Proc. Natl. Acad. Sci. USAVol. 93, pp. 4403-4408, April 1996Medical Sciences

Synthesis and targeted delivery of an azidothymidinehomodinucleotide conferring protection to macrophages againstretroviral infection

[di(thymidine-3'-azido-2',3'-dideoxy-D-riboside)-5'-5'-pl-p2-pyrophosphate (AZTp2AZT)/human immunodeficiency virusinhibition/feline and murine immunodeficiency/drug targeting/erythrocytes]

MAURO MAGNANI*, ANNA CASABIANCA*, ALESSANDRA FRATERNALE*, GIORGIO BRANDIt, SANDRA GESSANIt,ROSAMUND WILLIAMS§, MARCO GIOVINE¶, GIANLUCA DAMONTE¶, ANTONIO DE FLORA$, AND UMBERTO BENATrTI¶Institute of *Biochemistry "Giorgio Fornaini" and tHygiene, University of Urbino, Via Saffi 2, 61029, Urbino, Italy; tLaboratory of Virology, Istituto Superiore diSanita, Viale Regina Elena 299, 00161, Rome, Italy; §Lepetit Research Center, Gerenzano, 21040, Varese, Italy; and ¶Institute of Biochemistry, University ofGenoa, Viale Benedetto XV 1, 16132, Genoa, Italy

Communicated by Joseph Hoffmnan, Yale University, New Haven, CT, January 4, 1996 (received for review October 25, 1995)

ABSTRACT The infectivity and replication of human(HIV-1), feline (FIV), and murine (LP-BM5) immunodefi-ciency viruses are all inhibited by several nucleoside analoguesafter intracellular conversion to their triphosphorylated de-rivatives. At the cellular level, the main problems in the use ofthese drugs concern their limited phosphorylation in somecells (e.g., macrophages) and the cytotoxic side effects ofnucleoside analogue triphosphates. To overcome these limi-tations a new nucleoside analogue homodinucleotide, di(thy-midine-3 '-azido-2 ',3 '-dideoxy-D-riboside)-5'-5 '-pl-p2-pyro-phosphate (AZTp2AZT), was designed and synthesized.AZTp2AZT was a poor in vitro inhibitor of HIV reversetranscriptase, although it showed antiviral and cytotoxicactivities comparable to those of the parent AZT when addedto cultures of a HTLV-1 transformed cell line. AZTp2AZTencapsulated into erythrocytes was remarkably stable. Induc-tion of erythrocyte-membrane protein clusterization andsubsequent phagocytosis of AZTp2AZT-loaded cells allowedthe targeted delivery of this impermeant drug to macrophageswhere its metabolic activation occurs. The addition ofAZTp2AZT-loaded erythrocytes to human, feline, and murinemacrophages afforded almost complete in vitro protection ofthese cells from infection by HIVBa.L, FIV, and LP-BM5,respectively. Therefore, AZTp2AZT, unlike the membrane-diffusing azidothymidine, acts as a very efficient anti-retroviral prodrug following selective targeting to macro-phages by means of loaded erythrocytes.

Mononuclear phagocytes are important target cells for humanimmunodeficiency virus type 1 (HIV-1) infectivity and prop-agation in most tissues (1-6). Furthermore, because cytopathiceffects on these cells are minimal, they serve as major reser-voirs of HIV-1 with an important role in the pathogenesis ofAIDS.To be pharmacologically active, most antiviral nucleoside

analogues [3'-azido-3'-deoxythymidine (AZT), ddC, etc.]must first be phosphorylated by cellular kinases (7, 8). Theactivity levels of these enzymes depend on the cell type and thecell activation state (9, 10). Usually, quiescent cells have lowlevels of the enzymes responsible for nucleoside analoguephosphorylation, while activation (e.g., mitogen stimulation)results in increased activity levels. Macrophages phosphorylateAZT and ddC at a much lower rate than peripheral bloodmononuclear cells (11, 12). Furthermore, granulocyte/macrophage colony-stimulating factor strongly influences viralproduction and antiviral activity of dideoxynucleoside ana-

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

4403

logues (13). Thus, anti-retroviral drugs are expected to havedifferent effects in lymphocytes and monocytes/macrophagesand, ideally, each drug should be tested in both cell systems or,eventually, specifically designed for the effective protection ofat least one cell compartment.To overcome the low ability of resting macrophages in

phosphorylating the antiviral dideoxynucleosides, we havepreviously developed a drug-delivery system, based on the useof autologous erythrocytes, for the direct administration ofphosphorylated nucleoside analogues (14). Using this drug-delivery system we have shown that erythrocyte-encapsulatedddCTP protects in vitro human macrophages from HIV-1infectivity and both in vitro and in vivo murine and felinemacrophages from the infectivity of the murine immunodefi-ciency virus LP-BM5 and feline immunodeficiency virus(FIV), respectively (14-16). Furthermore, the anti-retroviralactivity of erythrocyte (RBC)-resealed ddCTP always ex-ceeded that of the nonphosphorylated nucleoside analogue(17).

