SHMT2 and the BRCC36/BRISC deubiquitinase regulate HIV-1 Tat K63-ubiquitylation and destruction by autophagy

HIV-1 Tat is a key regulator of viral transcription, however little is known about the mechanisms that control its turnover in T cells. Here we use a novel proteomics technique, called DiffPOP, to identify the molecular target of JIB-04, a small molecule compound that potently and selectively blocks HIV-1 Tat expression, transactivation, and virus replication in T cell lines. Mass-spectrometry analysis of whole-cell extracts from 2D10 Jurkat T cells revealed that JIB-04 targets Serine Hydroxymethyltransferase 2 (SHMT2), a regulator of glycine biosynthesis and an adaptor for the BRCC36 K63Ub-specific deubiquitinase in the BRISC complex. Importantly, knockdown of SHMT1,2 or BRCC36, or exposure of cells to JIB-04, strongly increased Tat K63Ub-dependent destruction via autophagy. Moreover, point mutation of multiple lysines in Tat, or knockdown of BRCC36 or SHMT1,2, was sufficient to prevent destruction of Tat by JIB-04. We conclude that HIV-1 Tat levels are regulated through K63Ub-selective autophagy mediated through SHMT1,2 and the BRCC36 deubiquitinase.


Introduction
Although proteins or misfolded protein aggregates destined for autophagy are often modified by high levels of K63-ubiquitylation (K63Ub) [25][26][27], the interaction of Tat with the autophagy factor SQSTM1/p62 was found to be independent of ubiquitin [21], raising the question of whether ubiquitin has a significant role in the control of Tat stability, and which factors act as adaptors for Tat lysosomal proteolysis and destruction.
Here we report that HIV-1 Tat protein levels in HeLa and T cells are strongly affected by the histone demethylase inhibitor, JIB-04 [28]. Unexpectedly, knockdown of putative histone demethylase targets did not recapitulate the effect of JIB-04 on Tat expression. To better understand how JIB-04 controls Tat stability, we used a novel mass-spectrometry approach, called DiffPOP, which can identify protein:drug conjugates via their differential solubility in methanol. Through this approach, we identified the serine hydroxymethyltransferase enzyme, SHMT2, as a direct target of JIB-04 in Jurkat 2D10 T cells. SHMT2, and the related SHMT1 protein, are metabolic enzymes required for nuclear thymidylate biosynthesis during S phase [29]. In addition, SHMT2 was recently shown to deliver substrates to the cytoplasmic BRCC36/BRISC complex specifically for removal of K63 ubiquitylation [30,31]. In this latter role, SHMT2 was found to stabilize the type 1 interferon (IFN) receptor chain 1 (IFNAR1) in interferon-signaling cells [32]. In particular, the removal of K63Ub by BRCC36 was sufficient to protect IFNAR1 from lysosomal degradation during autophagy. Here we show that SHMT1,2 and the BRCC36/BRISC deubiquitase complex are novel and selective regulators of Tat K63, but not K48, ubiquitylation, and its degradation through selective autophagy in T cells.

