Upstream Stimulatory Factors Regulate HIV-1 Latency and Are Required for Robust T Cell Activation

HIV-1 provirus expression is controlled by signaling pathways that are responsive to T cell receptor engagement, including those involving Ras and downstream protein kinases. The induction of transcription from the HIV-1 LTR in response to Ras signaling requires binding of the Ras-responsive element binding factor (RBF-2) to conserved cis elements flanking the enhancer region, designated RBE3 and RBE1. RBF-2 is composed minimally of the USF1, USF2, and TFII-I transcription factors. We recently determined that TFII-I regulates transcriptional elongation from the LTR through recruitment of the co-activator TRIM24. However, the function of USF1 and USF2 for this effect are uncharacterized. Here, we find that genetic deletion of USF2 but not USF1 in T cells inhibits HIV-1 expression. The loss of USF2 caused a reduction in expression of the USF1 protein, an effect that was not associated with decreased USF1 mRNA abundance. USF1 and USF2 were previously shown to exist predominately as heterodimers and to cooperatively regulate target genes. To examine cooperativity between these factors, we performed RNA-seq analysis of T cell lines bearing knockouts of the genes encoding these factors. In untreated cells, we found limited evidence of coordinated global gene regulation between USF1 and USF2. In contrast, we observed a high degree of genome-wide cooperative regulation of RNA expression between these factors in cells stimulated with the combination of PMA and ionomycin. In particular, we found that the deletion of USF1 or USF2 restricted T cell activation response. These observations indicate that USF2, but not USF1, is crucial for HIV-1 expression, while the combined function of these factors is required for a robust T cell inflammatory response.


Introduction
Despite decades of research, the etiological agent of acquired immunodeficiency syndrome (AIDS), the human immunodeficiency virus (HIV-1), remains a pressing global healthcare issue as nearly 40 million individuals are infected globally and close to a million deaths occur annually [1]. Current antiretroviral therapy (ART) typically limits HIV-1 replication in most individuals to a chronic asymptomatic infection [2]. However, ART treatment must be maintained for the lifespan of infected individuals to prevent a rebound of viral replication from the latently infected cells [3], and often altered due to the emergence of drug resistant HIV-1 variants [4]. The barrier to a cure for HIV-1/AIDS is the population of infected CD4 + T memory cells that establish proviral latency during early infection. This extremely long-lived latently-infected cell population provides a reservoir of viremia that evades host immune surveillance and is invulnerable to ART [4,5]. Various proposed strategies directed towards eliminating latently-infected cells involve modulating the expression of latent provirus [3]. Often referred to as "shock and kill" and "block and lock", these strategies employ latency reversing and latency promoting agents, respectively, and have been the subject of intense recent investigation [6,7]. However, initial trials using latency reversing agents in patients during antiretroviral therapy did not significantly reduce the pool of latently infected CD4 + T cells in vivo [3,8]. Consequently, it is recognized that treatment with latency modulating agents must be applied with consideration of effects on the global immune response, including the function of CD8 + T cells [9,10]. Overall, the development of effective therapies that are based on modulating provirus expression requires a more detailed understating of the mechanisms underpinning HIV-1 latency.
The 5 long terminal repeat (LTR) serves as the HIV-1 promoter and possesses numerous cis elements that bind host cell transcription factors [11]. Multiple cis elements within the 5 LTR bind transcription factors that are regulated downstream of T cell receptor signaling, causing provirus expression to be tightly linked to CD4 receptor engagement [12]. The Ras-responsive factor binding elements three and one (RBE3 and RBE1) are highly conserved on the HIV-1 provirus in patients who develop AIDS [13] ( Figure 1A) and were initially characterized as required for induction of the HIV-1 LTR in response to Ras and MAPK signaling [14]. These conserved elements bind the RBF-2 complex that is comprised minimally of the TFII-I, USF1, and USF2 proteins [15]. Additionally, RBF-2 was found to be associated with Yin Yang 1 (YY1) at the upstream RBE3 element, which plays a role in establishing and enforcing latency [16]. Mutations of RBE1 and RBE3 abrogate the binding of RBF-2 in vivo and render the provirus incapable of transcriptional induction in response to T cell signaling [15,[17][18][19]. This effect is mediated at least partially by the recruitment of the co-factor TRIM24 to the HIV-1 LTR by TFII-I, resulting in enhanced transcriptional elongation [20,21].
The upstream stimulatory factor (USF) was initially identified as a transcriptional regulator which binds an E-box of the adenovirus major late promoter (AdMLP) [22]. This factor predominately exists as a heterodimer of two structurally related helix-loop-helix leucine zipper (b-HLH-LZ) proteins of 43 kDa and 44 kDa, designated USF1 and USF2 [23]. The USF factors are ubiquitously expressed, although relative ratios in abundance vary depending on the tissue type [24,25]. USF1 and USF2 share 44% sequence identity overall, with the C-terminal b-HLH-LZ region displaying~70% conservation [25]. Given their similarities in sequence and their effect for activation of the AdMLP [23], in addition to USF1/USF2 heterodimers predominating within cells [26], the functional differences between these proteins have largely been overlooked. However, USF1 and USF2 knockout mice displayed unique phenotypes, likely representing differential gene expression [27], indicating that these factors have distinct functions. USF1 and USF2 are associated with both positive and negative effects on transcription that are dependent upon the target promoter context and cellular differentiation state [27,28], where the temporal and spatial nature of the USF function generates differential effects in specific cell types [29][30][31]. The activity of the USFs has been shown to be controlled by signaling pathways through phosphorylation [32] which regulates both the DNA binding function and protein-protein interactions [33,34]. The phosphorylation of USF2 by CDK5 is thought to contribute to carcinogenesis [35], while the CK2-dependent phosphorylation of USF1 causes altered expression of genes involved in metabolism [36]. Moreover, T cell activation induced by the phorbol ester PMA causes extensive phosphorylation of both USF1 and USF2, although the regulatory implications of these modifications have not been determined [15].
USF was among the first cellular factors shown to bind the HIV-1 promoter at an E-box element located at −160 on the LAI subtype B LTR [37,38]. This upstream E-box was found to be conserved on only~15% of LTRs from provirus in samples from HIV-1 patients who developed symptoms [13]. The effect of USF for HIV-1 expression was subsequently found to be predominately mediated as a component of RBF-2. The binding of USF1/2 to the RBE3 and RBE1 elements, positioned at −130 and −22 ( Figure 1A), is stimulated by interaction with TFII-I [18,19]. Mutations within the core RBE3 element (ACTGCTGA, Figure 1A) or immediate 3' sequences prevent reactivation of HIV-1 in response to T cell signaling, but also abrogates binding of at least four transcription factors to this region of the LTR, including USF1, USF2, TFII-I, and YY1 [15,16,18]. Consequently, the overall effect of USF1 and USF2 for the regulation of HIV-1 expression has not been clarified. Furthermore, although USF1/2 have been reported to be regulated by multiple different protein kinases [32], the role of these factors for the regulation of signal-responsive genes in T cells has not been characterized. As core components of the RBF-2 complex, we propose that the USF proteins play a critical role for regulation of HIV-1 expression, at least in part due to recruitment of TFII-I and TRIM24 to the highly conserved RBE1 and RBE3 LTR elements. Consequently, in this study, we examined the effect(s) caused by the genetic depletion of USF1 and USF2 on HIV-1 expression in T cells, and the effect on global gene expression in response to T cell signaling.
The 5′ long terminal repeat (LTR) serves as the HIV-1 promoter and possess merous cis elements that bind host cell transcription factors [11]. Multiple cis ele within the 5′ LTR bind transcription factors that are regulated downstream of T cell tor signaling, causing provirus expression to be tightly linked to CD4 receptor e ment [12]. The Ras-responsive factor binding elements three and one (RBE3 and RBE highly conserved on the HIV-1 provirus in patients who develop AIDS [13] (Figu and were initially characterized as required for induction of the HIV-1 LTR in respo Ras and MAPK signaling [14]. These conserved elements bind the RBF-2 complex comprised minimally of the TFII-I, USF1, and USF2 proteins [15]. Additionally, RBF found to be associated with Yin Yang 1 (YY1) at the upstream RBE3 element, which a role in establishing and enforcing latency [16]. Mutations of RBE1 and RBE3 ab the binding of RBF-2 in vivo and render the provirus incapable of transcriptional tion in response to T cell signaling [15,[17][18][19]. This effect is mediated at least partia the recruitment of the co-factor TRIM24 to the HIV-1 LTR by TFII-I, resulting in enh transcriptional elongation [20,21]. In this study, we demonstrate that HIV-1 provirus expression was inhibited upon the depletion of USF2, whereas the loss of USF1 had no significant effect. Additionally, the loss of USF2 resulted in a reduction in USF1 protein expression, an effect that did not correlate with USF1 mRNA abundance. Furthermore, we found that USF1 and USF2 were key factors required for T cell activation, as the loss of either protein caused global constraint of T cell signal-induced genes. The USF2 gene knockout limited T cell activation to a greater extent than the USF1 knockout, and additionally rendered T cells more restrictive to HIV-1 infection. These observations clarified the role of USF1 and USF2 for regulation of HIV-1 transcription and identified these proteins as key signal regulated factors that play a vital role in activation of the T cell inflammatory response.

