Báo cáo khoa học: Saporin and ricin A chain follow different intracellular routes to enter the cytosol of intoxicated cells pptx

Saporin and ricin A chain follow different intracellular
routes to enter the cytosol of intoxicated cells
Riccardo Vago
1,
*, Catherine J. Marsden
2,
*, J. Michael Lord
2
, Rodolfo Ippoliti
3
, David J. Flavell
4
,
Sopsamorn-U Flavell
4
, Aldo Ceriotti
5
and M. Serena Fabbrini
1,5
1 Dibit-S Raffaele Scientific Institute, Milan, Italy
2 Department of Biological Sciences, University of Warwick, Coventry, UK
3 Dipartimento di Biologia di Base ed Applicata, Universita
`
degli Studi di L’Aquila, Italy
4 The Simon Flavell Leukaemia Research Unit, University Department of Pathology, Southampton, UK
5 Istituto di Biologia e Biotecnologia Agraria, CNR, Milan, Italy
Protein toxins whose substrates are located within the
cytosol of mammalian cells must be able to cross an
intracellular membrane in order to exert their biologi-
cal activity. Following initial internalization, these tox-
ins must travel intracellularly to reach their molecular

targets [1]. Some bacterial toxins such Pseudomonas ae-
ruginosa Exotoxin A (PEA) carry a KDEL-like signal
for retrieval to the endoplasmic reticulum (ER) [2,3].
KDEL receptors, normally cycling between the Golgi
complex and the ER, can retrieve escaped ER-resident
proteins that carry KDEL ⁄ REDL (single amino acid
letter code) at their C-termini. In the ER, the presence
of a higher pH allows detachment of the retrieved
protein from the KDEL receptors [4]. The REDLK
Keywords
anticancer therapy; bacterial toxins;
intracellular trafficking; KDEL retrieval
sequence; plant ribosome-inactivating
proteins
Correspondence
M. S. Fabbrini, CNR, via Bassini 15,
20133 Milan, Italy
Fax: +39 223 699 411
Tel: +39 223 699 444
E-mail:
*Riccardo Vago and Catherine J. Marsden
contributed equally to this work.
(Received 18 May 2005, revised 11 July
2005, accepted 9 August 2005)
doi:10.1111/j.1742-4658.2005.04908.x
Several protein toxins, such as the potent plant toxin ricin, enter mamma-
lian cells by endocytosis and undergo retrograde transport via the Golgi
complex to reach the endoplasmic reticulum (ER). In this compartment the
catalytic moieties exploit the ER-associated degradation (ERAD) pathway
to reach their cytosolic targets. Bacterial toxins such as cholera toxin or

Pseudomonas exotoxin A carry KDEL or KDEL-like C-terminal tetrapep-
tides for efficient delivery to the ER. Chimeric toxins containing monomer-
ic plant ribosome-inactivating proteins linked to various targeting moieties
are highly cytotoxic, but it remains unclear how these molecules travel
within the target cell to reach cytosolic ribosomes. We investigated the
intracellular pathways of saporin, a monomeric plant ribosome-inactivating
protein that can enter cells by receptor-mediated endocytosis. Saporin toxi-
city was not affected by treatment with Brefeldin A or chloroquine, indica-
ting that this toxin follows a Golgi-independent pathway to the cytosol
and does not require a low pH for membrane translocation. In intoxicated
Vero or HeLa cells, ricin but not saporin could be clearly visualized in the
Golgi complex using immunofluorescence. The saporin signal was not evi-
dent in the Golgi, but was found to partially overlap with that of a late
endosome ⁄ lysosome marker. Consistently, the toxicities of saporin or sapo-
rin-based targeted chimeric polypeptides were not enhanced by the addition
of ER retrieval sequences. Thus, the intracellular movement of saporin
differs from that followed by ricin and other protein toxins that rely on
Golgi-mediated retrograde transport to reach their retrotranslocation site.
Abbreviations
ATF, amino-terminal fragment of urokinase; BFA, Brefeldin A; DT, diphtheria toxin; ER, endoplasmic reticulum; ERAD, ER-associated
degradation; huPAR, human urokinase receptor; LRP, LDL-receptor related protein; PEA, Pseudomonas aeruginosa Exotoxin A; RIP,
ribosome-inactivating protein; RTA, ricin A chain; SAP, saporin.
FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS 4983
sequence found at the C-terminus of PEA is essential
for the cytotoxicity of the endocytosed toxin, allowing
PEA to reach its site of action [2,5]. This implies that
PEA may interact with the KDEL receptor in order to
traffic from the Golgi to the ER.
The plant ribosome-inactivating protein ricin also
enters the endocytic pathway and travels backwards

from the Golgi complex to the ER where it is thought
to parasitize the ER-associated degradation (ERAD)
pathway [1,7–10] that normally disposes misfolded or
unassembled proteins to the cytosol for proteasomal
degradation [6]. Although ricin does not contain a
KDEL-like C-terminal sequence, addition of this ER
retrieval signal greatly enhances the toxicity of both
a reconstituted AB holotoxin and the A chain alone
[10–12].
Thus, the catalytic domains of different bacterial
and plant protein toxins, including ricin [8], PEA
[2,3] cholera toxin and Shiga toxin, can exploit the
ERAD pathways [1,4,14] to reach their targets in the
cytosol [15]. Here, most of them can irreversibly
inactivate protein synthesis [1,13] causing apoptotic
cell death [16,17]. PEA ADP-ribosylates elongation
factor 2 [2,13,14,20], whereas Shiga and ricin A chain
act by specifically depurinating 28S ribosomal RNA
[1,3,7,9]. Saporin is a monomeric plant polypeptide
that shows the same N-glycosidase activity as the
ricin A chain. Different isoforms can be found in
seeds and leaves of the soapwort Saponaria officinalis
and some have been expressed in Escherichia coli
and characterized biochemically [18,19]. The catalytic
subunits of protein toxins are used to construct toxic
chimeras selectively directed against tumor or meta-
static cells via specific targeting domains [20]. One
such recombinant chimera, preATF–SAP, targets
transformed cells expressing the human urokinase
receptor (huPAR) and contains the amino-terminal-

fragment (ATF) of human prourokinase fused to
the mature sequence of the ricin-related single-chain
ribosome-inactivating protein saporin (SAP) [19].
To allow correct folding of the ATF cell-binding
domain, which is essential for binding to huPAR
[21] and contains six disulfide bridges forming a
kringle and a growth factor-like domain, we
expressed a secretory version of the ATF–SAP chi-
mera in Xenopus laevis oocytes. Endogenously syn-
thesized preATF–SAP was highly cytotoxic to host
Xenopus laevis oocytes, but the oocytes could be pro-
tected from autointoxication by injecting neutralizing
antisaporin antibodies into the cytosol [22]. The
mechanism(s) underlying this cytotoxicity remains
unclear but these results clearly show that some pre-
ATF–SAP polypeptides reached the oocyte cytosol.
These observations raised the possibility that saporin
may also use ER dislocon channels to enter this
compartment.
We investigated the pathway followed by saporin in
exogenously intoxicated cells. Overall, our results
strongly indicate that, in spite of the structural similar-
ities with ricin A chain, saporin and derived chimeras
follow a different intracellular transport route(s).
Results
Vero and HeLa cells were treated with drugs known to
interfere with ricin holotoxin intracellular delivery and
thus cytotoxic activity. The fungal inhibitor Brefel-
din A (BFA) causes Golgi complex disassembly, pro-
tecting cells against both ricin and PEA intoxication

[23,24]. Furthermore, proteasomal inhibition prevents
the cytosolic degradation of catalytic A chains of ricin
following ER-to-cytosol transport, an effect exacerba-
ted in the case of a mutant (ricin-6K) with increased
lysine content [25]. Saporin is a lysine-rich protein and
proteasomal inhibitors would sensitize target cells if
the dislocation mechanism was similar to the one used
by ricin.
Ricin cytotoxicity was sensibly decreased by BFA
treatment, as expected [23], whereas it was slightly
increased by the proteasome inhibitor (Table 1). In
contrast, neither drug significantly affected saporin-
mediated toxicity (Table 1). In a second set of experi-
ments, HeLa cells were challenged with different
concentrations of saporin, ricin or RTAKDEL, either
in the absence or presence of BFA (Table 1). BFA
treatment led to a dramatic increase in the ID
50
sof
ricin and RTAKDEL but did not have any effect on
saporin ID
50
. The proteasomal inhibitor clasto-lacta-
cystin-b-lactone sensitizes HeLa cells toward the action
of mutated ricin-6K [25], as expected, but its effect on
ricin and saporin toxicity was only a two- to threefold
sensitization (Table 1). Thus, neither transport via the
Golgi to the ER nor dislocation as an unfolded poly-
peptide appears to contribute to the productive intoxi-
cation route followed by saporin.

