|
|
|
Original Paper
|
|
|||
|
Acta Biochim Biophys
Sin 2006, 38: 356-362 |
||||
|
doi:10.1111/j.1745-7270.2006.00170.x |
Effects of Heat Stress on
Yeast Heat Shock Factor-Promoter Binding In Vivo
Ning LI1,
Le-min ZHANG1*,
Ke-Qin ZHANG1, Jing-shi DENG1, Ralf PR�NDL2&, and Fritz SCH�FFL2
1
Laboratory for Conservation and Utilization of
Bio-resources, Yunnan University, Kunming 650091, China;
2
Zentrum f�r Molekularbiologie der Pflanzen - Allgemeine
Genetik, Eberhard-Karls-Universit�t T�bingen, 72076 T�bingen, Germany
Received: January
9, 2006
Accepted: March 22,
2006
This work was
supported by the grants from the National Natural Science Foundation of China
(No. 30560012), the Department of Science and Technology of Yunnan Province
(2003C0012R, 2005NG05), the project sponsored by SRF for ROCS, SEM; and the grants
from Deutsche Forschungsgemeinschaft (PR511/1-1; SFB446-A2)
& Present address:
SerCon GmbH, Heinrich-von-Brentano-Sre. 2, D-55130 Mainz, Germany
*Corresponding
author: Tel, 86-871-5031094; Fax, 86-871-5034838; E-mail, [email protected]
Abstract������� Heat shock factor-DNA
interaction is critical for understanding the regulatory mechanisms of
stress-induced gene expression in eukaryotes. In this study, we analyzed the in
vivo binding of yeast heat shock factor (HSF) to the promoters of target
genes ScSSA1, ScSSA4, HSP30 and HSP104, using
chromatin immunoprecipitation. Previous work suggested that yeast HSF is constitutively
bound to DNA at all temperatures. Expression of HSF target genes is regulated
at the post-transcriptional level. However, our results indicated that HSF does
not bind to the promoters of ScSSA4 and HSP30 at normal
temperature (23 �C). Binding to these promoters is rapidly induced by heat
stress at 39 �C. HSF binds to ScSSA1 and HSP104 promoters under
non-stress conditions, but at a low level. Heat stress rapidly leads to a
notable increase in the binding of HSF to these two genes. The kinetics of the
level of HSF-promoter binding correlate well with the expression of target
genes, suggesting that the expression of HSF target genes is at least partially
the result of HSF-promoter binding stability and subsequent transcription
stimulation.
Key words������� chromatin immunoprecipitation; heat shock factor; heat
shock gene; yeast heat shock factor-promoter binding
Cells respond to elevated temperature and other physiological stresses by dramatically increasing the expression of heat shock proteins (HSPs), a set of proteins functioning as molecular chaperones, which are involved in the folding, trafficking, maturation and degradation of proteins. An increased accumulation of HSPs is essential for the survival of cells exposed to various stresses [1,2]. In eukaryotes, the expression of heat shock genes which encode HSPs is regulated by the binding of heat shock factors (HSFs) to heat shock gene promoter elements. heat shock element (HSE) is composed of tandem inverted repeats of a short consensus sequence 5'-nGAAn-3' [1]. The HSF-HSE interaction is conserved from yeast to human, but there is wide variability in the number of HSF genes in nature. Plants and mammals harbor multiple genes encoding HSF isoforms, with Arabidopsis thaliana possessing 21 distinct HSF genes and mammals possessing three genes [3,4]. The existence of multiple HSF isoforms might have a specialized function and regulate different target genes. Yeast, however, harbors only a single HSF which is thought to play multiple roles that are shared with the isoforms in higher eukaryotes [5,6]. Recent reports demonstrated yeast HSF is essential for heat-inducible transcription of not only HSPs but also genes encoding proteins involved in diverse cellular processes including growth, development, disease, aging, and in the complex metabolic reprogramming that occurs in response to stresses [7,8].
