myosin subfragment 1. PARAQUAT (l,1'-DIMETHYL-4,4'-BIPYRIDYLIUM ... posure, occur in the lung (4, 6-9). The lung's ability to ...... by 4-ipomeanol, a product.
Paraquat
induces
actin
assembly
in depolymerizmg
conditions MILZANI,
ALDO Department Milan,
ISABELLA
of Biology,
University
DALLEDONNE, of Milan,
Lab
GIOVANNI of Biochemistry
and
Biophysics
AND
of
ROBERTO
Cytoskeletal
COLOMBO’
Structures,
20133
Italy
The molecular mechanism (or mechanisms) at the basis of paraquat (PQ) (a widely used herbicide) toxicity is far from being fully understood. Until now, two main points of view have emerged: 1) PQ-related cell injuries could be mediated by toxic oxygen free radicals coming from the metabolism of the herbicide by the microsomal enzyme system, and/or 2) PQ, by inducing initochondrial swelling and breakage, could cause troubles in cell energy charge, then driving the cell to death. Recently, some of cytoskeletal structures (microtubules and microfilaments) have been proposed as further PQ cell targets. The microfilament system in particular seems to be markedly affected by the herbicide, but so far no direct evidence associates PQ to actin damage. In this study, experimental data are presented concerning the direct effect of PQ on actin dynamics in solution. We demonstrate that actin selectively binds PQ; moreover, PQ induces the formation of actin sopramolecular structures in depolymerizing medium (G-buffer). Furthermore, by the interactions with F-actin crosslinking proteins (a-actinin and filamin), FITC-phalloidin, and myosin subfragment 1 (Si), it is demonstrated that PQ-induced actin aggregates are undoubtedly built up by F-actin. Electron micrographs showed that PQ-induced actin polymers are very short and tend to aggregate one to another. This mutual cohesion leads to the steric blockage of polymer growing ends as suggested by nucleated actin polymerization assays. Sonication, by releasing F-actin fragments from short polymer aggregates, allows actin polymer ends to regain their growing ability.-Milzani, A., DalleDonne, I., Vailati, G., Colombo, R. Paraquat induces actin assembly in depolymerizing conditions. FASEB J. 11, 261-270 (1997) ABSTRACT
Key words: actin polymerization F-actin actin cross-linking pro!ein.s FITC-phalloidin myosin subfragment 1 .
.
PARAQUAT (l,1’-DIMETHYL-4,4’-BIPYRIDYLIUM dichloride; PQ)2 is a widely used and extremely effective herbicide (1) with a broad spectrum of activity. PQ is quite toxic and its toxicity on animals and humans has been exhaustively
0892-6638/97/0011
VAILATI,
-0261 /$O1 .50 © FASEB
described (2). Many cases of human death have been attributed to PQ poisoning (3-5), whose main characteristic lies in the specificity of its anatomical target. In fact, PQrelated primary lesions, from both acute and chronic exposure, occur in the lung (4, 6-9). The lung’s ability to accumulate this herbicide (10-13) probably causes the selectivity of the PQ-related toxicity. The molecular mechanism (or mechanisms) responsible for the toxicity of this chemical is far from being fully understood. Several different possible pathways illustrating PQ biological activity have been outlined hut no general consensus exists to date. One possibility is that the PQ biological activity could be mediated by toxic oxygen free radicals (14-16), causing lipid peroxidation, which leads to cell death. Oxygen radicals are formed during PQ metabolism carried on by the microsomal enzyme system. The endoplasmic reticulum, in fact, was predicted to be structurally damaged by PQ (17) as well as by other microsome-metabolized chemical compounds (18-22). In contrast to the above-mentioned microsome theory, Wang et al. (17) suggested mitochondria as the initial toxic site of PQ. Mitochondrial swelling and the selective breackage of these organelles supported the idea of a mitochondriamediated cell injury by PQ. The interference of PQ with the oxidative activityof eukatyotic cells is well established, since treatment with PQ is often used to experimentally induce cell oxidative stress (23, 24). A third possibility has been recently advanced. Because cell shape changes associated with PQ cytotoxicity have been observed (both in vitro and in vivo experiments) (2527), cytoskeleton has been considered as a further probable intracellular target of PQ injury. Cell shape, in fact, is known to reflect the inside spatial organization of cytoskeletal elements (28, 29). Immunocytochemical data (30-32) revealed deep modifications in the cytoskeletal Correspondence:
Department
of Biology,
University
of Milan,
Lab
of Biochemistry and Biophysics of Cytoskeletal Structures. Via Celoria 26-20133 Milan, Italy. Abbreviations: G-actin. (globular) monomenc aetin; F-actin, (filamentous) polymeric actin; PQ. paraquat, 1.1 ‘-dimethyl-4.4’-hipyridyhum dichloride; BSA, bovine serum albumin; MTP, microtubule proteins (tubulins + coseditnetit-associated proteins); SDS-PAGE. sodium dodecyl sulfate-polyacrylamide gel electrophoresis; EDC, i-ethyl3-[3-(dimethyl-amino)propylj
phenylnxethylsulfonyl fluoride; cyanate-laheled phalloidin.