Intracellular tracing of a fluorescinated saporin in
both Vero cells (Fig. 1A) and HeLa cells (not shown)
revealed the presence of saporin in punctuate struc-
tures after exposure to an excess of the toxin.
Although we cannot exclude that some fluid-phase
uptake may also have occurred in these conditions, at
these time points, we did not observe any colocaliza-
tion of saporin with early endosome markers (anti-
EEA1), although the late endosomal marker Lamp2
was shown to partially overlap with CY3 saporin
fluorescence. Furthermore, unlike for ricin (Fig. 1B),
Saporin trafficking in intoxicated mammalian cells R. Vago et al.
4984 FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS
no Golgi localization of fluorescent saporin could be
detected using anti-(Golgin 97) serum. This finding is
fully consistent with BFA being ineffective in blocking
saporin toxicity (Table 1).
Toxins exploiting the ERAD pathway have a low
lysine content to avoid ubiquitination upon disloca-
tion to the cytosol [26]. A paradigm of a second class
of toxins with normal lysine content is the diphtheria
toxin (DT) that, while transiting in acidic endosomes,
undergoes a conformational change triggering forma-
tion of a pore through which the catalytic chain
escapes into the cytosol and inactivates protein syn-
thesis [27,28]. Therefore, in both Vero and HeLa cells,
we analyzed the effects of chloroquine, a lysosomal
caotropic drug that raises the pH in acidic compart-
ments and almost abolished DT toxicity, but could
not affect saporin-mediated cytotoxicity (Table 1).

Bafylomycin A1, an inhibitor of the H
+
ATPase
pump was able to protect cells from DT intoxication,
but again did not affect saporin-mediated cytotoxicity
(data not shown).
The low cytotoxic activity of saporin in HeLa cells,
with ID
50
in the micromolar range after 6 h of expo-
sure (Table 1), prompted us to verify that toxicity was
due to a genuine depurinating capability of the plant
toxin over the endogenous ribosomes. RNA was iso-
lated from cells treated with graded saporin concen-
trations, as indicated, and either treated or not treated
with acetic aniline. Figure 2 shows that saporin is
indeed able to reach and inactivate HeLa ribosomes,
as shown by the diagnostic aniline fragment vizualized
in the denaturing agarose gels. We predicted, based on
these observations, that appending an ER recycling
signal such as KDEL to the C-terminus of saporin
would not affect its cytotoxicity. We therefore com-
pared the killing activities of SAPKDEL with SAPwt,
independent of any targeting domain, and investigated
whether the cytotoxicity of saporin would be potenti-
ated by a KDEL motif, as previously shown for both
PEA [2] and ricin A chain [12]. Both SAPwt and
SAPKDEL (Fig. 3A) were expressed in bacteria and
purified to homogeneity as described previously [18]
(data not shown), and recombinant saporin-KDEL

was specifically immunoprecipitated by monoclonal
anti-KDEL serum before assaying its biological activit-
ies (Fig. 3B). In Table 2, the in vitro activities of
recombinant ricin and saporin polypeptides are com-
pared: RTA and SAP IC
50
values were in the pico-
molar range and were essentially the same as their
KDEL-extended versions (Table 2). Vero cells are
greatly sensitized to RTAKDEL [10–12] and were
therefore used to compare the cytotoxic activities of
the recombinant polypeptides (Table 2). The KDEL
sequence increased the cytotoxicity of RTA almost
20-fold (ID
50
of 1.8 nm for RTA-KDEL vs. 44 nm for
RTAwt). In contrast, addition of the KDEL sequence
did not potentiate the cytotoxic activity of saporin
after 4 h exposure (ID
50
of 5 nm for SAPKDEL vs.
3.7 nm for SAPwt) or even after longer exposure, indi-
cating that this effect was independent of the kinetics
Table 1. Saporin cytotoxicity is resistant to treatments affecting
ricin or diphtheria toxin toxicities. Cell-killing of Vero cells was per-
formed as described in the Experimental Procedures. The ID
50
val-
ues of plant intact ricin (A + B) and diphtheria toxin (DT) were
determined, using these same assays, and found to be around 1.7

and 2.5 p
M, respectively. Vero cells were exposed to either 9 nM
saporin or 5 pM ricin or 10 pM DT for 4 h in the presence or
absence of the Golgi disrupting drug BFA (0.5 lgÆmL
)1
) or a protea-
some inhibitor (MG-132, 10 l
M)or10lM chloroquine. The data
referred to in A, B and C show the percent of relative light units
(% RLU) referred to 100% luciferase expression in the untreated
samples ± S.E.M. n, number of independent experiments. Where
indicated, HeLa cells were pretreated for 15 min at 37 °C with
10 l
M BFA, 60 min with 20 lM proteasome inhibitor clasto-lacta-
cystin-b-lactone, 60 min with 100 l
M chloroquine. Cells were then
exposed to the various toxins for the indicated times. Residual pro-
tein synthesis was measured by incubating cells at 37 °C for
90 min in the presence of 1 lCi [
35
S]-methionine in NaCl ⁄ P
i
. The
ID
50
values obtained in the absence (–) or (+) presence of drugs are
reported.
Cell type Toxin exposures
BFA
–+

A
Vero % RLU (n ¼ 6) 5 p
M Ricin 30.4 ± 9.8 89.3 ± 7.1
9n
M Saporin 19 ± 5.6 34 ± 12
HeLa ID
50
(6 h) Ricin 3.3 pM >1700pM
RTA-KDEL 33.4 nM >1670nM
Saporin 2100 nM 1940 nM
Cell type Toxin exposures
Proteasome inihibitors
–+
B
Vero % RLU (n ¼ 3) 5 p
M Ricin 17.4 ± 3 6.8 ± 1
9n
M Saporin 16.8 ± 4.8 20.5 ± 7.3
HeLa ID
50
Ricin-6K (4 h) 2140 pM 33.4 pM
Ricin (18 h) 0.199 pM 0.0997 pM
Saporin (18 h) 17.6 nM 5.3 nM
Cell type Toxin exposures
Chloroquine
–+
C
Vero % RLU (n ¼ 3) 10 p
M DT 7.1 ± 0,9 60.5 ± 21.7
9n

