Tài liệu Báo cáo khoa học: The role of electrostatic interactions in the antitumor activity of dimeric RNases docx

The role of electrostatic interactions in the antitumor
activity of dimeric RNases
Eugenio Notomista
1
, Jose
´
Miguel Manchen
˜
o
2
, Orlando Crescenzi
3
, Alberto Di Donato
1
,
Jose
´
Gavilanes
4
and Giuseppe D’Alessio
1
1 Dipartimento di Biologia Strutturale e Funzionale, Universita
`
di Napoli Federico II, Napoli, Italy
2 Grupo de Cristalografı
´
a Macromolecular y Biologı
´
a Estructural, Instituto Rocasolano, Madrid, Spain
3 Dipartimento di Chimica, Universita
`

di Napoli Federico II, Napoli, Italy
4 Departamento de Bioquı
´
mica y Biologı
´
a Molecular I, Universidad Complutense, Madrid, Spain
The superfamily of pancreatic-type RNases [1] includes
several members capable of carrying out ‘special’
actions, i.e., actions other than catalytic, although
strictly dependent on their catalytic, RNA degrading
action [2]. These actions could be linked to physiologi-
cal functions, as in the case of the angiogenic action of
angiogenins [3], or due to the mere reflection in the
laboratory assays, mirrors proposed in the experiment,
of unknown functions, as may be the case of the anti-
fertility action of seminal RNase [4]. Particular atten-
tion has been given to the cytotoxic action of some
RNases, especially because they often appear to be
selective for tumor cells [5].
Many studies have been devoted to the mechanism
of action of these antitumor RNases. However, a con-
clusive understanding of why these RNases kill cells,
especially why some of them selectively kill tumor
cells, has not been obtained. The correlation has been
stressed [6] between the ability of certain RNases to
display a cytotoxic action and their ability to evade
the strong, neutralizing action of the cytosolic RNase
inhibitor (cRI), a 50-kDa protein containing 16 leu-
cine-rich repeat motifs [7,8]. In fact, onconase from
Rana pipiens eggs [9], and seminal RNase from

Bos taurus seminal vesicles [10], among the most stud-
ied natural cytotoxic RNases, do not bind cRI, and
are both totally resistant to its inhibitory action [8].
On the other hand, RNases with no cytotoxic action,
such as bovine or human pancreatic RNase, and a
very high affinity for cRI, acquire the ability to kill
Keywords
antitumor RNases; electrostatic interactions;
electrostatic interaction energy; RNases;
transport through membranes
Correspondence
G. D’Alessio, Dipartimento di Biologia
Strutturale e Funzionale, Universita
`
di Napoli
Federico II, Via Cinthia, I-80126 Napoli, Italy
Fax: +39 081 679159
Tel: +39 081 679157
E-mail:
(Received 12 April 2006, revised 31 May
2006, accepted 12 June 2006)
doi:10.1111/j.1742-4658.2006.05373.x
The cytotoxic action of some ribonucleases homologous to bovine pancre-
atic RNase A, the superfamily prototype, has interested and intrigued
investigators. Their ribonucleolytic activity is essential for their cytotoxic
action, and their target RNA is in the cytosol. It has been proposed that
the cytosolic RNase inhibitor (cRI) plays a major role in determining the
ability of an RNase to be cytotoxic. However, to interact with cRI RNases
must reach the cytosol, and cross intracellular membranes. To investigate
the interactions of cytotoxic RNases with membranes, cytotoxic dimeric

RNases resistant, or considered to be resistant to cRI, were assayed for
their effects on negatively charged membranes. Furthermore, we analyzed
the electrostatic interaction energy of the RNases complexed in silico with
a model membrane. The results of this study suggest that close correlations
can be recognized between the cytotoxic action of a dimeric RNase and its
ability to complex and destabilize negatively charged membranes.
Abbreviations
ANTS, 1,3,6-trisulphonate-8-aminonaphtalene; cRI, cytosolic RNase inhibitor; DMPG, dimyristoylphosphatidylglycerol; DMPS,
dimyristoylphosphatidylserine; DPH, 1,6-diphenyl-1,3,5-hexatriene; DPX, p-xylenebispyridinium bromide; ECP, eosinophil cationic protein;
EDN, eosinophil-derived neurotoxin; EIE, electrostatic interaction energy; HP-RNase, human pancreatic RNase; kT, product of the Boltzmann
constant by absolute temperature in K; PG, phosphatidylglycerol; RET, resonance energy transfer; SVT2, malignant murine fibroblasts.
FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS 3687
cells when they are engineered into cRI-resistant
RNases [6,11–14]. These results led to the description
of the cytosolic RNase inhibitor as a sentry to guard
cells from, and inactivate, potentially cytotoxic foreign
RNases [6].
Recently however, it has been reported [15] that
HeLa cells deprived through RNA silencing of any cRI
in their cytosol, are still not affected by noncytotoxic
RNases. This indicates that noncytotoxic, cRI-sensitive
RNases do not become cytotoxic when cRI, the postu-
lated ‘sentry’, is absent. Furthermore, artificial dimers
of RNase A, obtained through lyophilization from
acetic acid, have been found to bind with high affinity
the cytosolic RNase inhibitor [16], but also to be cyto-
toxic on malignant cell lines [17]. These results have
cast shadows on the prospect that the antitumor action
of RNases be based solely, or primarily, on its resist-
ance to the cytosolic RNase inhibitor.

