Limitations of dual‐fluorescent HIV reporter viruses in a model of pre‐activation latency

Abstract Introduction HIV latency can be established in vitro following direct infection of a resting CD4+ T cell (pre‐activation latency) or infection of an activated CD4+ T cell which then returns to a resting state (post‐activation latency). We modified a previously published dual‐fluorescent reporter virus seeking to track the establishment and reactivation of pre‐activation latency in primary CD4+ T cells. Methods A previously published dual‐fluorescent reporter virus was modified so that expression of enhanced green fluorescent protein (GFP) was under control of the elongation factor 1 alpha (EF1α) promoter to detect latent infection, and E2 crimson (E2CRM) was under control of the nef promoter to detect productive infection. NL4.3 that expressed GFP in place of nef was used as a positive control. We infected the Jurkat T‐cell line and primary CD4+ T cells that were either unstimulated or stimulated with either the chemokine CCL19 or phytohaemagglutinin (PHA)/IL‐2 and quantified the expression of both fluorescent proteins by flow cytometry. The study was carried out over a period of two years from September 2016 to October 2018. Results and Discussion Expression of both fluorophores was detected following infection of the Jurkat T‐cell line while only low levels of the latent reporter were observed following infection of primary CD4+ T cells. In unstimulated and CCL19‐treated CD4+ T cells, expression of the GFP latent reporter, increased after further activation of the cells with PHA/phorbol 12‐myristate 13‐acetate (PMA). Conclusions Our findings demonstrate that the EF1α promoter has poor constitutive expression in resting CD4+ T cells. Therefore, dual‐fluorescent reporter viruses with the EF1α promoter may underestimate the frequency of latent infection in resting CD4+ T cells.


| INTRODUCTION
The main barrier to finding a cure for HIV infection is the persistence of latently infected cells in HIV-infected individuals on antiretroviral therapy (ART) [1][2][3]. In vitro, latent infection can be established through multiple pathways: direct infection of resting CD4+ T cells (pre-activation latency [4]), infection of an activated cell which survives and reverts to a resting state (post-activation latency) [5] or infection of a cell transitioning from an effector memory to central memory phenotype [6,7]. Given the low frequency of latently infected cells in vivo [8] insights into latency are often obtained through the use of in vitro models of latent infection established in primary T cells or T-cell lines (reviewed in [9]).
In recent years, dual-fluorescent reporter viruses have emerged as a common tool to study latency in vitro [10][11][12] by including two fluorescent reportersa latent reporter that is under the control of an HIV LTR-independent promoter that is constitutively expressed following integration and a productive reporter that is expressed following transcription from the HIV LTR. Duo-Fluo is a previously reported virus that contains mCherry under the control of the elongation factor 1a (EF1a) promoter as the latent reporter and enhanced green fluorescent protein (GFP) as the reporter for productive infection, both inserted into the nef gene [11]. One potential limitation of the Duo-Fluo virus is the possibility of recombination of the two reporters during reverse transcription due to the matching amino and carboxyl termini [13]. This limitation was recently addressed in a new virus HIV GKO , whereby mCherry was replaced with Kusabira orange (mKO2) and GFP was switched to a codon-switched GFP protein (csGFP) [14].
In order to enhance infectivity in primary cells and to enhance the detection of latent virus established by direct infection (pre-activation latency), we designed a new dual-fluorescent reporter virus. We modified Duo-Fluo and restored env and vpr to wild type sequences and changed the reporter proteins to GFP (latent) and E2 crimson (productive infection), naming it DuoAdvance. Following direct infection of resting cells stimulated with CCL19 [4,15], we found low constitutive expression of GFP (the latent reporter), despite successfully establishing latent infection. We conclude that dual-fluorescent reporter viruses may underestimate the frequency of latently infected cells in resting CD4+ T cells.

| DuoAdvance plasmid generation
A 6668bp DNA plasmid insert containing the modified DuoAdvance sequence was synthesized by GenScript (Gen-Script, Piscataway, NJ). The DuoAdvance DNA plasmid insert was ligated into the Duo-Fluo plasmid using T4 DNA ligase. Plasmids were prepared using a Maxi Kit (QIAGEN, Hilden, Germany) and ligation confirmed using combined NdeI and PacI digestion for 2.5 hours at 37°C followed by 1 hour at 50°C with BssHII, resolved on a 1.2% gel.

