Controlled infection of Poecilia reticulata Peters (guppy) with Tetrahymena by immersion and intraperitoneal injection.

Tetrahymena is a protozoan parasite, which infects guppy, Poecilia reticulata Peters, and causes substantial economical losses in commercial farms worldwide. Studies of guppy infected by Tetrahymena require standardized infection protocols. The LD50 for Tetrahymena infection of guppies by intraperitoneal (IP) injection was calibrated, and the level obtained was 946 parasites per fish. Guppy infection with Tetrahymena by immersion, imitating the natural route of infection via the integument, was studied under normal or stress conditions. Exposure to cold and netting (CNI) and to cold only (CI) followed by immersion exposure to 10 000 Tetrahymena per mL resulted in 22.5% and 19.2% mortality, respectively, as compared to 14.2% and 10% in groups that were netted only (NI) or non-stressed (I). Histopathology revealed that immersion infection resulted in a systemic infection. Lysozyme levels, measured 3 weeks after infection, were significantly higher in the CNI group (288 µg per mg protein) compared with CI-, NI- and I-treated groups (94.5, 64 and 62.3 µg mg(-1), respectively). There was no evident parasite immobilization activity in body homogenates, suggesting no development of acquired immunity. Re-infection by IP injection revealed no increase in protection in any of the treatment groups, mortality range of 56.3-75%, higher than in the non-exposed control (40.6% mortality).

Attenuation of colon inflammation through activators of the retinoid X receptor (RXR)/peroxisome proliferator-activated receptor gamma (PPARgamma) heterodimer. A basis for new therapeutic strategies.

The peroxisome proliferator-activated receptor gamma (PPARgamma) is highly expressed in the colon mucosa and its activation has been reported to protect against colitis. We studied the involvement of PPARgamma and its heterodimeric partner, the retinoid X receptor (RXR) in intestinal inflammatory responses. PPARgamma(1/)- and RXRalpha(1/)- mice both displayed a significantly enhanced susceptibility to 2,4,6-trinitrobenzene sulfonic acid (TNBS)-induced colitis compared with their wild-type littermates. A role for the RXR/PPARgamma heterodimer in the protection against colon inflammation was explored by the use of selective RXR and PPARgamma agonists. TNBS-induced colitis was significantly reduced by the administration of both PPARgamma and RXR agonists. This beneficial effect was reflected by increased survival rates, an improvement of macroscopic and histologic scores, a decrease in tumor necrosis factor alpha and interleukin 1beta mRNA levels, a diminished myeloperoxidase concentration, and reduction of nuclear factor kappaB DNA binding activity, c-Jun NH(2)-terminal kinase, and p38 activities in the colon. When coadministered, a significant synergistic effect of PPARgamma and RXR ligands was observed. In combination, these data demonstrate that activation of the RXR/PPARgamma heterodimer protects against colon inflammation and suggest that combination therapy with both RXR and PPARgamma ligands might hold promise in the clinic due to their synergistic effects.

Treatment development for systemic Tetrahymena sp. infection in guppies, Poecilia reticulata Peters.

Antibacterial and antiparasitic agents and a cysteine protease inhibitor (E-64) were tested against Tetrahymena infection, a serious problem in guppy production worldwide. Chemicals were tested in vitro by a colorimetric assay for Tetrahymena survival. The most effective were niclosamide, albendazole and chloroquine, with 23%, 35% and 60% survival, respectively, following 2-h exposure to 100 ppm. Longer incubation periods resulted in greater reductions in survival. Niclosamide was further studied in vivo at different dosages, administered orally to Tetrahymena-infected guppies. Mortality rates were significantly lower in all treatment groups; in trial I, 30% and 33% mortality in 5 and 40 mg kg(-1) niclosamide-fed fish vs. 59% mortality in controls; in trial II, 35%, 13% and 10% in 50, 100 and 200 mg kg(-1) niclosamide-fed fish vs. 64% in controls. The effect of the cysteine protease inhibitor E64 was tested in tissue culture, by measuring histolytic activity of the parasite (Tet-NI) on a guppy-fin cell line, based on cell depletion. Tet-NI feeding activity was significantly reduced following pretreatment with E-64 relative to non-treated Tet-NI. E-64-pretreated Tet-NI was injected i.p. into guppies: recorded mortality rates were significantly lower (35%) than that in non-treated Tet-NI (60%), suggesting inhibition of the parasite's cysteine protease as a possible therapeutic approach.