M Saporin 19.2 ± 4.9 21.1 ± 3.3
HeLa ID
50
(4 h) DT 0.143 nM > 3.17 nM
Ricin 9.97 pM 8.30 pM
Saporin 2640 nM 2820 nM
R. Vago et al. Saporin trafficking in intoxicated mammalian cells
FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS 4985
of intoxication. Neither the human leukemic cell line
HSB-2 nor the Burkitt lymphoma Ramo cells showed
any potentiation of cytotoxicity by addition of the
KDEL sequence to the C-terminus of saporin (data
not shown). An intriguing result was the decrease in
cytotoxicity of SAPKDEL observed in U937 cells
(Table 2). However, recombinant SAPAARL assayed
as a control showed same ID
50
as SAPKDEL.
We then tested if a targeted saporin chimera such
as ATF–SAP, containing six disulfides, present in a
kringle and the huPAR-binding growth factor
domain [19,21,22] would need to partially unfold
and ⁄ or undergo reduction prior to membrane dislo-
cation. If these steps occurred in the ER, cytotox-
icity could potentially be enhanced by ER-retrieval
motifs. Figure 4 summarizes the secretory mutant
chimeras that were constructed and expressed in
Xenopus oocytes. When the terminal lysine residue of
ATFSAPREDLK (here in bold) is removed by extra-
cellular carboxypeptidase(s) normally present in cell

culture medium [5], does the REDL sequence behave
as an active KDEL-like motif. Therefore, this
mutant chimera should be initially efficiently secreted
by the protected oocytes and, as in the case of PEA,
when exposed to the target cell, would be endo-
cytically taken up and possibly retrieved to the ER.
As a control, we have also expressed preATF–
SAPKDEL.
Synthetic mRNAs encoding preATF–SAP or the
mutants were produced and in an in vitro translation
assay, these COOH-extended mutants could inactivate
reticulocyte lysate ribosomes (Fig. 4B), as shown for
preATF–SAP [22], by blocking their own translation.
Indeed, polypeptides translated from preATF–SAPwt
Fig. 1. Intracellular distribution of saporin in
intoxicated Vero cells. Vero cells were trea-
ted with 100 lgÆmL
)1
Cy-3-labeled saporin
(A) or Cy-3-labeled ricin (B) for 4 h before
methanol fixation and immunostaining using
the antibodies indicated. The scale bar rep-
resents 20 lm.
Saporin trafficking in intoxicated mammalian cells R. Vago et al.
4986 FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS
and either preATF–SAPKDEL or preATF–SAP-
REDLK cRNAs, but not those translated from con-
trol RNA (BMV, compare lanes 4 and 6), were seen
only when translated in the presence of anti-saporin
neutralizing immune Igs (Fig. 4B lanes 7, 10 and 13,

respectively). Thus, as previously shown for SAPK-
DEL, these C-terminal amino acid extensions to the
ATF–saporin chimera did not affect the in vitro toxic-
ity of the chimeras. In pulse-labeled, Ig-protected
Xenopus oocytes (Fig. 5A), the newly synthesized
polypeptides all showed the expected electrophoretic
mobility, those of the mutants being decreased com-
pared with the wild-type chimera. At the end of the
24 h chase period, most of the ATFSAPREDLK poly-
peptides (lane 8) were, as expected, secreted into the
culture medium, as was the wild-type polypeptide (lane
12) [22]. In contrast, KDEL mutant polypeptides were
mostly retained within the oocyte (compare lanes 3
and 4). Fewer than 20% of the newly synthesized
ATFSAPKDEL polypeptides were found in the oocyte
medium after 24 h of chase. Western blot analysis of
the 72 h oocyte incubation media with anti-(ATF krin-
gle domain) (Fig. 5B) and anti-SAP sera (Fig. 5C,
lower panel) showed that only the full-length poly-
peptides were secreted by the protected oocytes,
whereas blotting with a monoclonal anti-KDEL serum
(Fig. 5C, upper panel) indicated that an intact KDEL
sequence was still present in the corresponding KDEL-
secreted chimera.
The specific cytotoxicity of the chimeric proteins was
evaluated using standard cell-killing experiments. U937
cells express both human uPAR and endocytic recep-
tors belonging to the LDL-related receptor family
(LRP) that are required for the efficient targeting and
internalization of these toxic chimeras [19,33]. The

cytotoxic activity of the seed-extracted saporin in
U937 cells was almost three orders of magnitude lower
than that of the huPAR-targeted chimera [19,33]
(Fig. 6, compare ID
50
of SAP [35 nm] with that of
ATF–SAPwt [0.04 nm]). However, both mutant chime-
ras were slightly less active than the wild-type chimera
with REDL- and KDEL-extended versions showing an
ID
50
of 0.1 and 0.2 nm, respectively. This is consistent
with the decrease in cytotoxicity of SAPKDEL com-
pared with SAPwt observed in U937 monocytes
(Table 2).
Appending a KDEL sequence enhanced both RTA
[12] and PEA cytotoxicity [29]. The finding that
ER-retrieval sequences did not enhance the cytotoxici-
ty of either saporin or the ATF–SAP chimera was,
indeed, expected and confirms our initial observations
that the intracellular transport of saporin bypasses the
Golgi complex.
Fig. 2. Saporin cytotoxicity is a direct result of ribosome modifica-
tion. HeLa cells were treated with increasing concentrations of
saporin for 18 h. To ensure that the N-glycosidase activity seen
was entirely due to depurination of ribosomes during the saporin
exposures for the cytotoxicity assay, ribosomes were isolated by
denaturing all proteins upon lysis. After lysis of the cells, RNA was
isolated and aniline treated before running on denaturing agarose
gels. The arrow indicates the aniline band, which is diagnostic of

N-glycosidase activity. The spike sample received the highest con-
centration of saporin, added just prior to cell lysis, and a control
sample received no saporin (–).
A
B
Fig. 3. Schematic representation of the DNA constructs expressed
in E. coli and purifed from bacteria lysates. (A) Mature saporin
(SAPwt) (black and white bars) or mature saporin with a C-terminal
KDEL (SAPKDEL) or AARL (single amino acid letter code) were
expressed in BL21 (De3) pLys E. coli and recombinant toxins puri-
fied to homogeneity; RIP: ribosome-inactivating catalytic domain.
(B) SAPKDEL is immunoprecipitated by monoclonal anti-KDEL sera.
Immunoprecipitates were recovered on protein G–Sepharose beads
and polypeptides transferred on nitrocellulose were revealed with
an antisaporin serum and detected by enhanced-chemiolumines-
cence; H: heavy chains of the Igs. Molecular mass markers are
shown in the right. The arrow points to the position of SAPKDEL
(lane 3). Saporin wt or the mock (–) induced lysates gave no signal
(lanes1 and 2).
R. Vago et al. Saporin trafficking in intoxicated mammalian cells
FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS 4987
Discussion
The therapeutic use of saporin-based immunotoxins
[30,31] prompted us to investigate whether saporin
would follow the same route of entry into the cytosol as
the related plant toxin ricin [10–12] or the bacterial toxin
PEA [2,29]. This would imply that saporin-mediated
cytotoxicity should be increased by introducing KDEL-
like sequences at the C-terminus of this molecule.
Cell-surface binding of saporin is mediated, at least

in part, by members of the LDL-related family of
receptors [18,19,32,33] and LRP-minus MEF cells show
a 10-fold decrease in saporin sensitivity (our unpub-
lished results). LRP mediates internalization of the
ATF–saporin chimera through clathrin-coated pits
[19,33] and the binding and internalization of another
type I RIP, trichosanthin [34], and PEA [3,13]. Thus,
saporin is able to use the same internalization receptor
as PEA bacterial toxin. However, when Vero or HeLa
cells were treated with BFA although Golgi disassem-
bly clearly impaired ricin cytotoxicity, it did not signifi-
cantly affect saporin-mediated toxicity. We therefore
concluded that the Golgi complex is not a major intra-
cellular compartment for productive trafficking of sap-
orin. When we investigated the intracellular route of
a human prourokinase–saporin TRITC conjugate [33],
the fluorescence of the saporin chimera did not overlap
either with a fluorescinated ricin holotoxin or with the
Golgi marker NBD-ceramide. Toxins that use ERAD
pathways [1,7–10,14], such as ricin, PEA [13] and chol-
era toxin [1,10,14,35], must avoid proteasomal degrada-
tion to exert their toxic action [15,35] and their paucity
in lysine (but not in DT retrotranslocating from a dif-
ferent compartment) [27,28] helps avoid ubiquitination
and subsequent proteasome degradation [26]. Cholera
toxin essentially avoids ubiquitination [35] and, in
Table 2. Saporin with a KDEL C-terminal extension has similar
in vitro and Vero cell-killing activity as the wild-type saporin. (A)
Saporin RIP activities were compared using the cell-free system
reticulocyte lysate (in vitro) or by intoxicating Vero cells. The con-

centration inhibiting 50% of BMV RNA translation in vitro was
measured (IC
50
) in replicated samples and reported with the stand-
ard deviations (SD). Vero cells were exposed for 4 h to serial log
dilutions of each toxin before luciferase reporter transfection
(Experimental Procedures). Relative light units (RLU) were quanti-
fied in each sample in a luminometer and the dose of toxin that
inhibits reporter expression by 50% over the untreated controls
(ID
50
) was calculated with the SEM of at least two independent
experiments, each performed four times. Recombinant ricin A chain
was used as a control. (B) Comparison of saporin wild-type [19,22],
SAPKDEL and SAPAARL killing activities in promyelocytic human
U937 cell was carried out essentially as described previously
(Experimental Procedures and the legend to Fig. 6), following 48 h
exposure to serial dilutions of the toxins and measuring the remain-
ing protein synthesis with tritiated leucine incorporation. Mean ID
50
values are reported.
Toxin In vitro IC
50
±SD10
)12
M Vero ID
50
± SEM 10
)9
M