RNase catalytic activity is an absolute requirement
for the cytotoxic action of all cytotoxic RNases tested
[5]. Moreover, when the effects of cytotoxic RNases on
cellular RNA were studied, cytosolic rRNA [18] or
tRNA [19] were found to be the targets of cytotoxic
RNases. Thus, we can conclude that to exert its cyto-
toxic action, an RNase must reach the cytosol. Fur-
thermore, given that cRI is a cytosolic protein [6,7],
the characterization of an RNase as a cRI-resistant or
-sensitive RNase can only occur if the RNase reaches
the cytosol. These considerations lead to the conclu-
sion that to approach the cytosol, where their cyto-
toxic action is exerted, cytotoxic RNases must cross
not only the plasma membrane, but intracellular mem-
branes as well.
A study of the intracellular journey of cytotoxic,
dimeric seminal RNase has revealed that it is internal-
ized by various types of malignant cells, and has access
to endosomes, whereas noncytotoxic bovine pancreatic
RNase A is not internalized [18,20]. Interestingly,
when RNase A was made dimeric and cytotoxic by
site-directed mutagenesis, it gained access to endo-
somes [20]. It should be added that endocytosed sem-
inal RNase is found in endosomes both in malignant
and normal cells, but only in malignant cells does the
RNase reaches the trans-Golgi network, and then the
cytosol, where it degrades rRNA [18,20].
These data suggest that cell membranes can discrim-
inate RNases, and may play an important role in
determining the cytotoxicity of an RNase by simply

allowing, or not, an RNase to pass from one cell com-
partment to another. On the other hand, the RNase
itself is expected to present to the membrane which it
should permeate structural elements that determine
recognition and favor its access through that mem-
brane.
For a systematic study of the interactions between
membranes and cytotoxic RNases, we report here the
effects on the stability of artificial membranes of nat-
ural and engineered dimeric RNases with different
degrees of cytotoxic activity. Their electrostatic interac-
tions with a model membrane are then compared
through in silico analyses. The results of these investi-
gations indicate that close correlations can be recog-
nized between the ability of an RNase to destabilize
a membrane, its attraction to a negatively charged
membrane, and its cytotoxic action.
Results
The RNases
The RNases investigated in this study are described in
Table 1. They are cationic proteins with cytotoxic activ-
ity: BS-RNase [10] is a natural dimeric RNase; RNase-
AA [21], RNase-AA-G [21], and RNase-AA-GG [22]
are dimeric variants engineered from bovine pancreatic
RNase A; HHP-RNase [11] is a dimeric variant engine-
ered from human pancreatic RNase (HP-RNase).
Effects on vesicle aggregation
All RNases investigated are cationic proteins, with high
isoelectric point values. Thus we investigated the effects
Table 1. Natural and engineered RNases investigated in this study. IC

50
is the RNase concentration producing half-maximal cytotoxicity on
SVT2 malignant cells [5,11,14].
Abbreviations Natural or modified RNase Structure pI IC
50
(lgÆmL
)1
)
RNase A Bovine pancreatic RNase A Monomeric 9.8 > 200
RNase-AA A19P,Q28L,K31C,S32C-RNase A Dimeric 9.6 82
RNase-AA-G D38G-RNase-AA Dimeric 9.8 64
RNase-AA-GG D38G,E111G-RNase-AA Dimeric 9.9 27
HP-RNase Human pancreatic RNase Monomeric 10.1 > 200
HHP-RNase Q28L,R31C,R32C,N34K-HP-RNase Dimeric 9.9 26
BS-RNase Bovine seminal RNase Dimeric 10.3 22
Electrostatic interactions and antitumor RNases E. Notomista et al.
3688 FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS
of these RNases on the aggregation of negatively
charged dimyristoylphosphatidylglycerol (DMPG) and
dimyristoylphosphatidylserine (DMPS) vesicles. Aggre-
gation was measured from the increase of absorbance
at 360 nm due to the enhanced turbidity as engendered
by vesicle aggregation. The results with DMPG vesicles
are presented here, for the higher availability to outside
interactions of the negatively charged glycerol-linked
phosphate [23–25]. In DMPS vesicles, the bonding of
the serine carboxylate ion to the adjacent ammonium
group renders the phosphate negative charge less avail-
able [23–25]. In fact, when DMPS vesicles were tested
(data not shown), similar results were obtained in terms

of effects exerted by the RNases under investigation,
only to some extent less powerful.
As previously reported [26], seminal RNase (BS-
RNase), a naturally dimeric RNase, exerts a very
strong effect on the aggregation of DMPG vesicles,
whereas monomeric RNases, namely RNase A and a
monomeric derivative of BS-RNase, exert no signifi-
cant effects. However, the results illustrated in Fig. 1
indicate that the quaternary structure of an RNase is
not a prerequisite for its ability to aggregate vesicles,
as RNase A does not acquire this ability upon dimeri-
zation into RNase-AA.
The replacement in dimeric RNase-AA of one negat-
ively charged residue (Asp38) with an uncharged Gly
residue generates an RNase variant (RNase-AA-G)
capable of aggregating vesicles. The result is even more
striking when two negatively charged residues (Asp38
and Glu111) are replaced with Gly residues, as in
RNase-AA-GG. As shown in Fig. 1, the effects on
DMPG vesicles of RNase-AA-GG are similar to those
obtained with BS-RNase, the RNase with the most
powerful aggregating effect. When DMPS vesicles were
used, similar effects were observed, but were less evi-
dent (data not shown).
It should be added that CD spectra of the three
RNase A dimers (RNase AA, -AA-G, and -AA-GG)
in the far-UV and near-UV regions were virtually iden-
tical, with a slightly different spectrum for RNase-
AA-G in the near-UV region. Thermal denaturation,
measured through the variation of ellipticity at

218 nm, gave an identical midpoint of denaturation at
62 °C for all three RNase dimers, and identical far-UV
CD spectra of the thermally denatured states. These
data (not shown) indicate that the three proteins
apparently possess an identical folded structure under
the conditions used in the experiments with mem-
branes, hence their different effects on the model mem-
branes may not be ascribed to different conformations
of the three proteins.
Effects on bilayer fluidity
Figure 2 illustrates the effects of RNases on the phase
transition profile of DMPG vesicles labeled with 1,6-
diphenyl-1,3,5-hexatriene (DPH), a fluorescent probe
used for measuring fluorescence polarization. Bilayer
fluidity was measured by comparing the difference in
fluorescence anisotropy (Dr) as a function of pro-
tein : DMPG ratio.
Fig. 2. Effect of RNases on the thermotropic behavior of DMPG
vesicles. Symbols: (e) BS-RNase; (+) RNase AA-GG; (n)
HHP-RNase; (n) HP-RNase; (h) RNase AA-G; (m) RNase AA; (·)
RNase A.
Fig. 1. Aggregation of DMPG vesicles induced by RNases. Sym-
bols: (e) BS-RNase; (+) RNase AA-GG; (n) HHP-RNase; (n) RNase
AA-G; (h) RNase AA; (m) HP-RNase; (·) RNase A.
E. Notomista et al. Electrostatic interactions and antitumor RNases
FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS 3689
The data show that most investigated RNases affect
the bilayer fluidity of DMPG membranes, as evidenced
by a decrease of the amplitude of their thermotropic
transition. Only RNase A and RNase AA show no