DTT, dithiothreitol; FITC-phalloidin, fluorescein
carbodiimide;
PMSF, isothio-
261
organization of PQ-treated cultured cells. Both microtubules (MT) and microfilaments (MF) seem to be sensitive to PQ. Although MT injury by PQ is debated (30, 31), the PQ disarranging effect on cellular MF scaffoldings appears to be severe and irreversible (32). In particular, low doses of PQ resulted in the F-actin redistribution in the pennuclear area whereas higher treatments led to the marked loss of actin bundles. Because the direct interaction of PQ with cytoskeletalelements is not yet demonstrated, PQ activity on MT and MF has usually been considered as the reflection of PQ primary cell injuries on different cellular targets. Oxygen free radicals, for instance, markedly act on actin both in cultured cells (33, 34) and in solution (35) (R. Colombo, unpublished results), so PQ cytoskeletal injury could represent a side effect of the primary PQ influence on the microsomal drug-metabolizing enzyme cell system. Moreover, changes in the mitochondrial activity (eventually caused by PQ treatments), leading to the lowering of cell energy charge, could affect the spatial organization of cytoskeletal actin structures (36). To avoid complications associated with herbicide metabolism, we testedthe directeffectofPQ on actindynamics in solution. The results obtained are quite surprising.
A pyrene/actin molar ratio of 0.8-0.9 was usually obtained. Because high PQ concentrations interfere with pyrene-actin fluorescence emission. we used fluorescent actin only to perfonn “seeded” actin polymerization tests. 1n fact, the latter provide for a marked dilution (1:10 v/v) of “actin polymerization seed” aliquots with pyrene-laheled C-actiti. This consequently leads to dilution of PQ because the herbicide was present only in actin seed solutiotis. After dilution, the influence of PQ on pyrene fluorescence emission was negligible. Other
proteins
Stnooth muscle a-actinin and filamimi from chicken gizzards were prepared according to Feramisco and Burridge (44) and then stored at -20#{176}C in 20 mM Tris-HCI, pH 7.6, 20 mM NaCI. 0.1 mM EDTA, 1.5 mM -mercaptoethanol after the addition of 20% (v/v) glycerol. Before use, both protein preparations were dialyzed for 48 h against C-buffer; a-actinin was centrifuged for 60 mm at 100,000 X g. a-Actinin (subunit 100 kDa) (45) and filamin (dimer 500 kDa) (46) concentrations were determined by their 0.D. at different wavelength (278 nm for aactinin; 280 nm for filarnin). using E273, = 0.97 nlg’ ml cm’ (45) and E2, = 1.351 mg’ nil cm’ (47), respectively. Bovine serum albumin (BSA) was purchased from Sigma. Microtubule protein (MTP) was a generous gift of Dr. Khalid Islam (Lepetit Research Center, Merrill Dow Research Inst.). Fluorometric
assay to test actin
Seeded actin polymerizations hancement of trace quantities
MATERIALS
AND
Paraquat (methyl viologen 1.1 ‘-dimethyl-4.4’-bipyridvlium dichloride) was purchased from Sigma (St. Louis, Mo.). Paraquat stock solution (20 mM) was prepared weekly by dissolving the herbicide in distilled water. PQ concentrations we considered ranged around the highest ones Used to treat cultured cells (30). All other reagents were of analytical grade.
Light
scattering
Actin
polymerization
of actin
Light
assay for
actin
was followed,
polytnerization at 25#{176}C. by checking
the increase
scattering
assay for actin
filament
intensities
bundles
was prepared
Fluorescent
labeling
IG-atinl
= (A.5,
-
A,1/2.2
[pyrenel
Vol.