Considering the important role of HSF in the cellular homeostatic control, it is important to elucidate the regulatory mechanisms of HSF in cellular stress response. In plant, animal and mammalian cells, HSF is present in a latent, monomeric state under normal conditions; the HSF DNA-binding activity depends on the trimerization of monomeric HSF subunits upon heat stress, and the event of binding triggers the transcriptional activation of target genes [4,9,10]. In yeast, early studies using in vitro binding assay and genetic methods suggested that the stress inducing expression of HSF target genes is regulated at the post-transcriptional level, for instance, the heat-inducible phosphorylation of the HSE-bound HSF was suggested to be responsible for the activation of heat shock gene expression, whereas HSF binding to HSEs is strictly constitutive [5,11,12]. However, in vivo, protein-DNA binding is determined by additional factors which stimulate or inhibit binding. Later, in vivo footprinting was used to study protein-DNA interaction on the yeast HSP82 gene, and it was proposed that HSF binds to strong HSEs (three consensuses) constitutively, but binding to weak HSEs (poor matches to consensus) is dependent on heat stress [13,14]. However, footprinting does not identify the interacting protein.
In this report, we analyzed
the in vivo binding of HSF to target genes using cross-linking chromatin
immunoprecipitation (X-ChIP) which is thought to allow both the identification
of the protein and the interacting DNA sequence [15]. Our results showed that
HSF binds to different targets in a different manner. HSF binding to ScSSA4
and HSP30 is heat stress dependent. HSF binds to ScSSA1 and HSP104
under non-stress conditions, but at a low level; heat stress apparently increases
the binding level. The kinetics of binding correlate well with target gene
expression, suggesting that the binding plays a role in the transcriptional
stimulation of target genes
Materials and methods
Construction of expression
vectors
Plasmid pQE30 was isolated
from Escherichia coli TG1 (Qiagen, Carlsbad, USA) using a Nucleospin
Multi-8 plasmid kit (Machery-Nagel, D�ren, Germany). The yeast HSF
entire coding region DNA fragment (yHSF) was inserted, containing an adaptor of
the PstI restriction site located after the translation stop
codon, and an adaptor of the SacI restriction site located before
the translation start codon. They were produced by polymerase chain reaction
(PCR) using primer pair QE30-hsf/u (5'-GCACTGCAGCTATTTCTTAACTCGTTTGG-3')
and QE30-hsf/d (5'-GCAGAGCTCATGAATAATGCTGCAAATACA-3') at
annealing temperature 53 �C.
Plasmid pQE30 and the PCR
product of yHSF were digested with SacI, followed by PstI. After
ligation by T4 DNA ligase, the ligated construct pQE30/His-yHSF was transformed
to TG1 by electroporation. Positive colonies were identified by hybridization
with the HSF gene (coding sequence) probe which was synthesized using
the PCR product of yHSF as template. The constructs were verified by
sequencing. The colony with the verified construct was chosen for expressing
the recombinant peptides.
Preparation of
affinity-purified antibodies
Yeast His6-HSF
was expressed from TG1(pQE30/yHSF) and purified using denaturing-renaturing
purification according to the QIAexpressionist (Qiagen), verified by Western
blot analysis and used for the generation of antiserum against yeast HSF in
rabbit. For the generation of monospecific antibodies, His6-HSF
peptide was coupled to AminoLink Plus Coupling Gel (Pierce, Rockford, USA) and
the affinity purification of the antibodies was carried out using ImmunoPure
IgG Elution Buffer (Pierce).
Chromatin immunoprecipitation
Yeast cells (Saccharomyces
cerevisiae, strain Y190) were grown in 50 ml SD-Leu medium (26.7 g/L
minimal synthetic defined base, 0.69 g/L Leu drop out supplement, ph 5.8) (BD Biosciences Clontech, Palo
Alto, USA) to an optical density (OD600)
of 2 at 23 �C, then heated at 39 �C for 0, 10, 30, and 60 min. Cell cultures
were cross-linked with 1% formaldehyde and chromatin was isolated according to
the method described by Strahl-Bolsinger et al. [16]. Four hundred
microliters of the resulting chromatin from each sample was sonicated for 10�20 s (resulting in fragment
sizes 0.3-0.5 kb) using a Branson
Digital Sonifier (S-250D; Branson Ultrasonic, Danbury, USA) at 40% output
(keeping the sample on ice during sonication). The chromatin extracts were
centrifuged at 10,000 rpm at 4 �C for 20 min, and the supernatants were
filtered through a 0.22 mm filter.