| Preparation of virus stocks and determining
infectivity Viral stocks of DuoAdvance, Duo-Fluo and NL4.3-EGFP were prepared from transfection of HEK293T cells with each respective plasmid, concentrated and used in all experiments as previously described [15].
The infectivity of viral stocks was performed using a TZM-bl assay. TZM-bl cells were plated and infected with 10-fold dilutions of viral stocks in triplicate. Negative and positive control wells using media and a reference virus with known infectivity were also included respectively. 50% tissue culture infectious dose (TCID 50 ) was measured using luciferase readout and calculated using the Reed-Muench method [16].

| Infection of Jurkat T-cell line
Jurkat T cells were infected with concentrated DuoAdvance virus, Duo-Fluo virus, NL4.3-EGFP virus at a TCID 50 of 0.5 in an infection volume of 100 lL. The integrase inhibitor Raltegravir (1 lmol/L) or non-nucleoside reverse transcriptase inhibitor Nevirapine (10 lmol/L) (Selleck Chemicals, Houston, TX) were added to cells 30 minutes before infection and then maintained in cultures throughout the remaining experiment. Cells were infected for 2 hours, washed, cultured for three days and then fixed in 1% formaldehyde for flow cytometry. Samples were analysed on a BD LSR Fortessa using FACSDIVA software (BD Biosciences, San Jose, CA). Data were analysed using FlowJo v10 (FlowJo LLC, Ashland, OR).

| Isolation and infection of resting CD4+ T cells
Peripheral blood mononuclear cells (PBMCs) and resting CD4+ T cells were isolated from whole blood from healthy donors as described previously [4] and cultured in three different conditions; unactivated, 100 nmol/L recombinant human CCL19/MIP-3b (In Vitro Technologies, Vic, AUS) treated and PHA/IL-2 activated. Untreated and treated primary CD4+ T cells were typically infected with 100 µL of DuoAdvance or NL4.3-EGFP (both at TCID 50 0.5-1) per 1 9 10 6 cells for 2 hours. Cells were also spinoculated (12009g, 90 minutes, room temperature) followed by a further 30 minutes incubation. Cells were then washed and resuspended in RF10 plus 1 U/mL IL-2 and cultured for six days.
For reactivation studies, the E2CRMÀ fraction of cells were sorted out from CCL19-treated CD4+ T cells infected with DuoAdvance using a BD FACSAria Fusion (BD, Franklin Lakes, NJ) and then reactivated by the addition of 10 lg/mL PHA+10 nmol/L PMA, or DMSO diluent negative control, for three days in the presence of 1 lmol/L raltegravir. After reactivation, cells were fixed in 1% formaldehyde and run on a BD LSR Fortessa using bandpass filters for E2CRM (670/30 Yellow) and GFP (530/30 Blue) plus FACSDIVA software. Results were analysed using FlowJo v10.

| Data analysis
FlowJo results were graphed using GraphPad Prism (version 7, GraphPad Software, La Jolla, CA). Statistical testing was not appropriate as the small sample size (n < 6) precluded the use of non-parametric tests. Due to the small sample size, testing normality cannot be performed, thus parametric t-tests were also not used. Additionally, as the percentage fluorescence data points appeared skewed, this further suggests that a parametric t-test was not appropriate. Therefore, we have only reported the mean, range and fold-difference between means.

