The Nerve Growth Factor IB-like Receptor Nurr1 (NR4A2) Recruits CoREST Transcription Repressor Complexes to Silence HIV Following Proviral Reactivation in Microglial Cells

Human immune deficiency virus (HIV) infection of microglial cells in the brain leads to chronic neuroinflammation, which is antecedent to the development of HIV-associated neurocognitive disorders (HAND) in the majority of patients. Productively HIV infected microglia release multiple neurotoxins including proinflammatory cytokines and HIV proteins such as envelope glycoprotein (gp120) and transactivator of transcription (Tat). However, powerful counteracting silencing mechanisms in microglial cells result in the rapid shutdown of HIV expression to limit neuronal damage. Here we investigated whether the Nerve Growth Factor IB-like nuclear receptor Nurr1 (NR4A2), which is a repressor of inflammation in the brain, acts to directly restrict HIV expression. HIV silencing was substantially enhanced by Nurr1 agonists in both immortalized human microglial cells (hµglia) and induced pluripotent stem cells (iPSC)-derived human microglial cells (iMG). Overexpression of Nurr1 led to viral suppression, whereas by contrast, knock down (KD) of endogenous Nurr1 blocked HIV silencing. Chromatin immunoprecipitation (ChIP) assays showed that Nurr1 mediates recruitment of the CoREST/HDAC1/G9a/EZH2 transcription repressor complex to HIV promoter resulting in epigenetic silencing of active HIV. Transcriptomic studies demonstrated that in addition to repressing HIV transcription, Nurr1 also downregulated numerous cellular genes involved in inflammation, cell cycle, and metabolism, thus promoting HIV latency and microglial homoeostasis. Thus, Nurr1 plays a pivotal role in modulating the cycles of proviral reactivation by cytokines and potentiating the proviral transcriptional shutdown. These data highlight the therapeutic potential of Nurr1 agonists for inducing HIV silencing and microglial homeostasis and amelioration of the neuroinflammation associated with HAND. AUTHOR SUMMARY HIV enters the brain almost immediately after infection where it infects perivascular macrophages, microglia and, to a less extent, astrocytes. In previous work using an immortalized human microglial cell model, we observed that integrated HIV constantly underwent cycles of reactivation and subsequent silencing. In the present study, we found that the Nurr1 nuclear receptor is a key mediator of HIV silencing. The functional activation of Nurr1 by specific agonists, or the over expression of Nurr1, resulted in rapid silencing of activated HIV in microglial cells. Global gene expression analysis confirmed that Nurr1 not only repressed HIV expression but also regulated numerous genes involved in microglial homeostasis and inflammation. Thus, Nurr1 is pivotal for HIV silencing and repression of inflammation in the brain and is a promising therapeutic target for treatment of HAND.

49 that the Nurr1 nuclear receptor is a key mediator of HIV silencing. The functional 50 activation of Nurr1 by specific agonists, or the over expression of Nurr1, resulted in rapid 51 silencing of activated HIV in microglial cells. Global gene expression analysis confirmed 52 that Nurr1 not only repressed HIV expression but also regulated numerous genes 53 involved in microglial homeostasis and inflammation. Thus, Nurr1 is pivotal for HIV 54 silencing and repression of inflammation in the brain and is a promising therapeutic target 55 for treatment of HAND.

INTRODUCTION
147 Nurr1 agonists strongly induce HIV silencing in microglial cells 148 To study the role of nuclear receptors in the control of HIV expression in the 149 microglia, we used our immortalized human microglial (hµglia) cells [37], which were 150 infected with a recombinant HIV-1 reporter that carried an EGFP marker for "real-time" 151 monitoring of HIV latency and reactivation (Fig 1A). One representative clone, HC69 [37, 187 constitutively expressed Nurr1, as well as a very low level of Nor1 (Fig 2B) (Fig 3A) 210 western blots (Fig 3B).

