Identification of novel HIV-1 Rev-interacting proteins
To identify novel Rev-interacting proteins, we screened a library of cDNAs derived from the human Jurkat T-cell line with full-length Rev as bait in a yeast two-hybrid system. Repeated selection procedures led to isolation of two library plasmids (11.5.1 and 16.4.1) encoding specific interactors of Rev.
Sequence analyses and data base comparisons revealed that the 936 bp insert in plasmid 11.5.1 is identical with a segment of a 1543 bp cDNA encoding human DNA binding protein B (dbpB; NCBI accession number BC002411) [40
]. The predicted coding sequences in the 11.5.1 cDNA comprise the C-terminal 139 amino acids of the dbpB protein (324 amino acids; NCBI accession number M24070). Several biological activities have been attributed to dbpB, including binding to DNA [41
] and RNA [42
] and regulation of transcription [44
The other library plasmid 16.4.1 contained a 696 bp insert of which a region of over 450 nucleotides showed strong similarity to a sequence within a human fetal heart cDNA (NCBI accession number W67699). In the fetal heart cDNA the matching region encompasses a predicted open reading frame. Alignment of the 16.4.1 and the fetal heart cDNA sequences yielded a sequence encoding a hypothetical 171 amino acid 16.4.1 protein. Since interaction with Rev is the first biological activity associated with this gene product, we analysed interaction of Rev with the 16.4.1 protein in more detail.
To investigate which regions of Rev contribute to interaction with the 16.4.1 protein, we analysed the capacity of various known mutants of Rev to interact with 16.4.1 in the yeast two-hybrid assay. The amino acid exchanges in these mutants map to regions associated with major biological properties of Rev (Fig. ), including multimerization (RevM4 [45
] and RevSLT40 [46
]), RNA binding and nuclear localization/accumulation (RevM5 [45
]) and nuclear export of Rev (RevM10BL [47
]). Expression of LexA-Rev-mutant bait proteins in yeast transformants was confirmed by Western blot analysis with polyclonal antibodies against Rev (data not shown). As positive control for Rev interaction, interaction analysis was performed with LexA-Rev bait and B42AD-Rev prey, confirming oligomerization of wildtype Rev molecules with each other (data not shown).
While Rev mutants RevM4 and Rev M10BL were capable of interacting with 16.4.1, no interaction was observed with Rev mutants RevM5 and RevSLT40 (Fig. ). These results indicate that amino acid residues R38 or R39 of the ARM and I59 or L60 of the multimerization region II (MII) are required for interaction of Rev with the 16.4.1 protein. Furthermore, they suggest that the 16.4.1 interacting sequences in Rev are located between aa positions 38 and 60.
For more detailed study of the interaction of the 16.4.1 protein with Rev, yeast two-hybrid analysis was performed with various segments of the 16.4.1 cDNA as prey and wildtype Rev as bait (Fig. ). Amino acid regions of 16.4.1 extending from position 2 to 133 and from position 39 to 171 showed similar Rev-binding capacity as full-length 16.4.1 protein. In contrast, both the N-terminal region (2 to 38) and the C-terminal region (134 to 171) of 16.4.1 failed to interact with Rev. While 16.4.1 protein fragments from position 2 to 73 or position 74 to 171 clearly interacted with Rev, interactions were weaker than that of full-length 16.4.1. These results indicate that the Rev-interacting region of the 16.4.1 protein is located between amino acid positions 39 and 133 and that, within this region, sequences N- and C-terminal of position 73 contribute to interaction with Rev.
Interaction of the 16.4.1 protein with Rev, CRM1 and itself in human cells
The interaction of the 16.4.1 protein with Rev in yeast raises the question whether the 16.4.1 protein can also interact with Rev in human cells. It was also of interest whether 16.4.1 is capable of interacting with human CRM1, since CRM1 has been shown to interact with several Rev-associated factors (see Background).