In the past few years it has been shown that the nucleotideanalogues are responsible in the cell for antiviral activity andcell toxicity (18, 19). Based on these considerations, wethought that an antiviral drug designed to protect macrophagesshould ideally be administered in a phosphorylated form, yetnot as the nucleoside 5'-triphosphate, to avoid cell toxicity, butas a close precursor of it. We report here the chemicalsynthesis, characterization, macrophage-targeted delivery, andanti-retroviral activity of a new AZT analogue, an AZThomodinucleotide [di(thymidine-3'-azido-2',3'-dideoxy-D-riboside)-5',5'-pt,p2-pyrophosphate (AZTp2AZT)] that fitsthe above-mentioned requirements.

MATERIALS AND METHODSCells and Viruses. Human monocyte-derived macrophages

were prepared and cultured as in Magnani et al. (14). Adher-ence was for 10 days. Murine macrophages were obtained fromthe peritoneal cavity of C57BL/6 mice and cultured as in Rossiet al. (16). Cat monocyte-derived macrophages were preparedand cultured as in Magnani et al. (15).Human macrophages were infected with a preparation of

HIVBa-L for 5 h (4 ng p24/ml). Murine macrophages wereinfected with LP-BM5 viruses prepared as cell-free superna-tant from SC-1 cells (20). Exposure to the virus lasted 24 h at

Abbreviations: HIV, human immunodeficiency virus; FIV, fe-line immunodeficiency virus; AZT, 3'-azido-3'-deoxythymidine;AZTp2 AZT, di(thymidine-3'-azido-2',3'-dideoxy-D-riboside)-5',5'-p ,p2-pyrophosphate; RBC, erythrocyte(s); AZT-MP, 3'-azido-3'-deoxythymidine 5'-monophosphate; AZT-TP, 3'-azido-3'-deoxythymidine 5'-triphosphate.

4404 Medical Sciences: Magnani et al.

3 x 106 cpm/106 cells of reverse transcriptase. Feline macro-phages were infected with the Pisa-M2 isolate of FIV asdescribed in Magnani et al. (15, 21).

Synthesis and Characterization of AZTp2AZT. AZT-MPwas synthesized and characterized as described (22). Synthesisof Pl-AZT-5'-P2-diphenyl pyrophosphate was obtained ac-cording to Michelson (23). AZT-MP tri-n-octylammonium saltwas prepared by dissolving 0.25 mmol of AZT-MP-free acid(100 mg) in 3.0 ml of methanol and 0.45 ml of tri-n-octylamine.The mixture was stirred for 30 min at 25°C and vacuum dried.The residue was suspended in 2 ml ofN,N-dimethylformamideand dried under reduced pressure. The latter procedure wasrepeated three times.

Pl-AZT-5'-P2-diphenyl pyrophosphate (0.25 mmol) wasthen dissolved in 1.8 ml of anhydrous pyridine and added to0.25 mmol of the AZT-MP tri-n-octylammonium salt, pre-pared as above, together with 0.32 ml of hexamethylphospho-triamide. The mixture was dried under reduced pressure; 0.4ml of anhydrous pyridine was added to the residue and theresulting solution was stirred for 24 h at 25°C and then vacuumdried. The residue was suspended in 6 ml of H20, the pH wasadjusted to 8.0 with 1M NaOH, and the mixture was extractedthree times with 6 ml of diethyl ether. The lower aqueous layer,recovered by centrifugation, was then submitted to HPLCfractionation in 1-ml aliquots. A Bio-sil C18 HL 90-10 reverse-phase column (Bio-Rad; 210 x 10 mm, 10-,tm particle size)was used. Elution was carried out stepwise with aqueousethanol at increasing concentrations as follows: 3% (vol/vol)for 7 min, 5% for 10 min, 15% for 15 min, 100% for 16 min,and again 100% for 25 min at a constant flow of 6 ml/min. Theeluted compounds were detected by a Hewlett-Packard 1040A diode array spectrophotometric detector set at 260 nm. Thenucleotide eluted at 11 min was lyophilized and submitted tochemical analyses.Mass spectra were acquired using a single quadrupole

Hewlett-Packard engine 5989 A equipped with an electrosprayion source. The mass spectrum of the purified product ofsynthesis showed a molecular ion at m/z 675.9, consistent withthe [M-H]- ion of the expected AZTp2AZT molecule (Fig. 1Inset). The minor peak at m/z 697.7 corresponds to the[M-2H+Na]- ion of the same molecule. The UV absorptionspectrum of the same purified product, recorded in phosphate

