Response from the authors
Writing a paper for Nature implies that one has to deal
with very stringent space limitations. As a consequence we could present
only a part of the information we current have about an expansin from the
potato cyst nematode Globodera rostochiensis. Here I would like to
comment on some of your concerns:
> (DC) But is this really an expansin?
> Structural characteristics:
> 1. Gr-EXP1 has two domains. The first domain is resembles a carbohydrate
binding module family II (CBM-II). This domain is not found in plant
expansins.
As described in the Cosgrove nomenclature ("defined more explicitly - after
the publication of Qin et al"), expansins domain II is a grass pollen
group-2 allergen-like domain (SCOP 1.63 49590) - GPG2. This is in fact also
a carbohydrate-binding module being a cellulose-binding domain [Henrissat,
Teeri and Warren (1998) FEBS Lett, 425,352-354 and Barre and Rougé (2002)
BBRC 296,1346-1351].
[DC comments: the
Henrissat et al. publication says nothing about this domain and the second
article relies entirely on unsupported speculation by me (!)
regarding the function of this domain. To my knowledge there are no
published data showing that this domain has carbohydrate-binding activity
and it is not included in the list of 39 carbohydrate-binding modules in
Henrissat's CAZy database] Moreover sequence to
3D structure alignment shows that the closest template of Gr-EXP1-D1 is a
xylan-binding domain from Cellulomonas fimi (1xbd), with 17% sequence
identity, 25% sequence homology and an e-value of 0.005, the second template
is a cellulose binding domain [found exo-1,4-beta-D-glycanase from
Cellulomonas fimi (1exg)] with 16% sequence identity, 23% sequence
homology). Both belong to Cellulose-binding domain family II (according to
SCOP classification). This suggests that Gr-EXP1-D1 may – just like plant
expansin domain II - be involved in plant cell wall matrix polysaccharide
adhesion, performing at molecular level a similar function with GPG2.
> (DC - continuation “Structural
characteristics”) > The second domain has significant sequence similarity to
plant expansin domain I. > This domain is part of a large superfamily of
domains that are found in a wide > variety of proteins - most notably
glycoside hydrolase family-45 (GH45) which > includes a number of potent
endoglucanases. Members of GH45 have been found in > a variety of organisms,
especially fungi, but also a bacterium, a protist living in the > hindgut of
termites, and a mollusk. This domain gets around! It is identified as >
50685 in SCOP (ver. 1.63) and is part of many different kinds of proteins,
not just > expansins and GH45 enzymes, but also barwin/hevein-like/PR4
proteins (plant > defense proteins), a plant signaling protein known as PNP,
and a variety of genes > of unknown function in many phyla. The sequence of
Gr-EXP1 is so distant from > that of plant expansins, that if we accept it
as an expansin, then all the proteins > within this domain superfamily (SCOP
50685) might logically be classified as > expansins. But many of them have
distinctive functions.
Indeed, SCOP families are rather large as they define only
the overall topology of the protein and do not refer to details (such as:
the length of the sequence, the number and length of secondary structure
elements, the accessibility profile, the S-S bond pattern or the presence of
various local sequence motifs involved in function). However the statement
that "if we accept Gr-EXP1 as an expansin, then all the proteins within this
domain superfamily (SCOP 50685) might logically be classified as expansins"
is not justified. By taking in account the details mentioned below and
performing sequence to 3D structure alignments one can easily spot with high
confidence the real structural homologues. In our case, even by eye
inspection one could easily find the best templates within the SCOP family
(see attachment). More elaborate techniques, indicate that the closest
template of Gr-EXP1-D2 is indeed the pollen allergen phl p1 from Phleum
pratense, N-terminal domain (PDB code: 1n10, plant expansin domain 1) with
~25% identity and an e-value of 3.18e-06 (corresponding to more than 95%
fold recognition confidence). With ~14% identity and an e-value of only
1.57e-01 (corresponding to less 80% fold recognition confidence), the second
relative (the Barwin protein) comes way below 1n10 as possible structural
homologue. In addition Barwin lectin is rather a single domain protein and
similar sequences could not be found as part of multi-domain proteins.
On short, therefore, Gr-EXP1 has 2 domains with similar
functionality to the 2 domains of plant expansins: -1) a CBM and -2) a plant
expansin domain I. In contrast to the plant expansins however, the domains
are swapped and the CBM could possibly recognise xylan rather than cellulose
(additional experiments needed). Based on these structural arguments we
think that Gr-EXP and Pl-EXPs are similar, and that the structural
definition "Only two-domain proteins will be designated as members of the
expansin superfamily. The name "expansin" should not given to proteins that
are homologous to only one of the expansin domains, i.e. the GH45-like
domain I (SCOP 1.63 50685) and the grass pollen group-2 allergen-like domain
II (SCOP 1.63 49590)." is rather arbitrary, eliminating domains that could
perform similar (but slightly different) functions at molecular level - as
for example targeting different parts of the complex plant wall†.
[DC comments: “A rose by any other
name would smell as sweet.” The definition of expansins based on sequence
may not be completely congruent with that based on sequence. Because it is
not feasible to assay the activity of every protein, we are left with the
less-than-perfect method of classifying based on sequence. With multi-domain
proteins, the classification problem is severely aggravated. Crystal
structure data convincingly show that expansin domain 2 is a structural
homologue of GH45 enzymes, a subset of SCOP 50685. Our tests show that GH45
enzymes lack significant expansin activity, so clearly not all proteins with
GH45 domains have expansin activity. There is no natural structural
demarcation between the expansin domain 2 and the catalytic domain of GH45
enzymes. When one wants to classify proteins containing a domain as
widespread and divergent as the GH45-like domain, one must draw lines
somewhere, and sometimes well-intentioned people will disagree on where the
line should be drawn.]
(† This hypothesis is included in Cosgrove et al. (1995)
PNAS, 92, 9245-, p. 9249)
> (DC) Functional characteristics: >it is
also possible that the cell wall extension activity was not due to the
expansin-related domain, >but to the CBM-II domain”.
Sure, carbohydrate-binding modules may also induce disruptions of the
non-covalent interactions in cellulose microfibrils. Therefore, it could
indeed be argued that the cellulose-binding domain in Gr-Exp1 is responsible
for the cell wall-loosening activity observed in the extensometer assays.
However, transgenic tobacco expressing the CBM of Gr-Exp1 alone showed no
cell wall-loosening activity (not significantly different from the empty
vector control plants). Primary data will be included in a follow-up paper.
> (DC) Expansin ... (c) lack significant
wall hydrolytic activity
We checked recombinant Gr-EXP1, and - while this protein showed a potent
cell wall expansion activity - no endoglucanase activity could be detected
in standard assays.
Hans Helder & Andrei Petrescu
- Hans Helder (Lab. Nematology, Wageningen University, The
Netherlands), e-mail: Hans.Helder@wur.nl - Andrei Petrescu (Institute of
Biochemistry of the Romanian Academy, Bucharest, Rumania), e-mail:
ap@biochim.ro
03 March 2004 (revised)