Protective immunization against Tetrahymena sp. infection in guppies (Poecilia reticulata).

Systemic tetrahymenosis constitutes a serious problem in guppy (Poecilia reticulata) production worldwide and no therapeutic solution is available for this disease. Three immunization trials were conducted, testing the effectiveness of different Tetrahymena preparations applied by intraperitoneal injection (IP) with or without Freund's complete adjuvant (FCA) and with or without booster dose. In trial 1, immunization with the pathogenic Tet-NI 6 lysate and live attenuated Tet-NI 1 did not provide significant protection from infection, although infection rates were significantly lower in the Tet-NI 6-immunized group than in controls. In trial 2, mortality in Tet-NI 6 + FCA-immunized fish was 10%, significantly lower than in all other treatment groups, including Tet-NI 6 lysate, live attenuated Tet-NI 1 and controls (77, 67 and 73%, respectively). In trial 3, the lowest mortality rates were obtained in the Tet-NI 6 + FCA + booster-immunized group (15%). These levels were lower but not significantly different from the non-boostered Tet-NI 6-immunized group (28%) and the groups immunized with Tet-NI 1, with and without booster (32 and 34%, respectively). Mortality in these four groups was significantly lower than in controls, including adjuvant- and PBS-injected groups (72 and 81%, respectively). Body homogenates of immunized fish immobilized Tetrahymena in-vitro, as compared to no or very little immobilization in controls. Lysozyme levels in the Tet-NI 6 + FCA + booster group were significantly higher than in all other treatments in trial 2 and controls in trial 3. There was no significant difference in anti-protease activity among the differently immunized fish. We conclude that immunization with Tetrahymena lysates in FCA confers a high degree of protection from infection, suggesting this preparation as a basis for vaccine development.

Tetrahymena sp. infection in guppies, Poecilia reticulata Peters: parasite characterization and pathology of infected fish.

Tetrahymena sp. infection was diagnosed in guppies imported from Singapore. The parasite was isolated (Tet-NI) and optimally cultured in vitro in RM-9 medium. Cytological analyses [silver-staining and scanning electron microscopy (SEM)] revealed a pyriform-shaped, 64 x 41-microm holotrich ciliate without caudal cilium, containing a macro-nucleus (18.25 x 16.83 microm) and micro-nucleus (5.73 x 5.40 microm). Wet-mount examination and histological analyses of fish exposed to the parasite by co-habitation, immersion and infection by i.p. (intra-peritoneal) and i.m. (intra-muscular) injection revealed numerous ciliates on the skin, and in the gill and caudal fin blood vessels. Ciliates surrounded internal organs, the peri-orbital region of the eye, and were observed inside developing guppy embryos. Some muscle necrosis was associated with infection, but little or no inflammatory response. Immersion, co-habitation and i.m. injection caused relatively high infection rates and levels in the skin and tail, and lower infection in the gill blood vessels and internal organs; i.p. injection caused higher infection in the gill blood vessels and internal organs. Co-habited fish had relatively high infection levels in the hind-gut sub-mucosa. This is the first report of controlled systemic infection by Tetrahymena sp.

Agonists that suppress M-current elicit phosphoinositide turnover and Ca2+ transients, but these events do not explain M-current suppression.