JIB-04 inhibits HIV-1 Tat expression and transactivation
Based on reports that HIV-1 is regulated by histone demethylases [33,34], we examined the effect of histone demethylase inhibitors on HIV-1 Tat transactivation in 2D10 T cells, a Jurkatderived cell line containing a single integrated HIV-1 provirus reporter gene [5]. Among several compounds tested (JIB-04, GSKJ1, ML324, IOX1 and 2-PCPA), only the pyridine hydrazine JIB-04 [28] strongly blocked Tat expression at low concentrations ( Fig 1A and S1D Fig). Importantly, JIB-04 also potently blocked expression of the d2EGFP reporter gene inserted into the viral Nef open reading frame in this cell line (S1A Fig). Parallel experiments revealed that JIB-04 similarly suppressed induction of Tat in response to PMA/PHA in 2D10 cells ( Fig  1B), indicating that the effect of this compound is not stimulus or pathway-specific. Further analysis revealed that JIB-04 blocked Tat expression in a dose-dependent manner ( Fig 1C). By contrast, Tat expression in activated 2D10 cells was unaffected by a different histone demethylase inhibitor, GSKJ1 (only a minor reduction at 50 μM, Fig 1D). Detailed analysis of the effects of JIB-04 on Tat and eGFP protein expression established an approximate IC50 of 0.75 μM in 2D10 T cells (Fig 1E). Exposure of TNFα-treated 2D10 T cells to JIB-04 (3 μM) strongly blocked expression of HIV-1-encoded Env and d2EGFP mRNAs (Fig 1F and S1C Fig). By contrast, JIB-04 had no effect on the level of CDK9 mRNA, or induction of the CXCL10 gene by TNFα. Together, these findings suggest that JIB-04 strongly blocks Tat expression and transactivation of the integrated HIV-1 LTR in 2D10 T cells.
To assess how JIB-04 affects binding of transcription factors to the HIV-1 promoter, we carried out ChIP experiments at the integrated HIV-1 genome in 2D10 cells in the presence and absence of the drug (Fig 2 and S2 Fig). As expected, TNFα treatment led to the rapid recruitment of NF-κB and RNAPII (Fig 2A and 2C, pink line), whereas P-TEFb (CyclinT1 and CDK9 panels) occupancy was observed only at later times (Fig 2A, light blue line), following T cell activation and accumulation of Tat. As we showed previously [10], the binding of NF-κB  [7][8][9][10][11][12] overnight. Cyclin T1 served as loading control. (B) As in part A, except that 2D10 cells were activated with PHA (10 μg/mL) and PMA (50 ng/mL) for the times indicated above each lane. (C) Immunoblot analysis of HIV-1 Tat protein levels in 2D10 cells treated with different concentrations of JIB-04 (0.1 μM-50 μM) for 24 h, as indicated. Cyclin T1 and TAF4 (TFIID subunit) served as loading controls. (D) As in part C, except that cells were exposed to the histone demethylase inhibitor GSKJ1, at the concentrations listed above each lane. Cyclin T1 and p65 (NF-κB subunit) served as loading controls. (E) Quantification of Tat and eGFP protein levels in TNFα-stimulated 2D10 cells treated with the JIB-04 at the concentrations indicated in the X-axis. The signals were calculated by Image J and averaged from three independent experiments. (F) qRT-PCR analysis d2EGFP (HIV-1 reporter gene), HIV-1 Env, or host cell CXCL10 and CDK9 mRNAs extracted from 2D10 cells pre-treated with either DMSO or 3 μM JIB-04 for 16 h and stimulated by TNFα (10 ng/ml), for the times indicated below each graph. Values on the Y-axis for d2EGFP, Env, CXCL10 and CDK9 mRNAs prior to TNFα stimulation were normalized to 1. Differences between DMSO and RNAPII is accompanied by recruitment of the NELF negative elongation factor to establish the paused RNAPII complex (Fig 2A, pink line), and NELF is displaced at later times upon binding of Tat and P-TEFb (light blue line). JIB-04 did not affect the binding of NF-κB    Fig 2A and 2B, pink line). Levels of total RNAPII at the HIV-1 promoter did not decrease following TNFα stimulation in JIB-04 treated cells, as they do in DMSO-treated cells, indicating a block to elongation (compare Fig 2C and 2D, light blue line). Consistent with the rapid loss of Tat expression, JIB-04 treatment dramatically reduced P-TEFb occupancy at the HIV-1 promoter (Fig 2B, light blue line). Consistent with this possibility, dramatically reduced levels of elongation-competent RNAPII CTD-Ser2P and -Ser7P were observed within the coding region of the provirus in JIB-04 treated cells (compare Fig 2C and 2D, light blue line). Furthermore, the release of NELF-A from HIV-1 promoter did not occur in JIB-04 treated cells (compare Fig 2A and 2B, light blue line). Taken together, these data show that JIB-04 strongly blocks Tat-dependent transcription elongation in vivo.
