ANP32A and ANP32B are key factors in the Rev dependent CRM1 pathway for nuclear export of HIV-1 unspliced mRNA

The nuclear export receptor CRM1 is an important regulator involved in the shuttling of various cellular and viral RNAs between the nucleus and the cytoplasm. HIV-1 Rev interacts with CRM1 in the late phase of HIV-1 replication to promote nuclear export of unspliced and single spliced HIV-1 transcripts. However, other cellular factors that are involved in the CRM1-dependent viral RNA nuclear export remain largely unknown. Here, we identified that ANP32A and ANP32B mediate the export of unspliced or partially spliced viral mRNA via interacting with Rev and CRM1. We found that double, but not single, knockout of ANP32A and ANP32B, significantly decreased the expression of gag protein. Reconstitution of either ANP32A or ANP32B restored the viral production equally. Disruption of both ANP32A and ANP32B expression led to a dramatic accumulation of unspliced viral mRNA in the nucleus. We further identified that ANP32A and ANP32B interact with both Rev and CRM1 to promote RNA transport. Our data strongly suggest that ANP32A and ANP32B play important role in the Rev-CRM1 pathway, which is essential for HIV-1 replication, and our findings provide a candidate therapeutic target for host defense against retroviral infection.

consensus view of the completely spliced mRNA is presumably transported into the cytoplasm mediated by the 21 TAP/NXF1 pathway to translate the regulatory proteins Rev, Nef, and Tat. The Rev protein of HIV-1, a 116 22 amino acid accessory protein, has a nuclear localization signal (NLS) recognized by importins, and also a 23 leucine-rich nuclear export signal (NES) recognized by mammalian nuclear export factor Chromosomal 24 Maintenance 1 (CRM1); it therefore shuttles between the nucleus and cytoplasm. Rev can specifically bind to 25 the Rev response element (RRE) located in the env gene of the unspliced or partially spliced mRNA. In the late 26 phase of HIV-1 replication, with Rev accumulation in the nucleus the unspliced or partially spliced mRNAs are 27 exported to the cytoplasm via a Rev-CRM1 dependent export pathway to translate all structural viral proteins 28 (2-10). 29 The key factor in the export of viral mRNA is the Rev-CRM1 complex. In the nucleus, the multimerized 30 Rev recruits CRM1 through the Rev leucine-rich nuclear export signal (NES) located in the C-terminal domain 31 to assemble the ribonucleoprotein complex with , thus facilitating its exportation. However, a 32 number of host factors (including Matrin-3, DDX1, DDX21, DDX3, DDX5, MOV10, Sam68, and CBP80) have 33 been reported to interact with Rev and RRE and help viral mRNA export from the nucleus to the cytoplasm and 34 mRNA translation (14-23). Most of these factors have not been identified as interacting with CRM1. So far 35 there are only few proteins, such as DDX3 and Naf1, that are reported to interact with the Rev-CRM1 complex 36 to facilitate viral mRNA export. Whether other cellular proteins are involved in the Rev-CRM1 complex and 37 direct the viral RNA export from the nucleus to the cytoplasm remains largely unknown. 38 CRM1 is well known as an important receptor in the nuclear-cytoplasmic transportation of cellular RNA 39 complexes and many viral RNAs, and functions by interacting with different cellular proteins through various 40 mechanisms (24)(25)(26)(27). In this study, we investigated the functions of two proteins that interact with CRM1, 41 ANP32A and ANP32B, during the export of HIV-1 unspliced mRNA from the nucleus to the cytoplasm. The 42 ANP32 family member, initially identified as a 32kDa highly conserved acidic (leucine-rich) nuclear 43 phosphoprotein protein, is characterized by an N-terminal leucine rich repeat (LRR) domain and a C-terminal 44 low-complexity acidic (LCAR) region (28). It has been suggested that the LRR