1. Microbiology and Infectious Disease
Download icon

TRIM28 promotes HIV-1 latency by SUMOylating CDK9 and inhibiting P-TEFb

  1. Xiancai Ma
  2. Tao Yang
  3. Yuewen Luo
  4. Liyang Wu
  5. Yawen Jiang
  6. Zheng Song
  7. Ting Pan
  8. Bingfeng Liu
  9. Guangyan Liu
  10. Jun Liu
  11. Fei Yu
  12. Zhangping He
  13. Wanying Zhang
  14. Jinyu Yang
  15. Liting Liang
  16. Yuanjun Guan
  17. Xu Zhang
  18. Linghua Li
  19. Weiping Cai
  20. Xiaoping Tang
  21. Song Gao
  22. Kai Deng
  23. Hui Zhang  Is a corresponding author
  1. Sun Yat-sen University, China
  2. Shenyang Medical College, China
  3. Sun Yat-sen University Cancer Center, China
  4. Guangzhou Eighth People’s Hospital, China
Research Article
  • Cited 5
  • Views 1,776
  • Annotations
Cite this article as: eLife 2019;8:e42426 doi: 10.7554/eLife.42426

Abstract

Comprehensively elucidating the molecular mechanisms of human immunodeficiency virus type 1 (HIV-1) latency is a priority to achieve a functional cure. As current 'shock' agents failed to efficiently reactivate the latent reservoir, it is important to discover new targets for developing more efficient latency-reversing agents (LRAs). Here, we found that TRIM28 potently suppresses HIV-1 expression by utilizing both SUMO E3 ligase activity and epigenetic adaptor function. Through global site-specific SUMO-MS study and serial SUMOylation assays, we identified that P-TEFb catalytic subunit CDK9 is significantly SUMOylated by TRIM28 with SUMO4. The Lys44, Lys56 and Lys68 residues on CDK9 are SUMOylated by TRIM28, which inhibits CDK9 kinase activity or prevents P-TEFb assembly by directly blocking the interaction between CDK9 and Cyclin T1, subsequently inhibits viral transcription and contributes to HIV-1 latency. The manipulation of TRIM28 and its consequent SUMOylation pathway could be the target for developing LRAs.

https://doi.org/10.7554/eLife.42426.001

eLife digest

The human immunodeficiency virus-1, or HIV-1, infects certain human cells, including white blood cells. One reason the infection is incurable is because the virus can integrate its genetic information into its host, and essentially ‘sleep’ within the host cell, a process called latency. This helps to hide HIV-1 from the immune system and stops it getting destroyed.

Latency represents a critical challenge in treating and curing HIV-1. One proposed cure for HIV-1 involves ‘shocking’ the viruses out of latency so that they can be eliminated. Applying this so-called shock and kill approach means scientists need to understand more about how latency is maintained. Previous evidence shows that latency requires proteins known as histone deacetylases and histone methyltransferases. Certain gene-silencing proteins called transcription suppressors are also involved.

Ma et al. have now examined latent HIV-1 in several kinds of human cells grown in the laboratory. The cells were modified to make certain proteins at much lower levels than normal. The experiments showed that the loss of a protein called TRIM28 ‘wakes up’ latent HIV-1. TRIM28 attaches chemical marks called SUMOylations to gene regulators in the cell. These SUMOylations restrict the activity of HIV-1’s genes, which is important to maintain latency. Specifically, TRIM28 adds SUMOylations to a protein named CDK9 at three key positions.

Reducing the levels of TRIM28 made it easier to shock many HIV-1 in infected cells out of latency. With further investigation, targeting TRIM28 may one day be used to treat HIV-1 infection through a shock and kill method.

https://doi.org/10.7554/eLife.42426.002

Introduction

Despite the suppressive combined antiretroviral therapy (cART), the persistence of HIV-1 in the latent reservoirs is the major obstacle to achieve a cure (Chun et al., 1997; Finzi et al., 1997; Wong et al., 1997). To completely eradicate the reservoir, it needs almost 73.4 years of cART due to its long half-life in resting CD4+ T cells (Siliciano et al., 2003). Although over 200/106 resting CD4+ T cells contain proviruses, only 1/106 resting CD4+ T cells (or 1/200 of them) contain inducible replication-competent proviruses and 40/106 resting CD4+ T cells contain intact non-inducible proviruses (Eriksson et al., 2013; Ho et al., 2013). Most of the proviruses are defective, some of which can be induced to produce functional viral proteins and exposed to immunosurveillance (Ho et al., 2013; Pollack et al., 2017). Most of the integration sites locate in the intron of actively transcribed genes (Schröder et al., 2002). Some integration hotspots were found in latently infected clonally expanded CD4+ T cells in HIV-1 patients on cART (Cohn et al., 2015; Maldarelli et al., 2014; Wagner et al., 2014). To decrease the latent reservoirs, several functional cure strategies which are defined as a long-term control of HIV-1 replication and remission of the symptoms of HIV-1 infection without cART, have been proposed (Katlama et al., 2013). The latently infected resting CD4 +T cells do not produce sufficient viral antigens which are recognized by immune system. Thus, the infected cells can hardly be eradicated. To this end, the ‘shock and kill’ strategy, which is one of the functional cure strategies, has been introduced and extensively performed these years. (Deeks, 2012; Geng et al., 2016b; Liu et al., 2016; Liu et al., 2015). Based on the ‘shock and kill’ strategy, the inducible proviruses are ‘shocked’ out by latency reversing agents (LRAs). Then the immune surveillance system recognizes and ‘kills’ these HIV-1-expressing cells utilizing various ways which include CTL response and antibody-dependent cell-mediated cytotoxicity (ADCC). However, some infected cells harbor non-inducible proviruses which can hardly be reactivated by LRAs. Permanent silence of proviruses, accompanied by potent anti-HIV-1 immune surveillance, have been proposed as another strategy to inactivate proviruses in infected cells (Gallo, 2016; Kessing et al., 2017; Liu et al., 2015; Mousseau et al., 2012; Mousseau et al., 2015; Shan et al., 2012). Further elucidating the mechanisms of HIV-1 latency will help us to better understand the formation and maintenance of viral reservoirs and develop new therapeutic interventions.

Epigenetic regulations contribute to the establishment and maintenance of HIV-1 latency. Both histone deacetylases including HDAC1 and HDAC2, and histone methyltransferases including G9a, Suv39H1, GLP, EZH2 and SMYD2, have been found to be responsible for ‘’writing’ repressive marks on HIV-1 long terminal repeat (LTR) (Boehm et al., 2017; Ding et al., 2013; du Chéné et al., 2007; Friedman et al., 2011; Imai et al., 2010; Marban et al., 2007; Ruelas and Greene, 2013). Suppressive epigenetic marks are further maintained by ‘reader’ proteins HP1γ and L3MBTL1 (Boehm et al., 2017; du Chéné et al., 2007). In addition, multiple miRNAs including miR-28, miR-125b, miR-150, miR-223 and miR-382, and lncRNAs including NEAT1 and NRON, were also found to target viral RNAs and viral proteins to mediate transcriptional or posttranscriptional regulations of HIV-1 latency (Huang et al., 2007; Li et al., 2016; Zhang et al., 2013).

Apart from the above epigenetic mechanisms of HIV-1 latency, another barrier to successfully reactivate latent HIV-1 depends upon transcriptional control (Mbonye and Karn, 2014). In transcription initiation level, HIV-1 latency is contributed by both the insufficiency of transcription factors including NF-κB, Sp1, AP-1, NFAT and TFIIH, and the accumulation of transcription suppressors including LSF, YY1 and CTIP2 (Mbonye and Karn, 2017). For the escaped RNA Polymerase II (RNAP II) which passed through initiation, the absence of HIV-1 Tat and the presence of negative elongation factors NELF and DSIF facilitate promoter-proximal pausing of RNAP II on HIV-1 LTR (Ping and Rana, 2001; Razooky et al., 2015). To further escape from promoter-proximal pausing and turn to transcriptional elongation, RNAP II must be extensively phosphorylated at Ser2 residues by positive transcription elongation factor b (P-TEFb), which consists of cyclin-dependent kinase 9 (CDK9) and Cyclin T1 (Ott et al., 2011). However, the expression of Cyclin T1 is downregulated in latently infected cells (Budhiraja et al., 2013). CDK9 is also inactive because of the dephosphorylation of its T-loop at Thr186 and sequestered in the 7SK small nuclear ribonucleoprotein (snRNP) complex by HEXIM1 or HEXIM2 (Budhiraja et al., 2013; Nguyen et al., 2001; Yang et al., 2001). Another two studies indicate that CDK9 is acetylated at Lys44 by p300 to fully perform its kinase activity (Cho et al., 2010; Fu et al., 2007). Acetylation of Lys48 by GCN5 negatively regulates CDK9 activity (Sabò et al., 2008).

Although many work have unveiled the epigenetic and transcriptional mechanisms of HIV-1 latency, some important questions remain. For instance, there could be a versatile factor responsible for both mechanisms. The mechanism of promoter-proximal pausing has not been fully elucidated. In addition, how the P-TEFb is appropriately sequestered, released and targeted to HIV-1 promoter. More realistically, we have not yet found a powerful LRA which can efficiently reactivate the latent HIV-1 (Spivak and Planelles, 2018). To find more cellular factors as potential targets for LRAs, we designed and screened a custom siRNA library targeting multiple cellular epigenetic and non-epigenetic modification pathways in the nucleus. We found that a SUMOylation E3 ligase tripartite motif-containing protein 28 (TRIM28), also known as transcriptional intermediary factor 1β (TIF1β) and KAP1 (KRAB-associated protein-1), binds to CDK9 and mediates the SUMOylation of CDK9, resulting in the disassociation of CDK9 with Cyclin T1 and the inhibition of CDK9 kinase activity. Consequently, its depletion significantly reactivates HIV-1 transcription and reverses HIV-1 latency.

Results

TRIM28 suppresses HIV-1 expression and contributes to HIV-1 latency

To identify cellular targets which may contribute to HIV-1 suppression and latency, we started from the design and high-throughput screening of a custom siRNA library which targeted several cellular pathways within the nucleus including chromatin binding, epigenetic modification, chromatin remodeling, ubiquitination, SUMOylation, and chromosome organization (Supplementary file 1). We knocked down each gene in a TZM-bl cell line which harbors an integrated copy of luciferase under the control of HIV-1 promoter (Platt et al., 1998). We found that many proteins restricted the activity of HIV-1 promoter based on the expression level of luciferase upon knockdown each target (Figure 1A). The top hit proteins included HP1α, GLP, SUZ12 and CYLD, which have been identified to inhibit HIV-1 transcription (Ding et al., 2013; Khan et al., 2018; Manganaro et al., 2014). Intriguingly, we found that knockdown of two less-defined SUMOylation pathway genes TRIM28 and SUMO4 significantly upregulated HIV-1 promoter activity (Figure 1A, Figure 1—figure supplement 1A–B). The overexpression of TRIM28 inhibited the basal level of HIV-1 promoter activity and rescued HIV-1 repression in dose-dependent manner (Figure 1—figure supplement 1C). The upregulation was more significant when combined with HIV-1 Tat and TNFα (Figure 1—figure supplement 1D). We measured the expression of TRIM28 in different cells and found that TRIM28 is ubiquitously overexpressed in multiple cell lines and primary cells (Figure 1—figure supplement 1E). As a complemental experiment to search for latency contributors, we compared gene expression in unstimulated and PHA-stimulated primary CD4+ T cells utilizing RNA-Seq (Figure 1—figure supplement 1F). We found that TRIM28 was highly expressed in unstimulated primary CD4+ T cells and down regulated upon activation by PHA (Figure 1—figure supplement 1G). The expression of TRIM28 was upregulated again when the activated CD4+ T cells returned to resting status (Figure 1—figure supplement 1H, Figure 1—figure supplement 2). To test whether TRIM28 contributes to HIV-1 latency, we knocked down TRIM28 in HIV-1 latency cell line J-Lat 10.6 and found that the depletion of TRIM28 upregulated HIV-1 expression (Figure 1B–C) (Jordan et al., 2003). Besides, HIV-1 reactivation was enhanced much higher when supplemented with histone deacetylase (HDAC) inhibitor suberoylanilide hydroxamic acid (SAHA, vorinostat) or Bromodomain and Extra-Terminal (BET) domain inhibitor JQ-1, both of which have been widely described as LRAs (Spivak and Planelles, 2018). These results were well repeated in other latency model cell lines including J-Lat 6.3, 8.4, 9.2, and 15.4 (Figure 1—figure supplement 3A–E).

Figure 1 with 3 supplements see all
TRIM28 suppresses HIV-1 expression and contributes to HIV-1 latency.

(A) A siRNA library targeting 182 human genes was transfected into TZM-bl cell line, respectively. Three distinct siRNAs targeting each gene were transfected as a mixture. Forty-eight hours post-transfection, cells were harvested and the activity of luciferase from cell lysates was measured. Fold changes were calculated for each gene compared to negative control siRNA (siNC). (B–C) shRNA constructs were packaged into recombinant lentiviruses and infected J-Lat 10.6. The reactivation efficiency was measured by the GFP-positive percentage which was shown in the top right corner. SAHA and JQ-1 were used as positive controls. (D) Eight ChIP-qPCR primers targeting HIV-1 reporter provirus were designed. G5: Cellular DNA and viral 5’LTR junction; A: Nucleosome 0 assembly site; B: Nucleosome-free region; C: Nucleosome one assembly site; V5: Viral 5’LTR and gag leader sequence junction; L: Luciferase region; V3: Viral poly purine tract and 3’LTR junction; G3: Viral 3’LTR and cellular DNA junction. (E) ChIP assay with antibody against TRIM28 was performed in TZM-bl cell line. All the ChIP-qPCR DNA signals were normalized to siNC IgG of G5. (F–J) ChIP assays with antibodies against H3K9me2, H3K9me3, H3K4me3, H3K9Acetyl and H3K27me3 were performed in TZM-bl cell lines. Data represents mean ±SEM in triplicates. p-Values were calculated by Student’s t-test. *p<0.05, **p<0.01, ***p<0.001.

https://doi.org/10.7554/eLife.42426.003

TRIM28 was previously identified to inhibit HIV-1 integration by recruiting HDAC1 to deacetylate HIV-1 integrase (Allouch et al., 2011). However, its roles in the expression of integrated HIV-1 and HIV-1 latency have not been clearly elucidated. To this end, we performed chromatin immunoprecipitation (ChIP) assay of TRIM28 in TZM-bl and J-Lat 10.6 cell lines to examine its possible association with integrated HIV-1 DNA (Supplementary file 2). We found that TRIM28 was significantly enriched on HIV-1 LTR compared to the regions of host-provirus junction and viral coding region (Figure 1D–E and Figure 1—figure supplement 3F). The enrichment of TRIM28 on HIV-1 LTR was not influenced by TNFα signaling (Figure 1—figure supplement 3G). Because TRIM28 was identified as an epigenetic adaptor recruiting HP1, SETDB1 and NuRD complex to maintain suppressive epigenetic environment, we then tested whether the depletion of TRIM28 would influence the epigenetic status of HIV-1 LTR (Iyengar and Farnham, 2011). We observed significant decrease of H3K9me2 and H3K9me3, as well as significant increase of H3K4me3 and H3K9Ac after knocking down TRIM28 (Figure 1F–I, Figure 1—figure supplement 3H–J). The depletion of TRIM28 also induced slight H3K27me3 downregulation (Figure 1J, Figure 1—figure supplement 3K). These results indicate that TRIM28 suppresses HIV-1 expression and contributes to HIV-1 latency by manipulating suppressive epigenetic modifications.

TRIM28 SUMOylates many transcription factors and transferases

Having identified the suppressive epigenetic adaptor role of TRIM28 on HIV-1 latency, we next attempted to search for new mechanism(s) of TRIM28 by function-based mutation. TRIM28 is a muti-functional protein containing seven different domains (Ivanov et al., 2007). The C-terminal bromodomain (BR), which is SUMOylated by the adjacent plant homeodomain (PHD), recruits SETDB1 and NuRD complex in a SUMOylation-dependent manner. The N-terminal tripartite motif RBCC region is composed of a RING finger domain (RING), two B-box domians (BB), and a coiled-coil domain (CC). The RING of TRIM28 functions as an intermolecular SUMO E3 ligase, while PHD is important for the intramolecular SUMO E3 ligase activity (Ivanov et al., 2007; Liang et al., 2011; Neo et al., 2015).

We constructed different TRIM28 mutants by depleting each of the seven domains (Figure 2A). Then we knocked down the endogenous TRIM28 with siRNA targeting 3’UTR of TRIM28 mRNA and supplied with the wild-type TRIM28 construct and the mutants, respectively. Reactivation of HIV-1 expression by the knockdown of endogenous TRIM28 was re-suppressed to the basal level by the wild-type TRIM28 overexpression (Figure 2B). Theoretically, none of the HP1BD, NHD, or BR mutants, especially the mutant of PHD which harbor the intramolecular SUMO E3 ligase activity, was able to significantly rescue the suppression, but the results showed they did. Nevertheless, the mutant without RING or RBCC domains totally aborted the re-suppression, which might be due to the loss of the Krüppel-associated box domain zinc fingers (KRAB-ZNFs) binding ability. We tested a mutant containing only RBCC. Interestingly, it still resumed the suppression. We also tested whether the two E3 ligase domains contributed to the epigenetic suppression of HIV-1 promoter by knocking down endogenous TRIM28, followed by the overexpression of wild type or mutated TRIM28. The results showed that the wild-type TRIM28 was able to rescue the suppressive epigenetic marks H3K9me3 and H3K27me3 and suppress the active epigenetic mark H3K9Acetyl, however, the mutant without RING or PHD domain was only able to rescue partial of the suppressive marks (Figure 2C–E, Figure 2—figure supplement 1). As the RING within RBCC domain plays a key role for the intermolecular SUMO E3 ligase activity of TRIM28, we therefore hypothesize that TRIM28 may utilize the RING domain to SUMOylate cellular protein (s) which is (are) vital for HIV-1 expression (Liang et al., 2011).

Figure 2 with 1 supplement see all
Both RING and PHD domains E3 ligase activities are important for repressive epigenetic modifications.

(A) Schematic of wild-type TRIM28 and nine TRIM28 mutants. (B) Endogenous TRIM28 was knocked down by siRNA targeting 3’UTR in TZM-bl cells and re-expressed with wild type and different TRIM28 mutants. The luciferase activity was measured. Data represents mean ±SEM in triplicates. p-Values were calculated by Student’s t-test. *p<0.05, **p<0.01, ***p<0.001. (C–E) Endogenous TRIM28 in TZM-bl cells was knocked down by siRNA targeting 3’UTR of TRIM28 mRNA. Another three groups whose endogenous TRIM28 was knocked down were overexpressed with wild type TRIM28 construct or TRIM28 mutants without RING or PHD domain, respectively. ChIP assays with antibodies against H3K9me3, H3K9Acetyl and H3K27me3 were performed for each group. Data represents mean ±SEM in triplicates. p-Values were calculated by Student’s t-test. *p<0.05, **p<0.01.

https://doi.org/10.7554/eLife.42426.007

To identify candidate substrates SUMOylated by TRIM28, we conducted a modified global site-specific SUMOylation Mass Spectrometry (SUMO-MS) (Figure 3A). We generated SUMO1-Q92R, SUMO2-Q88R and SUMO4-Q88R mutants mimicking yeast SUMO Smt3 to enable efficient identification of SUMO-acceptor lysines by MS (Supplementary file 3) and co-expressed the SUMO mutants with TRIM28 and SUMO E2 UBC9 followed by the enrichment of SUMO conjugated substrates (Hendriks et al., 2014). To increase the coverage and mapping possibility of targeted proteins, we used SDS-PAGE to separate the enriched proteins and excised the entire gel lane into 16 slices which were subjected to separate in-gel digestions. The digested peptides were analyzed by nanoscale LC-MS/MS. Finally, we identified 1,329 SUMOyalted proteins at significance threshold below 10−7 (Supplementary file 4). Based on the STRING network analysis, the SUMOylated proteins exerted a large complex network at the interaction confidence of 0.7 (Figure 3B). We further performed MCODE analysis on SUMOylated proteins and found that the STRING core network could be clustered into 12 subclusters with interconnectivity scores ranging from 14 to 96 (Figure 3B, Figure 3—figure supplement 1A and Supplementary file 5). Through Gene Ontology (GO) analysis, we found that cellular and metabolic processes were the top two biological processes which the SUMOylated proteins could be involved in (Figure 3—figure supplement 1B and Supplementary file 6). Most SUMOylated targets have the catalytic activity and DNA binding function. Many transferases and transcription factors were also among the SUMOylated candidates. We specifically clustered the transferases and transcription factors by k-means clustering and visualized with STRING analysis. Interestingly, we found that many candidates were pivotal for HIV-1 expression, such as JUN, JUNB, JUND, mTOR, STAT3, Cyclin T1 (CCNT1) and CDK9 (Figure 3C). Especially, CDK9 and CCNT1 were also found in MCODE Cluster 8 (Figure 3—figure supplement 1A). Recently, it has been identified that the SUMOylation of transcription factor STAT5 was inactivated by benzotriazoles, resulting in the reactivation of latent HIV-1 (Bosque et al., 2017). SUMOylation may participate in transcription more generally. We further narrowed down the significance threshold below 10−8 to find the more extensively SUMOylated targets. CDK9 was still among the top protein candidates (Supplementary file 7). Then, we co-overexpressed SUMO system proteins (SUMO1, SUMO2, SUMO4, UBC9 and TRIM28) with 10 transcription factor candidates, respectively. Several transcription factors were SUMOylated, such as NFKB1A, RelA, CCNT1, CDK9, SKIP, MEN1 and JUN, which verified the reliability of our global site-specific SUMO-MS (Figure 3D). Nevertheless, the SUMOylation signals were much more significant for CDK9, which merited being further studied.