Cell and Virus Culture
Jurkat E6-1, Jurkat Tat mHIV-Luciferase, TZM-bl, and HEK293T cells were cultured under standard conditions of 37 • C and a 5% CO 2 atmosphere. The Jurkat cells were cultured in RPMI-1640 media while HEK293T and TZM-bl were cultured in DMEM, supplemented with 10% FBS, penicillin [100 units/]mL, streptomycin [100 g/]mL, and Lglutamine [2 mM]. The vesicular stomatitis virus G (VSV-G) pseudotyped viral stocks were produced by co-transfecting HEK293T cells with a combination of viral molecular clone, psPAX, and pVSV-G at a ratio of 8 µg:4 µg:2 µg. The transfections were performed using polyethylenimine (PEI) at a ratio of 6:1 (PEI:DNA) in Gibco Opti-MEM tm . The lentiviral infections were performed by plating 1 × 10 6 cells in 24-well plates with media containing 8 µg/mL polybrene and the amount of viral supernatant to give the desired multiplicity of infection (M.O.I.) as indicated. The plates were subsequently spinoculated for 1.5 h at 1500 rpm.

Luciferase Expression Assays
For the Jurkat Tat mHIV-Luciferase expression assays, 1 × 10 5 of cells were plated with 100 µL media in 96-well plates. The luciferase activity was measured after the indicated time of treatment. The measurements were performed using the Superlight™ luciferase reporter Gene Assay Kit (BioAssay Systems, Hayward, CA, USA) as per the manufacturer's instructions. The 96 well plates were read using a Victor™ X3 Multilabel Plate Reader.

Transfection Assays
A total of 6.67 × 10 5 WT or USF2 KO HEK293T cells were plated in 2 mL DMEM in sixwell plates. The following day, the cells were transfected with 0.5 ng of the indicated LAI LTR-GFP reporter construct and 1 µg of the indicated EF1α-RFP construct. The transfections were performed using polyethylenimine (PEI) at a ratio of 1:4 (DNA:PEI) in Gibco Opti-MEM™. At one day post-transfection, the cells were detached using trypsin-EDTA and analyzed using flow cytometry.

Flow Cytometry
The cells were treated as indicated in the Figure legends. Following the indicated treatment, Jurkat-derived cells were suspended in PBS while the HEK293T-derived cells were suspended in PBS containing 10% trypsin-EDTA to prevent aggregation. Flow cytometric analysis was performed using a BD Biosciences LSRII-561 instrument where the threshold forward scatter (FSC) and side scatter (SSC) parameters were set so that a homogenous population of live cells was counted ( Figure S8). FlowJo software (v10.8.1) (TreeStar, Hendersonville, TN, USA) was used to analyze the data and determine the indicated mean fluorescent intensity (MFI).

RT-PCR
RNA was extracted from Jurkat Tat mHIV-Luciferase cells following the indicated treatment using the RNeasy Kit (Qiagen). The RNA was analyzed using the Quant Studio 3 Real-Time PCR system (Applied Biosystems) and the Power SYBR ® Green RNA-to-CT™ 1-Step Kit (Thermo Fisher) as per the manufacturer's instructions. The RT-PCR data was normalized to GAPDH expression using the ∆∆Ct method. The primers used were USF1 Fwd 5 GCTCTATGGAGAGCACCAAGTC, Rev 5 AGACAAGCGGTGGTTACTCTGC; and GAPDH Fwd 5 TGCACCACCAACTGCTTAGC, Rev 5 GGCATGGACTGTGGTCATGAG.