These observations raised the question of whether JIB-04 inhibits histone demethylase enzymes required for HIV-1 transcription, or whether it can directly or indirectly inhibit P-TEFb similar to compounds like flavopiridol, which inhibits host cell transcription and induces a strong host cell stress response [35,36]. To assess this question, 2D10 cells were treated with JIB-04 in the presence or absence of TNFα, and total RNA was extracted for analysis by next-generation RNA sequencing (RNA-seq). As expected, the reads mapping to the HIV-1 genome and to eGFP were dramatically reduced in JIB-04 treated cells upon activation by TNFα (Fig 3A), consistent with the loss of Tat protein expression in these cells. By contrast, the effect of JIB-04 on host cell mRNA levels was quite modest. Overall, JIB-04 affected the expression of only 811 genes by more than 2-fold (p<0.05), out of 13546 genes monitored. Of the JIB-04-affected genes, only 413 (51%) were up-regulated and 398 (49%) were down-regulated ( S3A Fig and S1 File). Gene ontology analysis indicated that the most avidly repressed host cell genes included canonical histone genes, and that many of the genes activated by JIB-04 were associated with glycolysis ( Fig 3B, S3B (Fig 3C and S3C Fig). Taken together, these findings indicate that JIB-04 does not inhibit transcription broadly, nor does it affect signaling through the NF-κB pathway.
We next tested whether JIB-04 blocks virus replication in the HeLa P4.R5 MAGI. indicator cell line. This cell line contains an integrated β-galactosidase reporter under the control of the HIV-1 LTR promoter, and infected cells can be visualized by staining with an X-gal substrate, which turns blue in the presence of β-galactosidase. For this experiment, we analyzed JIB-04 at concentrations (1-5 μM) that did not strongly affect cell viability. Untreated cells or cells treated with JIB-04 were challenged with active Nef-deleted HIV-1 (pNL4-3), as described previously [37]. As shown in Fig 4A and 4B, JIB-04 significantly reduced the percentage of HIVinfected blue cells (from 73% to 22%) with little effect on HeLa P4.R5 MAGI. cell viability, indicating the compound strongly inhibits virus replication.
We next analyzed the effect of JIB-04 in primary CD4+ T cells isolated from peripheral blood from a healthy donor. Primary CD4+ T cells were infected with wild-type NL4-3 virus and treated with DMSO or JIB-04 at various concentrations (S4 Fig). Although JIB-04 efficiently inhibited HIV replication, we noted that cell viability decreased sharply, indicating that the compound is toxic to primary T cells. As expected, the integrase inhibitor Raltegravir blocked HIV-1 replication in this assay, without affecting cell viability. Taken together, these data indicated that JIB-04 is a potent inhibitor of Tat expression and activity in cell lines, but is not suitable for analysis in primary T cells. Based on these findings, and the observation that JIB-04 does not broadly affect host cell transcription or CDK9/ P-TEFb expression (Fig 3), we sought to identify additional targets for JIB-04 in 2D10 cells. To achieve this goal, we adapted a technique currently in development at The Salk Mass Spectrometry Core and Yates laboratory (TSRI, La Jolla, CA), to identify drug interaction targets in an unbiased manner. This approach, termed DiffPOP, is based on the differential solubility of drug:protein conjugates as compared to unconjugated proteins in the presence of methanol gradients. In brief, 2D10 whole-cell extracts were treated with either DMSO or JIB-04, and the soluble proteins were precipitated in a stepwise fashion in fractions containing increasing concentrations of methanol ( Fig 5A). Proteins present in the precipitates of each fraction were then identified using mass-spectrometry.