Figure 3 with 1 supplement see all
TRIM28 SUMOylates many transcription factors and transferases.

(A) Schematic of global site-specific SUMO-MS. His-tagged SUMO mutants were co-overexpressed with UBC9 and TRIM28. The SUMOylated proteins were enriched by His-tag beads and separated by SDS-PAGE. Gel fragments were excised and subjected to separate in-gel digestions. The digested peptides were desalted and analyzed by nanoscale LC-MS/MS. (B) SUMOylated proteins were analyzed with STRING. The network were further analyzed by MCODE. Twelve highly interconnected functional subclusters were extracted and shown in different colors. (C) Transferases and transcription factors were clustered by k-means clustering and visualized with STRING analysis. (D) Ten HA-tagged various transcriptional factors were overexpressed with Flag-tagged SUMO proteins, UBC9 and TRIM28. The targeted proteins were immunoprecipitated (IP) by anti-HA-tag beads followed by immunoblotting (IB) with anti-HA and –Flag antibodies. Asterisk (*) indicated the SUMOylated bands.

https://doi.org/10.7554/eLife.42426.009

CDK9 is SUMOylated by TRIM28

To further verify that CDK9 is SUMOylated by TRIM28, we conducted several in vivo and in vitro SUMOyaltion assays. In vertebrates, there are four well-studied SUMO paralogs, SUMO1, SUMO2, SUMO3, and SUMO4. Because SUMO2 and SUMO3 share highly sequence-homolog and have similar functions, they are often referred to as SUMO2/3 (Cubeñas-Potts and Matunis, 2013). It is worthy to note that the depletion of SUMO4 was able to upregulate the HIV-1 promoter activity more significantly than the depletion of the other SUMO paralogs in our siRNA library screening (Figure 1A). The upregulation was more significant when combined with HIV-1 Tat, the phenomenon of which was similar as we observed for TRIM28 (Figure 4A–B, Figure 4—figure supplement 1A). The knockdown or knockout of SUMO4 was able to reactivate latent pseudotyped HIV-1 in J-Lat 10.6 as well (Figure 4C). SUMO4 is also ubiquitously overexpressed in multiple cell lines and primary CD4+ T cells (Figure 4—figure supplement 1B). After PHA stimulation in primary CD4+ T cells, the expression of SUMO4 was downregulated (Figure 1—figure supplement 1G, Figure 4—figure supplement 1C). The expression SUMO4 returned to basal level when activated primary CD4+ T cells re-entered to resting status (Figure 4—figure supplement 1C). As the SUMOylation of TRIM28 and associated epigenetic modifiers participates in the regulation of epigenetic patterns, we next testified whether SUMO4 could influence the function of TRIM28 and the epigenetic status of HIV-1 promoter (Iyengar and Farnham, 2011). We found that more than half of TRIM28 was lost from HIV-1 LTR upon SUMO4 knockdown, which indicated that the enrichment of TRIM28 on HIV-1 LTR may be partially SUMOylation-dependent apart from the Krüppel-associated box domain zinc fingers (KRAB-ZNFs)–dependent binding (Figure 4D, Figure 4—figure supplement 1D–H). We also found that H3K9me, H3K9me2 and H3K9me3 were significantly decreased on HIV-1 LTR in the absence of SUMO4, as well as the H3K9 methylation ‘writer’ SETDB1 and ‘reader’ HP1α (Figure 4E–G, Figure 4K–L). Moreover, we observed significant upregulation of H3K9acetyl and H3K4me3 and downregulation of HDAC1, which was consistent with previous reports that TRIM28 recruited SETDB1, HP1α and HDAC1 in a SUMOylation-dependent manner (Figure 4H–I, Figure 4M) (Iyengar and Farnham, 2011). Besides, we found that the H3K27me3 was also decreased on HIV-1 LTR upon SUMO4 knockdown (Figure 4J). It is possible that some polycomb repressive complex 2 (PRC2) components such as EZH2 and SUZ12, the major ‘writers’ of H3K27me3, may be SUMOylated by SUMO4, resulting in the enhancement of modifier function.

Figure 4 with 1 supplement see all
SUMO4 suppresses HIV-1 expression and contributes to HIV-1 latency.

(A) SUMO4 in TZM-bl cells was knocked down by siRNAs targeting the coding sequence and 3’UTR of SUMO4 mRNA. The luciferase from clarified lysates was quantitated and normalized to siNC. Data represents mean ± SEM in triplicates. p-Values were calculated by Student’s t-test. **p<0.01, ***p<0.001. (B) SUMO4 in TZM-bl cells was knocked down by siRNAs or treated with siNC. HIV-1 Tat construct was co-treated with siRNAs. The luciferase from clarified lysates was quantitated and normalized to the siNC which had no additive. Data represents mean ±SEM in triplicates. p-Values were calculated by Student’s t-test. *p<0.05, **p<0.01. (C) shRNA or sgRNA constructs targeting luciferase (shluc), non-target (sgNT) and SUMO4 (shSUMO4 and sgSUMO4) were packaged into recombinant lentiviruses and infected J-Lat 10.6. The reactivation efficiency was measured by the GFP positive percentage. Data represents mean ±SEM in triplicates. p-Values were calculated by Student’s t-test. *p<0.05. (D–M) SUMO4 in TZM-bl cells was knocked down by siRNA targeting SUMO4 mRNA. ChIP assays with antibodies against TRIM28, H3K9me, H3K9me2, H3K9me3, H3K9Acetyl, H3K4me3, H3K27me3, SETDB1, HP1α and HDAC1 were performed for each group. Data represents mean ±SEM in triplicates. p-Values were calculated by Student’s t-test. *p<0.05, **p<0.01.

https://doi.org/10.7554/eLife.42426.011

As SUMO4 was able to mediate HIV-1 suppression and latency, possibly through the epigenetic control of HIV-1 promoter, we next attempted to identify the underlying mechanism by investigating its role in TRIM28-mediated CDK9 SUMOylation. We co-overexpressed CDK9 with SUMO1, SUMO2 and SUMO4, respectively. We found that CDK9 was mainly SUMOylated with SUMO1 and SUMO4 (Figure 4—figure supplement 1I). The SUMO4-CDK9 amount was much more abundant than the SUMO1-CDK9 amount. Besides, SUMO E3 ligase TRIM28 utilized more SUMO4 compared with SUMO1 and SUMO2 (Figure 4—figure supplement 1J). After the supplement of TRIM28, the SUMO-CDK9 amount turned to be more abundant. However, the SUMOyaltion did not increase if we only co-overexpressed CDK9 with TRIM28 but without SUMO E2 UBC9, which indicated that TRIM28-mediated SUMOylation was UBC9-dependent (Figure 5A). The SUMO-CDK9 amount was increased dose-dependently when the TRIM28 increased gradually (Figure 5B). We then conducted in vitro SUMOylation assay. Only when SUMO4, E1 SAE1/UBA2, E2 UBC9 and TRIM28 were supplied into the SUMO conjugation reaction buffer together, was SUMO4 conjugated to CDK9 (Figure 5C). After knocking down TRIM28 in HeLa cells, the SUMOylated CDK9 decreased (Figure 5D). In our previous siRNA screening, we noticed that the absence of several SUMO-specific isopeptidases (SENPs), which deSUMOylated substrates, prevented the expression of HIV-1, especially SENP3 (Figure 5—figure supplement 1A–B). We then co-overexpressed SENP3 with TRIM28 and found that SENP3 prevented TRIM28-mediated CDK9 SUMOylation (Figure 5E). To investigate whether TRIM28-mediated SUMOylation of CDK9 by SUMO4 exist in primary CD4+ T cells, we firstly confirmed that the conjugation of SUMO4 to cellular proteins frequently occurs (Figure 5—figure supplement 1C). We also immunoblotted the endogenous CDK9 in primary CD4+ T cells and found that a small portion of CDK9 was SUMOylated by SUMO4 (Figure 5—figure supplement 1D). The SUMO4-SUMOylated endogenous CDK9 increased significantly after the overexpression of SUMOylation components including SUMO4, UBC9 and TRIM28 (Figure 5—figure supplement 1E). Taken together, our data indicates that TRIM28 mediates the conjugation of SUMO4 to CDK9, which is reversed by SENP3.

Figure 5 with 1 supplement see all
CDK9 is SUMOylated by TRIM28.

(A) HA-tagged CDK9 was co-overexpressed with Flag-tagged SUMO4, UBC9 or TRIM28. CDK9 was IP with anti-HA-tag beads, followed by IB with anti-HA and –Flag antibodies. TRIM28, UBC9 and GAPDH in total samples were IB with specific antibodies targeting each proteins. (B) HA-tagged CDK9 was co-overexpressed with Flag-tagged SUMO4, Flag-tagged UBC9 and different amount of Flag-tagged TRIM28. Target proteins were IB as in (A). (C) In vitro purified CDK9, SUMO4, SAE1, UBA2, UBC9 and TRIM28 were co-cultured in SUMO conjugation reaction buffer. Proteins including SUMOylated CDK9 were IB with antibodies against each targets. (D) HA-tagged CDK9 was co-overexpressed with Flag-tagged SUMO4, Flag-tagged UBC9 or Flag-tagged TRIM28, and siNC. In the last group, CDK9 was co-overexpressed with SUMO4, UBC9 and siRNA against TRIM28. Target proteins were IB as in (A). (E) HA-tagged CDK9 was co-overexpressed with Flag-tagged SUMO4, Flag-tagged UBC9, Flag-tagged TRIM28 or two gradients of SENP3. Target proteins were IB as in (A).

https://doi.org/10.7554/eLife.42426.013

The RING domain of TRIM28 plays a key role in binding to and SUMOylating CDK9

To identify whether TRIM28 binds to CDK9, we used the super-resolution continuous STochastic Optical Reconstruction Microscopy (cSTORM) to investigate the three dimensional (3D) co-localization in the resolution of 20 nm. We found that TRIM28 existed in many small clusters and large bodies in the nucleus and co-localized with dotted SUMO4 (Figure 6A, first panel). From amplified view and 3D-cSTORM, we found that SUMO4 proteins were enriched by TRIM28 and shaped big spots (Figure 6A, second and third panels; Video 1). Although CDK9 existed in dispersed dots all within the nucleus, we still found that CDK9 co-localized with TRIM28 (Figure 6B, first panel). Similarly to SUMO4, CDK9 proteins were enriched by and surrounded TRIM28 bodies (Figure 6B, second and third panels; Video 2). The lateral resolution of cSTORM imaging can be up to 20 nm and the axial resolution is 50 nm, which is within the range to distinguish protein complexes, even single protein molecules (Lagache et al., 2015). Thus, we transformed the cSTORM-imaged protein molecules and complexes into small or large spots based on their diameter (Figure 6C–D, Figure 6—figure supplement 1A–B, left panel; Video 3, Video 4). The direct interaction between spots and spots was measured in compliance with the criterion of maximal distance of 10 nm (Figure 6C–D, Figure 6—figure supplement 1A–B, middle panel). The indirect interaction between complexes and spots was measured in compliance with the criterion of maximal distance of 100 nm (Figure 6C–D, Figure 6—figure supplement 1A–B, right panel). Finally, we found that nearly 80% of TRIM28 spots or complexes were co-localized with 94% of SUMO4 spots (Figure 6E). Similarly, 88% of TRIM28 spots or complexes were co-localized with 76% of CDK9 spots (Figure 6E).

Figure 6 with 1 supplement see all
TRIM28 co-localizes with SUMO4 and CDK9.

(A) cSTORM image of endogenous TRIM28 and SUMO4 in HEK293T cells. The first row: the original whole nucleus; the second row: one of the amplified region of the nucleus; the third row: the 3D-cSTORM image of the amplified region. Merged views of TRIM28 and SUMO4 were shown on the left column. Endogenous TRIM28 was shown in the middle column and colored green. Endogenous SUMO4 was shown in the right column and colored red. Of note, DAPI and Hoechst were not allowed to dye DNA according to cSTORM protocol. (B) cSTORM image of endogenous TRIM28 and CDK9 in HEK293T cells. Each row was shown as in (A). First column: merged view of TRIM28 and CDK9, yellow indicating co-localization; second column: endogenous TRIM28 which was colored green; third column: endogenous CDK9 which was colored red. (C–D) cSTORM-imaged protein molecules and complexes were transformed into small or large spots based on their diameter. The left panel of each figure showed the original transformation. The middle panel showed spots-spots co-localization in compliance with the criterion of maximal distance of 10 nm. The right panel showed complexes-spots co-localization in compliance with the criterion of maximal distance of 100 nm. Green spots indicated TRIM28 molecules. Red spots indicated SUMO4 or CDK9 molecules. (E) Quantitation of co-localization of TRIM28 with SUMO4 or CDK9. Both of total proteins-proteins, spots-spots and complexes-spots co-localizations were measured.

https://doi.org/10.7554/eLife.42426.015
Video 1
3D-cSTORM movie of the 3D co-localization of TRIM28 with SUMO4.

Green spots indicate TRIM28. Red spots indicate SUMO4.

https://doi.org/10.7554/eLife.42426.017
Video 2
3D-cSTORM movie of the 3D co-localization of TRIM28 with CDK9.

Green spots indicate TRIM28. Red spots indicate CDK9.

https://doi.org/10.7554/eLife.42426.018
Video 3
Transformed 3D-cSTORM movie of the 3D co-localization of TRIM28 with SUMO4.

Green spots indicate TRIM28. Red spots indicate SUMO4.

https://doi.org/10.7554/eLife.42426.019
Video 4
Transformed 3D-cSTORM movie of the 3D co-localization of TRIM28 with CDK9.

Green spots indicate TRIM28. Red spots indicate CDK9.

https://doi.org/10.7554/eLife.42426.020

Through co-immunoprecipitation (Co-IP) assay, we found that CDK9 bound to TRIM28, even in the presence of RNase (Figure 7—figure supplement 1A). To identify which region of TRIM28 bound to CDK9, we examined various TRIM28 deletion mutants to enrich CDK9. The depletion of RING aborted the binding of CDK9 as well as the SUMOylation of CDK9 (Figure 7A–B). Further, we co-transfected GFP-TRIM28 and several GFP-TRIM28 mutants with RFP-CDK9 in HEK293T cells and utilized the super-resolution Structured Illumination Microscopy (SIM) to investigate the co-localization. Exogenously expressed TRIM28 also co-localized with CDK9 with Pearson’s coefficient of 0.7336 and thresholded Mander’s coefficient of 0.5846, which indicated a highly co-localization. However, the mutant of RING domain deletion was not capable (Figure 7C–D). We also inspected the SUMOylation status of each TRIM28 mutants and found that all the mutants was SUMOylated, which coincided with previous reports that both the RING and PHD had the E3 ligase activity and enriched UBC9 (Figure 7—figure supplement 1B) (Ivanov et al., 2007; Liang et al., 2011). Collectively, our results indicate that TRIM28 binds to CDK9 and SUMOylates CDK9 through its RING domain.

Figure 7 with 1 supplement see all
The RING domain of TRIM28 plays a key role in binding to and SUMOylating CDK9.

(A) HA-tagged CDK9 was co-overexpressed with Flag-tagged full length TRIM28 or domain-truncated TRIM28 mutants. Flag-tagged proteins were IP, followed by IB with antibodies against HA-tag, Flag-tag and GAPDH. (B) HA-tagged CDK9 was co-overexpressed with Flag-tagged SUMO4, Flag-tagged UBC9, Flag-tagged full length TRIM28 or Flag-tagged domain-truncated TRIM28 mutants. CDK9 was IP with anti-HA-tag beads, followed by IB with antibodies against HA-tag, Flag-tag and GAPDH. (C) GFP-tagged TRIM28 or TRIM28-dRING mutant was co-overexpressed with RFP-tagged CDK9 in HEK293T cells. The samples were fixed and dyed according to the immunofluorescence procedure, then visualized in Nikon A1 N-SIM. DAPI was used to dye DNA which was colored into blue. (D) Quantitation of co-localization of TRIM28 or TRIM28-dRING with CDK9. The percentage of co-localization was indicated by percentage of target protein voxels above threshold co-localized voxels. Both Pearson’s coefficient and thresholded Mander’s coefficient were used to evaluate co-localization. For Pearson’s coefficient, a value of 1 represents perfect co-localization, 0 no co-localization, and −1 perfect inverse co-localization. For thresholded Mander’s coefficient, a value of 1 represents perfect co-localization and 0 no co-localization.

https://doi.org/10.7554/eLife.42426.021

CDK9 function is inhibited when SUMOylated by TRIM28

After confirming CDK9 is indeed SUMOylated by TRIM28, and also because the RBCC domain contributes to HIV-1 suppression, we next tried to examine whether the function of CDK9 is influenced by TRIM28-mediated SUMOylation. We firstly utilized ATAC-Seq to probe the chromatin accessibility of HIV-1 promoter upon TRIM28 elimination. We found that most of the increased accessible regions across the genome lied on promoters and distal intergenic regions upon the depletion of TRIM28 in J-Lat 10.6 or TZM-bl cell lines (Figure 8—figure supplement 1A–B). Through GO analysis and Clusters of Orthologous Groups of proteins (COGs) analysis, we found that the chromatin accessibility variation happened in genes related to various biological processes and cellular components upon TRIM28 depletion (Figure 8—figure supplement 1C–D). Most of the influenced general functional genes had the DNA or protein-binding abilities and catalytic activities (Figure 8—figure supplement 1C–D, Figure 8—figure supplement 2A–B). To inspect whether the chromatin accessibility of HIV-1 genome was influenced upon TRIM28 depletion, we separately aligned the sequencing reads to HIV-1 reference genome. We found that the accessible region indicated by transposable tag density increased on HIV-1 LTR when TRIM28 was knocked out from J-lat 10.6 cell lines, as well as when TRIM28 was knocked down in TZM-bl cell lines, which indicated significantly enhanced promoter activity (Figure 8A–B). The promoters of genes within which the integrated pseudotyped HIV-1 or HIV-1 reporter provirus located and housekeeping gene GAPDH were not influenced (Figure 8—figure supplement 2C–F). Alternatively, we also observed significant enrichment of CDK9 and Ser2 super-phosphorylated RNAP II on HIV-1 LTR upon the knockdown of either TRIM28 or SUMO4, which was in agreement with the results that the depletion of TRIM28 or SUMO4 reactivated HIV-1 expression (Figure 8C–D).

Figure 8 with 3 supplements see all
CDK9 function is reduced when SUMOylated by TRIM28.

(A–B) TRIM28-defective (sgTRIM28) J-Lat 10.6 cell line was generated by CRISPR-CAS9 technique. ATAC-Seq was conducted with sgNT and sgTRIM28 J-Lat 10.6 cell lines, as well as siNC and siTRIM28 TZM-bl cell lines. The tag reads of the HIV-1 pseudotyped virus/minigenome 5’LTR integration sites were counted and normalized to the total mapped reads, and represented as relative tag density. The highest tag density was set as 100. Figures showed 2 kb range centered the 5’LTR integration sites. (C–D) ChIP assays with antibodies against CDK9 and Ser2 Pho-Pol II were performed in TZM-bl cell lines which were treated with siNC, siSUMO4 and siTRIM28, respectively. (E) Cyclin T1 or GFP was co-overexpressed with CDK9 in the absence or presence of SUMO4, UBC9 and TRIM28. Cyclin T1 and GFP were IP followed by IB. (F) Fold change of kinase activity when CDK9 was SUMOylated. Data represents mean ±SEM in triplicates. p-Values were calculated by Student’s t-test. *p<0.05, **p<0.01.

https://doi.org/10.7554/eLife.42426.023

Interestingly, through Co-IP assay, we found that Cyclin T1 only bound to wild-type CDK9 to form P-TEFb complex, not the SUMOylated CDK9 (Figure 8E). To investigate whether TRIM28-mediated SUMOylation of CDK9 affects the kinase activity of CDK9, we conducted in vitro CDK9 SUMOylation assay followed by CDK9 kinase assay (Figure 8—figure supplement 3A). We found that the kinase activity of CDK9 significantly decreased when SUMOylated by TRIM28. However, the kinase activity of CDK9 was not influenced without the addition of TRIM28, although the other SUMOylation components have been added (Figure 8F). Collectively, TRIM28-mediated SUMOylation impairs both the binding ability of CDK9 to Cyclin T1 and the kinase activity of CDK9 to RNAP II, resulting in the dysfunction of transcription elongation.

The Lys44, Lys56 and Lys68 residues of CDK9 are SUMOylated with SUMO4

To elucidate the mechanisms that SUMOylation weakens the interaction between CDK9 and Cyclin T and the CDK9 kinase activity, we next attempted to identify the CDK9 SUMOylation sites which should occur on lysine residues. In order to narrow down the search scope, we equally grouped the sequence of CDK9 into three parts. Each part was given a mutant version that all the lysines were mutated to arginines. Then, we combined these six sequences and obtained eight constructs including the wild type CDK9 (Figure 9—figure supplement 1A and Supplementary file 3). The construct named CDK9-K0R, which contained the mutation that all the lysines were changed to arginines, totally aborted the capability of CDK9 to be SUMOylated (Figure 9—figure supplement 1B). However, the other CDK9 mutants still were able to be SUMOylated by TRIM28, which indicated that multiple SUMOylation sites might exist across the whole CDK9 sequence. To locate all the suspicious SUMOylation sites, we adopted reversing mutation strategy based on CDK9-K0R construct. Each of the 29 arginines of CDK9-K0R was mutated back to lysine separately (Supplementary file 3). Finally, we found that several lysines on CDK9 were significantly SUMOylated (Figure 9A). Among them, multiple SUMOylation sites were adjacent to CDK9 C-terminal autophosphorylation sites which have been reported to be required for high-affinity binding of Tat–P-TEFb to TAR RNA (Baumli et al., 2008; Garber et al., 2000). SUMOylation may decrease the binding ability through preventing the neighboring phosphorylation.