We addressed these issues with a mammalian two-hybrid assay, in which the interaction of a protein fused to the Gal4 DNA-binding domain with a second protein fused to the VP16-activator domain induces transcription of a luciferase reporter gene from a synthetic promoter (for details see Materials and Methods). Rev was fused to VP16 (VP16-Rev) to avoid unspecific interactions between the acidic VP16 domain [48
] and the basic Rev protein (estimated pI = 9.93; MacVector calculation). Functionality of VP16-Rev was demonstrated (data not shown) in a Rev-reporter assay [3
]. For interaction analysis, HEK293 cells were cotransfected with expression plasmids for VP16-Rev and Gal4-16.4.1 fusion proteins and the reporter plasmid pG5luc
. As shown in Fig. , a ≈11-fold mean induction of luciferase activity was observed in 14 independent transfection experiments. Assessment of interaction of 16.4.1 with human CRM1 in cells coexpressing Gal4-16.4.1 and VP16-hCRM1 revealed a ≈41-fold mean induction of luciferase activity export (n = 7) (Fig. ). Self-interaction of the 16.4.1 domain was analysed by coexpressing Gal4-16.4.1 and VP16-16.4.1, resulting in ≈12-fold mean induction of luciferase activity (n = 6).
Figure 2 Interaction of 16.4.1 with HIV-1 Rev, hCrm1 and with itself in human cells. 16.4.1 interactions in human cells were analysed with a mammalian two-hybrid assay in which the interaction of a protein fused to the Gal4 DNA-binding domain with a second protein (more ...)
In all three cases, induction of luciferase activity was significantly (p < 0.04) increased over induction levels obtained in control assays with unfused VP16 and Gal4-16.4.1 (3.3-fold; n = 7).
These results indicate that the 16.4.1 domain is capable of interacting with Rev as well as with the export receptor CRM1 and of forming homo-oligomers in human cells.
Cytoplasmic localization of 16.4.1 is CRM1/Exportin 1 dependent
Comparison of the sequence in the 16.4.1 cDNA with the fetal heart cDNA indicated that the 16.4.1 sequence was incomplete at its 5' terminus. To generate a full-length (171 aa) 16.4.1 coding sequence, nucleotides encoding the first 8 N-terminal amino acids derived from the predicted open reading frame of the fetal heart cDNA were inserted upstream of the 16.4.1 cDNA. To analyse subcellular localization of the 16.4.1 protein, cells were transfected with plasmids directing expression of fusion proteins containing full-length 16.4.1 or various segments of 16.4.1. Those fusion proteins contained either a N-terminal IgG1 tag or a C-terminal GFP tag. The full-length IgG1-16.4.1 fusion protein (IgG1-2-171) was located mainly in the cytoplasm of HeLa cells (Fig. ). IgG1 fusion proteins with 16.4.1 regions extending from amino acid position 2 to 133, 39 to 171 and 74 to 171 showed similar predominantly cytoplasmic localization (Fig. ). In contrast, IgG1 fusion proteins with the N-terminal region (2 to 38) or the C-terminal region (134 to 171) of 16.4.1 were apparent in both nucleus and cytoplasm, similar to unfused IgG1. These results demonstrate that the 16.4.1 protein is capable of cytoplasmic accumulation and suggest that sequences directing cytoplasmic localization of the 16.4.1 protein are located between amino acid positions 74 to 133.
Figure 3 CRM1-dependent cytoplasmic localization of 16.4.1. HeLa cells were transfected with plasmids directing expression of IgG1-16.4.1 or 16.4.1-GFP fusion proteins and subcellular distribution of tagged proteins analysed 24 hours later in fixed cells. (A) (more ...)
