# 2008 The Authors

Journal compilation # 2008 Blackwell Publishing Ltd

doi: 10.1111/j.1600-0854.2007.00696.xTraffic 2008; 9: 528–539Blackwell Munksgaard

Tuning the Transport Properties of HIV-1 TatArginine-Rich Motif in Living Cells

Francesco Cardarelli1,2,*, Michela Serresi1,

Ranieri Bizzarri1,2 and Fabio Beltram1,2

1Scuola Normale Superiore and Italian Institute ofTechnology, Piazza dei Cavalieri, 7, I-56126, Pisa, Italy2Scuola Normale Superiore and NEST INFM-CNR,via della Faggiola, 19, I-56126, Pisa, Italy*Corresponding author: Francesco Cardarelli,f.cardarelli@sns.it

A large body of work is currently devoted to the rational

design of new molecular carriers for the controlled

delivery of cargoes (e.g. proteins or nucleic acids) to

relevant subcellular domains, particularly the nucleus. In

this article, we show that rational mutagenesis of the

human immunodeficiency virus type 1 Tat-derived pep-

tide (YGRKKRRQRRR) affords variants with finely tuned

intercompartmental dynamics and controllable transport

mechanisms. Our findings are made possible by the

demonstration that the Tat peptide possesses two com-

peting functionalities capable of active nuclear targeting

and additional binding to intracellular moieties. By alter-

ing the competition between these two functions, we

show how to control cargo localization of Tat peptide

chimeras. Our investigation provides a unified, coherent

description of previous conflicting in vivo and in vitro

results and lets the true nature of the Tat peptide emerge.

Key words: active import, FRAP, NLS, passive diffusion,

Tat peptide

Received 6 July 2007, revised and accepted for publica-

tion 19 December 2007, uncorrected manuscript pub-

lished online 21 December 2007, published online 13

January 2008

Human immunodeficiency virus type 1 encodes Tat, an

essential regulatory protein active in the cell nucleus (1–3).

This protein is an unusual transcriptional transactivator

that dramatically enhances the processivity of transcrip-

tion directed by the viral long terminal repeat promoter

element (4–6). Tat function involves its direct interaction

with an RNA target site, termed the trans-acting respon-

sive (TAR) element, that is mediated by an arginine-rich

RNA-binding motif (ARM; sequence: YGRKKRRQRRR,

basic residues in bold) (7–10). The same domain was also

indicated as a nuclear localization signal (NLS) (11,12). The

Tat ARM exhibits an additional important property: it is

readily taken up by cells through interactions with heparan

sulfate proteoglycans displayed on the cell membrane

(13). By means of this mechanism, this Tat-derived pep-

tide can determine cellular uptake of nonpermeant mole-

cules and lead to the expected biological response,

consistently with the internalization of intact peptide–

cargo fusions (14,15).

In the search for new therapeutic strategies, a large body

of work is currently devoted to determine whether cell-

penetrating peptides (CPPs, such as Tat-derived peptides)

can act as molecular carriers for delivery of cargoes (e.g.

proteins or nucleic acids) to relevant subcellular domains,

particularly the nucleus. Indeed, nuclear localization is very

significant because it is one of the fundamental steps for

gene therapy approaches and can be exploited to probe

and modify cellular processes of utmost importance. The

majority of nuclear proteins are targeted to the nucleus by

basic, generally lysine-rich NLSs. Nuclear localization signals

appear to fall into several classes, including those homol-

ogous to the NLS of the simian virus SV40 large tumor

antigen consisting of a single stretch of basic residues and

those termed bipartite NLSs comprising two clusters of

basic amino acids separated by a spacer of 10–12 amino

acids as typified by Xenopus laevis nucleoplasmin (16). All

these NLSs share analogous transport processes and

cytosolic factors mediating this transport (17). In detail,

they are usually recognized by the cytoplasmic receptor

proteins Imp-a and Imp-b and actively translocated through

the nuclear pore channel (NPC) (18). The directionality of

nuclear import is thought to be conferred by the asym-

metric distribution of the GTP- and GDP-bound forms of

protein Ran between cytoplasm and nucleus (19).

Tat ARM was ascribed to a novel class of basic NLSs, and

different detailed molecular mechanisms for its nuclear

active import were described based on in vitro assays.

Truant and Cullen proposed an active nuclear import

mechanism involving the cytosolic factor Imp-b but not

Imp-a (11). In contrast, Efthymiadis et al. reported that

the Tat peptide can target a large heterologous protein

(b-galactosidase from Escherichia coli, 120 kDa) to the

nucleus in the absence of both Imp-b and Imp-a by a

novel carrier-independent, energy-consuming process

(12). More recently, contrary to these in vitro experiments,

we showed that in vivo, the dominant mechanism of Tat

peptide-mediated nuclear transport is passive diffusion

(20). In this latter work, we discussed the role played by

intracellular interactions in determining this behavior. In

particular, we emphasized the high affinity for RNA of the

Tat peptide (7–10). Also, Friedler et al. reported evidence of

this affinity and developed a functional backbone mimetic

of the Tat arginine-rich motif whose nuclear import/RNA-

binding properties were tested by in vitro assays (21).

Interestingly, for the homologous arginine-rich sequence

of the protein Rev, it was shown that the latter has a much

528 www.traffic.dk

higher affinity for RNA molecules than for the importins

(22) and that cellular RNAs are able to efficiently block its

Imp-b nuclear import (23). Additional Tat interactions must

also be considered, however. For example, de Mareuil

et al. (24) reported that Tat, through its residues 38–72,

interacts with the microtubule network (20).

In this article, we show that the Tat peptide sequence

possesses two competing functionalities capable of active

nuclear targeting and additional binding to intracellular

moieties. The role of the C-terminal basic ‘RRR’ stretch is

highlighted: it determines the passive diffusion behavior of

Tat peptide–cargo fusion proteins by inducing high-affinity

interactionswith intracellularmoieties such as RNAs. Based

on this knowledge, we show that the NLS function that is

suppressed in wild-type Tat peptides in vivo can be fully

recovered in engineered mutants. Finally, by scanning

mutagenesis, we show how to tune the Tat peptide-

sequence properties in terms of both its localization and

its dominant transport mechanism. Subcellular localization

in living cells was analyzed by confocal imaging, while

intracellular dynamics was investigated by fluorescence

recovery after photobleaching (FRAP) real-time imaging.

The impact of this tunability on Tat exploitation as a CPP

is also addressed and its implications discussed.

Results

Tat peptide sequence can be turned into a functional

NLS by scanning mutagenesis

In our previous work, we developed a method to screen

short peptide sequences for their ability to mediate nuclear

import based on in vivo studies and real-time imaging with

an array of cargoes and benchmark systems. Our approach

allowed us to conclude that passive diffusion drives

nucleus–cytoplasm trafficking of cargoes mediated by Tat

peptide (YGRKKRRQRRR, Table 1) (20). However, the

homologous basic domain of SV40 (YPKKKRKVEDP, Table

1) unambiguously showed different intracellular localization

and trafficking properties. Analysis of the two sequences

drew our attention to the last three C-terminal amino acids.

