Novel Compound Inhibitors of HIV-1NL4-3 Vpu

HIV-1 Vpu targets the host cell proteins CD4 and BST-2/Tetherin for degradation, ultimately resulting in enhanced virus spread and host immune evasion. The discovery and characterization of small molecules that antagonize Vpu would further elucidate the contribution of Vpu to pathogenesis and lay the foundation for the study of a new class of novel HIV-1 therapeutics. To identify novel compounds that block Vpu activity, we have developed a cell-based ‘gain of function’ assay that produces a positive signal in response to Vpu inhibition. To develop this assay, we took advantage of the viral glycoprotein, GaLV Env. In the presence of Vpu, GaLV Env is not incorporated into viral particles, resulting in non-infectious virions. Vpu inhibition restores infectious particle production. Using this assay, a high throughput screen of >650,000 compounds was performed to identify inhibitors that block the biological activity of Vpu. From this screen, we identified several positive hits but focused on two compounds from one structural family, SRI-41897 and SRI-42371. We developed independent counter-screens for off target interactions of the compounds and found no off target interactions. Additionally, these compounds block Vpu-mediated modulation of CD4, BST-2/Tetherin and antibody dependent cell-mediated toxicity (ADCC). Unfortunately, both SRI-41897 and SRI-42371 were shown to be specific to the N-terminal region of NL4-3 Vpu and did not function against other, more clinically relevant, strains of Vpu; however, this assay may be slightly modified to include more significant Vpu strains in the future.


Introduction
Since the beginning of the HIV epidemic, more than 75 million people have been infected with HIV and about 33 million people have died as a result. While there have been advances in treating HIV, approximately one million people per year die from complications of CD4-induced Env epitopes targeted by ADCC-mediating, non-neutralizing antibodies naturally present in the plasma from HIV-1-infected individuals [26,27]. Downregulation of BST-2/Tetherin, which otherwise traps viral particles at the surface of infected cells, contributes to limit the overall amount of Env on the surface of infected cells and to a lower ADCC response [27][28][29].
The impact of Vpu on viral spread suggests considerable therapeutic potential for a Vpu-specific inhibitor [30,31]. There have been several attempts to specifically target Vpu; however, many of the proposed compounds target the ion-channel activity of Vpu and have been shown to have a low selectivity index or non-Vpu-specific ion-channel targets [23,32,33]. Notably, the compound BIT225 is currently in Phase II of the Australian New Zealand clinical trials and has been shown to increase plasma derived activated CD4 + and CD8 + T cells and NK cells if coupled with ART when compared with patients being treated with ART alone [34]. However, it is important to note that BIT225 is not Vpu specific [33,35] and does not appear to affect the Vpu-mediated antagonism of BST-2/Tetherin [36]. There have been additional efforts to target the ubiquitin ligase machinery upstream of Vpu, such as with the compound MLN4924, but this compound does not rescue ADCC activity at tolerated levels in cells [37].
Finally, our lab has previously demonstrated that Vpu downmodulates Gibbon Ape Leukemia Virus Envelope (GaLV Env) through the targeting of the cytoplasmic tail region using the same SCF/βTrCP ubiquitin ligase [16,38,39]. Using this observation, we developed a novel assay for a gain-of-function high throughput screen for inhibitors of Vpu. Here, we identify a family of novel inhibitors specific to NL4-3 HIV-1Vpu. Two of the identified compounds, SRI-41897 and SRI-42371, were examined further and found to effectively block the Vpu downmodulation of GaLV Env, BST-2/Tetherin, CD4, and rescue ADCC killing of infected cells.

Plasmids
NL4-3 derived HIV-1-CMV-GFP was provided by Vineet Kewal Rammani (National Cancer Institute (NCI)-Frederick) and previously described [40]. This proviral vector lacks the accessory genes vif, vpr, nef, and env and contains a CMV promoter driven GFP in the place of nef. Both Vpu positive and Vpu negative constructs were used and described previously [38]. The infectious molecular clone NL4-3 was obtained through the NIH HIV Reagent Program, Division of AIDS, NIAID, NIH, contributed by M.A. Martin [41].
HIV-1 Gag-mRFP constructs were created in a two-step cloning process. First, the Nhe1-Xho1 region of NL4-3 derived HIV-1-CMV-GFP constructs (+/− Vpu) containing GFP were replaced with a Blasticidin Resistance gene gblock fragment ordered from Integrated DNA Technologies (IDT). Secondly, the BamH1-Xho1 region of a previously described Gag-mRFP construct gifted by Akira Ono (University of Michigan) was replaced with the BamH1-Xho1 region of the new constructs where Gag-mRFP replaces Pol [42].
The MLV/GaLV chimera construct used in infectivity assays was described previously [16,38,39]. The APOBEC3G-GFP was gifted by Daniel Salamango and described previously [43]. Plasmids for creation of stable cell lines were created by insertion of CDC25-GFP or MLV/GaLV into the retroviral transfer vector pQCXIH (Clonetech, Mountain View, CA, USA). The SHIV KU-1vMC33 strain of Vpu was gifted from Edward Stephens (University of Kansas), and the CH77 strain Vpu was inserted using a GBlock (IDT). Both Vpu strains were cloned into the NL4-3 derived HIV-1-CMV-GFP plasmid described above. The proline insertion mutant was created using PCR mutagenesis of NL4-3 Vpu and the subsequent infusion into the NL4-3 derived HIV-1-CMV-GFP plasmid described above.