11
March
0.127
1997
The formation of actin filament bundles was measured at 400 nm under a 90#{176} observation angle at 25#{176}C (49). The light scattering intensity was expressed in arbitrary units. High-shear
The
mM’
X A)/26.6 X
10’
M’
cm’
viscometry
increase
monitored, ml; buffer
in specific
viscosity
of polytsierizing
at 25 #{176}C, using Ostwald viscometers flow time: about 30 s) consideritig: il,,
=
(t,/t,,)
-
actin
samples
was
(sample
volume:
0.5
1
where t. is the samnple flow time and t,, is the buffer viscosity was expressed in centiStokes: 1 St = 10 Low-shear
of actin
Pyrene-actin was prepared as described by Tellam and Frieden (41). Pyrene-actin concentration and the pyrene/actin molar ratio were calculated from the absorbamwe at 290 and 344 run (42, 43), considering:
262
by the fluorescence en(10% of total actin) at
in the 90#{176} light scattering at 546 nm (48). Light scattering were expressed in arbitrary units (a.u.).
from rabbit skeletal muscles according to Spudich and Watt (37) and further purified by gel filtration on Sephadex C-150. Actin purity was greater than 99%. as checked by sodium dodecyl sulfate-polyacmylamide gel electrophoresis (SDS-PAGE) (38). When used within a few days. G-aetin solution was kept on ice in 2 mM TrisHCI. pH 7.5, 0.2 mM AlP. 0.5 mM Dli’, 0.2 niM CaCI2. 1.5 mM NaN3 (C-buffer). If stored, it was lyophilized after the addition of 2 mg sucrose/mng actin (39). The extinction coefficient of 0.617 mg tnl’ cm’, at 29() nm of wavelength (40). has been used to calculate C-aetin (43 kDa) concentration. Before use. small actin aggregates were carefully removed from monomeric actin solutions. by centnfugation at 100,000 X g. for 1 h. To prepare actin samples. protein solutions and buffers were filtered with 0.20 llm disposable filters and degassed. In the experiments presented here, actin concentration was 0.5 mg/mI (12 j.tM) unless stated otherwise. Actin
were followed of pyrene-actin
25#{176}C. Fluorescence measurements have been carried out with a Kontron SFM 25 spectrofluoromneter. Excitation and emission wavelengths were 365 and 407 tim, respectively. The spectrofluorotneter was equipped with a neutral density filter (50%) to avoid pvrene-actin photobleaching and with a cutoff filter (390 nm) to minimize light scattering.
METHODS
Materials
Preparation
assembly
Apparent
flow
time.
Specific
tn2s’.
viscotnetry viscosity
measurements
were
carried
out
with
a homemade
falling ball apparatus, as described by Pollard amid Cooper (50). Glass capillary tubes (100 .tl. micropipettes) with an internal diameter of 1.3 mm and stainless steel balls (material 440 C), diameter of 0.64 mm and density of 7.2 g cmn”. were used. For tneasurenients, the micropipettes were held at 15#{176} to the horizontal in a thermostatted water bath at 25#{176}C. Samples were drawn up in the micropipettes immediately after salt
The FASEB lournal
MILZANI
ET AL.
addition, so they were subjected to shear only at the beginning of polymerization. The apparent viscosity of samples is given in seconds. Electron
mnicroscopy
Protein samples were dispensed in 5 pi ahiquots on collodion filmcoated grids by truncated pipette tips to avoid filament fragmentation. After 60 s, grids were dried from the side with filter paper and negatively stained with the addition of 5 j.tI of 1% uranyl acetate. After 10 s, excess stain was removed with filter paper. Staining procedure was repeated for five times. Specimens were air dried and observed at 80 kV with a JEOL 100-SX electron microscope. Ultracentrifugation
Subtilisin Carlsberg (from Bacillus lichenformts, Sigma) digestion was carried out at an enzyme/protein mass ratio of 1:1500 for 5 mm. Digestion was stopped with 2 mM phenylmethylsulfonyl fluoride (PMSF). The samples were mixed with 3X SDS-PAGE and immediately incubated for 5 mm at 90#{176}C. cz-Chymotrypsin (from bovine pancreas, tosyllysine-chloromethanetreated, Sigma) digestion was carried out at an enzyme/protein mass ratio of 1:50 for 60 mm. Digestion was stopped with 2 mM phenylmethylsulfonyl fluoride (PMSF). The samples were mixed with 3X SDSPAGE and immediately incubated for 5 mm at 90#{176}C. After heating, samples were analyzed by SDS-PAGE (38) on 15% (w/v) polyacrylamide gels. Gels were stained with Coomassie brilliant blue G-250. Only trypsin-related digestion patterns are shown (see Fig. 3).
