Biochem. J. (2002) 367, 347357 (Printed in Great Britain)
Role of endocytosis in the internalization of spermidine-C2-BODIPY, a highly uorescent probe of polyamine transport
Denis SOULET*1, Laurence COVASSIN1, Mohammadi KAOUASS*2, Rene! CHAREST-GAUDREAULT, Marie AUDETTER and Richard POULIN*3
*Department of Anatomy and Physiology, Faculty of Medicine, Laval University, Quebec, Canada G1K 7P4, Molecular Endocrinology and Oncology Research Center, CHUL Medical Research Center (CHUQ), 2705 Laurier Blvd., Ste. Foy, Quebec, Canada G1V 4G2, Faculty of Pharmacy, Laval University, Quebec, Canada G1K 7P4, Department of Pharmacology, Laval University, Quebec, Canada G1K 7P4, RSt. Franc: ois dAssise Hospital Research Center (CHUQ), 10 rue de lEspinay, Quebec, Canada G1L 3L5, and Department of Medical Biology, Faculty of Medicine, Laval University, Quebec, Canada G1K 7P4
The mechanism of transmembrane polyamine internalization in mammalian cells remains unknown. A novel uorescent spermidine conjugate [Spd-C -BODIPY ; N-(4,4-diuoro-5,7# dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionyl)-Nh-oS[spermidine-(N%-ethyl)]thioacetylqethylenediamine] was synthesized from N%-(mercaptoethyl)spermidine by a simple, one-step coupling procedure. In Chinese-hamster ovary (CHO) cells, SpdC -BODIPY accumulation was inhibited by exogenous # putrescine, spermidine and spermine, was subject to feedback transport inhibition and was up-regulated by prior polyamine depletion achieved with a biosynthetic inhibitor. Probe internalization was decreased by about 85 % in a polyaminetransport-decient CHO mutant cell line. Using confocal laser scanning uorescence microscopy, internalized Spd-C -BODIPY # was concentrated in vesicle-like structures similar to the recycling endosomes observed with uorescent transferrin, which partly co-localized with the polyamine probe. In yeast, Spd-C # BODIPY uptake was stringently dependent on receptor-mediated endocytosis, as determined with a mutant defective in early-
endosome formation. On the other hand, Spd-C -BODIPY did # not mimic the substrate behaviour of natural polyamines in yeast, as shown by the lack of correlation of its uptake characteristics with the phenotypes of mutants defective in either polyamine transport or biosynthesis. These data suggest that endocytosis might be an integral part of the mechanism of polyamine transport in mammalian cells, and that the mammalian and yeast transport systems use qualitatively dierent transport mechanisms. However, the current data do not rule out the possibility that sequestration of the probe into vesicular structures might be secondary to its prior uptake via a classical plasma membrane carrier. Spd-C -BODIPY, a highly sensitive # probe of polyamine transport with biochemical parameters qualitatively similar to those of natural polyamines in mammalian cells, should be very useful for dissecting the pathway responsible for polyamine internalization. Key words : confocal microscopy, diuoromethylornithine, ow cytometry, membrane transport, yeast.

INTRODUCTION

Polyamines are small ubiquitous molecules involved in various functions, including macromolecular synthesis, ion channel gating and the post-translational modication of eukaryotic initiation factor eIF-5A [1,2]. Intracellular polyamine pools are narrowly controlled by the enzymes involved in their biosynthesis and degradation [1,2], as well as by internalization via specic plasma-membrane carriers [3,4]. High-anity polyamine transport can be demonstrated in a wide spectrum of tissues and cell lines, and is activated upon entry into the cell cycle, cellular transformation by oncogenes, and various hormonal signals [3,4]. Polyamine-transport activity is negatively regulated by intracellular polyamines, as shown by its up-regulation upon polyamine depletion by agents such as -diuoromethylornithine (DFMO) [3,5], a suicide substrate of ornithine decarboxylase [1]. Antizymes are major factors involved in the acute inhibition of polyamine transport by newly internalized polyamines [6,7], although the mechanism of their interaction with the polyamine
carrier(s) is currently unknown. Polyamine uptake is membranepotential-dependent, Na+-independent and requires bivalent cations such as Ca#+ or Mg#+ [8], but its actual contribution to polyamine homoeostasis under physiological conditions is still uncertain. Prokaryotic polyamine carriers have been extensively characterized, and vacuolar polyamine transporters have recently been identied in the yeast Saccharomyces cere isiae [9]. However, little is known about the molecular structure of mammalian polyamine carriers. Genetic evidence indicates that at least two loci control polyamine transport [10], and an as yet unidentied human gene for polyamine transport transfected into a Chinesehamster ovary (CHO) mutant cell line could restore the polyamine-transport defect present in these cells [11]. Although mammalian polyamine carriers are highly specic for putrescine, spermidine and spermine, the polyamine-binding site of these putative proteins can accommodate substantial modications of the basic polyamine structure, as shown by the utilization of paraquat (1,1h-dimethyl-4,4h-bipyridinium) and

Abbreviations used : CHO, Chinese hamster ovary ; MANT, N-methylanthranylic acid ; Spd-MANT, N-ospermidine-[N4-(3-aminopropyl)]q anthranylamide ; Spd-C2-BODIPY, N-(4,4-diuoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionyl)-Nh-oS-[spermidine-(N4-ethyl)]thioacetylq ethylenediamine ; BODIPY2 FL iodoacetamide, N-(4,4-diuoro-5,7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionyl)-Nh-iodoacetylethylenediamine ; -MEM, minimal essential medium with modication ; DFMO, -diuoromethylornithine ; MFI, mean uorescence intensity. 1 These authors contributed equally to this work. 2 Current address : Laboratory of Molecular Biology, Clinical Research Institute of Montreal, 110 Pine Avenue West, Montreal, Quebec, Canada H2W 1R7. 3 To whom correspondence should be addressed, at the Molecular Endocrinology and Oncology Research Center (e-mail rpoulin!drs.crchul.ulaval.ca). # 2002 Biochemical Society

