Nuclear restriction of HIV-1 infection by SUN1

Overexpression of the human Sad-1-Unc-84 homology protein 2 (SUN2) blocks HIV-1 infection in a capsid-dependent manner. In agreement, we showed that overexpression of SUN1 (Sad1 and UNC-84a) also blocks HIV-1 infection in a capsid-dependent manner. SUN2 and the related protein SUN1 are transmembrane proteins located in the inner membrane of the nuclear envelope. The N-terminal domains of SUN1/2 localizes to the nucleoplasm while the C-terminal domains are localized in the nuclear lamina. Because the N-terminal domains of SUN1/2 are located in the nucleoplasm, we hypothesized that SUN1/2 might be interacting with the HIV-1 replication complex in the nucleus leading to HIV-1 inhibition. Our results demonstrated that SUN1/2 interacts with the HIV-1 capsid, and in agreement with our hypothesis, the use of N-terminal deletion mutants showed that SUN1/2 proteins bind to the viral capsid by using its N-terminal domain. SUN1/2 deletion mutants correlated restriction of HIV-1 with capsid binding. Interestingly, the ability of SUN1/2 to restrict HIV-1 also correlated with perinuclear localization of these proteins. In agreement with the notion that SUN proteins interact with the HIV-1 capsid in the nucleus, we found that restriction of HIV-1 by overexpression of SUN proteins do not block the entry of the HIV-1 core into the nucleus. Our results showed that HIV-1 restriction is mediated by the interaction of SUN1/2N-terminal domains with the HIV-1 core in the nuclear compartment.

Overexpression of the human protein SUN2 was originally discovered to affect HIV-1 infection in a screen that tested the ability of interferon-stimulated genes to block HIV-1 infection 21 . We have previously shown that overexpression of the human SUN2 protein blocks HIV-1 infection in a capsid-dependent manner 22 . These experiments suggested that SUN2 might be interacting with the HIV-1 capsid protein, and that restriction is caused by this interaction.
SUN2 and the related protein SUN1 are transmembrane proteins located in the inner membrane of the nuclear envelope. The nuclear envelope separates the nucleoplasm from the cytoplasm and is composed of inner and an outer membrane. The outer nuclear membrane is contiguous with the endoplasmic reticulum. The two membranes are separated by a thin lumen known as the nuclear lamina or lamina propia. The N-terminal domains of SUN1 and SUN2 are located in the nucleoplasm whereas the C-terminal domains are located in the nuclear lamina. The conserved SUN domains, in SUN 1 and SUN2, interact with KASH-domain containing proteins inside the nuclear lamina to form bridges that span both nuclear membranes 23 . These protein bridges have been postulated to be the "Velcro" that links the nucleoskeleton with the cytoskeleton 23 .
We have previously shown that overexpression of the human SUN2 protein blocks HIV-1 infection in a capsid-dependent manner 22 . Here we tested whether overexpression of SUN1 and SUN2 proteins block HIV-1 and other retroviruses. To explore the role of capsid in the ability of SUN1 and SUN2 (SUN1/2) to block HIV-1 infection, we measured the ability of SUN1/2 to bind to the HIV-1 core. Our results demonstrated that SUN1/2 interacts with the HIV-1 capsid in cellular extracts. The use of deletion mutants showed that SUN1/2 proteins bind to the viral capsid using its N-terminal domain, which is located in the nucleoplasm. Furthermore, we found that the N-terminal domain of SUN1/2 governs the ability of the protein to bind HIV-1 capsid and to restrict HIV-1. Interestingly, the ability of SUN1/2 to restrict HIV-1 also correlated with perinuclear localization of these proteins. In addition, we found that restriction of HIV-1 by overexpression of SUN1/2 proteins do not block the entry of the HIV-1 core into the nucleus suggesting that this restriction occurs in the nucleus, which is in agreement with the cellular localization of the N-terminal domain of SUN1/2 proteins. Next, we tested whether cells that do not endogenously expressed SUN1 and SUN2 protein plays a role on HIV-1 infection. To this end, using the CRISPR/Cas9 system, we generated human HAP-1 cells that are knockout for the expression of SUN1, SUN2, or SUN1/SUN2. We found that HAP1 SUN1, SUN2, or SUN1/SUN2 knockout cells were permissive to HIV-1 infection, suggesting that this system is not providing evidence for a functional contribution of SUN1 or SUN2 to HIV-1 infection.

