Rhesus monkey TRIM5α protein SPRY domain contributes to AP-1 activation

TRIM5α is an important host restriction factor that could potently block retrovirus infection. The SPRY domain of TRIM5α mediates post-entry restriction by recognition of and binding to the retroviral capsid. Human TRIM5α also functions as an innate immune sensor to activate AP-1 and NF-κB signaling, which subsequently restrict virus replication. Previous studies have shown that the AP-1 and NF-κB signaling activation relies on the RING motif of TRIM5α. In this study, we have demonstrated that the SPRY domain is essential for rhesus macaque TRIM5α to activate AP-1 but not NF-κB signaling. The AP-1 activation mainly depends on all of the β-sheet barrel on SPRY structure of TRIM5α. Furthermore, the SPRY-mediated auto-ubiquitination of TRIM5α is required for AP-1 activation. This study reports that rhesus macaque TRIM5α mainly undergoes Lys27-linked and Met1-linked auto-polyubiquitination. Finally, we found that the TRIM5α signaling function was positively correlated with its retroviral restriction activity. This study discovered an important role of the SPRY domain in immune signaling and antiviral activity and further expanded our knowledge of the antiviral mechanism of TRIM5α.

TRIM5␣ is an important host restriction factor that could potently block retrovirus infection. The SPRY domain of TRIM5␣ mediates post-entry restriction by recognition of and binding to the retroviral capsid. Human TRIM5␣ also functions as an innate immune sensor to activate AP-1 and NF-B signaling, which subsequently restrict virus replication. Previous studies have shown that the AP-1 and NF-B signaling activation relies on the RING motif of TRIM5␣. In this study, we have demonstrated that the SPRY domain is essential for rhesus macaque TRIM5␣ to activate AP-1 but not NF-B signaling. The AP-1 activation mainly depends on all of the ␤-sheet barrel on SPRY structure of TRIM5␣. Furthermore, the SPRY-mediated auto-ubiquitination of TRIM5␣ is required for AP-1 activation. This study reports that rhesus macaque TRIM5␣ mainly undergoes Lys 27 -linked and Met 1 -linked auto-polyubiquitination. Finally, we found that the TRIM5␣ signaling function was positively correlated with its retroviral restriction activity. This study discovered an important role of the SPRY domain in immune signaling and antiviral activity and further expanded our knowledge of the antiviral mechanism of TRIM5␣.
TRIM5␣ is a member of the tripartite motif (TRIM) protein family, members of which are encoded by over 100 genes in humans (10). TRIM proteins have a common structure named RBCC that features three major motifs: N-terminal RING, B-box, and coiled-coil domains. Over half of the TRIMs contain a SPRY/B30.2 domain at the C terminus that mediates protein interactions. In most cases, the RING domain has E3 ligase activity (11,12) and activates Ubc13-ubiquitin (Ub) conjugate through its dimerization (13). The B-box and coiled-coil domains promote its oligomerization, which is required for the TRIM protein to form cytoplasmic or nuclear bodies (14 -18). Furthermore, the RBCC motif was found to contribute to TRIM assembly, which is critical for the TRIM protein's ubiquitination activity (19). The SPRY domain of TRIM5␣ mediates post-entry restriction by recognizing and binding to the retroviral capsid (20 -22). However, the SPRY domain has recently been considered to serve some unknown functions as well as binding to the capsid (23,24). For instance, four putative SUMO-interacting motifs were reported in the SPRY domain, and SUMO-interacting motifs are responsible for the antiviral activity of TRIM5␣ (25)(26)(27). The SPRY domain may have additional functions that require further investigation.
HuTRIM5␣ regulates immune signaling mainly by interacting with mitogen-activated protein kinase kinase kinase 7 (TAK1) and then activates downstream pathways, including activator protein-1 (AP-1) and nuclear factor-B (NF-B) signaling (6,9,28). TAK1 is an important MAP3K activated by unanchored polyubiquitin chains (29). Quite a few studies indicate that TAK1 polyubiquitination is involved in signaling pathways (30 -32). HuTRIM5␣ recruits E2 Ub-conjugating enzymes UBC13-UEV1A by the RING domain to generate free Lys 63 -linked polyubiquitination, resulting in TAK1 activation (9). However, the roles of the C-terminal function of TRIM5␣ in signaling activation and the mechanisms involved are still to be investigated.
This study was planned to evaluate (i) whether and how the SPRY domain of RhTRIM5␣ contributes to innate immune signaling and (ii) whether this function is correlated to its antiviral activity. In this study, it was demonstrated for the first time that the SPRY domain of RhTRIM5␣ was vital for activating the AP-1 signal, but not NF-B. The molecular basis of SPRY for  cro ARTICLE the activation of AP-1 by RhTRIM5␣ was mapped to the ␤-sheet barrel on the SPRY structure. Moreover, AP-1 activation of RhTRIM5␣ was found to be positively correlated with its auto-ubiquitination. A significant finding was that RhTRIM5␣ mainly undergoes Lys 27 -and Met 1 -linked auto-ubiquitination.