The hypothesis that acetylcholine, substance P, and LHRH suppress M-current by activating phospholipase C was tested. Each agonist caused turnover of phosphoinositide, as measured by release of inositol phosphates, and a modest transient rise in intracellular free Ca2+ ([ Ca2+]i), as determined with fura-2. Active phorbol esters depressed M-current only 50% and did not prevent further suppression by LHRH. M-current, its control by agonists, and its depression by phorbol esters were not affected by adding inositol trisphosphate or Ca2+ buffers with high or low Ca2+ to the whole-cell, voltage-clamp pipette. We conclude that phospholipase C activation does occur but does not mediate the suppression of M-current by agonists. Caffeine produced large [Ca2+]i transients and acted as an agonist to suppress M-current.

Low penetration of oseltamivir and its carboxylate into cerebrospinal fluid in healthy Japanese and Caucasian volunteers.

Oseltamivir is a potent, well-tolerated antiviral for the treatment and prophylaxis of influenza. Although no relationship with treatment could be demonstrated, recent reports of abnormal behavior in young individuals with influenza who were receiving oseltamivir have generated renewed interest in the central nervous system (CNS) tolerability of oseltamivir. This single-center, open-label study explored the pharmacokinetics of oseltamivir and oseltamivir carboxylate (OC) in the plasma and cerebrospinal fluid (CSF) of healthy adult volunteers over a 24-hour interval to determine the CNS penetration of both these compounds. Four Japanese and four Caucasian males were enrolled in the study. Oseltamivir and OC concentrations in CSF were low (mean of observed maximum concentrations [C(max)], 2.4 ng/ml [oseltamivir] and 19.0 ng/ml [OC]) versus those in plasma (mean C(max), 115 ng/ml [oseltamivir] and 544 ng/ml [OC]), with corresponding C(max) CSF/plasma ratios of 2.1% (oseltamivir) and 3.5% (OC). Overall exposure to oseltamivir and OC in CSF was also comparatively low versus that in plasma (mean area under the concentration-time curve CSF/plasma ratio, 2.4% [oseltamivir] and 2.9% [OC]). No gross differences in the pharmacokinetics of oseltamivir or OC were observed between the Japanese and Caucasian subjects. Oseltamivir was well tolerated. This demonstrates that the CNS penetration of oseltamivir and OC is low in Japanese and Caucasian adults. Emerging data support the idea that oseltamivir and OC have limited potential to induce or exacerbate CNS adverse events in individuals with influenza. A disease- rather than drug-related effect appears likely.

Multiple L double-stranded RNA species of Saccharomyces cerevisiae: evidence for separate encapsidation.

The L double-stranded (ds) RNA component of Saccharomyces cerevisiae may contain up to three dsRNA species, each with a distinct sequence but with identical molecular weights. These dsRNAs have been separated from each other by denaturation and polyacrylamide gel electrophoresis. The 3' terminal sequences of the major species, LA dsRNA, were determined. Secondary structural analysis supported the presence of two stem and loop structures at the 3' terminus of the LA positive strand. In strain T132B NK-3, both the LA and LC species are virion encapsidated. Two distinct classes of virions were purified from this strain, each with a different RNA polymerase activity and with distinct protein components. The heavy virions harbored LA dsRNA, whereas the LC dsRNA species co purified with the light virion peak. Thus, LA and LC dsRNAs, when present in the same cell, may be separately encapsidated.

Saccharomyces cerevisiae killer virus transcripts contain template-coded polyadenylate tracts.

The M double-stranded RNA component of type 1 killer strains of the yeast Saccharomyces cerevisiae contains an internal 200-base pair adenine- and uracil-rich region. The plus strands of this viral genomic RNA contain an internal adenine-rich region which allows these strands to bind to polyuridylate-Sepharose as tightly as do polyadenylated RNAs with 3'-terminal polyadenylated tracts of 70 to 100 residues. Internal template coding of an adenine-rich tract in positive polarity in vivo and in vitro transcripts of M double-stranded RNA may serve as an alternate method of transcript polyadenylation. The 3'-terminal residue of the in vitro m transcript is a non-template-encoded purine residue. The 5' terminus of this transcript is involved in a stem-and-loop structure which includes an AUG initiation codon, along with potential 18S and 5.8S rRNA binding sites. Except for the 3'-terminal residue, transcription in in vitro shows complete fidelity.

