Báo cáo khoa học: Crystal and solution structure, stability and post-translational modifications of collapsin response mediator protein 2 pdf

By a proteomic approach, we further identified a number of post-translational modifications in CRMP-2 from rat brain hippocampus and mapped them onto the crystal structure.. Results CD spe

post-translational modifications of collapsin

response mediator protein 2

Viivi Majava1, Noora Lo¨ytynoja1, Wei-Qiang Chen2, Gert Lubec2and Petri Kursula1

1 Department of Biochemistry, University of Oulu, Finland

2 Department of Pediatrics, Medical University of Vienna, Austria

Axonal growth cone guidance is a tightly regulated

process central to both nervous system development

and its repair after injury One of the proteins shown

to play a specific role in growth cone guidance is the

collapsin response mediator protein 2 (CRMP-2) [1–3],

also known as dihydropyriminidase-related protein 2

(DRP-2, DPYSL-2), unc-33 like protein 2 (Ulip-2) and

turned on after division 64 kDa (TOAD-64)

CRMP-2 is a member of the family of collapsin

response mediator proteins, which is comprised of five

related proteins in humans [4] The crystal structure of

human CRMP-2 has previously been determined to a

resolution of 2.4 A˚ [5] from crystals grown in the

pres-ence of calcium ions Structurally, CRMP-2 is

homolo-gous to the dihydropyriminidases (DHP), with a major

part of its 3D structure being formed as a (ba)8barrel, but it has no characterized catalytic activity of its own, nor have any specific small-molecule ligands for CRMP-2 been identified However, interactions between CRMP-2 and other proteins, such as tubulin [6], Sra-1 [7], Numb [8], a2-chimaerin [9] and phospho-lipase D [10], have been described

CRMP-2 is a homotetramer, but it can also form heterotetramers with other members of the CRMP⁄ DHP structural family [11] It is possible that its functional mechanism in neuronal development relates, at least partly, to its ability to bind to and regulate other homologous proteins in the family Recently, CRMP-2 has been highlighted as a target for drug development against nervous system disorders,

Keywords

divalent cations; nervous system; oligomeric

status; protein structure; small-angle X-ray

scattering

Correspondence

P Kursula, Department of Biochemistry,

University of Oulu, P.O Box 3000,

FIN-90014 Oulu, Finland

Fax: +358 8 5531141

Tel: +358 44 5658288

E-mail: petri.kursula@oulu.fi

Database

The coordinates and structure factors have

been deposited in the Protein Data Bank

under the accession code 2VM8

(Received 28 April 2008, revised 14 July

2008, accepted 17 July 2008)

doi:10.1111/j.1742-4658.2008.06601.x

The collapsin response mediator protein 2 (CRMP-2) is a central molecule regulating axonal growth cone guidance It interacts with the cytoskeleton and mediates signals related to myelin-induced axonal growth inhibition CRMP-2 has also been characterized as a constituent of neurofibrillary tangles in Alzheimer’s disease CD spectroscopy and thermal stability assays using the Thermofluor method indicated that Ca2+and Mg2+affect the stability of CRMP-2 and prevent the formation of b-aggregates upon heating Gel filtration showed that the presence of Ca2+ or Mg2+ promoted the formation of CRMP-2 homotetramers, and this was further proven by small-angle X-ray scattering experiments, where a 3D solution structure for CRMP-2 was obtained Previously, we described a crystal structure of human CRMP-2 complexed with calcium In the present study,

we determined the structure of CRMP-2 in the absence of calcium at 1.9 A˚ resolution When Ca2+ was omitted, crystals could only be grown in the presence of Mg2+ ions By a proteomic approach, we further identified

a number of post-translational modifications in CRMP-2 from rat brain hippocampus and mapped them onto the crystal structure

Abbreviations

CRMP-2, collapsin response mediator protein 2; DHP, dihydropyriminidase; SAXS, small-angle X-ray scattering.

such as epilepsy [12], Alzheimer’s disease and nerve

injury [3] Its association with depression and

schizo-phrenia has also been studied [13,14] CRMP-2 also

forms a part of the signal transduction cascade related

to axonal growth inhibition brought about by myelin

components, such as the myelin-associated

glycopro-tein [15], a mechanism that prevents the regeneration

of myelinated axons in the central nervous system

Non-neuronal expression for CRMP-2 has also been

reported [16] and, recently, it has been identified as a

putative marker for colorectal carcinoma [17]