Figure 9 with 1 supplement see all
The Lys44, Lys56 and Lys68 residues of CDK9 are SUMOylated with SUMO4.

(A) Different HA-tagged CDK9 reversing mutation constructs or wild type CDK9 were co-overexpressed with SUMO4, UBC9 and TRIM28, respectively. CDK9 and CDK9 mutants were IP with anti-HA-tag beads followed by IB. S4: SUMO4. (B) HA-tagged wild type CDK9 and 12 identified SUMOylation site reversing mutation constructs were co-overexpressed with Flag-tagged SUMO4 and Flag-tagged UBC9. The endogenous TRIM28 was knocked down with siRNAs. CDK9 and CDK9 mutants were IP with anti-HA-tag beads followed by IB. Asterisks represented the constructs whose SUMOylation bands disappeared upon TRIM28 knockdown. (C) Three angles of co-crystal structure of Cyclin T1 and CDK9 (PDB ID: 4EC8). Three SUMOylation sites Lys44, Lys56 and Lys68 were shown in ball-and-stick models. The two upper panels showed the ribbon models, while two lower panels showed the surface models. The inner six framed figures which numbered from I to VI represented the amplification views of Lys44, Lys56 and Lys68 sites.

https://doi.org/10.7554/eLife.42426.027

It was notable that, although the remove of endogenous TRIM28 significant downregulated SUMO-CDK9, slightly residual SUMO-CDK9 still occurred, implying that other CDK9 SUMOylation E3 ligases may exists and some of the SUMOylation sites are not the TRIM28 targets (Figure 5D). To further identify which sites are indeed SUMOylated by TRIM28 only, we knocked down the endogenous TRIM28 and tested the SUMOylation potential of the candidate sites identified above. We found that the SUMOylation signals of Lys44, Lys56 and Lys68 totally disappeared in the absence of endogenous TRIM28, further supporting that these sites are specifically SUMOylated by TRIM28 (Figure 9B). The target-specific SUMO-MS for directly analyzing the enriched SUMO-CDK9 also confirmed this result (Figure 9—figure supplement 1C–E). As the acetylation of Lys44 is required for its kinase activity, it is not surprising that the kinase activity of CDK9 was weakened when CDK9 was SUMOylated (Cho et al., 2010; Fu et al., 2007). Interestingly, other two SUMOylated sites Lys56 and Lys68 are within the interaction region of CDK9 and Cyclin T1 based on the co-crystal structure (PDB ID: 4EC8) (Baumli et al., 2012) (Figure 9C). Because SUMO protein is a polypeptide macromolecule, its presence can form steric hindrance which prevents the formation of P-TEFb complex.

TRIM28 depletion reactivates latent HIV-1 in cells from HIV-1-infected individuals

To verify whether TRIM28 could be a safe target for developing new LRAs, we firstly evaluated the possible toxicities associated with depleting TRIM28 in Hela cells, Jurkat cells as well as resting CD4+ T cells isolated from aviremic participants. We conducted several experiments which included cytotoxicity assay, cell viability assay, cell number counting and cell proliferation assay. The results showed that the depletion of TRIM28 was non-toxic to cell viability and proliferation (Figure 10—figure supplement 1, Figure 10—figure supplement 2). Afterwards, we tried to determine whether the knockdown of TRIM28 reactivated latent HIV-1 in resting CD4+ T cells from HIV-1-infected individuals who received suppressive cART for at least 6 months. Stimulation with αCD3/αCD28 significantly induced the expression of HIV-1 based on the quantitation of intracellular HIV-1 RNAs (Figure 10A and Figure 10—figure supplement 3A–B). The depletion of TRIM28 reactivated similar amount of HIV-1 RNA as suberanilohydroxamic acid (SAHA). After we combined the knockdown of TRIM28 with SAHA, the reactivation was more significant (Figure 10A). To provide evidence that SUMO4-mediated modification of CDK9 by TRIM28 is one of the mechanisms used by TRIM28 to contribute to HIV-1 latency in cells isolated from aviremic participants, we also tested whether the depletion of SUMO4 could reactivate latent HIV-1 in resting CD4+ T cells isolated from HIV-1-infected individuals. The results showed that the depletion of SUMO4 reactivated substantial productions of HIV-1 RNAs which were even slightly higher than those activated by SAHA. The combination use of SUMO4 knockdown and SAHA addition could reactivate more HIV-1 RNAs than those reactivated by them separately (Figure 10—figure supplement 4). We next examined whether the knockdown of TRIM28 reactivated more genetically-diversified HIV-1, as we described previously (Figure 10—figure supplement 3A) (Geng et al., 2016b). Although TRIM28 depletion alone reactivated similar amount of genetically diversified HIV-1 with SAHA, the combination of TRIM28 knockdown and SAHA reactivated much more genetically-diversified HIV-1 (Figure 10B). To determine whether the reactivated HIV-1 was replication-competent, we co-cultured the PHA-stimulated, SAHA-induced, or TRIM28-deficient resting CD4 +T cells from HIV-1-infected individuals, with PHA-activated CD4 +T cells from heathy donors (Figure 10—figure supplement 3A). The accumulating production of p24 antigen indicates the reactivated HIV-1 viral particles were replication-competent. The knockdown of TRIM28 reactivated replication-competent viruses in all the three samples (Figure 10C and Figure 10—figure supplement 3C). Similarly, the combination of SAHA with TRIM28 knockdown reactivated more replication-competent viral particles. These results indicate that TRIM28 contributes to HIV-1 latency in HIV-1-infected individuals. Targeting TRIM28 is well-tolerated for HIV-1-infected CD4+ T cells.

Figure 10 with 4 supplements see all
TRIM28 depletion reactivates latent HIV-1 in cells from HIV-1-infected individuals.

(A) shRNAs targeting luciferase and TRIM28 were packaged into lentiviruses and infected CD4+ T cells from HIV-1-infected individuals. Unstimulated CD4 +T cells were used as negative control (NC). Stimulation with αCD3/αCD28/IL-2 was used as positive control. Intracellular HIV-1 RNA was isolated and quantitated by qPCR. Experiments were conducted in three HIV-1-infected individuals. (B) The experiment setting was as in (A). Envelope V1 to V3 region from intracellular HIV-1 RNAs was reverse-transcribed and PCR-amplified. The PCR products were TA-ligated in pMD-18 T vector. At least 60 single clones were picked from each group and sequenced. The sequences from each group were aligned and the genetic diversity index was calculated and analyzed by Mann-Whitney U-test. The upper panel showed the statistical analysis results. The lower panel indicated the bootstrap consensus trees which were generated based on HIV-1 sequences. *p<0.05, **p<0.01, ***p<0.001. (C) Resting CD4+ T cells from HIV-1-infected individuals were isolated and nucleofected with siRNAs targeting negative control or TRIM28. Seventy-two hours later, PHA-stimulated uninfected CD4+ T cells were added into each group and co-cultured for another 27 days. The supernatants were collected and half-changed every 3 days. P24 antigens in supernatants were measured with ELISA and plotted in log10 scale. Dashed lines indicated the limit of detection (L.O.D.) of 50 pg/ml. Triplicates were represented by mean ±SEM. (D) Schematic of TRIM28-mediated HIV-1 latency.

https://doi.org/10.7554/eLife.42426.029

Discussion

TRIM28 has been found as an epigenetic adaptor which recruits multiple suppressive epigenetic modifiers to the LTRs of endogenous retroviruses (Rowe et al., 2010; Wolf and Goff, 2007). It is also identified to stabilize promoter-proximal pausing of RNAP II with some unsolved functions (Bunch et al., 2014). Furthermore, TRIM28 is also a SUMO E3 ligase which can mediate intramolecular SUMOylation of its bromodomain and intermolecular SUMOylation of IFN regulatory factor 7 (IRF7), resulting in the recruitment of epigenetic modifiers and the inhibition of IRF7 function, respectively (Ivanov et al., 2007; Liang et al., 2011) In this report, we identified that TRIM28 functions not only as a well-defined epigenetic adaptor but also as a SUMO E3 ligase to SUMOylate P-TEFb complex to significantly repress HIV-1 expression and contributes to HIV-1 latency. Based on our data, we propose a model of TRIM28-mediated HIV-1 latency (Figure 10D). In active status, P-TEFb complex is recruited by HIV-1 Tat to the partly transcribed HIV-1 RNA trans-activation response element (TAR). P-TEFb catalytic subunit CDK9 super-phosphorylates the Ser2 residues of RNAP II, facilitating the processivity of RNAP II on the transcribing HIV-1 RNA. In latent status, TRIM28 is recruited to HIV-1 LTR and SUMOylates CDK9 in Lys44, Lys56 and Lys68, resulting in the inhibition of CDK9 kinase activity and its disconnecting with Cyclin T1. Without the super-phosphorylation on Ser2, RNAP II promoter-proximal paused at LTR. Therefore, the latent status is maintained by both TRIM28-mediated CDK9 dysfunction and TRIM28-mediated suppressive epigenetic modification on nucleosome nuc-1 which lies precisely downstream of HIV-1 promoter (Verdin et al., 1993).

Nevertheless, previous works reported that CDK9 and P-TEFb regulatory subunit Cyclin T1 were recruited to HIV-1 LTR by TRIM28 through 7SK snRNP bridging, although some debates existed (D'Orso and Frankel, 2010; D'Orso et al., 2012; Mbonye and Karn, 2014; Mbonye and Karn, 2017; McNamara et al., 2016; Ott et al., 2011). In contrast, we found that TRIM28 was still able to enrich CDK9 in the presence of RNase (Figure 7—figure supplement 1A). Instead, our results showed that TRIM28 bound to CDK9 through its RING domain. Besides, our findings regarding the effect of TRIM28 upon HIV-1 transcription are inconsistent with the observations from D'Orso’s group. They found that TRIM28 facilitates RNAP II elongation by manipulating ‘on-site’ P-TEFb activation, resulting in quick response to stimulation and facilitating HIV-1 transcribing (McNamara et al., 2016). However, in a simple HIV-1 expression model, several HIV-1 latency models, and resting CD4+ T cells isolated from HIV-1-infected individuals, we all found that the depletion of TRIM28 results in HIV-1 transcriptional activation and TRIM28 functions as a latency contributor rather than a stimulator in our model. We also found that the enrichment of TRIM28 on HIV-1 promoter was unchanged upon TNFα stimulation, which indicated that TRIM28 might not be controlled by TNFα signaling (Figure 1—figure supplement 3G). Whether these controversies are caused by cell lines, cellular conditions, or various HIV-1 integration sites as hypothesized by them, still needs to be further confirmed.

SUMO enigma of TRIM28

Post-translational modifications of CDK9 have been studied extensively, most of which focus on phosphorylation and acetylation (Cho et al., 2010). Interestingly, many CDK9 SUMOylation sites which we identified here are highly related to phosphorylation and acetylation. The acetylation of Lys44 is vital for CDK9 phosphorylation activity on RNAP II. The SUMOylation of Lys44 masks the kinase activity. The acetylated Lys44 can also be deacetylated by NuRD complex which recruited by TRIM28.

Although multiple sites on CDK9 can be SUMOylated by TRIM28, the percentage of SUMOylated CDK9 is only a small proportion (less than 5%). This phenomenon has been observed for most of the identified SUMOylation targets (Gareau and Lima, 2010; Impens et al., 2014). How the small portion triggers extensive effect on target substrate remains a mystery. Two models have been proposed to explain the small fraction of SUMOylation mediated transcriptional suppression, respectively (Hay, 2005; Johnson, 2004). Both models suggest transcriptional suppression is initiated by SUMOylation. However, the maintenance of suppression is SUMOylation-independent. In our co-localization experiment, we found that CDK9 is extensively recruited to the sub-compartment shaped by TRIM28, although the SUMOylated CDK9 is only a small proportion based on the western blotting data. We propose that SUMOylation is a transient signal for CDK9 to enter to silent status or silent complex. The SUMOylated CDK9 may recruit other suppressive modifiers to stabilize the suppressive complex. After the remove of SUMO peptide by ubiquitous SENPs, CDK9 might be still sequestered in the suppressive complex. In recent years, TRIM28 was identified to form a large repressive complex with other epigenetic silencing complex such as the human silencing hub (HUSH) complex which also recruits SETDB1 to HIV-1 LTR to maintain H3K9me3 (Robbez-Masson et al., 2018; Tchasovnikarova et al., 2015). In rapid growing cells, 90% of P-TEFb is sequestered in suppressive complex 7SK snRNP (Zhou et al., 2012). Whether TRIM28 is part of 7SK snRNP and whether TRIM28 complex shares overlap with 7SK snRNP or other CDK9 suppressive complexes in primary CD4+ T cells need to be further elucidated.

TRIM28-mediated transcriptional-pausing

TRIM28 has previously been found to stabilize the RNAP II promoter-proximal pausing (Bunch et al., 2014). However, the detailed mechanism is largely unknown. Our findings here could potentially explain this phenomenon. The largest barrier for RNAP II to escape from transcriptional-pausing to effective elongation is the recruitment of P-TEFb to super-phosphorylate RNAP II. TRIM28 is bound to upstream of transcription start sites (TSSs) and SUMOylates the invaded CDK9, resulting in the disconnection of CDK9 with Cyclin T1 and inhibition of CDK9 kinase activity. This hypothesis is also consistent with our finding that the depletion of TRIM28 or SUMO4 induces more significant HIV-1 expression when combining the use of HIV-1 Tat. Without the constraint of TRIM28-mediated CDK9 SUMO4-SUMOylation, HIV-1 Tat utilizes more functional CDK9 to facilitate RNAP II on transcribing HIV-1 RNA. Another mechanisms which TRIM28 may manipulate is TRIM28-mediated suppressive epigenetic modifications on nucleosomes downstream of RNAP II pausing sites, which further stabilizes transcriptional-pausing. One report showed that SENP3 deSUMOylates RbBP5, one of the subunits of MLL1/MLL2 complexes, resulting in the complexes stabilization, H3K4me3 accumulation and RNAP II recruitment (Nayak et al., 2014). We found that SENP3 prevents TRIM28-mediated CDK9 SUMOylation, which facilitates the transcriptional-pausing release of recruited RNAP II. More work needs to further identify the upstream signaling pathway which determines when to release TRIM28-mediated transcriptional-pausing of RNAP II on HIV-1 LTR.