Abundance 675.9

1600

1200

800

400

697.7

mnlz 500 600 700 6000 0

H\N/^ \N CH, CH.- H

11 11 10o CH, P-° P-F O CH,o

C C OH OH C C

H /I H

H N, N, H

FIG. 1. Structure of AZTp2AZT. (Inset) Mass spectrum of thesynthesized molecule.

buffer (pH 4.9) and 40% (vol/vol) methanol in water, exhib-ited an absorption maximum at 265 nm. Finally, the 'H NMRspectrum recorded at 200 MHz in the temperature range from20°C to 30°C on a Varian Gemini spectrometer in 2H20showed the most significant chemical shifts at 1.639 Sppm(-CH3), 2.225 Sppm (-O-CH2), 3.943 8ppm (broad 1'-CH and2'-CH2), 4.275 Sppm (3'-CH), 5.955 /ppm (4'-CH), and 7.439Sppm (aromatic proton 6-CH). Taken together, these proper-ties are consistent with the structure expected from theabove-described procedure of synthesis-i.e., AZTp2AZT(Fig. 1).

Inhibition of HIV-1 Reverse Transcriptase. Human recom-binant HIV-1 reverse transcriptase and reverse transcriptaseassay kit were from Boehringer Mannheim. Poly(A) oli-go(dT)15 was used as template/primer, and dUTP was used asa labeled nucleotide substrate according to the manufacturer'sinstructions.HPLC Analyses ofAZTp2AZT Metabolites. Cells (RBC and

macrophages) were extracted with perchloric acid as in ref. 24.Neutralized extracts were then used for HPLC determinationof AZTp2AZT. A 5-,um, 100 x 4.6 mm, ODS-Hypersil C18reverse-phase column (Hewlett-Packard) was used. Solvent Awas 0.1 M KH2PO4 containing 5 mM tetra-n-butylammonium(pH 5.0); solvent B was solvent A containing 40% (vol/vol)methanol. The solvent program was a linear gradient (at a flowrate of 0.4 ml/min) starting at 100% solvent A and increasingto 100% solvent B in 30 min. The various compounds weredetected by a Hewlett-Packard 1040 A diode array spectro-photometric detector set at 260 nm. The elution time ofAZTp2AZT was 31 min. AZTp2AZT metabolism was alsoinvestigated under identical conditions in the absence of ATP.In this case AZT-MP and AZT were simultaneously deter-mined by the HPLC procedure described above, using thestandard compounds as described in ref. 21.

Targeting AZTp2AZT-Loaded RBC to Macrophages. Hu-man RBC were loaded with AZTp2AZT at a final concentra-tion of 0.4 ± 0.03 ,tmol/ml RBC as described (14). TheAZTp2AZT-loaded RBC were then modified to increase theirrecognition by macrophages. The procedure is described indetail elsewhere (14). The number of IgG bound per cell andphagocytosis of drug-loaded RBC was determined in vitro asdescribed (14).We also evaluated the amount of AZTp2AZT delivered by

carrier RBC to macrophages and its stability in macrophages.Human RBC were loaded with AZTp2AZT at a concentrationof 0.4 ,tmol/ml RBC and then modified to increase theirrecognition by macrophages as above. Modified RBC wereadded to macrophages at a ratio of 100 RBC per macrophageand phagocytosis was for 24 h. Noningested RBC were re-moved by extensive washing (three times) with RPMI 1640medium and then washed with 0.9% (wt/vol) ammoniumchloride. Macrophages' perchloric acid extracts were thenprepared at 0, 24, 48, and 72 h after RBC phagocytosis andneutralized with K2CO3. The extracts were then processed forsolid-phase extraction of AZTp2AZT and analyzed by HPLCas described above.Assays of Antiviral Activity. The anti-HIV activity of

AZTp2AZT and of the parent compound AZT was directlymeasured in vitro on a HTLV-1 transformed cell line, MT-4.The inhibition of the HIV-induced cytopathic effect was usedas the end point (25).For assays of anti-retroviral activity on infected macro-

phages, AZTp2AZT-loaded RBC (0.4 gmol/ml RBC) wereadded for 20 h at a ratio of 100 RBC per macrophage.Noningested RBC were removed with extensive washing withculture medium. As a control, macrophage cultures weretreated with "unloaded RBC" (i.e., RBC submitted to thesame procedure but without addition of AZTp2AZT), or 0.1,tM AZT in culture medium. Human macrophage culturesreceiving AZTp2AZT-loaded RBC, unloaded RBC, AZT, or

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 4405

Table 1. Recognition of AZTp2AZT-loaded RBC by human macrophagesRBC

AZTp2AZT AZTp2AZT-loadedNative loaded plus ZnCl2/BS3

IgG bound 20-40 20-40 -1500(molecules/cell)