Folding of the group I intron ribozyme from the 26S rRNA gene of Candida albicans.

Preincubation of the group I intron Ca.LSU from Candida albicans at 37 degrees C in the absence of divalent cations results in partial folding of this intron. This is indicated by increased resistance to T1 ribonuclease cleavage of many G residues in most local helices, including P4-P6, as well as the non-local helix P7, where the G binding site is located. These changes correlate with increased gel mobility and activation of catalysis by precursor RNA containing this intron after preincubation. The presence of divalent cations or spermidine during preincubation results in formation of the predicted helices, as indicated by protection of additional G residues. However, addition of these cations during preincubation of the precursor RNA alters its gel mobility and eliminates the preincubation activation of precursor RNA seen in the absence of cations. These results suggest that, in the presence of divalent cations or spermidine, Ca.LSU folds into a more ordered, stable but misfolded conformation that is less able to convert into the catalytically active form than the ribozyme preincubated without cations. These results indicate that, like the group I intron of Tetrahymena, multiple folding pathways exist for Ca.LSU. However, it appears that the role cations play in the multiple folding pathways leading to the catalytically active form may differ between folding of these two group I introns.

A novel insulin secretagogue is a phosphodiesterase inhibitor.

The arylpiperazine L-686,398 was described as an oral hypoglycemic agent and is shown to be an insulin secretagogue in vitro. The characteristics of its activity were similar to those of the incretin glucagon-like peptide I (GLP-I). We demonstrate that both the peptide and L-686,398 increase the accumulation of cAMP in isolated ob/ob mouse pancreatic islet cells, but by different mechanisms. Although GLP-I activates adenylate cyclase, the arylpiperazine has no effect on this enzyme or on the binding of 125I-labeled GLP-I to its receptor on RINm5F rat insulinoma cell membranes. However, L-686,398 inhibits the total cAMP phosphodiesterase (PDE) activity in homogenates of ob/ob mouse pancreatic islets with an EC50 of approximately 50 mumol/l. To determine the mechanism of PDE inhibition by the arylpiperazine and to examine its specificity, we studied the kinetics of arylpiperazine inhibition of two recombinant PDEs. The arylpiperazine is a competitive inhibitor of both a human heart type III PDE and a rat type IV-D PDE. Inhibition of the type III and IV isozymes are characterized by Ki values of 27 and 5 mumol/l, respectively. Although not extremely potent, the arylpiperazine does exhibit modest selectivity between these PDEs. The observation that L-686,398 acts as a PDE inhibitor suggests that exploration for beta-cell-specific PDE isoforms may reveal novel PDEs as targets for the development of therapeutically useful glucose-dependent insulin secretagogues.

Voltage-dependent gating of veratridine-modified Na channels.

Na channels of frog muscle fibers treated with 100 microM veratridine became transiently modified after a train of repetitive depolarizations. They open and close reversibly with a gating process whose midpoint lies 93 mV more negative than the midpoint of normal activation gating and whose time course shows no appreciable delay in the opening or closing kinetics but still requires more than two kinetic states. Like normal activation, the voltage dependence of the modified gating can be shifted by changing the bathing Ca2+ concentration. The instantaneous current-voltage relation of veratridine-modified channels is curved at potentials negative to -90 mV, as if external Ca ions produced a voltage-dependent block but also permeated. Modified channels probably carry less current than normal ones. When the concentration of veratridine is varied between 5 and 100 microM, the initial rate of modification during a pulse train is directly proportional to the concentration, while the rate of recovery from modification after the train is unaffected. These are the properties expected if drug binding and modification of channels can be equated. Hyperpolarizations that close modified channels slow unbinding. Allethrin and DDT also modify channels. They bind and unbind far faster than veratridine does, and their binding requires open channels.