Ability of SUN1 to block HIV-1 and other retroviruses.
To test the ability of SUN1 to block HIV-1 infection, we challenged human HT1080 cells stably expressing SUN1 with increasing amounts of HIV-1-GFP. As shown in Fig. 1, SUN1 potently blocks HIV-1 infection (eightfold). By contrast overexpression of SUN1 was not able to block HIV-1 viruses bearing the capsid mutation G208R. This is in agreement with previous results suggesting that capsid is the viral determinant for the ability of the related protein SUN2 to block HIV-1 infection 22 . Next, we tested whether other retroviruses are restricted by SUN1 overexpression. We found that other primate lentiviruses such as human immunodeficiency virus type 2 (HIV-2), and simian immunodeficiency virus (SIV mac ), were poorly restricted by SUN1 (Fig. 1). Similarly, feline immunodeficiency virus (FIV), bovine immunodeficiency virus (BIV), equine infectious anemia virus (EIAV), and B-tropic murine leukemia virus (B-MLV), were also poorly restricted by SUN-1 (Fig. 1). Overall, overexpression of SUN1 only restricted HIV-1. Next we decided to study the SUN1 determinants for HIV-1 restriction.
Capsid binding ability of SUN1. Our previous experiments suggested that the N-terminal domain of SUN1 is required for its ability to block HIV-1 infection, which is in agreement with the fact that HIV-1 can only interact with the N-terminus of SUN1 in the nucleoplasm. One hypothesis is that the N-terminal domain of SUN1 is interacting with the HIV-1 replication complex in the nucleus consequentially affecting HIV-1 infection. Because capsid mutations such as G208R abolished the ability of SUN1 to block HIV-1, we tested whether SUN1 binds to the HIV-1 capsid. To this end, we tested whether SUN1 binds to in vitro assembled HIV-1 CA-NC complexes, which recapitulates the surface of the HIV-1 core 24 . As shown in Fig. 3 and Table 1, SUN1 binds to in vitro assembled HIV-1 CA-NC complexes. Interestingly, deletions SUN1-∆(1-40), SUN1-∆(1-60) and SUN1-∆(1-80) bind to in vitro assembled HIV-1 CA-NC complexes similar to wild type SUN1( Fig. 3 and Table 1). By contrast, deletion SUN1-∆(1-100) lost its ability to bind in vitro assembled HIV-1 CA-NC complexes ( Fig. 3 and Table 1). These experiments showed that loss of binding correlates with loss of restriction. Ability of SUN2 to block HIV-1 and other retroviruses. Next we tested the ability of SUN2 to block HIV-1 infection. As we have previously shown 22 , SUN2 blocks HIV-1 infection (threefold). Although overexpression of SUN2 blocks HIV-1 infection (Fig. 4), we observed that this block is less potent when compared to SUN1. Like SUN1, SUN2 poorly affected the infection of HIV-1 bearing the capsid mutation G208R, which is in agreement with our previous results suggesting that capsid mutations overcome the HIV-1 restriction imposed by overexpression of SUN2 22 . Like in the case of SUN1, these experiments suggested that capsid is the viral determinant for the restriction of SUN2. Next we tested whether other retroviruses are restricted by SUN2 overexpression. We found that SUN2 modestly restricts HIV-1 and HIV-2 (threefold), but not SIV mac (Fig. 4).