211
To examine how overexpression of each of these nuclear receptors modulated HIV 212 proviral activation and silencing, we stimulated all four cell lines with high dose (400 pg/ml) 213 TNF- for 24 hr to induce HIV transcription through activation of NF-B [38], followed by 214 a 48 hr chase experiment in which TNF- was removed by washing the cells with PBS P a g e 11 | 60 215 followed by the addition of media lacking TNF- (Fig 3C). As shown by western blot in 216 Fig 3D, TNF- strongly induced the expression of HIV Nef protein, which we used as a 217 marker of HIV reactivation, in all cell lines at 24 hr. Notably, Nef expression decreased in 218 all four cell lines 48 hr after TNF- withdrawal. However, the reduction in Nef expression 219 was much more pronounced in HC69 cells that express 3X-FLAG-Nurr1, suggesting that 220 overexpression of Nurr1 enhanced silencing of active HIV in HC69 cells.

221
We rigorously confirmed these findings using the RNA-Seq data (Fig 3E) (Fig 3E). The level of HIV mRNA after withdrawal of high dose TNF- was 230 three times lower in Nurr1 overexpressing cells than in vector-infected cells (Fig 3E),  (Fig 4A) 237 Following the protocol described in Fig 4B, control and KD cells were activated with a P a g e 12 | 60 238 high dose (400 pg/ml) TNF- for 24 hr, followed by a 72 hr chase. Western blot analyses 239 confirmed the Nurr1 knock down efficiency (Fig 4C). The blots also showed that HIV Nef 240 protein, which is a measure of HIV transcription, was strongly induced at 24 hr post TNF- 241 stimulation in both the control and the Nurr1 KD cells. However, after the chase, Nef levels 242 decreased significantly in the control cells but remained high in Nurr1 KD cells (Fig 4C).

243
Similar results were obtained using flow cytometry (Fig 4D). Compared to cells 244 expressing control shRNA with 10.5% GFP+ cells, the Nurr1 KD cells displayed 58.8% 245 GFP + cells even before TNF- stimulation, which most likely resulted from failure of 246 silencing spontaneously reactivated HIV in these cells due to Nurr1 depletion (Fig 4D).
247 As expected, after exposure to high dose TNF- for 24 hr, both the control and Nurr1 KD 248 cell lines expressed equally high levels of GFP expression (Fig 4D), displaying 86.3% 249 and 91.2% GFP+ cells respectively. However, 72 hrs after TNF- withdrawal, GFP 250 expression decreased significantly in cells expressing the control shRNA (47.2% GFP+) 251 but remained high (74.6% GFP+) in the Nurr1 KD cells (Fig 4D). Finally, the overall mRNA 252 level of the HIV measured by RNA-Seq was about 1.7 times higher in Nurr1 KD at the 253 end of the chase experiment (Fig 4E).

254
Thus, both the overexpression and the reciprocal KD experiments confirmed an 255 essential role of Nurr1 in the silencing HIV in microglial cells.

257
Our RNA-Seq data also provided important insights into the cellular pathways that 258 were impacted by Nurr-1 over-and under-expression. We focused our attention on the 259 changes in cellular transcriptome during the chase step following TNF- induction since, 260 as described above, this is the stage where Nurr1 has the greatest impact on HIV gene 261 expression. As shown by the differential gene expression curves in Fig 5A, (Fig 5A & S1 Fig.). Pathways that showed the most statistically significant changes 265 in response to Nurr1 overexpression included the downregulation of key pathways with 266 critical roles in cellular proliferation and metabolism including: MYC, E2F and MTORC 267 signaling and G2M checkpoint (Fig 5B). By contrast, KD of Nurr1 by shRNA did not 268 selectively activate any major signaling pathways.

269
It is important to note that Nurr1 overexpression did not significantly interfere with 270 the TNF- signaling pathway during any step of these experiments (Fig 5B), suggesting 271 that the cellular proliferation pathways we have identified are directly regulated by Nurr1.
272 To further address this issue and determine whether Nurr1 simply accelerated the 273 reversal of the normal microglial response to TNF- stimulation during the chase, or if it 274 regulated a distinct set of genes and pathways, we performed a gene trajectory analysis 275 (Fig 6A, S2 Fig).