Cell Lines
The base HEK293FT (293FT) cell line used for stable expression was originally obtained from Invitrogen (Carlsbad, CA, USA). For the GaLV inhibition assay, cell lines stably expressing GaLV/MLV chimera construct were transduced with NL4-3 derived HIV-1-CMV-GFP virus pseudotyped with VSV-G.
Counter-screen stable cell lines (AOBEC3G-GFP and CDC25α-GFP) were cloned into a pQCXIH retroviral transfer vector packaged and transduced into HEK293FT cells using VSV-G gylocprotein, then selected with hygromycin. In both cases, a single cell isolate was selected and propagated for use.
The 293T mCAT-1 cell line expressing the ecotropic F-MLV Env receptor was provided by Walther Mothes.
TZM-GFP cells were gifted by Massimo Pizzato and previously described [44]. All 293FT based cell lines and TZM-GFP cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 7.5% fetal bovine serum, 2 mM L-glutamine, 1 mM sodium pyruvate, and 10 mM nonessential amino acids.
The cell lines for the ADCC assay were obtained and cared for as previously described [45]. The NK cell line (KHYG-1 cells) was obtained from the Japan Health Sciences Foundation and transduced with the V158 variant of human CD16. The target cells (CEM.NKR-CCR5) were obtained from the AIDS Reagent Program and were modified as described by transducing with a vector carrying an SIV LTR driven luciferase reporter gene.
Primary human peripheral blood mononuclear cells (PBMCs) and CD4 + T cells were isolated, activated, and cultured as previously described [46]. Briefly, PBMCs were obtained by leukapheresis and CD4 + T lymphocytes were purified from resting PBMCs by negative selection using immunomagnetic beads per the instructions of the manufacturer (StemCell Technologies, Vancouver, BC, Canada) and were activated with phytohemagglutinin-L (10 µg/mL) for 48 h and then maintained in RPMI 1640 complete medium supplemented with recombinant interleukin-2 (rIL-2) (100 U/mL). VSV-G-pseudotyped HIV-1 NL4-3 virus was produced and titrated. Viruses were then used to infect activated primary CD4 T cells from healthy HIV-1-negative donors by spin infection at 800× g for 1 h in 96-well plates at 25 • C. Ethics Statement: Written informed consent was obtained from all study participants and research adhered to the ethical guidelines of CRCHUM and was reviewed and approved by the CRCHUM institutional review board (ethics committee, approval number CE16.164-CA). Research adhered to the standards indicated by the Declaration of Helsinki. All participants were adults and provided informed written consent prior to enrolment in accordance with Institutional Review Board approval.

GaLV Assay/Screen/Flow
Infectivity (GaLV) Assay. Cell lines stably expressing GaLV/MLV chimera construct were transduced with an aliquot of NL4-3 derived HIV-1-CMV-GFP virus pseudotyped with Vesicular Stomatitis Virus protein G (VSV-G) in a 10 cm cell culture dish. 24 h posttransduction, cells were separated into 6 or 12 well plates, and treated with compound DMSO (Sigma-Aldrich Cat# D4540), or MLN4924 (Calbiochem Cat# 5.05477.0001) for 24 additional hours at 37 • C. Viral media was collected and frozen at −80 • C for between 2 and 24 h. Once thawed, the virus was spun at 3000 RCF for 5 min to clear any debris. Viral supernatant (NL4-3 derived HIV-1-CMV-GFP pseudotyped with GaLV ENV/MLV chimera) was then used to transduce mCAT-1 cells for 48 h at 37 • C and collected for flow cytometry. GFP positive mCAT-1 Cells (indicating a successful transduction) were measured.