domains of the ANP32 proteins 45 interact with several proteins, including CRM1 (29), PP2Ac (30), Ataxin-1 (31), Histone H3-H4 (32), and 46 Clip170 (33). The ANP32 proteins are also ascribed biochemical activities including chromatin modification 47 while at 24 and 36 h post infection, levels of viral gag mRNA in DKO cells were about 10~fold lower than in 93 WT cells (Fig. 2C). This result suggests the existence of a block to viral RNA processing after RNA generation 94 in DKO cells. ANP32A and ANP32B have been reported to bridge HuR and CRM1 to support FV RNA transport 95 (40) and also to contribute to cellular RNA transport (29,41). We next examined whether ANP32A and ANP32B 96 had any effect on HIV-1 unspliced mRNA nucleo-cytoplasmic shuttling. We fractioned the nucleus from the 97 cytosol, and the cytoplasmic and nuclear distributions of HIV-1 unspliced mRNA in WT or DKO cells were 98 analyzed by quantitative PCR at 12 h and 24 h post infection. At 12 h post infection, there is no obvious 99 difference in the distribution of viral RNA between the WT and DKO cells, in that most of the viral RNA in 100 both cell types was located in the nucleus. In contrast, at 24 h post infection, in the WT cells 85% of the RNA 101 was located in the cytoplasm, indicating that the viral RNA was largely exported from the nucleus into the 102 cytoplasm. However, in the DKO cells, most of the unspliced viral RNA had accumulated in the nucleus. As the 103 result of the viral RNA retention in the nucleus, the ratio of transported viral unspliced mRNA vs total RNA in 104 DKO cells at 24 h post infection was very low (Fig 2D). The transportation efficiency of viral gag mRNA from 105 nucleus to cytoplasm was largely reduced in DKO cells compared with WT ( Fig 2E). However, the completely 106 spliced tat mRNA is mostly located in the cytoplasm at 24 post infection in both WT and DKO cells. Only slight 107 alteration of the nuclear export of the tat mRNA was observed in DKO cells (Fig 2F). We used Lamin B1 and 108 tubulin protein as nuclear and cytoplasmic protein controls respectively to assess the purity of the nuclear and 109 cytoplasmic extracts (Fig. 2G). This result implied that ANP32A and ANP32B are required for the transport of 110 unspliced viral mRNA from the nucleus to the cytoplasm. Absence of ANP32A and ANP32B could lead to the 111 accumulation of gag mRNA in the nucleus and subsequently reduced gag mRNA production in trans ( Fig 2C). 112 To better illustrate this phenomenon, the subcellular distribution of unspliced gag transcripts were 113 visualized using fluorescence in-situ hybridization (RNA Scope) with RNA probes that specifically target gag-114 pol mRNA. Consistent with the above observations, at 24 and 36 h post infection, gag-pol mRNA in the WT 115 cells is mostly located in the cytoplasm (Fig. 3A), whereas most of the gag-pol mRNAs were in the nucleus of 116 the DKO cells (Fig. 3B), indicating a defect in the nuclear export of the unspliced transcripts in the absence of 117 ANP32A and ANP32B. The distribution of the gag-pol mRNA observed using the DeltaVision OMX 3D 118 structured illumination microscope (3D-SIM) (GE, USA) closely mirrored the confocal microscopy result. It 119 was clearly showed that in the WT cells, the gag-pol mRNA RNA is mostly located in the cytoplasm (Fig. 3C), 120 but in the DKO cells the unspliced mRNA was retained in the nucleus (Fig. 3D). These data strongly support 121 that ANP32A and ANP32B contribute to the efficient export of unspliced viral gag-pol transcripts from the 122 nucleus to the cytoplasm. 