Isolation of myocardial L-type calcium channel gating currents with the spider toxin omega-Aga-IIIA.

The peptide omega-agatoxin-IIIA (omega-Aga-IIIA) blocks ionic current through L-type Ca channels in guinea pig atrial cells without affecting the associated gating currents. omega-Aga-IIIA permits the study of L-type Ca channel ionic and gating currents under nearly identical ionic conditions. Under conditions that isolate L-type Ca channel currents, omega-Aga-IIIA blocks all ionic current during a test pulse and after repolarization. This block reveals intramembrane charge movements of equal magnitude and opposite sign at the beginning of the pulse (Q(on)) and after repolarization (Q(off)). Q(on) and Q(off) are suppressed by 1 microM felodipine, saturate with increasing test potential, and are insensitive to Cd. The decay of the transient current associated with Q(on) is composed of fast and slow exponential components. The slow component has a time constant similar to that for activation of L-type Ca channel ionic current, over a broad voltage range. The current associated with Q(off) decays monoexponentially and more slowly than ionic current. Similar charge movements are found in guinea pig tracheal myocytes, which lack Na channels and T-type Ca channels. The kinetic and pharmacological properties of Q(on) and Q(off) indicate that they reflect gating currents associated with L-type Ca channels. omega-Aga-IIIA has no effect on gating currents when ionic current is eliminated by stepping to the reversal potential for Ca or by Cd block. Gating currents constitute a significant component of total current when physiological concentrations of Ca are present and they obscure the activation and deactivation of L-type Ca channels. By using omega-Aga-IIIA, we resolve the entire time course of L-type Ca channel ionic and gating currents. We also show that L- and T-type Ca channel ionic currents can be accurately quantified by tail current analysis once gating currents are taken into account.

Electrophysiological comparison of insecticide and alkaloid agonists of Na channels.

Macroscopic currents in Na channels were recorded from adult frog skeletal muscle under voltage clamp as various toxins were added to the bathing medium. Veratridine, cevadine, and 3-(4-ethoxybenzoyl)-veracevine modified the Na channels in a use-dependent manner during depolarizations and held them open for 3, 2.4, and 1.2 s, respectively, at -90 mV. The three alkaloids modified channels in the same way. Activation gating was shifted about -100 mV by the modification, and reversible closing of the channels by strong hyperpolarizations slowed reversal of the modification. The synthetic insecticides deltamethrin, EDO, GH739, and GH414 also modified channels during depolarizations that opened channels. The modification lasted 3 s with deltamethrin, but only 3-5 ms with the others. Hyperpolarization speeded the shutting off of current in insecticide-modified channels, but no reversible activation gating could be demonstrated. The ionic selectivity, PNa/PNH4, of channels was decreased by all of the toxins. This ratio was 0.11 in normal channels, 0.26 in insecticide-modified channels, and 0.7-1.6 in veratrum-alkaloid-modified channels. During use-dependent modification, the veratrum alkaloids reduced the total Na current markedly, while deltamethrin did not. Thus, alkaloid and insecticide modifications share many features but differ in how much the conducting properties of the pore are changed and whether the channel can close reversibly while the toxin remains bound.

Dominant chromosomal mutation bypassing chromosomal genes needed for killer RNA plasmid replication in yeast.