tests FITC-phalloidin
In high-speed centrifugation tests, protein mixtures (1.6 ml) were centrifuged for 15 mm at 350,000 X g at 20#{176}C. Pellets were resuspended in 800 tl sample buffer and subjected to SDS-PAGE (7.5% gel) (38). Gels were stained with Coomassie brilliant blue R-250. Monitoring
of
PQ concentration
in aqueous
solutions
PQ in aqueous solutions was revealed by measuring the O.D. at 600 nm after the reduction of the herbicide with sodium dithionite in the presence of alkali (51). When PQ is reduced in the above-mentioned conditions, a stable blue-violet radical cation forms. This cation strongly absorbs at 396 nm but also exhibits a less intense broad absorption in the visible region (600 nm). Considering our experimental conditions, we preferred to record O.D. values at 600 nm. Protein/PQ
binding
intrinsic
fluorescence
(tryptophan
fluorescence)
Fluorescence data were obtained with a Kontron SFM 25 spectrofluorometer using 10 X 10 mm quartz cuvettes. Tryptophan fluorescence, in both native and PQ-treated actin samples, was excited at 300 nm, and intensity changes in the emission fluorescence were measured at all wavelengths ranging between 300 and 400 nm. Light scattering was observed at 300 nm (53). Spectra of PQ-treated actin samples were obtained taking into account the filtering effect of PQ on actin fluorescence intensity.
Peptide
All enzymatic digestions were carried out at the actin concentration of 24 p.M. at 25 #{176}C, in C-buffer. All the enzymes were solubilized just before use and kept on ice before the digestion. To evaluate possible undesired effects of paraquat on the activity of the proteolytic enzymes, each enzyme activity was assayed digesting 1 mg/ml BSA in the presence or absence 1 mM paraquat. No differences were observed in BSA digestion products either in control or paraquat-treated samples (not shown). Trypsin (from bovine pancreas, tosylphenylalanine-chloromethanetreated, Sigma) digestion was carried out at an enzyme/protein mass ratio of 1:25 for 15 mm. Digestion was stopped with 3X SDS-PAGE sample buffer; samples were immediately incubated for 5 mm at 90#{176}C.
AND
It is well known that phalloidin, from the poisonous mushroom Amanita phalloides, selectively interacts with F-actin (54). Considering this specific interaction, we set up a test in which actin samples (12 l.tM) in Cbuffer and incubated with different concentrations of PQ for 24 h were added to FITC-phalloidin solutions (1.2 l.tM, in C-buffer), whose previously determined fluorescence intensities (i.e., FITC-phalloidin concentration) were kept strictly similar (we did not use TRITC-phalloidin, since PQ influences its fluorescence emission). After mixing, actin/ phalloidin mixtures were immediately centrifuged (350,000 X g for 15 mm) and then fluorescent phalloidin concentrations were further tested in supernatants. As phalloidin markedly increases actin assembly in the presence of salts (55), this test is appropriate only in depolymerizing conditions and after very short times of phalloidin/actin interaction.
FITC-phalloidin
fluorescence
anisotropy
The time scale of fluorescence emission of excited FITC-phalloidin molecules is comparable to that of their rotational diffusion in solution. The polarization or anisotropy of the emission provides a measure of this process (56). When FITC-phalloidin molecules bind ordered, filament-like structures and are excited with vertically polarized light, the fluorophores are not able to rotate during the lifetime of the excited state. The emission is then polarized, usually in the vertical direction. The extent of polarization is most conveniently defined by the anisotropy (r):
r
-
I,
=
I + 2I.
where I refers to the intensities and subscripts indicate the parallel (II) or perpendicular (J..) component. A number of processes can result in the increase of anisotropy, the most common being the loss of rotational diffusion.
Actin/Si covalent cross-linkingby EDC
mapping
PARAQUAT
test
assay
The ability of proteins to bind PQ has been tested by gel filtration. Briefly, proteins (actin, a-actinin, filamin, microtubule proteins, and BSA). 1 mg/mI in suitable buffers, were incubated overnight at 25#{176}C with PQ (1 mM). Experimental samples (5 ml) were then gel-filtered on Sephadex G-25 columns (1.6 X 45cm) to separate unbound PQ. Eluted fractions (1.5 ml) were collected amid analyzed. The 0.D.2a;, O.D.595 [after Coomassie brilliant blue staining (52) of proteins] and 0.D.3,,,0 [after alkaline reduction (51) of the herbicide] elution profiles were recorded.
Actin
binding
ACTIN
The “rigor” complex occurring between actin and the myosin subfragment 1 (Si) could be stabilized by the cross-linking of its constituents with a zero-length cross-linker, 1-ethyl-3-[3-(dimethylamino) propyljcarbodiimide (EDC) (57). Actin sample (0.5 mg/mI in iO mM imidazole) assembly was induced by salt addition (native F-actin) or by overnight PQ treatment with different concentrations of the herbicide (PQ-mnduced formation of actin aggregates). Cross-linking has been carried on by the addition of 5 mM EDC to actinlSl mixtures (1:1, w/w). All samples contained 40 mM NaC1 in order to maintain the myosin head in solution. After 30 mm incubation at 25#{176}C, the cross-linking reaction was quenched by the addition of excess 3-mercaptoethanol. Cross-linked products were then analyzed by SDS-PAGE (38).