This approach established that JIB-04 has no effect on the solubility of the vast majority of proteins in the 2D10 cell extracts. However, a few proteins were found to precipitate differently in extracts treated with JIB-04, as determined by analysis of the Pearson's correlation coefficient ( Fig 5B and S2 File). The top host cell protein target identified by this technique was the host cell serine hydroxymethyltransferase enzyme, SHMT2 ( Fig 5C). SHMT2, and the related SHMT1 enzyme, are multifunctional proteins that regulates one-carbon metabolism and nucleotide biosynthesis. Relevant to our study, SHMT2 was also shown to stabilize the interferon receptor (IFNAR1), by targeting it to the cytosolic BRCC36 K63-specific deubiquitinase complex. The results of the DiffPOP analysis were verified manually by immunoblot analysis of 2D10 extract protein fractions precipitated at different methanol concentrations ( Fig 5D).
In the presence of JIB-04, the fractionation profile of SHMT2 was altered (indicated with arrows), whereas the solubility of HIV-1 Tat, and host cell factors Csn3 and Cyclin T1 was unchanged (peaks at fractions 1, 5 and 8, respectively). Taken together, these data strongly suggest that JIB-04 binds SHMT2 and alters its solubility in methanol.

JIB-04 promotes Tat protein ubiquitination and degradation
These findings raised the question of whether SHMT2 regulates Tat turnover, potentially by serving to target Tat to the BRCC36/BRISC K63-specific deubiquitinase (DUB) complex, as observed for IFNAR1 in interferon-treated cells [31]. The BRCC36 deubiquitinase is a subunit of both nuclear (BRCA1-A) and cytoplasmic (BRISC) complexes. Nuclear BRCA1-A complexes have been implicated in histone deubiquitylation during DNA repair [38], whereas cytosolic BRISC complexes deubiquitylates and stabilizes various substrate proteins. The results also show that the methanol solubility of HIV-1 Tat, Csn3, and Cyclin T1 proteins did not change in response to JIB-04. The peak of SHMT2 shifted from fractions 4-6 in the DMSO-treated fractions to fractions 2-5 in the extracts exposed to JIB-04.
https://doi.org/10.1371/journal.ppat.1007071.g005 SHMT2 and BRCC36 reverse HIV-1 Tat K63Ub to prevent destruction by autophagy Consistent with this possibility, we noted that JIB-04 affected HIV-1 Tat protein levels at early timepoints, prior to the decline of Tat mRNA or eGFP protein levels ( Fig 6A). Of note, each of these proteins has a short half-life of less than two hours (S6C Fig). We next asked whether JIB-04 would also control Tat protein levels in stably expressed cell lines. Interesting, incubation with JIB-04 markedly reduced FLAG-Tat expression in HeLa P4 cells, as well as HA-Tat levels in a Tet-on-Tat-off HeLa cell line (Fig 6B and 6C). JIB-04 also inhibited the activity of the HIV-Luciferase reporter gene in each cell line, without affecting the activity of SV40:Luciferase or CMV:β-galactosidase reporter genes (Fig 6B, 6C and S6B Fig). Moreover, JIB-04 reduced Tat protein levels, but not mRNA levels, in a dose-dependent manner (S6A Fig). Taken together, these data show that the primary effect of JIB-04 is at the level of protein Tat stability, rather than transcription. Consistent with this possibility, measurements in the presence of cycloheximide revealed that the half-life of the Tat strongly decreased in response to JIB-04 exposure in 2D10 T cells (S6C Fig).
We next asked whether JIB-04 affects HIV-1 Tat ubiquitylation. Interestingly, immunoprecipitation of Tat protein under stringent conditions revealed a dose-dependent increase in total ubiquitylation in response to JIB-04 in HeLa Tet-on-Tat-off cells (Fig 6D). Immunoblot analysis using K48Ub-and K63Ub-specific antisera revealed that JIB-04 selectively increased endogenous Tat-K63Ub, but not Tat-K48Ub, levels. Similarly, JIB-04 increased the fraction of Flag-Tat protein modified by HA-ubiquitin in HeLa P4 cells (Fig 6E). Transfection of HAtagged ubiquitin mutants that only allow either K63 or K48 linkages further established that JIB-04 selectively increased FLAG-Tat K63, but not K48, ubiquitylation. Taken together, these data indicates that JIB-04 exposure strongly increases HIV-1 Tat K63Ub and proteolytic turnover.