RNA-Seq
RNA was extracted from the wildtype, USF1 KO, or USF2 KO Jurkat Tat mHIV-Luciferase human T cells that were left untreated or stimulated with PMA/ionomycin for 4 h. Sample quality control was performed using an Agilent 2100 Bioanalyzer. The qualifying samples were prepped using the NEBnext Ultra ii Stranded mRNA (New England Biolabs) standard protocol. Sequencing was performed using an Illumina NextSeq 500 with Paired End 42 bp × 42 bp reads. The demultiplexed read sequences were uploaded to the Galaxy web platform [39] and aligned to the hg38 reference genome using STAR. Subsequently, the transcript assembly was accomplished using featureCounts and differential gene expression (DEG) was determined by DESeq2 analysis. The DEG was defined as having a fold change in the expression > 1.5 and p-value < 0.05 as to the compared condition. Volcano plots were generated on the Galaxy web platform and gene expression heatmaps were created in RStudio.

Statistics and Reproducibility
All the replicates represent independent biological samples, and the results are presented as mean values ± the standard deviation, as shown by error bars. The number of times an experiment was performed is indicated in the Figure legends. The p-values were determined by performing unpaired sample t-tests with the use of GraphPad Prism 9.0.0. The statistical significance is indicated as * p < 0.05, ** p < 0.01, or *** p < 0.001, with n.s. denoting a non-significant p ≥ 0.05.

Regulation of the HIV-1 Expression by USF1 and USF2
The 5 LTR serves as the HIV-1 promoter and contains several cis elements that bind USF1/2. The highly conserved RBE1 and RBE3 elements are both occupied by the RBF-2 complex that is minimally composed of USF1/USF2 and TFII-I [19]. Additionally, USF heterodimers bind an upstream E-Box present on the prototypical LAI LTR ( Figure 1A) [15]. Previous studies implicated USF1 and/or USF2 as regulators of HIV-1 expression, mostly based on transient transfection assays and LTR mutations that abolish binding of these factors [15,19,38,40,41]. However, interpretation of effects caused by mutations of the RBE1 and RBE3 elements is complicated by the fact that they also impair binding of at least TFII-I and YY1 [16,18,19]. Consequently, to examine the specific effects of USF1 and USF2 on HIV-1 expression, we produced USF1 and USF2 gene knockouts (KO) ( Figure 1C,D) in the Jurkat Tat mHIV-Luciferase T cell line bearing an integrated HIV-1 mini-virus where luciferase expression is driven by the 5 LTR ( Figure 1B) [42]. Surprisingly, we noted that USF1 protein abundance was reduced in all the USF2 knockout lines ( Figure 1D), whereas USF2 protein expression was not affected by the USF1 knockout ( Figure 1C). We examined the activation of HIV-1 expression in these knockout lines and found that the USF1 disruption tended to cause a slight increase in luciferase expression in unstimulated as well as activated conditions, as mediated by the T cell signaling agonist PMA ( Figure 1E). In contrast, provirus expression was diminished in both unstimulated and stimulated conditions in all the USF2 knockout lines ( Figure 1F). Consistent with results using the knockout lines, we found that shRNA mediated depletion of USF1 ( Figure S1A) had no effect on HIV-1 provirus expression ( Figure S1B), while shRNA knockdown of USF2 ( Figure S1C) inhibited HIV-1 expression in both untreated cells and cells stimulated with the phorbol ester PMA ( Figure S1D). Additionally, we found that TZM-bl cells were refractory to HIV-1 induction upon USF2 depletion ( Figure S2), confirming the requirement of USF2 for the full expression of chromosomally integrated HIV-1.
As USF1 protein expression was dependent upon USF2 in Jurkat T cells, we examined the effect of the USF2 gene knockout in HEK293T cells where we also observed a correlation between USF1 and USF2 protein expression ( Figure 2A). Next, we used an LAI-derived LTR-Tat-IRES-GFP reporter construct to assess the effect of the USF2 knockout on HIV-1 expression in these cells ( Figure S3A,B). Consistent with the above results, we observed significantly lower GFP expression in the HEK293T USF2 KO cells relative to the wildtype (WT) ( Figure 2B). Remarkably, we found that co-transfection of the HEK293T USF2 KO cells with vectors expressing either USF1 or USF2 ( Figure 2C) and the HIV-1 LTR-Tat-IRES-GFP reporter not only rescued LTR transcription but also caused a greater activation of GFP expression than that in wildtype cells ( Figures 2D and S3C). These observations suggest that any combination of USF1 and USF2 can activate transcription from the HIV-1 LTR, but because USF1 protein expression was decreased in the USF2 knockout lines while the USF2 expression was unaffected by the USF1 depletion, only the USF2 knockouts produce a significant effect on HIV-1 LTR transcription. Of note, the knockout of either USF1 or USF2 was not associated with any noticeable change in the cell phenotype, viability, or growth ( Figure S4).
to HIV-1 induction upon USF2 depletion ( Figure S2), confirming the requirement of USF2 for the full expression of chromosomally integrated HIV-1.
As USF1 protein expression was dependent upon USF2 in Jurkat T cells, we examined the effect of the USF2 gene knockout in HEK293T cells where we also observed a correlation between USF1 and USF2 protein expression ( Figure 2A). Next, we used an LAI-derived LTR-Tat-IRES-GFP reporter construct to assess the effect of the USF2 knockout on HIV-1 expression in these cells ( Figure S3A,B). Consistent with the above results, we observed significantly lower GFP expression in the HEK293T USF2 KO cells relative to the wildtype (WT) ( Figure 2B). Remarkably, we found that co-transfection of the HEK293T USF2 KO cells with vectors expressing either USF1 or USF2 ( Figure 2C) and the HIV-1 LTR-Tat-IRES-GFP reporter not only rescued LTR transcription but also caused a greater activation of GFP expression than that in wildtype cells ( Figures 2D and S3C). These observations suggest that any combination of USF1 and USF2 can activate transcription from the HIV-1 LTR, but because USF1 protein expression was decreased in the USF2 knockout lines while the USF2 expression was unaffected by the USF1 depletion, only the USF2 knockouts produce a significant effect on HIV-1 LTR transcription. Of note, the knockout of either USF1 or USF2 was not associated with any noticeable change in the cell phenotype, viability, or growth ( Figure S4).   Figure S3A). One day post-transfection, the GFP mean fluorescent intensity (MFI) of the RFP expressing population was determined using flow cytometry and normalized to the CMV-EV-EF1α-RFP (EV) transfected wildtype cells (n = 3, mean ± SD).