263
g
6
11 0.8
(0.6
U’ 0
I
I 1.2
1.2
11 g
0.8
0.8
t
(0.6 0I
g
c 0.4
I
0.2
jj
S
10
20
30
40
Recon
60
50
70
0
I
0.2 0
0 0
U’
0.6 0.4
0
SO
10
20
30
number
40
50
80
70
50
number
..-
1.2
11
g
0.8 0.6
U’ 0
0 jj
0.4
I
0.2 0 0
Figure (-) F;ht
1.
Coomassie
filamin
PQ/Protein profile brilliant
binding.
The
himiding
at 280 tim of wavelength blue
(52);
(C), MTP (D). and BSA
elution (E).
(0)
ability
10
20
30 40 50 Frctlon number
of PQ to different
binding
The binding of PQ to different proteins has been checked by gel filtration on Sephadex G-25 columns. Some cytoskeletal proteins (G-actin, microtubule proteins, a-actinin and filamin) and BSA (i.e., an extracytoskeletal protein) were incubated overnight with PQ at 25#{176}C. Different mixtures were then gel-filtered. Because of their different molecular sizes, proteins and unbound PQ are eluted from the chromatographic’ gel with different speeds. In eluted fractions, the presence of both proteins and PQ has been revealed as previously reported (see Material and Methods). Figure 1 clearly shows that actin selectively binds
264
Vol.
11
March
1997
proteins
70
80
has been
checked
by gel
filtration
on Sephadex
C-25
culumnns.
(both PQ and proteins absorb); (#{149}) elution profile at 595 urn of wavelength after protein staining with profile at 600 nm of wavelength after PQ reduction with sodium dithionite (51). Actin (A), U-actimimn (B),
RESULTS ProteinlPQ
80
The
PQ (Fig. 1A), in contrast to other tested polypeptides (Fig. 1B-D). In particular, MTP do not bind PQ (Fig. 1D), suggesting that, in PQ-treated cells, microtubule alterations (30) could represent the side effect of other, undetermined PQ-related cell damage. Effects
of PQ on actin
intrinsic
fluorescence
Ttyptophan fluorescence emission spectra of actin samples treated overnight with different PQ concentrations are shown in Fig. 2. Increasing PQ concentration gradually lowers fluorescence intensity and, in parallel, enhances light scattering but does not cause spectral shifts. These fluorescence patterns seem to mimic the intrinsic fluores-
FASEB Journal
MILZANI
ET AL.
sium-induced actin polymerization could mask a possible polymerizing activity of PQ. To investigate this, actin samples in depolymerizing conditions (G-buffer) were incubated for 24 h with different PQ concentrations; then light scattering intensity specific viscosity (Fig. 4B), and sedimenting actin (Fig. 4A) were measured. All tests (including electron microspy, Fig. 5) revealed that PQ tends to aggregate actin subunits in a dose-dependent fashion, even though this is slower and less pronounced than salt-induced actin polymerization. The PQ-induced increase in viscosity of actin samples, for instance, is about one-quarter of that usually reached by routinely assembled actin solutions.
120
100
C
80
C 60 C
40 0
20
U-
0 300
340 Emission
Figure
2. Actimi tryptophan
fluorescence
380 (nm) emission
spectra.
Emnission
spetra of aetin intrinsic fluorescence as a function of PQ treatments. Native C-actin (#{149}). native F-actin polymerized in 2 mnM MgCI2 + 100 mM KCI(#{149}). actin treated overnight with 0.1 mM PQ (). actin treated overnight with 0.5 mnM PQ (0). actin treated overnight with 1.0 mM PQ (A), and actin treated overnight with 2 mM PQ (0). Emission spectra of 0.1 mM PQ-treated actin and native G-actin are largely superitnposed as well as spectra related to 1 and 2 mM PQ-treated actins and native F-actin. Spectra were determine(l as described in Materials and Methnds.