These are positively charged Arg-Arg-Arg (RRR) residues in

the Tat sequence (hereafter, TatRRR) and acidic Glu-Asp-Pro

(EDP) residues in the NLS of SV40 (hereafter, NLSEDP).

We investigated the properties of the ‘RRR’ stretch and

its role in determining TatRRR molecular interactions and

nuclear transport properties. First, scanning mutagenesis

of the Tat peptide sequence was performed by mutating

RRR into three (not charged) glycine residues: the resulting

mutant (TatGGG sequence in Table 1) was analyzed by

in vivo imaging techniques.

Remarkably, TatGGG fused to a 110-kDa cargo [hereafter,

TatGGG–green fluorescent protein (GFP)4], well above the

threshold for passive diffusion across the nuclear pore

(25), was almost exclusively detected into the nucleus

(Figure 1A). This indicates that the TatGGG sequence does

determine active nuclear import, analogously to its

NLSEDP-tagged counterpart. Conversely, as previously

reported and shown in Figure 1B, the wild-type Tat peptide

is not able to drive the active nuclear import of the same

cargo and is excluded from the nucleus. The integrity of

GFP4 constructs was checked by means of fluorescence

resonance energy transfer (FRET) imaging (Figure 1D; see

Materials and Methods for further details).

The role of active processes in determining the nuclear

permeation driven by TatGGG sequence was further verified

by analyzing the subcellular localization of TatGGG–GFP

(single GFP cargo, capable of crossing the nuclear envelope

by passive diffusion) in response to energy depletion (see

Materials andMethods). Figure 2A shows that TatGGG–GFP

is predominantly localized in the nucleus, with a nuclear-to-

cytoplasmic ratio Keq ¼ 2.1 � 0.1 when expressed in cells

under normal growing conditions (þATP panel). A sizable

GFP fluorescence signal is also observed within the

cytoplasm owing to the nucleus-to-cytoplasm TatGGG–GFP

Table 1: Transport parameters derived from FRAP measurements

Sequence Protein Keq t(C/N)

(seconds)

kDN(mm3/second)

kAT(mm3/second)

DC

(mm2/second)

DN

(mm2/second)

pbN

— GFP �1 77 � 19 13.6 � 3.7 0 20 � 5 19.8 —

YPKKKRKVEDP- NLSEDP-GFP 3.3 60 � 3 5.2 � 0.6 10–17.1 11.9 � 2.6 17.8 —

YGRKKRRQRRR- TatRRR-GFP �1 410 � 125 2.7 � 1.5 0 5.9 � 1.5 6 1

YGRKKRRQRRG- TatRRG-GFP 1.1 172 � 25 3.2 � 1.2 0.3–3.5 6.1 � 0.7 ND 0.95

YGRKKRRQRGG- TatRGG-GFP 1.6 95 � 15 5.2 � 1.6 3.0–8.3 6.9 � 1 ND 0.77

YGRKKRRQGGG- TatGGG-GFP 2.1 66 � 9 7.5 � 1.2 7.1–15.7 9.8 � 1.3 14.4 0.56

YPKKKRKVGGG- NLSGGG-GFP 2.4 66 � 11 5 � 2 7.2–12 11.2 � 1.6 19.6 —

YPKKKRKVRRR- NLSRRR-GFP 1.2 262 � 36 1.2 � 0.3 0.2–1.4 4.5 � 0.8 4.3 —

This table summarizes the main transport parameters derived from FRAP analysis: Keq, ratio between nucleoplasmic and cytoplasmic

fluorescence (mean value); t(C/N), time constant of nuclear fluorescence recovery (mean � SD); kDN, passive diffusion parameter

(mean � SD); kAT, active transport parameter (range of reported values); DC, cytoplasmic diffusion coefficient (mean � SD); DN,

nucleoplasmic diffusion coefficient (values derived from the equation: RM ¼ DN/DC, as described in Materials and Methods); pbN, molar

fraction of bound molecules within nucleoplasm; ND, no data. Italicized values represent previously reported data (see (20)). Tag-

sequences are reported in one-letter code with conserved residues outlined and introduced mutations in bold, respectively.

Traffic 2008; 9: 528–539 529

In Vivo Study of Tat Peptide-Targeting Properties

passive diffusion. When active processes were blocked

by ATP depletion, TatGGG–GFP passive diffusion from the

nucleus to the cytoplasm successfully restored a homoge-

neous intracellular distribution (Figure 2B, �ATP). Notably,

this behavior is reversible: within a fewminutes after bringing

cells back to physiological conditions, fluorescence was

again predominantly localized in the nucleus (data not

shown). This assay reveals that TatGGG–GFP nuclear

accumulation is driven by an energy-consuming process,

analogous to what we observed for NLSEDP–GFP (20). We

obtained analogous results for TatEDP–GFP construct in

which the ‘RRR’ stretch was replaced with the SV40-

derived ‘EDP’ one (data not shown).

We compared these results with similar measurements

performed on an NLSEDP-derived mutant in which the

C-terminal ‘Glu-Asp-Pro’ residues had been replaced by

‘Gly-Gly-Gly’ ones (hereafter, NLSGGG). As expected, this

protein maintained the wild-type nuclear active import prop-

erties (Figure 2B, þATP), with only a slight decrease in

efficiency (compare Keq values in Table 1). Accordingly, we

observed relocalization upon energy depletion (Figure 2B,

�ATP). This behavior is not unexpected because the

functional core domain of the NLS of SV40 (YPKKKRKV)

is conserved in this mutant (26). Notably, when we

replaced the C-terminal ‘EDP’ residues of the wild-type

NLS of SV40 sequence with the three arginines derived

from the Tat sequence, the characteristic nuclear accumu-

lation was almost completely abolished and a marked

nucleolar staining emerged (NLSRRR–GFP in Figure 2A). On

the contrary, untagged GFPs did not change their homoge-

neous intracellular distribution upon energy depletion treat-

ment (data not shown). Taken together, these results

indicate that the ‘RRR’ stretch can impair the potential active

nuclear import mechanism. In order to establish the trans-

port parameters of these mutants, we performed a FRAP

analysis of their nucleus/cytoplasm shuttling. Data were

analyzed under the framework of a general model of

nucleus–cytoplasm exchange that considers not only the

fluorescent construct but also its complexes with import

carrier(s) and/or other biomolecules (Figure 7). This approach

is described in the supplementary information and yields the

quantitative determination of the passive diffusion out of the

nucleus (expressed by the parameter kDN) and the concom-

itant estimate of the active transport into the nucleus

(expressed by the variability range of the parameter kAT).

The construct encoding for TatGGG–GFP was investigated

first and compared with the corresponding passive diffu-

sion benchmark TatRRR–GFP. The fluorescence of the

nuclear compartment was photobleached by irradiating

a single point with high laser power for 10 seconds, the

Figure 1: TatGGG-mediated active nuclear accumulation of GFP4 cargo. A) As expected, for a 110-kDa protein, GFP4 is excluded from

the nucleus (cargo molecular weight above the limit for passive diffusion through the NPC). B) TatRRR sequence is not able to drive GFP4

nuclear localization, as shown previously (18). C) Notably, TatGGG-tagged GFP4 is almost exclusively localized within the nucleus,

demonstrating that this Tat-derived mutant is able to perform active nuclear import. D) Integrity of GFP4 construct is shown here by FRET

imaging (see Materials and Methods for further details). Scale bar: 10 mm. ex, excitation.