effects on the thermotropic behavior of the tested
membranes. Surprisingly, also monomeric HP-RNase,
with no effect on membrane aggregation (see above),
was found to affect to a certain extent the bilayer
fluidity (Fig. 2). Similar, albeit minor effects, were
observed when the experiments were carried out on
DMPS vesicles (data not shown).
Effects on membrane fusion
This parameter was evaluated by determining the inter-
mixing of DMPG phospholipids between unlabeled
vesicles and vesicles labeled with a donor ⁄ acceptor pair
of fluorescent probes. When the RNase under test pro-
moted intermixing, a decrease of the energy transfer
between the two probes was measured as resonance
energy transfer (RET). Under the test conditions (see
below), a RET value of 80% indicated that the RNase
under test did not promote any perturbations of the
vesicles bilayer.
Figure 3 shows that all the RNases that promoted
membrane aggregation also engendered membrane
fusion. Even native, monomeric HP-RNase, with no
effects on membrane aggregation, produced a modest
intermixing. RNase-AA-IG instead, which induced a
low, but significant degree of membrane aggregation,
did not show any effects on vesicle intermixing. No
effects of membrane fusion were observed with RNase
A and RNase-AA, or when DMPS vesicles were
tested.
Correlations between the effects of RNases on
membranes and their electrostatic charges

In Table 2 the results described above on the effects
on membrane stability induced by the investigated
RNases are normalized to those obtained with BS-
RNase, taken as 100%. For the normalization the
highest values of Dr, DA
360
and percentage RET were
used for effects on bilayer fluidity, membrane aggrega-
tion and fusion, respectively (see above and Figs 1–3).
When the effects of RNases on negatively charged
membranes were compared with the net charge values
of the RNases, the comparison suggested (Table 2)
that the ability of RNases to destabilize lipid vesicles is
due to an electrostatic component, as the higher posit-
ive net charge of an RNase appears to enhance its
ability to aggregate, fuse, or affect bilayer fluidity of
the negatively charged membranes. However, a hydro-
phobic component may not be excluded in the RNase–
membrane destabilizing interactions. This can be
deduced: (a) from the RNases aggregating effects,
which could be generated by the relief of hydration
repulsion among vesicles occurring upon adsorption of
the RNases to the membrane bilayer; (b) from the
thermotropic behavior of lipids in the presence of
RNases, which can be assigned to lipid molecules
removed from participating in their gel-to-liquid phase
transition.
To further investigate this aspect, values of electro-
static interaction energy (EIE) were calculated for the
interactions between the dimeric RNases and a model

membrane constructed with DMPG lipids (see below).
A series of EIE values were determined by varying the
orientation of each RNase towards the membrane
(Fig. 4). The results of these analyses revealed that a
single highest negative value of EIE (EIE
max
) could be
estimated for each dimeric RNase. This value was
found for all RNases at the unique orientation with
Table 2. Effects on membrane vesicles and properties of the
dimeric RNases investigated in this study.
RNase
Aggregation
(%)
Fluidity
alteration (%)
Fusion
(%)
Positive
net charge
EIE
max
(kT)
BS-RNase 100 100 100 18 )63.8
HHP-RNase 51 103 100 10 )67.3
RNase AA-GG 71 109 100 10 )44.3
RNase AA-G 28 58 0 8 )27.3
RNase AA 12 0 0 6 )14.0
Fig. 3. Lipid mixing of DMPG vesicles induced by RNases.
Symbols: (e) BS-RNase; (+) RNase AA-GG; (n) HHP-RNase; (n)

HP-RNase; (h) RNase AA-G; (m) RNase AA; (·) RNase A.
Electrostatic interactions and antitumor RNases E. Notomista et al.
3690 FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS
r ¼ 180° and i ¼ 90°. In this orientation, characterized
by the strongest electrostatic attraction of the RNases
with the membrane, all RNases pointed their N-ter-
minal regions towards the membrane. Figure 5 illus-
trates the results for some of the RNases (also
supplementary Fig. S1). It should be added that both
models of RNase A or HP-RNase dimers, either swap-
ping their N-terminal ends between subunits, or not
swapping dimers, led to virtually identical EIE values
(data not shown).
The analysis of the electrostatic field generated by
dimeric RNases showed that all dimers possess a posi-
tively charged end located at the N-terminal region
and a negative end located at the C-terminal one.
Hence, they can be described as dipoles with an orien-
tation parallel to the direction from the RNases C-ter-
minal surfaces (negatively charged) to the N-terminal
surfaces (positively charged). In the supplementary
material, Fig. S2 shows the shape of the electrostatic
field of BS-RNase seen from different directions.
Figure S3 compares the electrostatic field of different
Fig. 5. Plot of the EIE values as function of rotation (r) and inclination (i) angles of RNases. The hatched axis at the bottom and right of the
plots show schemes of some conformations with i ¼ 90° and r ¼ 180°, respectively.
Fig. 4. Scheme of the initial complex dimeric RNase–model mem-
brane used to generate the set of orientations for
DELPHI calcula-
tions through rotation of the RNase around its major axis (rotation

angle ‘r’) and variation of the inclination (inclination angle ‘i’) with
respect to the plane of the membrane (parallel to the xy plane).
Cter and Nter are the C-terminal and N-terminal structural regions,
respectively, of the two A and B monomers of the dimeric RNase.
The black circle between the monomers indicates the mass center
of the dimer.
E. Notomista et al. Electrostatic interactions and antitumor RNases
FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS 3691
RNases oriented with the negative end of the dipoles
toward the bottom, and the positive one toward the
top of the figure. Figure S4 shows instead the N-ter-
minal surfaces colored by charge. As expected, in the
orientation that provides the strongest attraction
between the RNase and the model membrane the
RNases dipoles are perpendicular to the negatively
charged membrane.
In Table 2 the EIE
max
values of the RNases investi-
gated in this study are tabulated and compared to the
RNases’ net charge values, and their ability to desta-
bilize negatively charged membranes. A satisfactory
correlation is evident not only between the RNases’
EIE
max
values and their positive net charges, but also
with their ability to destabilize a membrane.
We then considered the possibility of a correlation
between the cytotoxic action of the investigated dimeric
RNases and their ability to destabilize and permeate a