| Minimal constitutive expression of EF1a following infection of resting cells
We next used DuoAdvance to infect CCL19-treated and PHA/IL-2-activated CD4+ T cells. E2CRM was detected in both models and this declined when nevirapine was added to the cell culture prior to, during and after infection. As expected, the frequency of E2CRM+ cells was higher in PHA/ IL-2 activated (mean (range) = 2.55 (0.83 to 5.14)%) compared to CCL19-treated cells (mean (range) = 0.11 (0.01 to 0.28)%) (Figure 2A,C and D). Low frequencies of GFP (latent infection) were detected in both CCL19-treated (mean (range) = 0.01 (0.00 to 0.01)%) and PHA/IL-2-activated CD4+ T cells (mean (range) = 0.034 (0.02 to 0.05)%) (Figure 2A-B). These GFP frequencies were higher than mock infection and when infection was performed in the presence of nevirapine. Spinoculation of the DuoAdvance virus at the time of infection of primary T cells was also tested to enhance viral binding and infection [17]; however, we found little change in the number of infected cells following spinoculation ( Figure 2B,C, closed symbols) or when we increased the TCID 50 from 0.5 to 1 (data not shown) to enhance infection.
The low frequency of latently infected cells observed following infection of primary resting CD4+ T cells with DuoAdvance may be explained by inefficient expression of the EF1a-driven GFP latent reporter in primary resting CD4+ T cells, in contrast to expression in the Jurkat T-cell line ( Figure 1). Promoter silencing may account for the low expression levels in resting CD4+ T cells compared to the actively dividing Jurkat T-cell line. A key feature of resting CD4+ T cells is sequestration of transcription factors NF-jB and nuclear factor of activated T cell (NFAT) [18]. These transcription factors are important for the initiation of transcription of HIV and may also be important for initiation of transcription of the latent GFP reporter from the EF1a promoter. The downregulation of antiretroviral defence genes such as SAM domain and HD domain-containing protein 1 (SAMHD1) and guanylate-binding protein 5 (GBP5) in Jurkat T cells could further contribute to the success of viral infection in cell lines with these virus strains compared to primary CD4+ T cells, where these genes are not downregulated [19]. Additionally, the recognition of CG dinucleotides in viral RNA by the cellular zinc-finger antiviral protein (ZAP) could have contributed to depleted cytoplasmic viral RNA and impaired viral expression in primary resting CD4+ T cells [20].

