The HIV-1 Tat Protein Is Monomethylated at Lysine 71 by the Lysine Methyltransferase KMT7*

The HIV-1 transactivator protein Tat is a critical regulator of HIV transcription primarily enabling efficient elongation of viral transcripts. Its interactions with RNA and various host factors are regulated by ordered, transient post-translational modifications. Here, we report a novel Tat modification, monomethylation at lysine 71 (K71). We found that Lys-71 monomethylation (K71me) is catalyzed by KMT7, a methyltransferase that also targets lysine 51 (K51) in Tat. Using mass spectrometry, in vitro enzymology, and modification-specific antibodies, we found that KMT7 monomethylates both Lys-71 and Lys-51 in Tat. K71me is important for full Tat transactivation, as KMT7 knockdown impaired the transcriptional activity of wild type (WT) Tat but not a Tat K71R mutant. These findings underscore the role of KMT7 as an important monomethyltransferase regulating HIV transcription through Tat.

The HIV-1 epidemic remains a global health problem despite the growing availability of potent antiretroviral therapies. These therapies are not curative, as latent, transcriptionally silent virus can spontaneously reactivate from sanctuaries and rapidly rekindle viral infection after withdrawal of therapy (1)(2)(3). Therefore, the molecular mechanisms of the activation and suppression of HIV transcription are of great interest.
HIV encodes its own viral transactivator, Tat, which activates HIV transcription and facilitates its own production in a positive feedback loop. Tat is a small protein, typically found in a full-length form of ϳ101 amino acids (aa) 3 or as a splice variant (72 aa) encoded only by the first exon of the tat open reading frame. Both isoforms efficiently transactivate the HIV promoter in the 5Ј long terminal repeat (LTR). Tat interacts with the positive transcriptional elongation factor b (P-TEFb), and viral RNA through several well-characterized domains that can be found in its one-exon form: a cysteine-rich domain (aa [22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37] and a highly conserved core domain (aa [41][42][43][44][45][46][47][48], both of which participate in binding of P-TEFb (4 -6). The neighboring arginine-rich motif (ARM; aa 49 -57) interacts with a specific stem-loop RNA structure called transactivation response element (TAR) located in the 5Ј extremities of all viral transcripts (7)(8)(9)(10). Specific binding of the Tat ARM to TAR requires the coordinated binding of P-TEFb to Tat, as the cyclin T1 subunit of P-TEFb binds both the Tat cysteine-rich and core domains and loop sequences of TAR (11). C-terminal to these well-characterized domains is a glutamine-rich motif (aa 59 -72). When expressed as a peptide, this region adopts a conserved ␣-helical structure that is stabilized upon binding to TAR RNA and is implicated in T-cell apoptosis (12,13).
We previously reported that Lys-51, within the Tat ARM, is monomethylated (K51me) by the methyltransferase KMT7 (also called SET7/9) (17). Monomethylation at Lys-51 enhanced TAR RNA binding of Tat, and increased trimolecular complex formation between Tat, TAR, and P-TEFb. Interestingly, KMT7 itself was found to bind TAR RNA in band-shift assays, indicating that it could be recruited to the HIV promoter before Tat was produced (17). In in vitro methylation assays of ARM peptides (aa 44 -59), we observed that that KMT7 fails to methylate residues other than Lys-51 in Tat (17,18). However, when we conducted the same in vitro assays using purified oneexon Tat proteins (Tat72), we observed additional methylation on Tat indicating that one or more additional methylation sites existed outside of the ARM region. In this study, we sought to further understand the role of KMT7 in the Tat transcription cycle by identifying and characterizing the additional modification(s). We hypothesized that understanding the full extent of Tat methylation would bring new insight into the regulation of Tat function and the role of KMT7 as an HIV-1 transcriptional cofactor.  (Fig. 1, A and B). In Tat72 proteins with an acetyl group at Lys-50, [ 3 H]SAM incorporation was also slightly decreased. This is consistent with previous observations that this modification reduces access of KMT7 to Lys-51 (18).