membrane, which in turn appears to be linked to the
attraction of the RNase to a membrane, as evidenced
by the RNase negative EIE
max
values. As a measure of
the RNase cytotoxicity we used the reciprocal of the
IC
50
parameter. The values of IC
50
, the RNase concen-
tration at which 50% of the cytotoxic effect is dis-
played, were obtained [11,21,22] by assaying the RNase
cytotoxic activity on the same cell type, namely SVT2
malignant fibroblasts. When the EIE
max
values of
dimeric, cytotoxic RNases were plotted against the
IC
50
values reported for these RNases (Fig. 6), a direct,
linear correlation was found between the two sets of
values, with a significant correlation coefficient (0.88).
This strongly suggests a positive correlation between the
attraction to a membrane of a dimeric RNase and its
cytotoxic activity.
Discussion
Although many years have gone by since an RNase
was found to have an antitumor action [27], the struc-
tural and functional determinants which render certain

RNases capable of exerting this effect are mostly
obscure. As mentioned above, a primary if not exclu-
sive role of the cytosolic inhibitor in the mechanism of
antitumor RNases cannot be generalized. The only
firmly established fact is that the cytotoxic action of
RNases is based on their RNA degrading activity [5].
The cytosolic localization of target RNA(s), and previ-
ous data on the intracellular journey of antitumor sem-
inal RNase [20], indicate that cytotoxic RNases must
be capable of permeating specific membrane compart-
ments to reach the cytosol. Furthermore, antitumor
seminal RNase has been found to alter the stability of
negatively charged membrane vesicles, with no effects
on positively charged membranes [26].
It has long been known that subcellular membranes
from malignant tissues have a higher content of negat-
ively charged lipids when compared to normal tissues
[28]. More recently, it has been found [29] that some
C2 domains, folded modules from proteins involved in
signal transduction and vesicle trafficking, specifically
associate to negatively charged membranes.
As all these data are suggestive of a role of mem-
branes, in particular of negatively charged membranes,
in the mechanism of action of RNases, we investigated
the ability of a series of RNases to destabilize model
membranes as a representation of their potential to
cross intracellular membranes through their journey
to the cytosol.
We found that cytotoxic, dimeric RNases, natural or
engineered, are able to destabilize negatively charged

membranes, as they strongly affect membrane aggrega-
tion, fluidity and fusion. When the RNases were tested
in silico with a model membrane by determining the
electrostatic energy of their interaction, we found that
their ability to destabilize membranes is correlated with
their electrostatic attraction to the model membrane.
In particular, such property was found to be strictly
dependent on the ability of the RNases to approach
the membranes with their N-terminal regions. Any
other orientations in their contacts with the mem-
branes produced a less attractive electrostatic interac-
tion. Hence RNases that tumble about a membrane
Fig. 6. Correlation between antitumor activity on SVT2 cells and
the highest negative EIE values for covalent dimeric RNases identi-
fied by the abbreviations listed in Table 1. Antitumor activity was
plotted as the reciprocal of IC
50
, the RNase concentration produ-
cing 50% cytotoxicity [11,14]. The square correlation coefficient
(R
2
) is shown.
Electrostatic interactions and antitumor RNases E. Notomista et al.
3692 FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS
until their most membrane-attractive structural ele-
ment, namely their N-terminal regions, interact with
the membrane, are the most capable of affecting and
permeating membranes.
Furthermore, our results also reveal a stringent
correlation between (a) the ability of dimeric RNases

to destabilize membranes with their specific mem-
brane-attractive region, and (b) their cytotoxic
action. This finding strongly suggests that the elec-
trostatic surface potential of a defined 3D district of
these proteins has a role in determining their cyto-
toxic action.
It has been reported that when negative charges are
abolished in noncytotoxic RNases, these become cyto-
toxic [30,31]. This result, obtained with heterogeneous
protein products, only indicates that molecules with a
high content of positive charges are generally toxic to
cells. However, a high content of basic residues may
not be the only basis of RNase cytotoxicity. Two
RNases, homologous to the RNases investigated in
this study, have been isolated from human eosinophils,
namely eosinophil-derived neurotoxin (EDN) and eosi-
nophil cationic protein (ECP). They are RNases with a
high sequence identity (66%), a very high content of
basic residues and high isoelectric points (pI ¼ 11.9
and 10.4, for EDN and ECP, respectively). Yet only
ECP, not EDN, displays cytotoxic activity on eukary-
otic and bacterial cells [32].
Given the recently reported cytotoxic activity of the
noncovalent dimers of RNase A [17], intriguing when
compared to their sensitivity to the cytosolic inhibitor
(see above), we tested their interaction with the model
membrane described above for RNase covalent di-
mers, and calculated their EIE
max
values. Two distinct

noncovalent dimers have been described for RNase A,
in which the dimeric structure is stabilized through
the exchange between subunits of either their N-ter-
minal, or their C-terminal ends [33–35]. We found
that the behavior with the model membrane of the
dimer in which the C-terminal ends are exchanged
between subunits is perfectly superimposable to the
behavior described in this report for any covalently
dimeric RNase. Its highest negative value of EIE was
calculated to be )28 kT. The dimer, in which the sub-
units exchange their N-terminal ends, behaved differ-
ently. It did not present a single, unique value of
highest negative EIE, but two weak negative values of
)13 and )10 kT. We noted that by summing up the
two EIE
max
values, a value of EIE was obtained com-
parable to those calculated for the other dimeric
RNase. As shown in Fig. 7, the EIE
max
values
obtained for the two dimers satisfactorily correlate
with the dimers IC
50
values as reported by Matousek
et al. [17]. Thus, a role of electrostatic membrane
attraction has been identified also for noncovalent,
cRI sensitive RNases.
In conclusion, the data presented here lead to the
proposal that cytotoxic RNases must possess specific