123 124 ANP32A and ANP32B interact with Rev. Rev is the key viral protein mediates unspliced viral RNA export 125 from nucleus to cytoplasm. To investigate whether ANP32A and ANP32B interact with Rev, Rev-HA was 126 transfected into HEK293T cells with ANP32A-Flag or ANP32B-Flag and the formaldehyde cross-linking 127 immunoprecipitates (42,43) were analyzed by Western blotting in 24 h. The result showed that both ANP32A 128 and ANP32B co-immunoprecipitate with Rev protein (Fig. 4A). We then analyzed the subcelluar location of 129 Rev-HA and ANP32A-Flag or ANP32B-Flag in HeLa cells using confocal imaging. We observed that the 130 ANP32A and ANP32B proteins were mainly found in the nucleus of the transfected cells. As Rev is a nuclear-131 cytoplasmic shuttling protein, we would expect to see it both free and co-localized with the proteins it shuttles. 132 Our results suggest that a fraction of the Rev protein does indeed co-localize with ANP32A or ANP32B in the 133 nuclei of the HeLa cells ( Fig 4B). 134 To further confirm the interaction between ANP32 proteins and Rev, a bimolecular fluorescence 135 complementation (BiFC) assay was used. In our research, the function of ANP32B is similar to ANP32A, so in 136 the BiFC assay we just detected the interaction between ANP32A and Rev. This assay enables direct 137 visualization of protein interaction and the subcellular location in living cells. The BiFC assay is based on the 138 finding that two non-fluorescent fragments of a fluorescent protein can associate to produce a significantly 139 brighter fluorescent signal, and that a fluorescent signal can still be obtained if the association of the fragments 140 is adjacent, for example if they are fused to specifically interacting partners. N-terminal residues 1-173 (VN) 141 and C-terminal residues 174-239 (VC) of Venus were fused to the C-terminus of ANP32A and Rev protein, 142 respectively ( Fig. 4C). Firstly, we confirm that the Rev-VC was functional equally as wild type Rev in a RRE 143 dependent reporter system (data not shown). When ANP32A-VN and Rev-VC were co-transfected into Hela 144 cells, the Venus signal mostly located in the nucleus, while the human trim5α-VC plasmid was used as a negative 145 control (Fig. 4D). The Venus signal was observed in the nucleus only when ANP32A-VN and Rev-VC were co-146 transfected compared with single transfection of ANP32A-VN or Rev-VC (no Venus signal) (Fig. 4E). Thus, 147 together, these data indicate that both ANP32A and ANP32B can specifically interact with Rev in the nucleus 148 of Hela cells. 149 150 ANP32A and ANP32B play key role in the CRM1 dependent RNA export pathway. HIV-1 Rev/RRE 151 dependent mRNA transport relies on the CRM1 export pathway. We next investigated whether ANP32A and 152 ANP32B interact with CRM1. To do this, we overexpressed ANP32A-Flag or ANP32B-Flag, either with or 153 without a CRM1 express vector (CRM1-HA), into HEK293T cells. Formaldehyde cross-linking 154 immunoprecipitants with HA beads were performed. The result showed that both ANP32A and ANP32B co-155 immunoprecipitated with CRM1 (Fig. 5A). By confocal imaging analysis, we found that either ANP32A or 156 ANP32B protein co-localized with CRM1 in the nucleus and perinuclear region (Fig. 5B). To further confirm 157 the interaction between CRM1 and ANP32A, we performed a BiFC assay by using CRM1-VN and ANP32A-158 VC fusion proteins. We observed that ANP32A interacted with CRM1 at the nuclear periphery at 24 h post 159 transfection when CRM1-VN and ANP32A-VC were co-transfected into Hela cells (Fig. 5C). When CRM1-160 VN or ANP32A-VC was transfected individually into Hela cells, the CRM1-VN was mostly located in the 161 nuclear periphery and nucleus and ANP32A-VN protein was mostly located in the nucleus (Fig. 5C). These 162 results confirm that ANP32A and ANP32B interact with CRM1 proteins. 