| Inducible virus is detected in E2CRM negative cells
We next performed a reactivation experiment ( Figure 3A  To test for inducible virus, the E2CRMÀ population was isolated on day 6 from CCL19-treated infected cells and reactivated with PMA/PHA for three days. PMA/PHA led to an increase in the number of E2CRM+ cells 3.1-fold versus the DMSO diluent control ( Figure 3D). This revealed that the E2CRM cells sorted six days post-infection harboured inducible virus, even though these cells had low to absent GFP expression prior to stimulation. Interestingly, the level of GFP+E2CRMÀ latent cells also increased 2.4-fold with PMA/ PHA reactivation compared to the DMSO diluent control (Figure 3D), suggesting that the latent reporter may need the cell to be in an activated state to effectively express GFP protein.
We do not believe that this increase in GFP was a result of integration of unintegrated DNA following T cell activation as reactivation experiments were performed in the presence of an integrase inhibitor.
In our experiments, the expression of the EF1a-driven GFP clearly increased following T cell activation. It is worth noting that we were unable to determine if the inefficient expression of GFP in resting T cells was a feature of the EF1a promoter itself or a consequence of the EF1a promoter in the background of the DuoAdvance construct. However, we propose that the use of the EF1a promoter in a dual reporter virus might significantly underestimate the frequency of latently infected cells, particularly when using a pre-activation model of resting CD4+ T cell latency.
In activated primary CD4+ T cells, low expression of the latent reporter GFP was observed despite efficient expression of the HIV LTR-driven E2CRM (Figure 2) might be explained by promoter competition. This can occur when a promoter that has a higher affinity for a limiting factor competitively inhibits transcription from an adjacent promoter [21]. For example, once the HIV LTR becomes activated by Tat, there may be preferential recruitment of transcription factors to the HIV LTR promoter driving E2CRM expression at the expense of the EF1a promoter and GFP expression, thereby suppressing GFP and resulting in a higher proportion of E2CRM+ cells. Similar promoter competition has been reported in other model systems such as human embryonic stem cells transduced with a dual-promoter vector [22] and HeLa cells co-transfected with HIV and HIV Rev binding aptamer under the control of the human cytomegalovirus (CMV) promoter [23].
Collectively, these experiments demonstrate that while DuoAdvance can successfully infect the Jurkat T-cell lines and establish latent and productive infection, following infection of primary CD4+ T cells, only E2 Crimson expression was detected and there was poor constitutive expression of the EF1a promoter as detected by GFP expression. Expression of E2 Crimson and GFP could be induced following activation of sorted E2CRM-CCL19 CD4+ T cells suggesting integration did occur in these cells, but GFP had poor constitutive expression and was not expressed in all cells that contained inducible virus.
Interestingly, Chavez and colleagues noted using Duo-Fluo, that in activated T cells returning to a resting state, the expression of both latent and productive infection fluorescent markers were progressively lost over time [12]. The second generation HIV GKO variant of Duo-Fluo has only been used to infect activated T cells to date and expression of the new EF1a-driven mKO2 latency reporter has not yet been examined in resting CD4+ T cells [14]. Another dual-fluorescent reporter virus (HIfate-E) has some similarities to DuoAdvance [24]. In this construct the latency reporter was changed to GFP and the productive reporter to E2 Crimson and a composite EF1a-HTLV-1 promoter was used in place of the EF1a promoter to drive the GFP latency reporter. Following infection of activated CD4+ T cells with HIfate-E, insufficient latently infected cells were generated to assess the response to further activation [24], suggesting a very low yield of latently infected cells despite successful infection of activated CD4+ T cells similar to this study. Further investigation with these two new viruses, HIV GKO or HIfate-E, in direct infection of resting CD4+ T cells will elucidate whether these viruses are suited to studying HIV latency in pre-activation models.
Developing a sensitive and specific marker of latently infected cells using an in vitro model of a dual reporter virus to infect primary CD4 T cells is of high interest as this would allow for examination of host gene changes following latency using techniques such as scRNAseq as these studies cannot be performed with immortalized cell lines, and activation of the cell to express viral proteins changes the host transcriptome, as recently reported [25]. Using DuoAdvance, if latency is defined by GFP expression but not all latently infected cells express GFP, this would underestimate the frequency of latently infected cells. If these cells were being used to screen for LRAs, this would potentially overestimate the efficiency of a particular LRA. In addition, screening would miss latently infected cells that do not react to an LRA.
Adding an internal ribosome entry site (IRES) between the EF1a promoter and the latency reporter might lead to the improved expression of the latency reporter in resting CD4+ T cells by enhancing the reporter gene activity as previously CD4+ T cells were isolated from PBMC and left resting (unstimulated, Un) or stimulated with either CCL19 or PHA/IL-2 and then infected with either DuoAdvance (DA) virus or media only (mock) (A). After six days, expression of both fluorescent proteins was quantified by flow cytometry (B-D). The E2 Crimson negative cell population (cells that were not present in the E2CRM+ gate in B top panel) was sorted from the CCL19-treated samples six days after infection and then stimulated with either DMSO or PMA/PHA in the presence of 1 lM raltegravir and expression of both fluorescent proteins quantified by flow cytometry (D). Each symbol represents a different donor and the column represents the mean. described [26]. This approach could be pursued to drive better expression of the latency reporter in resting CD4+ T cells to study HIV latency and reactivation in resting CD4+ T cells [27].
Finally, a limitation of this study was the number of donors analysed. We examined six or fewer donors for each experiment and therefore could not determine whether the data was normally distributed. In preference to using parametric tests, as is often used for small samples, we have only reported the mean fold changes in different experimental conditions and not reported p values.

| CONCLUSIONS
We generated a modified dual-fluorescent reporter virus that was able to successfully infect a Jurkat T-cell line and generate latent and productively infected cell populations. However, DuoAdvance resulted in low levels of expression of the latency reporter in resting CD4+ T cells, using a pre-activation model of HIV latency. As current dual-fluorescent reporters are usually studied by infecting activated CD4+ T cells [11,12,14], and expression of latent and productive infection reporters can be progressively lost as activated cells return to a resting state [12], dual-fluorescent reporter viruses may significantly underestimate the frequency of latently infected cells in vitro.

A C K N O W L E D G E M E N T S
We thank Prof. Damain Purcell and Dr. Georges Khoury (The University of Melbourne) for assistance with generating the DuoAdvance construct. We thank Flinders Medical Centre, Prof. Ken Shortman (Walter and Eliza Hall Institute of Medical Research), Prof. Anthony Jaworowski (Royal Melbourne Institute of Technology) and the former Ralph Steinman lab for antibodies used in the antibody cocktail for resting CD4+ T cell isolation. We also thank the Doherty Institute Flow Cytometry Facility with assistance with cell sorting and flow cytometry and healthy donors from Red Cross Australia for providing blood cells. We would also like to thank Dr. David Price (The University of Melbourne) with assistance with statistical advice.

F U N D I N G
This work was funded by the National Health and Medical Research Council of Australia (SRL).