Previous studies have established that whereas K48Ub predisposes proteins to degradation through the proteasome, K63Ub of protein aggregates and inclusion bodies targets proteins for proteolytic destruction via autophagy [25,26]. Consistent with a prevalent role for autophagy in the destruction of Tat, our proteomics analysis of HA-Tat complexes by MudPIT (Multi-dimensional Protein Identification Technology) revealed the presence of several autophagy regulators, including HSC70, LC3, ATG3, and SQSTM1/p62 (Fig 7A). An interaction with SQSTM1/p62 was further confirmed by co-immunoprecipitation with HA-Tat (Fig 7B), and knockdown of SQSTM1/p62 in HeLa or 2D10 cells modestly increased Tat protein levels ( Fig 7C and 7D). Moreover, treatment of 2D10 cells with the autophagy inhibitor, hydroxychloroquine (HCQ) [39], strongly increased Tat protein levels in a dose-dependent manner (Fig 7E and S7A Fig)

SHMT2 and the BRCC36/BRISC deubiquitinase regulate Tat-K63Ub and turnover
We next analyzed whether SHMT2 and the BRCC36/BRISC K63Ub deubiquitinase control Tat K63Ub levels and stability in 2D10 cells. Importantly, knockdown of SHMT2 strongly reduced Tat protein levels as early as 24h following after siRNA transfection in 2D10 cells ( Fig  8A and S8A Fig). Similarly, knockdown of the BRCC36 K63Ub-specific deubiquitinase [31,32,40] led to a drastic decline in Tat protein levels in 2D10 or HeLa cells (Fig 8B and 8C). Additional knockdown experiments revealed that Tat stability is controlled by SHMT1, rather than SHMT2, in HeLa cells (Fig 8C). Consequently, the residual Tat expression in 2D10 cells depleted of SHMT2 might be due to redundant actions of SHMT1. Further analysis showed  S8D Fig), indicating that SHMT2 controls Tat stability. Additionally, JIB-04 did not affect Tat levels in cells depleted of either SHMT2 or BRCC36, confirming that both enzymes are required for the response to the drug (Fig 8F). Moreover, ectopic expression of FLAG-SHMT1 or FH-BRCC36 strongly increased FLAG-Tat protein levels and HIV-Luc activity in HeLa cells (Fig 8D). Importantly, depletion of either BRCC36 or SHMT1 in HeLa cells also selectively increased K63Ub, but not K48Ub, of the FLAG-Tat protein (Fig 8E). Collectively, these data strongly indicate that JIB-04 promotes Tat destruction by interfering with SHMT2 and the BRCC36 deubiquitinase.  It was previously shown that single lysine mutations were not sufficient to eliminate Tat-K63Ub in infected cells [19], and similarly, we found that individual lysine point mutations were not sufficient to stabilize Tat in JIB-04-treated cells (S8E and S8F Fig). Whereas the Tat ΔK mutant (all lysines substituted by arginine) was not stable in HeLa cells, the Tat ΔK (+K41) mutant, with eight lysines replaced with arginine [19], was relatively well expressed (S8G Fig). Most importantly, JIB-04 did not affect the stability of the Tat ΔK (+K41) mutant protein (Fig 8G). These findings establish that multiple lysine residues in Tat are required for turnover in response to JIB-04. Taken together, these data suggested a model through which SHMT2 and BRCC36 act to reverse K63Ub of Tat, thereby preventing its destruction by autophagy (Fig 8H).