Capsid binding ability of SUN2. Our previous experiments suggested that the N-terminal domain of
SUN2 is required for its ability to block HIV-1 infection, which is in agreement with the fact that the N-terminus of SUN2 is localized to the nucleoplasm. In the same manner as SUN1, we tested the hypothesis that the N-terminal domain of SUN2 is interacting with the HIV-1 replication complex in the nucleus consequentially affecting HIV-1 infection. Because capsid mutations such as G208R and P207S abolished the ability of SUN2 to block HIV-1 22 , we also tested whether SUN2 binds to the HIV-1 capsid. As shown in Fig Capsid Binding ability of SUN1 and SUN2 to mutants G208R and P207S. Next we tested the ability of SUN1 and SUN2 to bind capsid mutants G208R and P207S. In agreement with the inability of SUN1 and SUN2 to block infection of HIV-1 viruses bearing the capsid changes G208R or P207S, we saw that SUN1 and SUN2 showed decreased binding to capsid bearing G208R or P207S changes when compared to wild type (Fig. 7A). These experiments demonstrated that the inability of SUN1 and SUN2 to block HIV-1 viruses bearing capsid changes G208R and P207S is due to the lose of capsid binding. To control for the bona fide folding of the capsid mutants, we tested the ability of capsid mutants to interact with TRIMCyp in presence or absence of cyclosporine A (CsA). As shown in Fig. 7B, TRIMCyp was able to interact with capsid mutants at wild type levels. Overall these results correlate SUN1/2 binding with restriction.

Subcellular localization of SUN1 and SUN2 variants. Previous experiments have defined that SUN1
and SUN2 localized to the lamina propia 25 . To understand weather the subcellular localization of SUN1 and SUN2 are important for their ability to block HIV-1 infection, we analyzed subcellular localization of the different SUN1 and SUN2 variants. As shown in Fig. 8, SUN1 and SUN2 showed distinct perinuclear localization, which according to the literature, corresponds to the lamina propia. Interestingly, the localization of the deletion mutant SUN1-∆(1-20) resembles the subcellular localization of the wild type SUN1 protein (Fig. 8A), suggesting that this deletion did not lose wild type localization. Although all other SUN1 deletions showed a perinuclear localization, they also displayed a more intense nucleoplasm staining (Fig. 8A). SUN2 deletion constructs showed a perinuclear staining similar to wild type SUN2. In addition, deletion constructs showed a more intense nucleoplasm staining (Fig. 8B). Interestingly, all study deletions localized to the perinuclear region suggesting that these constructs were not majorly affected by the deletions, indirectly indicating that folding was not greatly disrupted. Perinuclear staining was quantified by examining 50 cells for SUN1-FLAG, SUN2-FLAG and mutants (Fig. 8A,B). Our results showed that for each sequential deletion mutant, there is some nucleoplasmic staining in addition to the pre-dominant peri-nuclear staining. Table 1. Phenotypes of SUN1 variants. a Restriction was measured by infecting cells expressing the indicated SUN1 variant with HIV-1-GFP. After 48 h, the percentage of GFP-positive cells (infected cells) was determined by flow cytometry. "++": indicates strong restriction, "+" indicates moderate restriction, "−": indicates absence of restriction. b Binding to the HIV-1 capsid complexes was determined for each SUN1 variant as described in "Methods". "+": indicates binding, "−": indicates no binding. c Wild type and mutant SUN1-FLAG proteins were assayed to determine localization. "PS" indicates Perinuclear staining, "NP" indicates Non-Perinuclear localization. "PS + NP" indicates both perinuclear and non-peri-nuclear staining is observed. www.nature.com/scientificreports/

Inhibition of HIV-1 infection by SUN1 does prevent the entry of capsid into the nucleus.