D. Soulet and others

methylglyoxal bis(guanylhydrazone) as substrates [3]. This property has previously been exploited for the biochemical characterization of the polyamine carrier. For instance, spermidine, spermine and norspermine were derivatized with "#&I-labelled 4azidosalicylic acid to generate photoreactive probes for the labelling of polyamine-binding proteins as potential candidates for the mammalian polyamine carriers [12]. The polyaminetransport system has also been used for the targeting of cytotoxic polyamine analogues [1] or chemotherapeutic agents conjugated to polyamines [13]. Conjugation of aromatic structures such as 4-azidosalicylic acid [12] and chlorambucil [13] to the polyamine core structure does not abrogate its strong binding to the polyamine carrier. This interesting property has recently been exploited for the design of uorescent probes to study polyamine transport. Monouoresceinyl adducts of spermidine and spermine substituted on one of the primary amino groups have been described by Aziz et al. [14], which displayed anity comparable with the parent polyamine and mostly accumulated in the cytoplasm of mammalian cells. Likewise, amidation of N-methylanthranylic acid (MANT) to an aminopropyl group of N%-(3-aminopropyl)spermidine led to an UV-excitable substrate for polyamine transport in various mammalian cells [13]. Interestingly, Nospermidine-[N%-(3-aminopropyl)]qanthranylamide (Spd-MANT) mainly accumulated into discrete subcellular structures reminiscent of endocytic vesicles [13]. Polyamines accumulated from exogenous sources exert regulatory eects on polyamine homoeostasis at levels representing only a minor fraction of the total polyamine pool [1517]. Likewise, indirect biochemical evidence suggests that the free (i.e. unbound) pools of spermidine and spermine would be restricted to less than 2 and 7 % of total content respectively [18]. Thus a better understanding of the mechanism of polyamine internalization and compartmentalization is required to understand the actual sites of action of these compounds under physiological conditions. In this report we describe the synthesis and characterization of a novel uorescent probe of polyamine transport, spermidine-C -BODIPY [Spd-C -BODIPY ; N-(4,4-diuoro-5,# # 7-dimethyl-4-bora-3a,4a-diaza-s-indacene-3-propionyl)-Nh-oS[spermidine-(N%-ethyl)]thioacetylqethylenediamine], obtained by grafting a highly sensitive uorophore to an N%-mercaptoethyl side arm extending from the spermidine backbone. Like SpdMANT [13], Spd-C -BODIPY mainly compartmentalizes into # vesicle-like intracellular structures and is excluded from the nucleus. Moreover, the internalization of Spd-C -BODIPY likely # proceeds via endocytosis of a complex formed between the probe and the polyamine transporter(s), rather than carrier-mediated inux.

Figure 1

Structures of Spd-C2-BODIPY and Spd-MANT
from NEN Life Science Products (Lachine, Quebec, Canada). Unless otherwise indicated, other biochemical and tissue culture reagents were from SigmaAldrich.
Synthesis of Spd-C2-BODIPY
N%-(Mercaptoethyl)spermidine was synthesized according to the method of Cohen et al. [19]. Spd-C -BODIPY (Figure 1) was # prepared by alkylation of the thiol group of N%(mercaptoethyl)spermidine with BODIPY FL iodoacetamide. Prior to conjugation of the uorophore, N%-(mercaptoethyl)spermidine was dissolved in an aqueous solution containing a 5-fold molar excess of dithiothreitol to ensure complete reduction of the thiol group. The mixture was left for 1 h at room temperature, and N%-(mercaptoethyl)spermidine was re-puried by ion-exchange chromatography as described above, followed by lyophilization. All following operations were carried out under subdued light. To 1 mol of a neutralized solution of N%-(mercaptoethyl)spermidine in 100 M Tris\HCl (pH 7.3) was then added 4 mol of BODIPY FL iodoacetamide (extemporaneously prepared as a 20 mM solution in DMSO), and the mixture was stirred overnight in the dark at room temperature. Solvents were removed by lyophilization, the foam was dissolved in deionized water, and the solution was passed through a PVDF syringe membrane lter (Millex-HV4, 0.45 m pore size ; Millipore, Bedford, MA, U.S.A.). The aqueous phase was washed with dichloromethane (HPLC grade) until the organic phase, containing unchanged BODIPY FL iodoacetamide, became colourless. Complete extraction of unchanged BODIPY FL iodoacetamide and formation of a polar BODIPYspermidine conjugate was conrmed by silica-gel TLC using dichloromethane\methanol (9 : 1, v\v) as an eluent. The aqueous phase containing Spd-C -BODIPY was reduced to a small volume # by lyophilization. The actual concentration of Spd-C -BODIPY # was then determined on an aliquot of the latter solution by derivatization with an o-phthaldialdehyde solution in water [400 mM potassium borate, 0.5 g\l o-phthaldialdehyde, 1 % methanol (v\v), 0.5 % (v\v) -mercaptoethanol and 0.1 % Brij 35 (v\v)] for 20 min in the dark at 37 mC [20], and determination of uorescence intensity using an SLM-AMINCO-Bowman AB2 spectrouorimeter (excitation wavelength, 350 nm), using spermidine as standard. Standardization of the solution was con-