163 Leptomycin B (LMB), a specific inhibitor that covalently binds to cys528 of CRM1 was used to block 164 CRM1 mediated export from the nucleus (44). We found that HIV-1 unspliced mRNA transportation from the 165 nucleus to the cytoplasm was significantly inhibited with LMB treatment (Fig. 5D). To further clarify the ability 166 of ANP32A and ANP32B in the CRM1 dependent RNA export pathway, pNL4-3-lucΔVifΔEnv were used to 167 transfect both WT and DKO cells either with or without LMB treatment. We found that in the WT cells, LMB 168 treatment significantly decreased the efficiency of HIV-1 gag expression. Interestingly, depletion of ANP32A 169 and ANP32B resulted in worse p55 expression, although the LMB treatment caused further reduction of gag 170 expression in DKO cells (Fig. 5E). Considering the interactions between ANP32A/B and Rev/CRM1 together, 171 we propose that ANP32A and ANP32B are required for the Rev/CRM1 dependent viral RNA export. 172 In the nucleus, Rev binds to the RRE and interacts with CRM1, facilitating nuclear export of viral RNAs. 173 In the above result, we confirmed that ANP32A/B interact with Rev and CRM1 protein. To determine the 174 functional relationship between ANP32A/B and Rev-CRM1 nuclear export pathway, we investigated whether 175 the ANP32A/B deletion influence the interaction between Rev and CRM1. 293T WT and DKO cells were co-176 transfected with FLAG-tagged Rev with or without HA-tagged CRM1, co-IP experiments were performed. Cell 177 lysates were formaldehyde cross-linking immunoprecipitated with an anti-HA beads, followed by western 178 blotting. The result showed that either in the WT or DKO cells Rev protein interacts with CRM1 (Fig. 6A). The 179 result was confirmed by the BiFC assay (Fig. 6B). This data demonstrated that deletion of ANP32A/B impaired 180 CRM1/Rev/RRE dependent RNA transport without disruption of the interaction between Rev and CRM1. 181 Together, we conclude that ANP32A and ANP32B are key cofactors of Rev, and they interact with CRM1 to 182 mediate export of unspliced or partially spliced viral RNA from nucleus to cytoplasm. of Rev binds to CRM1, therefore enabling the transportation of the RNA. Here we use a double knock out of 191 ANP32A and ANP32B, generated using the CRISPR/Cas9 system, to provide novel evidence that the cellular 192 factors ANP32A and ANP32B are both functional and essential components in the production of HIV- 1. 193 ANP32A and ANP32B bind to Rev and CRM1 and support the nuclear export of unspliced and partially spliced 194 mRNA. Furthermore, our studies show that inhibition of the function of CRM1 or deletion of ANP32A and 195 ANP32B, impaired unspliced viral RNA export and HIV-1 production. All these results demonstrated that 196 ANP32A and ANP32B play an important role in HIV-1 production and presumably are key members of the 197 Rev-CRM1 RNA export complex. 198 The evolutionarily conserved ANP32 family of proteins is characterized by an N-terminal variable number 199 leucine rich repeat domain (LRR) and a C-terminal acidic tail (28,47). ANP32A and ANP32B share a highly 200 conserved structure. Recent study has showed that chicken ANP32A is a major host factor affecting avian 201 influenza viral polymerase activity in human cells (38); ANP32B can function as a target of the M protein of 202 henipavirus to support virus replication (48). Moreover, ANP32A and ANP32B act as necessary cofactors in the 203 export of nuclear RNA of foamy virus without a Rev-like protein (40). The CRISPR/Cas9 system and siRNA 204 screen approaches enable us to achieve a clear background of certain protein express and have allowed the 205 identification of various host co-factors that mediate 50). In this study, we found that 206 double, but not single, deletion of ANP32A and ANP32B using the CRISPR/Cas9 system in an HEK293T cell 207 line significantly decreases HIV-1 production and export of unspliced viral RNA. Reconstitution of ANP32A or 208 ANP32B restores the gag expression. This result indicates that both ANP32A and ANP32B function similarly 209 and contribute equally to the export of viral RNA. 210 Many host factors have been reported to involve in the Rev dependent viral RNA export. Rev interacts 211 with transport