Ten micrograms of
affinity-purified anti-HSF antibodies was added to 400 ml
chromatin extract samples and incubated at 4 �C for 3 h, then 60 ml protein A-agarose preincubated with 1%
BSA was added to block non-specific binding, and the samples were incubated at
4 �C for 30 min. Subsequently, the resin was washed and the immunoprecipitated
material was eluted according to the method described by Strahl-Bolsinger et
al. [16]. Eighty microliters of the resulting elution material was digested
by adding 80 ml of proteinase K (1 mg/ml) in
proteinase K buffer (50 mM Tris, 10 mM EDTA and 0.3% sodium dodecylsulfate),
incubated at 60 �C for 16 h (from this point, 80 ml
chromatin extract was processed in parallel to obtain total genomic DNA),
extracted twice with phenol/CHCl3, and once with CHCl3.
DNAs were precipitated by ethanol and dissolved in 20 ml
TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0).
PCR analysis of
immunoprecipitated DNA
Immunoprecipitated DNA was
analyzed by PCR. One microlitre of immunoprecipitated material was added to 20 ml PCR reaction mixture (50 mM KCl, 20 mM
Tris-HCl, pH 8.3, 0.05% Tween-20, 0.01 mg/ml gelatin, 2.5 mM MgCl2,
2.5 mM each dNTP, 1 U Taq polymerase). PCR reaction was carried out for
30 cycles (unless elsewhere indicated) of 20 s at 94 �C, 20 s at 49-56 �C and 10 s at 72 �C. The PCR products
were separated on a 10% polyacrylamide gel and visualized by staining with
ethidium bromide. The following primer pairs were used to detect target genes:
ScSSA1-307u, 5'-TCAACTAAAATCTGGAGAAAA-3';
ScSSA1-314d, 5'-CGGAACGTTTAGAAGCTGTCATT-3',
53 �C; ScSSA1+1893d, 5'-GGTGCTCCTCCAGCTCCA-3'; ScSSA1+1899u, 5'-TCAACGGTTGGACCTTCA-3',
54 �C; ScSSA1+1193d, 5'-TGTCGCTCCATTATCCTT-3'; ScSSA1+1199u, 5'-CACCACCAGCAGTTTCAA-3',
54 �C; ScSSA4-301u, 5'-AACTCACCGGGCAAAAGA-3';
ScSSA4-308d, 5'-AATGTAATAGGTTTCAAAG-3',
49 �C; ScSSA4-364u, 5'-GAGAGTACATACCGGAATG-3';
ScSSA4-429d, 5'-ACTAAATTACGTTCATAGGG-3',
54 �C; ScSSA4+1807d, 5'-GTTAGATGCTTCGCAAGC-3'; ScSSA4+1814u, 5'-TCCTTGTATTCCTCGGTG-3',
54 �C; HSP30-503u, 5'-AAGCACGCTTTCGATGCG-3';
HSP30-514d, 5'-CGTAGGAGGATTCTCTCA-3',
52 �C; HSP104-131u, 5'-TGAGGCAAGATTACAATGC-3';
HSP104-143d, 5'-CTTATGCAACCTGCCAGA-3',
52 �C; YPL231W-1170u, 5'-AGTGACTTGCTTGCTCCTC-3';
YPL231W-1206d, 5'-GATTCTTTGCATAAGAGGCTA-3',
55�C. Primers were named according to the gene and the distance of the 3'
nucleotide relative to the ATG translation start codon. The annealing
temperatures are given.
Results
HSF binding site detected at
high resolution by cross-linking chromatin immunoprecipitation and PCR analysis
To detect HSF binding to
target in vivo, we employed the X-ChIP technique and PCR analysis for
immunoprecipitated DNA. We successfully detected the binding of yeast HSF to
the ScSSA4 gene in vivo and examined the binding site at high
resolution. Yeast cells were heat stressed at 39 �C for 10 min, followed by
formaldehyde cross-linking and immunoprecipitation using affinity-purified
yeast HSF antibodies. The immunoprecipitated DNAs were analyzed by PCR using
different primer pairs. The genes and distribution of primers are shown in Fig.