High Throughput Screen
Compound collection: Southern Research maintains a collection of 759,059 unique, non-proprietary compounds assembled from various commercial vendors (Enzo, Selleck, ChemBridge, Enamine, Life Sciences) for screening targets in HTS. Eight molecular properties were calculated for this collection using the Accelrys Pipeline Pilot application. An analysis showed that 89.7% of compounds have molecular properties matching all eight criteria for lead-like molecules to serve as starting points for a drug discovery effort (Molecular Weight ≤ 500; Heteroatom count ≤ 10; Number Rotatable Bonds ≤ 8; Number Aromatic Rings ≤ 4; A Log p ≤ 6; Molecular Polar Surface Area ≤ 200; H-bond acceptors < 10; H-bond donors < 5). Within this chemical space, the collection is diverse, containing: (1) 278,767 non-overlapping Murcko scaffolds with an average cluster size 2 to 3, (2) 9228 individual ring systems with unique substitution pattern (average frequency 218), and (3) 16,008 contiguous ring systems with unique substitution patterns (average cluster frequency 125). A subset of 674,336 compounds from this collection were tested in HTS format at a single concentration of 10 µg/mL or 30 µM depending on the compound library source.
Assay Method: Library compounds were diluted in assay medium (DMEM with 10% FBS, 1% PSG, 1% HEPES) to prepare a 3.5× concentrated dosing solution (35 µg/mL or 105 µM) and added to 384-well black clear bottom plates (Corning; Cat # 3764BC) in 10 µL (1/3.5 final well volume). Twenty-five µL of Producer cells (HEK 293 FT cells stably expressing GaLV Env and pNL4-3 ∆Env-GFP) at 800,000 Cells/mL was added to the plates for a final count of 20,000 Cells/well, a final compound concentration of 10 µg/mL or 30 µM, and a final DMSO concentration of 0.5%. Assay medium alone (at 0.5% DMSO) served as the negative control (columns 1 and 2 of each plate) and 0.5 µM MLN4924 (Activebiochem; Cat # MLN4924) (at 0.5% DMSO) as the positive control (columns 23 and 24 of each plate). The plates were incubated at 37 • C/5% CO 2 for 24 h in a humidified atmosphere. Ten µL of supernatant was transferred from each plate into a new 384-well black clear bottom plate, and 20 µL of Acceptor cells (293FT cells stably expressing mCAT-1 and the MLV Env receptor) at 300,000 Cells/mL added to the plates now containing 10 µL supernatant for a final count of 6000 Cells/well. The plates were returned to 37 • C/5% CO 2 for 72 h and GFP signal imaged on a Mirrorball plate reader (BMG Labtech). The fluorescence signals of the test wells were normalized to percent activation relative to the average of the positive control wells on each plate by the formula: % activation = 100 × (1 − (test well signal-average positive control signal)/(average negative control signal-average positive control signal)) Potential hits from this primary HTS were cherry-picked and tested at ten concentrations in a two-fold serial dilution over a concentration range of 50-0.1 µM or µg/mL, in both the main assay outlined above as well as in a counter screen assay.
HTS Counter Screen: Compounds were diluted in assay medium (DMEM with 10% FBS, 1% PSG, 1% HEPES) to prepare a 3.5× concentrated dosing solution and added to 384-well black clear bottom plates (Corning; Cat # 3764BC) in 10 µL (1/3.5 final well volume). Twenty-five µL of Acceptor cells (293FT cells stably expressing mCAT-1 and the MLV Env receptor) was added to the plates for a final count of 6000 Cells/well, a final compound concentration of 50-0.1 µM or 50-0.1 µg/mL, and a final DMSO concentration of 0.5%. Assay medium alone (at 0.5% DMSO) and 0.5 µM MLN4924 (at 0.5% DMSO) served as controls. The plates were incubated at 37 • C/5% CO 2 for 24 h in a humidified atmosphere, after which GFP signal was imaged on a Mirrorball plate reader (BMG Labtech).

Cellular Target Counter Screens
293FT cells stably expressing CDC25α-GFP or APOBEC3G-GFP (described above) were treated with 40 µM compound, DMSO, or MLN4924 for 24 h at 37 • C and collected for flow cytometry. Mean Fluorescence Intensity (MFI) was measured.

CD4 Surface Labeling in TZM-GFP
TZM-GFP cells were transduced with VSV-G pseudotyped packaged HIV-1 constructs containing either gag and env and lacking vif, vpr, pol and vpu or containing gag and env and lacking vif, vpr, pol and vpu. Both constructs contain mCHERRY in place of the Pol gene. The media was changed on day four post-infection, and cells were treated with 40 µM SRI-41897, SRI-42371, 1 µM MLN4924, or 0.1% DMSO for 24 h at 37 • C and 5% CO 2 . For surface staining, cells were lifted using TrypLE Express and transferred to round bottom 2 mL tubes in 1 mL of PBS. Cells were pelleted at 300 RCF for 3 min and the supernatant was removed. Cells were resuspended in a blocking solution containing 5% goat serum in PBS at 4 • C for 30 min. Cells were pelleted at 300 RCF for 3 min and the blocking solution was removed. Staining was done in a 1% goat serum and PBS solution containing 1:100 APC conjugate CD4 Stain (Life Technologies REF# MHCD0405) at 4 • C for 1 h in the dark. Cells were pelleted at 300 RCF for 3 min, and the stain was removed. Cells were washed 3 times in 1mL of PBS. Following washes, cells were fixed in 1% PFA in PBS. PFA solution was removed after centrifugation for 3 min at 300 RCF, and cells were resuspended in 400 µL of PBS and analyzed on n Accuri C6 flow cytometer.

Flow Cytometry
Cells were washed with PBS and lifted using Tryple Express (Gibco). Cells were removed from the plate and moved to 1.5 mL microcentrifuge tubes containing PBS and paraformaldehide (PFA) at a final concentration of 4% PFA for 10 min at room temperature. Next, cells were spun down at 800 RCF for 5 min and washed 2 times with PBS before the final resuspension in PBS and were analyzed using an Accuri C6 flow cytometer.

2G12 Surface Labeling
Surface Labeling 2G12 was previously described [47]. (p24 + ) was determined by gating the living cell population on the basis of viability dye staining (Aquavivid; Thermo Fisher Scientific). Samples were acquired on an LSR II cytometer (BD Biosciences), and data analysis was performed using FlowJo vX.0.7 (Tree Star, Ashland, OR, USA).