The best characterized function for CRMP-2 relates

to its interactions with cytoskeletal components,

espe-cially tubulin [6,18] CRMP-2 is able to regulate the

formation of microtubules and, accordingly, it is

highly concentrated in growing axons CRMP-2 binds

to tubulin heterodimers, and its overexpression in

neurons promotes axonal growth and branching [6]

Extensive post-translational modifications have been

detected for CRMP-2 [19–25]; mainly, this has

con-cerned phosphorylation of the C-terminal tail, which is

predicted to be unfolded In addition, CRMP-2 has

been characterized as a major target for oxidation in

the aging brain [23,26–29] Changes in CRMP-2

post-translational modifications have also been suggested to

play a role in Alzheimer’s disease [19–21,26–29]

In the present study, we describe the crystal

struc-ture of human CRMP-2 in the absence of calcium

Instead of calcium, magnesium ions were required to

grow the crystals, resulting in a 1.9 A˚ structure of

CRMP-2 being obtained The effects of Ca2+ and

Mg2+on the CRMP-2 structure were also analysed by

CD spectroscopy, the Thermofluor method,

small-angle X-ray scattering (SAXS) and gel filtration At a

20 mm concentration, both CaCl2 and MgCl2 stabilize

the protein and promote the formation of tetramers

The structure was further analysed by mapping

post-translational modifications, as detected using advanced

proteomics methods [30], onto the 3D structure of the

folded core domain of CRMP-2

Results

CD spectroscopic analysis of CRMP-2

conformation and stability in solution

To characterize the effects of divalent cations on

CRMP-2 structure and stability, CD spectroscopy was

carried out in the presence of NaCl, CaCl2and MgCl2

The CD spectra of CRMP-2 in all tested conditions

were similar, indicating the expected presence of both

a and b secondary structures However, in the presence

of Ca2+ and Mg2+, the CD signal was significantly

stronger, suggesting a higher average content of sec-ondary structure in solution NaCl had no effect on the CD spectrum (Fig 1A)

Temperature scans of the samples revealed an intriguing phenomenon upon denaturation In buffer alone, CRMP-2 underwent a structural transition at approximately 50C, which resulted in an increase in ellipticity at 220 nm (Fig 1B); this is opposite to the effect generally expected upon protein denaturation The ellipticity did not significantly decrease, even at temperatures approaching 100 C (data not shown) A

CD spectrum recorded at 90C shows that the transi-tion involved a complete loss of helical structure and a significant increase in the amount of b structure (Fig 1C) After cooling down, the spectrum at room temperature indicated that the observed structural transition into a b-aggregate was irreversible (data not shown) A sample analysed in the presence of 50 mm NaCl behaved essentially the same (data not shown)

In the presence of Ca2+ or Mg2+, however, heat denaturation proceeded as expected, with a sharp decrease in ellipticity at 220 nm at the melting temper-ature and complete loss of secondary structure (Fig 1B,C) Increasing the concentration of these ions from 20 to 200 mm resulted in a decrease by several degrees in the Tm (Fig 1B and Table 1), indicating destabilization of CRMP-2 by a high concentration of divalent cations

Test for heat stability of CRMP-2 using the Thermofluor method

A series of conditions were screened in 96-well format,

by measuring the fluorescence of SyproOrange, in order to further characterize the stability of CRMP-2

in the presence and absence of divalent cations The results clearly indicate that, although 20 mm CaCl2 and MgCl2 stabilize the protein slightly, a 200 mm concentration destabilizes the protein significantly (Fig 1D,E and Table 1) The same effect was observed

in two different buffers, phosphate and Hepes, both adjusted to pH 7.5 The results are summarized in Table 1

Divalent cations promote the tetramerization

of CRMP-2 The oligomerization status of CRMP-2 was analysed

by gel filtration in the presence and absence of 20 mm CaCl2 and MgCl2 It is evident that, in the absence of divalent cations, only approximately 50% of CRMP-2

is in the tetrameric state The rest of CRMP-2 is in its monomeric form In the presence of either Ca2+ or

Mg2+, however, the protein is almost completely

tetrameric (Fig 2A), with only a minor fraction of

monomer being detectable

The structure of CRMP-2 in the absence of

calcium ions

The replacement of calcium with magnesium in

CRMP-2 crystallization gave rise to a novel

ortho-rhombic crystal form, which diffracted X-rays

signifi-cantly better than the monoclinic form previously obtained in the presence of Ca2+[5] Thus, the human CRMP-2 structure could be refined to a resolution of 1.9 A˚ (Fig 2B and Table 2) As a slight drawback, a significant pseudotranslational symmetry component was present in the new crystal form (see below), which lead to unusually high R factors in refinement The anomaly of the crystal form is given rise to by the noncrystallographic symmetry axes present in the CRMP-2 tetramer

C

D E

Fig 1 Folding and stability of CRMP-2 (A)

CD spectra for CRMP-2 in different buffers.