530 Traffic 2008; 9: 528–539

Cardarelli et al.

subsequent recovery was recorded by time-lapse acquisi-

tion at 30-second intervals. Figure 2C shows the meas-

ured recovery curves: the characteristic TatGGG–GFP

recovery kinetics is considerably faster than that of

TatRRR–GFP. Table 1 summarizes our results for the whole

population of analyzed cells. We obtained t ¼ 66 � 9 sec-

onds for TatGGG–GFP and t ¼ 410 � 125 seconds for

TatRRR–GFP (20). Furthermore, by means of our kinetic

model, we can derive the TatGGG characteristic transport

parameter kAT ¼ 7.1 � 15.7 mm3/second for the active

transport component and kDN ¼ 7.5 � 1.2 mm3/second

for the nucleus-to-cytoplasm diffusion. Fluorescence

recovery after photobleaching analysis of NLSGGG–GFP

shows that substitution of the ‘EDP’ stretch does not alter

the kinetics of NLS-driven nuclear import (see recovery

curve in Figure 2D and transport parameters in Table 1).

Overall, the t values obtained for NLS-tagged proteins are

in good agreement with recently reported FRAP data in

COS-7 cells (27). At the same time, the mutant NLSRRR–

GFP showed a marked decrease in the nucleus/cytoplasm

shuttling rate (Figure 2D), confirming the hypothesis that

the C-terminal arginine-rich ‘RRR’ stretch has a dramatic

impact on active transport processes.

The role of the ‘RRR’ stretch in determining Tat

peptide interactions

Arginine-rich sequences are found in many RNA-binding

proteins and were proposed as the mediators of specific

RNA recognition processes: in particular, the number/

position of positively charged amino acids in the basic

domain of Tat is critical for its interaction with the specific

TAR RNA sequence (7–10). This prompted us to assess

the impact of the RRR/GGG replacement examined

above on Tat peptide-binding affinity toward RNAs

Figure 2: Influence of the ‘RRR’ determinant on nuclear localization and intercompartment dynamics of constructs. A) The

presence of the ‘RRR’ stretch is able to impair the active nuclear import driven by the NLS of SV40, leading to an intracellular localization

similar to that of wild-type Tat-tagged GFP (TatRRR–GFP). B) The corresponding mutants in which the ‘RRR’ residues were replaced by the

‘GGG’ ones (TatGGG–GFP and NLSGGG–GFP) showed a marked nuclear accumulation in normal growing conditions (þATP panels). Within

15 min after treatment with 2-deoxy-D-glucose and sodium azide, both constructs equilibrated between nucleus and cytoplasm (�ATP

panels). Scale bar: 10 mm. C) FRAP analysis of nucleus/cytoplasm exchange of expressed proteins. A typical TatGGG–GFP recovery curve

(filled red circles, t value of 53 � 1 seconds) is here compared with the slower passive diffusion kinetics of TatRRR–GFP (open red circles,

t value of 307 � 4 seconds). D) NLSGGG–GFP characteristic active import kinetics (filled dark-yellow circles, t value of 53 � 2 seconds)

compared with the slower recovery kinetics of NLSRRR–GFP (open dark-yellow circles, t value of 242 � 8 seconds). All recovery curves

were normalized by prebleaching values: the asymptotic value around 1 indicates that no immobile fraction of molecules can be detected

in the bleached compartment. Overall t values are reported in Table 1.

Traffic 2008; 9: 528–539 531

In Vivo Study of Tat Peptide-Targeting Properties

because a significant variation of this affinity could be at

the basis of the observed emergence of Tat–NLS proper-

ties in vivo.

We experimentally confirmed a large variation of the

binding affinity between Tat and RNAs when RRR is

replaced by GGG by an in vitro binding assay. In detail,

we used tetramethylrhodamine (TAMRA)-labeled peptides

and AlexaFluor647-labeled random oligo-RNAs. These

fluorophores were selected owing to their high FRET

efficiency. Increasing amounts of labeled peptide were

added to an RNA-containing solution, and the fluorescence

emission spectrum was recorded following excitation of

the ‘donor’ species (TAMRA-labeled peptide). In this way,

peptide–RNA interactions could be directly assessed

in vitro by FRET signal analysis (see eqn 1 in Materials

and Methods): this approach allowed us to quantitatively

compare TatRRR and TatGGG sequences for their ‘overall’

RNA-binding affinity. The substitution of the three

C-terminal arginines with glycines leads to a dramatic

decrease in FRET efficiency (Figure 3), i.e. in the Tat

peptide-binding affinity for RNA molecules. Ribonuclease

treatment (black box in Figure 3) brought the FRET signal

back to zero, further demonstrating the direct involvement

of RNAs in the observed interaction. As expected, addition

of a TAMRA-labeled glycine residue (black dots) did not

lead to a FRET signal. Our results are consistent with

analogous measurements reported so far in the literature

(7,8) and confirm the impact of the RRR/GGG substitu-

tion on RNA affinity of the peptide under study.

Single-point mutagenesis on TatRRR sequence:

modulation of nuclear import properties

The above set of experiments showed that the ‘RRR’

residues impair the potential nuclear-targeting properties

of the Tat sequence. Because the replacement of ‘RRR’

residues with ‘GGG’ allowed us to switch on the Tat

peptide nuclear import properties, we tried to tune them

by varying the number of positively charged C-terminal

arginine residues. To this aim, we engineered two novel

mutants in which the ‘RRR’ residues were replaced by

‘RRG’ and ‘RGG’, respectively (the corresponding se-

quences are given in Table 1). Hence, we studied their

intracellular localization and transport properties by confo-

cal imaging and FRAP analysis and compared them with

actively imported TatGGG–GFP (and NLSGGG–GFP) and

passively diffusing TatRRR–GFP (and NLSRRR–GFP). Nota-

bly, the one-by-one addition of glycine residues to the

wild-type Tat sequence gradually increases the effective

nuclear accumulation of constructs in comparison with

TatRRR, while it leads to a decrease in their nucleolar

localization (Figure 4A,B). We addressed the quantitative

aspects of the nucleus–cytoplasm exchange by means of

FRAP measurements: both bleaching time and sampling

Figure 3: Monitoring peptide–RNA interaction by FRET.

Increasing amounts of TAMRA-labeled peptide were added to

a solution containing AlexaFluor647-labeled random oligodeoxy-

nucleotides, and fluorescence emission spectrum was recorded,

as described in Materials and Methods. Complex formation

between Tat peptides and RNAs was monitored through FRET

calculation (eqn 1). Here, N_FRET is plotted against peptide

concentration. Wild-type Tat peptide (red dots) shows a much

higher affinity for RNA compared with TatGGG mutant (dark-yellow

dots), as highlighted by the hyperbolic fitting (solid lines). Ribonu-

clease addition abolishes the FRET signal (dots highlighted in the

black box), demonstrating the specificity of peptide–RNA interac-

tion. We performed similar measurements with TAMRA–glycine

(black dots): as expected, we observed no FRET signal.