ADCC
The ADCC assay was previously described [45]. Target cells were infected 4 days prior to each assay with either WT NL4-3 HIV-1 or NL4-3 HIV-1 ∆Vpu. CEM.NKR-CCR5 cells were infected by spinoculation for 2 h at 1200 RCF. After spinoculation, the virus was removed, and target cells were cultured in R10 medium (RPMI medium supplemented with 10% FBS, L-glutamine, and Primocin). Immediately before the assembly of ADCC assays, the infected target cells were washed three times in R10 medium. 24 h after infection, compounds were added to the infected target cells at a concentration of 25 µM for each independently. Forty-eight h after infection, target cells were washed then incubated with the NK cell line KHYG-1 at a ratio of 10:1 of 150,000 effector cells to 15,000 target cells in round-bottom, tissue culture-treated polystyrene 96-well plates. Assays were performed in R10 culture medium containing 10 U IL-2 per mL, with no CsA. For controls, each plate contained NK effector cells and uninfected target cells in the absence of antibody (resulting in a 0% relative light units (RLU)), and NK cells and infected targets in the absence of antibody (resulting in 100% RLU). Serial, 4-fold, triplicate dilutions of plasma or monoclonal antibody were added. Each compound was once again used to treat the cell mixture at 25 µM. Once targets, effectors, and serially diluted antibody were combined, assay plates were incubated for 8 h at 37 • C and 5% CO 2 . After an 8 h incubation, a 150 µM volume of cells was resuspended and mixed by pipette with 50 µM of the luciferase substrate reagent BriteLite Plus (Perkin Elmer, Waltham, MA, USA) in white 96-well plates. Luciferase activity was read approximately 2 min later using a Wallac Victor3 plate reader (Perkin Elmer). 50% ADCC titers were estimated as previously described for virus neutralization assays [46]. The 50% intercept was calculated using the adjacent%RLU values above and below 50% RLU.

High Throughput Screen for Vpu Inhibitors
Our lab previously showed that Vpu prevents infectious particle production in HIV-1 particles pseudotyped with a GaLV Env chimera consisting of a Murine Leukemia Virus (MLV) Envelope containing a GaLV Env c-terminal-domain (CTD) (henceforth referred to as GaLV Env) that is targeted by Vpu for downmodulation [38]. Using this observation, we developed a gain-of-function high throughput screen (HTS) for Vpu inhibitors ( Figure 1A). This screen uses HEK 293 FT cells stably expressing GaLV Env and pNL4-3 ∆Env-GFP to produce virus followed by the transduction into 293FT cells stably expressing mCAT-1, the MLV Env receptor, as target cells [39]. The cells produce HIV-1 viral particles pseudotyped with GaLV Env that successfully transduce mCAT-1 expressing cells in the absence of Vpu and produce few to no viral particles if the NL4-3 provirus contains a functioning Vpu ( Figure 1B). The compound MLN4924 is a Neddylation inhibitor that effectively blocks the function of the cullin-Ring group of ubiquitin ligases and therefore blocks the cellular machinery that Vpu depends on for much of its activity [15,16,37,48]. In the absence of Vpu or in the presence of an inhibitor blocking Vpu, such as MLN4924, infectious particle production is rescued ( Figure 1B).
This assay was used in the HTS to screen 674,336 commercially available compounds for Vpu inhibition ( Figure 1C,D). The average Z'-value calculated from the 32 positive and 32 negative control wells on each plate was 0.72, with a range from 0.51 to 0.85 for the 2145 assay plates run in the screen. A Z'-value > 0.5 indicates that the assay performance was adequate to detect active compounds when tested once [49]. A statistical analysis (mean + 3 × SD of all test compound data) identified 14.73% inhibition as the cutoff between inactive and active compounds. Based on this cutoff, a total of 461 compounds were identified as hits resulting in an overall hit rate of 0.07% for the primary screen. Of the 461 hit compounds identified, 202 were confirmed as active following retest at ten concentrations in the GFP reporter assay used in HTS. These compounds were also evaluated for concentration-response in counter screen assays measuring compound autofluorescence or non-Vpu mediated activation of GFP expression. Based on these results, 177 compounds were excluded for causing autofluorescence or having off-target effects. The remaining 25 compounds were selected for further testing in secondary assays, of which 21 were available for purchase as fresh powders. Further information for hits that are not further explored in this manuscript is available in Supplemental Document S1.
for Vpu inhibition ( Figure 1C,D). The average Z'-value calculated from the 32 positive and 32 negative control wells on each plate was 0.72, with a range from 0.51 to 0.85 for the 2145 assay plates run in the screen. A Z'-value > 0.5 indicates that the assay performance was adequate to detect active compounds when tested once [49]. A statistical analysis (mean + 3 × SD of all test compound data) identified 14.73% inhibition as the cutoff between inactive and active compounds. Based on this cutoff, a total of 461 compounds were identified as hits resulting in an overall hit rate of 0.07% for the primary screen. Of the 461 hit compounds identified, 202 were confirmed as active following retest at ten concentrations in the GFP reporter assay used in HTS. These compounds were also evaluated for concentration-response in counter screen assays measuring compound autofluorescence or non-Vpu mediated activation of GFP expression. Based on these results, 177 compounds were excluded for causing autofluorescence or having off-target effects. The remaining 25 compounds were selected for further testing in secondary assays, of which 21 were available for purchase as fresh powders. Further information for hits that are not further explored in this manuscript is available in Supplemental Document S1.