The spectra were measured at 23.4 C as

described in the Experimental procedures.

The samples are: 0 ⁄ 50 m M NaCl (thin ⁄ thick

black line); 200 ⁄ 20 m M CaCl2(thin ⁄ thick red

line); and 200⁄ 20 m M MgCl2(thin ⁄ thick blue

line) in 10 m M Hepes (pH 7.3) (B) Melting

curves based on the change in molar

elliptic-ity as a function of temperature Sample

colours are as described in (A) (C) Spectra

for CRMP-2 after heating denaturation,

measured at 90 C The samples contain

200 m M CaCl 2 (red), 200 m M MgCl 2 (blue),

or no additives (black) in 10 m M Hepes

(pH 7.3) In the absence of divalent cations,

CRMP-2 forms a b-aggregate (D)

Thermo-fluor stability assay Eight replicate samples

are shown at a single condition with 50 m M

phosphate buffer + 20 m M CaCl2 The

curves were normalized such that the

maxi-mum is 1 and the minimaxi-mum is 0 (E)

Super-position of averaged Thermofluor curves

from samples under the conditions:

phate buffer (thin black line);

phos-phate + 150 m M NaCl (thick black line);

phosphate + 20 m M CaCl 2 (thick red line);

phosphate + 200 m M CaCl 2 (thin red

line); phosphate + 20 m M MgCl2(thick

blue line); and phosphate + 200 m M MgCl2

(thin blue line).

As previously described [5], CRMP-2 forms a

homo-tetramer by an arrangement with 222 symmetry (i.e a

‘dimer of dimers’) (Fig 3B) We have also previously

suggested that this type of oligomeric assembly is

responsible for dimer formation between homodimers

of CRMP-1 and CRMP-2, resulting in a

hetero-tetramer [5] The protein used in both the current and

previous [5] structural studies contains a His-tag,

which is disordered in the crystal structure; thus, it is

highly unlikely that the oligomeric status or stability of

CRMP-2 would be affected by the affinity tag

Analysis of the packing

Pseudotranslation with the vector 0,0.175,0.5 (fraction

26.3%) was clearly observed in the data obtained from

the orthorhombic crystal form, indicating that a

frac-tion of the tetramers in the unit cell are related to each

other by this vector This was confirmed by the solved

structure, in which, for each of the four tetramers

in the unit cell, there is another one related by

pseudotranslation Pseudotranslation affects refinement

statistics by incorporating a large amount of weak

reflections and, thus, leads to the crystallographic R

factors during refinement being higher than generally

expected for the used resolution range Taking this

into account, the observed Rcrystand Rfreevalues from

this crystal form are acceptable An analysis of the

data using rstats software [31] also confirmed that

the weak reflections in the data had systematically high

crystallographic R factors (data not shown) A better quality indicator in such a case is the correlation coefficient, and these values indicate that the CRMP-2 model is accurate (Table 2)

Comparison with the structure in the presence

of calcium The high-resolution structure obtained from the ortho-rhombic crystal form grown in the presence of Mg2+, but in the absence of Ca2+, was compared with the

Table 1 T m values from CD spectroscopy and Thermofluor assays.

In both cases, the T m was taken as the point of steepest ascent of

the measured curve The number of replicates for each condition in

the Thermofluor assays is given in parentheses.

CD (all in 10 m M Hepes, pH 7.3)

Thermofluor

A

B

Fig 2 Divalent cations and the oligomeric structure of CRMP-2 (A) Analysis of oligomeric state by size exclusion chromatography The samples contained either no additives (black), 20 m M CaCl2 (red), or 20 m M MgCl 2 (blue) The elution volumes of molecular mass markers (in kDa) are indicated above the graph (B) The crys-tal structure of CRMP-2 and the locations of detected divalent cations Ca 2+ ions (from the previous structure) [5] are shown in red and Mg2+(from the current structure) are shown in blue, and the different subunits of CRMP-2 are colour-coded Only the ions bound to the two monomers in front are visible in this view.

previous structure determined in the presence of Ca2+.

