Human Immunodeficiency Virus Type 2 (HIV-2), the second causative agent of the acquired immunodeficiency syndrome (AIDS), is mainly present in West Africa, where it was discovered 1 and spread to Europe, Asia, and the Americas in a slow but continuous manner. Although the two types of HIV (1 and 2) share only 40% of their amino acid sequences, HIV-2 infected individuals in developed countries are treated with highly active anti retroviral therapy (HAART), following the same therapeutic protocols that have been defined for HIV-1 infection.

HAART targets two main viral enzyme activities, the Reverse Transcriptase and the Protease. The drugs inhibiting the Protease competitively bind the substrate binding site of the enzyme, thus abolishing the proteolytic maturation of the Gag and Gag-pol precursors, resulting in the production of immature, non infectious particles 2.

Many epidemiological studies on HIV-1 infected individuals have documented that following anti-retroviral treatments a number of resistant isolates emerge causing therapeutic failure 3. Resistant HIV-1 Proteases present a specific pattern of amino acid substitutions that are categorized as major or minor mutations 4.

Thus far the analysis of the data obtained from the few studies of the impact of Protease Inhibitors (PI), on HIV-2 infected individuals is still not consistent enough to clearly define the specific amino acid substitutions conferring resistance to these anti retroviral drugs. Nevertheless, the few reported data establish that i) the natural nucleotide polymorphism of the HIV-2 Protease includes amino acid substitutions that are associated with drug resistance in HIV-1 5, and ii) comparison between the Protease sequences of treated and untreated HIV-2 infected individuals reveals a number of mutations under some PI-selective pressure such as K7R, V20I/A, I36V, V46I, I54L/M, V62A/T, V71L, I82F, I84V/L, 90LM, and L99F 678. In a recent report, where five primary HIV-2 isolates obtained from two different infected individuals at different time points were tested for PI activity, it has been shown that neither I36V, V46I nor V71L modified the susceptibility of HIV-2 Protease to PIs 9. Results, obtained from functional test on the ability of those PI to inhibit viral replication showed that amino acid substitutions such as : T12P 10, G17Q 10, R72A 10, M76V 10, I54M 79, I82F 789, V47A 9, K45R 9 confer a resistant phenotype.

In this study we present a novel test to assess isolated Protease susceptibility to PIs. This assay is based on the expression of the viral Protease in yeast cells and the definition of IC₅₀ values for the PIs. As a proof of principle for this assay, we report the susceptibility pattern to Lopinavir and Saquinavir of HIV-2 Proteases from 7 infected individuals.

With the aim of designing a tool capable of defining the susceptibility of any viral Protease from HIV-2 treated or untreated infected individuals to different PIs in a cellular context, we exploited the yeast cell. Indeed Saccharomyces cerevisiae has been used for many decades as an experimental tool to study the functional role of several bacterial and viral proteins through the phenotype of the resulting transformed cells. Moreover, it has recently been shown that the HIV-1 Protease expressed in yeast induces cell lysis by a yet unknown mechanism unrelated to apoptosis 11.

The proteases encoded by the two HIV types are not identical, neither in their amino acid sequence nor in the specificity of peptide bond recognition 12. Furthermore since there is no information concerning the cellular molecular pathway involved in the HIV protease-specific lysis of yeast cells, we first tested the phenotype induced by the expression of the HIV-2 Protease in yeast transformed cells. For this purpose we sub-cloned the Protease gene of the ROD isolate of HIV-2 in the pRS316Gal1/10 vector 13 under the control of the galactose inducible Gal 1/10 promoter 14, as detailed in the Material and Methods section.

