Visualization of in Vivo Direct Interaction between HIV-1 TAT and Human Cyclin T1 in Specific Subcellular Compartments by Fluorescence Resonance Energy Transfer*

Human cyclin T1, a component of the P-TEFb kinase complex, was originally identified through its biochemical interaction with the Tat transactivator protein of human immunodeficiency virus type 1 (HIV-1). Current understanding suggests that binding of Tat to P-TEFb is required to promote efficient transcriptional elongation of viral RNAs. However, the dynamics and the subnuclear localization of this process are still largely unexplored in vivo. Here we exploit high resolution fluorescence resonance energy transfer (FRET) to visualize and quantitatively analyze the direct interaction between Tat and cyclin T1 inside the cells. We observed that cyclin T1 resides in specific subnuclear foci which are in close contact with nuclear speckles and that Tat determines its redistribution outside of these compartments. Consistent with this observation, strong FRET was observed between the two proteins both in the cytoplasm and in regions of the nucleus outside of cyclin T1 foci and overlapping with Tat localization. These results are consistent with a model by which Tat recruits cyclin T1 outside of the nuclear compartments where the protein resides to promote transcriptional activation.

The human immunodeficiency virus type 1 (HIV-1) 1 transactivator protein Tat is a small polypeptide (86 -101 amino acids, according to the viral strains) essential for efficient transcription of viral genes. The protein is a highly unusual transcription factor since, at the HIV LTR promoter, it interacts with a cis-acting RNA element (trans-activation-responsive region, TAR) present at the 5Ј-end of each viral transcript (1).
Through this interaction, Tat activates HIV-1 transcription by promoting the assembly of transcriptionally active complexes at the LTR by multiple protein-protein interactions (for a recent review, see Ref. 2). Over the last few years a number of cellular proteins have been reported to interact with Tat and to mediate or modulate its activity. These include general transcription factors, among which TBP, TAFII250, TFIIB, TFIIH (3-7); RNA polymerase II (8); transcription factor Sp1 (9); the transcriptional co-activators and histone acetyltransferases p300/CBP and P/CAF (10 -12); and the cyclin subunit of the positive transcription elongation factor complex (P-TEFb), cyclin T1 (13)(14)(15)(16).
The finding that Tat biochemically and functionally interacts with several cellular proteins raises some fundamental questions. Does Tat directly interact with its partners inside live cells? Which is the subcellular compartment of these interactions? Are they occurring simultaneously or consecutively? Some of these questions can be successfully addressed by taking advantage of fluorescence resonance energy transfer (FRET) measurements (17), allowing investigation of direct interaction of proteins labeled with optically matched fluorophores. FRET exploits radiationless energy transfer driven by dipole-dipole interaction occurring from a fluorophore (the donor) in the excited state to another fluorophore (the acceptor) when in close proximity; energy transfer is followed by acceptor fluorescence. The presence of FRET indicates actual proteinprotein interaction at distances in the range of the FRET length scale, the Förster radius (R 0 ), defined as the distance at which FRET efficiency (E T ) is 50%. E T is defined as the ratio between the sixth power of R 0 and the sum of the sixth power of R 0 and the sixth power of R. R is the actual distance among the donor and the acceptor fluorophores. E T dramatically decreases when R increases by a fraction of the nanometer (nm) around R 0 , which is commonly of the order of the nm for many pairs of matched fluorophores (18 -20). In particular, E T reaches 98 and 1.5% for donor-acceptor separations lower than 0.5 R 0 and higher than 2 R 0 , respectively. This implies that simple co-localization of two proteins is not sufficient to yield energy transfer; thus, the presence of FRET is a powerful indicator of physical protein-protein interaction.