In
To specifically identify additional KMT7 methylation sites in Tat, we performed in vitro methylation reactions using nonradiolabeled SAM and subjected modified Tat proteins to MALDI-TOF MS/MS analysis developed to analyze Tat (18). This analysis revealed monomethylation at a single additional site, Lys-71 ( Fig. 1, C and D, supplemental Table S-1) in two distinct peptides in reactions with KMT7, but not in reactions with a control enzyme KMT1E (also called SETDB1) or no enzyme (Fig. 1C). In one peptide, only Lys-71 was monomethy-  KTM7 Modifies Tat at Lys-71 in Vitro-Next, we tested whether pre-modification of Lys-71 in Tat72 peptides affects in vitro methylation by KMT7. Premethylation of Lys-71 markedly decreased [ 3 H]SAM incorporation while Lys-51 pre-methylation had a lesser effect. Finally, premethylation of both Lys-51 and Lys-71 abolished methylation of Tat, demonstrating that there are no additional targets for KMT7 in Tat ( Fig. 2A). These findings support Lys-51 and Lys-71 as the sole sites for KMT7 monomethylation in Tat72.
To better quantify dynamics of KMT7-mediated methylation of Tat72 proteins, we used pre-modified Tat proteins in a modified kinetic radiometric assay (24). After testing for linearity with time and KMT7 enzyme concentrations, kinetics assays were performed with various concentrations of Tat72 proteins (Fig. 2B). Methyltransferase activity was hyperbolic, and all Tat proteins except the Tat K51/K71me doubly modified form followed Michaelis-Menten kinetics. The doubly modified form had very little methylation, consistent with our observations using autoradiography ( Fig. 2A).
The catalytic turnover (k cat ) and Michaelis constant (K m ) of the Tat K51me proteins were 62% lower relative to the unmodified proteins. Interestingly, the k cat of the K71me proteins decreased by 83% but the K m decreased only 21%. Thus, the methylation efficiency (k cat /K m ) on K51me protein was similar to that of the unmodified protein, but the K71me was 78% less than that on control proteins. Furthermore, almost no methylation was detected in the K51/K71me protein. Together, these data support the model that Lys-71 is preferred over Lys-51 as a target for KMT7 in Tat (Fig. 2, B and C) and there are no additional sites in Tat72 for KMT7 monomethylation.

MS of Tat Purified from HEK293T and Jurkat A2 Cells Reveals Monomethylation at Lys-71 exists in Vivo-To examine
Lys-71 methylation in vivo, we coexpressed Tat101 bearing a C-terminal FLAG tag together with either a KMT7 or empty vector in HEK293T cells for 24 h. We first purified Tat from lysates by FLAG immunoprecipitation, followed by SDS-PAGE separation. The separated Tat protein was subjected to MALDI-TOF MS/MS, which showed that Lys-71 was monomethylated only when KMT7 was overexpressed (Fig. 3

, A and B;
supplemental Table S -2). This suggests that the fraction of Tat that is naturally monomethylated at Lys-71 in HEK293T cells may be small compared with the unmodified form.
Next, we analyzed Tat methylation in J-Lat A2 T cells, in which expression of FLAG-tagged Tat101 is controlled by its natural promoter and induced by stimuli such as tumor necrosis factor ␣ (TNF␣) (25). Tat expression was induced with TNF␣, FLAG-affinity-purified, and analyzed by MS. We identified a peptide fragment of 1032.627 Da, corresponding to Tat aa 70 -78 with a monomethyl group at Lys-71 (Fig. 3C). MS/MS confirmed the monomethylation at Lys-71 ( Fig. 3D, supplemental Table S-2). Together these findings show that Tat Lys-71 is monomethylated in Jurkat T cells under conditions mimicking natural HIV infection and that this modification can be induced by KMT7 overexpression in 293T cells.