EXPERIMENTAL Reagents

Bovine calf serum and Eagles minimal essential medium (with modication ; -MEM) were purchased from Wisent (St-Bruno, Quebec, Canada) and Gibco-BRL Life Technologies (Burlington, Ontario, Canada), respectively. ,--Diuoromethylornithine hydrochloride was generously provided by ILEX Oncology (San Antonio, TX, U.S.A.). o-Phthaldialdehyde was obtained from MAT Laboratories (Quebec City, Quebec, Canada). BODIPY2 FL iodoacetamide [N-(4,4-diuoro-5,7-dimethyl-4-bora-3a,4adiaza-s-indacene-3-propionyl)-Nh-iodoacetylethylenediamine] and Texas Red2- and BODIPY FL-labelled human serum transferrin as well as FITC were purchased from Molecular Probes (Eugene, OR, U.S.A.). [terminal-methylene-groups-$H(n)] Spermidine trihydrochloride (3.1i10& Ci\mol) was purchased
# 2002 Biochemical Society
Fluorimetry of polyamine transport with BODIPY-labelled spermidine
rmed by measuring the uorescence intensity of the BODIPY adduct (excitation wavelength, 493 nm), and the nal concentration of Spd-C -BODIPY was then adjusted to 1 mM. #
Synthesis of Spd-MANT and of monouoresceinylspermidine conjugates
Spd-MANT was synthesized by N,Nh-dicyclohexylcarbodiimide-coupled amide formation between anthranilic acid and N",N)-di(t-butoxycarbonyl)-N%-(3-aminopropyl)spermidine as described in [19]. A mixture of N"- and N)-(monouoresceinyl)spermidines was synthesized using FITC and puried by TLC on silica-gel plates [14].
twice with 1 ml of PBS. Coverslips were then inverted on a droplet of Hanks balanced salt solution. Intracellular uorescence was observed using an Axioskop microscope equipped with a Plan-Neouar 403\0.75 objective (Zeiss) and lter set (bandpass, 450490 nm ; lter threshold, 510 nm ; long pass, 520 nm). Images were obtained with a CCD-300T-RC camera (DAGE-MTI, Michigan City, IN, U.S.A.) coupled to the MetaMorph2 imaging system (version 3.5 ; Universal Imaging Corporation, West Chester, PA, U.S.A.).
FACS analysis of uorescent probe uptake
Both CHO cell lines were seeded at 2.5i10$ cells\well in 24-well culture plates. Cells were incubated for the indicated time interval with serum-free -MEM containing 1 M Spd-C -BODIPY or # 1 M Spd-MANT. Medium was then removed and cell monolayers were washed three times with 1 ml of PBS containing 1 mM spermidine. Cell cultures were then rinsed twice with 1 ml of PBS, and harvested after a 5-min incubation with pancreatin\EDTA\Hepes solution (2.5 g\l\1 mM\7.25 mM) in Hanks balanced salt solution. Cells were suspended in ice-cold -MEM and immediately processed for uorescence analysis by ow cytometry with an Epics Prole II cytouorometer (Coulter Corp., Miami, FL, U.S.A.). For Spd-C -BODIPY, the argon # cytouorimeter was tuned to 488 nm, using an FL1 photomultiplier (bandpass, 525p30 nm). For Spd-MANT, the He\Cd lamp of the cytouorimeter was tuned to 325 nm, using an FL2 photomultiplier. The mean value of uorescence intensity (MFI) distribution was recorded for the analysis of 1i10% cells\sample. MFI is the mean channel number recorded for the distribution of uorescence among the cells counted. To determine the eect of polyamine depletion and protein synthesis inhibition on Spd-C -BODIPY uptake, cells were rst # grown for 72 h in the presence of 5 mM DFMO or vehicle. Fresh serum-free -MEM containing 1 M Spd-C -BODIPY was then # added for a 4 h incubation, with or without 200 mM cycloheximide and\or 5 mM DFMO.

Cell lines and cell culture
The parental CHO cell line (CHO-TOR) and a polyaminetransport-decient subline (CHO-MG) isolated after chronic selection for growth resistance to methylglyoxal bis(guanylhydrazone) [10,21] were generously provided by Dr Wayne Flinto (Department of Microbiology and Immunology, University of Western Ontario, London, Ontario, Canada). Both lines were routinely grown in -MEM supplemented with 10 % CosmicTM calf serum (Hyclone, Logan, UT, U.S.A.) and antibiotics in a water-saturated 5 % CO atmosphere at 37 mC. # All yeast strains used in this study were haploid. The wild-type yeast (S. cere isiae) strain RH144-3D (MATa leu2 his4 ura3 bar1-1) and its RH266-1D (MATa leu2 his4 ura3 bar1-1 ts end3) mutant derivative defective in an early internalization step of endocytosis [2224] were kindly provided by Dr H. Riezman (Biozentrum, University of Basel, Basel, Switzerland). The DBY747 wild-type strain (MATa leu2 his3 ura3 trp1) and its mutant derivative DBY747spe2 (MATa leu2 his3 ura3 trp1 spe2-5 : : LEU2) carrying a chromosomal deletion of the SPE2 gene encoding S-adenosylmethionine decarboxylase have been described previously [23]. The W303 ptk2 : deletion mutant (MATa ade2 ura3 trp1 his3 leu2 ptk2 : : TRP1) was generated by the one-step disruption method [25]. Briey, the full PTK2 sequence, including 683 and 262 bp adjacent to the 5h and 3h ends of the coding region, respectively, was isolated as a BamHI fragment (3.4 kb) from a lambda PM-1436 S. cere isiae genomic clone from chromosome X (A.T.C.C. 70345) [26]. The BamHI fragment was subcloned into the same site of the Bluescript II KS(j) plasmid. The resulting plasmid pKSPTK2 was digested with HpaI and NdeI restriction enzymes to eliminate a 2631 bp fragment corresponding to the full sequence of the PTK2 open reading frame. A ScaINdeI fragment (3100 bp) bearing the TRP1 marker (from the pJG4-5 expression vector) was ligated into the blunt-end HpaI and NdeI sites of pKSPTK2. The resulting 3900 bp BamHI fragment, containing the TRP1 gene anked with PTK2 sequences ( ptk2 : : TRP1), was isolated and used to transform the wildtype haploid strain W303 (MATa ade2 ura3 trp1 his3 leu2) with subsequent selection on tryptophan-free synthetic dextrose (SD) medium [0.17 % yeast nitrogen base without amino acids or (NH ) SO , with 0.5 % (NH ) SO and 2 % dextrose added]. The %# % %# % yeast strains used in this study were routinely grown in YPD medium (1 % yeast extract, 2 % peptone and 2 % -glucose).