1. For positive controls, total genomic DNAs were used as the
templates for PCR amplification. For negative controls, we used
immunoprecipitates without prior formaldehyde cross-linking or without HSF
antibodies. The results are shown in Fig. 2. Two primer pairs, ScSSA4-364u/-429d
and ScSSA4-301u/-308d,
were used, which were located in the promoter region of ScSSA4. The
space between primers in ScSSA4-364u/-429d
and ScSSA4-301u/-308d
was 64 and 6 nucleotides, respectively. Both primer pairs properly produced
specific PCR products and gave similar results. Heat stressed samples gave
strong signals after 40 cycles of amplification, but there were weak
background signals for negative controls and weak signals for the sample
without heat stress. However, the signals of the heat stressed samples
were significantly stronger than those of other samples, indicating heat stress
induced HSF binding to the ScSSA4 gene. The weak bands of the
sample without heat stress were comparable with the bands for the negative
controls. With the PCR cycle reduced to 30 cycles, the heat stressed samples
still gave clear signals, whereas the signals were abolished for both
the negative controls and the sample without heat stress, indicating HSF did
not bind to ScSSA4 promoter under non-stress conditions, but the binding
occurred after heat stress.
To determine the spatial
specificity of HSF binding, we amplified immunoprecipitates using primer pair
YPL231W-1206d/-1170u
located in a randomly chosen YPL231W promoter. The results showed that
no signal was detected and HSF did not bind to the non-target promoter. To
further define the binding site, we analyzed whether HSF binds to the space
adjacent to the ScSSA4 promoter by using primer pair ScSSA4+1807d/+1814u
located in the coding region of the ScSSA4 gene, approximately 1800 bp
away from the start codon. No PCR signal was detected with this primer pair,
demonstrating the X-ChIP technique could detect the HSF binding site at high
resolution.
Effects of heat stress on
binding of HSF to target genes in vivo
Early experiments suggested that yeast HSF was localized in the nucleus and bound to the target gene promoter, even in the absence of heat stress. However, the above results showed that HSF binding to the ScSSA4 promoter is dependent on heat stress. This contradiction prompted us to further investigate the effects of heat stress on HSF binding to different target promoters.
Based on multiple independent chromatin immunoprecipitation experiments, we analyzed the dynamic association of HSF with the promoters of ScSSA1, ScSSA4, HSP30, and HSP104 genes in vivo. Yeast cells were heat shocked at 39 �C for 0, 10, 30 and 60 min, followed by formaldehyde cross-linking and immunoprecipitation. The PCR amplification of immunoprecipitated DNA using primer pairs located in different promoter regions are shown in Fig. 3. The results showed that HSF bound to ScSSA1 and HSP104 promoters under non-stress conditions, whereas no binding occurred to ScSSA4 and HSP30 without heat stress. Heat stress rapidly increased HSF binding to ScSSA1 and HSP104 promoters and induced HSF binding to ScSSA4 and HSP30 promoters. The level of HSF binding to these four promoters reached a peak after 10 min of heat stress and declined with continuous exposure to heat stress for up to 1 h, however, it was slightly higher than basal level in the cases of ScSSA1 and HSP104. The negative controls without formaldehyde cross-linking or without antibodies did not give PCR signals. Therefore we excluded the effect of the background signal. Two primer pairs located in the ScSSA1 coding regions and one primer pair situated in the YPL231W promoter region did not have PCR signals, indicating HSF did not bind to non-target DNA. These results suggested that HSF binds to different targets in a different manner, constitutively or heat stress-dependently, and heat stress apparently increases binding affinity.
The comparison of HSF binding
levels in the time-course experiment with the profiling of target gene
expression within published data indicates that HSF binding correlates well
with target gene expression, suggesting that induced or increased HSF binding
plays an important role in the regulation of yeast target gene activation.
Discussion
It was initially thought that
yeast HSF trimers constitutively bind to HSEs under both normal and heat stress
conditions. Later, based on in vivo footprinting experiments, Erkine et
al. [13] and Giardina and Lis [14] proposed that HSF binding to strong HSEs
of HSP82 is constitutive, but binding to weak HSEs is induced by heat
stress. However, our results indicate that heat stress significantly affects
binding behavior, and that HSF binds to different promoters in a different
manner. Binding to HSP30 and ScSSA4 genes is heat
stress-dependent. HSF does not bind to these two genes under non-stress
conditions, even the ScSSA4 promoter that contains strong HSE. HSF binds
to ScSSA1 and HSP104 promoters under non-stress conditions, but
at a low level. Heat stress rapidly leads to a notable increase in the binding
of HSF to these two genes. Interestingly, the HSF binding level is attenuated
by prolonged exposure to heat stress. A similar phenomenon was previously found
in higher eukaryotes and it was proposed that the binding stability is
negatively regulated by the association of HSF with other proteins, including
HSP70 and HSP90 [17,18].