The Ca rmsd between the new structure and the

pre-vious one is 0.2–0.35 A˚, depending on which chains of

the tetramer are compared The rmsd of Ca positions for the whole tetramer is 0.32 A˚, as determined by ssm [32]

The electron density clearly indicates that no large ion is bound to the previously observed calcium site [5]

in the new crystal form This observation further confirms the presence of a Ca2+ ion in the previous structure because the main difference between the crys-tallization conditions is the change of Ca2+ to Mg2+ The solvent environment of the new crystal structure was analysed to identify putative Mg2+ions The only location that shows an electron density reminiscent of hydrated Mg2+, as well as a suitable coordination environment, is approximately the same site where

Ca2+was bound in the earlier structure, at the mouth

of the central pocket in the (ba)8 barrel (Fig 2B) It should be noted that Mg2+ is not easy to distinguish from a water molecule based on electron density alone because it has the same number of electrons as oxygen

Table 2 Crystallographic data collection and structure refinement.

Data collection

Structure refinement

Correlation coefficient (free) 0.847

A

B

C

Fig 3 Small-angle X-ray scattering (A)

Superposition of the measured scattering

curves in the presence (red) and absence

(green) of 20 m M CaCl 2 , as well as the

theo-retical scattering curve calculated from the

crystal structure (black) using CRYSOL (B)

Distance distribution functions for CRMP-2

in the presence (red) and absence (green) of

CaCl2 (C) Ab initio models from DAMMIN

(spheres) superimposed on the crystal

struc-ture (ribbons) Red spheres indicate the

SAXS structure in the presence of Ca 2+ , and

green spheres indicate the structure in its

absence.

Considering the effects of Ca2+ on CRMP-2

oligo-merization and stability, the solvent shell of the earlier

CRMP-2 crystal structure (Protein Data Bank entry

2GSE) [5] was carefully re-analysed to locate any

calcium ions, as indicated by an electron density too

strong for a water molecule and a good coordination

environment for calcium, that might explain such

effects Indeed, two novel calcium sites were identified

in each CRMP-2 monomer These are located between

the backbone carbonyls of residues 128 and 464 and

near the side chain of Gln245, which is located at an

oligomerization interface (Fig 2B) In the structure

determined in the presence of Mg2+, these sites do not

contain Mg2+ions

Small-angle X-ray scattering

The oligomeric status and solution structure of

CRMP-2 were studied using synchrotron SAXS in

the presence and absence of Ca2+ In line with the

above results, SAXS indicated a tetrameric structure

for the sample measured in the presence of CaCl2

(Fig 3 and Table 3) The estimated molecular mass

of 191 kDa, based on a standard sample of BSA,

also clearly shows the predominant presence of a

tetrameric form (expected molecular mass of

220 kDa) The crystal structure could easily be fitted

into ab initio models built based on the distance

distribution function calculated from the SAXS data

(Fig 3B,C) Similar results were also obtained in the

absence of calcium, indicating that the major form

of CRMP-2 is a tetramer also in the absence of

Ca2+, in the high protein concentration conditions

employed (10 mgÆmL)1) However, when calculating a

theoretical scattering curve based on the crystal

struc-ture, the fit between the measured data and the crystal

structure is better in the presence of calcium (Fig 3A), indicating a possible heterogeneity in the sample with-out calcium Indeed, the software oligomer could fit the data without calcium better when given a combina-tion of calculated scattering curves for a tetramer and a dimer or a monomer This was not true for the sample containing CaCl2 (data not shown) In line with this,

ab initio model building with dammin resulted in a better fit to the crystal structure for the sample in the presence of CaCl2, indicating the possibility of subtle calcium-dependent movements of the subunits,

or the presence of dimeric or monomeric forms of CRMP-2 in the sample without calcium These observations could also explain the success of crystallization experiments only in the presence of Ca2+

or Mg2+

Analysis of ion binding by surface plasmon resonance

To obtain an idea of the extent and affinity of diva-lent cation binding by CRMP-2, we carried out a surface plasmon resonance experiment where