Effect of USF1 and USF2 on Response of HIV-1 to Latency Reversing Agents and Tat
We examined the effect of the USF1 and USF2 knockouts on reactivation of HIV-1 in response to various mechanistically diverse latency reversing agents (LRAs). In T cells, the phorbol ester PMA activates protein kinase C (PKC) and Ras-MAPK signaling, while ionomycin causes the release of intracellular calcium, which activates calcineurin. Consequently, treatment with the combination of PMA and ionomycin mimics T cell activation produced by CD4 receptor engagement [43]. We found that USF2 KO inhibited HIV-1 provirus activation in response to PMA or ionomycin, individually or in combination ( Figure 3A). In contrast, USF1 KO caused a slight increase in response to the combination of PMA/ionomycin or PMA alone ( Figure 3A). The effect of the USF1 and USF2 knockouts in response to PMA and ionomycin was most pronounced at 6 h post-treatment ( Figure 3B). As observed previously, we found that HIV-1 expression became exhausted upon prolonged treatment (20 h) with PMA/ionomycin in WT cells [15,17], and this effect was not significantly different in the USF1 and USF2 knockout lines ( Figure 3B). Interestingly, the loss of USF1 caused slightly decreased induction upon treatment with ionomycin alone ( Figure 3A), indicating that USF1 was not required for HIV-1 reactivation in response to PKC and Ras-MAPK signaling but may play a role for activation by calcium-calcineurin signaling. Ingenol 3-angelate (PEP005) also activated PKC signaling [44], and similar to the effect of PMA treatment, we observed a significantly lower response to PEP005 in the USF2 KO cells but no effect in the USF1 KO line ( Figure 3A). These results indicate that USF2 is required for reactivation of HIV-1 in response to the PKC, Ras-MAPK, and calcineurin T cell signaling pathways, whereas USF1 may cause some inhibition in response to MAPK activation.
The latent HIV-1 provirus can be reactivated independently of T cell signaling by altering the epigenetic composition or interaction of bromodomain-containing proteins with the LTR. Treatment with the histone deacetylase inhibitor (HDACi) suberanilohydroxamic acid (SAHA) [45] modestly increased HIV-1 expression, an effect that was limited by USF1 depletion and almost completely abrogated by the USF2 knockout ( Figure 3A). We also examined the effect of the USF knockouts on reactivation in response to the bromodomain inhibitors JQ1 and IACS-9571. JQ1 is a well-studied LRA that inhibits the BRD4 bromodomain, causing increased Tat transactivation [46], while IACS-9571 facilitates transcriptional elongation by targeting TRIM24 to the LTR [20,21]. We found that the USF2 knockout significantly limited the effect of JQ1, whereas USF1 KO had no effect ( Figure 3A). The USF2 knockout, and to a lesser extent USF1 KO, inhibited reactivation of HIV-1 expression in cells treated with IACS-9571 ( Figure 3A), suggesting that both USFs impact regulation of HIV-1 by TRIM24.
We also examined the effect of the USF2 gene knockout on activation of HIV-1 expression by the viral transactivator protein Tat. For this experiment, we transfected the USF2 knockout HEK293T lines (Figure 2A) with HIV-1 LTR-Tat-IRES-GFP (+Tat) or LTR-GFP reporter plasmids (-Tat) ( Figure S5A,B). We observed that GFP expression from both the HIV-1 LTR reporter plasmids was significantly lower in the USF2 knockout cells compared to the WT ( Figures 3C and S5C,D). These observations indicate that full activation of HIV-1 transcription by Tat requires USF2 protein, although USF2 also has a significant effect on HIV-1 expression in the absence of Tat.

USF2 Stabilizes USF1 Protein
We observed that depletion of USF2 in Jurkat T cells ( Figures 1D and S1C) or HEK293T cells (Figure 2A) invariably caused a corresponding decrease in USF1 protein abundance. To examine the mechanism by which USF1 protein abundance was dependent upon USF2, we generated a Jurkat USF2 KO cell line where Flag-tagged USF2 was expressed using a doxycycline-inducible promoter. We found that doxycycline-induced expression of USF2-Flag in this USF2 KO cell line rescued the expression of USF1 protein to greater levels compared to WT cells ( Figure 4A). We also observed that USF2 KO did not cause a decrease in USF1 mRNA abundance compared to WT cells ( Figure 4B), regardless of the T cell activation state ( Figure 4C), indicating that differences in USF1 expression in USF2 KO cells must relate to protein stability or effects on mRNA translation. Furthermore, stimulation with PMA/ionomycin did not alter USF1 or USF2 protein abundance in Jurkat T cells, and USF1 protein levels were decreased in the USF2 knockout cells regardless of the cellular activation state ( Figure 4D). The latent HIV-1 provirus can be reactivated independently of T cell signaling by altering the epigenetic composition or interaction of bromodomain-containing proteins with the LTR. Treatment with the histone deacetylase inhibitor (HDACi) suberanilohydroxamic acid (SAHA) [45] modestly increased HIV-1 expression, an effect that was limited by USF1 depletion and almost completely abrogated by the USF2 knockout ( Figure  3A). We also examined the effect of the USF knockouts on reactivation in response to the bromodomain inhibitors JQ1 and IACS-9571. JQ1 is a well-studied LRA that inhibits the BRD4 bromodomain, causing increased Tat transactivation [46], while IACS-9571 facilitates transcriptional elongation by targeting TRIM24 to the LTR [20,21]. We found that the

Effect of USF Knockouts on the Interaction of RBF-2 Proteins with the LTR
We examined the requirements of USF1 and USF2 for binding to the HIV-1 LTR using chromatin immunoprecipitation and quantitative PCR (ChIP-qPCR) analysis ( Figure 5A). Since we were unable to immunoprecipitate USF1 with antibodies against native protein, we generated mHIV-Luciferase Jurkat USF1 and USF2 KO cell lines that stably express Myc-tagged USF1 ( Figure 5B). We found that USF1-Myc was associated with the LTR in both the USF1 and USF2 KO cell lines ( Figure 5C). However, we did observe slightly higher amounts of USF1-Myc associated with all three sites on the LTR in the USF1 knockout line ( Figure 5C), which expressed endogenous USF2 and has slightly more USF1-Myc expression than the USF2 null cells ( Figure 5B). Similarly, we found that USF2 was associated with the LTR at equivalent levels in both the wildtype and USF1 knockout cells