Phagocytosis 0.1-0.2 0.1-0.2 >1 (1-1.3)(RBC/macrophage)

AZT dimer in macrophages ND ND 3 ± 0.3(pmol/106 cells)ND, not detectable. Human RBC were loaded with 0.4 ± 0.03 tumol of AZTp2AZT/ml. Phagocytosis

was for 24 h at 37°C at a ratio of 100 RBC per macrophage.

any addition were then infected with a macrophage-tropicHIV-lBa-L for 5 h. Cell cultures were extensively washed toremove any viral particle associated with macrophages. Cellcultures were then maintained at 37°C, 5% CO2 95% air for 3weeks. New medium was replaced every 3 days, and p24production in culture supernatants was determined using acommercial ELISA kit (DuPont).For each sample, total cellular DNA was isolated from the

macrophages 7 days after infection by lysis with 8 M urea, 0.3M NaCl, and 10 mM Tris-HCl (pH 7.5) for 60 min at 37°C.Extraction was with phenol/chloroform/isoamyl alcohol(25:24:1) and then with chloroform/isoamyl alcohol (24:1).The DNAs were precipitated with ethanol and stored at -20°Cuntil used in the PCR assays. PCR analyses of HIV-1 proviralDNA were done using a commercial kit (Perkin-Elmer GeneAmplimer).The PCR process was achieved using the Perkin-Elmer/

Cetus DNA thermal cycler for 30 cycles of denaturation for 1min at 95°C, annealing and extension for 1 min at 60°C,followed by final extension at 60°C for 10 min. The 115-bp PCRproducts were detected by the oligomer-hybridization proce-dure whereby a 32P-end-labeled SK19 probe with [y-32P]ATPhybridizes in solution to one strand of the amplified product.The probe-target duplex was separated from the unhybridized

3.5

3.0

2.5 \

2.0

1.5 . \

1.0-

0.5-

. * I *0 20 40 60 80

Hours

FIG. 2. AZTp2AZT stability in human macrophages. AZTp2AZT-loaded RBC were added to macrophages at a ratio of 100 RBC permacrophage for 24 h at 37°C. Noningested RBC were removed andmacrophage perchloric acid extracts prepared at different times asindicated. Results are mean ± SD of three experiments.

probe by gel electrophoresis on a 10% polyacrylamide gel andwas then autoradiographed.

Feline monocyte-derived macrophages and AZTp2AZT-loaded RBC were as above. Macrophage cultures receivingAZTp2AZT-loaded RBC, unloaded RBC, or any additionwere infected with 330 infectious doses (i.d.)/well of FIV (PisaM-2) (see ref. 26 for details). Cell cultures were then main-tained at 37°C and 5% CO2 for 4 days and then their total DNAwas isolated as described for human macrophages. Analysis ofFIV proviral DNA was performed by amplification of 498 bpof the viral p24 gag gene in two stages. For the first stage ofamplification, the primers used were 5'-GGCATATCCTAT-TCAAACAG-3' (sense) corresponding to nucleotides 1025-1044 in the viral sequence and 5'-AAGAGTTGCATTT-TATATCC-3' (antisense) corresponding to nucleotides 1680-1699. For the second stage of amplification, the primers usedwere 5'-TATGGTTTACTGCCTTCTCT-3' (sense) corre-sponding to nucleotides 1141-1160 in the viral sequence and5'-AGGGTACTTTCTGGCTTAAG-3' (antisense) corre-sponding to nucleotides 1619-1638.PCR, performed in a Perkin-Elmer thermocycler, was done

in a 50-,ul final volume containing 168 ng of genomic DNA, 50mM KCl, 10 mM Tris-HCL (pH 8.3), 1.5 mM MgCl2, 0.005%Tween 20,0.005% Nonidet P-40, 0.001% gelatin, 150 nM eachprimer, and 5 units of Replitherm DNA polymerase (Epicen-tre, Madison, WI). The reaction mixture, containing the firstpair of primers, was subjected to 40 cycles of denaturation at94°C for 1 min, annealing at 50°C for 1 min, and extension at72°C for 2 min followed by final extension at 72°C for 15 min.A 10-,l aliquot of this amplified mixture was reamplified withthe second pair of primers using exactly the same conditions.The PCR products were analyzed by electrophoresis on

1.5% agarose gel, transferred to a nylon membrane, andhybridized with 498 bp of the p24 gag gene probe cloned in aTA-cloning plasmid (Invitrogen). As an internal control thefeline hexokinase gene was amplified with the following prim-ers: 5'-ACATGGAGTGGGGGGCCTTTGG-3' (sense) cor-responding to nucleotides 2198-2219 and 5'-GTTGCGGAC-GATTTCACCCAGG-3' (antisense) corresponding to nucle-otides 2328-2349 in the human hexokinase type I cDNA.Detection was performed with a feline hexokinase probe.