Once integrated in the host cell chromosomes, transcription of HIV-1 genes reflects the complex processes that characterize endogenous mammalian gene expression. A critical event that regulates processivity of transcription from the HIV-1 LTR promoter is the phosphorylation of the carboxyl-terminal domain of cellular RNA polymerase II (RNP), an enzymatic modification carried out by different kinase complexes that regulate promoter clearance and counteract the effect of negative transcription elon-gation factors (for a recent review, see Ref. 21). One of the essential functions of Tat at the LTR promoter is to stimulate RNP processivity by specifically interacting with cyclin T1 (15). This cyclin and its cognate catalytic component (CDK9) constitute the P-TEFb kinase complex that phosphorylates the RNP carboxylterminal domain (13,14,16,22;reviewed in Ref. 23). Binding of Tat to cyclin T1 induces co-operative binding of P-TEFb onto nascent TAR RNA (24), which in turn increases Tat affinity for TAR (25). Biochemically, Tat appears to contact residues in the amino terminus of cyclin T1, which are not essential for binding of cyclin T1 to CDK9, through its trans-activation domain (24, 26 -30). Tat transactivation and HIV-1 replication are closely tied to the levels of P-TEFb inside the cell. Increased levels of cyclin T1 are found in activated lymphocytes and after the differentiation of promonocytic cells to macrophages (22,31). Furthermore, cyclin T1 is up-regulated by phorbol myristate acetate and phytohemagglutinin and the ability of peripheral blood lymphocytes to support HIV replication directly correlates with the levels of induced CycT1 and carboxyl-terminal domain kinase activity (32). To further corroborate the central role of cyclin T1 in Tatmediated transactivation is the observation that murine cyclin T1, that associate HIV-1 Tat only weakly, is also inefficient in supporting Tat-mediated transactivation (24,26). Recently, Herrmann and Mancini (33) reported that cyclin T1 is distributed in discrete foci inside the nucleus, which co-localize with regions enriched of splicing factors (nuclear speckles), and raised the possibility that these sites might constitute the regions of P-TEFb activity.
By using FRET, here we demonstrate that Tat and cyclin T1 physically associate inside the cells and visualize the subcellular compartments in which this interaction occurs. We show that FRET is maximum in regions of the nucleus outside of cyclin T1 foci and superimpose with the regions of Tat localization. According to these results, we derive a model by which Tat determines the redistribution of cyclin T1 outside of the compartments where the protein normally resides to promote transcriptional activation.

EXPERIMENTAL PROCEDURES
Cell Culture-HL3T1 cells, a HeLa derivative containing an integrated HIV-1 LTR driving the expression of a CAT reporter gene, were a kind gift of G. Pavlakis (National Cancer Institute-Frederick Cancer Research Facility, Frederick, MD). All other cell lines were obtained from the American Type Culture Collection (ATCC) and were cultured in Dulbecco's modified Eagle's medium with glutamax (Life Technologies, Inc.) supplemented with 10% fetal bovine (Life Technologies, Inc.), serum, and gentamicin (100 g/ml) at 37°C in a humidified 95% air, 5% CO 2 incubator.
Plasmids-pcDNA3-Tat-BFP and pcDNA3-Tat-EGFP plasmids were generated by polymerase chain reaction cloning of the Tat gene from HIV-1 HXBc2 (86 amino acids) fused at its 3Ј-end to BFP (Quantum, pQBI50-fc1) or EGFP (CLONTECH, pEGFP-N1) in pcDNA3 (Invitrogen). Similarly, pcDNA3-BFP-cyclin T1 and pcDNA3-EGFP-cyclin T1 were constructed by polymerase chain reaction amplification of the human cyclin T1 gene and contain BFP and EGFP fused at its 5Ј-end. Plasmid pCMV-HA-CycT1 contains the epitope for an anti-HA monoclonal antibody fused at the NH 2 terminus of the cyclin T1 cDNA. It was a kind gift of K. A. Jones (Salk Institute, La Jolla, CA). Plasmid pU3R-III, containing the HIV-1 LTR upstream of the CAT reporter gene, was a kind gift of J. Sodroski (Dana Farber Cancer Institute, Boston, MA). EGFP-SF2/ASF and Fibrillarin-EGFP were generously provided by T. Misteli (National Institutes of Health, Bethesda, MD).