Figure 4: Analysis of intracellular localization of Tat-derived

mutants. A) Ratio between nucleoplasmic and cytoplasmic

fluorescence (red symbols, corresponding to the Keq values

reported in Table 1) is here plotted together with the ratio between

nucleolar and nucleoplasmic fluorescence (black symbols). The

latter parameter indicates the extent of nucleolar accumulation of

the expressed protein. B) Intracellular distribution of Tat-derived

mutants imaged using the confocal microscope. Scale bar: 10 mm.

532 Traffic 2008; 9: 528–539

Cardarelli et al.

rate were tailored to the speed of fluorescence recovery of

the tested protein. Collected recovery curves (shown in

Figure 5A) and the derived shuttling time constants (Table

1) clearly show that the addition of glycine residues can

progressively accelerate the intercompartment mobility of

TatRRR-derived mutants. We shall argue that this behavior

stems from the gradual change from passive diffusion to

active import. kAT and kDN values provide a quantitative

description of this trend (Figure 5B,C; Table 1) and are

higher for actively imported species (TatGGG–GFP and

NLSGGG–GFP) and lower for the passively diffusing ones

(TatRRR–GFP and NLSRRR–GFP). Under the assumption

that kDN values of GFP and TatRRR–GFP reflect the passive

diffusion dynamics of the totally free and totally bound

proteins, respectively, the kDN values found for the other

Tat constructs were used to obtain the molar fractions of

bound proteins in the nucleus (Table 1).

Disk FRAP measurements: protein mobility

within the cytoplasm

These results provide strong indications of the relevance of

the TatRRR–peptide interactions in determining the overall

construct properties: in particular, we showed how the

mechanism of TatRRR peptide nuclear entry strongly de-

pends on the presence and the number of positively

charged C-terminal arginine residues.

Figure 5: Kinetic parameters for all Tat-derived mutants. A) Kinetics of nucleoplasmic FRAP for all Tat-derived mutants. t values

(mean � SD) are 53 � 1 seconds (TatGGG–GFP; the same of Figure 2C), 108 � 3 seconds (TatRGG–GFP), 190 � 5 seconds (TatRRG–GFP)

and 307 � 4 seconds (TatRRR–GFP, Figure 2C). All FRAP curves are normalized by prebleaching values; the asymptotic value of 1 indicates

that no immobile fraction of proteins is present in the nuclear compartment. t values for the whole population of analyzed cells are reported

in Table 1. B–D) Averaged kAT, kDN and DC values for Tat-derived mutants (red symbols) and SV40-derived mutants (yellow symbols). kATand kDN transport parameters have been derived from nuclear FRAP analysis, as described in Materials and Methods. The effective

cytoplasmic diffusion coefficients (DC) have been derived from disk bleaching measurements, as described in Materials and Methods.

Traffic 2008; 9: 528–539 533

In Vivo Study of Tat Peptide-Targeting Properties

To isolate the impact of these interactions from nuclear

envelope permeation, we investigated the intracytoplasmic

diffusion properties of constructs. Following the method

described in Cardarelli et al. (20), wemeasured the effective

cytoplasmic diffusion coefficient (DC) and the value of the

bleaching parameter K0 (related to the bleaching depth) for

all the tested proteins. As could be expected, the sub-

stitution of the C-terminal ‘RRR’ stretch with ‘GGG’ leads to

a considerable increase in Tat peptide intracytoplasmic

diffusivity: DC values are 5.9 � 1.5 mm2/second for

TatRRR–GFP and 9.8 � 1.3 mm2/second for TatGGG–GFP

(Figure 5D; Table 1). This large change confirms that the

presence of the ‘RRR’ stretch leads to increased molecular

interactions with RNAs and other intracellular moieties that

decrease protein diffusivity. At the same time, the almost

doubling of the diffusion coefficient of TatGGG–GFP with

respect to isolated GFP is consistent with the presence of

interactions with molecular components that are involved in

determining its nuclear active translocation. In fact, active

import benchmarks showed similar DC values (11.9 �2.6 mm2/second for NLSEDP–GFP and 10.8 � 1.6 mm2/sec-

ond for NLSGGG–GFP, respectively).

The one-by-one mutation of arginine residues to glycine

progressively increases protein intracytoplasmic diffusion:

we obtained DC ¼ 6.1 � 0.6 mm2/second for TatRRG–GFP

and 6.9 � 1 mm2/second for TatRGG–GFP (the overall data

plot is shown in Figure 5D).

Line FRAP method: seeking the mechanism

of nuclear entry

The uniform disk bleach method is a very effective tool for

measuring diffusion coefficients within the cytoplasm but

is not suitable for measurements in the nucleus. Indeed,

the lower volume of this compartment leads to consider-

able loss of overall fluorescence upon photobleaching and

prevents its use to evaluate diffusion parameters. In order

to do so, we took another approach based on the line scan

of the cell. In this experiment, we bleached a thin stripe

across both the nucleus and cytoplasm for 90 milliseconds

with laser light at full power and imaged the recovery

process at high speed along the same line (Figure 6A). The

characteristic recovery time was calculated by fitting

fluorescence recovery to double exponential equations

(Figure 6B) and taking the slower time constant as the

representative one, as described inMaterials and Methods.

This approach allowed us to compare the protein diffusion

properties in the nucleus and the cytoplasm. By taking the

ratio (RM parameter in Table 1) between nuclear and

Figure 6: Line FRAP measurement. A) The cell is repeatedly imaged along the dotted line at high frequency (400 Hz). Bleaching is

performed along a strip covering both the nucleus and cytoplasm for approximately 90 milliseconds using 488-nm laser at full power.

Fluorescence recovery is thenmonitored by imaging at 400 Hz for about 2.5 seconds. Collected data are analyzed as described inMaterials

and Methods. Scale bar: 10 mm. B) Nuclear (green dots) and cytoplasmic (red dots) fluorescence recovery from the cell shown in (A).

At this temporal resolution, the difference in mobility between the two compartments is clearly discernible: photobleaching within the

cytoplasm is larger and the corresponding recovery is slower, indicating lower mobility. C) Cumulative results. The ratio RM (mean � SD)

between nuclear and cytoplasmic recovery rates (slower time constant from double exponential fit, seeMaterials and Methods) is plotted

for all the analyzed constructs. Passively diffusing proteins are characterized by a RM value around 1, while actively imported proteins

yielded a RM value below 1.

534 Traffic 2008; 9: 528–539

Cardarelli et al.

cytoplasmic characteristic recovery times, we can esti-

mate the diffusion coefficient ratio of the two compart-

ments (DN/DC) and highlight the differences in protein

mobility between these two compartments. Figure 6C

shows that the proteins that are supposed to cross the

nuclear membrane by passive diffusion (i.e. GFP, TatRRR–

GFP and NLSRRR–GFP) showed no significant difference in

mobility between the nucleus and the cytoplasm, corres-

ponding to RM values around unity (Table 1). On the

contrary, the proteins that enter the nucleus by active

import (NLSEDP–GFP, NLSGGG–GFP and TatGGG–GFP)

unambiguously showed a larger diffusivity in the nucleus

than in the cytoplasm (RM < 1). These results will be

correlated to the molecular models proposed for the

nuclear import of the tested proteins (see Discussion).