CRMP-2 was immobilized and incubated in the presence of different concentrations of MgCl2, CaCl2, BaCl2 and KCl (Fig 4) The results indicate that the divalent cations produce a similar strong response, whereas potassium chloride only gives a weak signal The observed strong signal suggests the presence of many binding sites, the binding involving hydrated ions and⁄ or ordering of solvent on the protein surface The estimated overall Kd values for the binding of the cations onto the CRMP-2 protein surface are 15–20 mm for the divalent cations and > 100 mm for potassium

The detection of several isoforms of CRMP-2

in rat hippocampus using 2D electrophoresis and MS

When identifying all the protein spots from rat hippo-campus on 2D gels [30], 26 spots were identified as CRMP-2 (Fig 5) To shed light on the post-transla-tional modifications present in CRMP-2, all these spots were picked, digested with trypsin and analysed

by nano-LC-ESI-MS⁄ MS The results are summarized

in the Supporting information (Table S1) It is noteworthy that the deamidation of Asn356 was detected in eight spots, and that oxidation of methio-nines 64, 152, 168, 362, 375 and 437 was detected in at least five spots each We also detected phosphorylation

of Thr509 in three spots (one of these spots also showed phosphorylation of Ser522)

Table 3 SAXS results The calculated values were obtained from

samples of 10 mgÆmL)1, and the crystal structure values for Rg,

excluded volume and D max are as given by CRYSOL The theoretical

I(0) was calculated for the crystal structure based on the I(0) of

BSA The SAXS excluded volumes are the average volumes of the

ab initio models obtained from DAMMIN

D max (nm)

Excluded volume (nm 3 )

Molecular mass (kDa) CRMP-2 +

CaCl2

3.65 ± 0.001 716 ± 0.4 10.5 310 191

CRMP-2 3.75 ± 0.004 708 ± 0.4 11.5 284 189

Crystal

structure

(homotetramer)

Mapping of the post-translational modifications

onto the crystal structure of CRMP-2

A number of different post-translational modifications

were identified for the various isoforms detected for

adult rat brain hippocampal CRMP-2 (see above)

These modifications were mapped onto the 3D

struc-ture of human CRMP-2 (Fig 6) The sequence identity

between human and rat CRMP-2 is very high, with

only seven residues out of 572 being nonconserved,

and all the detected modification sites being conserved

All the post-translational modifications are located

outside the core of the (ba)8 barrel fold, mainly in

looplike structures on the protein surface The sites are

concentrated on specific regions in 3D space, especially

when considering the sites that were detected in several spots of the 2D gel (see Discussion)

Discussion

The effects of divalent cations on the stability and oligomeric status of CRMP-2

CRMP-2 forms homotetramers [5,11] and, in the present study, these assemblies are demonstrated to be stabilized

A

B

Fig 4 Surface plasmon resonance analysis of ion binding by

CRMP-2 (A) Kinetic analysis of Ca 2+ binding by immobilized

CRMP-2 For clarity, only injections with 1, 5, 10 and 50 m M CaCl 2

are shown (B) A comparison of the responses obtained with

differ-ent ions at 20 m M concentration Ca 2+ , red; Mg 2+ , blue; Ba 2+ , light

blue; K+, orange.

Fig 5 2D gel electrophoretic analysis of rat brain hippocampal proteins; spots identified as CRMP-2 are highlighted The details on the selected spots are given in Table S1 In the inset, the region containing the CRMP-2 spots is shown in more detail.

Fig 6 Mapping of post-translational modifications onto the 3D structure of CRMP-2 In this stereo view, the major oxidation sites are shown in magenta, minor oxidation sites are shown in yellow, major deamidation sites are shown in green, and minor deamida-tion sites are shown in blue.

by divalent cations The heat stability of CRMP-2 was

shown, by two independent methods, to decrease

signifi-cantly when the divalent cation concentration was raised

from 20 to 200 mm This indicates that the effect on

CRMP-2 stability is dual: at low concentration, both

Ca2+and Mg2+stabilize the protein, whereas, at higher

concentrations, the effect is the opposite

Divalent cations have previously been shown to

similarly affect protein stability in a

concentration-dependent manner [33–40] For example, reported

examples include glucose oxidase [35], RNase T1 [36]