Confocal laser scanning uorescence microscopy
CHO-TOR cells grown on coverslips were rst incubated for 4 h with 1 M Spd-C -BODIPY and Texas Redtransferrin con# jugate (3 g\ml) in -MEM at 37 mC. Culture medium was then removed, and cells were washed three times with ice-cold PBS containing 1 mM spermidine, and then twice with ice-cold PBS. Specimens were visualized with a Bio-Rad MRC-1024 confocal imaging system equipped with a Kr\Ar laser and mounted on a Diaphot-TMD inverted microscope (Nikon). A 60i oilimmersion objective lens with a 1.4 numerical aperture was used for imaging. The photomultiplier gain was set at maximum, and the confocal aperture was adjusted for maximum resolution. Analysis of uorophore co-localization was performed using the LaserSharp software (version 3.0 ; Bio-Rad) with background subtraction.
Competition of spermidine uptake by uorescent polyamines
The ability of Spd-C -BODIPY, Spd-MANT and N%-(mercapto# ethyl)spermidine to compete for [$H]spermidine uptake was determined in CHO-TOR cells by a 20 min uptake assay in the presence of increasing concentrations of competitor, using 5 M [$H]spermidine as a substrate, as described in [27]. Data were averaged from three separate experiments with triplicate determinations of the Ki value for each experiment. The Km value of spermidine transport was determined by LineweaverBurk analysis as described in [27]. Ki values for inhibition of spermidine

Epiuorescence microscopy

Both CHO cell lines were grown on sterile glass coverslips in sixwell plates for 24 h. Cells were then incubated in the presence of Spd-C -BODIPY (1 M) for 24 h. Medium was removed, and # cell monolayers were washed three times with 1 ml of ice-cold Ca#+\Mg#+-free PBS containing 1 mM spermidine and then
uptake were determined using the ChengPruso equation [28] from the IC value derived by iterative curve tting of the &! sigmoidal equation describing the velocity of spermidine uptake in the presence of the respective competitor [5,27].
Uptake of Spd-C2-BODIPY in yeast
The wild-type DBY747 strain and its spe2 mutant derivative were incubated at 30 mC for 96 h in H medium, an amine-free minimal medium [29] supplemented with 5 mg\ml tryptophan with or without 5 mg\ml leucine, respectively. Cell growth had completely stopped in the spe2 mutant at the end of the incubation period, indicating the complete depletion of spermidine and spermine [23,29]. DBY747 cells and the spe2 mutants were then pre-incubated for 10 min at 30 mC in the appropriately supplemented H medium at 3i10( cells\ml. The wild-type W303 strain and its ptk2 mutant derivative were grown until mid-exponential stage in YPD at 30 mC, and 3i10( cells were then transferred into 1 ml of H medium at 30 mC. The four strains were then incubated with 1 M Spd-C -BODIPY in # the dark in a shaking incubator at 250 cycles\min. At the indicated times, cells were processed for FACS analysis as described above. Wild-type RH144-3D cells and the end3 mutant were grown to mid-exponential stage in YPD at 25 mC, then rinsed and preincubated for 10 min at 37 mC in H medium. Cells were then incubated with either 1 M Spd-C -BODIPY for the indi# cated time interval, or 10 M [$H]spermidine (50 Ci\mol) for 20 min to determine the rate of high-anity spermidine uptake [23,26].

Figure 2 cells

Heterogeneous cytoplasmic labelling by Spd-C2-BODIPY in CHO
CHO-TOR cells grown on sterile glass coverslips were incubated for 4 h with 1 M Spd-C2BODIPY. Note the uneven distribution of uorescence in the cytoplasm in the form of granules or vesicle-like structures.