Discussion

Recently, we showed that the mechanism driving in vivo

TatRRR nuclear permeation is passive diffusion, while

interaction with import carriers plays no detectable role

(20). These findings appear in contrast with previous

in vitro studies that had highlighted the relevance of active

processes in TatRRR-driven nuclear import (11,12). In

particular, Truant and Cullen showed that the TatRRRpeptide directly interacts with Imp-b in vitro and that

Imp-b is necessary and sufficient for the nuclear import

of TatRRR into isolated nuclei (11). We discussed these

findings and emphasized the impact on in vitro studies of

the absence of a diverse range of biomolecules that are

physiologically present in live cells. In fact, among these

molecules, some may bind to the TatRRR peptide and

compete with and even prevent TatRRR interaction with

Imp-b (21).

The SV40 peptide is an accepted functional NLS, as seen

also in our in vivo studies where it was actually used as an

active transport benchmark (20). A careful comparison

between Tat and SV40 peptides does show significant

sequence similarities (Table 1) with a notable difference in

the last three C-terminal amino acids. In fact, the latter are

positively charged in the TatRRR peptide (RRR) but are

mainly hydrophobic in SV40EDP (EDP). Moreover, these

three residues are not an integral part of the NLS of SV40,

and the remaining eight amino acid sequence (YPKKKRKV)

fully retains active import properties (26). These consid-

erations prompted us to investigate the impact of these

three amino acids on the transport properties of the Tat

peptide. To this end, we engineered fluorescent mutants

in which the terminal ‘RRR’ of the Tat sequence was

replaced by the neutral triplet ‘GGG’, while the remaining

sequence (YGRKKRRQ) was left unaltered. As previously,

we investigated the intracellular localization of these

proteins by in vivo confocal imaging. Remarkably,

TatGGG–GFP4 was virtually exclusively localized within the

nucleus (Figure 1C), even if its molecular weight exceeds

the diffusion limit through the NPC. This behavior is in full

contrast with TatRRR–GFP4 (completely excluded from the

nucleus) and consistent with active transport being the

dominant nucleus-to-cytoplasm transfer mechanism for

the TatGGG peptide. Furthermore, TatGGG–GFP (similar to

the active import benchmark SV40EDP–GFP) was not im-

ported into the nucleus upon energy depletion (Figure 2B).

This is once again in contrast with the behavior of TatRRR–

GFP (and isolated GFP). In fact, our published data show

that subcellular localization of TatRRR–GFP is not affected

by the same treatment (i.e. relocalization was not ob-

served) (20). Altogether, these findings indicate that the

NLS function is suppressed in wild-type Tat peptide, but

can be made available in engineered mutants. The Tat

peptide NLS can be associated to the first eight amino

acids of the sequence. Remarkably, the ratio between

nuclear and cytoplasmic fluorescence (Keq) of TatGGG–GFP

is very close to that of SV40GGG–GFP that can be consid-

ered its active import benchmark. Consistently and sym-

metrically with these results, we observed that when the

C-terminal ‘EDP’ residues of the functional NLS of SV40

are replaced by the ‘RRR’ ones, active nuclear import is

almost completely suppressed (with a concomitant

increase in nucleolar stain) and nucleus/cytoplasm shut-

tling kinetics is considerably slowed down (Figure 2B,D).

These findings lead to the conclusion that the first eight

residues of Tat peptide can operate as a NLS, but the

remaining three arginine residues hinder active transport

by promoting binding to intracellular moieties. Among

these, RNAs play an important role. In fact, Tat is known

to interact with negatively charged biomolecules such as

RNAs (7–10). Here, by an in vitro binding assay (Figure 3),

we showed that the substitution of the three C-terminal

arginines with glycines leads to a dramatic decrease in

Tat peptide-binding affinity for a population of random

RNA molecules. Our data show the role of unspecific

interactions between Tat and RNAs that can take place

within a living cell where TAR RNA is not present. The

marked decrease found in affinity upon ‘RRR’ replace-

ment by ‘GGG’ shows that interaction with RNAs can

indeed play a significant role in the observed modulation

of intracellular localization and trafficking properties as

described above.

In agreement with our findings, Friedler et al. reported that

a functional backbone mimetic of the Tat arginine-rich motif

was capable to interact with both import carriers and RNA

by in vitro assays (21). RNAs, however, are not the only

intracellular moieties that can influence Tat peptide proper-

ties. For example, it has been reported that Tat peptide-

comprising truncations of full-length Tat are able to target

the microtubule network (24). Thus, tubulin may represent

another moiety that modulates Tat trafficking upon binding.

Preliminary results using the microtubule-damaging drug

nocodazole, however, showed no effect on TatRRR–GFP

subcellular localization and dynamics (data not shown).

Even if we cannot exclude that microtubule assembly and

Traffic 2008; 9: 528–539 535

In Vivo Study of Tat Peptide-Targeting Properties

Tat are related, our data suggest that Tat–tubulin interaction

is not relevant in determining Tat transport biology, in

agreement with other results on Tat–tubulin interaction (28).

The picture of two competing equilibria was further probed

by varying the number of arginine residues. As expected,

from our analysis, we were able to fully tune the nuclear

import mechanism of Tat peptide by replacing one by one

the positively charged arginine residues with uncharged

glycine ones. This induced an increase in nuclear accumu-

lation of TatRRR-derived mutants (Keq in Table 1) and pro-

gressively accelerated the rate of their nucleus–cytoplasm

exchange.

These findings confirm a model of nucleus–cytoplasm

exchange of fluorescent constructs that takes into account

the competing binding with active import carriers (e.g.

importins) and other biomolecules (e.g. RNA). The interac-

tion pattern is schematized in Figure 7: ‘P’ labels the free

construct, ‘PA’ the complex with active import carriers and

‘PB’ the complex with other intracellular moieties. Our

findings show that P and PB diffuse passively across the

nuclear envelope, each species with its own diffusion

characteristics. In contrast, PA is actively imported into

the nucleus where it dissociates into its two components,

in line with a characteristic NLS behavior (17,29). The

overall observed permeation phenomenology is the result

of these processes weighted according to the mole

fraction of the different species (for further details, see

Supplementary Information).

The analysis of FRAP data according to this model allows

us to calculate the transport parameters associated to

nucleus-to-cytoplasm passive diffusion (kDN) and active

nuclear import (kAT). Not surprisingly, we observed that

passing from TatGGG–GFP to TatRRR–GFP one residue at

a time, both kAT and kDN values decrease (Figure 5B,C).

This shows the loss of nuclear import capability and

a concomitant increased affinity for other molecular com-

ponents. Assuming GFP and TatRRR–GFP as references of

totally free and totally bound molecules, respectively, we

were also able to estimate the mole fraction of bound

protein in the nucleus (Table 1). Note that the mole fraction

is not linearly related to the number of arginine residues,

suggesting co-operative behavior of these charged amino

acids in determining molecular interactions.