HIV-2_ROD Protease transformed yeast grew on glucose containing plates, but not on a galactose carbon source (Fig 1A). When an HIV specific protease inhibitor, such as Saquinavir (SQV) at 200 μM concentration, was added to the culture media (Fig 1A), the lethal phenotype was no longer observed on galactose. The galactose induced cell death was found to be directly linked to the protease enzyme activity, since when the Asp residue of the active site of the viral enzyme was modified to Ala (D25A mutant), the inactive Protease expressed in transformed cells (Fig 1C, lane 3) did not induce cell death when grown on galactose (Fig 1B). This implies that the protease activity was responsible for cell death. Additional evidence supporting that it is indeed this viral enzymatic activity which kills the yeast cells, and not simply the production of the exogenous protein, was provided through the analysis of the expression of HIV-2_ROD Protease in transformed yeast which grew in galactose in the presence of 60 μM of the protease inhibitor Lopinavir (LPV). Western Blot analysis of cytoplasmic protein extracts from those cells showed an 11 Kd band recognized specifically by a monoclonal antibody raised against the viral Protease (Fig 1C, lane 2). As expected, this band was absent when the transformant was grown in glucose (Fig 1C, lane 1).

Based on this observation, we further studied the susceptibility of the Protease from the ROD isolate to LPV and SQV. Equal amounts of transformed cells were incubated in galactose, in the presence of increasing amounts of LPV or SQV. Cell growth was scored 48 h later by measuring the absorbance of the cell culture at 600nm. There was a tight correlation observed between the inhibitor concentration and cell growth which allowed determination of the corresponding IC₅₀ values, for LPV 16.6+/-2.5 μM, and for SQV 149.7+/-7.2 μM (Fig 2A). These IC₅₀ values were defined as the inhibitor concentration where cell growth reached 50% of regular growth in glucose. The growth arrest was not due to a toxic action of the PIs on the cells, since pRS316Gal1/10-HIV2PR transformed yeast incubated in glucose, in the presence or the absence of each PI (Fig 2B), or cells containing the pRS316Gal1/10 plasmid incubated in galactose, in the presence or the absence of PI (data not shown) grew identically.

As a first approach to validate this experimental system we measured the ability of LPV and SQV to inhibit HIV-2_ROD Proteases harbouring resistant mutations. Out of a library of mutated HIV-2_ROD Proteases that we created (see Material and Methods), we were able to select 3 mutants, containing at least one of the 8 amino acid substitutions which were previously defined on clinical samples through functional studies as conferring resistance 78910. The susceptibility to LPV and/or SQV of those mutants was defined by measuring cell growth of transformed cells in galactose, in the presence of either 150 μM of LPV or 1mM of SQV, that corresponds respectively to 9.0 and 6.7 fold of the IC₅₀ value of the wild type Protease. In these culture conditions, cell growth of HIV-2_ROD Protease transformed yeast was lower when compared to cell growth in the control media containing glucose. The percentage of cell growth was 91.5% in the presence of LPV, and 75% when incubated with SQV (Table 1). When this HIV-2_ROD Protease harbours amino acid substitutions known to be involved in HIV-2 PI resistance such as K45R, I54M, or M76V 910, cell growth, in spite of the presence of the inhibitors, was strongly arrested achieving in some cases only 4% of normal growth (Table 1) demonstrating a tight correlation between resistance, defined in physiological conditions, and loss of susceptibility, defined in our experimental system.