HTS Reveals Potent Vpu Inhibitors
One compound, SRI-41897, inhibited Vpu more potently than others. Interestingly, SRI-41897 had a very similar structure to another of the 21 HTS hits, SRI-40244, highlighting the potential for the structural family as Vpu inhibitors (Figure 2A,B). While multiple occurrences of similarly structured compounds are interesting, SRI-40244 had very low activity in the GaLV inhibition assay and was not pursued further (Supplemental Document S1). As we were interested in this structural class of compounds, we screened more than 80 additional commercially available analogs for SRI-41897. Most of the analogs contained the same background molecule with the addition, removal, or exchange of individual side chains. From the screened analogs, we found SRI-42371 that similarly inhibited Vpu in the GaLV Inhibition assay compared to the original compound ( Figure 2D). These two compounds became the focus of our study. The EC 50 of both compounds is in the µM range, 4.4 and 7.4 µM for SRI-41897 and SRI-42371, respectively, as measured in the GaLV inhibition assay ( Figure 2E).

HTS Reveals Potent Vpu Inhibitors
One compound, SRI-41897, inhibited Vpu more potently than others. Interest SRI-41897 had a very similar structure to another of the 21 HTS hits, SRI-40244, high ing the potential for the structural family as Vpu inhibitors (Figure 2A,B). While mu occurrences of similarly structured compounds are interesting, SRI-40244 had ver activity in the GaLV inhibition assay and was not pursued further (Supplemental ment S1). As we were interested in this structural class of compounds, we screened than 80 additional commercially available analogs for SRI-41897. Most of the analog tained the same background molecule with the addition, removal, or exchange of in ual side chains. From the screened analogs, we found SRI-42371 that similarly inh Vpu in the GaLV Inhibition assay compared to the original compound ( Figure 2D). two compounds became the focus of our study. The EC50 of both compounds is in th range, 4.4 and 7.4 M for SRI-41897 and SRI-42371, respectively, as measured in the inhibition assay ( Figure 2E).

Vpu-Independent Counter Screens
While the HTS is Vpu dependent, compounds that block the cellular machinery that Vpu depends on will also result in a positive hit. To ensure that the compounds are specific for Vpu, we developed a series of counter screens for the upstream cellular machinery maintaining MLN4924 as a positive control ( Figure 3A). tagged CDC25 were treated with the compounds DMSO or MLN4924 as a positive control, and MFI was recorded via flow cytometry ( Figure 3C). TrCP has been previously shown to downmodulate the cellular protein CDC25; therefore, a blockage of TrCP results in an increase of CDC25 expression and an increase in GFP as shown by MFI [54,55]. The addition of MLN4924 resulted in a clear increase in fluorescence; however, neither SRI-41897 nor SRI-42371 displayed an increase in fluorescence, indicating that they do not block TrCP function.  The first counter screen ensures that the compounds do not have any activity for the broader cullin family of ubiquitin ligases. In this assay, 293FT cells that are stably expressing both GFP tagged APOBEC3G and the HIV-1 accessory protein, Vif, were treated with compounds, either DMSO or MLN4924, and MFI was recorded via flow cytometry ( Figure 3B). Vif has previously been shown to downmodulate the cellular protein APOBEC3G using cullin5 machinery [50][51][52][53], so a blockage of cullin5 results in an increase of APOBEC3G expression and therefore an increase in GFP as shown by MFI. Addition of MLN4924 showed a clear increase in MFI of GFP, whereas neither SRI-41897 nor SRI-42371 displayed an increase in fluorescence, indicating that they do not block cullin5 or the larger cullin ubiquitin ligase family.
The second counter screen ensures that the compounds do not have any activity against the F-box protein, βTrCP 1 or 2. In this assay, 293FT cells stably expressing GFP tagged CDC25α were treated with the compounds DMSO or MLN4924 as a positive control, and MFI was recorded via flow cytometry ( Figure 3C). βTrCP has been previously shown to downmodulate the cellular protein CDC25α; therefore, a blockage of βTrCP results in an increase of CDC25α expression and an increase in GFP as shown by MFI [54,55]. The addition of MLN4924 resulted in a clear increase in fluorescence; however, neither SRI-41897 nor SRI-42371 displayed an increase in fluorescence, indicating that they do not block βTrCP function.

Compounds Rescue CD4 and BST-2/Tetherin Expression
To explore the impact of compounds on surface expression of CD4, TZM-GFP cells were transduced with a Gag-mRFP construct containing a non-functional Pol protein and a functional Env in the presence or absence of Vpu ( Figure 4A,B). CD4 levels of Vpu positive cells had a MFI average of 1764.8, which is 61% of Vpu negative cells that had a MFI average of 28,472.5 ( Figure 4A). After treatment with MLN4924 or compounds, CD4 surface expression was rescued by 47% or 25% on infected TZM-GFP cells, for SRI-41897 or SRI-42371, respectively, in comparison to the negative control, DMSO ( Figure 4B). While the compounds had no effect on CD4 surface expression in the absence of Vpu, it is interesting that MLN4924 reduced CD4 surface expression in these cells. This is likely why the rescue of CD4 surface expression with MLN4924 in the presence of Vpu was not significant.
in representative images of four total biological replicates.