A
SAPwt 25 ± 5 3.7 ± 2.3
SAPKDEL 25 ± 14 5 ± 3.5
RTAwt 180 ± 9 44 ± 12
RTAKDEL 150 ± 4 1.8 ± 0.9
Cell type Toxin ID
50
B
U937 SAPwt 55 n
M
SAPKDEL 200 nM
SAPAARL 200 nM
A
B
Fig. 4. Neutralizing antisaporin Igs are needed for efficient in vitro
translation of the wild-type and COOH-extended ATF–SAPorin
chimeras. (A) Schematic representation of the DNA chimeric con-
structs expressed in protected Xenopus oocytes. PreATF–SAPorin
wild-type (preATF–SAPwt) was obtained by substituting the serine-
protease domain of urokinase with the saporin RIP domain and pre-
ATF–SAPorin with a C-terminal KDEL (preATF–SAPKDEL) or REDLK
(preATF–SAPREDLK) sequence were obtained after introducing
synthetic oligonucleotides (see Experimental procedures) SP: signal
peptide, ATF: amino-terminal fragment for uPAR cell surface bind-
ing. (B) preATF–SAPwt cRNA or those encoding the COOH-
mutants (preATF–SAPKDEL or preATF–SAPREDLK) were translated
in the presence of tritiated leucine in nuclease-treated rabbit reticu-
locyte lysates, supplemented with goat antisaporin immune (i) or
nonimmune (ni) Igs or NaCl ⁄ P
i

(–). BMV RNA was also translated in
the same conditions (lanes 4–6), as control. At the end of the trans-
lation period (1 h) equivalent amounts of lysates were subjected to
a 15% polyacrylamide SDS ⁄ PAGE and fluorography.
Saporin trafficking in intoxicated mammalian cells R. Vago et al.
4988 FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS
agreement with this view, the addition of extra lysine
residues at selected positions in RTA drastically
reduced the cytotoxicity of the holotoxin without
affecting its catalytic activity [25]. Saporin is a mono-
meric protein whose three-dimensional structure can be
superimposed on RTA, despite the fact that their
amino acid identity is lower than 30% [36] and that
10% of the amino acids in saporin are lysine residues
that also confer an extremely high pI (almost 10) and
an unusual stability to this polypeptide [37]. Inhibition
of the proteasomes, however, did not lead to a large
increase in saporin cytotoxicity, suggesting that this
toxin may not use ERAD pathway(s) or may be not
subjected to an unfolding step prior to entry into the
cytosol, as it has recently been postulated for FGF [38].
That saporin is able to reach the cytosolic compartment
was confirmed, because isolated HeLa ribosomes were
depurinated in a dose-dependent fashion. Thus, this
toxin might well be able to escape through different
intracellular compartment(s). Raising the intracellular
pH of the endosomal compartment using chloroquine
or bafilomycin A1 resulted, as expected, in complete
protection from DT. There were, however, no substan-
tial differences between the effects on saporin or ricin

cytotoxicities. This lack of protection by chloroquine
and bafilomycin A1 as well, indicates that whatever the
translocation mechanism of saporin is, it is not low-pH
dependent, as for DT and would differ also from that
of FGF that was shown to be bafilomycin A1 sensitive
[39]. Hence, saporin does not appear to possess
A
B
C
Fig. 5. (A) KDEL mutant polypeptides are retained by Xenopus oocytes whereas polypeptides carrying REDLK are efficiently secreted.
Oocytes were coinjected with preATF–SAPwt cRNA (300 ngÆlL
)1
) or the same amount of synthetic cRNA encoding the COOH-mutants pre-
ATF–SAP-KDEL or preATF–SAP-REDLK together with goat neutralizing antisaporin Igs (3.25 lgÆlL
)1
) to protect oocytes from autointoxi-
cation. Control oocytes (not shown) were left uninjected. After overnight incubation at 19 °C, oocytes were labeled 2 h with S
35
Promix
(0 h chase) and some oocytes were then further chased for 24 h. Equivalent amounts of oocyte lysates (o) and incubation media (m) were
immunoprecipitated with rabbit antisaporin serum, and proteins analyzed by 15% polyacrylamide SDS ⁄ PAGE and fluorography. The arrow
indicates intracellular polypeptide accumulated in the KDEL mutant. (B) Properly folded, full length polypeptides are secreted by the oocytes.
Oocytes were injected as described in Fig. 5A and the unlabeled oocytes incubated at 19 °C for 72 h in the presence of 6% MBS-dialyzed
fetal calf serum. Equivalent amounts of wild-type or mutant ATF–SAP polypeptides were subjected to a nonreducing 15% polyacrylamide
SDS ⁄ PAGE, and the electroblotted polypeptides were immunodetected using anti-ATF conformational sera, followed by secondary HRP-goat
anti-(mouse epitope) Igs and detection by enhanced-chemioluminescence. Conditioned media containing two glycosylated COOH-mutant
ATF–SAP chimeras (our unpublished data) were also loaded (lanes 2 and 4), for comparison. Molecular mass markers (kDa) are shown on
the right. (C) Secreted ATF–SAPorin (lane 2), ATF–SAP-KDEL (lane 3) or ATF–SAP-REDLK (lane 4) polypeptides were also detected using
anti-KDEL sera or rabbit antisaporin. Microsomal membrane preparation (not shown) or SAPKDEL (asterisk) were used as positive controls.
The positions of molecular mass markers (kDa) are indicated on the right. Control oocyte media are loaded in lane 1.

R. Vago et al. Saporin trafficking in intoxicated mammalian cells
FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS 4989
putative translocation domain(s) and the cytotoxic
activity of the ATF–SAPorin chimera was even slightly
increased both by chloroquine or bafilomycin A1 treat-
ments, suggesting its passage though a putative proteo-
lytic compartment [19]. The fact that we trace saporin
passage through late endosomes, because we do see
colocalization with a late endosomal marker, is also
fully consistent with these previous data.
COPI-independent paths do exist to reach the ER,
as recently shown for Shiga-like toxin [3,40]. Disrup-
tion of COPI retrograde transport or deletion of
KDEL in the A2 chain of the cholera toxin could not
abolish toxin delivery to the ER [41]. Interaction with
the KDEL receptors may not be required to reach the
ER, but the KDEL motif might function by retrieving
from the Golgi any toxin that escapes by anterograde
transport [42,43]. We therefore evaluated the exogen-
ous toxicities in target cells, comparing recombinant
saporin with SAPKDEL, and exploited a cytosolic
immunization strategy [22] to extend this comparison
to secreted saporin chimeric polypeptides, carrying sig-
nals that confer ER retrieval along the endocytic route.
Recombinant SAPKDEL was immunoprecipitated by
a monoclonal anti-KDEL serum, indicating that its
KDEL sequence remains fully accessible to this anti-
body in solution and its RIP activity in reticulocyte
lysates was almost superimposable with that of
wild-type saporin. Nevertheless, unlike RTA, saporin