and papain [34], and the results can often be explained

by two different modes of binding of divalent cations

to proteins At lower concentrations, the hydrated ions

act mainly to order and stabilize the solvent shell

around the protein molecule At a higher

concen-tration, the ions will bind directly to the protein

surface, resulting in destabilization

The structure of CRMP-2

The crystal structure of human CRMP-2 was refined at

1.9 A˚ resolution, and the overall structure is very

simi-lar to the earlier lower resolution structure obtained in

the presence of calcium [5], also demonstrating the same

oligomeric assembly into a homotetramer We have also

shown by SAXS studies that the homotetrameric

struc-ture of CRMP-2 is maintained in solution, especially in

the presence of calcium The good fit between our

experimental solution and crystal structures indicates

that the crystal structure is an accurate representation

of the CRMP-2 tetramer in solution Moreover, the

conformation obtained by ab initio modelling resembles

the crystal structure more closely in the presence of

calcium than in its absence This hints at the possibility

of subtle conformational changes within the tetramer

that can be induced by divalent cations In principle, in

the presence of calcium, the CRMP-2 tetramer appears

to be slightly more compact than in its absence, as

demonstrated by the smaller Rgand Dmaxvalues for the

Ca2+-containing sample in conjunction with the I(0)

and excluded volumes, which are higher (Table 3)

These differences could also be caused, at least partly,

by the presence of small amounts of dimeric CRMP-2

in the absence of calcium

Using CD spectroscopy, we have shown that

CRMP-2 undergoes a transition to a b-aggregate upon

heating in the absence of divalent cations

Interest-ingly, CRMP-2 has been characterized as a constituent

of the paired helical filaments of neurofibrillary tangles

in Alzheimer’s disease [41,42] Future studies should

aim to investigate any possible relationships between

the folding and denaturation behaviour of CRMP-2

and neuronal fibril formation in Alzheimer’s disease brains

Post-translational modifications and previously characterized mutations of CRMP-2

A number of post-translational modifications were detected in CRMP-2 from rat brain, including several sites of deamidation, oxidation and phosphorylation (Table S1) Although both of the phosphorylation sites that were detected (i.e Thr509 and Ser522) were in the C-terminal tail region, which is predicted to be struc-turally disordered, most of the other modifications could be mapped onto the 3D structure Thr509 and Ser522 have been both characterized as major phos-phorylation sites in CRMP-2 [9,42] The reason why some of the other previously characterized phosphory-lation sites on the C-terminal tail were not detected could be due to the fact that hippocampal tissue from

an adult animal was used This aspect, however, was not investigated further in the present study because

we were specifically interested in the post-translational modifications that could be mapped onto the folded domain of CRMP-2 Previously, modifications of residues within the folded region have not been charac-terized in detail, except for the detection of one phos-phorylation site at Ser465 [43]

Three ‘hotspot’ regions for post-translational modifi-cations can be visualized in the 3D structure of

CRMP-2 (Fig 6) The first concerns the main deamidation site and four of the main oxidation sites The second one contains the remaining two main oxidation sites and two minor deamidation sites, and the third one a minor deamidation site and three minor oxidation sites The reasons for the concentration of the modifications to these areas in 3D space are currently not known, but they could include the accessibility of these areas towards modifications, their mobility and the relation

of such modifications towards CRMP-2 function

In light of the finding that CRMP-2 is one of the neuronal proteins that accumulate high levels of iso-aspartate [23], it is interesting to observe a number

of deamidated asparagine residues in CRMP-2 The formation of isoaspartate is an important source of protein damage under physiological conditions, and is linked to the deamidation of Asn residues The deami-dation of Asn356, detected in eight of the 26 character-ized CRMP-2 spots, could be related to isoaspartate accumulation on brain CRMP-2

The e204 mutation in the Caenorhabditis elegans unc-33 gene [44] results in a protein where a conserved aspartate residue is mutated into an asparagine; the corresponding residue in CRMP-2 is Asp71 [44] In a