SPRY is indispensable for TRIM5␣ activating AP-1 signaling
HuTRIM5␣ has been identified to activate AP-1 and NF-B. RhTRIM5␣ and huTRIM5␣ were compared for their NF-B and AP-1 signaling activation capacity. We found that RhTRIM5␣ stimulated both NF-B and AP-1 transcriptional reporters with magnitudes similar to those of human TRIM5␣ (Fig. 1, A-C). We then tested RhTRIM5␣C15/18A (C15/18A), a RING E3 Ub-ligase domain mutation construct (24). It was found that C15/18A had a reduced capacity to activate NF-B and AP-1 of around 5-6-fold (Fig. 1, D and E), which is similar to the findings of a previous study that used HuTRIM5␣ (9). C terminus-truncated RhTRIM5 with a deletion of the SPRY domain (RhTRIM5⌬S) was constructed to test and determine the functionality of the SPRY domain of TRIM5␣ in signal transduction, (Fig. 1A). Surprisingly, RhTRIM5⌬S did not alter the capacity to activate NF-B, indicating that the SPRY domain may not be necessary for NF-B signal activation

Macaca fascicularis TRIM5␣ is deficient in AP-1 activation
M. fascicularis, a closely related species of rhesus macaque, also encodes TRIM5␣ loci. Interestingly, it was found that M. fascicularis TRIM5␣ (FaTRIM5␣) was deficient in activating AP-1 ( Fig. 2A). By comparing the sequences of FaTRIM5␣ and RhTRIM5␣, 13 amino acid differences were identified, including one located in the N terminus, three in the coiledcoil domain, two in Linker-2, and the other seven in SPRY (Fig. 2B). We further explored whether a single amino acid substitution could alter the signaling activity of RhTRIM5␣. The results revealed that only the mutation of proline to serine at the 453 site on the SPRY (RhTRIM5␣-S453P) aborted its ability to activate AP-1 (Fig. 2C). To confirm this result, HuTRIM5␣-S453P and FaTRIM5␣-P453S were constructed. As expected, HuTRIM5␣-S453P lost the capacity to activate AP-1, whereas FaTRIM5␣-P453S acquired this function (Fig. 2D). These results demonstrated that the mutation of serine at 453 of TRIM5␣ significantly influences its AP-1 activation.
SPRY has a typical ␤-sheet barrel structure. To determine the key amino acids of TRIM5␣ for AP-1 activation, the SPRY structure was artificially marked into four parts: the biggest ␣-helix (red), the ␤-sheets on the same side of Ser 453 (orange), the ␤-sheets on the opposite side of Ser 453 (purple), and marginal small ␤-sheets (green) (Fig. 3E). Further, serial point mutations were constructed, named 1-30 (Fig. 3F), where the marked amino acids were mutated into arginine. Sites on the ␣-helix (marked in red) and the small ␤-sheets (marked in green) did not affect AP-1 activation (Fig. 3G). However, most

SPRY-dependent AP-1 activation
variants on the ␤-sheet barrel lost this function, except for the number 22 mutant (Fig. 3H), which is a smaller ␤-sheet. These results indicated that the whole ␤-sheet barrel of SPRY was vital for TRIM5␣-mediated AP-1 activation.