Effect of USF Knockouts on the Interaction of RBF-2 Proteins with the LTR
We examined the requirements of USF1 and USF2 for binding to the HIV-1 LTR using chromatin immunoprecipitation and quantitative PCR (ChIP-qPCR) analysis ( Figure 5A). Since we were unable to immunoprecipitate USF1 with antibodies against native protein, we generated mHIV-Luciferase Jurkat USF1 and USF2 KO cell lines that stably express Myc-tagged USF1 ( Figure 5B). We found that USF1-Myc was associated with the LTR in both the USF1 and USF2 KO cell lines ( Figure 5C). However, we did observe slightly higher amounts of USF1-Myc associated with all three sites on the LTR in the USF1 knockout line ( Figure 5C), which expressed endogenous USF2 and has slightly more USF1-Myc expression than the USF2 null cells ( Figure 5B). Similarly, we found that USF2 was associated with the LTR at equivalent levels in both the wildtype and USF1 knockout cells ( Figure 5D). Furthermore, PMA/ionomycin mediated T cell activation had no effect on USF2 binding at RBE3 or RBE1 ( Figure 5E), which is consistent with the previous observation that TFII-I is constitutively associated with the HIV-1 LTR [20]. These results indicate that USF1 and USF2 are both capable of binding specific cis elements on the LTR as homodimers in vivo, a result that is consistent with binding of recombinant USF1 and USF2 proteins in vitro [15,18]. Thus, although USF1 and USF2 were predominately shown to function as heterodimers in vivo [24,25], USF2 is capable of binding independently of USF1 at multiple sites on the HIV-1 LTR for the activation of transcription. In contrast, as indicated above, USF1 protein abundance is dependent upon USF2, and therefore is incapable of regulating HIV-1 transcription in the USF2 knockout cells.
Viruses 2023, 15, x FOR PEER REVIEW 12 of 28 ( Figure 5D). Furthermore, PMA/ionomycin mediated T cell activation had no effect on USF2 binding at RBE3 or RBE1 ( Figure 5E), which is consistent with the previous observation that TFII-I is constitutively associated with the HIV-1 LTR [20]. These results indicate that USF1 and USF2 are both capable of binding specific cis elements on the LTR as homodimers in vivo, a result that is consistent with binding of recombinant USF1 and USF2 proteins in vitro [15,18]. Thus, although USF1 and USF2 were predominately shown to function as heterodimers in vivo [24,25], USF2 is capable of binding independently of USF1 at multiple sites on the HIV-1 LTR for the activation of transcription. In contrast, as indicated above, USF1 protein abundance is dependent upon USF2, and therefore is incapable of regulating HIV-1 transcription in the USF2 knockout cells.  Next, we sought to examine requirements of the RBF-2 proteins for binding to the HIV-1 LTR. We used lentiviral-delivered shRNA to knockdown the expression of TFII-I (GTF2I) in the Jurkat mHIV-Luciferase cell line ( Figure 1B), where we observed a partial reduction in TFII-I protein ( Figure 6A). Consistent with previous results [20], even with a partial knockdown of TFII-I, we observed reduced expression of HIV-1 provirus in both the unstimulated and stimulated cells ( Figure 6B), an effect that was similar in the USF2 KO cells ( Figure 1F). Interestingly, we observed no significant change in the binding of USF2 to the RBE elements in cells with a partial TFII-I knockdown ( Figure 6C). In contrast, binding of TFII-I to the LTR was completely abrogated in the USF2 KO cells ( Figure 6D). Similarly, we also observed a complete loss of the co-activator TRIM24 at the LTR in the USF2 KO cells ( Figure 6E), which is consistent with our observation that this factor is recruited to the LTR by TFII-I [20]. Collectively, these results indicate that USF proteins are required for RBF-2 complex formation in vivo, including the recruitment of TFII-I to the LTR.
Next, we sought to examine requirements of the RBF-2 proteins for binding to the HIV-1 LTR. We used lentiviral-delivered shRNA to knockdown the expression of TFII-I (GTF2I) in the Jurkat mHIV-Luciferase cell line ( Figure 1B), where we observed a partial reduction in TFII-I protein ( Figure 6A). Consistent with previous results [20], even with a partial knockdown of TFII-I, we observed reduced expression of HIV-1 provirus in both the unstimulated and stimulated cells ( Figure 6B), an effect that was similar in the USF2 KO cells ( Figure 1F). Interestingly, we observed no significant change in the binding of USF2 to the RBE elements in cells with a partial TFII-I knockdown ( Figure 6C). In contrast, binding of TFII-I to the LTR was completely abrogated in the USF2 KO cells ( Figure 6D). Similarly, we also observed a complete loss of the co-activator TRIM24 at the LTR in the USF2 KO cells (Figure 6E), which is consistent with our observation that this factor is recruited to the LTR by TFII-I [20]. Collectively, these results indicate that USF proteins are required for RBF-2 complex formation in vivo, including the recruitment of TFII-I to the LTR.

Global Regulation of Transcription by USF1 and USF2 in Unstimulated T Cells
Given the surprising observation that Jurkat T cells with USF1 and USF2 gene knockouts have distinct effects on the HIV-1 expression, we sought to determine the degree to which these factors cooperatively regulate global gene expression. To examine this, we performed RNA-seq on cell lines bearing USF1 or USF2 gene knockouts. In this analysis, we identified 236 differentially expressed genes (DEG) upon USF1 depletion and 563 DEG following USF2 disruption ( Figure 7A,B). Of the genes affected by the loss of USF1, slightly more were downregulated than upregulated (146 to 90, respectively) ( Figure 7A), while the USF2 depletion primarily resulted in decreased expression of genes (410 downregulated to 153 upregulated) ( Figure 7B). Comparison of transcriptomes from the USF1 and USF2 KO cells demonstrated some overlap amongst the downregulated genes with 34% of the USF1 KO downregulated genes similarly regulated in the USF2 KO cells ( Figure 7C). However, we observed only 7% of upregulated genes upon USF1 depletion that were similarly upregulated in the USF2 knockout cells ( Figure 7D). Overall, the USF1 and USF2 knockouts caused a surprisingly small overlap in differential gene regulatory effects in unstimulated Jurkat T cells ( Figure 7E). Furthermore, gene set enrichment analysis (GSEA) indicated only one pathway with less than 25% false discovery rate (FDR) that was shared between the untreated USF1/2 KO cells, "TNFA signaling via NFκB", for which the target genes were downregulated ( Figure 7F,G). These results indicated that USF1 and USF2 share regulatory effects for a minor subset of genes in unstimulated T cells.