Binding can be experimentally monitored by measuring

intracytoplasmic diffusion properties. This was performed

by disk photobleaching FRAP measurements, a powerful

tool to investigate even small differences in diffusivity. We

found that sequential arginine replacement does modulate

the intracytoplasmic absolute diffusion coefficient (DC).

Consistent with our hypothesis of arginine-promoted bind-

ing properties, TatRGG, TatRRG and TatRRR sequences show

progressively and considerably reduced diffusion coeffi-

cient values (Figure 5D and Table 1).

A protein with NLS capability should be characterized by

a lower mobility in the cytoplasm than in the nucleoplasm

owing to the larger number of binding interactions [e.g.

with import carrier(s) and with other related biomolecules].

We investigated this property by means of the line FRAP

method that allowed us to estimate the ratio between

cytoplasmic and nuclear mobility RM (Figure 6; Table 1).

Interestingly, TatGGG–GFP, NLSEDP–GFP and NLSGGG–GFP

are characterized by RM values smaller than 1, as expected

for proteins undergoing receptor-mediated import to the

nucleus (17,29). Conversely, GFP, TatRRR–GFP and

NLSRRR–GFP show RM close to unity, in agreement with

their full passive diffusion behavior.

In conclusion, we have shown that the Tat peptide

sequence possesses a dual functionality. It includes

a NLS together with an additional competing functionality

leading to binding to intracellular moieties, including RNAs

(20,21). Our findings highlight the impact of the competi-

tion between two functions and explain contrasting results

reported in the literature. In this way, they provide a unified

and coherent interpretation of Tat peptide intracellular

transport properties. We believe that these results can

be useful for the rational design of new molecular carriers

Figure 7: Model of nucleocytoplasmic exchange. The free

NLS-tagged construct (P) in the cytoplasm can bind to both import

carriers (PA) and other biomolecules (PB). We assume that P and

PB shuttle across the nuclear envelope (NE) by reversible passive

diffusion (bidirectional black arrows), each species with its proper

diffusion characteristics. Conversely, the PA complex is actively

imported into the nucleus (red arrow) where it dissociates out into

the two components: this energy-consuming and irreversible

process leads to the accumulation of the NLS-tagged construct

in the nuclear compartment. The directionality of the active import

process is determined by asymmetric distribution of RanGTP

between compartments (19).

536 Traffic 2008; 9: 528–539

Cardarelli et al.

for the controlled delivery of cargoes to specific subcellular

domains, particularly the nucleus.

Materials and Methods

Synthetic compoundsThe C-terminus-labeled peptides used in this study were purchased from

Sigma-Genosys. Their amino acidic sequence is reported in Figure 3.

Tetramethylrhodamine was chosen for peptide labeling.

AlexaFluor647-labeled oligo-RNAs were purchased from Invitrogen–

Molecular Probes. These are random RNA sequences of nine oligodeoxy-

nucleotides labeled at the 50 end.

For all spectroscopic measurements (see below), peptides and RNAs were

resuspended in a suitable reaction buffer (70 mM NaCl, 10 mM Tris–HCl

pH ¼ 7.5, 0.2 mM ethylenediaminetetraacetic acid and 5% glycerol).

The reaction between the succinimidyl ester derivative of TAMRA (Invitrogen–

Molecular Probes) and glycine (Sigma) has been performed in citrate buffer

at pH ¼ 9.

In vitro spectroscopic FRET measurementsFluorescence spectra were recorded at 238C on a Cary Eclipse spectroflu-

orometer (Varian) by setting the excitation and emissionmonochromator slits

to 5 and 10 nm, respectively; the scanning speed was set to 120 nm/min,

the data resolution to 1 nm and the time collection average to each

wavelength interval to 0.5 seconds. Emission spectra were recorded by

exciting at 520 nm and collecting the fluorescence between 550 and

750 nm. First, a fluorescence emission measurement was performed on

a solution of labeled RNA alone. Thus, increasing amounts of TAMRA-labeled

peptide were added, and the emission spectrum was recorded. Complex

formation between Tat peptides and RNAs was monitored through FRET

calculation by using the emission counts collected at 666 nm after excitation

at 520 nm. Normalized FRET counts (N_FRET) were calculated by

N FRET ¼ ðF666 � FACC � FDONÞFACC

½1�

where F666 are the overall counts recorded at 666 nm, FACC are the counts

recorded when only the acceptor is present in solution at 666 nm and

FDON are the predicted counts because of donor emission at 666 nm,

respectively. FDON is calculated each step from the ratio between the

fluorescence emission at 578 and 666 nm detected for the donor alone.

Finally, N_FRET is plotted against the donor concentration in fluorescence

units (counts/mL), and fitted to an hyperbolic function, as shown in Figure 3.

The dilution factor was considered for the calculation of both N_FRET and

donor concentrations.

DNA manipulations: cloning and mutagenesisPlasmids encoding for TatRRR–EGFP and NLSEDP–EGFP were described in

our previous work (20). These plasmids were used as templates to create

all mutants mentioned in this work, and mutations were introduced by site-

directed mutagenesis using a QuickChange kit (Stratagene).

The following primers (Sigma-Genosys) were used for mutagenesis reac-

tions: 50 GAA GCG GAG ACA GGA AGA CCC AAA GCT TAT AGT GAG C 30

for TatEDP–GFP; 50 GAA GCG GAG ACA GGG AGG AGG AAA GCT TAT AGT

GAG C 30 for TatGGG–GFP; 50 GAA GCG GAG ACA GCG AGG AGG AAA GCT

TAT AGT GAG C 30 for TatRGG–GFP and 50 GAA GCG GAG ACA GCG ACG

AGG AAA GCT TAT AGT GAG C 30 for TatRRG–GFP. Antisense primers had

respective reverse complementary sequences.

The TatGGG–E1GFP–EGFP–tdTomato construct was obtained by subcloning

E1GFP–EGFP–tdTomato in TatGGG–GFP plasmid cut byHindIII–XbaI in order

to remove the EGFP sequence. The cloning strategies used for the

plasmids encoding for E1GFP (an EGFP containing the mutations T65S

T203Y), EGFP and the dimeric red fluorescent protein tdTomato were

described in detail in Cardarelli et al. (20).

NLSRRR–EGFP plasmid was constructed by multiple point mutation reac-

tions starting from Tat11–EGFP sequence. The primer used for multiple

mutagenesis was 50 GGA TCC ATG TAT CCC AAG AAG AAG CGG AAA

GTG CGA CGA AGA 30. Subsequently, to obtain NLSGGG–GFP construct,

we used NLSRRR–EGFP as template and the following sense primer: 50 GAA

GCG GAA AGT GGG AGG AGG AAA GCT TAT AGT GAG C 30. Antisense

primer had respective reverse complementary sequence.

The expression vector encoding for tdTomato protein was generated by

polymerase chain reaction (PCR) amplification of tdTomato template

(obtained from Roger Y. Tsien’s laboratory) and was inserted into EcoRI–

XbaI sites of pcDNA3.1(þ) vector (Invitrogen). To obtain the construct

pE1GFP, the EGFP template (pEGFP-N1; obtained from Clontech) was

amplified by PCR introducingHindIII site in 50 and EcoRI sites in 30 extremity

and it was inserted into pcDNA3.1(þ) vector. By using two distinct steps

of site-directed mutagenesis, we introduced the mutations T65S T203Y

using respectively the appropriate primers 50 GTG ACC ACC CTG TCC TAC

GGC GTG CAG 30 and 50 CAA CCA CTA CCT GAG CTA CCA GTC CGCCCT

GAG 30.