cence emission actin (53). Peptide
changes
mapping
after the assembly
of PQ-treated
of monomeric
viscometry, actin samples
and pelleting assays in depolyinerizing
Time course experiments (not shown) revealed that PQ does not significantly influence salt-induced actin assembly. As the data presented to date suggest that PQ is able to provoke actin structural modification from the monomeric, globular spatial configuration to F-actin type, we suspected that massive F-actin formation during magne-
PARAQUAT
AND
ACTIN
of PQ-induced actin aggregates with F-actin cross-linking proteins
If PQAAs are really F-actin. they should bind both aactininand filamin.a-Actinin and filamin were added to actin samples and exposed to different concentrations of PQ for 24 h. Sedimentable materials were subsequently analyzed by SDS-PAGE. Electrophoretic patterns (Fig. 6) show that PQAAs are able to bind 1)0th F-actin crosslinking proteins. Note that PQ (2 mM) does not increase the usual poor sedimentation of either a-actinin [Fig. 6A. (a,)I or filamin [Fig. 6B, (f,)]. These findings are supported
____
actin
The results from both binding assays and fluorescence emission spectra of PQ-treated actin samples suggest that actin selectively binds the herbicide, which causes actin structural changes resembling those occurring in G-actin during assembly. To further investigate this aspect, we performed the enzymatic digestion of actin samples, tleate(l with different concentrations of PQ, by several proteases. Digestion patterns were then analyzed by SDSPAGE. As shown in Fig. 3. PQ-treated actin is, in general, markedly less sensitive than native actin to protease activity. Actin resistence to enzymatic digestion increases with PQ concentration. As previously appreciated in both binding and intrinsic fluorescence assays, PQ-treated actin behavior to enzymatic digestion resembles that of polymeric actin. Perfect superimposition of electrophoretic patterns of digested BSA, in the presence or in the absence of PQ, indicates that the proteolytic activity of all proteases used is not influenced by PQ (not shown). Light scattering, on PQ-treated conditions
Interaction (PQAAs)
Actin
_A
-
U G12
1 O.5O.1 [PQ] (mM)
B
Actin-
UG12
1 O.50.11F [PQ] (mM)
Figure 3. Peptide mapping. SDS-PAGE of lrvpsin-digesled actins as a function of PQ treatments. A) Tryplic digestion of a(’tin samples treated with different concentrations of PQ for 10 mimi. Frotn left to right: undigested native G-actin (U). native G-actin (C); further limies have been indicated with the PQ concentratiomi (mM) used to treat the electrophorized actin sample. B) Tryptic digestion of aclin samples treated with different concentrations of PQ for 24 Ii. From left to right: undigested native C-actin (U). native G-actin (U). and native F-actin (F). Other lines, related to PQ-treated actins, ‘are as in panel 4. F’or further details, see Materials and Methods.
265
A
Actin
of PQ. We did not obtain
PQAA myosin
with
interaction
the subfragment
indications
1 (Si)
of
The main, “physiological” characteristic of F-actin lies in its ability to interact with myosin globular heads. According to Sutoh (57), two different segments of Si heavy chain are able to interact with actin and, consequently, two electrophoretically distinguishable complexes (165 and 175 kDa) could form. We incubated PQAAs with the subfragment 1 of myosin, and the molecular complexes that formed were covalently cross-linked with EDC (for details,
-
G12
actin in the presence supporting this.
1 O.50.11F [PQ] (mM)
see Materials and Methods). F-actinlSl comple forma
U
was checked by SDS-PAGE. As shown in Fig. 9, PQAAs interact with myosin heads, forming both 165 and 175 kDa complexes, indicative of F-actin formation. Traces of actin! Si complexes are also present in native G-actin samples. This is ascribable to the relatively high ionic strength (40 mM NaC1) of the buffer solution used to maintain Si in solution. These ionic conditions are able to promote a slow but appreciable formation of polymeric actin. However, significantly larger amounts of F-actinlSl complexes were formed in PQ-treated actin samples.
I 0.e
IPQI
1.2 (mu)
4. Pelleting, light scattering, and viscometry tests. A) Pelleting: SDS-PAGE of sedirnentable actin as a function of the overnight treatment with PQ at different concentrations. Native G-actin (C), native Factin (F); other central lines have been labeled with the concentration values (mM) of PQ used to treat the sample. B) Light scattering and viscosity. The graph shows the increase of both light scattering intensity (0) and specific viscosity (A) of G-actin samples after PQ treatments (further details in Materials and Methods). Figure
by results from light-scattering (Fig. 7A) and low-shear viscometry (Fig. 7B) assays in which both different signals, due to PQAA formation, markedly increase in the presence of a-actinin and filamin, respectively.
Interaction
of PQAAs
Vol. 11
March
1997
as actin
polymerization
seeds
We know that in the presence of preformed actin polymers (polymerization seeds), actin assembly loses its characteristic lag phase. PQAAs, if really composed by F-actin, should function as actin polymerization seeds, increasing the polymerization rate of polymerizing native-actin samples. Surprisingly, mechanically undisturbed PQAAs fail
with FITC-phalloidin
Phalloidin selectively interacts with F-actin (54). An immediate consequence of this specific interaction is that PQAAs should bind phalloidin only if showing a correct F-actin spatial configuration. Figure 8 (0) clearly shows that FITC-phalloidin binds PQAAs. To exclude the possibility that FITC-phalloidin should be simply trapped into sedimenting PQAAs, we monitored the anisotropy of the fluorescence from FITC-phalloidin/PQ -treated actin mixtures. Figure 8 (A) shows that the anisotropy of FITCphalloidin fluorescence emission increases with the presence of PQAAs. This, in turn, indicates that most of the FITC-phalloidin molecules lose their rotational diffusion by binding PQAAs. We conclude that FITC-phalloidinl PQAA interaction also occurs mnsolution and seems not to be forced by sedimentation. Note that Cappelletti et al. (24) suggested the unsuitability of phalloidin to bind F-
266
PQAAs
5. Electron microscopy. A) Negatively stained actin filaments formed in the presence of 2 mM MgCI2 + 100 mM KCI (Xi00,000). B) PQ-induced actin aggregates (PQAAs) formed in C-buffer (depolymerizing conditions) and in the presence of 2 mM PQ. Notice the cohesion of short, rod-like actin aggregates (X 100,000). Figure
The FASEB Journal
MILZANI
FT AL.