Generating Antibodies Specific for K71me-We confirmed these results using newly generated modification-specific polyclonal antibodies. We previously published methods to produce and characterize mono-, di-, and tri-methyl Tat-specific antibodies at Lys-51 (26). Following these methods, we focused on two chemically synthesized K71me 11-mer peptides (type 1 and type 2) (Fig. 4A). Peptide 1 (aa 63-72) ends with the first exon, and peptide 2 (aa 67-76) spans both Tat exons. Cysteine residues are added to the C-terminal end of the peptides for purification purposes as described (26). After antigen-purification of the modification-specific antibodies, we performed dotblot analysis of various chemically modified Tat proteins (Fig.  4B). Type 1 antibodies reacted with all synthetic Tat72 proteins (WT, K51me, K71me, K50Ac, and K51/K71me), but type 2 antibodies detected only the K71me Tat72 proteins. Similar results were obtained by SDS-PAGE and Western blotting (Fig.  4C). The type 2 antibodies were highly specific for Tat72 proteins carrying a monomethyl group at K71, and the type 1 anti- bodies recognized all Tat proteins. Type 2 antibodies detected several cellular bands in whole cell lysates (data not shown). Therefore, for further analysis, we used type 2 TatK71me antibodies on FLAG-immunopurified samples.
Detecting Tat K71me by Western Blot Analysis in Vivo-To confirm the specificity of the type 2 antibodies in vivo, we overexpressed WT or FLAG-tagged Tat101 proteins carrying mutations at Lys-71, Lys-51, or Lys-50 in 293T cells. After FLAG immunoprecipitation, we detected K71me Tat only in WT, Lys-50 and Lys-51 mutants, but not in Lys-71 mutants (Fig. 5A). This finding indicates that endogenous KMT7 is sufficient to methylate Lys-71 in cells. Furthermore, we were able to increase TatK71me by coexpressing Tat101 with WT KMT7 relative to coexpression with a catalytically inactive KMT7 mutant (H297A). Importantly, no change in methylation was observed on the K71R mutant (Fig. 5B).
Next, we determined whether the K71me antibodies are able to detect Tat72 in T cells. For this, we used an HIV-1 minigenome to generate Jurkat cell lines that stably express Tat72, as described for the Jurkat A2 cells (27). We sorted for cells with high-level GFP expression, and transduced them with lentiviruses containing one of two shRNAs targeting KMT7 (a and b) or a scrambled shRNA (Scr). The lentiviral constructs also contained a puromycin resistance gene, which allowed us to select for shRNA-expressing cells. After selection, shRNA-expressing cells were lysed and subjected to FLAG immunoprecipitation, and Tat K71me levels were determined by Western blotting (Fig. 5C). Knockdown of KMT7 was robust; in accordance, Tat K71me levels were reduced by 51% in cells treated with the KMT7b shRNA and by 28% in cells treated with KMT7a shRNA (Fig. 5B, bottom). These results underscore the role of KMT7 as an important K71 methyltransferase for both forms of Tat in T cells. Since a substantial amount of Tat remained methylated at Lys-71 despite the knockdown, we suspect that a small residual pool of KMT7 may be sufficient to modify a proportion of Tat proteins, or perhaps another enzyme is capable of monomethylating Tat in cells.