However, our experiments
suggested that HSE is not the sole determinant of HSF binding. For instance,
previously published work proposed that HSP30 is not under the control
of HSF because of the lack of a typical HSE in the HSP30 promoter [19].
Our results showed that HSF binds to HSP30 in a heat-induced manner. We
also verified the lack of HSF binding to the YPL231W promoter, which contains
a consensus HSE. Although HSE is known to be important for HSF binding, the
lack of HSF binding to the YPL231W promoter and the heat-induced binding
to HSP30 suggest that its presence is clearly not the sole determinant
of whether a promoter is bound and activated by HSF. It appears that HSF
binding behavior in yeast is more complex than previously thought.
In order to analyze the
putative effects of heat stress-induced or increased HSF binding on
transcriptional activities of target genes, we examined the gene expression in S.
cerevisiae within published data. The ScSSA4 mRNA level rapidly
increased after 15 min and 30 min of heat stress at 39 �C and declined after 60
min of heat stress [11]. HSP30 mRNA was not detectable under normal
conditions; however, the expression was induced by heat stress and other
stresses [19]. The work by Causton et al. [20] showed that the
expression level of ScSSA4, HSP104 and ScSSA1 was
increased by 95-, 7- and 3-fold over basal levels, respectively, after 15 min
of heat stress, and subsequently declined during continuous heat stress [20].
The timing of induction and the decline of the mRNA correlate well with HSF
binding to the target genes in our time-course experiment, suggesting that
expression of the HSF target genes is, at least in part, the result of the HSF-promoter
binding stability and the consequent transcription stimulation, similar to that
of higher eukaryotes [10]. This information, along with the findings on the
role of heat-inducible phosphorylation of HSF in the expression of target
genes, would promote our understanding of the complexity of yeast HSF
regulatory mechanisms.
In our experiments, the X-ChIP
technique was excellent for detecting HSF target DNA. However, it must be noted
that immunoprecipitated DNA is relative enrichment of targets. To exclude
background signals, it is essential to use negative controls in each experiment
and design the appropriate number of PCR cycles. Only the signals produced by
immuno-enriched targets, shown to be significantly stronger than the background
signals produced by negative controls, can be considered. Furthermore, by
carefully reducing the number of cycles, background signals were abolished,
whereas the signals of the heat shock samples still existed. Therefore we could
clearly determine the true binding signal. Our experiments also demonstrated
that PCR analysis of target DNA is accurate in detecting binding sites. Only
the primer pair located in the promoter region gave signals, whereas the primer
pair located in the adjacent coding region of the same gene did not. To
investigate the binding site and non-binding site at high resolution, we
designed primers that were closely spaced in each primer pair and resulted in a
PCR product with a short fragment (41-175 bp).
Because the PCR reactions were carried out in a short phase of synthesis (10 s
at 72 �C), the primer pairs listed in this report produced proper PCR products,
and the specificity of the PCR products was further verified by sequencing in
the analysis of a randomly chosen sample. However, we also found that some
primer pairs did not work well in the amplification of such a short fragment,
so it is necessary to test multiple primer pairs and choose the most suitable.
However, with the digestion of formaldehyde-fixed chromatin with nonspecific
proteinases, such as proteinase K, it is difficult to yield a completely
peptide-free DNA [21]. Other cross-linking agents, such as ultraviolet light,
also cause significant DNA damage [22]. These DNA modifications hamper the PCR
analysis as PCR techniques require the DNA integrity for primer extension. The
short space between primers would minimize the effect of DNA damage on PCR.