receptor CRM1 and other host factors, including Ran-GTP, DDX3, RMB14, and Naf-1, to 212 facilitate RNA export (11,22,51,52). Certain proteins have been identified that bind not only to Rev, but also to 213 CRM1. Examples include the naf-1 protein, which can interact with CRM1 to enhance mRNA nuclear transport 214 and HIV-1 virus production (52); RBM14, which interacts with CRM1 and Rev protein and supports Rev-215 mediated export of unspliced viral transcripts (19); and UPF1, which is required in the regulation of vRNA 216 nuclear export and which shuttles between the nucleus and the cytoplasm to form a multimeric complex 217 containing UPF1, DDX3, CRM1 and Rev protein(53). There are also some factors such as Sam68, matrin 3, 218 and DDX1, that interact with Rev and enhance export of viral RNA without evidence of binding to CRM1. 219 Sam68 expression is required for Rev function by directly regulating the CRM1-mediated Rev nuclear export 220 pathway but Sam68 itself does not actively shuttle between the nucleus and the cytoplasm (18,54,55). The 221 nuclear matrix protein matrin 3 is shown to interact with Rev protein and bind Rev/RRE-containing viral RNA 222 to increase cytoplasmic expression of these viral RNAs (56,57). DDX1 is known to physically interact with 223 both Rev and the RRE and it can act through RNA to promote HIV-1 Rev-RRE assembly (20,58). Our results 224 showed that ANP32A/B can bind to Rev and CRM1. Without ANP32A and ANP32B, hardly any gag mRNA 225 was detected in the cytoplasm. From these results, we propose a key role of ANP32A/B proteins in the 226 Rev/RRE-CRM1 RNA export pathway (Fig. 7). 227 The mechanism of function of the Rev protein, and the details of transport of RNA need further 228 investigation. In our work, using RNA Scope technology, we found that without either ANP32A or ANP32B, 229 most of the gag mRNA is restricted to the nucleus. In addition, we evaluated the contribution of CRM1 and 230 ANP32A/B in HIV-1 Rev dependent RNA transport. We found that blocking of CRM1 by LMB dramatically 231 reduces gag expression in WT 293T cells, which is consistent with results from previous studies. Surprisingly, 232 in DKO cells, HIV-1 has a low-level of virus production ability and inhibition of CRM1 by LMB showed only 233 a slight reduction in viral replication (Fig. 5E). This result indicated that without ANP32A/B, the CRM1-234 mediated export pathway for HIV-1 RNAs was blocked. Thus, all our experiments appear to support the 235 conclusion that either ANP32A or ANP32B is required to facilitate nuclear export of HIV-1 RNA though the 236 Rev/CRM1 pathway. 237 We have observed that viral unspliced RNAs at 24h post infection was predominantly located in the nucleus 238 in the DKO cell while most of it has transported to the cytoplasm in the WT cells. Over accumulation of the 239 RNA in the nucleus presumably blocked the process of RNA synthesis. In our study, we found that total gag 240 RNA is significantly lower compared with WT (Fig. 2C). Whether nuclear retention is accompanied of 241 degradation or over splicing of the unspliced mRNA needs to be investigated. 242 In summary, we found that ANP32A and ANP32B are essential host factors supporting HIV-1 virus 243 production. ANP32A and ANP32B specifically interact with Rev and CRM1 and both contribute to the export 244

DNA was extracted using DNeasy kits (Qiagen). Equal amounts of cellular DNA after a further treatment with 291