Future development of LRAs targeting both epigenetics and transcription

Until now, nearly all the shock agents have failed to decrease the latent HIV-1 reservoir based on several clinical trials (Spivak and Planelles, 2018). The only effective LRAs across multiple latency model cell lines and ex vivo patient cells are protein kinase C (PKC) agonists (Bullen et al., 2014). However, PKC agonists induce some degree of T cell activation which is toxic to global T cells. Several lines of evidence have shown that both epigenetic regulation and transcriptional control are two barriers which we need to overcome when we develop novel LRAs (Mbonye and Karn, 2017). Interestingly, we found that TRIM28 bridges both suppressive epigenetic modifications and RNAP II transcriptional-pausing to contribute to HIV-1 latency. Besides, LRAs which target the SUMOylation of transcription factor result in the reactivation of latent HIV-1 (Bosque et al., 2017). TRIM28-mediated RNAP II transcriptional-pausing on HIV-1 promoter is also SUMOylation-dependent as we have elucidated extensively above. Developing next-generation LRAs targeting TRIM28 may release both epigenetic and transcriptional restrictions, which also provides a new direction to search dual-function candidates.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or
reference
IdentifiersAdditional
information
Strain, strain
background
(Escherichia coli)
E.coli DH5α: F-,
φ 80dlacZ ΔM15,
Δ(lacZYA -argF )U169, deoR ,
recA1 , endA1 , hsdR17
(rK-, mK+), phoA, supE44 ,
λ-, thi −1, gyrA96 , relA1
TakaraCat#9057
Strain, strain
background
(Escherichia coli)
E. coli HB101: F-,
hsdS20(rB-, mB-),
recA13, ara-14, proA2, lacY1,
galK2, rpsL20 (str), xyl-5,
mtl-1,supE44, leuB6, thi-1.
TakaraCat#9051
Strain, strain
background
(Escherichia coli)
E.coli BL21: F-, ompT,
hsdSB (rB-mB-),
gal, dcm
TakaraCat#9126
Strain, strain
background
(Escherichia coli)
E.coli Stbl3: F-, mcrB,
mrr, hsdS20 (rB-, mB-),
recA13, supE44, ara-14,
galK2, lacY1, proA2, rpsL20
(StrR), xyl-5, λ- leu, mtl-1
ThermoFisherCat#C7381201
Cell line
(Homo sapiens)
HEK293TATCCCRL-3216;
RRID: CVCL_0063
female
Cell line
(Homo sapiens)
HeLaATCCCCL-2;
RRID: CVCL_0030
female
Cell line
(Homo sapiens)
TZM-blNIH AIDS
Reagent Program
Cat#8129female
Cell line
(Homo sapiens)
J-Lat 6.3PMID: 12682019NIH AIDS Reagent
Program Cat#9846
Dr. Eric Verdin
(The Buck Institute
for Research on
Aging, Novato,
CA, USA)
Cell line
(Homo sapiens)
J-Lat 8.4PMID: 12682019NIH AIDS Reagent
Program Cat#9847
Dr. Eric Verdin
(The Buck Institute for
Research on Aging,
Novato, CA, USA)
Cell line
(Homo sapiens)
J-Lat 9.2PMID: 12682019NIH AIDS Reagent
Program Cat#9848
Dr. Eric Verdin
(The Buck Institute for
Research on Aging,
Novato, CA, USA)
Cell line
(Homo sapiens)
J-Lat 10.6PMID: 12682019NIH AIDS Reagent
Program Cat#9849
Dr. Eric Verdin
(The Buck Institute
for Research on
Aging, Novato, CA, USA)
Cell line
(Homo sapiens)
J-Lat 15.4PMID: 12682019NIH AIDS Reagent
Program Cat#9850
Dr. Eric Verdin
(The Buck Institute
for Research on Aging,
Novato, CA, USA)
Biological sample
(Homo sapiens)
Blood samples
from healthy individuals
Guangzhou Blood
Center, Guangzhou
http://www.gzbc.org/
Biological sample
(Homo sapiens)
Blood samples
from HIV-1-
infected individuals
Department of
Infectious Diseases, Guangzhou
8th People’s
Hospital, Guangzhou
http://gz8h.com.cn/
AntibodyMouse Monoclonal
anti-TRIM28 Antibody
ProteintechCat#66630–1-Ig;
RRID: AB_2732886;
Lot#10006062
(1:1000)
AntibodyRabbit Polyclonal
anti-TRIM28 Antibody
ProteintechCat#15202–1-AP;
RRID: AB_2209890;
Lot#00051172
(1:1000)
AntibodyRabbit Polyclonal
Anti-Histone H3
(tri methyl K4) Antibody
AbcamCat#ab8580;
RRID: AB_306649;
Lot#GR273043-3
Use 2 µg for 25 µg
of chromatin
AntibodyRabbit Polyclonal
Anti-Histone H3 (acetyl
K9) Antibody
AbcamCat#ab4441;
RRID: AB_2118292;
Lot#GR270585-1
Use 2 µg for 25 µg
of chromatin
AntibodyMouse Monoclonal
Anti-Histone H3 (tri
methyl K27) Antibody
AbcamCat#ab6002;
Lot#GR275911-3
Use 5 µg for 25 µg
of chromatin
AntibodyNormal Rabbit
Anti-IgG Antibody
CSTCat#2729;
RRID: AB_1031062
Use 1 µg for 25 µg
of chromatin
AntibodyRabbit Polyclonal
Anti-UBE2I Antibody
AbclonalCat#A2193;
Lot#45473
(1:1000)
AntibodyRabbit Polyclonal
Anti-UBA2 Antibody
AbclonalCat#A4363(1:1000)
AntibodyRabbit Polyclonal
Anti-SAE1 Antibody
ProteintechCat#10229–1-AP;
RRID: AB_2182917;
Lot#00040591
(1:1000)
AntibodyRabbit Monoclonal
Anti-SUMO4 Antibody
AbcamCat#ab126606;
RRID: AB_11128131;
Lot#GR851138-12
(1:1000)
AntibodyRabbit Monoclonal
Anti-CDK9 (C12F7) Antibody
CSTCat#2316; Lot#6(1:1000)
AntibodyRabbit Polyclonal
Anti-SENP3 Antibody
ProteintechCat#17659–1-AP;
RRID: AB_2301618;
Lot#00025621
(1:1000)
AntibodyRabbit Polyclonal Anti-RNA
polymerase II CTD repeat
YSPTSPS (phosphor-Ser2)
Antibody
AbcamCat#ab5095;
RRID: AB_304749;
Lot#GR278215-1
Use 2 µg for 25 µg
of chromatin
AntibodyMouse Monoclonal
Anti-Histone H3 (di methyl
K9) Antibody
AbcamCat#ab1220;
RRID: AB_449854
Use 4 µg for 25 µg
of chromatin
AntibodyRabbit Polyclonal
Anti-Histone H3 (tri
methyl K9) Antibody
AbcamCat#ab8898;
RRID: AB_306848
Use 4 µg for 25 µg
of chromatin
AntibodyDonkey Anti-Mouse
IgG H and L (Alexa Fluor
647) Antibody
AbcamCat#ab150107;
Lot#GR311164-3
(1:200)
AntibodyDonkey Anti-Rabbit IgG
H and L (Alexa Fluor 647)
Antibody
AbcamCat#ab150075;
Lot#GR3174006-4
(1:200)
AntibodyDonkey Anti-Rabbit IgG
(H + L), Highly Cross-
Adsorbed, CF 568 Dye
Conjugates, Single Label
for STORM
BiotiumCat#20803–500 μl;
Lot#17C0626
(1:200)
AntibodyDonkey Anti-Mouse IgG
(H + L), Highly Cross-
Adsorbed, CF 568 Dye
Conjugates, Single Label
for STORM
BiotiumCat#20802–500 μl;
Lot#17C1004
(1:200)
AntibodyRabbit Anti-DDDDK Tag
Polyclonal Antibody,
Unconjugated
MBLCat#PM020;
RRID: AB_591224;
Lot#026
(1:1000)
AntibodyMouse Monoclonal
Anti-HA-Tag Antibody
MBLCat#M180-3;
RRID: AB_10951811;
Lot#008
(1:10000)
AntibodyMouse Monoclonal
Anti-His-Tag Antibody
ProteintechCat#66005–1-Ig;
RRID: AB_11232599;
Lot#00083246
(1:1000)
AntibodyRabbit Polyclonal
Anti-GAPDH Antibody
ProteintechCat#10494–1-AP;
RRID: AB_2263076;
Lot#00039889
(1:10000)
AntibodyIRDye 680RD Goat
anti-Mouse IgG (H + L),
0.5 mg Antibody
LI-COR BiosciencesCat#926–68070;
RRID: AB_10956588;
Lot#C70613-15
(1:10000)
AntibodyIRDye 800CW Goat
Anti-Rabbit IgG,
Conjugated Antibody
LI-COR BiosciencesCat#926–32211;
RRID: AB_621843;
Lot#C70620-05
(1:10000)
AntibodyPerCP-Cy 5.5
Mouse Anti-Human
CD45RO
BD BiosciencesCat#560607;
RRID: AB_1727500;
Lot#5338941
(1:1000)
AntibodyAPC/Cy7 anti-
human CD45RA
BioLegendCat#304127;
RRID: AB_10708419;
Lot#B164612
(1:1000)
AntibodyAnti-Human
CD69 PE-Cy7
ThermoFisherCat#25-0699-42;
RRID: AB_1548714;
Lot#E10154-1635
(1:1000)
AntibodyAnti-Human CD62L
PE-Cyanine7
ThermoFisherCat#25-0629-42;
RRID: AB_1257142;
Lot#4291471
(1:1000)
AntibodyAnti-Human CD4 FITCThermoFisherCat#11-0048-42;
RRID: AB_1633390;
Lot#E10526-1631
(1:1000)
AntibodyPE-Cy5 Conjugated
Amti-human CD25 (IL-2R)
ThermoFisherCat#15-0259-42;
RRID: AB_1944361;
Lot#E11289-102
(1:1000)
Recombinant
DNA reagent
VSV-G glycoprotein-
expression vector
PMID: 9306402Addgene Plasmid
#12259
Dr. Didier Trono
(School of Life Sciences,
Ecole Polytechnique
Fédérale de
Lausanne,
Lausanne, Switzerland)
Recombinant
DNA reagent
Lentiviral packaging
construct pCMVΔR8.2
PMID: 9306402Addgene Plasmid
#12263
Dr. Didier Trono
(School of Life Sciences,
Ecole Polytechnique
Fédérale de
Lausanne,
Lausanne, Switzerland)
Recombinant
DNA reagent
Lentiviral construct
vector pLKO.3G-RFP
This paperN/AProgenitor: pLKO.3G
Recombinant
DNA reagent
Lentiviral construct
vector lentiCRISPRv2
PMID: 25075903Addgene Plasmid
#52961
Dr. Feng Zhang
(Broad Institute of MIT
and Harvard)
Recombinant
DNA reagent
Plasmid: 10His-
SUMO1-Q92R
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 10His
-SUMO2-Q88R
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 10His-
SUMO4-Q88R
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 3HA-
CDK9-KKR
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 3HA-
CDK9-RRK
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 3HA
-CDK9-RKK
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 3HA-
CDK9-KRR
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 3HA-
CDK9-KRK
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 3HA-
CDK9-RKR
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmid: 3HA-
CDK9-K0R
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Recombinant
DNA reagent
Plasmids:
3HA-CDK9-K0R-RXK
(X represent mutation
position)
This paperSupplementary file 3Progenitor: pcDNA3.1(+)
Sequence-
based reagent
siRNA LibraryRiboBioSupplementary file 1;
http://www.ribobio.com/
Sequence-
based reagent
ChIP-qPCR PrimersThis paperSupplementary file 2
Sequence-
based reagent
siRNA targeting
TRIM28 3’UTR:5’-
GCTCTGTTCTCTGTCCTGT-3’
RiboBiohttp://www.ribobio.com/
Sequence-
based reagent
shRNA targeting
Luciferase:5’-
ACCGCCTGAAGTCTCTGATTAA-3’
PMID: 29863470N/A
Sequence-
based reagent
shRNA targeting
TRIM28 CDS:5’-
CCAGCCAACCAGCGGAAATGTGA-3’
PMID: 18082607N/A
Sequence-
based reagent
sgRNA targeting
Dummyguide
(sgNT):5’-
ACGGAGGCTAAGCGTCGCAA-3’
PMID: 25075903N/ADr. Feng Zhang
(Broad Institute
of MIT and Harvard)
Sequence-
based reagent
sgRNA targeting
TRIM28
CDS:5’-
CACCGATTGAGCTGGCAGTCTCGGC-3’
PMID: 25075903N/ADr. Feng Zhang
(Broad Institute
of MIT and Harvard)
Sequence-
based reagent
β-Actin qPCR
Forward Primer:5’-
GCATGGAGTCCTGTGGCA-3’
PMID: 27291871N/A
Sequence-
based reagent
β-Actin qPCR
Reverse Primer:5’-
CAGGAGGAGCAATGATCTTGA-3’
PMID: 27291871N/A
Sequence-
based reagent
TRIM28 qPCR
Forward Primer:5’-
CTACTCAAGTGCAGAGCCCC-3’
This paperN/A
Sequence-
based reagent
TRIM28 qPCR
Reverse Primer:5’-
GGGAAGACCTTGAAGACGGG-3’
This paperN/A
Sequence-
based reagent
HIVTotRNA Forward
Primer:5’-
CTGGCTAACTAGGGAACCCACTGCT-3’
PMID: 27291871N/A
Sequence-
based reagent
HIVTotRNA Reverse
Primer:5’-
GCTTCAGCAAGCCGAGTCCTGCGTC-3’
PMID: 27535056N/A
Sequence-
based reagent
1 st round Nest PCR
Forward Primer
(E00):5’-
TAGAAAGAGCAGAAGACAGTGGCAATGA-3’
PMID: 27434587N/A
Sequence-
based reagent
1 st round Nest PCR
Reverse Primer (ES8B):5’-
CACTTCTCCAATTGTCCCTCA-3’
PMID: 27434587N/A
Sequence-
based reagent
2nd round Nest PCR
Forward Primer
(E20):5’-
GGGCCACACATGCCTGTGTACCCACAG-3’
PMID: 27434587N/A
Sequence-
based reagent
2nd round Nest PCR
Reverse Primer (E115):5’-
AGAAAAATTCCCCTCCACAATTAA-3’
PMID: 27434587N/A
Chemical
compound, drug
(+)-JQ-1SelleckchemCat#S7110
Chemical
compound, drug
Vorinostat (SAHA)SelleckchemCat#S1047
Chemical
compound, drug
Formaldehyde solutionSigma-AldrichCat#F8775-25ML
Chemical
compound, drug
TRIzol ReagentThermoFisherCat#15596018
Chemical
compound, drug
4',6-Diamidino-2-
Phenylindole,
Dihydrochloride (DAPI)
ThermoFisherCat#D1306
Chemical
compound, drug
Cysteamine (MEA)Sigma-AldrichCat#30070–10G
Chemical
compound, drug
Glucose Oxidase from
Aspergillus niger, Type
VII, lyophilized powder,
≥100,000 units/g solid
Sigma-AldrichCat#G2133-250KU
Chemical
compound, drug
Catalase from bovine liver
, lyophilized powder,
≥10,000 units/mg protein
Sigma-AldrichCat#C40-1G
Chemical
compound, drug
Sodium
borohydride (NaBH4)
Sigma-AldrichCat#213462–25G
Chemical
compound, drug
16% Paraformaldehyde
(formaldehyde) Aqueous
Solution
Electron Microscopy SciencesCat#15710
Chemical
compound, drug
8% Glutaraldehyde
Aqueous Solution
Electron Microscopy SciencesCat#16019
Chemical
compound, drug
Normal Donkey
Serum (NDS)
Jackson ImmunoResearchCat#017-000-121
Chemical
compound, drug
Triton X-100Sigma-AldrichCat#T8787-50ML
Chemical
compound, drug
Protease Inhibitor
Cocktail (PIC)
Sigma-AldrichCat#P8340-1ML
Chemical
compound, drug
N-Ethylmaleimide (NEM)SelleckchemCat#S3692
Chemical
compound, drug
EZview Red
Anti-HA Affinity Gel
Sigma-AldrichCat#E6779-1ML
Chemical
compound, drug
EZview Red Anti-
FLAG M2 Affinity Gel
Sigma-AldrichCat#F2426-1ML
Chemical
compound, drug
Anti-His-tag AgaroseAbcamCat#ab1231
Chemical
compound, drug
Penicillin-Streptomycin,
Liquid
ThermoFisherCat#15140122
Chemical
compound, drug
L-Glutamine,
200 mM Solution
ThermoFisherCat#25030081
Chemical
compound, drug
Fetal Bovine Serum (FBS)ThermoFisherCat#10270–106
Chemical
compound, drug
Phytohemagglutinin
-M (PHA-M)
Sigma-AldrichCat#11082132001
Peptide,
recombinant protein
Recombinant
Human TNF-α
PeproTechCat#300-01A
Peptide,
recombinant protein
Recombinant
Human IL-2
R&D SystemsCat#202-IL-500
Peptide,
recombinant protein
Recombinant Human
SUMO Activating
Enzyme E1 (SAE1/UBA2)
R&D SystemsCat#E-315
Peptide,
recombinant protein
Recombinant
Human UBE2I/Ubc9
R&D SystemsCat#E2-645-100
Peptide,
recombinant protein
Recombinant
Human CDK9
AbcamCat#ab85603
Peptide,
recombinant protein
Recombinant
Human SUMO4
This paperN/A
Peptide,
recombinant protein
Recombinant
Human TRIM28
AbcamCat#ab131899
Commercial
assay or kit
SUMO Conjugation
Reaction Buffer Kit
R&D SystemsCat#SK-15
Commercial
assay or kit
Human Lymphocyte
Separation Kit
TBDsciencesCat#LTS1077
Commercial
assay or kit
BD IMag Human
CD4 + T Lymphocyte
Enrichment Set-DM
BD BiosciencesCat#557939
Commercial
assay or kit
Luciferase Assay SystemPromegaCat#E4550
Commercial
assay or kit
SimpleChIP Enzymatic
Chromatin IP Kit (Magnetic
Beads)
CSTCat#9003S
Commercial
assay or kit
TruePrep DNA Library
Prep Kit V2 for Illumina
VazymeCat#TD501
Commercial
assay or kit
HIV-1 p24 ELISA KitAbcamCat#ab218268
Commercial
assay or kit
ProteoSilver Plus
Silver Stain Kit
Sigma-AldrichCat#PROTSIL2
-1KT
Commercial
assay or kit
CDK9/CyclinK Kinase
Enzyme System
PromegaCat#V4104
Commercial
assay or kit
ADP-GloTM
Kinase Assay
PromegaCat#V6903
Commercial
assay or kit
Cell Counting Kit-8DojindoCat#CK04;
Lot#KT793
Commercial
assay or kit
Zombie Violet
Fixable Viability Kit
BioLegendCat#423113;
Lot#B256957
Commercial
assay or kit
CellTrace CFSE Cell
Proliferation Kit -
For Flow Cytometry
ThermoFisherCat#C34554
Software,
algorithm
Prism 5GraphPadhttps://www.graphpad.com/scientific-software/prism/
Software,
algorithm
MEGA 7MEGAhttps://www.megasoftware.net/
Software,
algorithm
Cytoscape (3.6.1)Cytoscape
Consortium
RRID:SCR_015784
Software,
algorithm
STRINGCytoscape
Consortium
RRID:SCR_005223
Software,
algorithm
MCODECytoscape
Consortium
RRID:SCR_015828
Software,
algorithm
BD LSRFortessa cell analyzerBD Bioscienceshttp://www.bdbiosciences.com/in/instruments/lsr/index.jsp
Software,
algorithm
FlowJo V10Tree Starhttps://www.flowjo.com/
Software,
algorithm
Odyssey
CLX Imager
LI-COR
Biosciences
https://www.licor.com/bio/products/imaging_systems/odyssey/
Software,
algorithm
Image Studio
Lite Ver 4.0
LI-COR Bioscienceshttps://www.licor.com/bio/products/software/image_studio_lite/
Software,
algorithm
CFX ManagerBIO-RADhttp://www.bio-rad.com/
Software,
algorithm
GloMax 96 Microplate
Luminometer
Software
(version 1.9.3)
Promegahttps://www.promega.com/resources/software-firmware/detection-instruments-software/promega-branded-instruments/glomax-96-microplate-luminometer/
Software,
algorithm
SkanIt SW for
Microplate Readers
ThermoFisherhttps://www.thermofisher.com/order/catalog/product/5187139?SID=srch-srp-5187139
Software,
algorithm
NIS-Elements
Advanced Research
microscope
imaging
software
Nikonhttps://www.nikoninstruments.com/Products/Software
Software,
algorithm
PyMOLSchrödingerRRID:SCR_000305
Software,
algorithm
FastQCBabraham
Institute
RRID:SCR_014583
Software,
algorithm
Hisat2PMID: 25751142RRID:SCR_015530
Software,
algorithm
DEGseqBioconductorRRID:SCR_008480
Software
, algorithm
gplotsR Foundationhttps://www.rdocumentation.org/packages/gplots/versions/3.0.1
Software,
algorithm
Bowtie2PMID: 22388286RRID:SCR_016368
Software,
algorithm
SamtoolsPMID: 19505943RRID:SCR_002105
Software,
algorithm
igvtoolsBroad Institutehttps://software.broadinstitute.org/software/igv/igvtools
Software,
algorithm
Imaris
(Version 9.2)
BITPLANERRID:SCR_007370

Study participants

Request a detailed protocol

Chronically HIV-1-infected participants sampled by this study were recruited from Department of Infectious Diseases in Guangzhou 8th People’s Hospital, Guangzhou. The Ethics Review Board of Sun Yat-Sen University and the Ethics Review Board of Guangzhou 8th People’s Hospital approved this study. All the participants were given written informed consent with approval of the Ethics Committees. The enrollment of HIV-1-infected individuals was based on the criteria of prolonged suppression of plasma HIV-1 viremia on cART, which is undetectable plasma HIV-1 RNA levels (less than 50 copies/ml) for a minimum of 6 months, and having high CD4+ T cell count (at least 350 cells/mm3). Blood samples from healthy individuals were obtained from Guangzhou Blood Center. We did not have any interaction with the healthy individuals or protected information, and therefore no informed consent was required.

Cell lines

Request a detailed protocol

HEK293T (CVCL_0063) and HeLa (CVCL_0030) cells which were obtained from ATCC, and TZM-bl (8129) cells, which were obtained from NIH AIDS Reagent Program, were cultured in DMEM supplemented with 1% penicillin-streptomycin (ThermoFisher), 1% L-glutamine (ThermoFisher), and 10% FBS (ThermoFisher). J-Lat 6.3, 8.4, 9.2, 10.6 and 15.4 cell lines, which were originally generated from Dr. Eric Verdin (The Buck Institute for Research on Aging, Novato, CA) Laboratory, were obtained from Dr. Robert F. Siliciano (Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, MD) Laboratory. All the J-Lat cell lines were cultured in RPMI 1640 supplemented with 1% penicillin-streptomycin, 1% L-glutamine, and 10% FBS. Peripheral blood mononuclear cells (PBMCs) and primary CD4+ T cells, which were isolated and purified from study participants, were cultured in RPMI 1640 supplemented with 1% penicillin-streptomycin, 1% L-glutamine, and 10% FBS. 1/1000 Recombinant human interleukin 2 (IL-2) (R and D) was supplied for primary CD4+ T cells to maintain proliferation. All cells have been tested for mycoplasma using a PCR assay and confirmed to be mycoplasma-free. All cells cultured in sterile incubator at 37°C and 5% CO2.

SiRNA library screening

Request a detailed protocol

SiRNA library targeting 182 human genes, negative control siRNA (siNC) and siRNA targeting TRIM28 3’UTR (5’-GCTCTGTTCTCTGTCCTGT-3’) were purchased from RiboBio (Guangzhou, China) (Supplementary file 1). Three siRNAs were synthesized for each gene. The siRNAs targeting each gene were transfected as a mixture and have been validated by company to insure that at least one siRNA was able to knock down target gene mRNA up to 70%. The siRNA library covered six cellular pathways within the nucleus, which were chromatin binding, epigenetic modification, chromatin remodeling, ubiquitination, SUMOylation, and chromosome organization. Evenly mixed TZM-bl cell suspension was added into each well of 96-well plates with a Tecan Freedom EVO150 (Tecan, Männedorf, Schweiz) to insure that the cell confluency was 60% when the cells were transfected. Twelve hours post-seeding, cells from each well were transfected with siRNAs targeting each gene using Lipofectamine RNAiMAX (ThermoFisher) according to the manufacturer’s instruction. Each gene was set three biological replicates. Forty-eight hours post-transfection, cell samples from each well were removed culture medium and washed twice with PBS. Fifty microliter passive lysis buffer (Promega) was added into each well and lysed for 30 min with shaking. The cell lysates were clarified with centrifugation at 12,000 g for 3 min. Luciferase in the cell lysates was measured with luciferase-reporter assay (Promega) using a multiwell plate luminometer with an auto-injector (Promega) and analyzed by GloMax 96 Microplate Luminometer Software (Promega). Fold changes were calculated for each gene compared with siNC according to the light units.

ShRNA-mediated knockdown and CRISPR-CAS9-sgRNA-mediated knockout

Request a detailed protocol

ShRNA targeting luciferase (shluc: 5’-ACCGCCTGAAGTCTCTGATTAA-3’) was set as negative control (Rousseaux et al., 2018). The shRNA target sequence against TRIM28 CDS was 5’-CCAGCCAACCAGCGGAAATGTGA-3’ (Ivanov et al., 2007). Target sequences were cloned into pLKO.3G-RFP which was derived from pLKO.3G. The GFP-tag was replaced with RFP-tag in pLKO.3G-RFP. Pseudotyped viral stocks were produced in HEK293T cells by co-transfecting 3 μg of VSV-G glycoprotein-expression vector, 6 μg of lentiviral packaging construct pCMVΔR8.2, and 6 μg shRNA-expression lentiviral construct using Lipofectamine 2000 (ThermoFisher) according to the manufacturer’s instruction. VSV-G glycoprotein-expression vector was abtained from Addgene (Addgene plasmid # 12259). pCMVΔR8.2 was a kindly gift from Dr. Didier Trono (School of Life Sciences, Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland) (Zufferey et al., 1997). Virus supernatants from each 10 cm dish were concentrated into 1 ml RPMI 1640 by PEG 6000. J-Lat 6.3, 8.4, 9.2, 10.6 and 15.4 cell lines were spin-infected with shRNA virus. Forty-eight hours later, infected cells were treated with 500 nM SAHA (Selleckchem) or 1 μM JQ-1 (Selleckchem). Another 24 hr later, the percentages of GFP positive cells from each group were determined by BD LSRFortessa cell analyzer (BD Biosciences) and analyzed by FlowJo V10 (Tree Star). The infection efficiency was measured based on the percentage of RFP-positive cells using flow cytometry. The knockdown efficiency was confirmed by both qPCR and western blot.

For knocking out TRIM28, CRISPR-CAS9 system was used. SgRNA targeting dummyguide (sgNT: 5’-ACGGAGGCTAAGCGTCGCAA-3’) was set as negative control (Sanjana et al., 2014). The sgRNA target sequence against TRIM28 CDS was 5’-CACCGATTGAGCTGGCAGTCTCGGC-3’ (Sanjana et al., 2014). Target sequences were cloned into lentiCRISPRv2 (Sanjana et al., 2014). Pseudotyped viruses were produced and concentrated as shRNA viruses. J-Lat 10.6 cells were spin-infected with sgRNA virus and cultured for 48 hr followed by puromycin (Sigma-Aldrich) selection. Three days post-selection, the supernatant of infected cells was replaced with fresh RPMI 1640 and infected cells were went on culturing for 2 to 7 days. The knockout efficiency was confirmed both western blot. The percentages of GFP-positive cells were determined by flow cytometry.