Labeling of the DNA probes was by random primer DNA-labeling kit from Bio-Rad.Murine macrophages were obtained from the peritoneal

cavity of C57BL/6 mice and cultured as described above.AZTp2AZT-loaded RBC and infection were as described forFIV with the following modifications: the LP-BM5 was pre-pared as a cell-free supernatant from SC-1 cells according tostandard procedures (20). The amplification of defective virusgenome (BM5d) was performed as reported elsewhere (17).

RESULTSTargeting AZTp2AZT-Loaded RBC to Macrophages. Pre-

liminary experiments carried out on a HTLV-1-transformed

Medical Sciences: Magnani et al.

4406 Medical Sciences: Magnani et al.

cell line, MT-4, which is highly susceptible to and permissivefor HIV infection, indicated that the anti-HIV activity (IC50 =0.031 ,/M) and the cytotoxic activity (CC50 > 100 /tM) of thesynthesized AZTp2AZT were comparable to those of theparent nucleoside analogue AZT (IC50 = 0.028 /M and CC50= 70 ,uM, respectively). This close similarity can be explainedby degradation ofAZTp2AZT by the hydrolytic enzymes presentin the fetal calf serum of culture medium.AZT-TP is the pharmacologically active metabolite ofAZT

responsible for powerful inhibition of viral reverse tran-scriptase (7). In our in vitro assays, human recombinant HIV-1reverse transcriptase was competitively inhibited by AZT-TP,with a Ki of 0.05 ,M. Comparatively, AZTp2AZT showed a Kiabove 10 ,tM. Therefore, the antiretroviral activity ofAZTp2AZT is not directly afforded by the homodinucleotideanalogue but rather by its metabolites, most likely by AZT-TP.AZTp2AZT stability was first investigated in human RBC

lysates (21) in the presence of 1 mM ATP. After 2 h at 37°C,90% ofAZTp2AZT was still present (not shown). AZTp2AZTwas then encapsulated in RBC with the aim of constructing acell-based system for the selective targeting of the new antiviralmolecule to macrophages (14). Murine, feline, and humanRBC were used to this purpose following the procedure ofhypotonic dialysis, isotonic resealing, and reannealing, whichyields normally viable engineered red cells (14-16). Up to 1/umol of AZTp2AZT per ml of RBC was encapsulated andanalysis of the biochemical properties of the loaded RBCrevealed no significant alterations by the internalized dinucle-otide. Stability experiments indicated that as much as 90% ofAZTp2AZT was still present after incubation in isotonicconditions of the loaded human RBC for 24 h at 37°C (notshown). This remarkable stability of AZTp2AZT within thecarriers allowed us to introduce the membrane alterations

previously demonstrated to induce opsonization of the loadedRBC and their consequent recognition and phagocytosis bymacrophages (14). The features of this process are reported inTable 1, which shows that a concentration of 3 + 0.3 pmol ofAZTp2AZT per 106 macrophage cells was eventually obtainedby using loaded RBC containing 0.4 ± 0.03 g/mol ofAZTp2AZT per ml. In macrophages, AZTp2AZT showed aprogressive decay until becoming undetectable after 3 dayssince onset of phagocytosis of the loaded RBC (Fig. 2).These data are consistent for metabolism of AZTp2AZT

taking place largely in macrophages rather than in the carrierRBC, thus fulfilling a major requirement of the newly synthe-sized molecule.Comparable results in terms of AZTp2AZT stability in the

carrier red cells and in the macrophages were obtained withboth murine and feline homologous systems (not shown).Thus, for instance, >90% of AZTp2AZT was found afterincubating the loaded murine and feline RBC for 24 h at 37°C.Following uptake of the AZTp2AZT-loaded homologousRBC by the murine and feline macrophages, the decay ofAZTp2AZT takes 3 days to be completed.

Antiviral Activity of AZTp2AZT-Loaded RBC. Humanmonocyte-derived macrophages were cultured for 10 days andthen treated with AZTp2AZT-loaded RBC, unloaded RBC, or0.1 ,uM AZT in culture medium for 20 h. All cell cultures werethen washed with new medium and infected with a macroph-age-tropic strain HIV-lBa-L. HIV-1 proviral DNA content andp24 production after 1 week were then determined. As shownin Fig. 3, the AZTp2AZT-loaded RBC inhibited HIV-1 pro-viral formation by 93-97% (Fig. 3A, range of three experi-ments) and p24 production by 80% (Fig. 3B). AZT at 0.1 ,uMin culture medium for 20 h had no effect on viral infectivity andproduction, while AZT at 0.1 /LM maintained in culture medium

13A

L-

N

+

2

I-i

a

4.