cytotoxicity was not increased by the addition of a
C-terminal KDEL. The KDEL sequence can substitute
for the PEA C-terminus increasing PEA toxicity and
that of chimeric PEA toxins ending with KDEL [29].
The PEA C-terminal sequence binds to KDEL recep-
tors [2] and overexpression of this receptor makes cells
more susceptible to the PEA [3]. The terminal lysine
residue found in the natural PEA C-terminus is nor-
mally removed by extracellular carboxypeptidase(s)
present in cell culture medium [5] leaving a REDL
sequence which behaves as an active KDEL-like
sequence during internalization. We therefore
expressed preATF–SAPREDLK mutant polypeptides
and found they were, as expected, efficiently secreted
by the oocytes, whereas when appended to ATF–sapo-
rin the KDEL sequence was recognized by oocytes
causing the KDEL-bearing chimera to accumulate in-
tracellularly (Fig. 5). This reinforces our assumption
that a KDEL sequence appended to the saporin mole-
cule should behave as an effective signal if this poly-
peptide is able to reach the Golgi complex. The
Xenopus oocyte expression system was leaky, allowing
recovery of some ATFSAPKDEL polypeptides from
the conditioned medium. Blotting with anti-KDEL
confirmed the presence of this C-terminal sequence in
the secreted polypeptides. Secretion and cell-surface
expression of chaperones [44] and proteins carrying
KDEL has been already observed in different cell sys-
tems, and ATFSAPKDEL secretion by the oocytes
might not, therefore, be surprising. C-terminal exten-

ded chimeras showed slightly lower activity against
U937 cells. A similar difference in activity was also
seen when comparing SAPKDEL and SAPAARL with
saporin wt (Table 2). Extra C-terminal sequences
might interfere with endocytosis, slightly decreasing
the efficiency of internalization of KDEL ⁄ REDL sapo-
rins in the human promyelocytic cells. However, a 20
amino acid COOH-extended mutant chimera showed
an ID
50
of 0.07 nm in U937 cells, closer to that of the
wild-type toxin (our unpublished results). Therefore,
these data strongly support our assumption that sapo-
rin and the derived chimeras do not travel through the
Golgi complex to the ER after internalization.
Ricin may also bypass the Golgi apparatus, which
has been vesiculated by depletion of epsilon-COP, but
still reaches the ER [45]. Our data do not completely
exclude the possibility that the plant monomeric toxin
saporin might also exploit the ER for its retrotranslo-
cation. It has recently been postulated that some poly-
peptides such as DHFR or GFP may be able to
undergo ER dislocation without the need for an
unfolding step [46,47]. Our data and the literature
Fig. 6. Comparison of the cytotoxicities of ATF–SAPwt and
KDEL ⁄ REDLK chimeras in U937 monocytes. Acid-washed cells
were exposed 48 h at 37 °C to equivalent serial logarithmic dilu-
tions of the secreted ATF–SAP chimeras, wild-type (filled squares)
REDLK (middle, filled cones) or KDEL (filled triangles) and seed-
extracted saporin (empty triangles). Cells were then pulse-labeled

with
L
-
[4,5-
3
H]leucine and radioactivity incorporation measured after
harvesting cells onto glass fibre filters. Cytotoxicities were calcula-
ted by measuring the dose of toxin that inhibits protein synthesis in
treated cells by 50% (dashed line) and compared with untreated
control cells exposed to equivalent dilutions of the conditioned
medium of goat anti saporin injected oocytes. The dose–response
curves are shown with standard deviations. x-axis: percent total
leucine incorporation.
Saporin trafficking in intoxicated mammalian cells R. Vago et al.
4990 FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS
indicate the existence of multiple intracellular path-
way(s) and delivery mechanisms to reach the cytosolic
compartment. In addition to the great potential in
anti-cancer therapies, saporin should be useful in strat-
egies exploiting disarmed toxins as peptide carriers for
MHC class I presentation [48–50].
Experimental procedures
Cytotoxicity assays using HeLa cells
For 4 or 6 h assays HeLa cells were seeded at
1.5 · 10
5
cellsÆmL
)1
into 96-well plates and grown over-
night at 37 °C. For 18 h assays cells were seeded at

2.5 · 10
5
cellsÆmL
)1
into 96-well plates and grown at 37 °C
for 8 h. Cells were incubated with 100 lL of media (Dul-
becco’s modified Eagle’s medium [DMEM] supplemented
with 5% fetal calf serum and 2 mm glutamine) containing
increasing concentrations of native purified saporin (Sigma,
St Louis, MO, USA) or native purified ricin, recombinant
DT, recombinant ricin-6K [25], recombinant ricin A-chain-
KDEL (RTA-KDEL) used as controls for the different
drug treatments. After the appropriate time of incubation
at 37 °C residual protein synthesis was measured by incuba-
ting cells at 37 °C for 90 min in the presence of 1 lCi of
[
35
S]-methionine in 50 lL of NaCl ⁄ P
i
per well. Labeled pro-
teins were precipitated by washing with 5% TCA followed
by NaCl ⁄ P
i
and, after the addition of 200 lL of Optiphase
‘SuperMix’ scintillation fluid (Wallac, PerkinElmer-LAS,
UK) per well, plates were counted in a MicroBeta 1450
Trilux counter (PerkinElmer-LAS, UK). Experiments using
drug treatments were carried out by exactly the same
method except that HeLa cells were pretreated for 15 min
with 10 lm BFA, 60 min with 20 lm clasto-lactacystin-

b-lactone, 60 min with 100 lm chloroquine or 30 min with
500 nm bafilomycin A1 prior to the addition of toxin dilu-
tions. In all cases the appropriate drug was maintained at
the same concentration in both the toxin dilutions and the
labeling mix.
RNA extraction and depurination
Cells were grown to 80% confluence in 175 cm
2
flasks in
DMEM supplemented with 5% fetal calf serum and 2 mm
glutamine before incubating for 18 h with increasing con-
centrations of saporin. Cells were removed from the plates
by treating with 5 mm EDTA for 10 min at 37 °C before
pelleting through 30 mL of media for 5 min at 500 g. Cell
pellets were resuspended in 1 mL of Trizol (Invitrogen,
Carlsbad, CA, USA) and passed through a needle
(0.6 · 25 microlance) three times before pelleting at
12 000 g for 10 min at 4 °C. The supernatants were
removed and incubated at room temperature for 5 min
before adding 0.2 mL chloroform, vortexing briefly and
spinning at 12 000 g for 2 min at room temperature. We
added 0.5 mL of propan-2-ol to the aqueous layer and after
incubation at room temperature for 15 min the samples
were spun at 12 000 g for 15 min at 4 °C. The pellets were
washed with 1 mL of 75% ethanol prior to vacuum drying
and quantitation. Four micrograms of isolated RNA was
treated with 20 lL of acetic aniline for 2 min at 60 °C, pre-
cipitated using 0.1 vol. of 7 m ammonium acetate and
2.5 vol. of 100% ethanol and pelleted at 12 000 g for
30 min at 4 °C. Pellets were washed with 1 mL of 75% eth-

anol prior to vacuum drying and the RNA was resuspend-
ed in 20 lL of 60% formamide in 0.1 · TPE (3.6 mm
Tris ⁄ HCl pH 8.0, 3 mm sodium dihydrogen phosphate,
0.1 mm EDTA) and electrophoresed on a denaturing form-
amide gel (1.2% agarose, 50% formamide, 0.1 · TPE).
Labeling with Cy3
Saporin or ricin was labeled with Cy3 using Cy3-reactive
dye-pack. Briefly, saporin or ricin in 0.1 m sodium carbon-
ate buffer (pH 8.5) was incubated with the dye for 1 h at
room temperature and Cy3–saporin was separated from
free dye on a PD-10 column (Amersham Pharmacia Biotech
Italia, Milan, Italy) before concentrating in microcon cen-
trifugal filters (Millipore-Amicon, Madison, WI, USA). A
molar ratio between 1.7 and 2.2 mol of Cy3 per mole of
saporin was incorporated. The cytotoxicity of the Cy3–sap-
orin was assayed and was unchanged as compared to the
native saporin, used for the labeling.
Saporin uptake and intracellular
immunofluorescence
Green monkey kidney Vero cells were seeded at 5 · 10
5
cell-
sÆmL
)1
onto coverslips in 12-well plates and grown over-
night at 37 °C in DMEM supplemented with 5% fetal
calf serum and 2 mm glutamine. Cy-3-labeled saporin
(100 lgÆmL
)1
) or ricin was added to the cells for 1 or 4 h