Overexpression of SUN1 in human cells potently inhibit HIV-1 infection after reverse transcription but prior to integration 22 . Because the occurrence of reverse transcription correlates with entry of capsid into the nucleus [26][27][28] , and the SUN1 domain that interacts with capsid is in the nucleoplasm; we tested whether SUN1 inhibition of HIV-1 allows the entry of capsid into the nucleus. For this purpose, using our published methodology 26 , we tested whether capsid is imported into the nucleus of HIV-1 restricted cells by SUN1. To this end, human HT1080 cells expressing SUN1 were challenged with HIV-1 using an MOI = 2. Eight hours post infection, cells were separated into cytosolic (C) and nuclear (N) fractions. C and N fractions were analyzed for capsid content using anti-P24 antibodies (Fig. 9). To ensure bona fide origin of the cellular fractions, we performed Western blot analysis using anti-Nopp140 and anti-tubulin as nuclear and cytosolic markers, respectively. As shown on Fig. 9, we found that expression of SUN1 does not inhibit the entry of capsid into the nucleus. As controls, we used the small molecule PF74 that potently blocks the entry of capsid into the nucleus 26 . These experiments suggest that HIV-1 restriction by SUN1 occurs in the nucleus, which is in agreement with findings showing that the region of SUN1 that interact with capsid is in the nucleoplasm. Overall these experiments suggest that the restriction of HIV-1 by overexpression of SUN1 occurs in the nuclear compartment. To directly address this question, we knockout the expression of SUN1 and/or SUN2 in human haploid HAP-1 cells by using the CRISPR/Cas9 system. As shown in Fig. 10A, sequencing of the SUN1 and/or SUN2 alleles revealed deletions that resulted in the generation of premature stop codons (Fig. 10A). Furthermore, we tested expression of endogenous SUN1 and SUN2 in the KO HAP-1 cells. As shown in Fig. 10B, HAP-1 cell lines containing the premature stop codons did not expressed SUN1/2 proteins. Next we tested whether HAP-1 cells that do not express SUN1 and/or SUN2 were infected by HIV-1-GFP (Fig. 10B). As a control, we used a clone that showed an intact sequence for SUN1 and SUN2 (HAP-1 CTRL). Interestingly, knocking out the expression of these genes did not affect HIV-1 infection (Fig. 10B,C). These experiments suggested that expression of SUN1 and/or SUN2 is not important for HIV-1 infection of HAP-1 cells.

Discussion
This work and others have established that overexpression of SUN1 and SUN2 blocks HIV-1 infection, and this process depends upon an intact capsid protein 21,22,29 . The present work showed that SUN1/2 is interacting with capsid in the nuclear compartment to achieve HIV-1 restriction. To this end, we tested the ability of SUN1 and SUN2 to bind to in vitro assembled HIV-1 CA-NC complexes. Our data showed that SUN1/2 proteins in cell lysates interact with the HIV-1 core, in agreement with previous findings 29 . This interaction is likely to occur in the nucleoplasm where the N-terminal of SUN1/2 is located. One possibility is that the interaction between the N-terminal domain of SUN1 or SUN2 with the pre-integration complex negatively affects infection by changing the localization of the core in the nuclear compartment. Since the N-terminal of SUN1 or SUN2 is the only segment of the protein that might be interacting with HIV-1, we performed SUN1/2N-terminal deletions and tested for restriction. In agreement with the hypothesis that the N-terminal domain of SUN1/2 are important for restriction, we found that deletion of N-terminal residues renders SUN1/2 inactive against HIV-1 infection. Our mapping experiments revealed that residues 20-100 and 30-60 are important for HIV-1 restriction of SUN1 and SUN2, respectively. These experiments demonstrated that the N-terminal region of SUN1/SUN2 is important for the ability of the protein to block HIV-1. Similar results for SUN1 were observed by others 29,30 .
Next, we investigated whether a correlation between restriction and capsid binding exists. Interestingly, N-terminal deletion mutants that did not bind capsid did not restrict HIV-1. In addition, we demonstrated that capsid mutations G208R and P207S prevent the ability of capsid to bind to SUN1/2 proteins; this is in agreement with the inability of SUN1/2 to restrict HIV-1 viruses bearing changes G208R and P207S. These results correlated binding to capsid with HIV-1 restriction. Overall this is in agreement with the hypothesis that the N-terminal domain of SUN1/2 is interacting with capsid during infection and that this interaction is important for restriction. SUN1/2 proteins mainly localize to the lamina propia. To understand whether the localization of SUN1/2 correlates with restriction, we investigated the subcellular localization of SUN1/2N-terminal deletion mutants and correlated this to restriction. Our observations revealed that intact restriction is only achieved by an intact localization of SUN1/2. Although N-terminal deletion mutants remain in the lamina propia (perinuclear localization), we observed that HIV-1 restriction by SUN1 and SUN2 correlates with perinuclear localization. Deletion mutants that exhibit perinuclear and nonperinuclear localization were less or none restrictive to HIV-1. Therefore, SUN1/2 restriction correlates with capsid binding and perinuclear localization. Similar localization for SUN1 deletion mutants was observed by Schaller et al. 29 .