Using this assay, the susceptibility of HIV-2 Proteases from infected individuals was determined. We amplified the protease from DNA extracted from peripheral blood lymphocytes from 7 infected individuals either undergoing successful antiretroviral treatment or experiencing treatment failure (Table 2), as described in Material and Methods. For correct expression in yeast, an ATG start codon was introduced before the first Protease amino acid, and a TAA stop codon preceded by a Leu codon (as in HIV-2_ROD Protease) was introduced after the 98^th codon. The DNA fragments obtained were sub cloned into the pRS316Gal1/10 expression vector, and the resulting plasmids were used to transform BY4741 yeast cells. Yeast transformants expressing HIV-2 Proteases were tested for their ability to grow in galactose in the presence of LPV and SQV (Table 3). We observed that at a concentration of 150 μM of LPV, one yeast strain presented cell growth similar to the ROD isolate (expressing MRT-29 viral Protease), three presented cell growth comparable to growth observed in glucose containing media (expressing MRT-7, -22, -25 viral Proteases), two (expressing MRT-8, -20 viral Proteases) showed a slightly lower growth than the ROD isolate, and only the yeast transformant expressing the MRT-1 Protease was impaired in its growth. When the same transformed cells were incubated in the presence of 1mM SQV, four of them showed the same phenotype as the ROD isolate (MRT-1, -20, -22, -29), two showed a cell growth comparable to the one obtained in glucose containing media (MRT-7, -25), and only yeast transformant containing the MRT-8 Protease grew to 55% compared to the control (Table 3). We further defined the IC₅₀ values of the viral enzyme activity from those patients (Table 3). Interestingly, among the different samples analysed, only MRT-8 presented an IC₅₀ to SQV 4.2 fold greater than the reference strain. Although definition of specific sensitive/resistant cut-off values (IC₅₀ patient isolate/IC₅₀ sensitive-strain) for each PI can only be obtained after a clinical study, the IC₅₀ ratio we found for MRT-8 expressing Protease correlates to reported findings where loss of sensitivity corresponded to a higher IC₅₀ value of about 4 fold, compared to the sensitive wild type isolate 15. It should be remarked that :i) the HIV-2 Protease isolated from patient MRT-25 who never received SQV, showed a hypersensitivity towards this inhibitor (IC₅₀MRT-25/IC₅₀HIV-2_ROD = 0.03), and ii) the Protease from patient MRT-1 showed a loss in sensitivity to LPV which might not be related to resistance since the IC₅₀ ratio was found to be lower than 4.

In an attempt to define the amino acid mutation/s specifically responsible/s for the SQV loss of sensitivity phenotype of the MRT-8 Protease, we sequenced the coding DNAs of the 7 different Proteases and compared the sequences obtained to that of HIV-2_ROD Protease (Fig. 3). Among the 9 mutations observed in the MRT-8 Protease (4 conservative and 5 non-conservative), we focused on the 90^th residue since the L90M substitution has already been classified as a major mutation in HIV-1 which confers resistance to SQV and LPV 4, and has been suggested to be associated with resistance to a yet undefined PI in HIV-2 Protease 816. The HIV-2_ROD Protease was mutated and the resultant ROD L90M mutant was tested for its susceptibility to LPV and SQV in yeast as was performed for HIV-2 Proteases from infected individuals. At the same time we created and tested the ROD L99F mutant, since this mutation was proposed to confer PI resistance in a study that scored the emergence of mutations in infected individuals failing an anti-protease containing regime 6, and no functional data of this suggested resistance is available. The results obtained, presented in Table 4, show that the L90M mutant have lost sensitivity to SQV but not to LPV (IC₅₀HIV-2_ROD L90M/IC₅₀HIV-2_RODwt = 3.7 and 0.75 respectively). Conversely, the Phe residue in position 99 did not modify the respondent character of the Protease to either of the tested PIs (IC₅₀HIV-2_ROD L99F/IC₅₀HIV-2_RODwt = 0.67 for LPV and 0.98 for SQV).

Through the data presented we have showed that the loss of HIV-2 Protease sensitivity to LPV and SQV can easily be evaluated by defining the IC₅₀on protease-expressing yeast cells. Developing similar strategies to measure the IC₅₀ values for other PIs in yeast constitutes the next step necessary to clearly define the overall resistance phenotype of any HIV-2 Protease.

The causal relationship between the increase of the viral load in plasma and the emergence of HIV isolates that are resistant to protease inhibitors and other anti retroviral compounds has been well established for HIV-1 infected individuals experiencing therapeutic failure 81718. In relation to HIV-2 infection, there is still a lack of information concerning the amino acid substitutions that confer resistance to various PIs. Current phenotyping methods for defining the sensitivity of an isolate to PIs, are founded on the production of a virus that contains the PCR amplified gene encoding the Protease isolated from the infected individual in a wild type genetic backbone [reviewed in 19]. This chimeric virus is then tested for its infectivity in the presence of PIs, and the IC₅₀ values are obtained. This technology is complex and requires a significant infrastructure which is not well adapted for screening an increasing number of HIV-2 Proteases from infected individuals. As a way to define the biochemical resistance phenotype of viral proteases, we set up a simple and accurate experimental yeast cellular system to evaluate Protease sensitivity to antiretroviral compounds.