before washing with NaCl ⁄ P
i
. Cells were fixed and permea-
bilized in ice-cold methanol for 4 min prior to immuno-
staining. Nonspecific antibody binding was blocked by
incubating with 3% BSA in NaCl ⁄ P
i
for 30 min before incu-
bating with the indicated primary antibody followed by the
appropriate secondary antibody (Alexafluor, Invitrogen-
Molecular Probes, Eugene, OR, USA) each for 1 h. Cover-
slips were mounted and viewed by confocal microscopy
(Leica Microsystems, Mannheim, Germany).
Cytotoxicity experiments
Vero cells were used to compare the cytotoxic activities of
SAPwt and SAPKDEL to ricin A chain (RTAwt) or to
RTAKDEL or to the ricin holotoxin and DT, used as con-
trols for the different drug treatments. The cells were plated
R. Vago et al. Saporin trafficking in intoxicated mammalian cells
FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS 4991
into 48-well plate at a density of 1.6 · 10
4
cellsÆwell
)1
and
exposed to serial logarithmic dilutions of each protein
toxin, in quadruplicate. At the end of the exposure period
(4 or 30 h), cells were washed with NaCl ⁄ P
i
and transfected

with a reporter plasmid encoding firefly cytosolic luciferase
under a cytomegalovirus promoter and allowed to express
luciferase over the following 18 h. Lysates were read in a
luminometer following manufacturer’s instructions (Pro-
mega, Milan, Italy). The luciferase activity was quantitated
in each sample and results were expressed in relative light
units (RLU), as a percentage of that seen in untreated cells.
The RLUs per mg of total protein lysate served as an inter-
nal control to assess efficiency of transfections. Results are
referred to 100% luciferase expression in the untreated con-
trol samples and the dose of toxin inhibiting luciferase
activity by 50% relative to controls corresponds to the
ID
50
. Vero cells are efficiently killed by ricin holotoxin and
by DT [27]. In our assays, ricin showed an ID
50
between 1
and 2 pm in agreement with previous published data [8,12].
The ID
50s
of RTA, RTAKDEL were compared with those
of SAPwt and SAPKDEL. Kinetics of intoxication was
analyzed for SAPwt and SAPKDEL toxins and 4 h intoxi-
cation period was chosen for the comparison with
RTA ⁄ RTAKDEL using the different drug treatments. BFA
(Sigma-Aldrich, Milan, Italy), MG-132 proteasome inhib-
itor (Calbiochem, San Diego, CA, USA) and chloroquine
(Sigma) were first tested at different concentrations to min-
imize the inhibition of luciferase expression caused by the

drug itself. We observed a high toxicity, in particular when
using BFA observing a dose-dependent inhibition of pro-
tein synthesis. A final set of experiments in the presence or
absence of BFA (0.5 lgÆmL
)1
), MG-132 (10 lm) or chlo-
roquine (10 lm) with a fixed amount of toxins for a 4 h
total exposure was performed. Each experiment was per-
formed in quadruplicate samples and in several independent
replicates, as indicated in the results section.
Standard cytotoxicity assays were performed testing
increasing concentrations of each recombinant toxin and the
secreted ATF–SAP wild-type or COOH-mutant polypep-
tides in U937 human promyelocytic leukemia cells and for
recombinant saporins also in human T-cell leukemia cell line
HSB-2 or in the Burkitt lymphoma cell line Ramos, treated
as described in Flavell et al. [51]. U937 cells were plated in
96-well plates at a cell density of 2 · 10
4
cellsÆwell
)1
and trea-
ted as previously described [22]. Cells were typically incuba-
ted 48 h at 37 °C in the presence of serial logarithmic
dilutions (prepared in tissue culture medium) of the recom-
binant polypeptides (ATF–SAP chimeras from the 72 h con-
ditioned medium of protected oocytes, see below). At the end
of the exposure period, the cells were washed with NaCl ⁄ P
i
,

pulse-labeled for 4 h with 0.5 l CiÆ well
)1
l-[4,5-
3
H] leucine
(37 TBqÆmmol
)1
, Amersham Pharmacia Biotech, Piscata-
way, NJ) and total incorporation of radioactivity into pro-
tein was measured by liquid scintillation counting after
harvesting cells on glass fiber filters. Cytotoxicity was calcula-
ted by measuring the dose of toxin inhibiting by 50% incor-
poration of untreated cells (ID
50
). At least two independent
experiments were conducted, each in triplicate.
Construction of COOH-mutant ATF–SAPorin and
saporin expression plasmids
Synthetic oligonucleotides were purchased from Genset.
pBSpAS, a preATF–SAP-containing vector [22] was muta-
genized, inserting a Aat1 ⁄ Stu1 site into the original stop
codon. This was achieved by amplifiying ~ 500 bp HpaI–
EcoRI DNA fragment with the Pfu thermostable-poly-
merase (Stratagene, La Jolla, CA, USA) and the following
oligonucleotides: forward Aat1: 5-GAGTTAACCGC
CCTTTTCCCAGAGGCCACAA-3OH; (bold, HpaI
sequence); reverse Aat1: 5-CGGAATTCGCCTCGTTTGA
GGCCTTTGGTT-3OH; (bold, EcoRI sequence). This
Aat1 ⁄ stop minus HpaI–EcoRI-restricted DNA fragment
was substituted to wild-type ATF–SAP HpaI–EcoRI DNA

in pBSpAS yielding the recipient vector pBSpAS- (stop-).
Complementary synthetic oligonucleotides with Aat1-
and EcoRI-compatible ends were synthesized which enco-
ded an in frame KDEL or REDLK COOH amino acid
sequence followed by a stop codon. Sense oligonucleotides
were phosphorylated with T4-polynucleotide kinase [52]
and subsequently annealed with each complementary oligo-
nucleotide before ligation into Aat1–EcoRI-digested
pBSpAS(stop-). DNA sequencing was performed using the
Thermo sequenase (Amersham Pharmacia Biotech), follow-
ing manufacturer’s instructions. The ApaI–NotI fragments
from preATF–SAP and from each of the confirmed positive
clones were purified and ligated into ApaI–NotI digested
pSP64TA ⁄ N (courtesy of Giovanna Chimini, CNRS, Mar-
seille, France) yielding, respectively, pSP64TA ⁄ N-pAS
encoding the pATF–SAP wild-type or the pATF–SAP
mutants, those carrying KDEL-like sequences termed
pATF–SAPREDLK (REDLK) or pATF–SAPKDEL
(KDEL) (Fig. 4). All the COOH-extended mutants also
share three extra amino acids (Ala, Ser and Glu) introduced
by this cloning strategy.
The construct encoding SAP with a KDEL C-terminal
extension was obtained by substituting in pet-11d-SAP-3
the BamHI–EcoRI fragment encoding saporin wild-type
with an equivalent one derived from BamHI–EcoRI diges-
tion of pATF–SAPKDEL, as in Fabbrini et al. [18], giving
rise to pet-11d-SAPKDEL. As a control, recombinant SAP
with an extended C-terminus encoding SEARRL was also
obtained, by annealing complementary synthetic oligonucleo-
tides to Aat1–EcoRI-opened pet-11d-SAPKDEL. Expression

and purification of the recombinant proteins was performed
essentially as described in Fabbrini et al. [18]. Recombinant
SAPARRL polypeptides retain the same RIP activity in vitro
as saporin wild-type (data not shown). Purified SAPKDEL
polypeptides were specifically immunoprecipitated with
Saporin trafficking in intoxicated mammalian cells R. Vago et al.
4992 FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS
monoclonal anti-KDEL sera (StressGen Bioreagents, Vic-
toria, Canada, 1 : 500). Some of the recombinant proteins
were sequenced to confirm their C-terminal extensions.
In vitro transcription and translation assays were essen-
tially performed as described in Fabbrini et al. [22]. An
ELISA [19] was used to determine the concentration of
each chimera secreted in the oocyte medium (see below).
The cell-free specific RIP activity of each toxin was evalu-
ated by an in vitro inhibition translation assay, using BMV
RNA as reporter, as described in Fabbrini et al. [18], assay-
ing serial dilutions made in NaCl ⁄ P
i
of SAPwt, SAPKDEL,
RTAwt, RTAKDEL or of each secretory chimera. Data
are expressed as the concentration inhibiting the 50% BMV
translation (IC
50
). IC
50s
of the COOH-extended chimeras
were comparable with that of E. coli-expressed ATF–sapo-
rin [19], between 25 and 30 pm (data not shown).
Xenopus laevis oocyte preparation, micro-