electrostatic features and structural element(s) for
destabilizing and eventually permeating membranes,
and that intracellular membranes play a decisive role
in defining the antitumor action of an RNase.
Experimental procedures
Assays with membranes
Synthetic DMPG and DMPS were purchased from Avanti
Polar Lipids (Alabaster, AL). Egg yolk phosphatidyl-
glycerol (PG) was from Sigma (St Louis, MO). Unless
indicated, lipid vesicles were prepared at 1 mgÆmL
)1
phos-
pholipid concentration in the indicated buffer, by extrusion
through two 0.1-lm polycarbonate filters (Nuclepore,
Costar, Cambridge, MA) in an Extruder (Lipex Biomem-
branes Inc., Vancouver, Canada), as described previously
[26,36–38].
Protein concentration was determined from absorbance
measurements on a Beckman DU-7 (Palo Alto, CA) or
Uvikon 930 (Kontron Instruments, Milan, Italy) spectro-
photometers using the following extinction coefficients:
Fig. 7. Correlation between antitumor activity on HL-60 cells (n)
and ML-2 cells (h) and the highest negative EIE values for
BS-RNase (BS) and RNase A noncovalent dimmers. N
d
, the dimer
with N-terminal swapping between subunits; C
d
, the dimer with
C-terminal swapping between subunits. Antitumor activity was plot-

ted (see legend to Fig. 6) as the reciprocal of IC
50
reported values
[17]. The square correlation coefficients (R
2
) are shown.
E. Notomista et al. Electrostatic interactions and antitumor RNases
FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS 3693
0.695 (E
0.1%
, 280 nm) for RNase A and its variants, 0.465
(E
0.1%
, 278 nm) for BS-RNase.
The absorbance variation at 360 nm produced by the
addition of the proteins to a vesicle suspension was con-
tinuously measured on a Beckman DU-640 spectrophoto-
meter equipped with a high performance temperature
controller. In all assays, controls with no protein were
always carried out.
Intermixing of membrane lipids was analyzed by fluores-
cence RET assays [39]. The RET assay monitors the relief
of fluorescence energy transfer between a donor ⁄ acceptor
pair as the two probes dilute from labeled into unlabeled
bilayers. A vesicle population containing 0.6% (v ⁄ v) N-(lis-
samine rhodamine sulphonyl)-diacylphosphatidylethanol-
amine as acceptor and 1% (v ⁄ v) N-(7-nitro-2-1,3-
benzoxadiazol-4-yl)-dimyristoylphosphatidylethanolamine
as donor (Avanti Polar Lipids) was mixed with unlabeled
vesicles at 1 : 9 molar ratio. The increase of the fluores-

cence emission at 530 nm (donor emission) upon excitation
at 450 nm (4 nm slit width for both excitation and emission
beams), was continuously recorded on a SLM-Aminco 8000
spectrofluorimeter (Urbana, IL). Measurements were per-
formed in a thermostated cell holder at 37 °C. The percent-
ages of RET were calculated according to the equation
(%RET) ¼ (1–F ⁄ F
0
) · 100 where F and F
0
are the fluores-
cence emission intensities at 530 nm upon addition of the
corresponding protein, and the value measured with vesicles
lacking the acceptor probe, respectively. Normalization of
the measurements was performed by independent experi-
ments in the presence of 1% (v ⁄ v) Triton X-100 [26,36,39].
Fluorescence depolarization measurements were per-
formed on a SLM-Aminco 8000 spectrofluorimeter equip-
ped with 10 mm Glan-Thompson polarizers (Urbana, IL).
Cells of 0.2-cm optical path were used. Slit widths were
4 nm both for excitation and emission beams. Labeling of
the vesicles with DPH (Aldrich, Milwaukee, WI) was per-
formed as previously described [40]. The degree of polariza-
tion of the fluorescence emission of DPH was measured at
425 nm upon excitation at 365 nm, after equilibration of
the sample at the required temperature. Independent experi-
ments demonstrated a negligible contribution of the protein
to the degree of polarization of the fluorescence probe.
Leakage of vesicle aqueous contents induced by the
RNases was measured by the 1,3,6-trisulphonate-8-amino-

naphtalene (ANTS) ⁄ p-xylenebispyridinium bromide (DPX)
system [26,38,41]. PG vesicles contained 12.5 mm ANTS,
45 mm DPX, 20 mm NaCl and 10 mm Tris buffer pH 7.5.
Unencapsulated material was separated from the
ANTS ⁄ DPX containing vesicles by gel filtration on a
Sephadex G-75 column (Sigma) equilibrated in 10 mm Tris
buffer pH 7.5 containing 0.1 m NaCl and 1 mm EDTA.
Fluorescence emission at 510 nm was measured upon exci-
tation at 386 nm on a SLM-Aminco 8000 spectrofluorime-
ter. Internal calibration of the assays was as follows: 0%
leakage corresponded to the fluorescence signal of the
vesicles before protein addition (F
0
), and 100% leakage
was the fluorescence intensity measured upon detergent
addition [1% (v ⁄ v) Triton X-100] (F
max
). Percentages of
leakage were calculated as (%L) ¼ (F–F
0
) ⁄ (F
max
–F
0
) · 100.
Protein fluorescence emission was measured on a
SLM-Aminco 8000 spectrofluorimeter at 25 °C with 0.2 cm
optical path cells. CD measurements were performed on
a Jasco J-715 spectropolarimeter (Tokyo, Japan) with
thermostated cylindrical cells.