Discussion
In this study, we show that HIV-1 Tat protein is marked by K63Ub for selective autophagy and coupled lysosomal destruction. This study began with the fortuitous discovery that the histone demethylase (HDM) inhibitor JIB-04 markedly down-regulates Tat protein levels and HIV-1 LTR transactivation in T cell lines. The JIB-04 compound strongly decreased Tat protein stability in 2D10 cells, where native Tat is expressed from the HIV-1 genome, as well as in other cell lines where tagged Tat proteins are expressed from heterologous promoters. ChIP experiments at the integrated HIV-1 LTR confirmed a selective block to RNAPII transcription elongation in cells treated with JIB-04. RNA-seq studies further established that despite the dramatic effect of JIB-04 on viral transcripts, the compound had only limited effects on host cell transcription and NF-κB signaling. Thus JIB-04 acts differently from compounds like flavopiridol, which inhibits the P-TEFb elongation factor [35,41] and affects host cell transcription broadly.
Although JIB-04 is known to inhibit histone demethylases (HDM) [28], knockdown of multiple HDMs did not recapitulate the effect of the drug on Tat stability. To identify other factors that may be sensitive to JIB-04, we adapted a novel mass-spectrometry technique, called DiffPOP, which identifies proteins based on their altered solubility in methanol in the presence of a drug. Using this approach, we identified SHMT2 (serine hydroxymethyltransferase-2) as a novel target of JIB-04 in 2D10 cell extracts. Importantly, knockdown of SHMT2 in 2D10 cells, or the related SHMT1 protein in HeLa cells, strongly decreased Tat protein expression, without affecting mRNA levels. Thus, depletion of SHMT2 was able to recapitulate the effects of JIB-04 on Tat protein levels, and validated the results of the DiffPOP approach. Of note, the histone demethylases targeted by JIB-04 were not detected by the DiffPOP approach because they remained soluble in the highest levels of methanol tested in our experiments. Because SHMT2, and the related SHMT1 enzyme, supplies methyl groups for one-carbon pools used for nucleotide biosynthesis and histone and DNA methylation [42][43][44], JIB-04 might also modulate histone methylation indirectly, through inhibition of SHMT2.
In addition to its role in one-carbon metabolism, SHMT2 was previously shown to stabilize the IFNAR1 receptor by targeting it to the cytosolic BRCC36 K63-specific deubiquitinase (DUB) complex in interferon-treated cells [31]. The BRCC36 deubiquitinase is a subunit of distinct nuclear (BRCA1-A) and cytoplasmic (BRISC) complexes. Nuclear BRCA1-A complexes are predominantly involved in histone deubiquitylation during DNA repair [38], selectively support only K63 or K48 ubiquitylation, and Tat proteins were monitored using anti-HA antisera. (F) Immunoblot analysis of the effect of JIB-04 on Tat protein levels in 2D10 cells depleted of SHMT2 or BRCC36. (G) The effect of JIB-04 on wildtype or Tat ΔK (+K41) STREP-tagged Tat proteins was shown by immunoblot. TAF4 served as loading control. (H) Model of the role of SHMT2 and BRCC36 in the release of Tat-K63Ub from destruction through chaperone-mediated autophagy or SQSMT1/p62-dependent macroautophagy. https://doi.org/10.1371/journal.ppat.1007071.g008 whereas cytosolic BRCC36/BRISC complexes deubiquitylate and stabilize protein substrates, including the IFNAR1 interferon receptor [31], the inflammasome component NLRP3 [45] and components of the mitotic spindle [46]. Importantly, knockdown of BRCC36, like SHMT2, led to a dramatic loss of Tat protein levels, concomitant with a sharp increase of Tat-K63, but not -K48, ubiquitylation. These findings strongly implicate K63 ubiquitylation of Tat as a mechanism to target it for selective destruction through autophagy, as has been seen with other autophagy substrates [25,26]. Consistent with these findings, we show that Tat protein levels increase dramatically in cells with the autophagy inhibitor hydroxychloroquine, which permeabilizes lysosomal membranes, blocking both macroautophagy as well as chaperonemediated autophagy. Taken together, these data strongly suggest that SHMT2 and BRCC36/ BRISC cooperate to rescue Tat from destruction through removal of K63Ub. Importantly, previous studies have shown that SHMT2 levels are massively upregulated upon T cell activation [47]. These findings indicate that T cell activation not only increases Tat protein biosynthesis, through upregulation of Cyclin T1 protein and enhanced HIV-1 LTR transactivation, but also decreases Tat turnover due to increased expression of SHMT2 and removal of Tat-K63Ub.