The 16.4.1-GFP fusion protein showed similar cytoplasmic localization as IgG1-16.4.1 (Fig. ). Quantitative evaluation of subcellular distribution of GFP fluorescence [49
] revealed that only 25% of total fluorescence was contained in the nuclei of 16.4.1-GFP expressing cells. This localization is comparable to that of GFP fusion proteins containing PKIα (PKIα-GFP) or the carboxyterminal half of Rev (Rev(52–116)-GFP), which localize to 23% and 25%, respectively, in the nucleus (Fig. ). PKIα and the carboxyterminal half of Rev contain well-characterized recognition signals for CRM1/Exportin 1-dependent export [36
]. Similar cytoplasmic localization of 16.4.1-GFP and interaction of 16.4.1 with CRM1/Exportin 1 in human cells (Fig. ) raised the possibility that cytoplasmic localization of 16.4.1-GFP at steady state may involve nuclear export of 16.4.1 by CRM1/Exportin 1. Therefore we analysed the effect of Leptomycin B (LMB), an inhibitor of CRM1-dependent nuclear export [11
] on subcellular distribution of 16.4.1-GFP. LMB treatment significantly increased the nuclear proportion of 16.4.1-GFP from 25% to 44%. LMB-induced nuclear redistribution was similar in cells expressing PKIα-GFP and Rev(52–116)-GFP, whose nuclear proportion increased to 49% and 46%, respectively. Quantitative analysis demonstrated that 45% of unfused GFP localized to the nucleus, in agreement with its known capacity to diffuse throughout the cell [50
]. LMB had no significant effect on subcellular distribution of unfused GFP.
These results indicate that cytoplasmic localization of 16.4.1 involves nuclear export by CRM1/Exportin1. Amino acid region 74 to 133 of 16.4.1 seems to be crucial for those transport processes.
Identification of a candidate nuclear export signal in 16.4.1
To further characterize the involvement of the amino acid region 74–133 in cytoplasmic localization of 16.4.1, we assessed subcellular distribution of GFP fusion proteins containing this region of 16.4.1. Cells expressing a GFP fusion protein with a single copy of aa 74–133 of 16.4.1 contained a higher proportion of nuclear fluorescence (38%, Fig. ) than cells expressing GFP fusion proteins with full-length 16.4.1 (25%, Fig. ). However, GFP-fusion proteins containing two copies of region 74 to 133 of 16.4.1 in tandem showed similar cytoplasmic localization (Fig. ; 27% nuclear proportion) as full-length 16.4.1-GFP (Fig. ; 25%). Treatment of cells with LMB raised nuclear proportions of GFP fusion proteins with one or two copies of 16.4.1 region 74–133 to similar levels as full-length 16.4.1-GFP. These results suggest that the region between amino acid positions 74 and 133 contains a CRM1/Exportin 1 dependent nuclear export signal, which can act in a cumulative manner.
Figure 4 Analysis of CRM1-dependent nuclear export of amino acid region 74 to 133 of 16.4.1. HeLa cells were transfected with plasmids directing expression of GFP fusion proteins containing 16.4.1 region 74–133 in single copy or in tandem (74–133) (more ...)
Examination of the hypothetical amino acid sequence of region 74 to 133 revealed a clustering of leucine and isoleucine residues between amino acid 86 and 105 (Fig. , shaded in grey). To analyse whether region 86 to 105 of the 16.4.1 protein functions as a nuclear export signal, we compared its translocation capacities with the Rev-NES in a previously described microinjection assay [51
] (Fig. ). In this assay, peptides bearing the candidate transport sequences are linked to fluorescently labeled bovine serum albumin (BSA). These potential transport substrates are coinjected into the nucleus with unlinked BSA labeled with a different fluorescent color that serves as injection control. Two hours later, cells are fixed and the percentage of each fluorescent label in the nuclear compartment of individual cells determined. The relative translocation activity signifies the ratio of fluorescence of the transport substrate to the fluorescence of the injection control. Selective export of the transport substrate from the nucleus yields relative translocation activities < 1, as demonstrated for a transport substrate containing the NES of Rev (Fig. and [51
]). A substrate containing the 16.4.1-derived sequence also yielded a relative translocation activity < 1 (Fig. ). These results indicate that region 86 to 105 of 16.4.1 sequence can function as a nuclear export signal.