Statistical analysis

The statistical signicance of dierences between means was assessed at the 5 % level by unpaired Students t tests or by the multiple-range DuncanKramer test [30]. Unless otherwise indicated, results are expressed as the meanpS.D. from three separate experiments with duplicate or triplicate determinations for each experiment.
[13]. The Ki of Spd-C -BODIPY (46p13 M) was about 30-fold # higher than the Km of spermidine (1.5p0.4 M), consistent with the weak inhibitory potency also measured for N%(mercaptoethyl)spermidine (Ki l 25p1 M). Spd-MANT (Ki l 0.4p0.07 M) was a much stronger competitor of spermidine uptake than Spd-C -BODIPY, exhibiting an even stronger # anity than spermidine for the polyamine transporter.
Labelling of intact cells with Spd-C2-BODIPY
As shown in Figure 2, the accumulation of Spd-C -BODIPY # added in trace amounts (1 M) to the growth medium was readily detectable by epiuorescence microscopy in wild-type CHO (CHO-TOR) cells. Intracellular uorescence was clearly distributed in a discrete fashion, leading to a granular pattern, with a much less intense staining of the cytosol. To evaluate the suitability of Spd-C -BODIPY as a probe for polyamine trans# port, its ability to dierentially label CHO-TOR and polyaminetransport-decient mutant (CHO-MG) cells was next assessed. It had been established previously that CHO-MG cells are completely decient in diamine and polyamine transport [10,21]. Cells were pre-treated for 72 h with DFMO and were co-incubated with cycloheximide to maximize probe accumulation (see below). Whereas CHO-TOR cells were intensely labelled with Spd-C # BODIPY, only a faint green intracellular uorescence could be detected in the CHO-MG cells (Figure 3). These data indicate that conjugation of the hydrophobic BODIPY moiety to spermidine via a long side chain preserves the specicity of substrate binding to the polyamine carrier, and that Spd-C # BODIPY is a reliable probe of high-anity polyamine uptake. Polyamine uptake is an active-transport process and is thus strongly dependent on temperature [3,5]. MFI was determined by ow cytometry in CHO-TOR and CHO-MG cells incubated for 1 h at either 4 or 37 mC with Spd-C -BODIPY (1 M) in # serum-free -MEM, after pre-incubating cells at the respective temperature for 60 min in serum-supplemented growth medium.
RESULTS Design and synthesis of the BODIPY-spermidine conjugate
Spermidine analogues derivatized at the N% position are better substrates for the mammalian polyamine-transport system than those substituted on primary amino groups [31]. Moreover, spermidine and norspermidine dimers cross-linked through alkylation of the secondary amino group with aliphatic or aromatic linkers display low Ki values for competition against polyamine uptake [32]. In order to generate a sensitive probe of polyamine transport, we thus conjugated the iodoacetamide derivative of the highly uorescent dye BODIPY FL [33] to the thiol group of N%-(mercaptoethyl)spermidine [19]. The availability of a thiol as a unique reactive group obviated the need for protecting the amino groups of the spermidine derivative before conjugation to the uorophore. The resulting conjugate, Spd-C -BODIPY # (Figure 1) could thus be prepared using a simple, one-step reaction while avoiding the use of extreme pH conditions under which the uorophore is unstable [34]. The ability of Spd-C -BODIPY to interact with the polyamine# transport system was determined by measuring competition by the probe against spermidine uptake in CHO-TOR cells. For comparison, the Ki value against spermidine uptake was also determined for N%-(mercaptoethyl)spermidine and Spd-MANT (Figure 1), a recently described uorescent polyamine which bears a smaller hydrophobic uorophore than Spd-C -BODIPY #

Figure 3

Epiuorescence microscopy of Spd-C2-BODIPY accumulation in CHO-TOR and CHO-MG cells
CHO-TOR (A, B) or CHO-MG (C, D) cells grown on coverslips were incubated for 4 h with 1 M Spd-C2-BODIPY in the presence of 5 mM DFMO and 200 M cycloheximide. The same elds were examined by epiuorescence microscopy (A, C) and phase-contrast microscopy (B, D).
The uptake of Spd-C -BODIPY was strongly temperature# dependent, being decreased from 2.8p0.2 and 0.6p0.01 channel number units at 37 mC to 0.25p0.01 and 0.05p0.01 channel number units at 4 mC in CHO-TOR and CHO-MG cells, respectively. We next determined the ability of natural polyamines to compete against intracellular labelling by the uorescent probe. At a 1000-fold molar excess, putrescine, spermidine and spermine abolished 89p16 %, 82p9 % and 82p9 % of the MFI in CHOTOR cells incubated for 4 h with 1 M Spd-C -BODIPY. On the # other hand, spermidine (1 mM) did not signicantly decrease intracellular uorescence measured in the CHO-MG mutants incubated under the same conditions (14p4 and 17p8 relative uorescence units for cells labelled in the absence or presence of spermidine respectively ; meanspS.D. from three independent experiments). Thus cell labelling with Spd-C -BODIPY largely # depends on the binding of the spermidine-like portion of the conjugate to the polyamine carrier. FACS analysis of intracellular labelling by Spd-C -BODIPY # was performed over a 6 h period in wild-type and CHO-MG cells. The rate of uptake of the probe was initially linear in CHOTOR cells for about 2 h and then steadily decreased (Figure 4). In contrast, Spd-C -BODIPY accumulated very slowly in CHO# MG cells, and no net uptake could be detected 4 h after addition of the probe. Spd-C -BODIPY uptake in the transport mutant # represented about 1520 % of that detected in wild-type cells throughout the incubation period, and could not be decreased by co-incubation with a 1000-fold excess of spermidine (results not shown). The rapid decrease in the rate of Spd-C -BODIPY accumu# lation observed after a 2 h incubation was reminiscent of the
Figure 4 Time course of the uptake of Spd-C2-BODIPY in wild-type (CHOTOR) and polyamine-transport-decient mutant (CHO-MG) cells
Cells were incubated in medium containing 1 M Spd-C2-BODIPY for the indicated interval prior to FACS analysis. Data are meanspS.D. of mean uorescence emission values from three separate experiments.
onset of feedback repression of polyamine transport induced by polyamine internalization in mammalian cells [5]. This feedback inhibition results from the rapid induction of antizyme synthesis