Polyubiquitination of TAK1 was not sufficient for TRIM5-mediated AP-1 signaling
TAK1-binding protein 2 (TAB2) is involved in the direct interaction between the TAK1 complex and TRIM5␣ for AP-1 signaling activation. To explore the molecular mechanism of SPRY-mediated AP-1 activation, RhTRIM5␣ was compared with two selected inactive mutations in two nearby ␤-sheets of SPRY: RhTRIM5␣S453P and RhTRIM5␣VFVD (number 14 mutant). We found that RhTRIM5␣S453P and RhTRIM5␣VFVD degraded TAB2 to the same level as RhTRIM5␣ (Fig. 4A), indicating that TAB2 degradation is independent of TRIM5␣-mediated AP-1 activation.

SPRY-dependent AP-1 activation
TAK1 polyubiquitination (Fig. 4C). These data indicate that polyubiquitination of TAK1 was not sufficient for TRIM5-mediated AP-1 activation, suggesting that other unknown mechanisms may exist.

Auto-ubiquitination of TRIM5␣ mediated by SPRY is correlated with AP-1 signal activation
Auto-regulated polyubiquitination of many members of the TRIM family has been reported to be involved in regulating innate immune signaling (33,34). In our study, the heavily polyubiquitinated TRIM5␣ was also detected by both HA antibody (Fig. 5A) and TRIM5␣-specific polyclonal antibody (Fig.  5B), as per previous studies (11,(35)(36)(37), whereas the enzymatically inactive mutation, RhTRIM5␣C15/18A, dramatically lost its polyubiquitination (Fig. 5C). This result indicates that TRIM5␣ ubiquitination is auto-regulated. It was further investigated whether RhTRIM5␣S453P and RhTRIM5␣VFVD also lost their auto-ubiquitination activities. As shown in Fig. 5D, RhTRIM5␣ was well auto-ubiquitinated, contrary to RhTRIM5␣S453P and RhTRIM5␣VFVD. This result indicates that RhTRIM5␣ auto-ubiquitination may be responsible for AP-1 activation. To confirm the role of TRIM5␣ auto-ubiquitination in AP-1 signaling, 29 variants of SPRY were tested for auto-regulating polyubiquitination. We found that mutations that activated AP-1 could auto-ubiquitinate (with the single exception of variant 9), whereas mutations that did not activate AP-1 failed to auto-ubiquitinate sufficiently (Fig. 6,  A-D). These findings suggested that TRIM5␣ auto-ubiquitination was positively correlated with TRIM5␣-mediated AP-1 activation.

RhTRIM5␣ is modified with Met 1 -linked and Lys 27 -linked poly-Ub chains in HEK293T cells
It is reported that HuTRIM5␣ is modified with Lys 63 -linked poly-Ub by Ube2W and Ube2N/Ube2V2. To analyze the type of ubiquitin modification of RhTRIM5␣ under conditions of its overexpression, we utilized K63-Ub (Lys 63 only), K48-Ub, K63R (all residues except Arg 63 ), and K48R mutant ubiquitin plasmids and found that RhTRIM5␣ was modified by both Lys 63 -and Lys 48 -linked ubiquitin chains (Fig. 7A). Ubiquitin

SPRY-dependent AP-1 activation
has five lysine residues, which can be ubiquitinated to give rise to isopeptide-linked polyubiquitin chains with the exception of Lys 48 and Lys 63 . To further identify other lysines' contribution to RhTRIM5␣ polyubiquitination, K6-, K11-, K27-, K29-, and K33-Ub plasmids were used to characterize the polyubiquitin linkages within the Ub n -RhTIM5␣. It was clear that RhTRIM5␣ could be heavily polyubiquitinated in the presence of all of these Ub plasmids (Fig. 7B). In addition, Fig. 7C indicated that polyubiquitination of RhTIM5␣ was normal after expression of K0-Ub (seven Lys residues mutated to Arg, Lysless). The results above indicated that RhTRIM5␣ undergoes "atypical" Met 1 -linked polyubiquitination. Intriguingly, formation of Ub n -RhTIM5␣ was efficiently blocked by expression of M-R-K0-Ub (seven Lys residues and Met 1 mutated to Arg) (Fig. 7D). These results revealed that the polyubiquitin linkages within Ub n -RhTIM5␣ contained Met 1linked ubiquitin. Fig. 7E indicates that RhTIM5␣ undergoes other Lys-linked ubiquitination. We designed the experiments with a set of Ub plasmids containing only one Lys and no N-terminal Met 1 , because the mutant ubiquitin plasmids used in Fig. 7 (A and B) contained the N-terminal Met 1 (the ORF of Ub), which could also be ubiquitinated. Fig. 7F shows that RhTIM5␣ undergoes Lys 27 -linked ubiquitination. Together, these results indicate that RhTRIM5␣ is modified with "atypical" Met 1 -linked and Lys 27 -linked poly-Ub chains in HEK293T cells.