Figure 5 Functional analysis of a nuclear export signal in 16.4.1. (A) Depicted is the sequence (nucleotides and predicted amino acids) of the 16.4.1 protein investigated here. The sequence encoding amino acids 1–8 are derived from the fetal heart cDNA (more ...)
To further characterize this nuclear export signal in 16.4.1 we took advantage of a collection of weight matrices (M1-M7) derived for recognition of NES by bioinformatics (Blossom similarity matrix). These matrices recognized 48 out of 75 signals of a published NES database [36
] at a default threshold of 0.84 in the context of their native proteins. No match was obtained upon scanning of the 16.4.1 amino acid sequence with these matrices at default threshold. This indicates that the 16.4.1 sequence is distinct from the 48 NES represented by the matrices. However, rescanning of the 16.4.1 sequence at a lower threshold (0.74) yielded a single match for matrix M5 (0.78), comprising amino acids 92–99 of 16.4.1 (core NES). At default threshold the same matrix recognized a specific group of NES that includes the NES of Stat1 and p65RelA (Fig. ). However this matrix did not recognize the NES of PKIα or Rev, which were recognized by different matrices. An artificial 16.4.1 NES sequence containing leucine instead of isoleucine residues at positions 99 and 101 was recognized by matrix M5 above default score (0.86) but by no other matrices, even at reduced thresholds.
Figure 6 Influence of the nuclear export signal on cytoplasmic localization of 16.4.1. (A) Sequences identified in 16.4.1 and in proteins contained in a database of experimentally verified NES by a common weight matrix. Protein sequences in NESbase version 1  (more ...)
Finally we investigated whether the candidate transport signal also shows nuclear export activity in the context of the complete 16.4.1 protein. As shown in figure , the leucine and two isoleucine residues of the 16.4.1 core NES were changed to Alanin and the subcellular distribution of the 16.4.1(NESmut)-GFP was compared to the wildtype 16.4.1 fused to GFP. The mutant 16.4.1-GFP fusion protein localized to significantly higher levels in the nucleus than wildtype 16.4.1-GFP (34% versus 27%). However, the nuclear proportion of the mutant 16.4.1-GFP remained below that of unfused GFP (Fig. ), indicating residual nuclear export of the mutant 16.4.1-GFP.
In summary, combined computational and functional analyses indicate that amino acid residues 86 to 105 act as a nuclear export signal, with amino acids 92 to 99 constituting a potential core NES. Mutational analysis indicates that the leucine/isoleucine of the 16.4.1 core NES contribute to but are not sole determinants of cytoplasmic localization of 16.4.1.
Colocalization of 16.4.1 and Rev
This report demonstrates interaction of 16.4.1 and Rev in yeast and mammalian two-hybrid assays (Figs. and ). In these approaches, candidate interaction partners are artificially targeted to the nucleus to measure interaction-dependent reporter gene expression.
To analyse whether 16.4.1 and Rev interact under conditions in which they retain their natural localization behavior, we analysed cells coexpressing 16.4.1 and Rev for colocalization of both proteins.
To this end, we first established a HeLa cell line stably expressing 16.4.1-GFP and a corresponding control cell line expressing unfused GFP. The expression of 16.4.1-GFP for more than 20 passages did not affect cell growth monitored by measurement of growth curves and did not lead to cell toxicity detectable as release of lactate dehydrogenase (LDH) or ATP into cell culture supernatants (data not shown). Furthermore, long-term expression did not alter the predominantly cytoplasmic localization of 16.4.1-GFP.