ChIP-qPCR

Request a detailed protocol

Chromatin immunoprecipitation (ChIP) was performed according to the manufacturer’s instruction (CST). Approximately 4 × 106 cells were prepared for each immunoprecipitation (IP). Briefly, TZM-bl cells were treated with siNC, siTRIM28 or TNFα (PeproTech) for 48 hr followed by crosslinking proteins to DNA with 1% formaldehyde (Sigma-Aldrich) for 10 min at room temperature. The fixation was quenched with 125 mM glycine for 5 min at room temperature followed by centrifuging at 1,500 rpm for 5 min at 4°C. The supernatants were removed immediately. Cell pellets were resuspended in ice-cold Buffer A (CST) supplemented with DTT and Protease inhibitor cocktail (PIC) and incubated on ice for 10 min. The nuclei were enriched by centrifugation at 3000 rpm for 5 min at 4°C and resuspended in ice-cold Buffer B (CST) supplemented with DTT. Nuclei pellets were centrifuged again, removed supernatants and resuspended in 100 μl Buffer B supplemented with DTT and 0.5 μl micrococcal nuclease (CST) per IP preparation. The digestion was conducted at 37°C for 20 min. Incubation tubes were inverted several times per 5 min. After digestion, the reaction was stopped by adding 50 mM EDTA followed by centrifugation at 13,000 rpm for 1 min at 4°C. Nuclei pellet was resuspended in 100 μl ChIP Buffer (CST) supplemented with PIC per IP preparation and incubated for 10 min on ice. The nuclei pellet was further lysed by sonication with 3 sets of 20 s pulses at 40% amplitude. Pellet was incubated on ice for 30 s between pulses. The lysates were clarified by centrifugation at 10,000 rpm for 10 min at 4°C. The supernatants which contained digested chromatin were transferred into new tube. One-tenth of the chromatin sample was proceeded to analyze the size and concentration. Briefly, 50 μl chromatin sample was removed RNA by RNase A (CST) and reversed cross-linking by 200 mM NaCl and Proteinase K (CST). DNA from samples were purified by DNA purification spin columns (CST). Concentration was determined by measuring OD260. The size range was analyzed by electrophoresis on a 1% agarose gel, which should be between 150 and 900 bp.

For each IP preparation, approximately 10 μg chromatin was diluted into ChIP Buffer. Ten microliter diluted chromatin, which was 2% input sample, was transferred to a new tube and stored at −20°C. Immunoprecipitation antibodies normal rabbit IgG (CST, 2729), anti-TRIM28 antibody (Proteintech, 15202–1-AP), anti-H3K9me2 antibody (Abcam, ab1220), anti-H3K9me3 antibody (Abcam, ab8898), anti-H3K4me3 antibody (Abcam, ab8580), anti-H3K27me3 antibody (Abcam, ab6002), anti-H3K9Acetyl antibody (Abcam, ab4441), anti-CDK9 antibody (CST, 2316), and anti-RNA polymerase II CTD repeat YSPTSPS (phospho Ser2) antibody (Abcam, ab5095) were separately added to siNC and siTRIM28 groups, respectively. The immunoprecipitation was carried out overnight at 4°C while rotating. ChIP-Grade Protein G Magnetic Beads (CST) were added to the each IP reaction and incubated with IP samples for another 2 hr at 4°Cwhile rotating. The protein G magnetic beads were pelleted by placing the IP tubes in a magnetic separation rack and washed with 3 times low-salt washes and one time high-salt wash. Each wash was conducted at 4°C for 5 min while rotating. DNA enriched by protein G magnetic beads was eluted by ChIP Elution Buffer (CST). All the DNA samples including 2% input samples were reversed cross-linking with 200 mM NaCl and Proteinase K and purified as above.

ChIP primers targeting the HIV-1 mini-model in TZM-bl cell line were used to quantitate each target by Real-Time Quantitative PCR. The quantitation regions were shown below. G5: Cellular DNA and viral 5’LTR junction; A: Nucleosome 0 assembly site; B: Nucleosome free region; C: Nucleosome one assembly site; V5: Viral 5’LTR and gag leader sequence junction; L: Luciferase region; V3: Viral poly purine tract and 3’LTR junction; G3: Viral 3’LTR and cellular DNA junction. Primers which amplified each region were shown in Supplementary file 2. All the ChIP-qPCR DNA signals were normalized to siNC IgG of G5. ChIP-qPCR in J-Lat 10.6 cell line was conducted as in TZM-bl cell line. In J-Lat 10.6, G5’ represented cellular DNA and viral 5’LTR junction; E represented envelop; G3’ represented viral 3’LTR and cellular DNA junction; A, B, C, V5 and V3 represented as in Figure 1D.

RNA isolation, reverse transcription and qPCR

Request a detailed protocol

The identities of unstimulated primary CD4+ T cells, PHA-stimulated primary CD4+ T cells and resting CD4+ T cells were confirmed by flow cytometry with antibodies against human CD4 (ThermoFisher, 11-0048-42), CD45RA (BioLegend, 304127), CD45RO (BD Biosciences, 560607), CD62L (ThermoFisher, 25-0629-42), CD69 (ThermoFisher, 25-0699-42) and CD25 (ThermoFisher, 15-0259-42). RNAs from indicated numbers of cells were isolated with TRIzol reagent (ThermoFisher) and proceeded to cDNA synthesis with PrimeScript RT reagent Kit (Takara). For the samples which quantitated the expression of TRIM28, Real-time PCR was performed with SYBR Ex-taq premix (Takara) in a CFX96 Real-time PCR Detection System (Bio-Rad). Human β-actin mRNA was measured as internal control (Li et al., 2016). Primer pairs were shown as below: β-actin qPCR Forward Primer: 5’-GCATGGAGTCCTGTGGCA-3’, β-actin qPCR Reverse Primer: 5’-CAGGAGGAGCAATGATCTTGA-3’; TRIM28 qPCR Forward Primer: 5’-CTACTCAAGTGCAGAGCCCC-3’, TRIM28 qPCR Reverse Primer: 5’-GGGAAGACCTTGAAGACGGG-3’. The relative expression of each gene was calculated as 2[Ct(Control-TRIM28)-Ct(Control-β-Actin)]-[Ct(Treatment-TRIM28)-Ct(Treatment-β-Actin)]. For the quantitation of HIV-1 expression, a specific reverse primer was used to reversely transcribe HIV-1 RNA: 5’- GCTTCAGCAAGCCGAGTCCTGCGTC-3’. QPCR was performed for specific reverse-transcribed HIV-1 cDNA with primer pairs: HIVTotRNA Forward Primer: 5’-CTGGCTAACTAGGGAACCCACTGCT-3’ and HIVTotRNA Reverse Primer: 5’-GCTTCAGCAAGCCGAGTCCTGCGTC-3’ (Liu et al., 2016). After quantitation, an in vitro transcribed HIV-1 RNA was used as the external control for measuring cell-associated viral RNAs. The Ct of each group was converted to mass and further converted to copies. The final expression of intracellular HIV-1 RNA was represented as 103 copies viral RNA per million CD4 +T cells.

Global site-specific SUMO-MS

Request a detailed protocol

His-tagged SUMO mutants SUMO1-Q92R, SUMO2-Q88R and SUMO4-Q88R were co-overexpressed with E2 UBC9 and E3 TRIM28 in HeLa cells. Forty-eight hours post-transfection, cell pellets were lysed by guanidine lysis buffer (6 M guanidine-HCl, 100 mM sodium phosphate, and 10 mM Tris, buffered at pH 8.0). Lysates were sonicated for 15 s with 5 s pulse at a power of 30 W. Subsequently, prewashed anti-His Ni-NTA agarose beads (QIAGEN), 50 mM imidazole and 5 mM β-mercaptoethanol were added into the lysates and tumbled overnight at 4C. After overnight incubation, beads were centrifuged at 500 r.c.f. and washed for 30 min at 4C with the following wash buffers in order: wash buffer A (6 M guanidine-HCl, 0.1% Triton X-100, 10 mM imidazole, 5 mM β-mercaptoethanol, 100 mM sodium phosphate, and 10 mM Tris, buffered at pH 8.0), wash buffer B (8 M urea, 0.1% Triton X-100, 10 mM imidazole, 5 mM β-mercaptoethanol, 100 mM sodium phosphate, and 10 mM Tris, buffered at pH 8.0), wash buffer C (8 M urea, 10 mM imidazole, 5 mM β-mercaptoethanol, 100 mM sodium phosphate, and 10 mM Tris, buffered at pH 6.3), wash buffer D (8 M urea, 5 mM β-mercaptoethanol, 100 mM sodium phosphate, and 10 mM Tris, buffered at pH 6.3), and wash buffer E (same as wash buffer D). After washing, proteins were eluted three times from beads with elution buffer (7 M urea, 500 mM imidazole, 100 mM sodium phosphate, and 10 mM Tris, buffered at pH 7.0) for 30 min at 4C. All the eluates were combined together and filtered with 0.45 μm filter (Millipore). The clarified proteins were concentrated with a 10 kDa-cutoff filter (Millipore) and washed with PBS for three times. Concentrated proteins were transferred to new tubes and boiled with 4 × protein SDS-PAGE loading buffer (Takara) at 100C for 15 min. Samples were separated with 4–12% protein gel (ThermoFisher). The gel was dyed with silver stain kit (Sigma-Aldrich). Sixteen gel slices were cut out and proceeded to in-gel digestion.

Briefly, gel slices were destained and treated with 10 mM DTT followed by the treatment of 55 mM iodoacetamide. The gels were washed with 25 mM NH4HCO3 and 25 mM NH4HCO3 in 50% ACN followed by desiccation with vacuum. One hundred nanogram trypsin (ThermoFisher) which was dissolved in 25 mM NH4HCO3 was added to each gel and incubated overnight at 37C. Twenty four hours later, digested peptides were extracted with the following extraction solutions in order: 50% ACN containing 5%TFA, 75% ACN containing 0.1% TFA, and 100% ACN. The extracts were subjected to vacuum for 3 hr to remove the solvent. The peptides were desalted and enriched by C18 ZipTip (Millipore), and redissolved in 50% ACN containing 0.1% TFA, followed by vacuum to remove the solvent. Twelve microliter of 0.01% formic acid was used to resolve the peptides and proceeded to nanoscale LC-MS/MS with an EASY-nLC system (ThermoFisher) connected to a Q-Exactive (ThermoFisher) with higher collisional dissociation (HCD) fragmentation. Peptide were separated by 20-cm-long analytical columns (ID 75 μm, Polymicro Avantes) packed in house with Luna 3.0u C18 (2) 100A (Phenomenex) with a 90-min gradient from 3% to 90% acetonitrile in 0.1% formic acid and a flow rate of 300 nL/min. Data-dependent acquisition mode with a top-ten method was used to operate the mass spectrometer. Full-scan MS spectra were obtained with a target value of 3E6, a resolution of 70,000, with a scan range from 300 to 1,800 m/z. HCD tandem MS/MS spectra were obtained with a target value of 1E6, a resolution of 17,500, and a normalized collision energy of 25%. Unknown charges, or charges lower than two and higher than eight were rejected.

Target-specific SUMO-MS

Request a detailed protocol

To confirm the SUMOylation sites on CDK9 by SUMO-MS, two different tagged SUMO4 mutants were used to co-overexpressed with HA-tagged CDK9, respectively, which were Flag-tagged SUMO4-Q88R and His-tagged SUMO4-Q88R. Anti-HA-tag beads (Sigma-Aldrich) were used to immunoprecipitate CDK9 and corresponding SUMO-CDK9. Enriched target proteins were eluted from beads by boiling with 4 × protein SDS PAGE loading buffer at 100°C for 15 min. The supernatants containing target proteins were transferred to new tubes after centrifugation at 12,000 rpm for 3 min. One part of the samples was proceeded to western blot with antibodies against HA-tag, Flag-tag and His-tag to determine the SUMOylation efficiency. The left samples were separated with 4–12% SDS-PAGE protein gel and developed with silver staining. Stained bands which indicated the SUMOylated CDK9 were cut out and proceeded to in-gel digestion as above. LC-MS/MS was used to analyze the SUMOylated peptides as we have described in Global site-specific SUMO-MS.

Co-immunoprecipitation and western blot

Request a detailed protocol

For all the SUMOylation-related co-immunoprecipitation (Co-IP), different tagged protein-expression constructs were transfected into Hela cells which were cultured in 6 cm dishes. Forty-eight hours post-transfection, cells were washed twice with PBS and lysed with NP-40 lysis buffer (10 mM Tris-HCl buffered at pH 7.5, 150 mM NaCl, 0.5% NP-40, 1% Triton X-100, 10% Glycerol, 2 mM EDTA, 1 mM NaF, 1 Mm Na3VO4) supplemented with 1/100 protease inhibitor cocktail (PIC) (Sigma-Aldrich) and 2 M N-Ethylmaleimide (NEM) (Selleckchem) for 30 min on ice. Every 10 min, the incubation tubes were inverted several times. The lysates were clarified by centrifugation at 12,000 rpm for 10 min at 4°C, followed by incubating with anti-HA-tag beads (Sigma-Aldrich), anti-Flag-tag beads (Sigma-Aldrich) or anti-His-tag beads (Abcam) for 4 hr to overnight at 4°C while rotating. The next day, proteins which were enriched by beads were washed for five times with ice-cold STN IP wash buffer (10 mM Tri-HCl buffered at pH 7.5, 150 mM NaCl, 0.5% NP-40, 0.5% Triton X-100) and eluted by boiling with 4 × protein SDS-PAGE loading buffer at 100°C for 15 min. The supernatants containing target proteins were transferred to new tubes after centrifugation at 12,000 rpm for 3 min, followed by western blot with antibodies against HA-tag (MBL, PM020), Flag-tag (MBL, M180-3), His-tag (Proteintech, 66005–1-Ig) or other indicated antibodies. GAPDH (Proteintech, 10494–1-AP) was set as internal reference. 680RD goat anti-mouse IgG antibody (LI-COR Biosciences, 926–68070) and 800CW goat anti-rabbit IgG antibody (LI-COR Biosciences, 926–32211) were used as secondary antibodies. The western blot membranes were developed with Odyssey CLX Imager (LI-COR Biosciences) and analyzed by Image Studio Lite Ver 4.0 (LI-COR Biosciences).

SUMOylation and in vitro SUMOylation assay

Request a detailed protocol

For a given protein, the SUMOylated form is only a small proportion. To enhance the SUMOylation signals, we conducted several SUMOylation assay by co-overexpression target proteins with SUMOylation system components which were SUMOs, E1 SAE1/UBA2, E2 UBC9, and E3 TRIM28. In vertebrates, there are four well-studied SUMO paralogs, SUMO1, SUMO2, SUMO3, and SUMO4. Because SUMO2 and SUMO3 share highly sequence identity and have similar functions, they are referred to as SUMO2/3. In preliminary data, we found the overexpression of E1 had little influence on the SUMOylation due to the high expression of endogenous E1. Therefore, we omitted E1 in the following SUMOylation assays. Besides, there are lots SUMO-specific isopeptidases (SENPs) which deSUMOylate substrates. Thus we used mature SUMO polypeptides instead of immature ones. For CDK9 SUMOylation assay, 2 μg HA-tagged wild type or mutated CDK9-expression plasmids, 4 μg Flag-tagged SUMO4-expression plasmids, 500 ng Flag-tagged UBC9 and 500 ng Flag-tagged TRIM28 were co-transfected into Hela cells which cultured in 6 cm dishes. Forty-eight hours post-transfection, cells were harvested in NP-40 lysis buffer containing 2 M NEM which was used to prevent deSUMOylation. Co-IP and western blot against HA-tagged CDK9 was performed according to the procedure which we mentioned above. For SENP3-mediated deSUMOylation assay, 500 ng or 1 μg SENP3-expression plasmids were additionally co-overexpressed with indicated amount of CDK9, SUMO4, UBC9 and TRIM28. Specific antibodies against SENP3 (Proteintech, 17659–1-AP) was used in western blot to confirm the expression.

In vitro SUMOylation assay was performed by co-culturing in vitro-purified 1 μg CDK9 (Abcam) with in vitro-purified 4 μg SUMO4 (This paper), 500 ng E1 (SAE1/UBA2) (R and D), 500 ng UBC9 (R and D) or 500 ng TRIM28 (Abcam) in SUMO conjugation reaction buffer (R and D). The reaction was initiated by adding 1 mM Mg-ATP solution and incubated for 3 hr at 30°C, followed by adding stop buffer to terminate the reaction. Samples were boiled with SDS-PAGE loading buffer supplemented with 1 M DTT for 15 min at 100°C and proceeded to western blot with specific antibodies against CDK9 (CST, 2316), SUMO4 (Abcam, ab126606), SAE1 (Proteintech, 10229–1-AP), UBA2 (Abclonal, A4363), UBC9 (Abclonal, A2193), and TRIM28 (Proteintech, 15202–1-AP).

SIM and STORM imaging

Request a detailed protocol

For samples used for super-resolution Structured Illumination Microscopy (SIM) imaging, HEK293T cells were plated into Lab-Tek II chambered coverglass (ThermoFisher) which was pretreated with poly-lysine (Sigma-Aldrich). Twelve hours later, cells were transfected with GFP-tagged TRIM28 or GFP-tagged TRIM28-dRING with RFP-tagged CDK9. Twenty-four hours post-transfection, cells were washed with PBS once and fixed with 3% paraformaldehyde (Electron Microscopy Sciences)/0.1% glutaraldehyde (Electron Microscopy Sciences) for 10 min at room temperature (RT). Fixed samples were reduced with 0.1% NaBH4 (Sigma-Aldrich) for 7 min at room temperature while shaking, followed by washing with PBS for 3 times at room temperature, 5 min per wash. Cells were further permeabilized with 0.2% Triton X-100 (Sigma-Aldrich) for 15 min and blocked with 10% normal donkey serum (NDS) (Jackson ImmunoResearch)/0.05% Triton X-100 for 90 min at RT. After blocking, samples were washed with 1% NDS/0.05% Triton X-100 for 15 min at RT for five times. Then, samples were wash with PBS once for 5 min, followed by post-fixation for 10 min with 3% paraformaldehyde/0.1% glutaraldehyde. After post-fixation, samples were washed with PBS for three times, 5 min per wash. 4', 6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI) (ThermoFisher) solution was added into samples to dye DNA for 10 min while shaking. Finally, samples were washed with PBS for three times and imaged on an Eclipse Ti inverted microscope equipped with a CFI Apo TIRF objective (NA 1.49, oil immersion) and NIS-Elements AR software, an sCMOS camera (Hamamatsu Flash 4.0, 6.5 μm × 6.5 μm pixel size), and four lasers named SIM 405, SIM 488, SIM 561 and SIM 647. The original images were acquired with 512 × 512 resolution and reconstructed to form the SIM images with 1024 × 1024 resolution. The lateral resolution of the SIM image is 115 nm and the axial resolution is 300 nm. Z-step size was set to 0.20 μm. For each focal plane, 15 images (five phases, three angles, 3D-SIM mode) were captured with the NIS-Elements software. SIM images were reconstructed and analyzed with the N-SIM module of the NIS-Elements Advanced Research software (Nikon). For the quantitation of co-localization, SIM images were further analyzed with Imaris software (Version 9.2) (BITPLANE) using Coloc toolbar. Percentages of each channel voxels above threshold co-localized were calculated. Both Pearson`s coefficient and thresholded Mander’s coefficient were calculated to indicate the qualities of co-localization. For Pearson’s coefficient, a value of 1 represents perfect co-localization, 0 no co-localization, and −1 perfect inverse co-localization. For thresholded Mander’s coefficient, a value of 1 represents perfect co-localization and 0 no co-localization.

For samples used for super-resolution continuous STochastic Optical Reconstruction Microscopy (cSTORM) imaging, cells were plated, fixed, reduced, permeabilized, blocked and washed as in SIM samples preparation. After blocking, primary antibodies against TRIM28 (Proteintech, 66630–1-Ig), SUMO4 (Abcam, ab126606) and CDK9 (CST, 2316) were incubated with cells for 60 min at RT in 5% NDS/0.05% Triton X-100. Samples were washed for five times with 1% NDS/0.05% Triton X-100 at RT, 15 min per wash. Then, cells were incubated with secondary antibodies diluted in 5% NDS/0.05% Triton X-100 for 30 min at RT while shaking. Two sets of secondary antibody pairs were used to confirm the specificity, which were: Donkey Anti-Mouse IgG H and L (Alexa Fluor 647) Antibody (Abcam, ab150107) combining with Donkey Anti-Rabbit IgG H and L (CF 568) Antibody (Biotium, 20803–500 μl), Donkey Anti-Rabbit IgG H and L (Alexa Fluor 647) Antibody (Abcam, ab150075) combining with Donkey Anti-Mouse IgG H and L (CF 568) Antibody (Biotium, 20802–500 μl). After incubation, cells were washed as above followed by another wash with PBS for 5 min. Post-fixation was performed with 3% paraformaldehyde/0.1% glutaraldehyde for 10 min without shaking. Then, cells were washed with PBS for three times, 5 min per wash, followed by washing with water for two times, 3 min per wash. Of note, DAPI and Hoechst were not allowed to dye DNA according to cSTORM protocol. cSTORM imaging buffer was freshly prepared as below. GLOX solution was compounded by mixing 100 μl of 70 mg/ml Glucose Oxidase (Sigma-Aldrich) diluted in Buffer A (10 mM Tris-HCl buffered at pH 8.0, 50 mM NaCl) with 25 μl of 17 mg/ml Catalase (Sigma-Aldrich) diluted in Buffer A. One mole per liter of Cysteamine (MEA) (Sigma-Aldrich) was compounded by diluting 77 mg of MEA into 1 ml 0.25 N HCl. On ice, cSTORM imaging buffer was compounded by mixing 7 μl of GLOX, 70 μl of 1M MEA, and 620 μl of Buffer B (50 mM Tris-HCl buffered at pH 8.0, 10 mM NaCl, 10% Glucose). Each well of Lab-Tek II chambered coverglass was added 700 μl of imaging buffer which was able to be used for 2 hr. Samples were imaged under a Nikon N-STORM super-resolution microscope equipped with a high-numerical-aperture (high-NA) 100 × oil immersion objective (Nikon CFI SR Apochromat TIRF 100 × oil, 1.49 NA), a high-sensitivity and high-resolution sCMOS camera (Hamamatsu Flash 4.0, 6.5 μm × 6.5 μm pixel size, and an 0.4 × relay lens to match the pixel size under STORM mode), and four lasers with excitation wavelengths of 405, 488, 561 and 647 nm. For cSTORM which we used here, 405 nm laser was used as activation laser. 488 nm, 561 nm and 647 nm lasers were used as reporter lasers. The lateral resolution of the cSTORM image is 20 nm and the axial resolution is 50 nm. The z position was maintained during the acquisition by a Nikon ‘perfect focus system’. 20,000 to 25,000 frames were taken for each image. Single molecule localization was obtained by Gaussian fitting using the STORM plug-in of NIS-Elements Advanced Research software taking into account both drift and chromatic aberrations. For the quantitation of co-localization, cSTORM images were further analyzed with Imaris software (Version 9.2) (BITPLANE) by measuring the distance of spots-spots center. cSTORM-imaged protein molecules and complexes were transformed into small or large spots based on their diameter. The spots-spots co-localization was defined by the criterion of maximal distance of 10 nm. The complexes-spots co-localization was defined by the criterion of maximal distance of 100 nm. The percentages of co-localization were calculated for both total proteins-proteins co-localization, spots-spots co-localization and complexes-spots co-localization for each protein.

