A modification of the method of Kunicki et al. (8) was used to assess cell morphology. Samples were examined using a Nikon phase-contrast microscope (Microphot FX). A 10- to 20-p1 sample was placed on a microscope slide and covered with a coverslip. Samples were examined under phase objectives at x800. A total of 100 cells was counted while the cells were in motion. Cells were categorized as follows: rank 0, dead cells, rings, and bal-

5554

Medical Sciences: Goodrich et al.

loons; rank 1, pseudopods; rank 2, spherical cells; rank 4, discoid cells. Total morphology was calculated using the following formula:

Score = (no. dead x 0) + (no. pseudopods x 1) + (no. spherical x 2) + (no. discoid x 4).

The maximum possible score is 400. A score of 200 or greater is generally acceptable. pH Measurements. Sample pH was measured on an aliquot of the sample using a Corning pH meter model 240. Viral Inactivation Measurements. The level of viral reduction was determined using the plaque-forming assays as described by Maniatis et al. (9). The model phage 06 (ATCC 21781-Bi) was used for all experiments. The host for the phage is Pseudomonas syringae (ATCC 21781). All bacterial cultures and viral samples were obtained from the American Type Culture Collection stock supplies. Samples were cultured using nutrient broth yeast extract agar (Difco).

RESULTS Equilibrium Dialysis Experiments: Preferential Binding Studies. Results from equilibrium dialysis experiments are given in Table 1. Examination of the data indicates several interesting trends. Sensitizer 1, with three acidic hydrogens, shows a pronounced ability to bind to lipid membranes as evidenced by the increase in the sensitizer concentration of 21.8% + 1.4% above the equilibrium concentration. This same observation was made with sensitizer 2, which has one acidic hydrogen group. In this case as well, there is a direct indication of increased binding of the sensitizer to lipid relative to nucleic acids by the increase in the sensitizer concentration in the lipid chamber over that observed in the nucleic acids chamber. This behavior is in direct contrast to that of sensitizer 3, which does not possess an acidic hydrogen. In this case, there is a greater accumulation of the sensitizer in the chamber containing the nucleic acids. Comparison of the ratios of the increased concentrations in the lipid versus nucleic acids chambers shows that whereas sensitizers 1 and 2 show a marked preference for the lipid, Table 1. Equilibrium dialysis experiments for determination of preferential binding Sensitizer 2 1 n 3 Parameter ± ± ± % increase lipid 21.8 1.4 17.0 3.0 7.2 1.8 5 % increase DNA 5.0 ± 0.6 7.9 ± 2.6 11.2 ± 3.5 5 4.4 2.2 0.7 Lipid/DNA* % increase poly(A) 9.6 ± 2.3 6.1 ± 0.6 5.9 ± 2.4 3 % increase DNA 12.0 + 0.3 12.8 ± 3.6 10.7 ± 0.2 3 0.5 0.8 0.6 Poly(A)/DNA* % increase BSA 0.11 ± 0.2 0.15 ± 0.3 4 0 % increase DNA 20.3 + 3.8 16.4 ± 5.6 21.2 ± 7.4 4 0 0.01 0.01 BSA/DNA* The % increase values represent the percentage of sensitizer present above the equilibrium concentration of the sensitizer. The amount of sensitizer present was determined by direct reading of the absorbance at the A.. of emission of the sensitizer. Absorbance of the sensitizer in each compartment was determined, and the sum was used to obtain the total amount of sensitizer present. Values represent means ±1 SD for a minimum of three experiments. For competition experiments with nucleic acids, values from experiments with DNA and poly(A) were combined. n, number of experiments. *A value >1.0 represents preferred binding to the agent in the numerator. A value 0.05) in platelet quality'as measured by ADP aggregation, collagen aggregation, HSR, or morphology was observed at days 1, 3, and 4. Values for day 2 of storage showed significant deviation from control values. This may be due in part to the low sample number used for this study. During storage, a significant drop in pH relative to the control was observed. This drop in pH was independent of the presence or absence of the sensitizer (data not shown) and appears to be associated with UV light exposure alone. ±

DISCUSSION A major objective of efforts to inactivate viruses in the presence of blood components is to do so without compromising the integrity of the blood components needed for therapeutic applications. To accomplish this, one must selectively and preferentially target viruses present in the milieu of cellular and noncellular components present in blood collections. In this paper, we have examined a class of molecules that are capable of targeting nucleic acids. The rationale for their use in this application is the specificity that nucleic acid targeting can confer on samples treated with such agents. Red blood cells, plasma proteins, and platelets do not contain genomic nucleic acid. By selecting agents that are capable of binding nucleic acids, it is theoretically possible to selectively target these agents. Viruses, although they may differ in structural morphology and composition, are uniform in the feature of possessing nucleic acid. A technique that targets this common element to all viruses therefore has the potential for uniformity in its ability to kill wide varieties and types of viruses.