Compounds Rescue CD4 and BST-2/Tetherin Expression
To explore the impact of compounds on surface expression of CD4, TZM-GFP cells were transduced with a Gag-mRFP construct containing a non-functional Pol protein and a functional Env in the presence or absence of Vpu ( Figure 4A,B). CD4 levels of Vpu positive cells had a MFI average of 1764.8, which is 61% of Vpu negative cells that had a MFI average of 28472.5 ( Figure 4A). After treatment with MLN4924 or compounds, CD4 surface expression was rescued by 47% or 25% on infected TZM-GFP cells, for SRI-41897 or SRI-42371, respectively, in comparison to the negative control, DMSO ( Figure 4B). While the compounds had no effect on CD4 surface expression in the absence of Vpu, it is interesting that MLN4924 reduced CD4 surface expression in these cells. This is likely why the rescue of CD4 surface expression with MLN4924 in the presence of Vpu was not significant.
To determine if SRI-41897 and SRI-42371 can rescue the expression of CD4 and Tetherin in human peripheral blood mononuclear cells (PBMCs), cells were infected with HIV-1NL4-3, treated with compound for 24 h, and surface labeled. Surface levels of CD4 and BST-2/Tetherin were measured via flow cytometry. Tetherin expression was restored an average of 38% and 29% and CD4 expression was restored an average of 115% and 88% when treated with 25 M SRI-41897 and SRI-42371, respectively, in PHA-activated, CD8-depleted PBMCs (Figure 4C,D).  To determine if SRI-41897 and SRI-42371 can rescue the expression of CD4 and Tetherin in human peripheral blood mononuclear cells (PBMCs), cells were infected with HIV-1 NL4-3 , treated with compound for 24 h, and surface labeled. Surface levels of CD4 and BST-2/Tetherin were measured via flow cytometry. Tetherin expression was restored an average of 38% and 29% and CD4 expression was restored an average of 115% and 88% when treated with 25 µM SRI-41897 and SRI-42371, respectively, in PHA-activated, CD8-depleted PBMCs ( Figure 4C,D).

Compounds Rescue ADCC Response
Antibody Dependent Cellular Toxicity (ADCC) is a cellular response to infection that results in cell killing through lysis. In recent years, several groups have shown that Vpu protects HIV-1 infected cells from killing by ADCC, and that this effect is mostly due to downmodulation of CD4 and BST-2/Tetherin (reviewed in [24,56,57]). We wanted to test if the compounds could rescue ADCC as they rescued the expression of CD4 and partial expression of BST-2/Tetherin (Figure 4). Before testing ADCC, we examined Env expression using a conformationally-independent anti-HIV-1 Env antibody 2G12 expression [58,59]. Primary CD4 + T cells were infected with WT NL4-3 virus, treated with compounds at various concentrations, and probed with the anti-HIV-1 Env antibody, 2G12 ( Figure 5A). We observed a dose dependent response in 2G12 binding, indicating that in the presence of compound there is more Env present on the surface of the cell. Next, target cells derived from CEM.NKR-CCR5 CD4+ T cells expressing a Tat-inducible luciferase reporter gene were infected with either WT NL4-3 HIV-1 or NL4-3 HIV-1 ∆Vpu, co-cultured with natural killer (NK) cells and reciprocal anti-HIV-1 Immune Globulin (HIVIG) dilutions [45] ( Figure 5B-D). In this assay, an increase in ADCC killing is shown through a decrease in luciferase signal (decrease in actively infected cells). In the absence of compound, ADCC levels are significantly higher in NL4-3 ∆Vpu infected cells when compared with NL4-3 WT infected cells ( Figure 5B); however, the introduction of compound brought ADCC levels against NL4-3 WT ADCC close to ∆Vpu levels ( Figure 5C,D), indicating that the presence of compounds rescue ADCC killing in NL4-3 WT infected cells.  (100% RLU). Serial, 4-fold, triplicate dilutions of plasma or monoclonal antibody were added. A luciferase signal was detected in each well. The 50% intercept was calculated using the adjacent %RLU values above and below 50% RLU. Error bars represent standard deviation.

Compounds Are NL4-3 Strain Specific
To test the breadth of Vpu strains that are inhibited by SRI-41897 and SRI-42371, we tested inhibition against a wide variety of available Vpu strains and discovered a strain very similar to NL4-3 Vpu that was resistant to both compounds; the strain is from a simian-human immunodeficiency virus strain, SHIVKU-1vMC33, containing a Vpu originally  (100% RLU). Serial, 4-fold, triplicate dilutions of plasma or monoclonal antibody were added. A luciferase signal was detected in each well. The 50% intercept was calculated using the adjacent %RLU values above and below 50% RLU. Error bars represent standard deviation.