CDK9 kinase assay

Request a detailed protocol

In vitro SUMOylation assay was performed for CDK9 as described above. Five groups were set: Group 1 (G1): CDK9 only; Group 2 (G2): CDK9 and SUMO4; Group 3 (G3): CDK9, SUMO4 and E1 (SAE1 and UBA2); Group 4 (G4): CDK9, SUMO4, E1 and E2 (UBC9); Group 5 (G5): CDK9, SUMO4, E1, E2 and E3 (TRIM28). The reaction was terminated by stop buffer. To initiate the CDK9 kinase assay, CDK9 substrate PDKtides and ATP were added into each samples according to the manufacturer’s instruction (Promega). The reaction was incubated for 120 min at room temperature followed by ADP-Glo kinase assay (Promega). Briefly, ADP-Glo reagent was added into the reaction to deplete the remaining ATP. Samples were incubated at room temperature for 40 min. After ATP depletion, kinase detection reagent was added into samples to convert the ADP which was consumed during CDK9 kinase assay to ATP. This reaction was performed by incubating samples at room temperature for 30 min. Finally, the newly synthesized ATP was quantitated using luciferase/luciferin reaction. The luminescence generated during luciferase/luciferin reaction was recorded with integration time of 0.5 to 1 s. The relative light units were calculated by normalizing to untreated wild-type CDK9 group.

Toxicity assay

Request a detailed protocol

TRIM28 in Hela cells and HIV-1-infected CD4+ T cells was knocked down by siRNA targeting TRIM28. ShRNA and sgRNA lentiviruses targeting TRIM28 were used to knock down TRIM28 and knock out TRIM28 in J-Lat 10.6, respectively. The cytotoxicity assay was conducted by incubating Cell Counting Kit-8 (CCK-8) reagents (Dojindo, CK04) with wild type and TRIM28-deficient cells for 3 hr followed by measuring the absorbance at 450 nm using a microplate reader. The cell viability assay was conducted by measuring the percentage of amine-reactive fluorescent dye (BioLegend, 423113) non-permeant cells, which indicated the percentage of viable cells. Cell numbers were recorded every 2 days for both wild-type and TRIM28-deficient cells. The proliferation assay was conducted by staining live cells with CFSE (ThermoFisher, C34554). On Day 0, cells from each group were stained with CFSE. The percentage and mean fluorescence intensity (MFI) of CFSE-positive cells were analyzed by flow cytometry every 2 days.

Virus out-growth assay

Request a detailed protocol

Resting CD4 +cells were isolated from HIV-1-infected individuals who underwent suppressive cART for at least 6 months with undetectable plasma HIV-1 RNA (less than 50 copies/ml) and high CD4+ T cell count (at least 350 cells/mm3) (Human Lymphocyte Separation Kit, TBDsciences; BD IMag Human CD4+ T Lymphocyte Enrichment Set-DM, BD Biosciences). These CD4+ T cells were nucleofected with siRNAs targeting negative control and TRIM28 respectively, and cultured in Super T Cell Medium (STCM) consisting of RPMI 1640 supplemented with 1% penicillin-streptomycin, 1% L-glutamine, 10% FBS, 100 U/ml IL-2, and 2% T-cell growth factor (TCGF) from the supernatants of mitogen-activated healthy PBMCs treated with 2 μg/ml PHA-M and 5 ng/ml PMA for 4 hr. Six hours post-transfection, supernatants were replaced with new culture medium. Twenty-four hours later, half of siNC-treated cells were separated and supplemented with 0.5 μg/ml PHA-M (Sigma-Aldrich). Half of siTRIM28-treated cells were separated and supplemented with 500 nM Vorinostat (SAHA) (Selleckchem). Another 24 hr later, supernatants from each group were changed with fresh culture medium to prevent the toxicity of the PHA-M or SAHA. Seventy-two hours post-transfection, cells were exposed to 20 Gy X-ray irradiation for 5 min and supplemented with PHA-activated healthy CD4 +T cells. The supernatants were collected and half-changed with fresh STCM every 3 days. Cell suspension was half-changed with PHA-stimulated healthy CD4+ T cell suspension every 6 days. All the supernatants from each time points and each groups were measured for the presence of HIV-1 antigen with HIV-1 p24 ELISA kit (Abcam) according to the manufacturer’s instruction by SkanIt SW for Microplate Readers (ThermoFisher).

Genetic diversity analysis

Request a detailed protocol

The genetic diversity of HIV-1 quasispecies under different conditions was evaluated by sequencing the envelope V1-V3 region. HIV-1 RNAs from each group were reverse-transcribed by specific primer ES8B: 5’-CACTTCTCCAATTGTCCCTCA-3’. Two rounds of nested PCR were performed to amplify V1-V3 region with the following primer pairs: 1st round Nest PCR Forward Primer (E00): 5’-TAGAAAGAGCAGAAGACAGTGGCAATGA-3’, 1st round Nest PCR Reverse Primer (ES8B): 5’-CACTTCTCCAATTGTCCCTCA-3’; 2nd round Nest PCR Forward Primer (E20): 5’-GGGCCACACATGCCTGTGTACCCACAG-3’, 2nd round Nest PCR Reverse Primer (E115): 5’-AGAAAAATTCCCCTCCACAATTAA-3’ (Geng et al., 2016b). For each PCR reaction, Phanta Max Super-Fidelity DNA Polymerase (Vazyme) was used to amplify the V1-V3 region of HIV-1 envelope in order to ensure the fidelity. The amplification error rate of Phanta Max is 53-fold lower than that of Taq and 6-fold lower than that of Pfu according to the manufacturer’s instruction. After two rounds of nested PCR utilizing Phanta Max, the PCR products were proceeded to deoxyadenosine (A)-tailing at the 3'-end of the PCR products utilizing Ex Taq DNA polymerase (Takara) without thermal cycling as follows: 95°C, 5 min; 72°C, 30 min; 4°C hold. The A-tailed PCR products were TA-ligated into pMD-18T vector. To minimize the sampling bias, single genome amplification method was performed by obtaining 30 independent PCR products from each sample. At least 60 single clones were picked from each group and proceeded to Sanger sequencing. The sequences from each group were aligned using MUSCLE. The sequences with ambiguous positions were removed. The average genetic distance between one give clone and the relevant entire population were calculated by MEGA seven and represented as genetic diversity index. The Mann-Whitney U-test was performed to compare the genetic diversity indexes between different groups using Prism 5. The phylogenetic bootstrap consensus trees were generated for each samples using neighbor-joining method with 1000 bootstrap replications implemented in MEGA seven to depict the global landscape of HIV-1 diversity.

RNA-Seq and ATAC-Seq

Request a detailed protocol

Freshly isolated CD4+ T cells were stimulated with PHA for 2 days or left untreated. Total RNAs from each group were extracted by TRIzol Reagent (ThermoFisher) according to the manufacturer’s instruction. The quality of RNA samples were evaluated by Nanodrop 2000 (ThermoFisher) and BioAnalyzer 2100 (Aglient). The RNA-Seq library were built with TruSeq Stranded mRNA Library Prep Kit (Illumina) and sequenced with HiSeq X Ten (Illumina) at BioMarker (Beijing, China) under the PE150 protocol. RNA-Seq reads were trimmed, filtered and quality-controlled by FastQC (Babraham Institute) tool. The reads were aligned to human reference genome NCBI build 38 (GRCh38) by Hisat2 (Kim et al., 2015), followed by calculating the reads per kilobase per million mapped reads (RPKM). Differentially expressed genes were filtered by DEGseq (Bioconductor) tool with log2FC of 1 and PvalueFDR cutoff of 0.05, and plotted as heatmap or volcanoplot by gplots (R Foundation).

TRIM28-defective (sgTRIM28) J-Lat 10.6 cell line was generated by CRISPR-CAS9 technique. ATAC-Seq was conducted with sgNT and sgTRIM28 J-Lat 10.6 cell lines, as well as siNC and siTRIM28 TZM-bl cell lines. The ATAC-Seq library was built with TruePrep DNA Library Prep Kit V2 (Vazyme) as previously described (Buenrostro et al., 2013). Briefly, approximately 30,000 cells were harvested, washed with ice-cold PBS, and lysed with 50 μl of ice-cold lysis buffer (10 mM Tris-HCl buffered at pH 7.4, 10 mM NaCl, 3 mM MgCl2, 0.1% Igepal CA-630) for 10 min on ice. The lysates were centrifuged for 5 min at 500 G, 4°C. The supernatants were carefully removed. Transposition reaction mix, which consisted of 10 μl of 5 × TTBL, 5 μl of TTE Mix V50 and 35 μl of ddH2O, was used to resuspend nuclei pellet and incubated at 37°C for 30 min. The transposed DNA was purified by VAHTS DNA Clean Beads (Vazyme) and PCR-amplified with the following mixture: 24 μl of purified DNA, 10 μl of 5 × TAB, 5 μl of PPM, 5 μl of N5 primer, 5 μl of N7 primer, and 1 μl of TAE. Thermal cycle was as follows: 72°C for 3 min; 98°C for 30 s; and thermocycling at 98°C for 15 s, 60°C for 30 s and 72°C for 3 min; following by 72°C 5 min. The amplified ATAC-Seq library was purified with VAHTS DNA Clean Beads and eluted with 30 μl ddH2O. The library quality was evaluated by Qubit 3.0 Fluorometer (ThermoFisher) and BioAnalyzer 2100 (Aglient), and sequenced with HiSeq X Ten (Illumina) at BioMarker (Beijing, China) under the PE150 protocol. ATAC-Seq reads were trimmed, filtered and quality-controlled by FastQC tool. Then the reads were aligned to GRCh38 by Bowtie2 (Langmead and Salzberg, 2012), followed by rearranging with Samtools (Li et al., 2009). The reads were also separately aligned to HIV-1 reference genome K03455, M38432 (Version K03455.1) by Bowtie2, followed by rearranging with Samtools. Igvtools (Broad Institute) was used to visualize the tag peaks. Specific gene loci was amplified. Tag density from different groups was calculated by normalizing to the total mapped reads. The highest tag density was set as 100. Relative tag densities of two kilobases range centered HIV-1 5’LTR integration sites were calculated and compared with sgNT or siNC.

Statistical analysis

Request a detailed protocol

Triplicates data were presented as mean ±SEM. A value of p<0.05 was considered to be statistically significant and represented as asterisk (*). Value of p<0.01 was considered to be more statistically significant and represented as double asterisks (**). Value of p<0.001 was considered to be the most statistically significant and represented as triple asterisks (***). For the comparison of ChIP, the GFP-positive percentages and qPCR experiments, standard t test was used. For the comparison of genetic diversity index experiment, Mann-Whitney U-test was used. Statistical analyses were conducted with Prism 5 (GraphPad). The network analysis and clustering analysis were conducted with STRING and MCODE in Cytoscape (Cytoscape Consortium). Co-crystal structure of Cyclin T1 and CDK9 (PDB ID: 4EC8) were reconstituted in PyMOL (Schrödinger) (Baumli et al., 2012). Both ribbon models and surface models were used to present the structure.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
    Effects of CCR5 and CD4 cell surface concentrations on infections by macrophagetropic isolates of human immunodeficiency virus type 1
    1. EJ Platt
    2. K Wehrly
    3. SE Kuhmann
    4. B Chesebro
    5. D Kabat
    (1998)
    Journal of Virology 72:2855–2864.
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83

Decision letter

  1. Jeremy Luban
    Reviewing Editor; University of Massachusetts Medical School, United States
  2. Wenhui Li
    Senior Editor; National Institute of Biological Sciences, China

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "TRIM28 Inhibits the Function of P-TEFb Complex by SUMOylating CDK9 and Leads to HIV-1 Latency" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Wenhui Li as the Senior Editor. The following individual involved in review of your submission has agreed to reveal his identity: Guangxia Gao (Reviewer #1).

Summary:

Your manuscript was evaluated by a dedicated group of reviewers, who have discussed the reviews with one another. The Reviewing Editor has drafted this decision to help you prepare a revised submission. We all agree that you have described an important body of work and wish to see your manuscript go forward.

Essential revisions:

There are two points that were highlighted in the comments of at least two reviewers:

1) Please provide evidence for the importance of SUMO4. Does SUMO4 disruption alter the chromatin status of the HIV DNA? Please provide evidence that SUMO4 is important in the primary cells of interest; minimally you should demonstrate that SUMO4 is expressed in the cell type of interest. Without this additional data we feel that you need to remove claims about the importance of SUMO4.

2) Please clarify whether or not the ATAC-Seq effect that you report is specific for the HIV LTR. Does this change in chromatin state occur with promoters more generally?

Reviewer #1:

It is well documented that TRIM28 (also known as KAP1) mediates transcriptional silencing of MLV in stem cells and endogenous retroviruses. In this manuscript, Ma et al. reported that the SUMO E3 activity of TRIM28 is important for the silencing of HIV-1 DNA in reporter cell lines (TZM-bl, J-Lat) and primary resting T cells from patients. The major findings are interesting and important and the biochemistry results of TRIM28-mediated Sumoylation of CDK9 are pretty clear and convincing. Some clarifications are required to further support the conclusions.

Major concerns:

1) As shown in Figure 1, TRIM28 is responsible for the repressive epigenetic marks deposited on HIV-1 DNA. Is the E3 activity of important for these repressive epigenetic modifications? The authors can knockdown SUMO4 (the major SUMO used by TRIM28), maybe also CDK9, and examine the repressive epigenetic marks on HIV-1 DNA (with ChIP antibodies H3K9me3, H3K27me3, H3ac, SETDB1, HP1, TRIM28). Given that the authors have all the antibodies and primers, this should be an easy experiment.

2) Discussion: paragraph two, the authors claimed that "TRIM28 was still able to enrich CDK9 in the presence of RNase (data not shown)". This is an important piece of data; it should be shown, at least in the supplementary information.

Reviewer #2:

Ma et al. reports a mechanism by which TRIM28 inhibits the reactivation of HIV-1. Their primary finding is the ability of TRIM28 to SUMOylate CDK9, reducing the activity of CDK9 required for a functional pTEF-b complex. The authors primarily demonstrate this interaction utilizing high-resolution microscopy, proteomics, and mutagenesis in cell line models. Authors also demonstrate this functionality is relevant in vivo by knocking down TRIM28 and showing an increase in HIV-1 reactivation in patient cells. Overall this is a very interesting study using multiple approaches to test the function of TRIM28 in HIV-1 latency, which is novel and highly significant in the field. This manuscript could be strengthened with the following revisions.

Major suggestions

1) Western blot: In Figures 2F, 3A, 3D, 3E, 4C, 4D, Figure 4—figure supplement 1I, J, Figure 7—figure supplement 1B, and Figure 9—figure supplement 1, the authors performed immunoblot for multiple proteins on a single Western blot. The authors should demonstrate anti-HA, anti-Flag and anti-GAPDH antibody blots separately. For example, when GAPDH (detected by a rabbit polyclonal antibody) is shown in the same blot, along with anti-HA and anti-Flag (tagging SUMO4, UBC9 and TRIM28), it is hard to tease out specific detection of each target.

2) ATACseq: In Figure 5, the authored showed ATACseq density near the HIV-1 integration site. The curve does not look continuous on the left panel. Authors should note whether the HIV-1 genome itself was analyzed for accessibility, and if so how they differentiated between their viral vector and the cell line genomes in regions of homology (LTRs, PSI, RRE, etc.). Authors should also include a supplemental figure indicating similar accessibility at ectopic loci (such as the promoter of the gene in which HIV-1 is integrated, and a housekeeping gene) between the two samples (sgNT vs sgTRIM28) to demonstrate comparable transposition between samples.

3) HIV-1 genetic diversification and viral outgrowth: In Figure 6, the authors analyzed the genetic diversity of HIV-1 reactivated upon different shRNA knockdown and stimulation. The phylogenetic analysis showed a diversity (>100 clones) of cell-associated HIV-1 RNA. Since the authors used TOPO cloning of PCR products instead of direct sequencing of the bulk PCR product, the so called "diversity" of HIV-1 RNA is reflecting PCR errors identified in TOPO cloning, not real HIV diversity. The authors should only focus on HIV-1 RNA levels (Figure 7A) and remove Figure 7B.

4) Viral outgrowth: In Figure 7C, all viral outgrowth culture p24 readouts have to be plotted in log scale, not linear scale. The "viral outgrowth assay" shown is mainly a yes-no viral outgrowth instead of a "quantitative" viral outgrowth measurement, as all outgrowths are positive. This does not test the hypothesis whether TRIM28 affects latency reversal (unless it's quantitative with limiting dilution). The authors should remove Figure 7C.

Reviewer #3:

In this manuscript, Ma et al. identified TRIM28 as a negative regulator of HIV transcription. Mechanistically, the authors suggest that TRIM28 post-translationally modify CDK9 by SUMO4 through its SUMO E3 ligase activity. This modification reduces both kinase activity and binding of CDK9 to cyclin T1, an essential partner of the elongation factor pTEF-b.

In spite of the elegant biochemical analysis, there are some concerns regarding whether the results completely demonstrate their conclusions regarding the role of SUMO4 in regulating pTEFb.

– All the experimental observations that SUMO4 can modify CDK9 are only evaluated in the context of over-expression systems. It will be important to address whether this modification happens in primary CD4 T cells, the main latent reservoir, under endogenous expression of TRIM28, CDK9 and SUMO4. It will be also important to address whether SUMO4 is expressed in CD4 T cells in their RNASeq data.

– Figure 3 is misleading. SUMO has a Flag epitope but not WB against FLAG is done in any of the IP membranes to ensure that the bands marked as SUMO-CDK9 are actually SUMOylated CDK9.

–The activity of CDK9 is also controlled by phosphorylation. Does SUMOylation affect CDK9 phosphorylation?

– Based on Figure 4, there is not a strong co-localization between endogenous CDK9 and TRIM28 in 293T cells, suggesting that the interaction proposed may be an artifact of the over-expression system.

– co-IP experiments shown in Figure 5E do not demonstrate that SUMOylation of CDK9 reduces binding to Cyclin T1. The figure seems mislabeled in the IP section and no reduction on binding is observed when UBC9/TRIM28/SUMO4 are co-transfected, invalidating their proposed working model in Figure 7D.

– SUMO4 can also strongly modify TRIM28 independent of CDK9. Does SUMO4 modification of TRIM28 modify its activity?

– ATAC-seq reveals a more accessible chromatin around the HIV LTR. Is this particular of the LTR or is it a global alteration of other promoters? This will be important when addressing targeting TRIM28 as potential LRA as its targeting may have multiple pleiotropic effects.

– It is important to note that reduction of TRIM28 levels both in transformed cell model of latency as well as cells isolated from aviremic participants does seem to reactivate latent HIV, however whether this is through SUMO4-mediated modification of CDK9 by TRIM28 is not fully supported by the experimental data. Furthermore, it will be important to address what it is the toxicity associated with targeting TRIM28 as well as specificity to the HIV promoter.

https://doi.org/10.7554/eLife.42426.043

Author response

Essential revisions:

There are two points that were highlighted in the comments of at least two reviewers:

1) Please provide evidence for the importance of SUMO4. Does SUMO4 disruption alter the chromatin status of the HIV DNA? Please provide evidence that SUMO4 is important in the primary cells of interest; minimally you should demonstrate that SUMO4 is expressed in the cell type of interest. Without this additional data we feel that you need to remove claims about the importance of SUMO4.