276
For the trajectory analysis we included RNA-Seq data from cells that were treated 277 with the low dose of TNF- (20 pg/ml), to simulate a sub-optimal activation signal. A 278 pseudo-trajectory was defined containing three steps: Step 1 defines the changes in gene 279 expression following stimulation with low dose TNF- compared to untreated cells.
Step For each of these 283 steps we calculated whether the expressed protein-coding genes were either upregulated 284 (designated as "u"), downregulated (designated as "d") or did not show differential 285 expression in a statistically significant manner (designated as "n"). Genes that showed 286 similar patterns of changes during each step were placed in the same category and 287 named according to their pattern of change during these treatment steps. For example, 288 those that did not show a change after low dose TNF- stimulation (thus marked as n for 289 Step 1), but were downregulated after high dose TNF- treatment compared to cells 290 treated with low dose TNF- (marked as d for Step 2), and showed upregulation during 291 the chase study compared to cells treated with high dose TNF- (marked as u for step 292 3), were therefore designated as ndu.

293
Most genes did not show any change in their expression following the above 294 treatments (designated as the "nnn" group) in both control (vector) and Nurr1-295 overexpressing cells (Fig 6A, S2 Fig), and as expected, control cells had higher numbers 296 of nnn group genes than Nurr1 overexpressing cells.

297
Among those genes that showed an expression change in Nurr1 overexpressing 298 cells, the majority belonged to genes that were not differentially expressed after either a 299 low or high dose TNF- treatment and exclusively changed their expression profiles 300 during the chase step (i.e., nnu and nnd trajectories, Fig 6A, S2 Fig). We also noted that 301 the number of genes in these two trajectories were markedly higher in Nurr1 302 overexpressing cells compared to control cells (i.e., over 800 and 1400 genes for nnu and 303 nnd trajectories, respectively) while the number of genes in other trajectories with the 304 exception of nnn differed by less than 100 genes (Fig 6A, S2 Fig).

319
In order to define the functional impact of this Nurr1-specific set of genes, we 320 performed pathway analysis on the subset of genes that had either nnu or nnd trajectories 321 in Nurr1 overexpressing cells, and a nnn trajectory in control cells (Fig 6B). Strikingly, 322 these Nurr1-induced changes in gene expression during the chase step once again 323 highlighted the downregulation of several key proliferative pathways, including: MYC, E2F 324 and MTORC signaling, G2M checkpoint regulation, metabolic pathways such as oxidative 325 phosphorylation, and inflammatory pathways such as IFN- and IFN-γ response 326 pathways (Fig 6B).

327
Heat maps of the differentially expressed genes further emphasized that the vast 328 majority of genes in each pathway were downregulated in Nurr1 overexpressing cells.     (Fig 8A). However, in the presence of SAHA, UNC0638, or 406 GSK343, the numbers of GFP+ cells remained higher (i.e., 77.8%, 85.5%, and 84.7% 407 respectively), indicating that functional inhibition of these epigenetic silencers prevented 408 active HIV from reverting to latency.

409
To confirm the role of these epigenetic silencers, we generated HC69 cell lines 410 stably expressing CoREST-specific shRNA or CRISPR/Cas9/guide RNA (gRNA) for G9a 411 or EZH2. We confirmed successful KD or knock out (KO) of these proteins in these cell 412 lines by Western blot analysis (Fig 8B & C). The genetically modified cells were activated 413 with a high dose of TNF- (400 pg/ml) for 24 hr, followed by culturing the cells in the 414 absence of TNF- for 48 hr and measurement of GFP expression. CoREST KD 415 substantially increased GFP expression (80.1% GFP+ vs. 25.8% in control cell) even 416 without TNF- stimulation (Fig 8D).  (Fig 8E).  (Fig 1A). About 50% of the iMG became GFP+ two days 430 after HIV infection (Fig 9A). We then treated the infected iMG with 6-MP and another 431 Nurr1 agonist, amodiaquine (AQ) [56, 67], for four days. Both 6-MP and AQ decreased 432 the number of GFP+ cells in a dose-dependent manner (Fig 9B & C) and lowered the 433 levels of HIV un-spliced transcripts (Fig 9D). Both agonists also dose-dependently 434 reduced MMP2 mRNA in iMG (Fig 9E). Collectively, results from both hµglia and iMG 435 strongly suggested an important role for Nurr1 in HIV silencing in microglial cells.    (Fig 1 A)   PPARγ/SIRT6/FoxO3a pathway after subarachnoid hemorrhage in rats. J