Compounds Are NL4-3 Strain Specific
To test the breadth of Vpu strains that are inhibited by SRI-41897 and SRI-42371, we tested inhibition against a wide variety of available Vpu strains and discovered a strain very similar to NL4-3 Vpu that was resistant to both compounds; the strain is from a simian-human immunodeficiency virus strain, SHIV KU-1vMC33 , containing a Vpu originally from HIV-1 HXB2 [60,61]. SHIV KU-1vMC33 Vpu contains only four amino acid changes from NL4-3 Vpu (Figure 6A,C). Additionally, another more clinically relevant strain of Vpu, CH77, was cloned into a NL4-3 backbone. While approximately 74% identical, there are several differences between CH77 and NL4-3 Vpu ( Figure 6D). The NL4-3, CH77, and SHIV KU-1vMC33 strains of Vpu were examined in the GaLV inhibition assay (introduced in Figure 1A) in the presence and absence of compound or MLN4924 ( Figure 6A). Interestingly, both compounds had no inhibition of CH77 Vpu or SHIV KU-1vMC33 , but the positive control, MLN4924, still functioned to inhibit both Vpu strains. the 5th amino acid from both SHIVKU-1vMC33 Vpu and CH77 Vpu, a proline and tyrosine, respectively, to determine if this deletion could rescue the effects of compounds on these two Vpu strains ( Figure 6C). While deletion of Y5 in CH77 Vpu did not rescue compounds, deletion of P5 in SHIVKU-1vMC33 Vpu rescued SRI-41897 to similar levels as NL4-3 and rescue effects of SRI-41897 to a higher level than seen with NL4-3 Vpu. According to the Los Alamos database, less than 1% of Vpu contain single amino acid deletion between two side-by-side isoleucine residues at the N terminus like with NL4-3 Vpu. This suggests that the compounds would not likely be functional on most strains of Vpu.

Discussion
While still not fully elucidated, the inability to completely clear an HIV-1 infection could largely be due to accessory proteins, such as Vpu, that aid in immune system evasion through mechanisms such as the downmodulation of CD4, BST-2/Tetherin, and inhibitory effects on ADCC. Currently, there are no approved therapies targeting HIV-1 Vpu.
Here we described two small molecules, SRI-41897 and SRI-42371, identified from an HTS, that inhibit the HIV-1NL4-3 Vpu protein. The HTS took advantage of our previous finding that Vpu targets the cytoplasmic tail of GaLV Env using the same SCF/TrCP ubiquitin ligase machinery it uses to downmodulate other targets [16,38,39]. This HTS has the advantage of being a gain-of-function assay reducing the chance of false-negative hits. Additionally, in a gain-of-function cellular screen, compounds must be able to cross the The SHIV KU-1vMC33 Vpu and CH77 Vpu sequences differ in the same place from NL4-3 Vpu in only two places, AA5 and AA61 ( Figure 6D). To examine the effect of this position on SRI-41897 and SRI-42371, a single proline, the same amino acid in position 5 of SHIV KU-1vMC33 was inserted into NL4-3 Vpu after amino acid 4 and tested alongside NL4-3 WT Vpu in the GaLV inhibition assay ( Figure 6B). The single proline insertion was sufficient to completely abrogate the function of both compounds against Vpu. We also deleted the 5th amino acid from both SHIV KU-1vMC33 Vpu and CH77 Vpu, a proline and tyrosine, respectively, to determine if this deletion could rescue the effects of compounds on these two Vpu strains ( Figure 6C). While deletion of Y5 in CH77 Vpu did not rescue compounds, deletion of P5 in SHIV KU-1vMC33 Vpu rescued SRI-41897 to similar levels as NL4-3 and rescue effects of SRI-41897 to a higher level than seen with NL4-3 Vpu. According to the Los Alamos database, less than 1% of Vpu contain single amino acid deletion between two side-by-side isoleucine residues at the N terminus like with NL4-3 Vpu. This suggests that the compounds would not likely be functional on most strains of Vpu.