Yeast strains carrying a double-stranded RNA plasmid of 1.4-1.7 x 10(6) daltons encapsulated in virus-like particles secrete a toxin that kills strains lacking this plasmid. The plasmid requires at least 24 chromosomal genes (pets, and mak1 through mak23) for its replication or maintenance. We have detected dominant Mendelian mutations (called KRB1 for killer replication bypass) that bypass two chromosomal genes, mak7 and pets, normally needed for plasmid replication. Strains mutant in mak7 and carrying the bypass mutation (mak7-1 KRB1) are isolated as frequent K(+)R(+) sectors of predominantly K(-)R( -) segregants from crosses of mak7-1 with a wild-type killer. All KRB1 mutations isolated in this way are inherited as single dominant centromere-linked chromosomal changes. They define a new centromere. KRB1 is not a translational suppressor. KRB1 strains contain a genetically normal killer plasmid and ds RNA species approximately the same in size and amount as do wild-type killers. Bypass of both mak7 and pets by one mutation suggests that these two genes are functionally related. Two properties of the inheritance of KRB1 indicate an unusually high reversion frequency: (1) Heat or cycloheximide (treatments known to cure strains of the wild-type killer plasmid) readily induce conversion of mak7-1 KRB1 strains from killers to nonkillers with concomitant disappearance of KRB1 as judged by further crosses, and (2) mating two strains of the type mak7-1 KRB1 with each other yields mostly 2 K(+)R(+): 2 K(-)R(-) segregation, although the same KRB1 mutation and the same killer plasmid are present in both parents.

Two chromosomal genes required for killing expression in killer strains of Saccharomyces cerevisiae.

The killer character of yeast is determined by a 1.4 X 10(6) molecular weight double-stranded RNA plasmid and at least 12 chromosomal genes. Wild-type strains of yeast that carry this plasmid (killers) secret a toxin which is lethal only to strains not carrying this plasmid (sensitives).--We have isolated 28 independent recessive chromosomal mutants of a killer strain that have lost the ability to secrete an active toxin but remain resistant to the effects of the toxin and continue to carry the complete cytoplasmic killer genome. These mutants define two complementation groups, kex1 and kex2. Kex1 is located on chromosome VII between ade5 and lys5. Kex2 is located on chromosome XIV, but it does not show meiotic linkage to any gene previously located on this chromosome.--When the killer plasmid of kex1 or kex2 strains is eliminated by curing with heat or cycloheximide, the strains become sensitive to killing. The mutant phenotype reappears among the meiotic segregants in a cross with a normal killer. Thus, the kex phenotype does not require an alteration of the killer plasmid.--Kex1 and kex2 strains each contain near-normal levels of the 1.4 x 10(6) molecular weight double-stranded RNA, whose presence is correlated with the presence of the killer genome.

The [KIL-d] cytoplasmic genetic element of yeast results in epigenetic regulation of viral M double-stranded RNA gene expression.

[KIL-d] is a cytoplasmically inherited genetic trait that causes killer virus-infected cells of Saccharomyces cerevisiae to express the normal killer phenotypes in a/alpha cells, but to show variegated defective killer phenotypes in a or alpha type cells. Mating of [KIL-d] haploids results in "healing" of their phenotypic defects, while meiosis of the resulting diploids results in "resetting" of the variegated, but mitotically stable, defects. We show that [KIL-d] does not reside on the double-stranded RNA genome of killer virus. Thus, the [KIL-d] effect on viral gene expression is epigenetic in nature. Resetting requires nuclear events of meiosis, since [KIL-d] can be cytoplasmically transmitted during cytoduction without causing defects in killer virus expression. Subsequently, mating of these cytoductants followed by meiosis generates spore clones expressing variegated defective phenotypes. Cytoduction of wild-type cytoplasm into a phenotypically defective [KIL-d] haploid fails to heal, nor does simultaneous or sequential expression of both MAT alleles cause healing. Thus, healing is not triggered by the appearance of heterozygosity at the MAT locus, but rather requires the nuclear fusion events which occur during mating. Therefore, [KIL-d] appears to interact with the nucleus in order to exert its effects on gene expression by the killer virus RNA genome.

The [KIL-d] element specifically regulates viral gene expression in yeast.