USF1 and USF2 Cooperatively Regulate Global Gene Expression in Activated T Cells
USF1 and USF2 were reported to be signal-regulated transcription factors [33,34,36,47] that are phosphorylated upon T cell activation [15]. Consequently, we also examined global gene expression in the USF1 and USF2 knockout lines using RNA-seq from cells stimulated with PMA and ionomycin co-treatment. This analysis revealed a greater effect on global expression in the stimulated cells compared to unstimulated cells, with 1846 and 2667 DEG upon loss of USF1 and USF2, respectively ( Figure 8A,B). Importantly, depletion of either USF1 or USF2 resulted in the majority of DEG having decreased expression rather than increased expression ( Figure 8A,B). These results support the contention that USF1 and USF2 are T cell signal regulated factors that predominately behave as transcriptional activators.
Additionally, we observed a significantly greater proportion of genes that were similarly affected by USF1 KO and USF2 KO in activated T cells. In stimulated cells, we found that 68% of the USF1 KO downregulated and 38% of USF1 KO upregulated genes were similarly mis-regulated upon USF2 KO ( Figure 8C,D). Furthermore, a minority of the USF1 DEG were found to be mis-regulated opposite that of the USF2 null cells ( Figure 8E). We performed GSEA analysis on the USF1 and USF2 KO T cell transcriptomes to determine the predominant affected pathways, which revealed genes downstream of "TNFA signaling via NFκB", "hypoxia", and "interferon gamma response" as being repressed ( Figure 8F), while genes associated with "oxidative phosphorylation" and "Myc targets V1" were upregulated in both the USF1 KO and USF2 KO cell lines ( Figure 8G). Although genes from multiple pathways were similarly affected in the USF1 and USF2 KO lines, we observed an asymmetrical regulation between these factors for several pathways, including "mTORC1" and "Notch signaling", which were downregulated solely in the USF1 and USF2 knockout cells, respectively ( Figure 8F). Interestingly, USF1 was previously implicated in regulating the UV and tanning response [47][48][49]. However, we found that genes involved in "UV response down" were repressed in cells that were deficient in either USF1 or USF2 ( Figure 8F), suggesting that both factors mediate this process. Taken together, these results indicate that USF1 and USF2 display a high degree of cooperativity for global gene regulation during T cell activation, but that specific subsets of genes appear to be regulated independently by these factors. As noted previously, the loss of USF2 results in a corresponding decrease in USF1 expression, and therefore it is possible that USF2 may partially compensate for the loss of USF1 protein in both unstimulated and stimulated T cells.

USF1 and USF2 Cooperatively Regulate Global Gene Expression in Activated T Cells
USF1 and USF2 were reported to be signal-regulated transcription factors [33,34,36,47] that are phosphorylated upon T cell activation [15]. Consequently, we also examined global gene expression in the USF1 and USF2 knockout lines using RNA-seq from cells stimulated with PMA and ionomycin co-treatment. This analysis revealed a greater effect on global expression in the stimulated cells compared to unstimulated cells, with 1846 and 2667 DEG upon loss of USF1 and USF2, respectively ( Figure 8A,B).

USF1 and USF2 Facilitated T Cell Activation
Having observed that loss of either USF1 or USF2 generated an abundance of downregulated genes upon T cell activation relative to wildtype cells, we examined the potential role of these factors for regulation of the inflammatory response. From the RNA-seq analysis, we identified 2790 downregulated and 3481 upregulated genes in response to T cell activation ( Figure 9A). Notably, the genes that were induced in response to stimulation displayed a much greater range of expression compared to genes that were inhibited ( Figure 9A). We then examined the effect of the USF1 and USF2 gene knockouts on expression of genes that were activated or repressed upon T cell stimulation. Here, we found that 755 (22%) and 1114 (32%) genes induced upon T cell stimulation showed significantly lower expression in the absence of USF1 and USF2, respectively ( Figure 9B). The genes that were repressed upon T cell activation were much less affected by the USF disruption, with only 187 (7%) and 272 (10%) displaying greater expression upon the loss of USF1 or USF2, respectively ( Figure 9B). These results are consistent with the above observations, indicating the role of USF1 and USF2 as transcriptional activators.
T cell activation was previously shown to alter the expression of numerous genes with a wide degree of variance [50]. This is consistent with our results indicating that nearly 50% of genes induced upon stimulation showed less than a three-fold change in their expression ( Figure 9C). Consequently, it is possible that genes displaying a large variance in induction are regulated by different mechanism(s). To examine this with respect to USF1 and USF2, we segmented the genes induced upon T cell activation into three groups by the level of induction, designated low (fold change in expression between 1.5 and three), mid (fold change in expression between three and eight), and high (fold change greater than eight) ( Figure 9C). By stratifying the genes in this way, it became apparent that genes that were highly expressed in response to T cell stimulation were more potently inhibited in the USF1 or USF2 KO cells ( Figure 9D). Furthermore, we found that more of the genes that were downregulated upon USF1 or USF2 depletion were highly induced in response to T cell activation ( Figure 9E). Additionally, we found that the USF1 and USF2 knockout cells displayed a partially defective activation of genes downstream of key T cell signaling pathways, including TNFα and Ras ( Figure S6). Collectively, these results indicate that USF1 and USF2 are signal-regulated factors required for full induction of genes upon T cell activation, a finding that is consistent with the evolved capability of the HIV-1 LTR to appropriate transcription factors downstream of T cell signaling [51].