J"3

2DlS

ci

GJ>

6

r,

[IIV-1115 bp 4

c..lc-r4-

Inf'ected I - I + 1i- AZT I AZT( 1) Unloaded AZT-dimecr (120 hrs) (1 week)

IBC

FIG. 3. Inhibition of HIV-1 proviral DNA formation (A) and p24 production (B) in human macrophages infected with HIVBa-L and treatedwith AZTp2AZT-loaded RBC. Macrophages were cultured for 10 days; then untreated, treated with AZTp2AZT-loaded RBC, unloaded RBC,or 0.1 /M AZT; and then infected with a macrophage-tropic HIVBa-L for 5 h. The cells were then extensively washed and cultured at 37°C and5% CO2 for 1 week for the detection of proviral DNA formation and for the determination of p24 production using a commercial ELISA kit(DuPont). When AZT was present for the duration of the experiment, this was added in culture medium. A 115-bp gag region of HIV-1 genomewas amplified by primers SK38/39 from 400 ng of genomic DNA. Positive control was a plasmid DNA containing 300 copies of the genome ofHIVZ6 isolate (Perkin-Elmer/Cetus). Negative control was DNA from human placenta (4.2 ng). Detection was with a 32P-end-labeled SK19 probeaccording to the instructions of the manufacturer (Perkin-Elmer Gene Amplimer HIV-1 reagents). Values are mean ± SD of three experiments.Simultaneous addition of unloaded RBC and 0.1 ,M AZT for 20 h provided results similar to the addition of 0.1 ,/M AZT alone for 20 h (notshown).

Proc. Natl. Acad. Sci. USA 93 (1996)

C;

1

I,

.J

Proc. Natl. Acad. Sci. USA 93 (1996) 4407

for all the duration of the experiment inhibited p24 production by58%.

Similar experiments were performed with feline macro-phages that were infected with FIV. Again, inhibition ofproviral DNA formation was 95% ± 2% (three experiments)(Fig. 4 Left). We next tested the ability of AZTp2AZT-loadedRBC to protect peritoneal murine macrophages from a murineretrovirus that causes a disease known as murine AIDS. Asshown in Fig. 4 Right, inhibition of proviral BM5d DNAformation was 71% ± 1.2%.

DISCUSSIONThe development of new therapeutic systems against AIDS,based on nucleoside analogues and designed to achieve anti-retroviral activity in specific cell types like monocytes/macrophages, requires a combination of different features. (i)Pharmacologically active triphosphorylated nucleoside ana-logues can display unacceptable cell toxicity due to inhibitionof enzyme targets in the host cell other than viral reversetranscriptase (18, 19). Therefore, ddCTP, AZT-TP, and sim-ilar compounds cannot be safely delivered as such to infectedcells. (ii) Accordingly, triphosphorylated nucleoside analoguesshould be formed in the desired cell type at low concentrationssuch as to inhibit viral reverse transcriptase selectively; there-fore, metabolically suitable prodrugs (21, 27-29), structurallytailored to be converted by endogenous enzyme systems, arenecessary. (iii) Targeting of such prodrugs to macrophages,where their bioconversion is required, can be greatly favoredby exploiting the natural property of erythrophagocytosis (14).(iv) The selected prodrugs should be stable enough in thecarrier RBC to be metabolically converted to the nucleosidetriphosphate analogues within infected macrophages only.

All of the above requirements seem to be fulfilled by thehomodinucleotide of azidothymidine synthesized and charac-terized in this study. Although a dinucleotide pyrophosphataseis present in human red cells, its activity seems to be rather low,probably because of its susceptibility to ATP inhibition (30), as

I

-

r oC:Ce-i-

cra;-I

4-

FIV498 bp " _. U

shown by a very poor turnover of NAD+/NADH andNADP+/NADPH in intact RBC (31). On the other hand,AZTp2AZT, while being stable in the loaded red cells, disap-pears in the macrophages within time intervals that certainlyexceed those required to complete erythrophagocytosis (Fig.2). This property is consistent with metabolism ofAZTp2AZTlargely occurring in macrophages. That such metabolism isrelated to bioactivation-i.e., that AZTp2AZT behaves as a

prodrug in macrophages-is demonstrated by the anti-retroviral activity of this system with three different viralmodels including HIV (Figs. 3 and 4) and by poor efficiencyof AZTp2AZT itself as a reverse transcriptase inhibitor. Thelatter finding might suggest that AZTp2AZT bioactivationinvolves the intracellular synthesis of AZT-TP.Homo- and heterodimers of nucleoside analogues have been

synthesized in different laboratories (32-35). However, thesedimers were nonhydrolyzable because of the presence of amethylene spacer (35) or due to a methylphosphonate linkage(32-34). In some reports (33, 34), the nucleotide dimers werelinked by a 5',5'-phosphate bound and displayed antiviralactivity similar to or better than the individual nucleosideanalogues. However, it is not clear whether these dimersfunction as such or following conversion to the correspondingmoieties since the cleavage of the phosphodiester bond is verylikely to take place. In these studies, the presence of intactdimer in cells was never demonstrated.