The cytoplasmically inherited [KIL-d] element epigenetically regulates killer virus gene expression in Saccharomyces cerevisiae. [KIL-d] results in variegated defects in expression of the M double-stranded RNA viral segment in haploid cells that are "healed" in diploids. We report that the [KIL-d] element is spontaneously lost with a frequency of 10(-4)-10(-5) and reappears with variegated phenotypic expression with a frequency of > or =10(-3). This high rate of loss and higher rate of reappearance is unlike any known nucleic acid replicon but resembles the behavior of yeast prions. However, [KIL-d] is distinct from the known yeast prions in its relative guanidinium hydrochloride incurability and independence of Hsp104 protein for its maintenance. Despite its transmissibility by successive cytoplasmic transfers, multiple cytoplasmic nucleic acids have been proven not to carry the [KIL-d] trait. [KIL-d] epigenetically regulates the expression of the M double-stranded RNA satellite virus genome, but fails to alter the expression of M cDNA. This specificity remained even after a cycle of mating and meiosis. Due to its unique genetic properties and viral RNA specificity, [KIL-d] represents a new type of genetic element that interacts with a viral RNA genome.

Relevance of DMS-III depressive subtype and chronicity of antidepressant efficacy in atypical depression. Differential response to phenelzine, imipramine, and placebo.

One hundred ninety-four nonmelancholic depressed outpatients with features of atypical depression took part in a 6-week randomized trial of imipramine hydrochloride, phenelzine sulfate, and placebo. Their courses of illness were also rated for chronicity. Significantly more patients responded to phenelzine (71%) than to imipramine (48%), which benefited significantly more patients than placebo (26%). Both chronicity and DMS-III diagnosis predicted response on several outcome measures. For example, patients with dysthymic disorder responded better to treatment than did those with major depression, suggesting that dysthymic disorder can be treated with medication. Placebo response correlated inversely with chronicity, regardless of DMS-III diagnosis. Atypical depression and longitudinal course of illness may add to the usefulness of DMS-III depressive diagnosis as a predictor of antidepressant response.

Sequencing of protease inhibitor therapy: insights from an analysis of HIV phenotypic resistance in patients failing protease inhibitors.

OBJECTIVE: To characterize the pattern of HIV-1 susceptibility to protease inhibitors in patients failing an initial protease inhibitor-containing regimen. DESIGN: A cross-sectional analysis of antiretroviral susceptibility. SETTING: HIV clinics in six metropolitan areas. PATIENTS: Eighty-eight HIV-infected adults with HIV RNA > 400 copies/ml after > or = 6 months of antiretroviral therapy, including the use of one protease inhibitor for > or = 3 months. MEASUREMENTS: The frequency and magnitude of decreased susceptibility, measured with a phenotypic assay using recombinant constructs, to five protease inhibitors. Decreased susceptibility was defined as > 2.5-fold increase in the 50% inhibitory concentration (IC50) compared with drug sensitive control virus. RESULTS: At study entry, patients were being treated with nelfinavir (63%), indinavir (25%), or another protease inhibitor (11%). HIV isolates from these patients were susceptible (fold change < 2.5) to all five protease inhibitors in 18% of patients and to none in 8%. Isolates from patients receiving nelfinavir were less likely to have reduced susceptibility to other protease inhibitors than isolates from patients treated with indinavir (P < 0.001) or one of the other three agents (P < 0.001), even after adjustment for the duration of prior protease inhibitor use. Reduced susceptibility to saquinavir and amprenavir was observed significantly less frequently than for the other protease inhibitors. CONCLUSION: The frequency of protease inhibitor cross-resistance and the magnitude of changes in susceptibility varied according to the initial protease inhibitor used in the failing treatment regimen. Significantly less protease inhibitor cross-resistance was demonstrated for isolates from patients failing a nelfinavir-containing regimen compared with those from patients receiving other protease inhibitors.