______________
160
A
S U
120
a-actinifl80
Actin
-
-“
1 0.5 0.11 F
2
a
40
Ut
[P0] (mM)
-J
0
_______________ Filamin
G1 2
1 0.5 0.1 F
[P0] (mM) teins.
6. The binding The
U-actinin obtained treated
affinity
between
of PQAAs
and filamnin)
has been
by centrifugation actin
PQAAs
vs. actin estimated
of protein
and one of two considered
and actin
cross-linking
cross-linking
proteins
by the SDS-PAGE
mixtures actin
containing cross-linking
pro(such
as
of pellets
overnight proteins.
PQA)
Native G-actimi + a-actinin (C), native F-actin + U-actinin (F). aactinin alone (aC). and a-actinin + 2 mM PQ (aj. Other lines refer to PQ-treated actin samples + a-actinin, and have been indicated with the concentration values (mnM) of PQ used to treat actin. B) Native Cactin + filamnin (C), native F-actin + filamin (F), fitamin alone (f), and filamin + 2 mM PQ (f1). Other lines refer to PQ-treated aetin samples + filamin and are labeled as in panel A. For details, see Materials and Methods.
to affect actin assembly. Their addition to polymerizing actin samples, in fact, does not influence the actin polymerization rate (Fig. bA). In contrast, previously sonicated PQAAs markedly nucleate actin assembly (Fig. lOB). This unexpected behavior seems to resemble that of end-blocked actin filaments that regain their growing ability after mechanical fragmentation (i.e., after the formation of new growing free ends). In the electron micrographs (Fig. 5), ii is evident that PQAAs appear to be filamentlike structures. These filaments are not continuous but seem to he built up by the aggregation of many very short fragments (Fig. 5B). Therefore, we suggest that PQAAs are constituted by the cohesion of very short actin polymers. This cohesion could imply the steric blockage of the growing ends of most of the actin fragments. Sonication, by disrupting the cohesion among short actin polymers, favors the regaining of polymer growth.
DISCUSSION Until now. redox cell systems have been considered to be the main target of PQ cytotoxicity (17), but severe altera-
PARAQUAT
AND
ACTIN
0.5
1.5
0
0.5
1.5
LPQI (mM) Figure 7. Bundles an(l network formnation. A) The a-actinin-induced bundle formation in PQ-treated actin samples was checked by following the light scattering increase, at 400 nm of wavelength (49), in PQtreated actin samples after the addition of a-actinimi. Light scattering enhances with the increase of PQ concentration. B) The filarnin-induced network formation was monitored by low-shear viscometry. After the addition of filamin to PQ-treated actin samples. apparent viscosity in(reases dramatically only in actin samples treated with PQ concentrations higher than 1 mM.
ii
-
0
B
Actin-
Figure
I
C
tions of the dynamic components of cytoskeleton (microtubules, MT; microfilaments, MF) have also been described (24, 30-32). All data regarding cytoskeleton sensitivity to PQ agree on the PQ-related disruption of cytoskeletal elements (30-32). Reconsidering their previous results (31, 32), Cappelletti et al. (24) have recently suggested that PQ does not act as an actin depolymerizing factor, but that it should be able to increase cell F-actin content. It is therefore clear that the mechanism (or mechanisms) at the basis of PQ cytotoxicity, and in particular, the PQ influence on cytoskeletal structures, remains quite obscure. To clarify this matter, we tested the effect of PQ on actin dynamics in solution. As shown in Fig. 1, actin selectively binds PQ (Fig. 1A) whereas other tested proteins seem to
5 U
8
I 0
0.4
0.8 (PQ
1.2
1.6
2
(mM)
Figure 8. PQAAs/FITC-phalloidin binding. The binding between PQinduced actin aggregates (PQAAs) and FITC-phalloidin was revealed by both FITC-phalloidin fluorescence intensity in supernatants after pelleting of PQAAs/FITC-phalloidin complexes (0) and fluorescence anisotropy increases as a function of PQAA formation (A). Considering that PQAA formation is enhanced with the increase of PQ concentration (see Fig. 4A), both sedimenting FITC-phalloidin and fluorescence anisotropy increase in parallel.