FRET-Cells were transiently transfected with expression plasmids for Tat and cyclin T1 fused to the different fluorescent proteins by the calcium phosphate method in LabTek II four-chamber glass slides (Nalgene). Cells were fixed in 4% paraformaldehyde after 48 h and mounted directly in 70% glycerol for FRET analysis. FRET measurements were carried out by an epifluorescence Axioskop 2 Zeiss microscope mounting a 103 W HBO lamp, a 100 ϫ 1.3 N.A. oil-immersion Plan-Neofluar objective, and Nomarski optics. FRET analysis was performed in two steps. First, EGFP emission was collected by integrating the fluorescence signal around 520 nm (band width 40 nm) under EGFP excitation at 480 nm (wavelength selection was obtained by 40 nm band-pass filters, excitation power was 5 W/cm 2 ). Second, EGFP emission in the same frequency range was measured after excitation at 350 nm (power density 2 W/cm 2 and band width 60 nm). Background was detected out of the cell under study for each frame and subtracted from the relevant fluorescent signal. Following this procedure, the ratio between the two measured EGFP emissions (data taken following excitation at 350 nm divided by those at 480 nm) provides the FRET efficiency. Fluorescence was collected by a PentaMax 512-EFT intensified CCD camera with detection times of the order of 0.1 s (in particular, for data taken under excitation at 350 nm they were 5 times longer than for those relative to 480 nm excitation). Data acquisition and analysis were performed with Metamorph software (Universal Imaging Corp.). When evaluating FRET ratios, emission intensities were scaled to take into account the different detection times.
For the quantification of subcellular FRET, the boundaries of individual subcellular compartments (cytoplasm, nucleus, nucleolus, and cyclin T1 foci) were first drawn on the saved Nomarski image of the cells under analysis. These profiles were then superimposed on the corresponding fluorescent images collected by illuminating the same cells at 480 and 350 nm; FRET was calculated according to the ratios between the averages of the two signals within the regions defined by these boundaries.
Immunofluorescence-Cells were transiently transfected by the calcium phosphate method and observed for indirect immunofluorescence after 20 h. Following paraformaldehyde fixation, cells were washed with 100 mM glycine and permeabilized with 0.1% Triton X-100 for 5 min. Primary antibodies against HA (Roche Molecular Biochemicals, 1/100) and SC35 (Sigma, 1/500) were incubated at 37°C for 1 h in a humidified chamber in phosphate-buffered saline additioned with 1% bovine serum albumin and 0.1% Tween 20. Secondary conjugated antibodies (Sigma) were diluted at 1/50 and incubated as described above. Slides were mounted in Vectashield (Vector) and observed with a Zeiss Axiovert X100 confocal microscope. Images were acquired with the LSM510 software.

Visualization of Intracellular Tat-Cyclin T1 Interaction Us-
ing FRET-To explore the interaction of Tat with cyclin T1 using FRET, we obtained fusion constructs of the two proteins with either the blue fluorescent protein (BFP) or the enhanced green fluorescent protein (EGFP) (34) (Fig. 1A). This fluorescent protein pair has excitation and emission properties favorable for FRET, since the emission wavelength of BFP partially overlaps with the excitation wavelength of EGFP (35,36). To ascertain that fusion of the fluorescent proteins did not interfere with the transcriptional functions of Tat or cyclin T1, we studied HIV-1 LTR transactivation in co-transfection experiments. As shown in Fig. 1B, both Tat-EGFP and Tat-BFP were able to transactivate the HIV-1 LTR indistinguishably from wt Tat. Similarly, both EGFP-Cyclin T1 and BFP-Cyclin T1, while inactive per se on LTR transcription (not shown), were able to synergize with wt Tat for transactivation. More important for the application to FRET, both the Tat-EGFP:BFP-Cyclin T1 and the Tat-BFP:EGFP-Cyclin T1 pairs were equally efficient in their synergistic activation of transcription as the corresponding wt proteins.