Figure 5

Confocal laser scanning microscopy of Spd-C2-BODIPY and Texas Redtransferrin accumulation in CHO-TOR cells
Cells were incubated for 4 h with either 1 M Spd-C2-BODIPY or 3 g/ml Texas Redtransferrin (Texas Red-Tf) and then examined by confocal laser microscopy as described in the Experimental section. Three representative elds (AC) are shown to illustrate the extent of partial co-localization (arrows) and the pattern of reticular perinuclear staining (pn). For each eld shown, Spd-C2BODIPY uorescence is shown in green, Texas Redtransferrin in red and the superimposition of the two images is shown on the right.
by promotion of a j1 frameshifting in the translation of antizyme mRNA by polyamines [6]. We thus determined the eect of protein synthesis inhibition on the accumulation of the polyamine probe. Furthermore, we examined the eect of prior polyamine depletion on Spd-C -BODIPY uptake by pre-incubating cells # with DFMO, a treatment which up-regulates polyamine-transport activity [3,5,11]. DFMO signicantly (P 0.05) increased the apparent uptake of Spd-C -BODIPY by about 2-fold (MFI, # 108p27 and 45p6 in DFMO-treated and control cells, respectively), as previously observed for the transport of natural polyamines [11]. Moreover, cycloheximide signicantly (P 0.05) increased MFI of wild-type CHO cells incubated with Spd-C -BODIPY to 149p6 and 196p39 in the absence and #
presence of DFMO, respectively, consistent with the relief of feedback inhibition of Spd-C -BODIPY uptake by pre-empting # stimulation of de no o synthesis of antizyme by the internalized polyamine [5,35].
Spectrometric properties of Spd-C2-BODIPY and Spd-MANT
During the course of this work, the characteristics of two other uorescent polyamine probes, namely monouoresceinylspermidine [14] and Spd-MANT [13], have been reported. We thus compared the properties of the latter probes with those of Spd-C -BODIPY. In our hands, monouoresceinylspermidine # [14] (which is in fact a mixture of the N"- and N)-uoresceinyl-

Figure 6

Receptor-mediated endocytosis is required for Spd-C2-BODIPY accumulation in yeast
(A) The time course of high-anity spermidine uptake activity was determined in wild-type yeast cells as well as their end3-1 mutant derivative after transfer to the non-permissive temperature (37 mC) using 10 M [3H]spermidine as a substrate. (B) The same strains were incubated for the indicated period with 1 M Spd-C2-BODIPY after transfer to the non-permissive temperature (37 mC) prior to FACS analysis. Data are the meanspS.D. of intracellular radioactivity or MFI values from three separate experiments.

substituted isomers) very faintly labelled CHO cells at 1 M, whereas cells were readily stained with an equimolar concentration of Spd-MANT (results not shown) [13]. We next compared the relative sensitivity of Spd-C -BODIPY and Spd# MANT to generate a uorescent signal upon internalization by CHO-TOR cells. For this purpose, CHO-TOR cells were incubated with equimolar amounts of each probe (1 M) for 7.5 h, and the intensity of both total uorescence and autouorescence was then determined by ow cytometry. At 1 M, Spd-C # BODIPY exhibited an 18-fold larger specic signal than SpdMANT, whereas autouorescence was similar at the respective excitation wavelength used (0.93 and 1.23 for Spd-C -BODIPY # and Spd-MANT respectively). The dierence observed between the intensity of the two probes was consistent with their respective molar absorption coecients (8.1i10% and 5i10$ cm":M" for Spd-C -BODIPY and Spd-MANT respectively). Thus, despite # the approx. 50 100-fold higher apparent anity of Spd-MANT, a much stronger signal is generated by Spd-C -BODIPY at # equimolar working concentrations, owing to the favourable optical properties of the uorophore.
Compartmentalization of Spd-C2-BODIPY as studied by confocal laser scanning microscopy
The discrete distribution of internalized Spd-C -BODIPY into # vesicle-like entities noted by epiuorescence microscopy (see Figure 1) was very similar to that observed with Spd-MANT [13]. This pattern was suggested to result from receptor-mediated endocytosis of polyamines upon binding to the membranebound polyamine carrier [13]. To better understand the compartmentalization of Spd-C -BODIPY, and to assess the # hypothesis that the probe is accumulated by receptor-mediated endocytosis, we used confocal laser scanning microscopy to visualize the simultaneous accumulation of Spd-C -BODIPY # and Texas Redhuman transferrin conjugate. Internalization of labelled transferrin has been widely used as a marker of vesicle recycling through receptor-mediated endocytosis, and to visualize the early steps of endosome formation in mammalian cells [36]. Thus extensive co-localization of Spd-C -BODIPY and Texas #
Redtransferrin would indicate that Spd-C -BODIPY is co# internalized with transferrintransferrin-receptor complexes via receptor-mediated endocytosis. As shown in Figure 5, uorescence of both Spd-C -BODIPY # and Texas Redtransferrin was virtually excluded from the nucleus, and was similarly concentrated mainly as vesicles of various shapes and diameters, often displaying a perinuclear staining pattern. Faint cytoplasmic staining was observed for Spd-C -BODIPY, but not Texas Redtransferrin, suggesting # that free Spd-C -BODIPY may reach the cytosol to a detectable # extent. However, co-localization of Spd-C -BODIPY and Texas # Redtransferrin was only partial, as determined by computer image analysis, and was mostly limited to large vesicle-like structures. These data strongly suggest that Spd-C -BODIPY is # internalized via a pathway morphologically similar to receptormediated endocytosis, but that partly diverges from that used by the transferrintransferrin-receptor complexes. To rule out the possibility that the defect in Spd-C -BODIPY # accumulation found in CHO-MG mutants might be due to general alteration of receptor-mediated endocytosis in the latter cell line, we determined the accumulation of a BODIPY FL transferrin conjugate (5 g\ml) by FACS analysis. MFI associated with BODIPY FLtransferrin accumulation was 1.36p0.08 and 1.67p0.11 for the parental and CHO-MG cells respectively. Thus receptor-mediated endocytosis is normal in CHO-MG cells, and defective internalization of Spd-C # BODIPY in polyamine-transport-decient cells probably cannot be attributed to a defect in the early steps of endocytosis.