Previous studies have shown that Tat K48Ub can stimulate transactivation without affecting Tat turnover [14,19]. However, the potential role of Tat K63Ub in its destruction has been unclear, in part because mutation of single lysine residues does not increase Tat protein stability, and because Tat does not bind the SQSMT1/p62 autophagy factor through the ubiquitininteraction domain [21]. Several lines of evidence indicate that Tat turnover by autophagy is dependent on K63Ub. First, we show that increased Tat turnover in cells exposed to JIB-04, or in cells depleted of SHMT2 or BRCC36, is accompanied by a large increase in Tat-K63Ub levels, with no effect on Tat K48Ub. Because BRCC36 is capable of removing only K63Ub chains, it does not require specific targeting of substrates through the K63Ub modification. Second, we show that point mutation of multiple lysine residues in Tat is sufficient to stabilize protein levels, and renders Tat protein levels insensitive to proteolysis induced by JIB-04. Third, the residual levels of Tat that remain following depletion of either SHMT2 or BRCC36 are not sensitive to JIB-04. Taken together, these data provide strong evidence that Tat levels are regulated by selective autophagy through a K63Ub-dependent mechanism.
Selective protein destruction can occur either through canonical macroautophagy or through chaperone-mediated autophagy (CMA), or both, however the mechanisms of each pathway are quite distinct [48,49]. For CMA, client proteins interact through a KFERQ-like motif with HSC70 chaperones and are targeted directly to the LAMP2A lysosomal membrane receptor for destruction by proteases in the lysosomal lumen. For macroautophagy, K63Ub client proteins interact with the ubiquitin-binding domain of SQSMT1/p62. Previous studies have shown that Tat interacts with p62 and LC3 for destruction by macroautophagy [21]. However, Tat proteins in the cytoplasm have also been found to strongly co-localize with the LAMP2A lysosome receptor in neuronal cells [50]. Our proteomics analysis of Tat complexes reveals high levels of the CMA-specific chaperone, HSC70, in addition to the macroautophagy proteins p62 and LC3, which suggests that Tat may be destroyed through either process, depending on the specific environment of the cell. Therefore further studies will be needed to distinguish the relative contributions of canonical (macroautophagy) and non-canonical (CMA) pathways to Tat turnover during latency and active infection, where both pathways are known to be functional [51], and to assess whether or not SHMT2 and BRCC36 function selectively in one pathway.
In summary, the data presented here show that Tat-K63Ub proteins can be rescued from autophagy through the combined actions of SHMT2 and the BRCC36 deubuiquitinase, which may be important for the rapid induction of HIV-1 transcription upon T cell activation. Different parameters that affect Tat half-life have important considerations for establishment and escape from viral latency [7]. Further examination of the mechanisms that control Tat K63Ub levels and its in turnover by autophagy may enable a more robust induction of Tat expression for escape from latency to potentiate the effectiveness of antiviral regimens.