A first set of FRET experiments was performed by transfection of human HL3T1 cells, a HeLa derivative cell line carrying an integrated LTR-CAT cassette (37), with the constructs expressing the Tat-EGFP:BFP-Cyclin T1 protein pair. FRET image analysis of cells transfected with Tat-EGFP and BFP-Cyclin T1 and controls are shown in Fig. 2A, panels a1 to c4. Panels in row b show the intracellular distribution of fluorescence around 520 nm (this represents the optimum wavelength for EGFP detection) under excitation at 480 nm. In these conditions, most cells transfected with Tat-EGFP (panels b1 to b4) showed the characteristic pattern of overexpressed Tat, consisting of diffuse nucleoplasmic fluorescence with intense nucleolar staining (38,39). FRET analysis was performed by illuminating the same cells at 350 nm (to excite BFP) and recording at 520 nm (panels in row c), thus allowing compari-son of EGFP emission following BFP excitation with that following direct EGFP excitation. In these conditions, only samples expressing both Tat-EGFP and BFP-Cyclin T1 scored positive for fluorescence, indicating direct interaction between the two proteins (panels c1 and c2).
Detailed quantitative analysis of several cells transfected with the two proteins or with controls are presented in Fig. 2B, which shows the experimental FRET signal and its distribution. Most cells transfected with Tat-EGFP plus BFP-Cyclin T1 showed FRET values that were clearly higher than those detected in control transfections (p Ͻ 0.001). The distribution of these values, however, was disperse, with at least two cells out of 10 in the series of Fig. 2B showing FRET within control values. This variability is likely to be related to the different efficiency of expression and stability of the two proteins (data not shown). To overcome this problem we also analyzed FRET in cells transfected with the reciprocal constructs, namely Tat-BFP and EGFP-Cyclin T1. As shown in Fig. 3A, also this protein pair showed intracellular FRET, with acceptor (EGFP) fluorescence in the nucleus (including the nucleolus) after excitation at 350 nm (panels c1 and c2). In contrast, when Tat-BFP was co-transfected with fibrillarin-EGFP (a protein that specifically localizes in nucleoli, panels in column 3) or with EGFP-SF2/ASF, which marks sites of mRNA splicing, panels in column 4, no FRET was observed. Altogether these FRET results indicate that direct in vivo association between Tat and cyclin T1 occurs within the cells. In this respect, it is worth mentioning that results which are superimposable to those described above were also obtained in human U2OS cells, which do not contain any integrated proviral DNA (not shown). This indicates that the presence of the TAR RNA element is not a prerequisite nor does it influence Tat:cyclin T1 association.
Subcellular Localization of Cyclin T1-Analysis of fluorescence distribution showed that EGFP-Cyclin T1 exhibited a dotted distribution inside the nucleus, particularly evident when transfected in the absence of Tat (see, for example, Fig.  3A, panel b5). Moreover, also a fraction of the cells expressing EGFP-Tat showed a similar dotted distribution rather than the characteristic nucleoplasmic and nucleolar staining proper of Tat (one of these cells is shown in Fig. 2A, panel b2). Notably, this unusual dotted distribution of Tat was never observed when EGFP-Tat was expressed alone or together with proteins other than cyclin T1. This suggests that the modified Tat localization may be caused by its association with cyclin T1. The experiments were performed by calcium-phosphate transfection of the indicated constructs (Tat plasmids: 100 ng; cyclin T1 plasmids: 1 g) together with 0.5 g of a Tat-responsive CAT reporter (plasmid pU3R-III) in Chinese hamster ovary cells. In these cells, Tat transactivation is poorer than in human cells, since the endogenous cyclin T1 gene lacks a critical cystein residue required for high affinity Tat interaction (26,49,50). As a consequence, co-transfection of human cyclin T1 greatly enhances Tat activity. CAT activity was measured after 24 h as previously described (51).

FIG. 2. In vivo analysis of the interaction between Tat-EGFP and BFP-Cyclin T1 measured by FRET.
A, visualization of FRET in human HL3T1 cells. The plasmid constructs indicated on top of each column were transfected in HL3T1 cells; transfected cells were visualized by transmitted light in Nomarski configuration (panels in row a), by excitation at 480 nm and collection at 520 nm, showing EGFP fluorescence after direct EGFP excitation (panels in row b), and by excitation at 350 nm and collection at 520 nm, showing EGFP fluorescence after BFP excitation, indicating FRET (panels in row c). B, quantification of FRET between Tat-EGFP and BFP-Cyclin T1. Fluorescent emission at 520 nm from individual cells transfected with the indicated constructs was recorded after excitation at 350 or 480 nm, and integrated intensities over the whole cell were evaluated. Plotted values (indicated by dots) represent the ratio between these two measurements: higher values indicate more efficient resonant energy transfer between BFP and EGFP. Ten consecutively analyzed cells were considered for each transfection; both their individual fluorescence ratios and their percentile box plot distribution are shown. Horizontal lines from top to bottom mark the 10th, 25th, 50th, 75th, and 90th percentiles, respectively.