Mechanism of Spd-C2-BODIPY internalization in yeast
The evidence presented above points to a major role of receptormediated endocytosis for the internalization of Spd-C -BODIPY # in mammalian cells. We had previously determined that receptormediated endocytosis is also involved in the regulation of highanity polyamine transport in the yeast S. cere isiae, and contributes to the acute down-regulation of polyamine carrier activity by newly internalized spermidine [23].

Figure 7

Spd-C2-BODIPY accumulation does not respond to known phenotypic alterations of polyamine transport in yeast
Yeast mutants deleted for the PTK2 (A) or SPE2 (B) genes, as well as their respective parental strains, were incubated for the indicated time intervals with 1 M Spd-C2-BODIPY prior to FACS analysis, as described in the Experimental section. Prior to the experiment shown in (B), cells were grown for 48 h in amine-free medium to achieve total polyamine depletion in the spe2 cells. Data are the meanspS.D. of MFI values from three separate experiments.
To assess the involvement of receptor-mediated endocytosis in the uptake of Spd-C -BODIPY in yeast, we determined the eect # of an end3 mutation on probe internalization. The product of the END3 gene is involved in actin cytoskeleton organization and at an early step of receptor-mediated endocytosis, and the temperature-sensitive end3-1 mutant exhibits a defect in the endocytic internalization of various plasma membrane receptors and permeases [23,37]. As expected [23], high-anity [$H]spermidine uptake was increased markedly by the presence of the end3 mutation at the non-permissive temperature (37 mC ; Figure 6A). This eect has been attributed to an increased number of polyamine carriers due to their decreased recruitment by recycling endosomes [23]. However, defective END3 function strongly inhibited the intracellular accumulation of Spd-C # BODIPY (Figure 6B). We next determined Spd-C -BODIPY uptake in mutants # deleted for the PTK2 gene, encoding a serine\threonine protein kinase that positively regulates polyamine transport [4,26]. As shown in Figure 7(A), ptk2 mutants accumulated Spd-C # BODIPY at the same rate as wild-type cells, whereas [$H]spermidine uptake was decreased by 7080 % in these mutants (results not shown), as previously reported [26]. We also assessed the eect of polyamine depletion on the uptake of the uorescent polyamine derivative in mutants (spe2) deleted for the gene encoding S-adenosylmethionine decarboxylase. Upon incubation in amine-free medium, spe2 cells exhibit a severalfold up-regulation of polyamine transport due to the relief from feedback inhibition of the high-anity polyamine carrier by endogenous polyamines [23]. Indeed, pre-incubation of spe2 cells under polyamine-free conditions led to a signicant (P 0.01) increase in the apparent rate of Spd-C -BODIPY # accumulation as compared with wild-type cells (Figure 7B). However, maximal probe accumulation observed in spe2 cells was only 40 % greater than that found in wild-type cells after 3 h, and dierential accumulation per cell mass in the mutant strain was actually less than the observed uptake, since a substantial population (2530 %) of polyamine-depleted spe2 cells exhibited a signicant increase in cell volume (results not shown), as previously reported [29]. Taken together, these data strongly

suggest that Spd-C -BODIPY accumulation in yeast proceeds # via a pathway that is limited by endocytosis, unlike the natural polyamines, leading to divergent dynamic behaviour as a substrate of the high-anity polyamine-transport system.

DISCUSSION

FACS analysis based on the dierential uptake or display of specic uorescent probes has been successfully used for the expression cloning of various membrane proteins [38,39]. Moreover, microscopic analysis of intracellular tracking of transport substrates such as organic cations [40] or folate conjugates [41] has helped to elucidate their mode of delivery and\or compartmentalization. We have thus designed Spd-C -BODIPY to obtain # insight into the fate of internalized polyamines, and as a tool for monitoring the expression of functional polyamine transporters in our current eort to identify DNA sequences encoding these proteins. Spd-C -BODIPY accumulation reliably detects # polyamine-transport activity in mammalian cells since (i) it is suppressed by co-incubation with natural polyamines, (ii) it is 56-fold lower in polyamine-transport-decient cells and (iii) it is up-regulated upon prior polyamine depletion or concomitant suppression of feedback transport inhibition by the internalized polyamine. In our uorescence-microscopy assays with monouoresceinated spermidines, the weak labelling intensity did not allow clear visualization of the internalized polyamine, suggesting poor uptake of the probe. The reason for the discrepancy with results by Aziz et al. [14] is not clear. A potential disadvantage of uoresceinated spermidine derivatives is the fact that the published synthetic procedure involves using a mixture of the N"and N)-(monouoresceinyl)spermidine isoforms, as well as an unknown proportion of the 5- and 6-FITC isomers [14]. Thus lack of stereochemical denition of the mixture used might lead to variable results with the use of uoresceinated spermidine conjugates. On the other hand, we conrmed the results obtained by Cullis et al. [13] for Spd-MANT, and found that it could also be used as a probe for polyamine transport. Thus, despite the considerably longer and bulkier side arm of Spd-C -BODIPY, #