K71me Does Not Affect Tat Half-life-A number of substrates modified by KMT7 are altered in their stability by downstream changes in their polyubiquitylation (28). Because Lys-71 has previously been reported as a site for polyubiqitination (23), we examined whether Lys-71 methylation by KMT7 affects the stability of Tat in Jurkat cell lines expressing fluorescent Tat101-Dendra fusion proteins generated for this purpose (25). The Tat-Dendra system has been used to determine the Tat half-life in flow cytometry and microscopy studies (29,30). To knock down KMT7, we transduced Tat-Dendra cells with the lentiviral vectors expressing scrambled or KMT7b shRNAs used for Fig. 5B. After puromycin selection, we monitored effects of KMT7 knockdown on Tat half-life only in successfully transduced cells. Knockdown of KMT7 was confirmed by Western blotting (Fig. 6B). Expression of WT or K71R Tat-Dendra proteins was induced with TNF␣. After 16 -20 h of TNF␣ treatment, cycloheximide was added to inhibit de novo production of Tat. Cells were fixed at various times, and Tat expression was determined by flow cytometry of Dendra (Fig.  6A). Tat stability did not differ in cells transduced with control and KMT7 shRNAs, excluding any prominent effect of K71 methylation on Tat stability (Fig. 6C). In cells expressing scrambled shRNA, the K71R Tat mutant had a slightly longer half-life than WT Tat (10.9 versus 9.0 h, p ϭ 0.043). However, time-lapse single-cell microscopy of Tat-Dendra cells revealed no statistical difference in the half-lives of WT and K71R Tat in the absence of shRNAs (Fig. 6D). Together, these data affirm previous findings that Lys-71 modifications are not involved in Tat protein stability (29).
Lys-71 Methylation Enhances Tat Transactivation-Previously, we found that KMT7-mediated monomethylation of Tat Lys-51 enhances interactions between Tat and TAR RNA and P-TEFb and activates HIV gene expression (17). To assess the effect of Lys-71 monomethylation on Tat transactivation, we transfected plasmids expressing WT or mutant Tat101 (K51R, K71R, or K51/71R) into TZMBL cells, which express firefly luciferase from the integrated HIV LTR when functional Tat is expressed (31). Transactivation was ϳ50% lower in the Tat K51R and Tat K71R mutants than in WT Tat at three different plasmid concentrations, confirming that these residues are important for Tat transactivation (17,23). Transactivation was almost completely lost when both Lys-51 and Lys-71 were mutated, underscoring their combined importance in Tat transactivation. All Tat proteins were expressed at similar levels as confirmed by Western blot analysis (Fig. 7A).
To test whether monomethylation at Lys-71 by KMT7 contributes to Tat transactivation, we knocked down KMT7 expression using KMT7b shRNAs in TZMBL cells before transfection with WT Tat or K71R mutant Tat101. Knockdown of KMT7 decreased the transcriptional activity of Tat WT but had no effect on K71R mutant Tat. Evidently, KMT7 activates Tat transactivation, at least in part, through Lys-71 monomethylation (Fig. 7B). Knockdown of KMT7 was confirmed by Western blotting (Fig.7B).ThesefindingsindicatethatLys-71monomethylation, rather than regulating Tat stability, plays a positive role in the transcriptional activity of Tat.

Discussion
HIV-1 Tat is a potent viral transactivator that undergoes extensive post-translational modifications. Here, we expand the role of KMT7 in Tat function by identifying Lys-71 as a second monomethylation site. We found that Lys-71 is monomethylated in both functional HIV-1 Tat101 and Tat72  splice variants, underscoring its importance throughout the HIV-1 life cycle. We identified KMT7 as a robust Tat Lys-71 monomethyltransferase in both in vitro and in vivo assays. Although KMT7 substrates often have perturbed stability, mutation of Lys-71 or knock down of KMT7 did not affect the stability of Tat. Instead, Lys-71 monomethylation is important for the transcriptional activity of Tat through a yet unresolved mechanism.
Notably, in vitro methylation and enzymology experiments indicate that the preferred methylation site of KMT7 is Lys-71 rather than Lys-51, previously identified as a target of KMT7. Although KMT7 has no known stringent site specificity, two consensus sequences in KMT7 targets have been described:  (32,33). Comparing the sequences of K71me (S Ϫ3 L Ϫ2 S Ϫ1 Kme 0 Q ϩ1 -P ϩ2 ) and K51me (G Ϫ3 R Ϫ2 K Ϫ1 Kme 0 R ϩ1 R ϩ2 ), Lys-71 is more closely aligned with the newer consensus sequence than Lys-51, possibly explaining why Lys-71 was the preferred target in our in vitro studies (32).