These observations prompted us to analyze in more detail the subnuclear localization of cyclin T1. Both EGFP-Cyclin T1 (Fig.  4A, panel a) and a HA-tagged cyclin T1 construct probed with anti-HA antibody (Fig. 4A, panel b) were visualized in a nuclear dotted pattern. Recently reported evidence suggest that these sites of cyclin T1 accumulation inside the nucleus coincide with nuclear speckles (33), which contain pre-mRNA splicing factors and are commonly defined by the localization of the non-RNP protein SC35 (40). To confirm that cyclin T1 was localized in nuclear speckles, we studied localization of EGFP-Cyclin T1 and SC35, the latter visualized by immunofluorescence with a specific antibody. As shown in Fig. 4B for U2OS cells and Fig. 4C for HL3T1 cells, we found that most of the cyclin T1 nuclear foci were actually associated with SC35 speckles, but that only in few cases was their distribution exactly overlapped. These results suggest that cyclin T1 foci are possibly distinct from nuclear speckles, but that in most cases these two domains are spatially juxtaposed.
Tat-Cyclin T1 Interaction in Different Subcellular Compartments-As reported above, a fraction of cells transfected with Tat-EGFP and BFP-Cyclin T1 showed Tat fluorescence distributed in discrete foci instead than diffuse and nucleolar ( Fig. 2A,  panel b2). Conversely, EGFP-Cyclin T1 in cells co-transfected with Tat-BFP showed a less pronounced dotted pattern and an unusual staining of the nucleolus (compare Fig. 3A, panels b1  and b2, with Fig. 3A, panel b5). These observations indicate a reciprocal influence of the two proteins on their subcellular localization. The issue of Tat relocalization by cyclin T1 was also experimentally tested in a Chinese hamster ovary derivative cell line which stably expresses Tat-EGFP (41). In this cell line, in which Tat is constitutively expressed at low levels, EGFP fluorescence is found diffuse in the nucleoplasm (Fig. 5,  panel a). Transfection of these cells with an expression vector for cyclin T1 determined a remarkable re-distribution of Tat-EGFP to a dotted pattern resembling that of cyclin T1 bodies (Fig. 5, panels b-d). These considerations prompted us to measure the interaction between cyclin T1 and Tat in the different subcellular compartments. FRET results for 10 HL3T1 cells transfected with EGFP-Cyclin T1 and Tat-BFP are shown in Fig. 6A. In all cells, FRET highly above average of controls was observed in all analyzed subcellular compartments, including cytoplasm, nucleus, nucleolus, and cyclin T1 foci. Thus, high FRET intensities were observed also in compartments where cyclin T1 does not normally reside such as the nucleolus. This clearly results from its direct interaction with Tat, since fibrillarin, another protein that physiologically accumulates in the nucleolus, scored negative for FRET when tested together with Tat (see Fig. 3B). Interestingly, the highest FRET values were observed in the cytoplasm, suggesting that Tat-cyclin T1 interaction also occurs in this compartment. Inside the nucleus, FRET values were higher in the nucleolus, significantly lower inside the cyclin T1 bodies, and intermediate in the remaining regions (summarized by percentile distribution in Fig. 6B). This is consistent with the recruitment of cyclin T1 out of its nuclear bodies when Tat is expressed inside the cell. Fig. 2A. Controls obtained by co-transfection of plasmids expressing Tat-BFP with fibrillarin-EGFP, a protein that accumulates in nucleoli, and EGFP-SF2/ ASF, a protein which co-localizes with nuclear speckles are also shown. B, quantification of FRET between Tat-BFP and EGFP-Cyclin T1. Graphic representation of FRET is as in Fig. 2, panel B. Neither control protein EGFP-SF2/ASF nor fibrillarin-EGFP showed resonant energy transfer with Tat-BFP.