For colocalization studies, HeLa 16.4.1-GFP cells and control HeLa-GFP cells were transfected with a plasmid directing expression of Rev-CFP (cyan fluorescent protein) fusion proteins. Transfected cells were subjected to epifluorescence microscopy and Z-stacks were collected. Images were processed by deconvolution and multichannel unmixing, allowing separate evaluation of the spatial distribution of GFP and CFP signals. Over 25 cells were analyzed. Multichannel unmixing is a recently developed technique for separate detection of fluorochromes that exhibit significant spectral overlap in conventional fluorescence microscopy setups, such as CFP and GFP (for a review see [52
]). Fig. shows examples of cells expressing 16.4.1-GFP either alone (white arrow) or together with Rev-CFP (red arrow). 16.4.1-GFP was only visible in the nucleoli of cells co-expressing Rev-CFP but not in cells lacking Rev-CFP (see also Fig. ). Cells coexpressing 16.4.1-GFP and Rev-CFP showed stronger nucleoplasmic GFP fluorescence than HeLa 16.4.1-GFP cells lacking Rev-CFP. Rev-CFP retained typical nuclear/nucleolar localization [49
] when coexpressed with 16.4.1-GFP, indicating that 16.4.1-GFP does not influence localization of Rev-CFP. Control imaging of HeLa cells expressing GFP either alone (Fig. , top panel) or together with Rev-CFP (Fig. , bottom panel) showed that presence of Rev-CFP did not influence the GFP signal and that the CFP signal was apparent only in cells expressing Rev-CFP. These results verified separation of Rev-CFP and GFP signals by the multichannel unmixing routine and confirmed that the CFP-tag in Rev-CFP does not affect localization of GFP.
Figure 7 Nucleolar colocalization of 16.4.1-GFP and Rev-CFP. HeLa 16.4.1-GFP cells and control HeLa GFP cells were transiently transfected with a Rev-CFP expression plasmid. Nuclei were counterstained with Hoechst 33342. 3D-Maximum image projections of Z-stacks (more ...)
These results indicate that Rev is capable of directing 16.4.1 to nucleoli and provide further evidence for interaction of Rev and 16.4.1 in human cells.
Influence of 16.4.1 on Rev functions
To investigate the influence of 16.4.1 on Rev function, we analysed the effect of IgG1-16.4.1 and 16.4.1-GFP fusion proteins on transactivation capacity of Rev using a previously described Rev-reporter assay [3
]. The mRNA synthesized from the reporter gene in this assay contains a region coding for red fluorescent protein (RFP) and a non-coding region with HIV-1 derived sequence elements mediating Rev-responsiveness. These consist of multiple INS from the HIV-1 gag
gene and the RRE from the HIV-1 env
gene. Rev activity is measured by quantification of RFP reporter positive cells by flow cytometry using the gating strategy depicted in Fig. (for further details see figure legend). Experiments were performed in 293T cells because of the high transfection efficiencies achieved in these cells.
Figure 8 16.4.1 influences transactivation capacity of Rev in human cells. Panel (A) depicts the reduction of Rev activity obtained by cotransfection of 0.5 μg plasmids directing expression of 16.4.1 fusion proteins and 0.1 μg rev expression plasmids. (more ...)
The transactivation capacity of Rev in the absence of exogenous 16.4.1 was set at 100%. The result of 5 independent experiments demonstrate an approximately 50% reduction of Rev activity by coexpression of 16.4.1 fusion proteins (Fig. ). A dose-dependent effect of 16.4.1-GFP expression on Rev activity was observed (Fig. ). No effect was observed for numerous other gene products of a human cDNA-library tested in this assay (Wolff et al, manuscript in preparation).
In the experiment above we showed that overexpression of 16.4.1-GFP exhibited a negative effect on the transactivation capacity of HIV-1 Rev in human cells. Isolation of 16.4.1 from a human cDNA library suggests that 16.4.1 proteins may be produced in human cells. To target expression of native 16.4.1 we decided to use RNA interference. To identify inhibitors of 16.4.1 expression we analysed several candidate siRNAs targeted to sequences within the 16.4.1 coding region and a negative control siRNA that recognizes sequences located upstream of the 16.4.1 coding region. An exemplary experiment is shown in Fig. . HeLa 16.4.1-GFP cells were transfected with siRNAs and the effects on expression of 16.4.1-GFP monitored by flow cytometry (Fig. and ). 16.4.1-GFP expression levels in RNAi transfected cells were determined relative to those in untransfected cells in 40.000 cells by FACS analysis. siRNA-16.4.1 reduced mean relative expression levels of 16.4.1-GFP to 36%. A similar effect was observed for a positive control siRNA that silences GFP (data not shown). The negative control siRNA (siRNA-nsp) only moderately diminished mean relative expression of 16.4.1-GFP to 81%. A similarly moderate reduction was observed for mock-transfected cells (data not shown) indicating that this is caused by the RNA transfection procedure. Analysis of the inhibitory effect of siRNA-16.4.1 on 16.4.1-GFP expression in three additional experiments yielded a mean relative expression of 16.4.1-GFP of 30.7% + 4.7 (standard deviation), confirming the inhibitory effect of this siRNA on 16.4.1-GFP.