Determination of EIE
The EIE for RNase–lipidic membrane model complexes
was carried out using delphi version 4 [42,43].
A model membrane in a crystal-like state was modeled
containing 192 palmityl-stearyl-phosphatidylglycerol mole-
cules per layer arranged in a parallelepiped with dimen-
sions 117(l) · 108(w) · 54(h) A
˚
. The phospholipids were
arranged in a compact hexagonal lattice with the fatty acid
tails and the terminal glycerol moiety completely extended.
Each layer was composed of seven rows of 15 molecules
separated by six rows of 16 molecules. Within each row the
distance between the P atoms of two adjacent phosphates
was 8.88 A
˚
, whereas the minimum distance between the P
atoms of adjacent rows was 8.98 A
˚
. The minimum distance
between two P atoms of phosphates of the two different
layers was 48.96 A
˚
.
For each RNase several RNase–lipidic membrane model
complexes were generated through an automated procedure
which included the following steps:
(a) the model of the membrane was oriented with each
layer parallel to the xy plane so that all the phosphorus
atoms of each layer had the same z coordinate and the

z axis was perpendicular to the center of the membrane
(Fig. 4);
(b) the major axis of the dimeric RNase was made coinci-
dent with the z axis of the system, hence orthogonal to the
membrane plane (Fig. 4);
(c) the dimeric RNase was rotated around the axis z (by
rotation angle r); values of EIE were recorded at 30° inter-
vals of r from 0 to 360°;
(d) the inclination of the dimer major axis with respect to
the z axis, hence to the membrane plane, was changed (by
inclination of angle i); EIE values were recorded at 30°
intervals of i from 0 to 180°;
(e) steps (c) and (d) were repeated until i reached the maxi-
mum value (180°).
It should be noted that after each rotation or inclination
the RNase dimer was translated: (a) to keep the mass cen-
tre of the protein (marked with a black dot in Fig. 4) onto
the axis z, and (b) to conserve the minimal distance
between protein and membrane at 3.5 A
˚
.
The EIE, i.e., the electrostatic contribution to the binding
energy (DG
bindingR ⁄ M
) of RNase (R) to a lipidic membrane
Electrostatic interactions and antitumor RNases E. Notomista et al.
3694 FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS
(M) model complexes was calculated by the method of the
grid energy differences [42] through the equation:
EIE ¼ DG

bindingR=M
¼ G
R=M
À G
R
À G
M
where G
R ⁄ M
,G
R
and G
M
are the grid energies of the com-
plex, of the RNase alone and of the membrane alone,
respectively.
The following delphi parameters were used for the calcu-
lations of the grid energy values: gsize ¼ 250; scale ¼ 1.2;
indi ¼ 4.0; exdi ¼ 80.0; prbrad ¼ 1.4; salt ¼ 0.1; ionrad ¼
2.0; bndcon ¼ 2; maxc ¼ 0.0001; linit ¼ 800. Standard for-
mal charges at pH 7.0 were attributed to protein atoms,
whereas a charge of ) 0.5 was attributed to the nonester
oxygen atoms of each phosphate group in the membrane
(total charge of each phosphate group ¼ ) 1).
RNase structures and modeling
Protein–lipidic membrane complexes of BS-RNase, RNase
A, RNase AA-N, RNase AA-C and HP-RNase were pre-
pared using the crystallographic structures of these proteins
(PDB codes 1BSR, 7RSA, 1A2W, 1F0V and 1Z7X, respect-
ively). As in the 1Z7X structure of HP-RNase Lys1 is lack-

ing, this residue was modeled using the 7RSA structure as
template. Protein–lipidic membrane complexes of RNase
AA, RNase AA-G, RNase AA-GG, and HHP-RNase were
prepared using models of these dimers. The models of
RNase A covalent dimers in the nonexchanging conforma-
tion were prepared using the 7RSA structure and the struc-
ture of nonexchanging BS-RNase (1R3M) as template.
Briefly, two molecules of RNase A were superimposed to
the subunits of BS-RNase using the fit tools of swiss-
pdbviewer ( hence residues
19, 28, 31, 32, and when necessary 38 and 111 of RNase A
molecules were mutated to the corresponding residues of
BS-RNase, forcing the mutated side-chains to adopt the
conformation present in BS-RNase. The models were opti-
mized for energy minimization using the gromos imple-
mentation of swiss-pdbviewer (50 cycles of steepest descent
followed by 50 cycles of conjugate gradients and 50 cycles
of steepest descent). No clashes were detected at the dimer
interface of the models or in the surroundings of the
mutated residues. The models of RNase A covalent dimers
in the exchanging conformation were prepared using the
7RSA structure and the structure of exchanging BS-RNase
(1BSR) as template. The server swissmodel [44–46] was
used to prepare the homology model of exchanging RNase
AA dimer based on the exchanging structure of BS-RNase.
The hinge loop, i.e., the loop which adopts different confor-
mations in the exchanging and nonexchanging BS-RNase
dimers, was cut from the optimized homology model and
ligated to the models of nonexchanging RNase A dimers as
described above. The gromos implementation of swiss-pdb-

viewer was used to optimize geometry of the hinge loop
after ligation. Also in this case no clashes were detected at
the dimer interface of the models or in the surroundings of
the hinge loop residues. Likewise, the models of nonex-
changing and exchanging HHP-RNase were prepared start-
ing from 1Z7X. The programs pymol (DeLano Scientific
LLC, San Francisco, CA), and swiss-pdbviewer were used
to inspect RNase structures and models.
Acknowledgements
The authors are indebted to Dr Renata Piccoli, for
critical reading the manuscript and helpful suggestions;
and to Dr Valeria Cafaro, Aurora Bracale, Antonella
Antignani and Sonia Di Gaetano for preparing some
of the RNase variants used in this study.
This work was supported by the Italian Association
for Cancer Research, the Ministry of University and
Research (Italy) and the Ministerio de Educacion y
Ciencia (Spain).
References
1 Beintema JJ, Breukelman HJ, Carsana A & Furia A
(1997) Evolution of Vertebrate Ribonucleases: Ribonu-
clease A Superfamily. In Ribonucleases: Structures and
Functions (D’Alessio G & Riordan JF, eds), pp. 245–
269. Academic Press, San Diego.
2 D’Alessio G (1993) New and cryptic biological messages
from ribonucleases. Trends Cell Biol 3, 106–109.
3 Riordan JF (1997) Structure and Function of Angioge-
nin. In Ribonucleases: Structures and Functions (D’Ales-
sio G & Riordan JF, eds), pp. 446–466. Academic
Press, San Diego.