Discussion
While still not fully elucidated, the inability to completely clear an HIV-1 infection could largely be due to accessory proteins, such as Vpu, that aid in immune system evasion through mechanisms such as the downmodulation of CD4, BST-2/Tetherin, and inhibitory effects on ADCC. Currently, there are no approved therapies targeting HIV-1 Vpu.
Here we described two small molecules, SRI-41897 and SRI-42371, identified from an HTS, that inhibit the HIV-1 NL4-3 Vpu protein. The HTS took advantage of our previous finding that Vpu targets the cytoplasmic tail of GaLV Env using the same SCF/βTrCP ubiquitin ligase machinery it uses to downmodulate other targets [16,38,39]. This HTS has the advantage of being a gain-of-function assay reducing the chance of false-negative hits. Additionally, in a gain-of-function cellular screen, compounds must be able to cross the cell membrane to result in a positive hit, and highly toxic compounds will result in a negative signal.
Of the 21 HTS hits, two compounds were structurally similar, and one of them, SRI-41897, had a strong signal. Over 80 commercially available analogs of SRI-41897 were screened in the GaLV inhibition assay, and a second compound, SRI-42371 more potently inhibited Vpu (Figure 2). While SRI-42371 inhibited Vpu more potently than SRI-41897 in the GaLV inhibition assay, the EC 50 of SRI-42371 was higher. Furthermore, the nitro group present on SRI-41897 raises a concern of potential toxicity (Reviewed in [62]). Thus, SRI-42371 may be considered a better hit for optimization.
Vpu specificity has been a limitation in production of Vpu inhibitors, so it was important to us to test potential off-site effects of SRI-41897 and SRI42371. To do this, compounds of interest were counter screened in a Vpu-independent manner for auto fluorescence (data not shown) and inhibition of the SCF/βTrCP ubiquitin ligase machinery that Vpu depends on at two different levels, broader cullin-family inhibition and the inhibition of cellular βTrCP 1 and 2 ( Figure 3B,C). Both SRI-41897 and SRI-42371 had a negative signal for auto-fluorescence, cullin-family inhibition, and βTrCP inhibition, suggesting that they are specific to Vpu. This is a strength of SRI-41497 and SRI-42371 in comparison to the compound BIT225, which has been shown to increase plasma derived activated CD4 + and CD8 + T cells and NK cells if coupled with ART [34], and is not Vpu specific [33,35].
Next, we showed that SRI-41897 and SRI-42371 rescue CD4 and partial BST-2/Tetherin expression in PBMCs infected with HIV-1 NL4-3 in a Vpu-dependent manner ( Figure 4A,B). This result suggests that these two compounds hinder the functional activity of Vpu, which is an important factor in evasion of the immune system during a HIV-1 infection and is another limitation of the compound BIT225 [36]. Furthermore, in recent years, Vpu has been shown to be an important factor in protecting infected cells from killing by ADCC (reviewed in [25]). Excitingly, we also showed that SRI-41897 and SRI-42371 increase the surface Env expression and killing of infected cells by ADCC ( Figure 5). This finding is important, as the clinical trial of the HIV-1 vaccine ALVAC/AIDSVAX (RV144), despite being only about 30% effective, did not impact viral load or CD4+ cell counts in vaccinated individuals who became infected [63]; however, upon further analysis, it was reported that ADCC may have contributed to the~30% protection reported in the trial because of a reduced risk of infection in vaccinated individuals that had lower IgA titers [64]. The significance of heightened ADCC levels in HIV-1 infection control is also supported by higher ADCC levels reported in elite controllers not on ART [65,66], and by higher ADCC levels being correlated to a lower transmission rate through breastfeeding [67]. For this reason, for a Vpu inhibitor to be successful in a clinical setting, it would likely be necessary for the inhibitor to rescue the ADCC response. It is important to note that part of the effect of Vpu on ADCC protection is linked to its ability to downregulate BST-2/Tetherin [27][28][29]; however, the ability of Vpu to downregulate CD4 + also had a significant effect on ADCC susceptibility [26,68]. This further highlights the importance of a Vpu inhibitor that can rescue at least partial expression of both BST-2/Tetherin and CD4 if it were to be used in a clinical setting.
The positive hit in the HTS, the negative hit in both counter screens, and some CD4, BST-2/Tetherin, and ADCC rescue all suggest a dependence on Vpu. Unfortunately, we discovered that SRI-41897 and SRI-42371, specific to the NL4-3 strain of Vpu are presumed to interact primarily with the N-terminus of Vpu around proline 4 and 5 ( Figure 6). The addition of a proline after isoleucine 4 abrogated the effect of SRI-41897 and SRI-42371. While disappointing, the additional data showing NL4-3 specificity and dependance on a specific amino-acid sequence strongly suggest that both compounds are directly interacting with Vpu towards the N-terminus rather than any other cellular or HIV-1 targets. The finding that SRI-41897 and SRI-42371 interact with the N-terminus before or right at the transmembrane region is interesting because much of the activity of Vpu has been attributed to the cytoplasmic tail, specifically Serine 52/56. However, the transmembrane region downstream of where we hypothesize compounds are interacting, specifically A14/18, has been shown to be important for complete BST-2/Tetherin downmodulation, but this is not the case for CD4 downmodulation [16][17][18][19]. An NMR structure of Vpu has been previously solved, and it is interesting to note that the group that changes in the transmembrane region of Vpu altered the angle of the cytoplasmic tail region [69]. Such alterations could alter the overall structure of Vpu and alter membrane trafficking and/or co-factor interactions. While further studies would be necessary, this is a possible explanation as to why a compound interacting at the N terminus of Vpu may disrupt functions associated with the cytoplasmic tail as changes in the cytoplasmic tail angle could impact binding of the ubiquitin ligase complex.
Even though SRI-41897 and SRI-42371 are NL4-3 specific, these two compounds have the potential to help further the understanding of Vpu and the effect its inhibition has on HIV-1 infection in both cell culture and animal models. Most importantly, we demonstrate here a solid proof-of-concept for high throughput screening for Vpu inhibitors. We were able to pull several small-molecule candidates from a large library of compounds, counter screen candidate compounds, demonstrate functional inhibition, and locate important residues for these compounds to interact with Vpu. This gain-of-function assay may easily be adapted to screen inhibitors for other, more clinically relevant, strains of Vpu.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/v14040817/s1, Excel Document S1: AB Table Excel Tab: Contains the Supplier, Supplier ID number, and, when applicable, the assigned SRI number. 10K Excel Tab: Contains additional information on the first 10K compound screen hits including activity information. 660K Excel Tab: Contains additional information on compound activity, off-target activity, autofluorescence, and CD4 levels. Funding: This work was partially funded by NIH R21 AI136611 and also conducted, in part, by Southern Research using federal funds from the Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health under contract HHSN272201400010I entitled "In Vitro Testing Resource for HIV Therapeutics and Topical Microbicides". This work was partially supported by CIHR foundation grant 352417 to A.F. and by NIH R01 AI148379 to D.T.E. and A.F. A.F. is the recipient of a Canada Research Chair on Retroviral Entry (RCHS0235).

Informed Consent Statement:
Written informed consent was obtained from all study participants and research adhered to the ethical guidelines of CRCHUM and was reviewed and approved by the CRCHUM institutional review board (ethics committee, approval number CE16.164-CA). Research adhered to the standards indicated by the Declaration of Helsinki. All participants were adults and provided informed written consent prior to enrolment in accordance with Institutional Review Board approval.