Proc. Natl. Acad. Sci. USA 91 (1994)

5555

As we have observed here, however, these assumptions about specificity and selectivity must be subjected to direct evaluation. In equilibrium dialysis experiments, we have observed that variations in the side chain substituents of the psoralen derivatives can greatly affect the interaction of these molecules with phospholipid membranes. The presence of acidic hydrogens on the ammonium group can lead to enhanced binding of the sensitizers to membranes. These 8-methoxypsoralen derivatives were originally constructed with the intention of using the ammonium substituent to simply increase water solubility. All compounds are positively charged; thus, it is unlikely that the interaction with the membranes is due to this factor. Each side chain differs in nature', however, due to the presence or absence of acidic hydrogens on each ammonium group. Those sensitizers possessing one or more acidic hydrogens (sensitizers 1 and 2) bind to phospholipid membranes in preference to doublestranded DNA. This may in turn lead to the accumulation of sensitizers 1 and 2 in platelet membranes and subsequent damage upon photolysis. The compound (sensitizer 3) lacking in acidic hydrogens preferentially binds to DNA relative to membranes. This may explain the ability of sensitizer 3 to photoinactivate 06 in platelet concentrates in plasma in the presence of oxygen and in the absence of quenchers frequently used to suppress membrane damage. The presence or absence of acidic hydrogen atoms on the ammonium side chain provides for the possibility of hydrogen bonding between the ammonium group and the phosphate segment of the phospholipids (Fig. 1). Such interaction can thermodynamically favor the accumulation of the sensitizers in membranes. It is interesting to note that this potential interaction has been proposed for the enhanced binding of other compounds bearing ammonium groups, such as 4'aminomethyl-4,5',8-trimethylpsoralen (AMT), to DNA phosphate groups (11). Only in the absence of membranes for competitive binding can such enhancement of interaction with native DNA be anticipated. Our results imply that AMT should not be a selective antiviral agent. This may explain the previously unaccounted for necessity to use quenching agents to protect cellular membranes from the action of AMT and light (4). It is very worthwhile to note that the results from the equilibrium dialysis experiments are consistent and predic-

Table 3. Irradiation of full units of platelet concentrates in the presence of sensitizer 3 (quaternary ammonium derivative) Parameter ADP agg. HSR pH Day Collagen Morphology 1 57 ± 11 72 ± 6 Control 74 ± 5 7.48 ± 0.08 298 ± 15 55 ± 10 Treated 68 ± 8 66 ± 9 7.48 ± 0.07 259 ± 38 P NS NS NS NS 0.023 2 Control 47 ± 14 60 ± 8 70 ± 8 7.46 ± 0.04 242 ± 28 Treated 34 ± 8 55 ± 6 60 ± 10 7.31 ± 0.05 215 ± 26 P 0.034 0.021 NS 0.002 0.013 54 8 3 Control 31 6 68 ± 4 7.42 0.06 224 ± 23 Treated 27 6 55 11 65 ± 3 7.16 0.08 206 ± 12 P NS NS NS 0.0004 NS 4 24 9 Control 47 ± 11 69 ± 8 7.39 0.04 194 ± 24 Treated 25 11 49 ± 12 66 ± 3 6.97 0.11 177 ± 24 P NS NS NS 0.0003 NS Single units of platelet concentrates were irradiated with UVA light. Samples were irradiated in a Cryocyte bag (PL 269) for 8 min (3.36 J/cm2) in the presence of sensitizer 3 (30 tg/ml). Each sample was spiked with 6 loglo units of 46 bacteriophage. After treatment, the samples were transferred to a regular platelet storage container (Fenwal PL-732) and stored with agitation at 220C for 4 additional days. In vitro properties were monitored over the course of the storage interval. Data represent the mean of five separate experiments with paired controls. A total of 5.12 + 0.4 logio units of viral kill was obtained under these treatment conditions. Less than 0.2 logio unit of viral kill was obtained upon irradiation under the same conditions in the absence of the sensitizer. Concentrations of 20 MM ADP and collagen at 20 ,g/ml were used for the aggregation (agg.) response tests. Statistical analysis was performed using Student's t test for paired data. Values of P > 0.05 are denoted as NS (not significant variation from control).

5556

Medical Sciences: Goodrich et al.