4 Dostal J & Matousek J (1973) Isolation and some che-
mical properties of aspermatogenic substance from bull
seminal vesicle fluid, J Reprod Fertil 33, 263–274.
5 Youle RJ & D’Alessio G (1997) Antitumor RNases. In
Ribonucleases: Structures and Function (D’Alessio G &
Riordan JF, eds), pp. 491–514. Academic Press, San
Diego.
6 Haigis MC, Kurten EL & Raines RT (2003) Ribonu-
clease inhibitor as an intracellular sentry. Nucleic Acids
Res 31, 1024–1032.
7 Hofsteenge J (1997) Ribonuclease inhibitor. In Ribonuc-
leases: Structures and Functions (D’Alessio G & Riordan
JF, eds), pp. 621–658. Academic Press, San Diego.
8 Dickson KA, Haigis MC & Raines RT (2005) Ribonu-
clease inhibitor: structure and function. Prog Nucleic
Acid Res Mol Biol 80, 349–374.
9 Ardelt W, Mikulski SM & Shogen K (1991) Amino acid
sequence of an anti-tumor protein from Rana pipiens
oocytes and early embryos. Homology to pancreatic
ribonucleases. J Biol Chem 266, 245–251.
10 D’Alessio G, Di Donato A, Mazzarella L & Piccoli R
(1997) Seminal Ribonuclease: The Importance of
E. Notomista et al. Electrostatic interactions and antitumor RNases
FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS 3695
Diversity. In Ribonucleases: Structures and Functions
(D’Alessio G & Riordan JF, eds), pp. 383–423.
Academic Press, San Diego.
11 Piccoli R, Di Gaetano S, De Lorenzo C, Grauso M,
Monaco C, Spalletti-Cernia D, Laccetti P, Cinatl J,
Matousek J & D’Alessio G (1999) A dimeric mutant of

human pancreatic ribonuclease with selective cytotoxi-
city toward malignant cells. Proc Natl Acad Sci USA
96, 7768–7773.
12 Leland PA, Schultz LW, Kim BM & Raines RT (1998)
Ribonuclease A variants with potent cytotoxic activity.
Proc Natl Acad Sci USA 95, 10407–10412.
13 Suzuki M, Saxena SK, Boix E, Prill RJ, Vasandani
VM, Ladner JE, Sung C & Youle RJ (1999) Engineer-
ing receptor-mediated cytotoxicity into human ribonu-
cleases by steric blockade of inhibitor interaction. Nat
Biotechnol 17, 265–270.
14 Antignani A, Naddeo M, Cubellis MV, Russo A &
D’Alessio G (2001) Antitumor action of seminal ribo-
nuclease, its dimeric structure, and its resistance to the
cytosolic ribonuclease inhibitor. Biochemistry 40, 3492–
3496.
15 Monti DM & D’Alessio G (2004) Cytosolic RNase
inhibitor only affects RNases with intrinsic cytotoxicity.
J Biol Chem 279, 39195–39198.
16 Naddeo M, Vitagliano L, Russo A, Gotte G, D’Alessio
G & Sorrentino S (2005) Interactions of the cytotoxic
RNase A dimers with the cytosolic ribonuclease inhibi-
tor. FEBS Lett 579, 2663–2668.
17 Matousek J, Gotte G, Pouckova P, Soucek J, Slavik T,
Vottariello F & Libonati M (2003) Antitumor activity
and other biological actions of oligomers of ribonu-
clease A. J Biol Chem 278, 23817–23822.
18 Mastronicola MR, Piccoli R & D’Alessio G (1995)
Key extracellular and intracellular steps in the antitu-
mor action of seminal ribonuclease. Eur J Biochem

230, 242–249.
19 Lin JJ, Newton DL, Mikulski SM, Kung HF, Youle RJ
& Rybak SM (1994) Characterization of the mechanism
of cellular and cell free protein synthesis inhibition by
an anti-tumor ribonuclease. Biochem Biophys Res
Commun 204, 156–162.
20 Bracale A, Spalletti-Cernia D, Mastronicola M, Cast-
aldi F, Mannucci R, Nitsch L & D’Alessio G (2002)
Essential stations in the intracellular pathways of
cytotoxic bovine seminal ribonuclease. Biochem J 362,
553–560.
21 Di Donato A, Cafaro V & D’Alessio G (1994) Ribo-
nuclease A can be transformed into a dimeric ribonu-
clease with antitumor activity. J Biol Chem 269,
17394–17396.
22 Cafaro V, Bracale A, Di Maro A, Sorrentino S,
D’Alessio G & Di Donato A (1998) New Muteins of
RNase A with enhanced antitumor action. FEBS Lett
437, 149–152.
23 Boggs JM (1980) Intermolecular hydrogen bonding
between lipids: influence on organization and function
of lipids in membranes. Can J Biochem 58, 755–770.
24 Boggs JM (1986) Effect of lipid structural modifica-
tions on their intermolecular hydrogen bonding inter-
actions and membrane functions. Biochem Cell Biol
64, 50–57.
25 Boggs JM (1987) Lipid intermolecular hydrogen bond-
ing: influence on structural organization and membrane
function. Biochim Biophys Acta 906, 353–404.
26 Mancheno JM, Gasset M, Onaderra M, Gavilanes JG