AP-1 activity of RhTRIM5␣ was correlated with its antiviral activity
TRIM5␣ proteins have species-specific activities to inhibit retroviruses. RhTRIM5␣ does not inhibit the simian immunodeficiency virus of macaque strains, but it strongly inhibits HIV-1, feline immunodeficiency virus, and equine infectious anemia virus (39 -41). To identify whether RhTRIM5␣ signal transduction activity was correlated with its antiviral activity, we performed a post-entry inhibition assay against HIV-1 using HeLa cells, which stably express all of the RhTRIM5␣ variants. The results indicated that RhTRIM5␣ could significantly inhibit HIV-1, whereas (with the exception of variant 30) the corresponding variants that failed to activate AP-1 also lost their antiviral activities (Fig. 8, A-C). To further understand the

SPRY-dependent AP-1 activation
correlation between the capability of AP-1 activation and restriction of HIV-1, we plotted -fold change of AP-1 activities as a function of the restriction abilities of different RhTRIM5␣ variants. As shown in Fig. 8D, RhTRIM5␣-mediated AP-1 activation and its anti-HIV-1 activity were positively correlated, R 2 ϭ 0.6369. These results demonstrated that the antiviral activities of RhTRIM5␣ variants have a significant positive correlation with their capacities to activate AP-1 when stably expressed in HeLa cells. This finding is consistent with the fact that AP-1-dependent gene up-regulation can result in antiviral activity.

Discussion
Accumulating evidence suggests that the E3-ubiquitin ligase activity of the TRIM family of proteins plays multiple roles in innate antiviral immunity (42)(43)(44). Many TRIM proteins are reported to affect the ubiquitination levels of several signaling adaptors in the NF-B and AP-1 signaling pathways that are dependent on their RING domains (43,45). For example, TRIM21 negatively regulates DNA sensor signaling by enhancing Lys 48 -linked DDX41 ubiquitination (46), whereas TRIM4 promotes type I IFN production against virus infection by targeting RIG-I for Lys 63 -linked ubiquitination (47). Both regulations are dependent on their RING domains. TRIM5␣ has been reported to activate AP-1 and NF-B signaling pathways dependent on the RING domain (9). In this study, we demonstrated that AP-1, but not NF-B, signal activation by TRIM5␣ is dependent on the SPRY domain in addition to the RING domain.
It has been well documented that the SPRY domain is used by RhTRIM5␣ to recognize and interact with HIV-1 CA. Previous studies showed that the SPRY domain serves unknown functions uncoupled from binding retroviral CA (23,24). In this study, we have demonstrated that the SPRY domain functions to regulate RhTRIM5␣-mediated AP-1 signaling. Moreover, we mapped the key amino acids of SPRY for AP-1 signal and illustrated that the whole ␤-sheet barrel of the SPRY domain was Lines in red represent RhTRIM5␣ variants that intensely activate AP-1 signaling. All experiments were performed three times, and a representative result is shown. C, HeLa cells stably expressing all RhTRIM5␣ variants were lysed and used for IB analysis with the indicated antibodies. D, the relative -fold change of AP-1 activity of different RhTRIM5␣ SPRY domain variants was calculated as described in the legend to Fig. 3. Similarly, the capability of the different variants against HIV-1 was calculated as described in A and B. Dots in red represent RhTRIM5␣ variants that intensely activate AP-1 signaling. Correlation analyses indicated that RhTRIM5␣-mediated AP-1 activation and its anti-HIV-1 activity were positively correlated (R 2 ϭ 0.6369).