injections, pulse-chase labeling, homogenization
and storage
Microinjections ( 40 nLÆoocyte
)1
) of cRNAs together with
goat anti-saporin Igs and pulse-labeling of the oocytes were
performed as described previously [22]. Oocyte lysates or
conditioned media were stored in aliquots at )80 °C until
use. In some of the experiments, unlabeled oocytes were
also used. Control oocytes were injected with immune Igs
alone. The 72 h oocyte incubation media of oocytes co-
injected with goat immune Igs together with cRNA coding
for the different chimeric toxins were collected and the
amount of secreted chimera measured by ELISA using a
biotinylated anti-ATF conformational monoclonal serum
and human prourokinase as a standard [19]. The concentra-
tions of the saporin-based chimeras in the 72 h medium
were in the micromolar range.
Immunoprecipitation, western blotting and
SDS ⁄ PAGE analyses
An anti-(saporin rabbit epitope) serum, precleared against
Xenopus oocyte extracts [22], was used to immunoprecipitate
equivalent amounts of lysate (oocyte) or oocyte incubation
medium. In some experiments, phaseolin polypeptides were
coexpressed and an anti-(phaseolin rabbit epitope) serum
was used for immunoselections, as in Fabbrini et al. [22]. The
polypeptides either immunoprecipitated or from the unlabe-
led oocyte incubation media were analyzed by SDS ⁄ PAGE
(15% acrylamide, 0.075% bisacrylamide) using the system of
Laemmli. Gels were treated for fluorography as described

previously [22] and some films were subjected to densito-
metry analysis. Alternatively, polypeptides were electro-
blotted onto 0.2 lm pore nitrocellulose membranes and
membranes were blocked using TBST buffer (100 mm NaCl,
10 mm Tris ⁄ HCl, pH 7.2, containing 0.1% v ⁄ v Tween-20)
and incubated for 60 min with either rabbit anti-SAP serum
(1 : 2000) or anti-ATF (conformational monoclonal anti-
body 5B4, 1 : 500) or with a monoclonal anti-KDEL serum
(StressGen, 1 : 500) then incubated with the specific HRP-
conjugated Igs either goat anti-rabbit at 1 : 30 000 or anti-
mouse in TBST ⁄ 5% BSA followed by chemioluminescence
detection (Supersignal, Rockford, IL, USA).
Acknowledgements
We thank Alexandre Mezghrani, Jez Simpson, Roberto
Sitia and Tatiana Solda
`
for the critical reading of
this manuscript. We are grateful to Serena Camerini
(Dibit-HSR, Milan) for the amino acid sequencing of
some recombinant protein, to Marco Colombatti for
generous gift of reagents and to Lucia Monaco for
precious suggestions and continuous support. This
research was supported by CNR, AIRC, COFIN, Uni-
versity of L’Aquila and Leukaemia Busters grants and
work at Warwick was supported by the UK Biotech-
nology and Biological Sciences Research Council grant
88 ⁄ B16355 and the Wellcome Trust Programme Grant
([063058 ⁄ Z ⁄ 00 ⁄ Z]).
References
1 Sandvig K & van Deurs B (2002) Transport of protein

toxins into cells: pathways used by ricin, cholera toxin
and Shiga toxin. FEBS Lett 529, 49–53.
2 Kreitman RJ & Pastan I (1995) Importance of the glu-
tamate residue of KDEL in increasing the cytotoxicity
of Pseudomonas exotoxin derivatives and for increased
binding to the KDEL receptor. Biochem J 307, 29–37.
3 Jackson ME, Simpson JC, Girod A, Pepperkok R,
Roberts LM & Lord JM (1999) The KDEL retrieval
system is exploited by Pseudomonas exotoxin A, but not
by Shiga-like toxin-1, during retrograde transport from
the Golgi complex to the endoplasmic reticulum. J Cell
Sci 112, 467–475.
4 Wilson DW, Lewis MJ & Pelham HR (1993) pH-depen-
dent binding of KDEL to its receptor in vitro. J Biol
Chem 268, 7465–7468.
5 Hessler JL & Kreitman RJ (1997) An early step in
Pseudomonas exotoxin action is removal of the terminal
lysine residue, which allows binding to the KDEL recep-
tor. Biochemistry 36, 14577–14682.
6 Plemper RK & Wolf DH (1999) Retrograde protein
translocation: ERADication of secretory proteins in
health and disease. Trends Biochem Sci 24, 266–270.
7 Lord JM (1996) Go outside and see the proteasome.
Curr Biol 6, 1067–1069.
8 Rapak A, Falnes PO & Olsnes S (1997) Retrograde
transport of mutant ricin to the endoplasmic reticulum
R. Vago et al. Saporin trafficking in intoxicated mammalian cells
FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS 4993
with subsequent translocation to cytosol. Proc Natl
Acad Sci USA 94, 3783–3788.

9 Lord J & Roberts L (1998) Toxin entry – retrograde
transport through the secretory pathway. J Cell Biol
140, 733–736.
10 Wesche J, Rapak A & Olsnes S (1999) Dependence of
ricin toxicity on translocation of the toxin A-chain from
the endoplasmic reticulum to the cytosol. J Biol Chem
274, 34443–34449.
11 Wales R, Chaddock JA, Roberts LM & Lord JM
(1992) Addition of an ER retention signal to the ricin A
chain increases the cytotoxicity of the holotoxin. Exp
Cell Res 203, 1–4.
12 Wales R, Roberts LM & Lord JM (1993) Addition of
an endoplasmic reticulum retrieval sequence to ricin A
chain significantly increases its cytotoxicity to mamma-
lian cells. J Biol Chem 268, 23986–23990.
13 Fitzgerald D (1996) Why toxins! Semin Cancer Biol 7,
87–95.
14 Sandvig K, Spilsberg B, Lauvrak SU, Torgersen ML,
Iversen TG & van Deurs B (2004) Pathways followed
by protein toxins into cells. Int J Med Microbiol 293,
483–490.
15 Teter K, Jobling MG & Holmes RK (2003) A class of
mutant CHO cells resistant to cholera toxin rapidly
degrades the catalytic polypeptide of cholera toxin and
exhibits increased endoplasmic reticulum-associated
degradation. Traffic 4, 232–242.
16 Wu YH, Shih SF & Lin JY (2004) Ricin triggers apop-
totic morphological changes through caspase-3 cleavage
of BAT3. J Biol Chem 279, 19264–19275.
17 Jenkins CE, Swiatoniowski A, Issekutz AC & Lin TJ

(2004) Pseudomonas aeruginosa exotoxin A induces
human mast cell apoptosis by a caspase-8 and
-3-dependent mechanism. J Biol Chem 279 , 37201–
37207.
18 Fabbrini MS, Rappocciolo E, Carpani D, Solinas M,
Valsasina B, Breme U, Cavallaro U, Nykjaer A, Rovida
E, Legname G et al. (1997) Characterization of a sapo-
rin isoform with lower ribosome-inhibiting activity.
Biochem J 322, 719–727.
19 Fabbrini MS, Carpani D, Bello-Rivero I & Soria MR
(1997) The amino-terminal fragment of human uroki-
nase directs a recombinant chimeric toxin to target cells:
internalization is toxin mediated. FASEB J 11, 1169–
1176.
20 Frankel A, Kreitman RJ & Sausville E (2000) Targeted
toxins. Clin Cancer Res 6, 326–334.
21 Olson D, Pollanen J, Hoyer-Hansen G, Ronne E,
Sakaguchi K, Wun T-C, Appella E, Dano K & Blasi F
(1992) Internalization of the urokinase-plasminogen
activator inhibitor type-1 complex is mediated by the
urokinase receptor. J Biol Chem 267, 9129–9133.
22 Fabbrini MS, Carpani D, Soria MR & Ceriotti A
(2000) Cytosolic immunization allows the expression of
preATF-saporin chimeric toxin in eukaryotic cells.
FASEB J 14, 391–398.
23 Yoshida T, Chen CC, Zhang MS & Wu HC (1991) Dis-
ruption of the Golgi apparatus by brefeldin A inhibits
the cytotoxicity of ricin, modeccin, and Pseudomonas
toxin. Exp Cell Res 192, 389–395.
24 Sandvig K, Prydz K, Hansen SH & van Deurs B (1991)