USF2 Deficient Cells Are Refractory for HIV-1 Infection
HIV-1 infection of resting CD4 + T cells is inefficient compared to infection of activated cells [52][53][54]. Although Jurkat cells are actively dividing, they too are preferentially infected when activated [55]. As HIV-1 favors the infection of stimulated T cells, and that the loss of USF1 and USF2 dampens T cell activation, we sought to examine the role of these factors for HIV-1 infection. For this purpose, we created USF1 and USF2 knockouts in Jurkat T cells that did not contain a previously integrated provirus. Consistent with the results shown above, we observed the loss of USF1 expression in both the USF1 and USF2 knockout lines ( Figure 10A). These cell lines were infected with the Red-Green-HIV-1 (RGH) [56] or HIV GKO [57] dual-reporter viruses where GFP is expressed from the 5 LTR and the internal constitutive promoters express mCherry or mKO2 in RGH and HIV GKO , respectively ( Figure 10B). These reporter HIV-1 derivatives allow for the differentiation of productive and latent infections using flow cytometry ( Figure 10C,D) [56]. We found that infection of the USF1 knockout cells produced a small but not significant decrease in RGH ( Figure 10E) or HIV GKO ( Figure 10F) infectivity compared to the wildtype, as determined 4 days post-infection. In contrast, the USF2 knockout cells were infected at a rate roughly half that of the wildtype cells by either reporter virus ( Figure 10C-F). Additionally, infection of the USF knockout Jurkat Tat mHIV-Luciferase cell lines ( Figure 1C,D) with a Red-Blue-HIV-1 (RBH) reporter virus similarly indicated that USF2 depletion caused the cells to be more resistant to infection ( Figure S7). Interestingly, although USF2 was required for reactivation of latent provirus, we found that USF2 KO did not affect the proportion of cells that established latency four days following infection ( Figure 10G,H). Taken together, these observations indicate that USF2 is important for infection of T cells by HIV-1 through a process that may involve the cellular activation state. tial role of these factors for regulation of the inflammatory response. From the RNA-s analysis, we identified 2790 downregulated and 3481 upregulated genes in response to cell activation ( Figure 9A). Notably, the genes that were induced in response to stimu tion displayed a much greater range of expression compared to genes that were inhibit ( Figure 9A). We then examined the effect of the USF1 and USF2 gene knockouts on e pression of genes that were activated or repressed upon T cell stimulation. Here, we foun that 755 (22%) and 1114 (32%) genes induced upon T cell stimulation showed significant lower expression in the absence of USF1 and USF2, respectively ( Figure 9B). The gen that were repressed upon T cell activation were much less affected by the USF disruptio with only 187 (7%) and 272 (10%) displaying greater expression upon the loss of USF1 USF2, respectively ( Figure 9B). These results are consistent with the above observation indicating the role of USF1 and USF2 as transcriptional activators.

USF2 Promoted Productive HIV-1 Infection
We found that the USF2 knockout inhibited reactivation of latent HIV-1 provirus (Figure 1) but did not affect the proportion of latently infected cells at four days postinfection ( Figure 10G,H). To further examine the role of USF for regulating latency, we monitored the proportion of productively infected cells for three weeks following infection of wildtype, USF1 KO, and USF2 KO cells. Previous observations indicated that the internal constitutively expressed mCherry and mKO2 reporters in cells infected with RGH and HIV GKO becomes silenced following extended culture [56,57]. Consequently, we isolated the infected populations using fluorescent activated cell sorting (FACS) two days postinfection ( Figure 11A). Then, the WT, USF1 KO, and USF2 KO infected cells were analyzed using flow cytometry over the following 22 days (Figure 11B,C). Similar to the results with unsorted infected cells (Figure 10), we found that the WT, USF1, and USF2 deficient cells produced a nearly equivalent proportion of productively infected cells at 3 and 4 days post-infection ( Figure 11D,E). However, after 4 days of infection, the USF2 knockout cells accumulated latent HIV-1 infections at a higher proportion compared to the WT or USF1 KO cells ( Figure 11D,E). These results indicate that USF2 promotes the expression of HIV-1 and encourages productive replication upon infection.