Figure 8

Two models for the intravesicular accumulation of N 4-substituted polyamines
In model A (left), the polyamine (sphere) enters into the cytosol via a bona de plasma-membrane transporter, and a fraction of it can then be sequestered into pre-formed intracellular vesicles (a subpopulation of which mixed with the recycling endosomes) possibly via a putative H+/polyamine (PA) antiporter, perhaps similar to the vesicular monoamine transporters of synaptic vesicles and chroman granules [45]. However, direct insertion into these vesicles by passive diusion of the hydrophobic moiety of the probe (broken arrow) cannot be ruled out completely. In model B (right), the polyamine rst binds to a putative plasma-membrane receptor that undergoes endocytosis as a polyaminereceptor complex. Acidication of the endosome by insertion of vacuolar (V-) ATPases (shown as truncated cones) might favour dissociation of the latter complex, and promote export of the polyamine towards the cytosol, perhaps via the action of a H+polyamine symporter similar to the NRAMP metal transporters [44].

both probes are taken up by mechanisms basically similar to those of natural polyamines. On the other hand, the current results still allow for the possibility that Spd-C -BODIPY or Spd# MANT might be initially taken up by a classical polyamine transporter, and then be sequestered into the observed vesiclelike structures (Figure 8, model A). Such a secondary sequestration might occur via active carriers similar to H+-dependent vesicular monoamine transporters [45]. Alternatively, the free base form of the N%-substituted spermidine probes could passively diuse into small acidic compartments, as in the case of other amphipathic amines such as chloroquine [46] or quinacrine [40], although such a mechanism should be limited by the fact that the probes are much stronger bases than the latter amines. Notwithstanding the pathway of polyamine internalization, a uorescent probe with highly favourable spectral and biochemical properties such as Spd-C -BODIPY should be very useful in # delineating the various steps in polyamine transport and to identify proteins associated with the latter process(es). This work was supported by a grant (MT-12551) from the Medical Research Council of Canada. We are indebted to Mr Maurice Dufour and E; ric Pellerin for their invaluable expertise in FACS analysis and confocal laser scanning microscopy, ' respectively. We also thank Ms Genevieve Pare! for her contribution to the initial characterization of Spd-C2-BODIPY.
uorescently labelled transferrin, a marker of recycling endosomes [36]. However, the only partial co-localization of Spd-C # BODIPY and Texas Redtransferrin indicates that divergent pathways of endocytic internalization might exist for the two ligands. A rst possibility is that Spd-C -BODIPY bound to the # polyamine carrier is sorted from transferrin receptors at the level of sorting endosomes [36], corresponding to the larger vesicles where both probes were found to co-localize. A second hypothesis is that two similar populations of recycling endosomes containing polyamine carriers and transferrin receptors diverge due to the existence of specialized lipid microdomains according to the raft hypothesis [42]. Alternatively, polyamine-transporterenriched vesicles might exist as specialized endosomes recycling between the plasma membrane and an intracellular compartment, similar to the latent pool of GLUT4 hexose transporters present in insulin-responsive cells [43]. These hypotheses are currently being assessed biochemically using specic protein markers of the various endosomal compartments and detailed kinetics of Spd-C -BODIPY internalization. # Our data demonstrate that, in yeast, endocytosis is a ratelimiting step for the internalization of Spd-C -BODIPY, but not # for spermidine transport. Thus Spd-C -BODIPY cannot be # considered as a mimic of the natural substrates of the yeast polyamine permease, as further supported by the fact that its accumulation was not aected by disruption of the PTK2 gene that is required for polyamine transport [4,26], and was only marginally responsive to polyamine depletion. These marked qualitative dierences between the yeast and mammalian polyamine plasma-membrane transporters remain to be elucidated, since molecular information is lacking for either carrier(s). The fact that Spd-C -BODIPY is a good surrogate of the # natural polyamines for the biochemical parameters of polyamine transport in mammalian cells, but not in yeast, might suggest that receptor-mediated endocytosis is an integral part of the mechanism of polyamine transport in higher eukaryotes. Only a few substrates are known to rely mainly on an endocytic pathway for intracellular delivery, e.g. the Fe(II)\transferrin complex [44] and folates [41,42]. Such pathways involve binding of the substrate to a membrane receptor, followed by endocytosis of the substratereceptor complex, and transfer of the sequestered substrate from the acidic endosome to the cytosol. Spd-C # BODIPY indeed reaches the cytosolic compartment, as shown by the induction of feedback repression of its own accumulation and by the presence of diuse intracellular staining by the probe. Thus, if polyamine internalization initially proceeds via receptormediated endocytosis, polyamine uptake would require two steps, i.e. initial binding to a receptor-like membrane protein, followed by endocytosis and release of the polyamine via a putative endosomal polyamine exporter (see Figure 8, model B). The eciency of polyamine delivery through such a mechanism would clearly be limited by the rate of recycling of the putative polyamine receptor to the cell surface, since the substrate and the polyamine transporter would most likely be internalized with a 1 : 1 stoichiometry. The present data do not rule out the possibility that the accumulation of Spd-C -BODIPY or Spd-MANT into vesicle# like structures may not be representative of the fate of natural polyamines. For instance, a substantial fraction of Spd-C # BODIPY might be endocytosed as a probe\transporter complex, but remain sequestered into endosomes without an actual translocation of the probe into the cytosol because of steric constraints due to the bulky uorophore moiety. However, the fact that SpdMANT, which has a pattern of uptake similar to that of Spd-C # BODIPY, is intracellularly accumulated at a rate comparable with that of spermidine import [13], supports the notion that