Figure 9 Influences of endogenous 16.4.1 proteins on transactivation capacity of Rev in human cells. (A and B) Downregulation of 16.4.1-GFP expression in HeLa cells by siRNAs. HeLa cells stably expressing 16.4.1-GFP were transfected with a pre-synthesized siRNA (more ...)
Subsequently we investigated the effect of siRNA-16.4.1 and the negative control siRNA-nsp in 293T cells in the Rev-reporter assay described above (Fig. ). The negative control siRNA (siRNA-nsp) had no effect on Rev transactivation capacity compared to mock transfected cells (1% increase of Rev activity). In contrast, siRNA-16.4.1 increased Rev transactivation capacity by 17% compared to mock-transfected controls. A specific enhancing effect of siRNA-16.4.1 was observed in three independent experiments. These results indicate that endogenous 16.4.1 gene products are capable of modulating Rev activity.
Expression of 16.4.1 proteins
Database searches identified several cDNAs of various lengths that contain the complete 16.4.1 sequence within a predicted open reading frame [For overview see additional file 1
: Figure A1 and additional file 2
: Table A1]. These are derived from various human tissues and cells. The predicted molecular masses of the hypothetical proteins encoded by these cDNAs range from ~145 kDa to 18.5 kDa. This suggests existence of several human 16.4.1 protein species encoded by various cDNAs, rather than a single 16.4.1 protein generated from a single cDNA.
Production of 16.4.1 proteins has not been reported so far and is currently under investigation in our laboratory (Kramer-Hämmerle et al., in preparation). In the context of this ongoing study, a monoclonal antibody (Mab) was generated against recombinant 16.4.1. Indirect immunofluorescence of HeLa cells transfected with the IgG1-16.4.1 expression plasmid revealed a cytoplasmic staining pattern that was indistinguishable from that obtained with antibodies against IgG1 [see additional file 3
: Figure A2, A) a, for image]. The 16.4.1 Mab, but not the secondary antibody, also stained untransfected HeLa cells, yielding a cytoplasmic, granular pattern [see additional file 3
, Figure A2, A) b for image]. In Western Blot analysis of HeLa-16.4.1-GFP cells, the 16.4.1 Mab recognized a band with the predicted molecular mass of ~45 kDa for the 16.4.1-GFP fusion protein as well as additional proteins [see additional file 3
, Figure A2, B]. These results confirm specific recognition of 16.4.1 antigens by the 16.4.1 Mab and indicate expression of endogenous 16.4.1 proteins.
The 16.4.1 Mab has also been used to analyze cells from different lines and primary tissues, including brain and peripheral blood mononuclear cells. A staining pattern similar to that in HeLa cells was observed for 4 out 5 cell lines analyzed by indirect immunofluorescence (not shown). Western blot pagesanalysis of the cell lines/tissues investigated so far yielded a total of 4 distinct bands, ranging in size from > 150 kDa to < 30 kDa [for summary see additional file 3
, Table A2]. The occurrence of these bands depended on the cell line/tissue investigated. These results confirm expression of 16.4.1 proteins in human cells and tissues and suggest cell-specific expression patterns of 16.4.1 proteins.
Future studies are directed at identifying the full range of 16.4.1 protein species with a panel of antibodies and characterizing 16.4.1 expression patterns on cDNA and protein levels in various cell types.