Cell culture and transfectionsChinese Hamster Ovary (CHO-K1) cells were provided from the American

Type Culture Collection (CCL-61; ATCC) and were grown in Ham’s F12K

medium supplemented with 10% of fetal bovine serum (FBS) at 378C and

in 5% CO2.

Transfections were carried out using Lipofectamine reagent (Invitrogen)

according to the manufacturer’s instructions. For live imaging, 12 � 104

cells were plated 24 h before transfection onto 35-mm glass bottom dish

(WillCo-dish GWSt-3522).

For energy depletion studies, cells were incubated for 30 min at 378C in

glucose-free DMEM (Invitrogen) containing 10% FBS and supplemented

with 10 mM sodium azide and 6 mM 2-deoxy-D-glucose (Sigma). After this

treatment, energy depletion medium has been replaced by DMEM used in

normally growing conditions.

Fluorescence microscopy and image analysisCell fluorescence was measured using a Leica TCS SP2 inverted confocal

microscope (Leica Microsystems AG) interfaced with a diode laser (Hama-

matsu Photonics) for excitation at 403 nm, with an Ar laser for excitation at

458, 476, 488 and 514 nm and with a HeNe laser for excitation at 543 and

633 nm. Glass bottom Petri dishes containing transfected cells were

mounted in a thermostated chamber at 378C (Leica Microsystems) and

viewed with a � 40 1.25 NA oil immersion objective (Leica Microsystems).

The images were collected using low excitation power at the sample (10–

20 mW) andmonitoring the emission by means of an Acousto Optical Beam

Splitter (AOBS)-based built-in detectors of the confocal microscope. The

following collection ranges were adopted: 400–450 nm (Hoechst), 500–

550 nm (EGFP and E1GFP) and 580–680 nm (tdTomato). All images have

been subtracted by background signal. Data were analyzed by LEICA IMAGING

package version 2.61, by ORIGIN-PRO7 package and by an implementation of

IGOR-PRO software package.

In vivo FRET measurementsIntegrity of the 110-kDa GFP4 cargo construct was verified in living cells by

FRET technique. We took E1GFP as the donor fluorophore and tdTomato

as the acceptor fluorophore (the corresponding imaging settings were

described above). FRET signal was measured by collecting the tdTomato

fluorescence after excitation at 403 nm. We generated an image of

sensitized emission FRET (corrected pixel-by-pixel for donor and acceptor

bleed-through) with the IMAGEJ PixFRET plug-in (30).

Traffic 2008; 9: 528–539 537

In Vivo Study of Tat Peptide-Targeting Properties

FRAP experiments

Single-point bleach to study nucleus/cytoplasm

shuttlingEach FRAP experiment started with a four-time line-averaged image

(prebleach) of the cell followed by a single-point bleach (nonscanning)

near the center of the nucleus with laser pulse at full power (150 mW) for

the minimum time required to photobleach most of the nuclear fluores-

cence. Fluorescence recovery was measured by starting a time-lapse

acquisition within few milliseconds from the end of bleaching. Each

image of the recovery was four times line averaged, image size was

512 � 512 pixels and scan speed was usually set to 400 Hz. The pinhole size

was set to the optimal value of 1.0 Airy (corresponding to a 81.44-mm confocal

aperture).

According to the mathematical description of our nucleocytoplasmic

exchange model (Supplementary Information), collected FRAP curves for

both compartments were fitted to the monoexponential equations (eqns 15

and 16 in Supplementary Information):

FC ¼ FNC þ ðF0

C � FNC Þ � exp ð�t=tÞ

FN ¼ FNN þ ðF0

N � FNN Þ � exp ð�t=tÞ

where FC and FN are the fluorescence values at time t of cytoplasm and

nucleoplasm, respectively. The three fitting parameters in each equation

refer to the asymptotic fluorescence (FNC , FN

N ), the fluorescence immedi-

ately after bleaching (F0C , F

0N) and the time constant of exponential recovery

(t). Before fitting, the experimental values of fluorescence were normalized

by the fluorescence of the entire cell at the same time in order to minimize

the effect of cell motility and defocusing on the recovery curves and to

correct for bleaching caused by imaging (31). Moreover, data were

normalized by prebleach fluorescence values in order to verify the presence

of an immobile fraction of fluorescent molecules within the bleached

compartment.

The ratio between the protein concentration in the nucleoplasm and

cytoplasm at equilibrium (Keq) was determined by taking the ratio between

the nuclear (FpbN ) and cytoplasmic fluorescence (Fpb

C ) before bleaching, i.e.

Keq ¼ FpbN =Fpb

C . Nuclear volume (VN) was calculated by assuming an

ellipsoid shape for the nucleus with semiaxes dx, dy and dz by means of

the formula VN ¼ (4/3) � p � dx � dy � dz. The three axes were estimated

from confocal images of nucleus, and in most cases, we set dz equal to dy,

the smallest semiaxis in the horizontal plane.

Readers are referred to the Supplementary Information for equations 14,

17, 18 and 19 that were used to obtain the cytoplasmic volume VC, the

passive permeation parameters kDN and the variability range of the active

permeation parameter kAT from t, Keq and VN.

Uniform disk bleach to study protein mobility

within cytoplasmFRAP measurements were carried out as described in Cardarelli et al. (20).

Briefly, a circle area of 3.5-mm radius within the cytoplasm was photo-

bleached using 476-, 488- and 514-nm light at maximum power (87, 350,

350 mW, respectively), and fluorescence recovery was then monitored at

220-millisecond intervals for about 8 seconds using 488-nm excitation at

10 mW. Recovery curves have been normalized by a prior time-lapse

imaging of the same region without the bleaching step to verify the

presence of an immobile fraction of fluorescent molecules within cyto-

plasm. Collected data were analyzed as described in Cardarelli et al. (20)

from which we derived the effective diffusion coefficient in the cytoplasm

(DC) and the bleaching parameter [K0, related to the specific bleach rate of

the fluorophore and zoom setting of the Confocal Scanning Laser Micro-

scope (CSLM) (32)].

Line bleach to study differences in protein mobility

between compartmentsIn these measures, we performed the bleach along a line (covering both

nucleus and cytoplasm) for 90 milliseconds with 488-nm laser pulse at full

power. Fluorescence recovery was imaged along the same line at high

frequency (400 Hz). Collected data were corrected for background and

normalized by a prior time-lapse imaging of the same region without the

bleaching step. The recovery of all imaged cells was fitted to a double

exponential equation to provide an empirical quantification of the time

dependence of the process. We took the slower time constant value as the

representative dynamic parameter of intracompartment mobility because

it was calculated on a larger number of reliable recovery points. The dif-

ference in mobility between compartments was calculated by taking the

ratio between nuclear and cytoplasmic characteristic recovery rates (RM

parameter). Under the assumption that the mobility and diffusivity are

inversely related, we stated RM ¼ DC /DN where DN is the diffusion co-

efficient in the nucleus.