yeast two-hybrid assay, the mutation prevented unc-33

oligomerization [45] The mutant was originally

isolated by identifying its involvement in paralysis

resulting from defective axon growth [44], and similar

effects on axonal growth have subsequently been

found with the corresponding D71N mutant of

CRMP-2 [7] In the 3D structure, Asp71 is buried

within the folded protein, close to the interface

between the small and large lobes of CRMP-2, but far

from the tetramerization interfaces Its side chain is

buried, being close to that of Arg361, but no hydrogen

bond⁄ salt bridge is formed between the two side

chains These two buried charges apparently neutralize

each other and, in the D71N mutant, the charged

Arg361 is expected to have an isolated buried charge,

which may destabilize the protein

Conclusions

CRMP-2 is a central regulator of axonal growth cone

guidance, and its function is likely to involve both

homo- and heterotetramerization, as well as

post-trans-lational modifications Using a number of biochemical

and biophysical methods, we have gained a more

detailed view than was previously available for the

structure and properties of the CRMP-2 protein, both

in solution and in the crystal state Divalent cations

have a drastic effect on both the stability and

oligo-meric state of CRMP-2, and a number of different

post-translationally modified isoforms of CRMP-2

could be identified in the brain Our data, including

the mapping of post-translational modifications onto

the structure of CRMP-2, open up new possibilities for

studying the function and interactions of CRMP-2,

which still remain enigmatic

Experimental procedures

Protein purification

The purification of human CRMP-2 has been described

previously [5] The protein batches for the present study

were obtained from the SGC Stockholm laboratory A

con-struct containing residues 1–490 was used for the surface

plasmon resonance experiments, and other experiments

were carried out with a construct containing residues

13–490 of CRMP-2 The His tags were not removed prior

to any experiments

CD spectroscopy

CRMP-2 was extensively dialysed against 10 mm Hepes

(pH 7.3); subsequent dilutions were made in the dialysis

buffer, which was also used for the measurement of a buffer control by CD CD spectra were measured in the wavelength range 195–250 nm, on a Jasco J-810 spectro-polarimeter (Jasco, Tokyo, Japan) The protein concen-tration was 3 lm, and a 1 mm cuvette was used After measuring each spectrum at 23.4C, a temperature scan at

a fixed wavelength of 220 nm was run between 30 and

70C to obtain a melting curve After the temperature scan, CD spectra were further recorded for the samples, both at 90C and after cooling back to room temperature The effects of the following additives on the behaviour of CRMP-2 were studied: 20 or 200 mm CaCl2, 20 or 200 mm MgCl2, and 50 mm NaCl CD spectra and a temperature scan were recorded using exactly the same parameters and procedures for all samples Of note, the sample gained a gel-like appearance upon heating in the absence of divalent cations, but not in their presence

Stability analysis by the Thermofluor method The thermal stability of CRMP-2 was analysed in 96-well format by following the fluorescence from SYPRO Orange (Invitrogen, Carlsbad, CA, USA) as a function of tempera-ture (i.e the so-called Thermofluor or thermal shift assay method) [46,47] The experiment was carried out using a

7500 Real Time PCR System apparatus (Applied Biosys-tems, Foster City, CA, USA) and the temperature was scanned from 20 to 90C with 1 C increments, with moni-toring of fluorescence at 542 nm Eight replicates each from

12 different conditions were randomly placed on the 96-well PCR plate Occasionally, curves with abnormal shapes were observed; such curves were excluded from the analysis, most likely resulting from incomplete sealing of an individ-ual well on the 96-well PCR plate

Size exclusion chromatography The oligomeric status of CRMP-2 was analysed by gel fil-tration on a Superdex 200 HR 10⁄ 30 column coupled to an A¨KTApurifier (GE Healthcare, Uppsala, Sweden) An iden-tical sample (0.67 mgÆmL)1 of 200 lL) was run with the same protocol in 10 mm Hepes (pH 7.5), 100 mm NaCl, and in the same buffer containing either 20 mm CaCl2 or

20 mm MgCl2 Sample elution was followed at 280 nm Molecular masses were estimated by comparing the sample elution volumes with those observed for standard proteins run on the same column

Crystallization, data collection and structure solution

Human CRMP-2 was crystallized essentially as described previously [5], while simultaneously screening for substitutes for CaCl2that could promote crystallization In the absence

of CaCl2, the only tested additive producing crystals was

MgCl2 Crystals (crystal form 1) without CaCl2were grown

at 20C in hanging drops over a well solution consisting of

15% poly(ethylene glycol) 10 000, 0.1 m Tris (pH 9), 0.2 m

MgCl2 and 10 mm MgI2 Another crystal form (crystal

form 2) was similarly grown in the absence of MgI2

(pH 8.5)