Discussion
The upstream stimulatory factors (USFs) were among the first proteins identified as sequence-specific DNA binding factors required for activation of transcription in eukaryotes [22,24], and among the first human factors discovered to bind the 5 HIV-1 LTR [38]. Previous observations indicated that USF1 and USF2 regulate transcription in response to multiple signaling pathways involving protein kinases, including at least MAPK/ERKs [32,58], casein kinase 2 (CK2) [34], CDK2 [59], protein kinase C (PKC) [60], DNA-dependent protein kinase (DNAPK) [61,62], protein kinase A (PKA) [63], and glycogen synthase kinase-3 (GSK-3) [62]. These protein kinases represent a significant fraction of the signaling networks in human cells, which implicate USF1 and USF2 as key regulatory factors involved in coordinating gene expression in response to diverse cellular stimuli. USF1 and USF2 are ubiquitously expressed in human cells [25], and consequently, regulation of their function is likely cell-type specific. However, mechanistic details regarding regulation of USF1/2 function by signaling mechanisms are sparse. Nevertheless, our results are consistent with these observations, and indicate that these factors play a significant role in the regulation of the signal-responsive gene expression during T cell activation. USF2, in particular, is required for the induction of a significant proportion of highly induced genes in response to T cell signaling ( Figure 9). USF1/2 were found to bind the prototypical LAI HIV-1 LTR at an E-box element in the upstream modulatory region at −160 to −165 from the transcriptional start site [63], where it was shown to activate transcription in cooperation with Ets-1 [41]. This cis element is not well conserved on LTRs from infected individuals who developed symptoms of AIDS [13]. USF1 and USF2 were subsequently found to form part of the factor RBF-2, which binds two highly conserved cis elements at −130 (RBE3) and −20 (RBE1) [15,19]. The RBE3/1 sites were initially identified due to their requirement for activation of HIV-1 transcription by activated Ras in T cells [14]. The RBF-2 complex was subsequently purified from Jurkat nuclear extracts using oligonucleotide affinity [64], and was found to contain USF1, USF2, and TFII-I [15]. The RBE3 element is not typical of other characterized binding sites for USF1/2 (ACTGCTGA, Figure 1A), and the interaction of USF with this site in vitro is stimulated upon addition of TFII-I [15,18]. In contrast, the RBE1 site (CAGCTGC) [14] positioned near the transcriptional start resembles a typical E-box element (CANNTG, Figure 1A), and consequently, binding of USF1/2 to this site is less dependent upon TFII-I [19]. Mutations of these elements significantly inhibit activation of HIV-1 expression in response to T cell receptor cross linking or the combination of PMA and ionomycin [15,18,19]. Consistent with these previous observations, we found that the disruption of USF2 inhibited the expression of HIV-1 in response to T cell signaling as well as a variety of latency reversing agents (Figures 1 and 3). Furthermore, the USF2 disruption limited the initial incidence of HIV-1 infection ( Figure 10) and reduced the proportion of infected cells that underwent productive gene expression ( Figure 11). We note that our analysis of the effects of USF1 and USF2 knockouts were predominately examined in the Jurkat T cell line, and it will be important to examine effects caused by depletion of these factors in normal CD4 + T cells.
The mechanism by which the factors bound at RBE3 and/or RBE1 cause activation in response to T cell signaling has not been elucidated. However, TFII-I, in association with the serum response factor (SRF), was shown to regulate expression of the c-fos promoter in response to Ras-stimulated MAPK signaling by direct interaction with ERK1 [65,66]. Stimulation of T cells by antigen presentation causes activation of multiple pathways downstream of the T cell receptor, including the Ras-Raf-Mek-Erk (MAPK) pathway [17]. Consequently, the role of USF1/2 for activation of HIV-1 transcription likely includes, at least in part, interaction with TFII-I for binding the conserved RBE1 and RBE3 sites on the LTR [18]. Consistent with this possibility, we observed loss of TFII-I binding to the LTR in cells with the USF2 disruption ( Figure 6). Additionally, as mentioned above, USF1 and USF2 were reported to be regulated by multiple protein kinases, and it seems likely that these proteins mediate additional function(s) to regulate HIV-1 activation in response to T cell stimulation. Accordingly, we found that USF1, USF2, and TFII-I become hyperphosphorylated in T cells stimulated with PMA and ionomycin [15]. Interestingly, however, we note that in USF2 knockout cells, the HIV-1 provirus remained significantly responsive to the PKC agonists PMA and PEP005 (Figures 1F and 3A). This indicates that additional factors capable of binding to the LTR, potentially including AP1, GABP, and Ets [11], may at least partially compensate for the loss of RBF-2 components, demonstrating a particular resilience of this viral promoter for maintaining coordinate regulation with T cell signaling. Despite that USF1 and USF2 are thought to predominately exist as heterodimers in most cell types [25,27], we found that disruption of their genes caused significantly different alterations in the transcriptome and where we observed a greater regulatory overlap in stimulated cells relative to unstimulated cells (Figures 7 and 8). Interestingly, we note distinct factor-specific effects on genes involved in mTORC1 and Notch signaling in activated T cells ( Figure 8F). This is consistent with previous observations indicating discrete functions for USF1 and USF2. For example, USF2 was associated with progression of leukemia through upregulation of HOXA9, but disruption of USF1 alone had minimal effects [31]. Overexpression of USF2 was linked to rheumatoid arthritis that is refractory to treatment through upregulation of proinflammatory cytokines [30], while stabilization of USF2 protein by CDK5-dependent phosphorylation promotes prostate cancer proliferation [36]. Additionally, differential binding of USF1 and USF2 to the plasminogen activator inhibitor type-1 promoter regulates expression during the G 0 to G 1 cell cycle transition [67,68]. The C-terminal DNA binding domain of USF1 and USF2 share~70% sequence identity. However, the N-terminal regions have limited sequence homology [25,27]. Consequently, since these factors are capable of binding the same E-box element [23], either alone or as heterodimers, the specific independent function(s) of these factors has not been fully characterized. Importantly, analysis of the differential effects of USF1 and USF2 for expression of specific genes is complicated by the fact that USF2 is required for USF1 protein expression, at least in the T cell and HEK293T lines. Considering this, it is remarkable that the USF1 and USF2 gene knockouts produced notably different transcriptome phenotypes. It will be interesting to characterize the mechanistic differences producing these distinct effects for gene regulation, which may involve differential functions conferred by binding USF2 homodimers relative to USF1/USF2 heterodimers.
We found that USF1 protein levels were significantly reduced upon USF2 knockout or shRNA knockdown in Jurkat T cells and HEK293T cells. USF2 depletion did not reduce USF1 mRNA expression in either unstimulated or stimulated T cells (Figure 4), and consequently, the loss of USF2 must cause USF1 protein instability or inhibit translation of its mRNA. Reintroduction of Flag-tagged USF2 in USF2 null cells restored USF1 protein levels ( Figure 4A). We propose that since USF1 and USF2 predominately exist as heterodimers [25], this interaction stabilizes the USF1 protein, and consequently, loss of USF2 may result in USF1 protein instability. We tried to examine this possibility, but this analysis was complicated by the finding that USF1 stability was not enhanced by the proteasome inhibitor MG132, or expression of the C-terminal region of USF2 containing the b-HLH leucine zipper motif (not shown). Nevertheless, such a mechanism regulating stability of heterodimer partners has been described previously for the yeast MATa and α2 proteins, which are mutually stabilized by their heterodimer formation [69].
Our results have revealed novel properties of the ubiquitously expressed USF proteins and clarify their role for regulation of HIV-1 provirus expression. Aberrant expression of USF2 was previously linked to additional human diseases, including rheumatoid arthritis [30] and cancer [32]. Consequently, a more detailed understanding of the function and regulation of these factors will provide significant insight towards mechanisms that sculpt the transcriptome in response to extracellular signaling. USF proteins have been implicated in the regulation of chromatin modifications and insulator functions [70,71]. Therefore, it will be particularly important to understand the role of USF2 for epigenetic regulation and chromatin organization at the HIV-1 LTR and cellular target genes, as well as the relative contribution of associated factors, including TFII-I for the signal-responsive gene expression.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/v15071470/s1, Figure S1: shRNA knockdown of USF2 inhibits HIV-1 expression; Figure S2: USF2 knockout limits proviral induction in TZM-bl cells; Figure S3: Reincorporation of USF1 or USF2 rescues LTR transcription in USF2 KO cells; Figure S4: Loss of USF1 or USF2 does not affect cellular growth; Figure S5: USF2 regulates HIV expression regardless of Tat; Figure S6: Loss of either USF1 or USF2 impairs expression of genes downstream of multiple T cell activation pathways; Figure   Data Availability Statement: All the data supporting the findings of this study are available within the article or from the corresponding author upon reasonable request (I. Sadowski, ijs.ubc@gmail.com). All the high-throughput RNA-seq data have been deposited with NCBI GEO, accession GSE227850.