DpnⅠ were then subjected to real-time PCR in order to measure early and late viral reverse transcripts, as 292 described previously (61). β-globin DNA was measured as an endogenous control. The primers used are as 293 follows: early oHC64-F, 5-TAACTAGGGAACCCACTGC-3, early oHC64-R, 5-294 GCTAGAGATTTTCCACACTG-3; late RT MH531-F, 5-TGTGTGCCCGTCTGTTGTGT-3, late RT MH532-295 R, 5-GAGTCCTGCGTCGAGAGAGC-3; β-globin-F, 5-CCCTTGGACCCAGAGGTTCT -3, β-globin-R, 5-296 CGAGCACTTTCTTGCCATGA-3. To quantify the mRNA in the WT and DKO cells, total RNA from the cells 297 was extracted using a Bio-fast simply P RNA extraction kit (catalog # BSC60S1, Bioer). Nuclear and 298 cytoplasmic RNA fractions were purified with equal volume of elution buffer by using a PARIS Protein and 299 RNA isolation kit (Ambion, Life Technologies) following the manufacturer's instructions. Equal volume RNA 300 was reverse transcribed into cDNA using the PrimeScript RT Reagent Kit with gDNA Eraser (Takara) according 301 to the manufacturer's instructions. Real-time PCR was performed using the SYBR green PCR mixture to 302 calculate the absolute quantification with standard curves for unspliced RNA (gag) and completely spliced 303 RNA (tat). The primers used are as follows: Gag-forward, 5-GTGTGGAAAATCTCTAGCAGTGG-3, Gag-304 reverse, 5-CGCTCTCGCACCCATCTC-3 (52); Tat-forward, 5-CAGCCTAAAACTGCTTGTAC-3, Tat-305 reverse, 5-GGAGGTGGGTTGCTTTGATA-3. 306 307 Formaldehyde cross-linking, Co-immunoprecipitation and Western Blotting. To examine the interactions 308 between proteins in cells, HEK293T cells were transfected with the indicated plasmids using the PEI 309 transfection reagent. Immunoprecipitations were performed using an anti-HA affinity gel (catalog NO. A2095; 310 Sigma-Aldrich) following the manufacturer's instructions. Briefly, cells co-transfected with indicated plasmids 311 were harvested after 48 h and washed with PBS. Formaldehyde cross-linking was subsequently performed. 312 Firstly, cells were crosslinked with 1% formaldehyde for 20 min at RT and quenched by the addition of 0.25M 313 glycine for 5 min. Then the cells were lysed with lysis buffer containing protease inhibitor cocktail (catalog NO. 314 P8340; Sigma-Aldrich), sonicated and centrifuged at 13,000 g for 10 min. 30 μl of the anti-HA affinity gel was 315 washed with ice-cold PBS and incubated with the cell lysates overnight at 4°C. Following the incubation, beads 316 were washed with lysis buffer, and bound proteins were eluted by mixing and heating the beads in sample 317 loading buffer for 5 min at 98°C. Samples were loaded on a 12% Tris gel (Bio-Rad). The gels were transferred 318 to nitrocellulose membranes and blocked with 5% nonfat dry milk (NFDM) for 1 h and incubated with the 319 appropriate primary antibodies. After extensive washing with TBST, the gels were further incubated with the 320 relevant secondary antibody for 1 h at RT. Target protein bands were detected and analysis performed using 321 Licor Odyssey (USA).

346
All experiments were performed independently at least three times, with a representative experiment being shown. All 347 statistical analyses were performed in the GraphPad Prism using student's t test for pairwise and one-way analysis of 348 variance (ANOVA)for multiple comparisons. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, NS, not significant 349 (p > 0.05).

ACKNOWLEDGEMENTS 351
The authors wish to thank W. Barclay, Dr. Hui Zhang, Dr. Yong-Hui Zheng, and Dr. Jianhua Wang for provision 352 of the plasmids. 353

FUNDING 354
This study was supported by grants from the National Natural Science Foundation of China to Xiaojun Wang 355 (81561128010 and 31222054) 356

CONFLICT OF INTEREST 357
The authors declare that they have no conflicts of interest with the contents of this article. transcripts. HIV-1 pseudotype virions were first treated with DNase and then used to infect WT and DKO cells. 540 2, 6, and 18 h later, cellular DNAs were extracted from these infected cells, and the viral early (A) and late 541 reverse transcripts (B) were quantified by real-time PCR using specific primers and probes. Globin mRNA was 542 measured as an endogenous control. (C) Total RNA was extracted from transfected cells and analyzed by real-543 time PCR using primers specific for unspliced mRNA. β-actin mRNA was measured as an endogenous control.