FIG. 4. Subnuclear localization of cyclin T1.
A, cyclin T1 is found in dotted compartments within the nucleus. HL3T1 cells were transfected with pcDNA3-EGFP-cyclin T1 (panel a) or with pCMV-HA-CycT1 (panel b); in the latter case, cyclin T1 was visualized with a fluorescein isothiocyanate-labeled anti-HA antibody. B and C, cyclin T1 domains are juxtaposed to SC35 (splicing) nuclear speckles. Human U2OS cells (panels B, a-c) and HL3T1 cells (panels C, a and b) were transfected with pcDNA3-EGFP-cyclin T1; after 24 h, cells were probed with an anti-SC35 antibody and a tetramethylrhodamine isothiocyanatelabeled secondary antibody, followed by visualization of EGFP and tetramethylrhodamine isothiocyanate fluorescence by laser scanning confocal microscopy. Panels in B show both individual EGFP-Cyclin T1 and SC35 fluorescence as well as their merged image; panels in C show only merged images. Green color indicates EGFP, red color, SC35; and yellow color, co-localization of the two proteins. In most cases, cyclin T1 foci juxtapose to SC35 nuclear speckles, with only partial overlap.

DISCUSSION
The association of HIV-1 Tat with P-TEFb through the interaction with cyclin T1 represents a key event for transcriptional activation of the viral LTR. Although a great deal of information is available on the molecular details of this interaction in vitro, little is known on its spatial and temporal correlates in vivo. To explore binding between Tat and cyclin T1 inside the cells we exploited fluorescence resonance energy transfer using BFP and EGFP proteins. This fluorescent protein pair has excitation and emission properties favorable for FRET with R 0 Ϸ 4 nm, and has already been used to detect FRET in other biological settings (35,36). To perform the analysis we fused EGFP or BFP to COOH terminus of Tat and to the NH 2 terminus of cyclin T1. As show in Fig. 1B, the fusion proteins performed as well their wild type counterparts in promoting HIV-1 LTR transcriptional activation. FRET experiments were conducted by co-transfecting cells with different pairs of plasmids expressing EGFP and BFP fusion constructs. Fixed cells were excited at 350 or 480 nm and emission was recorded at 520 nm. While excitation at 480 nm induces only EGFP fluorescence (BFP absorption is absent), under illumination at 350 nm three signals at 520 nm must be considered: BFP emission, directly excited EGFP emission, and EGFP emission due to FRET. However, the first contribution is negligible because of the negligible BFP fluorescence in the spectral range of interest. As a consequence one can express the two measured emissions after illumination at about 480 and 350 nm respectively as F 480 ϭ K ⌽ A,480 ␣ A,480 P 480 and F 350 ϭ K ⌽ A,350 ␣ A,350 P 350 ϩ K n A ⌽ A,480 E T ␣ D,350 P 350 , where A and D label acceptor and donor, ␣ and ⌽ are the extinction coefficient, and the quantum yield after excitation at the indicated wavelength, E T is the FRET efficiency, n A the fraction of EGFP molecules coupled in BFP:EGFP complexes, P the incident photon flux in the indicated spectral range, and K, a constant taking into account the set-up collection efficiency and other factors that do not vary in the two equations. We assumed that EGFP quantum yield after FRET is equal to that after optical excitation at 480 nm. The first term of the expression for F 350 represents EGFP emission under direct excitation at about 350 nm and is the main one responsible for the signal detected in the control samples. The second term is due to FRET. Our FRET intensity data are given by F 350 /F 480 . Higher ratios indicate more effective FRET. Furthermore, when the FRET ratio is much higher of the average control value (about 4 times greater), one can neglect the first term appearing in the expression of F 350 , thus approximating the FRET ratio as n A E T ␣ D,350 P 350 /␣ A,480 P 480 . In particular, with our experimental settings, P 350 /P 480 Ϸ 0.3 and ␣ D,350 /␣ A,480 Ϸ 0.5 (42), resulting in a FRET ratio of about 0.15 n A E T .