Proc. Nad. Acad. Sci. USA 91

0

~~t~~~>0 0- CH, -~

soOCH~ H Membrane phospholipids ~

0

CHi-o-p- OCH2-CH,-R 0

H O(CH,)- N-R R

0

0

Sensitizers 1, 2 Br

0

CH,

CHr

__0 IH Membrane phospholipids

CH-,0-pII

N0

)3NI )RN

possible explanation for this observation is that the accumulation of the sensitizers in the membrane and subsequent activation with UV light leads to chemistry occurring between the sensitizer and proteins or lipids in the cellular membranes (13). Such chemistry can lead to a breakdown in the membrane integrity or structure. Active oxygen species created by the action of the sensitizer may also induce cell damage. These interactions may be expected to produce alterations severe enough to greatly compromise a cell's ability to survive the circulatory system and carry out its primary function in hemostasis. Finally, this work indicates that, by taking into account these features of structural design of sensitizers and the impact on conferred selectivity, it is possible to produce compounds capable of specifically targeting viral agents present in blood products and activating them upon exposure to UV light of the appropriate wavelength to produce selective kill of bacteriophage. Preliminary experiments indicate that sensitizer 3 also effectively inactivates human immunodeficiency virus (R.P.G., N.R.Y., S. Coker, and M.S.P., unpublished observations). Results obtained with sensitizer 3 in these studies show the potential promise of this approach in providing a method for the sterilization of blood products while not compromising the quality of the blood component being treated. Such methods can provide a means for the development of active methodologies for the safeguarding of the integrity of the blood supply.

OCH2-CH2-R

R O(CH,

(1994)

R

0

Sensitizer3\3

Br

FIG. 1. Diagram of potential interactions occurring between phospholipid head groups and the ammonium group of the sensitizers. The hydrogen bonding is possible only for sensitizers 1 and 2, which possess a free hydrogen. This is not possible for sensitizer 3. This interaction can lead to the accumulation of sensitizers in membranes of cells. Upon activation, these agents can damage cellular membranes and proteins.

tive of the data obtained from platelet concentrate samples treated with each of the sensitizers. In this case, those sensitizers bearing acidic hydrogen atoms show enhanced damage to the platelets. This is manifested primarily in a decreased HSR, a primary indicator of platelet membrane integrity. Although such cells still maintain aggregation potential to agents such as collagen, which are primarily receptor-mediated aggregation responses, the loss of the HSR in these cells is indicative of severe membrane alteration. One

1. Menitove, J. E. (1991) in Transfusion Transmitted Infections, eds. Smith, D. M. & Dodd, R. Y. (Am. Soc. Clin. Pathologists, Chicago), p. 10. 2. Wagner, S. J., Friedman, L. I. & Dodd, R. Y. (1991) Transfusion Med. Rev. 1, 18-32. 3. O'Brien, J. M., Gaffney, D. K., Wang, T. P. & Sieber, F. (1992) Blood 1, 277-285. 4. Margolis-Nunno, H., Williams, B., Rywkin, S., Geacintov, N. & Horowitz, B. (1992) Transfusion 32, 541-547. 5. Lamberson, H. V. (1991) in Transfusion Transmitted Infections, eds. Smith, D. M. & Dodd, R. Y. (Am. Soc. Clin. Pathologists, Chicago), pp. 23-53. 6. Rai, S., Kasturi, C., Grayzar, J., Platz, M. S., Goodrich, R., Yerram, N. R., Wong, V. & Goodrich, B. (1993) Photochem. Photobiol. 58, 59-65. 7. Armitage, W. J., Parmar, N. & Hunt, C. J. (1985) J. Cell. Physiol. 123, 241-248. 8. Kunicki, T. J., Tuccelli, M., Becker, G. A. & Aster, R. H. (1975) Transfusion 15, 414-421. 9. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual (Cold Spring Harbor Lab. Press, Plainview, NY). 10. Schneider, H.-J. & Blatter, T. (1992) Agnew. Chemie Int. Ed. Engl. 31, 1207-1208. 11. Isaacs, S. T., Shen, C. J., Hearst, J. E. & Rappoport, H. (1977) Biochemistry 16, 1058-1064. 12. Produz, K. N., Lytle, C. D., Keville, E. A., Budacz, A. P., Vargo, S. & Fratantoni, J. C. (1990) Transfusion 31, 415-422. 13. Midden, W. R. (1988) in Psoralen DNA Photobiology, ed. Gasparro, F. P. (CRC, Boca Raton, FL), Vol. 2, pp. 6-15.