We thank the editors and reviewers for bringing up the importance of SUMO4 and do apologize for missing this gap in our results, which we have now corrected. We would like to emphasize our revisions by the following points:

1) Within the section of “CDK9 Is SUMOylated by TRIM28”, we present a number of new chromatin immunoprecipitation data to indicate SUMO4 is the major HIV-1 latency contributor compared with the other SUMO paralogs. As we have shown in the newly-added fifteen figures in Figure 4D-M and Figure 4—figure supplement 1D-H, SUMO4 was enriched on HIV-1 LTR and mediated more than half of the enrichment of TRIM28 on HIV-1 LTR. This indicated that the enrichment of TRIM28 on HIV-1 LTR may be partly SUMOylation-dependent in addition to the Krüppel-associated box domain zinc fingers (KRAB-ZNFs)–dependent protein binding. Importantly, the absence of SUMO4 significantly reduced the enrichment of SETDB1, HP1α and HDAC1 on HIV-1 LTR, as well as the suppressive marks on HIV-1 LTR, such as H3K9me, H3K9me2 and H3K9me3. These data are consistent with previous findings that TRIM28 recruited SETDB1, HP1α and NuRD complex (the major component of which is HDAC1) in a SUMOylation-dependent manner. Correspondingly, the activation marks including H3K4me3 and H3K9acetyl were also increased upon SUMO4 knockdown. Intriguingly, we also noticed the decrease of H3K27me3 on HIV-1 LTR upon SUMO4 knockdown, which was compatible with our finding that the depletion of TRIM28 induced the decrease of H3K27me3 on HIV-1 promoter. Whether TRIM28 utilizes SUMO4 to influence the function of H3K27me3 mediators such as polycomb repressive complex 2 (PRC2) components EZH2 and SUZ12 is an interesting point, which deserves further investigation. However, the epigenetic roles of TRIM28 and SUMO4 are not the major points of our manuscript, which focuses on transcriptional regulation. Therefore, we have not profoundly extended on this area.

2) Within the section of “TRIM28 Suppresses HIV-1 Expression and Contributes to HIV-1 Latency” and “CDK9 Is SUMOylated by TRIM28”, we utilized four distinct siRNAs targeting the coding sequence and 3’UTR of TRIM28 or SUMO4 mRNAs to downregulate TRIM28 or SUMO4, which significantly upregulated HIV-1 promoter activity respectively, especially in combination with HIV-1 transactivator Tat. These results have been shown in four newly-added Figure 1—figure supplement 1A, Figure 1—figure supplement 1D and Figure 4A-B. To further demonstrate the importance of SUMO4 in primary CD4+ T cells, we firstly compared the expression of SUMO4 in different cells. We found that SUMO4 was ubiquitously expressed in several cell lines and primary CD4+ cells (newly-added Figure 4—figure supplement 1B). Besides, we also indicated the expression of SUMO4 in the volcanoplot of RNA-Seq data in CD4+ T cells which was shown in newly-added Figure 1—figure supplement 1G. The expression of SUMO4 mRNA was quantitated within unstimulated, PHA-stimulated and resting CD4+ T cells as we have done for the expression of TRIM28 mRNA (newly-added Figure 4—figure supplement 1C). Finally, we tested whether the depletion of SUMO4 could reactivate latent HIV-1 in resting CD4+ T cells isolated from HIV-1-infected individuals. The newly-added Figure 10—figure supplement 4 indicated that the depletion of SUMO4 reactivated substantial productions of HIV-1 RNAs which were even slightly higher than those activated by SAHA. The combination use of SUMO4 knockdown and SAHA addition could reactivate more HIV-1 RNAs than those reactivated by them separately. The result was consistent with that caused by TRIM28 depletion. Moreover, through immunoblotting the endogenous SUMO4, we confirmed that SUMOylation of cellular targets with SUMO4 are ubiquitous in primary CD4+ T cells, the result of which has been shown in newly-added Figure 5—figure supplement 1C. We have been trying very hard to monitor the endogenous SUMOylation of CDK9 in primary CD4+ T cells, Jurkat cell line, HeLa cell line and HEK293T cell line. However, we were unable to identify significant bands of SUMOylated CDK9. Instead, we only able to immunoblot a small portion of SUMOylated CDK9, which is in consistence with the other, previously reported SUMOylated substrates. Nevertheless, we conducted semi-endogenous SUMOylation assay. We overexpressed TRIM28, UBC9 and SUMO4 in primary CD4+ T cells, and immunoblotted the endogenous CDK9. The result showed that the endogenous CDK9 was also SUMOylated in the presence of exogenously expressed SUMOylation system components (newly-added Figure 5—figure supplement 1D-E).

2) Please clarify whether or not the ATAC-Seq effect that you report is specific for the HIV LTR. Does this change in chromatin state occur with promoters more generally?

TRIM28-mediated the increase of ATAC-Seq tag density is not specific for the HIV-1 LTR. The chromatin accessibilities of many TRIM28-regulated genes were also increased (newly-added Figure 8—figure supplement 1-2). This result is also consistent with the data shown in public database, which indicate that TRIM28 regulates lots of genes involved in cellular differentiation, DNA damage repairing, as well as the suppression of human cytomegalovirus (HCMV) and other human endogenous retroviruses (HERVs) in stem cells. The phenomenon that TRIM28 bound to and regulated HERVs was also found in human CD4+ T cells (Turelli et al., 2014, Genome Research, PMID: 24879559). Interestingly, we found that the tag densities of many corepressors of TRIM28, especially zinc finger proteins (ZNFs), were also significantly changed. This result is also consistent with a recently published paper showing that the knockout of TRIM28 induced the overexpression of several ZNFs (Tie et al., 2018, EMBO Reports, PMID: 30061100). Besides, we conducted several functional analysis and found that most genes which had upregulated ATAC-Seq density upon TRIM28 depletion were functional proteins with binding activity, catalytic activity, nucleic acid binding transcription factor activity and protein binding transcription factor activity. Few genes belonged to structural genes or housekeeping genes. More than forty percent of ATAC-Seq peaks lied in gene promoters. Forty-nine percent of ATAC-Seq peaks lied in distal intergenic regions that were enriched with HERVs and distal regulation elements. The increased accessibility enhanced the corresponding promoter activity, such as those transcription factor promoters.

It is notable that Figure 8A-B indicates that the HIV-1 promoter activity was inhibited by TRIM28 by SUMOylating CDK9. We are fully aware of the risk of pleiotropic effects caused by TRIM28 depletion. TRIM28 has long been identified as a multifunctional protein involving in transcriptional regulation, cellular differentiation and proliferation, DNA damage repair, viral suppression, and apoptosis. Also as we showed here, the depletion of TRIM28 could increase the chromatin accessibility of many functional genes. Some genes could even enhance the anti-HIV-1 activity. However, as far as we known, all of LRAs tested do not specifically target HIV-1. SAHA, the widely tested in pilot clinical trials to date, targets histone deacetylase (HDAC). JQ-1 targets the Bromodomain and Extra-Terminal (BET) family of bromodomain proteins. Disulfiram depletes the intracellular protein PTEN and consequently activates the Akt signaling pathway. Bryostatin-1, the PKC agonist, directly induces T cell activation. More LRAs and corresponding side effects have been well elucidated in a review paper published on the Annual Review of Medicine by Spivak and Planelles (PMID: 29099677). To test the toxicities associated with depleting TRIM28, we have conducted several toxicity experiments which included cytotoxicity assay, cell viability assay, cell number counting and cell proliferation assay in Jurkat cells, HeLa cells as well as resting CD4+ T cells isolated from aviremic participants. The newly-added Figure 10—figure supplement 1-2 indicates that the cytotoxicity, viability, cell number and cell proliferation abilities were not affected upon TRIM28 knockdown or knockout. Therefore, we propose that TRIM28 is a safe target for developing new LRAs.

Reviewer #1:

[…] Major concerns:

1) As shown in Figure 1, TRIM28 is responsible for the repressive epigenetic marks deposited on HIV-1 DNA. Is the E3 activity of TRIM28 important for these repressive epigenetic modifications? The authors can knockdown SUMO4 (the major SUMO used by TRIM28), maybe also CDK9, and examine the repressive epigenetic marks on HIV-1 DNA (with ChIP antibodies H3K9me3, H3K27me3, H3ac, SETDB1, HP1, TRIM28). Given that the authors have all the antibodies and primers, this should be an easy experiment.

We thank the reviewer for the insightful comment and agree that we should provide several lines of evidence to show the influence of SUMOylation on repressive epigenetic modifications. As the reviewer kindly suggested, we addressed this vital point in two major aspects as shown below.

1) We provided new evidence that the E3 activity of TRIM28 was important for the repressive epigenetic modifications. For a clear logic flow, we have summarized the results within the section of “TRIM28 SUMOylates Many Transcription Factors and Transferases” as Figure 2C-E. In this part, we have proved that both the intermolecular SUMOylation domain RING and intramolecular SUMOylation domain PHD of TRIM28 meditated the suppression of HIV-1 LTR. To test whether these two domains influenced the epigenetic status of HIV-1 LTR, firstly we knocked down the endogenous TRIM28 with siRNA targeting the 3’UTR of TRIM28 mRNA. Then we overexpressed the wild type TRIM28 construct and TRIM28 mutants without RING or PHD domain, respectively. The mRNAs expressed by these constructs were able to resist the degradation because the siRNA we used here targeting the 3’UTR of TRIM28 mRNA and the TRIM28 constructs we transfected here only expressed the coding sequences. Further, we investigated the epigenetic status of HIV-1 LTR among these different groups. We found that the absence of endogenous TRIM28 resulted in the decrease of suppressive epigenetic marks H3K9me3 and H3K27me3, as well as the increase of active epigenetic marks H3K9acetyl. The addition of exogenous wild type TRIM28 was able to rescue the suppressive epigenetic marks. However, the mutants without RING or PHD domain were only able to rescue the partial suppressive epigenetic marks, which indicated that both SUMOylation domains participated in the epigenetic regulation of HIV-1 LTR. To fully elucidate the detailed mechanisms behind, we conducted global site-specific SUMO-MS to identify the SUMOylation substrates as we have done within the section of “TRIM28 SUMOylates Many Transcription Factors and Transferases”.

2) We provided new evidence that SUMO4 increases the repressive epigenetic marks on HIV-1 LTR. We utilized several siRNAs targeting three coding regions and one 3’UTR region of TRIM28 or SUMO4 mRNA respectively and confirmed that SUMO4 suppressed the HIV-1 LTR activity (newly-added Figure 1—figure supplement 1A and Figure 4A). Then, we examined the suppressive epigenetic marks on HIV-1 LTR upon SUMO4 knockdown. We firstly examined the enrichment of TRIM28 on HIV-1 LTR when we knocked down SUMO4. We found that more than half of TRIM28 was lost from HIV-1 LTR upon SUMO4 knockdown, which indicated that the enrichment of TRIM28 on HIV-1 LTR may be partially SUMOylation-dependent apart from the Krüppel-associated box domain zinc fingers (KRAB-ZNFs)–dependent binding. H3K9me, H3K9me2 and H3K9me3 were significantly decreased on HIV-1 LTR in the absence of SUMO4, as well as the H3K9 methylation “writer” SETDB1 and “reader” HP1α. Moreover, we observed significant upregulation of H3K9acetyl and H3K4me3 and downregulation of HDAC1, which was consistent with previous reports that TRIM28 recruited SETDB1, HP1α and HDAC1 in a SUMOylation-dependent manner (Iyengar and Farnham, 2011). Besides, the H3K27me3 was also decreased on HIV-1 LTR upon SUMO4 knockdown. We suspected that some polycomb repressive complex 2 (PRC2) components such as EZH2 and SUZ12, the major “writers” of H3K27me3, may be SUMOylated by SUMO4, resulting in the enhancement of modifier function. However, we feel that the interactions between TRIM28 and SUMO4 with other epigenetic modifiers are beyond our scope. Therefore, we did not conduct further research on this area. The above results have been shown in several newly-added Figure 4D-M and Figure 4—figure supplement 1D-H. The reviewer also suggested that we might knock down CDK9 to examine the repressive epigenetic marks on HIV-1 DNA. However, we found that the knockdown of CDK9 was very toxic to the cells (data not shown). Besides, we feel that the SUMO4-mediated epigenetic marks change belongs to the epigenetic regulation of HIV-1 latency, while the SUMO4-mediated CDK9 SUMOylation belongs to the transcriptional regulation of HIV-1 latency. Similarly, the possible CDK9-meditated epigenetic modifications are also beyond the scope of our study.

2) Discussion: paragraph two, the authors claimed that "TRIM28 was still able to enrich CDK9 in the presence of RNase (data not shown)". This is an important piece of data; it should be shown, at least in the supplementary information.

We apologize for not showing it in the original submission. We have now shown it in newly-added Figure 7—figure supplement 1A, which indicates that TRIM28 was able to enrich CDK9 even in the presence of RNase.

Reviewer #2:

[…] Major suggestions

1) Western blot: In Figures 2F, 3A, 3D, 3E, 4C, 4D, Figure 4—figure supplement 1I, J, Figure 7—figure supplement 1B, and Figure 9—figure supplement 1, the authors performed immunoblot for multiple proteins on a single Western blot. The authors should demonstrate anti-HA, anti-Flag and anti-GAPDH antibody blots separately. For example, when GAPDH (detected by a rabbit polyclonal antibody) is shown in the same blot, along with anti-HA and anti-Flag (tagging SUMO4, UBC9 and TRIM28), it is hard to tease out specific detection of each target.

Our apologies for the confusion. We have separated each band of the target proteins in our revised manuscript.

2) ATACseq: In Figure 5, the authored showed ATACseq density near the HIV-1 integration site. The curve does not look continuous on the left panel. Authors should note whether the HIV-1 genome itself was analyzed for accessibility, and if so how they differentiated between their viral vector and the cell line genomes in regions of homology (LTRs, PSI, RRE,etc.). Authors should also include a supplemental figure indicating similar accessibility at ectopic loci (such as the promoter of the gene in which HIV-1 is integrated, and a housekeeping gene) between the two samples (sgNT vs sgTRIM28) to demonstrate comparable transposition between samples.

We apologize for the incomplete explanation of the analysis strategy for ATAC-Seq data, which we have carefully modified in the Materials and methods section of “RNA-Seq and ATAC-Seq”.

1) Yes, the HIV-1 genome itself was analyzed for accessibility. The reads were aligned to HIV-1 reference genome K03455, M38432 (Version K03455.1) by Bowtie2, followed by rearranging with Samtools. The human genome was aligned to human reference genome GRCh38. The integration sites of HIV-1 pseudotyped viruses in J-Lat 10.6 and TZM-bl have been identified by genome walking strategy. Thus, the read density centered HIV-1 5’LTR could be calculated and combined together. The density curves were indeed continuous. However, because we only showed the normalized tag densities of two kilobases range centered HIV-1 LTR, which was also one of the peak center of HIV-1, the other tag densities outside two kilobases have been truncated. LTR, PSI, and RRE sequences of integrated pseudotyped HIV-1 proviruses are different from those of human endogenous retrovirus (HERV) although they share very high homology. We still could distinguish them when we aligned the reads to HIV-1 reference genome or human reference genome.

2) The reviewer also kindly suggested us to include a supplemental figure indicating similar accessibility at ectopic loci. We now have made newly-added Figure 8—figure supplement 2C-F to show the accessibilities at the promoters of the genes which the pseudotyped HIV-1 proviruses are integrated as well as at the promoters of housekeeping gene GAPDH. The results showed that the chromatin accessibilities of these genes were similar between wild type cells or TRIM28 depletion cells.

3) HIV-1 genetic diversification and viral outgrowth: In Figure 6, the authors analyzed the genetic diversity of HIV-1 reactivated upon different shRNA knockdown and stimulation. The phylogenetic analysis showed a diversity (>100 clones) of cell-associated HIV-1 RNA. Since the authors used TOPO cloning of PCR products instead of direct sequencing of the bulk PCR product, the so called "diversity" of HIV-1 RNA is reflecting PCR errors identified in TOPO cloning, not real HIV diversity. The authors should only focus on HIV-1 RNA levels (Figure 7A) and remove Figure 7B.

We thank the reviewer for the suggestion. However, we still think that the HIV-1 genetic diversity assay is vital for supporting our hypothesis that TRIM28 could be the target for developing dual functional LRAs releasing both epigenetic and transcriptional restrictions. We hope to keep Figure 7B (Figure 10B in revised manuscript).

The reviewer concerns that the PCR errors which were introduced by TOPO cloning might overestimate the HIV-1 diversity. However, the PCR/cloning method which we used was not TOPO cloning. We apologize for having omitted some essential experimental description, which we have added in the revised manuscript. The PCR/cloning method, which we used to ligate the PCR products, was a little different from the standard TOPO cloning which used Taq DNA polymerase to amplify DNA fragments. For each PCR reaction, we firstly used Phanta Max Super-Fidelity DNA Polymerase (Vazyme) to amplify the V1-V3 region of HIV-1 envelope in order to ensure the fidelity. The amplification error rate of Phanta Max is 53-fold lower than that of Taq and 6-fold lower than that of Pfu according to the manufacturer’s instruction. After two rounds of nested PCR utilizing Phanta Max, the PCR products were proceeded to deoxyadenosine (A)-tailing at the 3'-end of the PCR products utilizing Ex Taq DNA polymerase (Takara) without thermal cycling as follows: 95℃, 5 min; 72℃, 30 min; 4℃ hold. The A-tailed PCR products were TA-ligated into pMD-18T vector. To minimize the sampling bias, we also obtained 30 independent PCR products for each sample. We also picked at least 60 single clones for each PCR products to detect low frequency mutations. Although some groups used single genome sequencing (SGS) method to analyze the HIV-1 genetic diversity, we found that we could hardly get enough PCR amplicons when we serially diluted the cDNA due to the low viral cDNA. The HIV-1 genetic diversities in infected individuals might be underestimated if we used SGS. We also carefully referred to papers comparing these two methods. Based on the paper of Michael R. Jordan et al., 2010, Journal of Virological Methods (PMID: 20451557), both PCR/cloning and SGS measures intra-patient HIV-1 genetic diversity similarly. The intrapopulation average pairwise distance (APD), the sampling bias and the systematic position specific bias, which are measured by each method, show no differences.

The reason we suggested to keep Figure 7B is that the HIV-1 genetic diversity assay is a method firstly described by our group to assess the quality of LRAs and corresponding cellular targets (Geng et al., 2016). We propose that the viral production only is unable to reflect the true size of the HIV-1 latent reservoir. Although some LRAs such as PMA/ionomycin could activate substantial HIV-1 RNA, the genetic diversities of these activated viral RNAs were quite low. On contrary, an attenuated Tat Protein, Tat-R5M4 which was screened out by our lab, could reactivate more genetically-diversified HIV-1 than PMA/ionomycin, although the amount of reactivated viral RNA was lower than that reactivated by PMA/ionomycin. Thus, we have proposed that different LRAs reactivate different amounts of genetically-diversified HIV-1 (Author response image 1).

Author response image 1
Attenuated Tat Protein Tat-R5M4 can reactivate more genetically-diversified HIV-1.

(Geng et al., 2016, Molecular Therapy, PMID: 27434587).

The underlying rationale is that there are viral quasispecies within an HIV-1-infected individual. After HIV-1 replicates for months or years, the genetically-diversified viruses could persistently convert into integrated proviral DNA in the CD4+ T cells and some of them become latent viral reservoir. For such a genetically-diversified viral reservoir, some LRAs such as PMA/ionomycin could reactivate more HIV-1 RNA production but with lower genetic diversity, indicating that only a few integrated proviruses are activated and the viral RNA are generated from these small amount of integrated viral DNA merely with a high transcriptional efficiency. Some LRAs such as Tat-R5M4 or SAHA can reactivate more genetically-diversified HIV-1, even though with a lower amount. Given that the major purpose of LRA is to expose more latently-infected CD4 cells to immunosurveillance, it is quite important that proviruses at more integration sites in more latently-infected cells are activated. Therefore, we believe that genetic diversity of activated viral RNA would be an important biomarker to access the quality of LRAs (Author response image 2). Downregulation of TRIM28 result in the activation of more genetically-diversified RNA, suggesting that it can activate more integrated proviruses and merits being an ideal target for LRA development.

Author response image 2
Different LRAs reactivate genetically-diversified HIV-1 at different integration sites.

4) Viral outgrowth: In Figure 7C, all viral outgrowth culture p24 readouts have to be plotted in log scale, not linear scale. The "viral outgrowth assay" shown is mainly a yes-no viral outgrowth instead of a "quantitative" viral outgrowth measurement, as all outgrowths are positive. This does not test the hypothesis whether TRIM28 affects latency reversal (unless it's quantitative with limiting dilution). The authors should remove Figure 7C.

We thank the reviewer for the suggestion. We have modified Figure 7C (Figure 10C in revised manuscript) by showing the viral outgrowth culture p24 readouts in log scale instead of linear scale. The reviewer also suggested to remove Figure 7C. However, we hope to keep this figure, which is also very important to support our hypothesis. Our purpose to conduct this experiment is not to measure the size of viral latent reservoir but determine the viruses we reactivated are the replication-competent rather than the dead viruses. The accumulating production of p24 indicated the ongoing HIV-1 replication. The time course study would measure the kinetics of viral replication and therefore reflect the viral infectivity, which is much better than the yes-no viral outgrowth. Our methodology to determine replication-competent HIV-1 between different experimental conditions has been well-established and repeated in several of our published papers (Huang et al., 2007; Li et al., 2016; Geng et al., 2016; Liu et al., 2016) and many works from other HIV-1 latency labs as well. Besides, although standard QVOA could be better to test our hypothesis, we are greatly limited by the shortage of large samples (more than 180 mL of blood for a single experimental group) from the study participants. Moreover, viral reservoir measured with VOA merely stands for a quite small amount of latent viral reservoir (Ho et al., 2013).

Reviewer #3:

In this manuscript, Ma et al. identified TRIM28 as a negative regulator of HIV transcription. […] In spite of the elegant biochemical analysis, there are some concerns regarding whether the results completely demonstrate their conclusions regarding the role of SUMO4 in regulating pTEFb.