Commercial antisera used in this study
Specific antisera were obtained from the following sources: HIV-1 Tat (ab42359 & ab43014

JIB-04 target identification using DiffPOP (Differential Precipitation of Proteins)
Five 15-cm plates of 2D10 T cells (Jurkat T cells, clone 2D10, as described in [5,  ). An aliquot of 0.5 ml of whole cell lysate was centrifuged at 18000 x g for 10 min on ice, and the supernatant was collected as the initial lysate. To this lysate we added 2 μl of DMSO or JIB-04 (50 mM in DMSO), respectively. Proteins were differentially precipitated in ten fractions by the stepwise addition of 90% methanol 1% acetic acid. Protein pellets were then washed in cold acetone, solubilized and digested in a standard tryptic digest. Digested samples were analyzed by mass spectrometry. The differential distribution of cellular factors precipitated from extracts treated with DMSO or JIB-04 were calculated using the Pearson's correlation. The most negative values of the Pearson's correlation represent those proteins whose solubility is most altered by exposure to JIB-04, which implies a direct drug-protein interaction. For verification, protein fractions were analyzed directly by immunoblot using an SHMT2-specific antibody.

RNA extraction, qRT-PCR, RNA-seq and bioinformatic analysis
Total RNA was extracted using RNeasy Plus Mini Kit (Qiagen). RNA was then quantitated by NanoDrop. About 1 μg of total RNA was subjected to reverse transcription using SuperScript III First-Strand Synthesis System kit from Invitrogen. About 0.5 to 1 μl of final cDNA products were used as PCR templates in each well of 96-well plate (ABI). The CTs from the real time PCR using ABI 7300 system were analyzed by Excel (The PCR efficiency was set to 2). For RNA-seq, purified RNA samples were first treated with Dnase I and purified by Zymo RNA purification kit (Zymo Research). HiSeq SE50 was performed on samples by the Next Generation Sequencing Core Facility at Salk Institute. The raw RNA-seq data can be accessed through NCBI GEO database (Accession: GSE109460). Sequencing reads were aligned to the human hg19 and HIV genomes using STAR and human gene-level read counts were obtained with feature Counts. Genes with less than 10 reads in all samples were ignored in further analysis. The reads were separately mapped to HIV-1 and d2EGFP sequences with BWA. From the uniquely-aligned read counts, the differential gene expression was calculated using DESeq2 and a design formula that accounted for sample batch, TNFα treatment, and JIB-04 treatment. For differentially expressed genes with at least a 2-fold change with adjusted p-value of 0.05 or less, Gene Ontology enrichment for biological process terms was carried out using clusterProfiler. The ten most significantly enriched terms per sample were included, but only terms with 200 associated genes or fewer were shown. For heatmap, all raw read counts were first transformed using a regularized logarithm (rlog) to generate log2-scaled values. The 100 genes with the largest differences between minimum and maximum rlog values were plotted after meancentering. (See S1 Supporting Methods_PDF for other Methods).  3). (A) Pie-chart of the 811 genes (out of 13546 genes with reads >10) altered more than 2-fold by JIB-04 (p<0.05). These 811 genes were listed in S1 File. (B) The top ten Gene Ontology enrichment biological process terms for DMSO vs JIB-04 and TNFα 0 h vs 6 h. (C) qRT-PCR results for the indicated genes randomly-selected from the top 100 heatmap for histones and JIB-04 activated genes, respectively. The significant differences between DMSOtreated and JIB-04-treated samples were analyzed by Student's T-test ( ÃÃÃ = p<0.0005). (D) Immunoblot analysis of histone H2B and H3 protein levels in 2D10 cells that were exposed to JIB-04 (0-10 μM) for 24 h. Csn3 served as loading control. HIV-LTR-Luc was normalized to SV40-Renilla-Luc. Middle, Luc assay analysis for SV40-Renilla-Luc at the same condition. Renilla-Luc activity was normalized to total protein concentrations. Bottom, CMV-β-Gal assay for CMV-β-Gal at the same condition. CMVβ-Gal activity was normalized to Renilla-Luc activity. Activity from cells treated by 10 μg/ml doxycycline was normalized to 1. The significant differences between luciferase and β-Gal activity for DMSO and JIB-04 treated samples were calculated by Student's T-test (ns = nonsignificant, Ã p<0.05).