Cell-targeted synthesis of pharmacologically active deriva-tives ofAZT seems to be optimal for display of anti-retroviralactivity, even starting from quite low intracellular concentra-tions of the AZTp2AZT prodrug in macrophages and therebypreventing cell toxicity. Therefore, the RBC-based system oftargeted delivery of precursors of triphosphorylated nucleo-side analogues in virus-infected macrophages proves to be apromising way to inhibit HIV replication in these cells. Theversatility of the chemical procedure for synthesizingAZTp2AZT should allow extension to other homodinucleo-tides (e.g., of ddC, ddA, and ddI) or heterodinucleotides

L.

i+-

+

LP-BM5/d141 bp 4

la

cIdei

m

xaiI-r-0

I-

:r .~ _ F

Feline _ _ _hexokinase 4 4_B̂ lk41

G(6PD203 bp 4 A AD1O/-

FIG. 4. PCR analysis of FIV and BM5d proviral DNA in feline macrophages (Left) and murine macrophage (Right) treated withAZTp2AZT-loaded RBC. Feline macrophages were infected with 330 i.d. per well of FIV (Pisa M-2) for 8 h, then washed six times to remove anyviral particle associated with macrophages, and cultured for 4 days at 37°C and 5% CO2. For detection of FIV proviral DNA, 498 bp of the viralp24 gag gene was amplified in two stages by specific primers, as described. As an internal control, a feline hexokinase gene was also amplified.Detection was with a 32P-labeled 498-bp probe for FIV or a specific feline hexokinase. Murine macrophages from the peritoneal cavity of C57BL/6mice were infected and cultured as described for FIV. A total of 141 bp of the viral pl2 gag gene was amplified by specific primers (as reportedin ref. 17) from 40 ng of genomic DNA. As an internal control, 203 bp of the G6PD gene was amplified. Detection was with a 32P-labeled D30probe for BM5d or a specific G6PD probe. Inhibition of FIV proviral DNA formation was 95% ± 2.0%, and inhibition of BM5d proviral DNAformation was 71% ± 1.2%. These values were obtained by laser scan densitometry of the autoradiographic films and are mean ± SD of threeexperiments. The partial inhibition of proviral DNA formation by unloaded RBC is due to nonspecific macrophage activation, as previouslydocumented (21).

Medical Sciences: Magnani et aL.

4408 Medical Sciences: Magnani et al.

carrying nucleoside analogues of comparable anti-retroviralactivity.

We thank Dr. P. Jolicoeur (Montreal) for the LP-BM5 D30 probe,Dr. R. Yetter (Baltimore) for providing the LP-BM5 virus, and Dr. M.Bendinelli (Pisa, Italy) for providing the FIV. This work was partiallysupported by Ministero della Saniti, Istituto Superiore di Sanita,Rome, progetto AIDS 1995 and Gruppo Lepetit S.p.A., Lainate (MI),Italy.

1. Gartner, S., Markovits, P., Markovits, D. M., Kaplan, M. H.,Gallo, R. C. & Popovic, M. (1986) Science 233, 215-219.

2. Ho, D. D., Rota, T. R. & Hirsch, M. S. (1986) J. Clin. Invest. 77,1712-1715.

3. Koenig, S., Gendelman, H. E., Orenstein, J. M., Dal Canto,M. C., Pezeshjpour, G. H., Yungbluth, M., Janotta, F., Aksamit,A., Martin, M. A. & Fauci, A. S. (1986) Science 233, 1089-1093.

4. Tschachler, E., Groh, V., Popovic, M., Mann, D. L., Konrad, K.,Safai, B., Eron, L., diMarzo Veronese, L., Wolff, K. & Stingl, G.(1987) J. Invest. Dermatol. 88, 233-237.

5. Armstrong, J. A. & Home, R. (1984) Lancet ii, 370-372.6. Salahuddin, S. Z., Rose, R. M., Groopman, J. E., Markham, P. D.

& Gallo, R. C. (1986) Blood 68, 281-284.7. Mitsuja, H., Yarchoan, R. & Broder, S. (1990) Science 249,

1533-1544.8. De Clercq, E. (1991) J. AIDS 4, 207-218.9. Gao, W., Cara, A., Gallo, R. & Lori, F. (1993) Proc. Natl. Acad.

Sci. USA 90, 8925-8928.10. Gao, W., Agbaria, R., Driscoll, J. S. & Mitsuja, H. (1994) J. Biol.