SPRY-dependent AP-1 activation
required for this function. We conclude that the ␤-sheet barrel of the SPRY domain influences the AP-1 pathway by regulating the auto-ubiquitination of TRIM5␣.
TRIM5␣ has been reported to activate the AP-1 and NF-B signaling pathways by recruiting the E2 Ub-conjugating enzymes UBC13-UEV1A to synthesize free Lys 63 -linked polyubiquitin chains, which could activate TAK1 in the presence of TAB2 and TAB3. TAB2 and TAB3 are required for TAK1 activation as redundant receptors to preferentially bind to Lys 63linked polyubiquitin chains (48,49). Many signal transduction molecules have been reported to negatively regulate TAB2-dependent NF-B and AP-1 signaling via degradation of TAB2, including TRIM22 (50), mouse TRIM30␣ (51), and TRIM38. TRIM38 inhibits IL-1␤-and TNF␣-triggered signaling by mediating the degradation of TAB2/3, which depends on the SPRY domain but not the RING domain (52). However, there were two controversial findings for RhTRIM5␣ in 2011 (24,53). One study indicated the potential of RhTRIM5␣ to promote TAB2 degradation, resulting in the repression of HIV-1 LTR promoter activity by the negative regulation of NF-B activation (53); another study demonstrated that the capacity of TRIM5␣ to negatively regulate TAB2 levels is present in human and mouse TRIM5␣ but not in rhesus TRIM5␣ (24). Our data supported the former finding that RhTRIM5␣ intensely degrades TAB2 (Fig. 4A). It also identified that this function is independent of the SPRY domain when RhTRIM5␣ is overexpressed (data not shown). Interestingly, and similar to TRIM38, RhTRIM5␣ overexpression activates NF-B and AP-1 in reporter assays regardless of the fact that it intensely degrades TAB2 (44,52). Therefore, NF-B and AP-1 activation by TRIM5␣ will increase when TAB2 is compensated (44,52). TRIM5␣ may also exhibit an unknown function in other pathways, as well as the pathway sensing the retrovirus CA lattice. This result is similar to the action of TRIM21, which has been reported to show differential regulation in different cells or with different stimuli (54 -58). The role of TRIM5␣ in other signaling pathways needs to be further investigated.
Several reports have suggested that TAK1 polyubiquitination is involved in TAK1-mediated activation of the NF-B and AP-1 signaling pathways (29,32). Many molecules have been reported to regulate innate immunity by ubiquitinating TAK1. For instance, TRIM8 enhances TNF␣and IL-1␤-triggered NF-B activation by targeting TAK1 for Lys 63 -linked polyubiquitination (31). In addition, TRAF6-catalyzed Lys 63 -linked polyubiquitination sufficiently activates NF-B and AP-1 by activating TAK1 (32). Likewise, Pertel et al. (9) provided evidence for a model in which TRIM5␣ interacts with TAK1 to activate NF-B and AP-1 signaling by catalyzing unanchored Lys 63 -linked polyubiquitination. In this study, we also found that RhTRIM5␣ interacts with TAK1 for polyubiquitination. However, the inactive mutants RhTRIM5␣S453P and RhTRIM5␣VFVD reserved their capacities to bind and polyubiquitinate TAK1. These results revealed that TAK1 polyubiquitination is not sufficient for TRIM5␣-mediated AP-1 activation.