Acknowledgments

We thank Roger Y. Tsien (University of California, CA, USA) for kindly making

available the tdTomato template. Authors also gratefully acknowledge partial

financial support by the Italian Ministry for University and Research (FIRB no.

RBLA03ER38) and by Fondazione Monte dei Paschi di Siena.

Supplementary Material

Supplementary Information: Mathematical model of nuclear/cytoplasmic

exchange in a FRAP experiment.

Supplemental materials are available as part of the online article at http://

www.blackwell-synergy.com

References

1. Cullen BR. HIV-1 auxiliary proteins: making connections in a dying cell.

Cell 1998;93:685–692.

2. Emerman M, Malim MH. HIV-1 regulatory/accessory genes: keys to

unraveling viral and host cell biology. Science 1998;280:1880–1884.

3. Marcello A, Cinelli R, Ferrari A, Signorelli A, Tyagi M, Pellegrini V, Beltram

F, Giacca M. Visualization of in vivo direct interaction between HIV-1

TAT and human cyclin T1 in specific subcellular compartments by fluo-

rescence resonance energy transfer. J Biol Chem 2001;276:39220–

39225.

4. Raha T, Cheng SW, Green MR. HIV-1 Tat stimulates transcription

complex assembly through recruitment of TBP in the absence of TAFs.

PLoS Biol 2005;3: e44.

5. Brady J, Kashanchi F. Tat gets the ‘‘green’’ light on transcription

initiation. Retrovirology 2005;2:69.

6. Marcello M, Lusic M, Pegoraro G, Pellegrini V, Beltram F, Giacca M.

Nuclear organization and the control of HIV-1 transcription. Gene 2004;

326:1–11.

7. Calnan BJ, Biancalana S, Hudson D, Frankel AD. Analysis of arginine-

rich peptides from the HIV Tat protein reveals unusual features of RNA-

protein recognition. Genes Dev 1991;5:201–210.

8. Delling U, Roy S, Sumner-Smith M, Barnett R, Reid L, Rosen C,

Sonenberg N. The number of positively charged amino acids in the

basic domain of Tat is critical for trans-activation and complex formation

with TAR RNA. Proc Natl Acad Sci U S A 1991;88:6234–6238.

538 Traffic 2008; 9: 528–539

Cardarelli et al.

9. Weeks KM, Crothers DM. RNA recognition by Tat-derived peptides:

interaction in the major groove? Cell 1991;66:577–588.

10. Tao J, Frankel AD. Electrostatic interactions modulate the RNA-binding

and transactivation specificities of the human immunodeficiency virus

and simian immunodeficiency virus Tat proteins. Biochemistry 1993;

90:1571–1575.

11. Truant R, Cullen BR. The arginine-rich domains present in human im-

munodeficiency virus type 1 Tat and Rev function as direct importin beta-

dependent nuclear localization signals. Mol Cell Biol 1999;19:1210–1217.

12. Efthymiadis A, Briggs LJ, Jans DA. The HIV-1 Tat nuclear localization

sequence confers novel nuclear import properties. J Biol Chem 1998;

273:1623–1628.

13. Tyagi M, Rusnati M, Presta M, Giacca M. Internalization of HIV-1 tat

requires cell surface heparan sulfate proteoglycans. J Biol Chem 2001;

276:3254–3261.

14. Brooks H, Lebleu B, Vives E. Tat peptide-mediated cellular delivery:

back to basics. Adv Drug Deliv Rev 2005;57:559–577.

15. Duchardt F, Fotin-Mleczek M, Schwarz H, Fischer R, Brock R. A

comprehensive model for the cellular uptake of cationic cell-penetrating

peptides. Traffic 2007;8:848–866.

16. Dingwall C, Laskey RA. Nuclear targeting sequences – a consensus?

Trends Biochem Sci 1991;16:478–481.

17. Gorlich D. Transport into and out of the cell nucleus. EMBO J 1998;

17:2721–2727.

18. Pemberton LF, Paschal BM. Mechanisms of receptor-mediated

nuclear import and nuclear export. Traffic 2005;6:187–198.

19. Izaurralde E, Kutay U, von Kobbe C, Mattaj IW, Gorlich D. The asymmet-

ric distribution of the constituents of the Ran system is essential for

transport into and out of the nucleus. EMBO J 1997;16:6535–6547.

20. Cardarelli F, Serresi M, Bizzarri R, Giacca M, Beltram F. In vivo study of

HIV-1 Tat arginine-rich motif unveils its transport properties. Mol Ther

2007;15:1313–1322.

21. Friedler A, Friedler D, Luedtke NW, Tor Y, Loyter A, Gilon C.

Development of a functional backbone cyclic mimetic of the HIV-1

Tat arginine-rich motif. J Biol Chem 2000;275:23783–23789.

22. Henderson BR, Percipalle P. Interactions between HIV Rev and nuclear

import and export factors: the Rev nuclear localisation signal mediates

specific binding to human importin-beta. J Mol Biol 1997;274:693–707.

23. Fineberg K, Fineberg T, Graessmann A, Luedtke NW, Tor Y, Lixin R,

Jans DA, Loyter A. Inhibition of nuclear import mediated by the Rev-

arginine rich motif by RNA molecules. Biochemistry 2003;42:2625–2633.

24. deMareuil J, CarreM, Barbier P, Campbell GR, Lancelot S, Opi S, Esquieu

D, Watkins JD, Prevot C, Braguer D, Peyrot V, Loret EP. HIV-1 Tat protein

enhances microtubule polymerization. Retrovirology 2005;2:5.

25. Paine PL, Moore LC, Horowitz SB. Nuclear envelope permeability.

Nature 1975;254:109–114.

26. Kalderon D, Roberts BL, Richardson WD, Smith AE. A short amino acid

sequence able to specify nuclear location. Cell 1984;39:499–509.

27. Roth DM, Moseley GW, Glover D, Pouton CW, Jans DA. A microtubule-

facilitated nuclear import pathway for cancer regulatory proteins.

Traffic 2007;8:673–686.

28. Chen D, Wang M, Zhou S, Zhou Q. HIV-1 Tat targets microtubules to

induce apoptosis, a process promoted by the pro-apoptotic Bcl-2

relative Bim. EMBO J 2002;21:6801–6810.

29. Nigg EA. Nucleocytoplasmic transport: signals, mechanisms and

regulation. Nature 1997;386:779–787.

30. Feige JN, Sage D, Wahli W, Desvergne D, Gelman L. PixFRET, an

ImageJ plug-in for FRET calculation that can accommodate variations

in spectral bleed-throughs. Microsc Res Tech 2005;68:51–58.

31. Phair RD, Misteli T. High mobility of proteins in the mammalian cell

nucleus. Nature 2000;404:604–609.

32. Braeckmans K, Peeters L, Sanders NN, De Smedt SC, Demeester J.

Three-dimensional fluorescence recovery after photobleaching with the

confocal scanning laser microscope. Biophys J 2003;85:2240–2252.

Traffic 2008; 9: 528–539 539

In Vivo Study of Tat Peptide-Targeting Properties