Chem. 269, 12633-12638.11. Perno, C. F., Yarchoan, R., Cooney, D.A., Hartman, N. R.,

Gartner, S., Popovic, M., Hao, Z., Gerrard, T. L., Wilson, Y. A.,Johns, D. G. & Broder, S. (1988) J. Exp. Med. 168, 1111-1125.

12. Richmann, D. D., Kombluth, R. S. & Carson, D. A. (1987)J. Exp.Med. 166, 1144-1149.

13. Pemo, C.F., Yarchoan, R., Cooney, D.A., Hartman, N. R.,Webb, D. S. A., Hao, Z., Mitsuya, H., Johns, D. G. & Broder, S.(1989) J. Exp. Med. 169, 933-951.

14. Magnani, M., Rossi, L., Brandi, G., Schiavano, G. F., Montroni,M. & Piedimonte, G. (1992) Proc. Natl. Acad. Sci. USA 89,6477-6481.

15. Magnani, M., Rossi, L., Fratemale, A., Silvotti, L., Quintavalla,F., Piedimonte, G., Matteucci, D., Baldinotti, F. & Bendinelli, M.(1994) AIDS Res. Hum. Retroviruses 10, 1179-1186.

16. Rossi, L., Brandi, G., Fraternale, A., Schiavano, G. F., Chiaran-tini, L. & Magnani, M. (1993) J. AIDS 6, 1179-1186.

17. Magnani, M., Casabianca, A., Rossi, L., Fraternale, A., Brandi,G., Silvotti, L. & Piedimonte, G. (1995) Antiviral Chem. Che-mother. 6, 312-319.

18. Parker, W. B. & Cheng, Y. C. (1994) J. NIH Res. 6, 57-61.19. Lewis, W. & Dalakas, M. C. (1995) Nat. Med. 1, 417-422.20. Mosier, D. E., Yetter, R. A. & Morse, H. C., III (1987) J. Exp.

Med. 165, 1737-1742.21. Piedimonte, G., Montroni, M., Silvestri, G., Silvotti, L., Donatini,

A., Rossi, L., Borghetti, A. F. & Magnani, M. (1992)Arch. Virol.130, 463-469.

22. Magnani, M., Giovine, M., Fratemale, A., Damonte, G., Rossi,L., Scarfi, S., Benatti, U. & De Flora, A. (1995) Drug Delivery 2,57-61.

23. Michelson, A. M. (1964) Biochim. Biophys. Acta 91, 1-13.24. Magnani, M., Rossi, L., Bianchi, M., Cucchiarini, L. & Stocchi,

V. (1989) J. Chromatogr. 491, 215-220.25. Pauwels, R., Balzarini, J., Baba, M., Snoeck, R., Schols, D.,

Herdwijn, P., Desmyter, J. & De Clercq, E. (1988) J. Virol.Methods 20, 309-321.

26. Bendinelli, M., Pistello, M., Lombardi, S., Poli, A., Garzelli, C.,Matteucci, D., Ceccherini-Nelli, L., Malvoldi, G. & Tozzini, F.(1995) Clin. Microbiol. Rev. 8, 87-112.

27. De Flora, A., Zocchi, E., Guida, L., Polvani, C. & Benatti, U.(1988) Proc. Natl. Acad. Sci. USA 85, 3145-3149.

28. Magnani, M., Bianchi, M., Rossi, L. & Stocchi, V. (1989)Biochem. Biophys. Res. Commun. 164, 446-452.

29. Gasparini, A., Giovine, M., Damonte, G., Tonetti, M., Grandi, T.,Mazzei, M., Balbi, A., Silvestro, L., Benatti, U. & De Flora, A.(1994) Biochem. Pharmacol. 48, 1121-1128.

30. Zocchi, E., Guida, L., Franco, L., Guerrini, M., Benatti, U. & DeFlora, A. (1993) Biochem. J. 295, 121-130.

31. Jaffe, E. R. & Neuman, G. (1965) Transfusion 5, 412-420.32. Busso, M., Mian, A. M., Hahn, A. & Resnick, L. (1988)AIDSRes.

Hum. Retroviruses 4, 449-455.33. Puech, F., Gosselin, G., Balzarini, J., Good, S. S., Rideout, J. L.,

De Clerq, E. & Imbach, J. L. (1990) Antiviral Res. 12, 11-23.34. Schinazi, R. F., Sommadossi, J. P., Saalmann, V., Cannon, D. L.,

Xia, M. Y., Hart, G. C., Smith, G. A. & Hahn, E. F. (1990)Antimicrob. Agents Chemother. 34, 1061-1067.

35. Velazquez, S., Alvarez, R., San-felix, A., Jimeno, M. L., De Clerq,E., Balzarini, J. & Camarasa, M. J. (1995) J. Med. Chem. 38,1641-1649.

Proc. Natl. Acad. Sci. USA 93 (1996)