TRIM5␣ auto-ubiquitination has been reported previously. There are many E2 ubiquitin ligases involved in TRIM5␣ autoubiquitination, including UbcH5B, Ubc2W, Ube2N/Ube2V2, and Ube2D3 (11,35,58). The RING and B-box 2 domains have been reported to contribute to auto-ubiquitination of TRIM5␣ (36,59). Furthermore, previous studies have concluded that auto-ubiquitination is a critical event in the TRIM5␣ restriction mechanism (11,36). Coincidentally, in our studies, we revealed that auto-ubiquitination also plays an important role in AP-1 activation. It was also found that AP-1 activity of RhTRIM5␣ was correlated with its antiviral activity, which is consistent with the work of the Luban laboratory on six murine Trim5 homologues restricting retroviruses (60). Therefore, at least a portion of the antiviral activity of RhTRIM5␣ is mediated through AP-1 activation. We further detected that RhTRIM5␣ auto-ubiquitination and AP-1 activation are both affected by the SPRY domain and especially by the ␤-sheet barrel of the SPRY structure. These results suggested that the SPRY domain is vital for RhTRIM5␣ auto-polyubiquitination to further activate AP-1.
It has been reported that HuTRIM5␣ employed Ube2W and Ube2N/Ube2V2 to anchor and elongate the Lys 63 -linked poly-Ub chains, respectively, in a process of HuTRIM5␣ autopolyubiquitination in cells and in vitro (38). Our studies, however, have revealed that RhTRIM5␣ was not modified with Lys 63 -linked poly-Ub chains when overexpressed in 293T cells. A surprising find was that RhTRIM5␣ undergoes "atypical" Met 1 -linked and Lys 27 -linked auto-ubiquitination in HEK293T cells. Although the reason for this difference has not been confirmed, it is now clear that the Lys 63 -linked and Met 1 -linked ubiquitin chains can mediate signal transduction by non-degradative mechanisms (61). Several key signal transduction molecules have so far been identified to be Met 1 -Ub substrates, including NEMO (62), RIPK1 (63), RIPK2 (64) and others. Our study is the first to state that RhTRIM5␣ is a new target in cells for Met 1 -Ub linkage, although we have no in vitro experiments for polyubiquitination of RhTRIM5␣. However, few substrates are known for chains linked through Lys 27 , and this area is poorly understood and little investigated. Several studies suggest that Lys 27 -linked chains may play roles in the DNA response and recruitment of proteins (65)(66)(67), but to date, the role of Lys 27linked chains of RhTRIM5␣ is unidentified.
It has been demonstrated that the auto-polyubiquitination of TRIM protein functions as a platform that facilitates proteinprotein interactions in signal transduction (68,69). TRIM14 recruits the NF-B essential modulator (NEMO) to the MAVS signalosome via Lys 63 -linked polyubiquitin chains (70); Lys 63linked auto-polyubiquitination of TRIM9S (short splice variant) serves as a platform between GSK3␤ and TBK1, leading to the activation of IRF3 signal, and auto-ubiquitination of TRIM11 is required for its binding of p62, resulting in the suppression of AIM2 inflammasome. However, the molecular details about the role of anto-ubiquitination of RhTRIM5␣ in the innate immunity require further study.
Overall, our results have revealed that the SPRY domain of RhTRIM5␣ is critical for AP-1 signaling, and this function may correlate with auto-ubiquitination of RhTRIM5␣. We conclude that RhTRIM5␣ was modified with Met 1 -linked and Lys 27linked polyubiquitination in 293T cells. In addition, we discovered that the RhTRIM5␣-mediated signal activity was positively correlated with its antiviral activity.