Prior to data collection, the crystals were briefly soaked

in a cryoprotectant solution (well solution supplemented

with 20% glycerol) and cooled to 100 K in a stream of

gaseous nitrogen Diffraction data were collected at

MAX-Lab (Lund, Sweden) beamline I911-2 [48] Data processing

was performed using xds [49] and xdsi [50] An analysis

of the data revealed the presence of pseudotranslational

symmetry in crystal form 1 (26.3%, fractional translation

0,0.175,0.5) However, space group determination for

crystal form 2 was not straightforward, and this

mono-clinic crystal was found to be pseudomerohedrally

twinned; for this reason, crystal form 2 was discarded

from the analysis

The previously determined structure of human CRMP-2

[5] was used as a template in molecular replacement Four

monomers of CRMP-2 were found within the asymmetric

unit using molrep [51], and refinement and model building

were carried out iteratively using refmac [52], phenix

refine[53,54] and coot [55] The coordinates and structure

factors were deposited in the Protein Data Bank under the

accession code 2VM8

Surface plasmon resonance

The binding of Ca2+, Mg2+, Ba2+ and K+ by CRMP-2

was analysed by surface plasmon resonance on a Biacore

3000 apparatus (Biacore AB, Uppsala, Sweden) by

immo-bilizing CRMP-2 on a CM5 chip and passing 0, 1, 5, 10,

20, 50 and 100 mm solutions of CaCl2, MgCl2, BaCl2

and KCl over the chip In the immobilization, 10 mm

socium acetate buffer (pH 4.5) was used During the

binding experiment, the running buffer contained 10 mm

Hepes (pH 7.5), 100 mm NaCl and 0.004% surfactant

P20 (Biacore AB), in addition to the salts being tested

The experiments were carried out at 25C, with a flow

rate of 30 lLÆmin)1 For regeneration of the surface

between injections, the ions were allowed to dissociate

freely into the binding buffer A control channel on

the chip was similarly treated, with the exception that no

protein was immobilized onto it The data were fitted

against a 1 : 1 binding model using biaevaluation

software (Biacore AB)

Small-angle X-ray scattering

For SAXS, CRMP-2 was dialysed into a buffer

contain-ing 10 mm Hepes (pH 7.5) and 100 mm NaCl SAXS

data for CRMP-2 in the presence and absence of 20 mm

CaCl2, at concentrations in the range 1–10 mgÆmL)1, were measured on the EMBL Hamburg⁄ DESY beamline X33, and the corresponding buffer was always used for a blank experiment Programs from the atsas software package [56] were used for data analysis, essentially as described previously [57] The measured data were further processed using primus [58] The molecular mass was estimated by comparing the forward scattering I(0) with that of a standard solution of BSA The distance distri-butions were obtained using gnom [59] and further used for ab initio modelling in dammin [60] An averaged model was generated from several runs using damaver [61], and the SAXS model and the crystal structure were superimposed with supcomb [62] The possible oligomeric assemblies were also studied using oligomer [58], after evaluating the solution scattering of each possible component using crysol [63]

2D electrophoresis and MS Materials

Immobilized pH gradient strips and buffers were purchased from Amersham Biosciences, a part of GE Healthcare (Mil-waukee, WI, USA) Reagents for polyacrylamide gel prepa-ration were purchased from Bio-Rad Laboratories (Hercules, CA, USA) Chaps was obtained from Roche Diagnostics (Mannheim, Germany), urea was obtained from AppliChem (Darmstadt, Germany), thiourea was ob-tained from Fluka (Buchs, Switzerland), 1,4-dithioerythritol and EDTA were obtained from Merck (Darmstadt, Germany) and tributylphosphine was obtained from Pierce Biotechnology (Rockford, IL, USA)

Sample preparation Twenty-three-month-old rat hippocampus tissue was pow-derized in liquid nitrogen and suspended in 2 mL of sam-ple buffer [20 mm Tris, 7 m urea, 2 m thiourea, 4% Chaps, 10 mm 1,4-dithioerythritol, 1 mm EDTA and

1 mm phenylmethanesulfonyl fluoride, containing 1 tablet Complete (Roche Diagnostics) and 0.2% (v ⁄ v) phos-phatase inhibitor cocktail (Calbiochem, San Diego, CA, USA)] The suspension was sonicated for approximately

30 s and centrifuged at 15 000 g for 60 min at 12C Desalting was performed with an Ultrafree-4 centrifugal filter unit (Millipore, Bedford, MA, USA), with a molec-ular mass cut-off of 10 kDa The protein concentration

of the supernatant was determined by the Bradford assay

2D gel electrophoresis Samples of 750 lg of protein were applied on immobilized nonlinear pH gradient (pH 3–10) strips Focusing started at