267
Si Heavy
-
-
.
Chain Actin
--
-
-
-
-
-
175K =465 K 95K 42K
G12 1 O.5F [PQ] (mM) 9. PQAAs/S1 interaction. The subfragment I of myosin was added to actin samples. incubated overnight with different concentrations of PQ. Then PQAAs/S1 complexes were cross-linked by EDC (see Materials and Methods). Cross-linked products were analyzed by SDSPAGE. Native C-actin + Si (C); native F-actin + Si (F); central lines refer to PQ-treated actin samples + SI and have been labeled by the concentration values of PQ (mM) used to treat actin. Figure
be refractory to the herbicide. In particular, the MTP failure in binding PQ (Fig. 1D) supports the findings of Cappelletti et al. (32) suggesting the insensitivity of cytoplasmic MT to PQ. The specific actin/PQ interaction causes a conformational change in the actin molecule as revealed by tryptophan fluorescence emission spectra of actin samples (Fig. 2) treated with increasing concentrations of PQ (from 0.1 to 2.0 mM). In comparing fluorescence emission spectra of treated actin samples with those of both native G-actin and F-actin, no spectral shifts have been detected. The maximum of fluorescence intensity decreases when actin assembly occurs (53) as well as when PQ concentration increases (Fig. 2). Considering also that light scattering is appreciable both in native F-actin and in PQ-treated actin samples, PQ-related modifications of actin intrinsic fluorescence strongly resemble those occuring during G/F transition. Similarly, digestion patterns of PQ-treated actins resemble that of native F-actin (Fig. 3). Even if the incubation is shortened to 10 mm, “F-actinlike digestion patterns” are inducible (Fig. 3A). PQ does not interfere with the proteolytic activity of the enzymes used here (not shown). By binding G-actin, PQ (per se) is able to induce conformational changes in the protein spatial structure, thus favoring monomeric actin aggregation in G-buffer (depolymerizing conditions). This raises the question of whether PQAAs are really built up by correctly structured F-actin. As clearly shown in Fig. 4B, 24 h exposure to PQ of actin samples (in G-buffer) results in an appreciable increase in both light scattering and specific viscosity. Moreover, PQAAs sediment (under centrifugation conditions routinely used for F-actin) (Fig. 4A) and are detectable by electron microscopy (Fig. 5). PQAAs bind F-actin cross-linking proteins, such as aactinin anti filamin (Fig. 6). The observed interactions seem to occur correctly, since both actin bundles and networks probably form. In fact, Fig. 7A suggests that after the addition of a-actinin to PQ-treated actin samples, the light scattering intensity of protein mixtures [at 400 nm,
268
Vol. 11
March
1997
which according to Wilkins and Lin (49) is selective for actin bundles] significantly increases with the enhancement of PQ concentration. Note that most of actin crosslinking proteins binds the subdomain 1 of filament actin subunits. a-Actinin, in particular, binds the subdomain 1 of two adjacent actin monomers along the long-pitch helix (58) of the actin filament (59). This suggests that PQAAs show a correct alignment of actin subunits suitable for the binding of a-actinin. The increase in the apparent viscosity of overnight PQ-treated actin samples, after the addition of filamin, revealed the probable formation of PQAA/ filamin networks (Fig. 7B). Viscosity enhancements are appreciable only in actin samples exposed to PQ concentrations ranging from 1.0 to 2.0 mM (where the formation of PQAAs is more pronounced). PQAAs/cross-linking protein interaction experiments imply the “F-actin nature” of PQAAs. Further tests regarding the binding of PQAAs with both phalloidin and the subfragment 1 (Si) of myosin, along with the PQAA nucleating activity on actin polymerization, should be conclusive. When FITC-phalloidinisadded to actinsamples (inGbuffer), treated overnight with increasing concentrations of PQ, and then immediately centrifuged, fluorescent phalloidin sediments along with PQAAs (Fig. 8, 0). Phalloidin sedimentation is not due to a simple trapping, but reflects its specific binding to PQAAs. After fluorescent phalloidin addition to PQ-treated actin samples, in fact, fluorescence anisotropy increases (Fig. 8, A), revealing that in solution, also, most of the fluorescent probe is orderly immobilized on PQAAs. The Si of myosin represents the “physiological” ligand for F-actin. We added Si to overnight PQ-treated actin samples, and then the PQAA/Si complexes formed were
7 ‘1 C 0
U
S S
U C C
U .