The interaction between Tat and cyclin T1 was quantitatively analyzed by observing fluorescence of a series of individual cells. As shown in Fig. 3B, cells transfected with Tat-BFP plus EGFPcyclin T1 showed average FRET values of 0.20 Ϯ 0.08, which were about 5 times higher than those expressing Tat-BFP and fibrillarin-EGFP (0.040 Ϯ 0.007) or Tat-BFP and EGFP-SF2/ASF (0.036 Ϯ 0.023; p Ͻ 0.001 in both cases). These high FRET signal values imply that the product n A E T between the fraction of EGFPs coupled to a BFP and the FRET efficiency is very close to unity. In light of the fact that by definition the two factors have one as an upper bound, this indicates that E T Ϸ 1. Given the peculiar ␤-can structure of GFPs, we are led to conclude that the interacting fluorophores are juxtaposed along the ␤-sheet outer shield with core-to-core distance of ϳ3 nm (43).
The distribution of cyclin T1 in the cell nucleus evoked that of certain subnuclear compartments rich in pre-mRNA processing factors known as nuclear speckles (44). When an antibody against the non-snRNP splicing factor SC35 was used to visualize nuclear speckles, we observed that in most cases cyclin T1 foci were in close vicinity of nuclear speckles, even if overlap between the two domains was only partial. This observation, which is in agreement with the data recently reported by Herrmann and Mancini (33), raises the possibility that the cyclin T1 nuclear domain might be independent but functionally related to nuclear speckles. Other proteins have been shown to accumulate in particular conditions at the periphery of nuclear speckles. Among these, nuclear bodies (defined as the sites of accumulation of the promyelocytic leukemia protein) have been suggested to often localize at the periphery of the SC35 domain, with transcription occurring in close proximity of this association (45). Further experiments are clearly required to clarify whether this might be case also of cyclin T1.
The measurement of the intensity of FRET in different subcellular compartments also offers an interesting tool to understand the kinetics of protein-protein interaction inside the cell. In agreement with biochemical data which showed that cyclin T1 can be also found in the cytoplasm of human cells (46), physical interaction between the Tat and cyclin T1 was also detected in this compartment. This cytoplasmic interaction is also in agreement with observations on some Tat mutants which lacked nuclear localization and were still able to act in a transdominant manner (47,48), possibly through their binding to cyclin T1 in the cytoplasm. Inside the nucleus, visualization of the interaction between Tat and cyclin T1 showed a distribution of FRET matching the localization of Tat (compare Fig. 2A, panel 1c, and Fig. 3A, panels 1c and 2c, with the localization of Tat-EGFP alone in Fig.  2A, panel 3b). Only a subset of cells (Ͻ5%) showed a different dotted localization ( Fig. 2A, panel 2c) similar to the distribution of EGFP-cyclin T1 alone (Fig. 3A, panel 5b, and Fig. 4A). When the analysis was conducted in Chinese hamster ovary cells expressing Tat-EGFP at low levels, transfection of cyclin T1 in excess of Tat redistributed the latter in nuclear foci resembling those of cyclin T1 (Fig. 5). Thus, the association of Tat with cyclin T1 is capable of redirecting the localization of each partner, depending on their relative concentrations. When we measured FRET in the different compartments we observed the lowest values in cyclin T1 foci and the highest in the nucleolus and in the cytoplasm (Fig. 6). These differences in FRET measurements are related to the relative local concentration of Tat and cyclin T1, which influences the probability of direct interaction between the two proteins. In the nucleolus (a site of Tat accumulation, but not of cyclin T1), FRET is higher since all cyclin T1 molecules recruited to this compartment are likely to be associated with Tat. Indeed its typical values (0.25) require nA ϳ1, i.e. virtually all EGFP molecules are paired with BFPs. In contrast, accumulation of unbound cyclin T1 in its foci prevails on Tat:cyclin T1 pairs, thus lowering FRET values. These observations are in favor of a model by which Tat redirects cyclin T1 outside of the compartment where the protein normally resides, suggesting that this event might be important for transcriptional activation. Visualization of the sites of HIV transcription inside the nucleus will permit us to directly address this issue.