Ricin transport in brefeldin A-treated cells: correlation
between Golgi structure and toxic effect. J Cell Biol
115, 971–981.
25 Deeks ED, Cook JP, Day PJ, Smith DC, Roberts LM
& Lord JM (2002) The low lysine content of ricin A
chain reduces the risk of proteolytic degradation after
translocation from the endoplasmic reticulum to the
cytosol. Biochemistry 41, 3405–3413.
26 Hazes B & Read RJ (1997) Accumulating evidence
suggests that several AB-toxins subvert the
endoplasmic reticulum-associated protein degradation
pathway to enter target cells. Biochemistry 36, 11051–
11054.
27 Falnes PO & Sandvig K (2000) Penetration of protein
toxins into cells. Curr Opin Cell Biol 2, 407–413.
28 Collier RJ (2001) Understanding the mode of action of
diphtheria toxin: a perspective on progress during the
20th century. Toxicon 39, 1793–1803.
29 Seetharam S, Chaudhary VK, FitzGerald D & Pastan I
(1991) Increased cytotoxic activity of Pseudomonas exo-
toxin and two chimeric toxins ending in KDEL. J Biol
Chem 266, 17376–17381.
30 Flavell DJ (1998) Saporin immunotoxins. Curr Top
Microbiol Immunol 234, 57–61.
31 Fabbrini MS, Flavell DJ & Ippoliti R (2003) Plant pro-
tein toxins: structure, function and biotechnological
applications. In Bacterial, Plant and Animal Toxins
(Ascenzi P, Polticelli F, Visca P, eds), pp. 69–99.
Research Signpost, Kerala, India.
32 Rajagopal V & Kreitman RJ (2000) Recombinant toxins

that bind to the urokinase receptor are cytotoxic with-
out requiring binding to the alpha (2)-macroglobulin
receptor. J Biol Chem 275, 7566–7573.
33 Ippoliti R, Lendaro E, Benedetti PA, Torrisi MR,
Belleudi F, Carpani D, Soria MR & Fabbrini MS
(2000) Endocytosis of a chimera between human pro-
urokinase and the plant toxin saporin: an unusual inter-
nalization mechanism. FASEB J 14, 1335–1444.
34 Chan WL, Shaw PC, Tam SC, Jacobsen C, Gliemann J
& Nielsen MS (2000) Trichosanthin interacts with and
enters cells via LDL receptor family members. Biochem
Biophys Res Commun 270, 453–457.
35 Rodighiero C, Tsai B, Rapoport TA & Lence WI
(2002) Role of ubiquitination in retro-translocation of
cholera toxin and escape of cytosolic degradation.
EMBO Report 3, 1222–1227.
36 Savino C, Federici L, Ippoliti R, Lendaro E &
Tsernoglou D (2000) The crystal structure of saporin
Saporin trafficking in intoxicated mammalian cells R. Vago et al.
4994 FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS
SO6 from Saponaria officinalis and its interaction with
the ribosome. FEBS Lett 470, 239–243.
37 Santanche
`
S, Bellelli A & Brunori M (1997) The un-
usual stability of saporin, a candidate for the synthesis
of immunotoxins. Biochem Biophys Res Commun 234,
129–132.
38 Wesche J, Wiedlocha A, Falnes PO, Choe S & Olsnes S
(2000) Externally added aFGF mutants do not require

extensive unfolding for transport to the cytosol and the
nucleus in NIH ⁄ 3T3 cells. Biochemistry 39, 15091–
15100.
39 Malecki J, Wiedlocha A, Wesche J & Olsnes S (2002)
Vesicle transmembrane potential is required for translo-
cation to the cytosol of externally added FGF-1. EMBO
J 21, 4480–4490.
40 Girod A, Storrie B, Simpson JC, Johannes L, Goud B,
Roberts LM, Lord JM, Nilsson T & Pepperkok R
(1999) Evidence for a COP-I-independent transport
route from the Golgi complex to the endoplasmic reticu-
lum. Nat Cell Biol 1, 423–430.
41 Chen A, Hu T, Mikoryak C & Draper RK (2002) Retro-
grade transport of protein toxins under conditions of
COPI dysfunction. Biochim Biophys Acta 1589, 124–139.
42 Johannes L, Tenza D, Antony C & Goud B (1997)
Retrograde transport of KDEL-bearing B-fragment of
Shiga toxin. J Biol Chem 272, 19554–19561.
43 Fujinaga Y, Wolf AA, Rodighiero C, Wheeler H, Tsai
B, Allen L, Jobling MG, Rapoport T, Holmes RK &
Lencer W (2003) Gangliosides that associate with lipid
rafts mediate transport of cholera and related toxins
from the plasma membrane to endoplasmic reticulm.
Mol Biol Cell 14, 4783–4793.
44 Mezghrani A, Courageot J, Mani JC, Pugniere M,
Bastiani P & Miquelis R (2000) Protein-disulfide isomer-
ase (PDI) in FRTL5 cells. pH-dependent thyroglobulin ⁄
PDI interactions determine a novel PDI function in the
post-endoplasmic reticulum of thyrocytes. J Biol Chem
275, 1920–1929.

45 Llorente A, Lauvrak SU, van Deurs B & Sandvig K
(2003) Induction of direct endosome to endoplasmic
reticulum transport in Chinese hamster ovary (CHO)
cells (LdlF) with a temperature-sensitive defect in epsi-
lon-coatomer protein (epsilon-COP). J Biol Chem 278,
35850–35855.
46 Fiebiger E, Story C, Ploegh H & T (2002) Visualization
of the ER-to-cytosol dislocation reaction of a type I
membrane protein. EMBO J 21, 1041–1053.
47 Tirosh B, Furman MH, Tortorella D & Ploegh HL
(2003) Protein unfolding is not a prerequisite for endo-
plasmic reticulum-to-cytosol dislocation. J Biol Chem
278, 6664–6672.
48 Smith DC, Lord JM, Robert LM, Tartour E &
Johannes L (2002) 1st class ticket to class I: protein tox-
ins as pathfinders for antigen presentation. Traffic 3,
697–704.
49 Koopmann JO, Albring J, Huter E, Bulbuc N, Spee P,
Neefjes J, Hammerling GJ & Momburg F (2000) Export
of antigenic peptides from the endoplasmic reticulum
intersects with retrograde protein translocation through
the Sec61p channel. Immunity 13, 117–127.
50 Sandvig K & van Deurs B (2000) Entry of ricin and
Shiga toxin into cells: molecular mechanisms and medi-
cal perspectives. EMBO J 19, 5943–5950.
51 Flavell D, Boehm D, Noss A, Warnes S & Flavell S
(2001) Therapy of human T-cell acute lymphoblastic
leukaemia with a combination of anti-CD7 and anti-
CD38-SAPORIN immunotoxins is significantly better
than therapy with each individual immunotoxin. Br J

Cancer 84, 571–578.
52 Sambrook T, Fritsch EF & Maniatis T (1989) Molecular
Cloning. A Laboratory Manual. Cold Spring Harbor
Laboratory Press, Cold Spring Harbor, NY.
R. Vago et al. Saporin trafficking in intoxicated mammalian cells
FEBS Journal 272 (2005) 4983–4995 ª 2005 FEBS 4995