References
1�� Mager WH, De Kruijff AJ. Stress-induced
transcriptional activation. Microbiol Rev 1995, 59: 506-531
2�� Morimoto RI. Regulation of the heat shock
transcriptional response: cross
talk between a family of heat shock factors, molecular chaperones, and negative
regulators. Genes Dev 1998, 12: 3788-3796
3�� Nover L, Bharti K, Doring P, Mishra SK,
Ganguli A, Scharf KD. Arabidopsis and the heat stress transcription
factor world: how many heat
stress transcription factors do we need? Cell Stress Chaperones 2001, 6: 177-189
4�� Sarge KD, Zimarino V, Holm K, Wu C, Morimoto
RI. Cloning and characterization of two mouse heat shock factors with distinct
inducible and constitutive DNA-binding ability. Genes Dev 1991, 5: 1902-1911
5�� Sorger K, Pelham HRB. Yeast heat shock factor
is an essential DNA-binding protein that exhibits temperature-dependent
phosphorylation. Cell 1988, 54: 855-864
6�� Wiederrecht G, Seto D, Parker CS. Isolation
of the gene encoding the S. cerevisiae heat shock transcription factor.
Cell 1988, 54: 841-853
7�� Hahn JS, Hu Z, Thiele DJ, Iyer V. Genome-wide
analysis of the biology of stress responses through heat shock transcription
factor. Mol Cell Biol 2004, 24: 5249-5256
8�� Yamamoto A, Mizukami Y, Sakurai H.
Identification of a novel class of target genes and a novel type of binding
sequence of heat shock transcription factor in Saccharomyces cerevisiae.
J Biol Chem 2005, 280: 11911-11919
9�� Westwood JT, Wu C. Activation of Drosophila
heat shock factor: Conformational change associated with a monomer-to-trimer
transition. Mol Cell Biol 1993, 13: 3481-3486
10� Zhang L, Lohmann C, Prandl R, Schoffl F. Heat
stress-dependent DNA binding of Arabidopsis heat shock transcription
factor HSF1 to heat shock gene promoters in Arabidopsis suspension
culture cells in vivo. Biol Chem 2003, 384: 959-963
11� Hashikawa N, Sakurai H. Phosphorylation of the
yeast heat shock transcription factor is implicated in gene-specific activation
dependent on the architecture of the heat shock element. Mol Cell Biol 2004,
24: 3648-3659
12� Jakobsen BK, Pelham HR. Constitutive binding
of yeast heat shock factor to DNA in vivo. Mol Cell Biol 1988, 8: 5040-5042
13� Erkine AM, Magrogan SF, Sekinger EA, Gross DS.
Cooperative binding of heat shock factor to the yeast HSP82 promoter in
vivo and in vitro. Mol Cell Biol 1999, 19: 1627-1639
14� Giardina C, Lis JT. Dynamic protein-DNA
architecture of a yeast heat shock promoter. Mol Cell Biol 1995, 15: 2737-2744
15� Orlando V. Mapping chromosomal proteins in
vivo by formaldehyde-crosslinked-chromatin immunoprecipitation. Trends
Biochem Sci 2000, 25: 99-104
16� Strahl-Bolsinger S, Hecht A, Luo K, Grunstein
M. SIR2 and SIR4 interactions differ in core and extended telomeric
heterochromatin in yeast. Genes Dev 1997, 11: 83�93
17� Abravaya K, Myers MP, Murphy SP, Morimoto RI.
The human heat shock protein hsp70 interacts with HSF, the transcription factor
that regulates heat shock gene expression. Genes Dev 1992, 6: 1153-1164
18� Zou J, Guo Y, Guettouche T, Smith DF, Voellmy R.
Repression of heat shock transcription factor HSF1 activation by HSP90 (HSP90
complex) that forms a stress-sensitive complex with HSF1. Cell 1998, 94: 471-480
19� Seymour IJ, Piper PW. Stress induction of
HSP30, the plasma membrane heat shock protein gene of Saccharomyces
cerevisiae, appears not to use known stress-regulated transcription
factors. Microbiology 1999, 145: 231-239
20� Causton HC, Ren B, Koh SS, Harbison CT, Kanin
E, Jennings EG, Lee TI et al. Remodeling of yeast genome expression in
response to environmental changes. Mol Biol Cell 2001, 12: 323-337
21� Solomon MJ, Varshavsky A.
Formaldehyde-mediated DNA-protein crosslinking: a probe for in vivo chromatin structures. Proc Natl
Acad Sci USA 1985, 82: 6470-6474
22� Zhang L, Zhang K, Prandl R, Schoffl F.
Detecting DNA-binding of proteins in vivo by UV-crosslinking and
immunoprecipitation. Biochem Biophys Res Commun 2004, 322: 705-711