& D’Alessio G (1994) Bovine seminal ribonuclease
destabilizes negatively charged membranes. Biochem
Biophys Res Commun 199, 119–124.
27 Matousek J (1973) The Effect of Bovine Seminal
Ribonuclease (AS RNase) on Cells of Crocker Tumour
ion Mice. Experientia 29, 858–859.
28 Bergelson LD, Dyatlovitskaya EV, Torkhovskaya TI,
Sorokina IB & Gorkova NP (1970) Phospholipid
composition of membranes in the tumor cell. Biochim
Biophys Acta 210, 287–298.
29 Murray D & Honig B (2002) Electrostatic control of
the membrane targeting of C2 domains. Mol Cell 9,
145–154.
30 Futami J, Maeda T, Kitazoe M, Nukui E, Tada H,
Seno M, Kosaka M & Yamada H (2001) Preparation of
potent cytotoxic ribonucleases by cationization:
enhanced cellular uptake and decreased interaction with
ribonuclease inhibitor by chemical modification of
carboxyl groups. Biochemistry 40, 7518–7524.
31 Ilinskaya ON, Dreyer F, Mitkevich VA, Shaw KL, Pace
CN & Makarov AA (2002) Changing the net charge
from negative to positive makes ribonuclease Sa
cytotoxic. Protein Sci 11, 2522–2525.
32 Snyder MR & Gleich GJ (1997) Eosinophil-Associated
Ribonucleases. In Ribonucleases: Structures and Func-
tions (D’Alessio G & Riordan JF, eds), pp. 426–440.
Academic Press, San Diego.
33 Liu Y, Hart PJ, Schlunegger MP & Eisenberg D (1998)
The crystal structure of a 3D domain-swapped dimer of
RNase A at 2.1 A

˚
resolution. Proc Natl Acad Sci USA
95, 3437–3442.
34 Liu Y, Gotte G, Libonati M & Eisenberg D (2001) A
domain-swapped RNase A dimer with implications for
amyloid formation. Nat Struct Biol 8, 211–214.
35 Sorrentino S, Barone R, Bucci E, Gotte G, Russo N,
Libonati M & D’Alessio G (2000) The two dimeric
forms of RNase A. FEBS Lett 466, 35–39.
36 Mancheno JM, Gasset M, Lacadena J, Ramon F,
Martinez del Pozo A, Onaderra M & Gavilanes JG
(1994) Kinetic study of the aggregation and lipid mixing
produced by alpha-sarcin on phosphatidylglycerol and
phosphatidylserine vesicles: stopped-flow light scattering
and fluorescence energy transfer measurements. Biophys
J 67, 1117–1125.
Electrostatic interactions and antitumor RNases E. Notomista et al.
3696 FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS
37 Mancheno JM, Gasset M, Albar JP, Lacadena J,
Martinez del Pozo A, Onaderra M & Gavilanes JG
(1995) Membrane interaction of a beta-structure-form-
ing synthetic peptide comprising the 116–139th sequence
region of the cytotoxic protein alpha-sarcin. Biophys J
68, 2387–2395.
38 de los Rios V, Mancheno JM, Lanio ME, Onaderra M
& Gavilanes JG (1998) Mechanism of the leakage
induced on lipid model membranes by the hemolytic
protein sticholysin II from the sea anemone Stichodac-
tyla helianthus. Eur J Biochem 252, 284–289.
39 Struck DK, Hoekstra D & Pagano RE (1981) Use of

resonance energy transfer to monitor membrane fusion.
Biochemistry 20, 4093–4099.
40 Gasset M, Martinez del Pozo A, Onaderra M &
Gavilanes JG (1989) Study of the interaction between
the antitumour protein alpha-sarcin and phospholipid
vesicles. Biochem J 258, 569–575.
41 Ellens H, Bentz J & Szoka FC (1985) H
+
- and Ca
2+
-
induced fusion and destabilization of liposomes.
Biochemistry 24, 3099–3106.
42 Rocchia W, Alexov E & Honig B (2001) Extending the
Applicability of the Nonlinear Poisson-Boltzmann
Equation: Multiple Dielectric Constants and Multiva-
lent Ions. J Phys Chem B 105, 6507–6514.
43 Rocchia W, Sridharan S, Nicholls A, Alexov E,
Chiabrera A & Honig B (2002) Rapid grid-based con-
struction of the molecular surface and the use of
induced surface charge to calculate reaction field ener-
gies: applications to the molecular systems and geo-
metric objects. J Comput Chem 23, 128–137.
44 Schwede T, Kopp J, Guex N & Peitsch MC (2003)
SWISS-MODEL: An automated protein homology-
modeling server. Nucleic Acids Res 31, 3381–3385.
45 Peitsch MC (1996) Promod and Swiss-Model: Internet-
based tools for automated comparative protein model-
ling. Biochem Soc Trans 24, 274–279.
46 Guex N & Peitsch MC (1997) Swiss-Model and the

Swiss-PDB-viewer: an environment for comparative
protein modelling. Electrophoresis 18, 2714–2723.
Supplementary material
The following supplementary material is available
online:
Fig. S1. Structure of the BS-RNase–model membrane
complex with the highest negative EIE value. Lipids
are shown as van der Waals spheres and colored
according to the atom type. Solvent accessible surface
is shown for BS-RNase. The two subunits are colored
blue and red. In (A) the solvent accessible surface is
transparent to show secondary structure elements.
Fig. S2. Electrostatic field of BS-RNase at pH 7.0.
Blue meshes correspond to potential values ¼ +2
kTÆe
)1
and red meshes to potential values ¼ )2kTÆe
)1
.
Secondary structure elements are shown in green.
Fig. S3. Electrostatic field at pH 7.0 of monomeric and
dimeric RNases. The proteins have the same orienta-
tion with the ‘basic surface’ toward the top of the fig-
ure. Cyan meshes correspond to potential values ¼
+2 kTÆe
)1
and red meshes to potential values ¼ )2
kTÆe
)1
. The solvent accessible surfaces of the RNases

are shown in light gray.
Fig. S4. Electrostatic potential at pH 7.0 mapped to
the accessible surface of monomeric and dimeric
RNases. The proteins have the same orientation with
the ‘basic surface’ toward the observer. Blue color
becomes saturated at +2 kTÆe
)1
and the red color at
)2kTÆe
)1
. In RNase-AA the ‘basic surface’ is much
less basic owing to the four carboxylates of residues
Asp38 and Glu111. In RNase-AA-GG the ‘basic sur-
face’ is more similar to that of BS-RNase because of
the removal of the four carboxylates. The ‘basic sur-
face’ of the mutant HHP-RNase is very similar to that
of RNase-AA-GG as the presence of Glu111 is coun-
terbalanced by basic residues not present on RNase-
AA-GG.
This material is available as part of the online article
from
E. Notomista et al. Electrostatic interactions and antitumor RNases
FEBS Journal 273 (2006) 3687–3697 ª 2006 The Authors Journal compilation ª 2006 FEBS 3697