Our finding that KMT7 methylates Tat at two sites is consistent with reports that KMT7 often has multiple targets within individual substrates. For example, the RelA subunit of the NF-B transcription factor has three KMT7 monomethylation sites (K37me, K314me, K315me) and the PCAF acetyltransferase has two (K78me, K89me) (34 -36). Since the addition of a monomethyl group is a rather subtle modification (14 Da), it is likely that multiple monomethylation sites act in concert to mediate appropriate regulation of the substrate protein by KMT7.

JOURNAL OF BIOLOGICAL CHEMISTRY 16245
Two main consequences of KMT7-mediated methylation have emerged: alteration of protein stability and the regulation of interactions between nucleic acids and proteins (32). The oncogene p53 and estrogen receptor ␣ are both stabilized upon monomethylation by KMT7 (37,38). In contrast, RelA of NF-B and DNMT1 are both destabilized upon monomethlyation of one or more sites by KMT7 (35,39). One review regarded KMT7 as a bona fide protein stability modifier, proposing the presence of a methyl/ubiquityl switch that can regulate the stability of substrates (28). However, polyubiquitination of Tat at Lys-71 has been linked to enhanced transcriptional activity, not degradation (23). Indeed, we found that KMT7 knockdown did not significantly affect the stability of WT or K71R Tat.
Early structural studies of Tat isolates suggested that the glutamine-rich motif (aa 59 -72) has a degenerate ␣-helical structure, consistent across different Tat isolates and adopted in the presence of TAR RNA (12). This structure is adopted partially through conserved glutamine-RNA hydrogen bonds (Gln-60, Gln-63, Gln-66, Gln-72) and an electrostatic interaction between Lys-71 and the TAR phosphodiester backbone at nucleotides 31-35 (12,13). Lys-71 monomethylation could thus enhance these interactions by stabilizing the electrostatic interaction between Lys-71 and the TAR loop (13,17). Previously, we showed that methylation of Lys-51 by KMT7 activates appropriate Tat/TAR/P-TEFb binding (17). Lys-51 lies in the ARM of Tat, which binds TAR RNA in the bulge region. Therefore, methylation of Lys-51 and Lys-71 by KMT7 might coordinately enhance the binding of TAR RNA to multiple residues in Tat, potentially positioning the RNA properly with respect to Tat and P-TEFb. Future experiments are necessary to explore this model and elucidate the mechanism by which KMT7 activates Tat transactivation by monomethylating Lys-71.
In Vitro Methylation-Reactions were carried out with 3 g of synthesized Tat peptides and 0, 1, or 2 g of purified KMT7 in a solution of 0.1 M Bicine, pH 8.2, 60 M [ 3 H]SAM. After incubating the mixtures for 18h at room temperature, the methylation reactions were run on 15% SDS-PAGE gels, which were stained with Coomassie Blue, and destained overnight. Methylation was detected by autoradiography (22-h exposure).
Mass Spectrometry-In vitro modified and in vivo purified Tat peptides were analyzed by MALDI-TOF tandem mass spectrometry (MS) as described (18).