Our initial observation in this study was that the expression of the Protease encoded by HIV-2 in yeast cells induces cell death, as previously shown for the HIV-1 homologue 11. This is not the only example of viral proteases that arrest cell growth in S. cerevisiae. Previous reports show that the 2A proteases from poliovirus and human rhinovirus 2 2021, both species belonging to the Picornavidae genus, produce cell growth arrest leading to cell death 10 h after their expression in yeast. These studies did not elucidate the specific protease-induced molecular cascades involved in cell death, but suggest the inhibition of RNA synthesis but not of translation as a key mechanism 2021. Although HIV-1 Protease possesses unique structural and functional properties that distinguish it from its cellular aspartic counterparts 22, several cellular proteins have been found to be efficient substrates. Among those, are proteins of the intermediary filaments, cytoskeleton components such as vimentin, desmin and glyal fibrillary acidic protein or cytoskeletal proteins as actin, troponin, tropomyosin 2324, and microtubule-associated proteins 25, or bcl-2 26, and precursors of NF-κB 27. Human cells expressing the HIV-1 Protease die via apoptosis 26, however the lethal effect of this enzyme activity in yeast is likely to involve other death pathways since there is no evidence of apoptosis in S.cerevisiae 28. It can thus be hypothesized that S. cerevisiae cell lysis produced by the viral Protease might occur through a drastic modification and/or degradation of the cellular cytoskeleton, as it does in higher eukaryote cells.

The mechanism by which HIV Proteases induce cell death in yeast is still an issue to be clarified, however since inhibition of Protease enzyme activity restores cell growth in HIV-2 Protease transformed yeast cells, we were able to define the IC₅₀ values of the Protease from HIV-2_ROD to LPV and SQV: 16.6+/-2.5 μM and 149.7+/-7.2 μM respectively. The difference between these values and those already reported in the literature, 27 nM and 11 nM respectively 929, might come from the nature of the yeast cell architecture. Indeed, the plasma membrane of the yeast cell has a non-human lipid composition that accounts for a different cell permeability. Moreover, the nature of the yeast cell wall might be a barrier for the entry of molecules that are hardly soluble in aqueous solutions, as the PIs.

The novel experimental system we present in this study allowed us to define the biochemical resistance phenotype to LPV and SQV for isolates from HIV-2 infected individuals either experiencing therapy failure or not. Moreover for the first time we were able to clearly define that the L90M substitution is responsible for SQV resistance but does not alter LPV sensitivity.

Our study also points out the importance of defining a PI susceptibility phenotype, as opposed to relying on the genotype, for definition of PI resistance/sensitivity in HIV-2 Protease isolates. The L99F substitution has previously been proposed to be involved in PI resistance, based on comparison of HIV-2 Protease sequences from PI untreated and treated patients 6. However, in our phenotyping system, the L99F substitution did not modify susceptibility either to LPV or to SQV. This, highlights the limitations of comparative genotyping procedure in determining susceptibility of HIV-2 Proteases to PI and in predicting the ability of a given PI to inhibit isolates containing particular mutations. Establishing PI susceptibility phenotypes is the most accurate method to determine whether an HIV-2 virus strain is sensitive to a specific inhibitor used in a therapeutic protocol.

Our work is a proof-of-concept setting up and evaluating a phenotypic test to define protease susceptibility to PIs for HIV-2. The next step will include clinical validation to establish correlations with clinical outcomes and evaluate the predictive value for therapeutic failure.

Lopinavir (LPV), Saquinavir (SQV), Protease Inhibitor (PI), Highly active anti retroviral therapy (HAART).

The author(s) declare that they have no competing interests.

NBM and GA performed the experiments. PG wrote the manuscript and participated with NBM and GA in the experimental design and data interpretation. DR provided the clinical specimens and participated in data interpretation. All authors read and approved the final manuscript.