Protein extraction and Western blotting
After the cells were transfected for 24 -36 h, they were washed once with ice-cold PBS and subsequently lysed in icecold radioimmune precipitation assay (RIPA) lysis buffer (CST, 9086) containing a protease inhibitor mixture (Sigma, P8340). The lysates were centrifuged at 12,000 ϫ g for 10 min at 4°C to remove the cell debris. The proteins were separated on a 12% Tris/glycine gel or 4 -12% BisTris gel (Genscript) and blotted onto PVDF membranes (Millipore). The filters were blocked in 5% BSA in Tris-buffered saline (TBS) and were then probed with the indicated primary antibodies followed by DyLight 680 or 800 -labeled secondary antibody (KPL). Signals were visualized using a LI-COR Odyssey imaging system with both 700 and 800 channels.

Cloning and plasmids
RhTRIM5␣ and FaTRIM5␣ were cloned into pcDNA3.1(ϩ) at the EcoRI and XhoI sites and fused to an N-terminal HA tag using PCR. All mutants, including the truncated variants, were introduced by site-directed mutagenesis, and the mutations were confirmed by sequencing. Human TRIM5␣, AP-1-luc, pRL-TK (Renilla luciferase internal control reporter plasmid), and pCDNA-TAK1 were gifts from Dr. Jeremy Luban. NF-Bluc plasmid was from Beyotime. FLAG-TAB2 was cloned into the pEF/V5-HisB vector from pCDNA-TAB2 (a gift from Dr. Jeremy Luban) using BamHI and XhoI sites, including the N-3ϫFLAG tag and the TAB2 ORF. Ub plasmids were constructed with an HA or FLAG tag at their N termini in-house. For creation of TRIM5-expressing HeLa cell lines, pLPCX, the MLV Gag polymerase expression vector pCGP, and pVSV-G were purchased from Clontech. pFUGW is an HIV-1-based transfer vector with enhanced green fluorescent protein expression, and psPAX2 and pMD2.G encode HIV-1-gag-pol and the vesicular stomatitis virus glycoprotein, respectively.

Luciferase assays
HEK293T cells were plated on 96-well culture plates (Corning) at a concentration of 5 ϫ 10 4 cells/well 14 h before transfection. Cells were transfected with poly Jet (SignaGen Laboratories), using 0.3 l of poly Jet/well, with 1 ng of the internal control reporter plasmid pRL-TK, 10 ng of firefly luciferase experimental reporter plasmid AP-1-luc or NF-B-luc, and 90 ng of individual TRIM plasmids or empty pcDNA3.1 as a control, following the manufacturer's instructions. Each experimental condition was performed in quadruplicate. After the cells were transfected for 24 h, luciferase activity was measured with a Dual-Glo luciferase assay system (Promega) using an EIA/RIA plate (Corning, 3693). Firefly luciferase values were normalized to Renilla luciferase values, and the data are represented as -fold inductions compared with empty pcDNA3.1(ϩ). Three independent transfection experiments were performed.

Immunoprecipitation
Thirty hours after transfection, cells were washed once with ice-cold PBS and subsequently lysed in 300 l of ice-cold RIPA lysis buffer (CST, 9086) containing a protease inhibitor mixture (Sigma, P8340) and PMSF. Cell lysates were scraped off the surface of the plate and transferred to prechilled 2-ml microcentrifuge tubes. The lysates were centrifuged at 12,000 ϫ g for 10 min to remove cell debris. Then 50 l of the clarified lysate was diluted in 5ϫ loading buffer, boiled at 100°C for 5 min, and stored at Ϫ80°C, and the remaining 250 l of cell lysate was incubated with monoclonal anti-HA/anti-FLAG-agarose (A2095/A2220, Sigma) at 4°C overnight. The beads were then washed four times with prechilled PBS and resuspended in 50 l of 2ϫ loading buffer. The immunoprecipitated proteins and input lysates were detected by Western blotting.

In vivo TAK1 polyubiquitination
HEK293T cells were co-transfected with the HA-tagged TRIM expression vector, FLAG-tagged TAK1, and HA-tagged Ub expression vector at a 1:1:1 ratio using a normal calcium phosphate method. Protein was extracted in 300 l of ice-cold RIPA lysis buffer 30 h post-transfection. The TAK1-ubiquitin complexes were immunoprecipitated using anti-FLAG antibody M2-conjugated beads (A2220, Sigma) and immunoblotted with anti-HA antibody to detect ubiquitinated proteins.

In vivo TRIM5 auto-ubiquitination
HEK293T cells were co-transfected with plasmids encoding HA/FLAG-tagged WT or mutant TRIM5 together with a plasmid expressing FLAG/HA-tagged ubiquitin at a 1:1 ratio using a normal calcium phosphate method. Thirty-six hours later, cells were harvested and lysed in RIPA buffer. Extracts were immunoprecipitated using anti-HA-agarose or anti-FLAGagarose (the tag of ubiquitin). Immunoprecipitated proteins were Western blotted and separately probed with an anti-FLAG or anti-HA antibody to detect polyubiquitinated TRIM5.

Infection with HIV-1 expressing GFP
Pseudotyped HIV-1 expressing GFP was prepared as shown before (71). For infections, HeLa cells stably transduced with SPRY-dependent AP-1 activation TRIM5 genes of interest or empty vector were seeded at a density of 1.5 ϫ 10 5 cells/well in a 24-well plate, incubated with this pseudotyped virus (described above) for 48 h. The cells were collected and then analyzed using flow cytometry (Beckman, FC500MCL/MPL).