Our results showed that the nuclear membrane protein SUN1/2 requires its N-terminal domain to interact with HIV-1 capsid. This interaction is likely to take place in the nucleus and it is important for the ability of the protein to block HIV-1 infectivity. Interestingly, our experiments demonstrated that the SUN1 block to HIV-1 infection is after the viral core has been transported into the nuclear compartment, which is in agreement with the notion that SUN1 interacts with the HIV-1 core in the nucleus. Previous observations have shown that Table 2. Phenotypes of SUN2 variant. a Restriction was measured by infecting cells expressing the indicated SUN2 variant with HIV-1-GFP. After 48 h, the percentage of GFP-positive cells (infected cells) was determined by flow cytometry. "++": indicates strong restriction, "+" indicates moderate restriction, "−": indicates absence of restriction. b Binding to the HIV-1 capsid complexes was determined for each SUN2 variant as described in "Methods". "+": indicates binding, "−": indicates no binding, "+/−" decreased binding. c Wild type and mutant SUN2-FLAG proteins were assayed to determine localization. "PS" indicates Perinuclear staining, "NP" indicates Non-Perinuclear localization. "PS + NP" indicates both perinuclear and non-peri-nuclear staining is observed.

SUN2
Restriction of HIV- www.nature.com/scientificreports/ Figure 6. Capsid Binding ability of wild type and mutant SUN2 proteins. HEK293T cells were transiently transfected with plasmids expressing wild type and mutant SUN2-FLAG proteins. Thirty-six hours after transfection, cells were lysed. The lysates were incubated with in vitro assembled HIV-1 CA-NC complexes at room temperature for 1 h. The mixtures were applied onto a 70% sucrose cushion and centrifuged as described in methods. INPUT represents the lysates analyzed by Western blotting before being applied to the 70% cushion. The INPUT fraction was analyzed by Western blotting using anti-FLAG antibodies. The pellet from the 70% cushion (BOUND) was analyzed by Western blotting using anti-FLAG and anti-p24 antibodies. The blots for a representative experiment and the standard deviation for three independent experiments are shown. www.nature.com/scientificreports/ overexpression of SUN1/2 proteins blocks HIV-1 infection after reverse transcription but before integration 22 . Because reverse transcription occurs in the nuclear compartment the interaction of SUN1 with the HIV-1 core in the nucleus is likely to disrupt the subsequent steps of replication [26][27][28] . One possibility is that SUN1 binding to capsid sequesters the viral complexes away from critical sites for replication such as open chromatin sites. Overexpression of SUN1 or SUN2 blocks HIV-1 infection; however, the role of endogenously expressed SUN1 and SUN2 in HIV-1 replication is not clear. For this purpose, we generated knockouts for the expression of SUN1 and/or SUN2 in human haploid HAP-1 cells, and investigated whether HIV-1 infection is affected. To our surprise, complete depletion of SUN1 and/or SUN2 by using CRISPR/Cas9 did not affect HIV-1 replication suggesting that endogenous SUN1 and SUN2 does not have a role in HIV-1 replication of haploid HAP-1 cells. In agreement with our results, knocking out the expression of SUN1 using CRISPR/Cas9 in THP-1 cells did not affect HIV-1 infection 29 . However, depletion of SUN2 using CRIPSR/Cas9 in THP-1 cells mildly affected HIV-1 infection. More recent results, depleting SUN1/2 using siRNA in HEK293T cells dramatically decreased infectivity of HIV-1 30 . To further complicate matters, depletion of SUN2 in human primary T cells affects proliferation and viability 31 . Although several laboratories have attempted depletion of this protein further experiments will be necessary to establish whether endogenous expression of SUN1/2 contribute to HIV-1 infection.
Overall our work showed that HIV-1 restriction by overexpression of SUN1/2 is mediated by the interaction of the N-terminal domain of SUN1/2 with the HIV-1 core occurs in the nuclear compartment.