We thank the reviewer for supporting this study. We have addressed each comment according to the reviewer’s suggestion. Especially, we further elucidated the role of SUMO4 on HIV-1 latency.

– All the experimental observations that SUMO4 can modify CDK9 are only evaluated in the context of over-expression systems. It will be important to address whether this modification happens in primary CD4 T cells, the main latent reservoir, under endogenous expression of TRIM28, CDK9 and SUMO4. It will be also important to address whether SUMO4 is expressed in CD4 T cells in their RNASeq data.

We thank the reviewer for pointing out these absences. As we have explained in the Discussion, the percentage of SUMOylated CDK9 is only a small proportion, less than 5%. The phenomenon is also observed for most of the previously identified SUMOylation targets. Through immunoblotting the endogenous SUMO4, we confirmed that SUMOylation of cellular targets with SUMO4 are ubiquitous in primary CD4+ T cells, the result of which has been shown in newly-added Figure 5—figure supplement 1C. We have been trying very hard to monitor the endogenous SUMOylation of CDK9 in primary CD4+ T cells, Jurkat cell line, HeLa cell line and HEK293T cell line. However, we were unable to identify significant bands of SUMOylated CDK9. Instead, we only able to immunoblot a small portion of SUMOylated CDK9, which is in consistence with the other, previously reported SUMOylated substrates (newly-added Figure 5—figure supplement 1D). Nevertheless, we conducted semi-endogenous SUMOylation assay. We overexpressed TRIM28, UBC9 and SUMO4 in primary CD4+ T cells, and immunoblotted the endogenous CDK9. The result showed that the endogenous CDK9 was also SUMOylated in the presence of exogenously expressed SUMOylation system components, the result of which has been shown in newly-added Figure 5—figure supplement 1E.

To confirm that SUMO4 is expressed in the cells we used, we compared the expression of SUMO4 in different cells to show that SUMO4 is ubiquitously expressed in several cell lines and primary cells (newly-added Figure 4—figure supplement 1B). We also indicated the expression of SUMO4 in the volcanoplot of RNA-Seq data which we showed in Figure 1—figure supplement 1G. Besides, we quantitated the expression of SUMO4 mRNA within unstimulated, PHA-stimulated and memory CD4+ T cells as we have conducted for the expression of TRIM28 mRNA, the result of which has been shown in newly-added Figure 4—figure supplement 1C.

– Figure 3 is misleading. SUMO has a Flag epitope but not WB against FLAG is done in any of the IP membranes to ensure that the bands marked as SUMO-CDK9 are actually SUMOylated CDK9.

Our apologies for the confusion. All the IP membranes were actually immunoblotted with both antibodies against HA and Flag. We missed the statement of “IB: Flag” on the left side of the figures. The SUMO-CDK9 represented the results of SUMOylated CDK9 with both epitopes. In order to present the results more precisely, we have separated the anti-HA and anti-Flag blots in our revised manuscript.

–The activity of CDK9 is also controlled by phosphorylation. Does SUMOylation affect CDK9 phosphorylation?

This is an interesting point raised by the reviewer. Although we did not conduct systematic experiments to elucidate the possibility of the effect of SUMOylation on CDK9 phosphorylation, we expect that SUMOylation could affect CDK9 phosphorylation. In our effort to locate all the SUMOylation sites on CDK9, we surprisingly found that several lysines on CDK9 were SUMOylated. Among them, multiple SUMOylation sites: Lys274, Lys276, Lys280, Lys325 and Lys345 were adjacent to CDK9 C-terminal autophosphorylation sites which have been reported to be required for high-affinity binding of Tat–P-TEFb to TAR RNA (Baumli et al., 2008; Garber et al., 2000). SUMOylation may decrease the binding ability by preventing the neighboring phosphorylation. However, in our further effort to identify which sites were indeed SUMOylated by TRIM28, we found that only the Lys44, Lys56 and Lys68 residues were specifically SUMOylated by TRIM28. The SUMOylation sites adjacent to CDK9 C-terminal autophosphorylation sites were not SUMOylated by TRIM28. Other CDK9 SUMOylation E3 ligases may exist to mediate their SUMOylation. Because other E3 ligases are out of our research scope, we have not conduct further experiments to study the effect of SUMOylation on CDK9 phosphorylation. The reviewer’s question is really helpful for the comprehensiveness of our work. Thus, we have added the following statements: “Among them, multiple SUMOylation sites were adjacent to CDK9 C-terminal autophosphorylation sites which have been reported to be required for high-affinity binding of Tat–P-TEFb to TAR RNA (Baumli et al., 2008; Garber et al., 2000). SUMOylation may decrease the binding ability by preventing the neighboring phosphorylation.”

– Based on Figure 4, there is not a strong co-localization between endogenous CDK9 and TRIM28 in 293T cells, suggesting that the interaction proposed may be an artifact of the over-expression system.

We apologize for not explaining the imaging figures clearly. The results we presented in Figure 4A and 4B (Figure 6A-B in revised manuscript) are the cSTORM images of endogenous TRIM28 with endogenous SUMO4 and CDK9. We did not overexpress any of them. The single molecule localization is obtained by Gaussian fitting. The co-localization of cSTORM is measured by the center of two points, which is slightly different from confocal, which measures the overlapping of two different protein clusters. From the amplified view, 3D-cSTORM and related videos, we easily found that dotted SUMO4 and CDK9 proteins were enriched by TRIM28 and shaped big spots. In Figure 4E (Figure 7C in revised manuscript) which was SIM image of overexpressed system, we used exogenously expressed GFP-tagged wild type TRIM28, GFP-tagged TRIM28 mutant and RFP-tagged CDK9 to investigate the co-localization due to that we have not constructed the endogenously expressed TRIM28 mutant so far. We still can easily notice the significant co-localization of GFP-TRIM28 and RFP-CDK9. We have conducted two statistical analyses, shown in the newly-added Figure 6E and Figure 7D in our revised manuscript. The quantitation strategies were elucidated within the method of “SIM and STORM imaging”. (Figure 4 has been made two newly-added Figure 6 and Figure 7.)

– co-IP experiments shown in Figure 5E do not demonstrate that SUMOylation of CDK9 reduces binding to Cyclin T1. The figure seems mislabeled in the IP section and no reduction on binding is observed when UBC9/TRIM28/SUMO4 are co-transfected, invalidating their proposed working model in Figure 7D.

We apologize for mislabeling and the lack of proper explanation. We have specifically labeled “SUMO-CDK9” which represented SUMOylated CDK9 in the upper part of Figure 5E (Figure 8E in revised manuscript). From the IP section of the result, we can easily notice the significant reduction of SUMOylated CDK9, which was not enriched by Cyclin T1. Instead, the wild type CDK9 was able to bind to Cyclin T1 was unchanged. Because the system we used here was the exogenous overexpression, it is very difficult to observe the reduction of wild type CDK9, although some CDK9 has been SUMOylated.

– SUMO4 can also strongly modify TRIM28 independent of CDK9. Does SUMO4 modification of TRIM28 modify its activity?

This is an interesting question. We did find that the function of TRIM28 was influenced by SUMO4 modification of TRIM28. The PHD domain of TRIM28 can mediate the intramolecular SUMOylation of TRIM28. The RING domain of TRIM28 is not only able to mediate the intermolecular SUMOylation of other proteins (Figure 3D), but also be able to mediate the intramolecular SUMOylation of TRIM28 (Figure 7—figure supplement 1B). As we have shown in Figure 7—figure supplement 1B, TRIM28 was strongly self-SUMOylated in the presence of SUMO4 and UBC9.

When we knocked down SUMO4 in the TZM-bl cell line, we found that more than half of TRIM28 was lost from HIV-1 LTR, which indicated that the enrichment of TRIM28 on HIV-1 LTR was partially SUMOylation-dependent, which is apart from the Krüppel-associated box domain zinc fingers (KRAB-ZNFs)–dependent binding. Some other HIV-1 LTR binding proteins may harbor SUMO-interacting motifs (SIM) mediate the enrichment of SUMOylated TRIM28. The result mentioned above has been shown in newly-added Figure 4D.

Furthermore, we found that the H3K9 methylation “writer” SETDB1 and “reader” HP1α were significantly decreased on HIV-1 LTR in the absence of SUMO4. Moreover, we observed a significant upregulation of H3K9acetyl and H3K4me3 and a downregulation of HDAC1. These results were in consistence with previous reports that TRIM28 recruited SETDB1, HP1α and HDAC1 in a SUMOylation-dependent manner. The result mentioned above has been shown in newly-added Figure 4E-M.

– ATAC-seq reveals a more accessible chromatin around the HIV LTR. Is this particular of the LTR or is it a global alteration of other promoters? This will be important when addressing targeting TRIM28 as potential LRA as its’ targeting may have multiple pleiotropic effects.

We thank the reviewer for the vital concern. The specificity of TRIM28 to HIV-1 LTR is not our study priority, however, the concern of which is very important when addressing targeting TRIM28 as potential LRA. The phenomenon that TRIM28 contributes to HIV-1 latency and the mechanism that TRIM28 SUMOylates CDK9 to mediate transcriptional control are two key findings of our study and we wish that TRIM28 could be a candidate target to develop LRAs. As we have carefully addressed in Essential revision 2 to editors and reviewers, we would like to comprehensively elucidate the question from three points as shown below:

1) TRIM28-mediated the increase of ATAC-Seq tag density is not specific for the HIV-1 LTR. The chromatin accessibilities of many TRIM28-regulated genes were also increased. The newly-added Figure 8—figure supplement 1-2 show the above result. This result was also consistent with the data shown in public database, which indicate that TRIM28 regulated lots of genes involved in cellular differentiation, DNA damage repairing, as well as the suppression of human cytomegalovirus (HCMV) and other human endogenous retroviruses (HERVs) in stem cells. The phenomenon that TRIM28 bound to and regulated HERVs was also found in human CD4+ T cells (Turelli et al., 2014, Genome Research, PMID: 24879559). Interestingly, we found that the tag densities of many corepressors of TRIM28, especially zinc finger proteins (ZNFs), were also significantly changed, the result of which was also consistent with a recently published paper showing that the knockout of TRIM28 induced the overexpression of several ZNFs (Tie et al., 2018, EMBO Reports, PMID 30061100). They proposed a model that the depletion of TRIM28 could reactivate some HERVs and ZNFs.

2) Besides, we conducted several functional analysis and found that most genes which had upregulated ATAC-Seq density upon TRIM28 depletion were functional proteins with binding activity, catalytic activity, nucleic acid binding transcription factor activity and protein binding transcription factor activity. Few genes belonged to structural genes or housekeeping genes. More than forty percent of ATAC-Seq peaks lied in gene promoters. Forty-nine percent of ATAC-Seq peaks lied in distal intergenic regions that were enriched with HERVs and distal regulation elements. The increased accessibility enhanced the corresponding promoter activity, such as those transcription factor promoters. Thus, these transcription factors promoted the HIV-1 reactivation by enhancing transcription.

3) The result we showed in Figure 5A-B (Figure 8A-B in revised manuscript) is to prove that the HIV-1 promoter activity was inhibited by TRIM28 through SUMOylating CDK9. We were fully aware of the risk of pleiotropic effects caused by TRIM28 depletion. TRIM28 has long been identified as a multifunctional protein involving in transcriptional regulation, cellular differentiation and proliferation, DNA damage repair, viral suppression, and apoptosis. Also as we showed here, the depletion of TRIM28 could increase the chromatin accessibility of many functional genes. Some genes could enhance the anti-HIV-1 activity. However, as far as we have known, all of LRAs tested so far do not specifically target HIV-1. SAHA, the widely tested in pilot clinical trials to date, targets histone deacetylase (HDAC). JQ-1 targets the Bromodomain and Extra-Terminal (BET) family of bromodomain proteins. Disulfiram depletes the intracellular protein PTEN, resulting in activating the Akt signaling pathway. Bryostatin-1, the PKC agonist, directly induces T cell activation. More LRAs and corresponding side effects have been well elucidated in a paper published on the Annual Review of Medicine by Spivak and Planelles (PMID: 29099677).

– It is important to note that reduction of TRIM28 levels both in transformed cell model of latency as well as cells isolated from aviremic participants does seem to reactivate latent HIV, however whether this is through SUMO4-mediated modification of CDK9 by TRIM28 is not fully supported by the experimental data. Furthermore, it will be important to address what it is the toxicity associated with targeting TRIM28 as well as specificity to the HIV promoter.

Thank you for pointing out these points of confusion in the original manuscript. If we understand correctly, this comment addressed three different issues. The first issue is the concern of the contribution of the SUMO4-mediated modification of CDK9 by TRIM28 on HIV-1 latency. The second issue is to address the toxicity associated with targeting TRIM28. The third issue is the specificity of targeting TRIM28 to the HIV-1 promoter.

1) We firstly would like to address that SUMO4-mediated modification of CDK9 by TRIM28 is one of the mechanisms used by TRIM28 to contribute to HIV-1 latency in cells isolated from aviremic participants. As we have carefully elucidated in Essential revision 1 to editors and reviewers, within the section of “TRIM28 Suppresses HIV-1 Expression and Contributes to HIV-1 Latency”, we utilized four distinct siRNAs targeting the coding sequence and 3’UTR of TRIM28 and SUMO4 mRNAs to downregulate TRIM28 and SUMO4, which significantly upregulated HIV-1 promoter activity, especially in combination with HIV-1 transactivator Tat. These results have been shown in four newly-added Figure 1—figure supplement 1A, Figure 1—figure supplement 1D and Figure 4A-B. To further demonstrate the importance of SUMO4 in primary CD4+ T cells, we firstly compared the expression of SUMO4 in different cells. We found that SUMO4 was ubiquitously overexpressed in several cell lines and primary CD4+ cells (newly-added Figure 4—figure supplement 1B). Besides, we also indicated the expression of SUMO4 in the volcanoplot of RNA-Seq data in CD4+ T cells shown in newly-added Figure 1—figure supplement 1G. The expression of SUMO4 mRNA was quantitated within unstimulated, PHA-stimulated and memory CD4+ T cells as we have done for the expression of TRIM28 mRNA (newly-added Figure 4—figure supplement 1C). Finally, we tested whether the depletion of SUMO4 could reactivate latent HIV-1 in resting CD4+ T cells isolated from HIV-1-infected individuals. The newly-added Figure 10—figure supplement 4 indicated that the depletion of SUMO4 reactivated substantial productions of HIV-1 RNAs which were even slightly higher than those activated by SAHA. The combination use of SUMO4 knockdown and SAHA addition could reactivate more HIV-1 RNAs than those reactivated by them separately. The result was consistent with that caused by TRIM28 depletion. Moreover, through immunoblotting the endogenous SUMO4, we confirmed that SUMOylation of cellular targets with SUMO4 are ubiquitous in primary CD4+ T cells, the result of which has been shown in the newly-added Figure 5—figure supplement 1C. We have been trying very hard to monitor the endogenous SUMOylation of CDK9 in primary CD4+ T cells, Jurkat cell line, HeLa cell line and HEK293T cell line. However, we were unable to identify significant bands of SUMOylated CDK9. Instead, we only able to immunoblot a small portion of SUMOylated CDK9, which is in consistence with the other, previously reported SUMOylated substrates (newly-added Figure 5—figure supplement 1D). Nevertheless, we conducted semi-endogenous SUMOylation assay. We overexpressed TRIM28, UBC9 and SUMO4 in primary CD4+ T cells, and immunoblotted the endogenous CDK9. The result showed that the endogenous CDK9 was also SUMOylated in the presence of exogenously expressed SUMOylation system components, the result of which has been shown in newly-added Figure 5—figure supplement 1E.

2) To address the possible toxicities associated with targeting TRIM28, we conducted several experiments including cytotoxicity assay, cell viability assay, cell number counting and cell proliferation assay. We used siRNAs to knock down TRIM28 in HeLa cells and the resting CD4+ T cells isolated from aviremic participants. We also used shRNA constructs to knock down TRIM28 and sgRNA constructs to knock out TRIM28 in Jurkat cells. The cytotoxicity was measured by comparing the amounts of dehydrogenases between wild type cells and TRIM28 knockdown or knockout cells. The cell viability was measured by comparing the percentages of live cells between wild type cells and TRIM28 deficiency cells. The cell numbers were measured by counting cells every days upon TRIM28 knockdown or knockout. The cell proliferation abilities were measured by CFSE staining. The results have been shown in newly-added Figure 10—figure supplement 1-2. We found that upon TRIM28 knockdown, the cytotoxicity, viability, cell number, and cell proliferation abilities were not influenced compared with wild type cells. However, previous reports found that targeting HDACs with SAHA had some toxicity to cell viability, although the toxicities might come from the side effects of LRAs. Our finding here provided another safe target to develop new LRAs, which was TRIM28.

3) The reviewer’s third concern that the specificity of targeting TRIM28 to the HIV-1 promoter has been explained in the ATAC-Seq data interpretation section. In summary, HIV-1 LTR was not the only target of TRIM28. All of the reported targets of LRAs were not specific to HIV-1 LTR only as well. We have conducted several toxicities experiments to prove that targeting TRIM28 is non-toxic. We do provide another novel target to develop LRA, although other target may also be influenced. We will carefully examine the other toxicity and carcinogenic potency associated with TRIM28 deficiency when we develop LRAs.

https://doi.org/10.7554/eLife.42426.044

Article and author information

Author details

  1. Xiancai Ma

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4934-4221
  2. Tao Yang

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Validation, Investigation, Methodology
    Competing interests
    No competing interests declared
  3. Yuewen Luo

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  4. Liyang Wu

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  5. Yawen Jiang

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  6. Zheng Song

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  7. Ting Pan

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  8. Bingfeng Liu

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  9. Guangyan Liu

    College of Basic Medical Sciences, Shenyang Medical College, Shenyang, China
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  10. Jun Liu

    1. Institute of Human Virology, Sun Yat-sen University, Guangzhou, China
    2. Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Software, Methodology
    Competing interests
    No competing interests declared
  11. Fei Yu

    Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  12. Zhangping He

    Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  13. Wanying Zhang

    Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  14. Jinyu Yang

    State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Sun Yat-sen University Cancer Center, Guangzhou, China
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  15. Liting Liang

    Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  16. Yuanjun Guan

    Core Laboratory Platform for Medical Science, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Data curation, Methodology
    Competing interests
    No competing interests declared
  17. Xu Zhang

    Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Methodology, Project administration
    Competing interests
    No competing interests declared
  18. Linghua Li

    Department of Infectious Diseases, Guangzhou Eighth People’s Hospital, Guangzhou, China
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  19. Weiping Cai

    Department of Infectious Diseases, Guangzhou Eighth People’s Hospital, Guangzhou, China
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  20. Xiaoping Tang

    Department of Infectious Diseases, Guangzhou Eighth People’s Hospital, Guangzhou, China
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  21. Song Gao

    State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Sun Yat-sen University Cancer Center, Guangzhou, China
    Contribution
    Resources, Software, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7427-6681
  22. Kai Deng

    Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Conceptualization, Resources, Methodology
    Competing interests
    No competing interests declared
  23. Hui Zhang

    Key Laboratory of Tropical Disease Control of Ministry of Education, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    zhangh92@mail.sysu.edu.cn
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3620-610X

Funding

National Special Research Program of China for Important Infectious Diseases (2018ZX10302103)

  • Hui Zhang

National Special Research Program of China for Important Infectious Diseases (2017ZX10202102)

  • Hui Zhang

National Natural Science Foundation of China (81730060)

  • Hui Zhang

National Natural Science Foundation of China (81561128007)

  • Hui Zhang

Joint-innovation Program in Healthcare for Special Scientific Research Projects of Guangzhou (201803040002)

  • Hui Zhang

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the members of Zhang, Deng and Gao laboratories for good discussions. We also thank Dr. Juntao Gao from TsingHua University for suggestions on super resolution imaging processing. We also thank clinicians from department of infectious diseases of Guangzhou Eighth People’s Hospital for recruiting study participants and collecting samples. We also honor and thank all the HIV-1-infected participants, whose supports impel us feel obligated to go on HIV-1 research.

Ethics

Human subjects: Chronically HIV-1-infected participants sampled by this study were recruited from Department of Infectious Diseases in Guangzhou 8th People's Hospital, Guangzhou. The Ethics Review Board of Sun Yat-Sen University and the Ethics Review Board of Guangzhou 8th People's Hospital approved this study. All the participants were given written informed consent with approval of the Ethics Committees. The enrollment of HIV-1-infected individuals was based on the criteria of prolonged suppression of plasma HIV-1 viremia on cART, which is undetectable plasma HIV-1 RNA levels (less than 50 copies/ml) for a minimum of six months, and having high CD4+ T cell count (at least 350 cells/mm3). Blood samples from healthy individuals were obtained from Guangzhou Blood Center. We did not have any interaction with the healthy individuals or protected information, and therefore no informed consent was required. The statement was also included in the Materials and Methods section.

Senior Editor

  1. Wenhui Li, National Institute of Biological Sciences, China

Reviewing Editor

  1. Jeremy Luban, University of Massachusetts Medical School, United States

Publication history

  1. Received: September 28, 2018
  2. Accepted: January 16, 2019
  3. Accepted Manuscript published: January 17, 2019 (version 1)
  4. Version of Record published: February 4, 2019 (version 2)

Copyright

© 2019, Ma et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,776
    Page views
  • 475
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Computational and Systems Biology
    2. Microbiology and Infectious Disease
    Guillaume Witz et al.
    Research Article Updated
    1. Immunology and Inflammation
    2. Microbiology and Infectious Disease
    Chung Park, John H Kehrl
    Research Article