Enzymology-Methyltransferase activity was measured with a modified radiometric assay (24). Kinetic assays were performed with various concentrations of synthetic Tat72 peptides (unmodified, Tat K71me, Tat K51me, and Tat K51/K71me). The 20-l reactions contained Tat peptides, 0.1 M Bicine, pH 8.2, 60 M [ 3 H]SAM (3.4Ci/mmol), and 1 g of purified fulllength KMT7 (40). The reactions were incubated for 1 min at 37°C. To terminate the reaction and ensure full precipitation of the substrates, 0.5 ml of 10% TCA and 5 l of a bovine serum albumin (BSA) solution (10 g/100 ml of water) were added. After this addition the mixture was then vortexed, incubated on ice for 3 min, and centrifuged at 14,000 rpm for 3 min. To process the reaction, the protein pellets were washed with 150 l of 0.1 M NaOH, re-precipitated in TCA, vortexed and centrifuged at 14,000 rpm for 3 min. The pelleted proteins were dissolved in 50 l of formic acid, diluted by half with water, mixed with 1.25 ml of Bio-Safe II scintillant (Research Products International), and subjected to liquid scintillation. Activity was calculated after correcting for methylation in control reactions lacking either the enzyme or the substrate. The assays were performed in triplicate, and the data were plotted and fitted with SigmaPlot 11 (Systat Software) to the Michaelis-Menten equation.
Tat K71me Antibody Purification-Peptides for antibody synthesis (K71me1 Type 1, K71me1 Type 2; Fig. 4A) were synthesized by PSL Peptide Specialty Laboratories and injected into rabbits and antibodies were purified from serum with antigenic peptides as described (26). Purified antibodies were eluted under acidic conditions, resuspended in 1% BSA and 0.1% sodium azide, and stored at Ϫ80°C.
Dot Blot Analysis of K71me Antibodies-Tat peptides were serially diluted in water and spotted on a 0.2 M Hybond ECL membrane. Membranes were air-dried, and nonspecific binding was blocked with nonfat dry milk (5 g/100 ml in TBST consisting of 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.1% Tween 20) for 1 h at room temperature. Membranes were incubated with K71me antibodies diluted in blocking buffer for 1 h, washed five times with TBST, then incubated with HRP-conjugated anti-rabbit IgG (Jackson ImmunoResearch) at a concentration of 16 ng/ml in blocking buffer, washed 3-5 times with TBST, and analyzed using standard ECL substrate.
Measuring the Half Life of Tat-WT LTR-Tat101-Dendra (LTD) constructs were subjected to site-directed mutagenesis to generate the K71R mutation in Tat. WT or K71R LTD constructs, a lentiviral construct (pCMV-⌬R8.91) and VSV-G pseudotyped envelope plasmid were co-transfected into HEK293T cells to produce lentivirus, as described (41). Jurkat T cells were infected with either WT or K71R LTD lentiviral vectors to generate polyclonal J-Lat populations as described (25). After resilencing of the LTR, LTD jurkats were infected with KMT7b shRNAs and selected with puromycin for at least 1 week. For protein stability experiments, Tat expression was induced with TNF␣ (10 ng/ml) for 16 h followed by treatment with cycloheximide (10 g/ml). Cells were fixed at various times in 2% paraformaldehyde (Alfa Aesar), incubated at 4°C for at least 1 h, and analyzed on a FACSCalibur DxP8 (Cytek). Flow cytometry data were analyzed by FlowJo X. Data were normalized to 100% at time 0 and natural log transformed to produce a linear half-life curve, from which the slope was calculated. Two-tailed Z-tests were performed on averaged slope values from at least four independent experiments. The equa-tion t (1/2) ϭ ln(2)/-slope was used to generate half-life values. Time-lapse microscopy experiments were performed as described (29).
Luciferase Assays-TZMBL cells (1 ϫ 10 5 ) were transfected with a total of 100 ng of DNA containing 1, 2, 5, or 10 ng of Tat-expressing plasmids (WT, K51R, K71R, K51/71R) or empty vector using X-tremeGENE 9 (Roche Diagnostics) as recommended by the manufacturer. The cells were incubated for 48 h and lysed in 1ϫ Promega Passive Lysis Buffer. Luciferase assays were processed with the Promega Dual-Luciferase Reporter Assay System and measured on a Monolight 2010 luminometer. Experiments were conducted with four independent biological replicates with technical duplicates and the statistical significance of differences was determined with one-sided t tests.