Conclusions
This work provides mechanistic understanding on how the overexpression of SUN1/2 proteins block HIV-1 infection in human cells. Our experiments showed that HIV-1 restriction is mediated by the interaction between the N-terminal domain of SUN1/2 proteins and the HIV-1 core in the nuclear compartment. Protocol was used as outlined 26 . Cells were harvested using trypsin and were washed twice with cold PBS by centrifugation at 4 °C. Supernatant was discarded and cell pellet was re-suspended in PBS. A fraction of the cell  www.nature.com/scientificreports/ suspension was centrifuged and cell pellet was re-suspended in lysis buffer and incubated for 1 h on ice then centrifuged for 1 h at 4 °C. The resulting supernatant was mixed with Laemmli buffer and used to measure the total amount of capsid. The remaining aliquot of the cell suspension was pelleted and re-suspended in lysis buffer and incubated on ice. Subsequently, the sample was centrifuged at 4 °C. The resulting supernatant and pellet correspond to cytosolic and nuclear fractions, respectively. Next, a fraction of the supernatant was mixed with Laemmli buffer and used as cytosolic fraction. The nuclear pellet was washed twice using lysis buffer without NP-40 by gently inverting the tube several times. The sample was then centrifuged and the nuclear pellet was re-suspended in extraction buffer, and incubated on ice. Subsequently, the sample was centrifuged at 4 °C. A fraction of the supernatant was mixed with Laemmli buffer and used as nuclear fraction. Proportional amounts of total, cytosolic, and nuclear fractions were analyzed by western blot using anti-p24, anti-Nopp140, anti-atubulin, or anti-GAPDH antibodies detailed below.
Generation of SUN1, SUN2, and SUN1/SUN2 double KO. Hap-1 SUN1, SUN2, and SUN1/SUN2 double knockout (KO) cell lines were generated by using the clustered regularly interspaced short palindromic repeat (CRISPR)-Cas9 gene system. The CRISPR genomic guide RNA sequences for human monoploid fibroblast HAP-1 cells SUN1 CRISPR KO was created using a 1 bp (GRNA sequence: TGT CTC CCT GAA GAA CCG AG) deletion in exon 2. CRISPR KO of SUN2 in HAP-1 cells was designed to target exon 4. Sanger sequencing Figure 9. Nuclear Import of capsid in SUN1 overexpressed cells. Human HT1080 cells, stably expressing SUN1-FLAG or containing the LPCX empty vector control, were infected with wild-type HIV-1-GFP at an MOI of 2 in the presence of 10uM PF74, or DMSO vehicle control for 8 h. Cells were separated into nuclear and cytosolic fractions and analyze for capsid content by western blotting using anti-p24, anti-Nopp140 and antiα-Tubulin antibodies. The ratio of nuclear to cytosolic capsid for three independent experiments with standard deviations is shown. **p < 0.001; ***p < 0.0005; NS, not significant (as determined by unpaired t test).  Antibodies. Rabbit Monoclonal clone EPR6557 antibody against SUN2 was obtained from Millipore. Rabbit polyclonal antibody against SUN1 was obtained from Genetex. Anti-FLAG rabbit polyclonal, Anti-GAPDH rabbit polyclonal, anti-mouse Alexa 594, and anti-rabbit Alexa 488 were obtained from Sigma.
Immunofluorescence. Transfections of cell monolayers were performed using PIE reagent. Transfections were incubated at 37 °C for 24 h. Indirect immunofluorescence microscopy was performed as previously described 34,35 . Transfected monolayers grown on coverslips were washed twice with PBS1X (NaCl 137 mM, KCl 2.7 mM, Na2HPO4·2H2O 10 mM, KH2PO4 mM) and fixed for 15 min in 3.9% paraformaldehyde in PBS1X. Fixed cells were washed twice in PBS1X, permeabilize for 4 min in permeabilizing buffer (0.5% Triton X-100 in PBS), and then blocked in PBS1X containing 2% bovine serum albumin (blocking buffer) for 1 h at room temperature. Cells were then incubated for 1 h at room temperature with M2-Flag primary antibodies diluted in blocking buffer. After three washes with PBS, cells were incubated for 30 min in anti-mouse secondary antibody. Cells were subsequently washed with PBS and stained with 1 μg/ml of 49,69-diamidino-2-phenylindole (DAPI). Samples were mounted for fluorescence microscopy by using the ProLong Antifade Kit (Molecular