Crystal structure of arabidopsis thaliana cyclophilin 38 (atcyp38 2

4.2 EXPRESSION AND PURIFICATION OF SELENOMETHIONYLATED WILD TYPE ATCYP38 4.2.1 Expression Feedback inhibition of methionine biosynthesis Doublie, 1997 at the protein expression stage

CHAPTER 4 RESULTS AND DISCUSSION

4.1 EXPRESSION AND PURIFICATION OF NATIVE WILD TYPE

Figure 19 SDS-PAGE showing the expression of soluble wt

AtCyP38 M – Low Molecular Weight Marker, Lane 1 – Cell

lysate before induction, Lane 2 – Whole cell lysate 4hrs after

induction, Lane 3 – Soluble protein 4 hrs after induction

4.1.2 Affinity chromatographic purification

The produced AtCyP38-GST fusion protein was purified using sepharose beads (Amersham Biosciences), Fig 20 Due to the slow kinetics of GST

glutathione-through Hence the flow through from the first step was subjected to a second round

of purification process to retrieve the unbound fusion protein

Figure 20 SDS-PAGE showing the affinity chromatographic

purification of the wt AtCyP38-GST fusion protein M – Low

Molecular Weight Marker, Lane 1 – Soluble protein before

affinity chromatography, Lane 2 – Flow through from affinity

chromatography column, Lane 3 – Wash 1, Lane 4 – Wash 2,

Lane 5 – Protein bound to glutathione sepharose beads

to the linker that connects the tag and the protein, at its N-terminus The presence of this linker with the sequence ‘GSPGISGGGGGILL’ is a peculiar feature of the pGEX-KG vector, which was used for protein expression This additional glycine-rich region is expected to improve the thrombin cleavage and probably favored the complete cleavage However, probably due to these extra residues from the linker, the

band of the tag-removed protein appears at a slightly higher molecular weight than its original 39 kDa weight in SDS-PAGE, Fig 21

14.4 kDa

Figure 21 SDS-PAGE showing the thrombin cleavage of the wt

AtCyP38-GST fusion protein M – Low Molecular Weight Marker,

Lane 1 – Protein on resin before cleavage, Lane 2 – Flow through from

the column, after cleavage, Lane 3 – Wash 1, Lane 4 – Wash 2, Lane 5

– GST, bound to the glutathione sepharose beads after cleavage

The cleaved AtCyP38 protein came out along with the flow through and the first wash But it still had some amount of contaminant proteins The two fractions were pooled together and subjected to another round of purification by size exclusion chromatography

4.1.4 Size exclusion chromatography

HiLoad 16/60 Superdex-75 column (Amersham Biosciences) was used for the size exclusion chromatographic experiment Pure AtCyp38 protein was eluted out as a single peak, Fig 22, and the corresponding fractions were pooled together SDS-PAGE of the pooled fractions, Fig 23, had no other contaminant proteins

Figure 22 Profile of size-exclusion chromatographic purification of

wt AtCyP38 (83-437) The higher peak at around 60 ml corresponds to

Figure 23 SDS-PAGE of size-exclusion chromatography fraction of

purified AtCyP38 (83-437) M – Low Molecular Weight Marker,

Lane 1 – Protein before size-exclusion chromatography, Lane 2 –

Pooled fractions of the protein after size-exclusion chromatography

4.1.5 Analyses for purity and homogeneity

The position of the peak from size-exclusion chromatography indicated the protein to be a monomer of about 40 kDa size, Fig 18 This was confirmed on

Native-PAGE, Fig 24, and the protein appeared as a single band that corresponded to

Figure 24 Native-PAGE of purified wt AtCyP38 (83-437) M –

High Molecular Weight Native Marker, Lane 1 – Purified wt

AtCyP38

The theoretical molecular weight of wt AtCyP38 (83-437) protein is 39.25 kDa MALDI-TOF mass spectrometry of the recombinant wt AtCyP38, carried out using an Applied Biosystems 4700 Proteomics Analyzer 86, showed a mass of about 40389.69 + 1.217, Fig 25 The additional stretch of 14 residues at the N-terminus of the protein makes its theoretical molecular weight 40.38 kDa The molecular weight determined by mass spectrometry is quite close to this value

A Dynamic Light Scattering (DLS) experiment showed a dispersity index

of 0.16 and this assured that the protein was monodispersed The molecular weight predicted by DLS was 39.6 kDa, which again is quite close to the actual molecular weight and hence the protein was confirmed to be a monomer

40389.69

Figure 25 Mass spectrometry for wt AtCyP38 (83-437)

4.2 EXPRESSION AND PURIFICATION OF SELENOMETHIONYLATED

WILD TYPE ATCYP38

4.2.1 Expression

Feedback inhibition of methionine biosynthesis (Doublie, 1997) at the protein expression stage was used for the incorporation of selenomethionine into the AtCyP38 protein The expression of selenomethionylated protein in the minimal medium M9 resulted in slightly lower yield compared to the native protein which was expressed in nutrient-rich LB medium

4.2.2 Purification

2 mM DTT was maintained throughout in all the buffers to avoid the oxidation

of selenomethionine during the purification process The protocol that was used for native AtCyP38 purification was used for the purification of selenomethionylated

AtCyP38 MALDI-TOF analysis on the derivatized protein gave a mass of 40839.95 + 1.205 The mass difference between the native and selenomethionylated proteins indicates that 9 out of the 10 methionine residues of AtCyp38 have been replaced by selenomethionine

4.3 CRYSTALLIZATION AND DATA COLLECTION FOR WILD TYPE

ATCYP38

4.3.1 Crystallization

Initial crystallization screening of the native AtCyP38 protein by the hanging drop vapor diffusion method yielded crystals in two different conditions within 2 days Poly ethylene glycol (PEG) was the common precipitant in both the conditions and in addition to PEG, the conditions also had a volatile precipitant solution One condition had 10% isopropanol and the other condition had 2.5% t-butanol The presence of these volatile precipitants was indispensable for crystal formation However, their volatile nature caused disintegration of the crystals as soon as the cover-slip was opened Optimization of these hanging drop conditions did not show any improvement

The vapor batch method of crystallization was attempted in order to avoid the problem of excessive evaporation of volatile precipitants, as suggested by Mortuza and co-workers (2004) The volatile precipitant was provided in the reservoir and was allowed to diffuse slowly into the drops that were overlaid with paraffin oil Crystal formation took slightly longer time than in the hanging drop method, probably due to the slow diffusion rate Crystals were larger in size, Fig 26, and fewer in number The presence of oil prevented the fast evaporation rate of volatile precipitants out of the

drop when the cover was opened The paraffin oil on the top of the drop also served as

an additional cryo-protectant during the cryo-cooling of crystals for data collection

Figure 26 Crystal of AtCyP38 obtained by vapor batch method in a

condition having PEG6000 and t-butanol as precipitants

Figure 27 Crystal of selenomethionylated AtCyP38 obtained by vapor

batch method in a condition having PEG6000 and t-butanol as

precipitants

Crystals of selenomethionylated AtCyP38 protein was obtained in the same condition as that of native AtCyP38 by the vapor batch method, Fig 27 The native protein was required at an initial concentration of 5 mg ml-1 for crystallization, whereas, the selenomethionylated protein was required only at 3 mg ml-1, probably due to its lower solubility

A few large crystals were picked-up, washed thoroughly in the crystallization buffer, dissolved in water and subjected to MALDI-TOF mass spectrometric analysis The determined mass for the native and selenomethionylated crystals were the same

as that of the corresponding purified protein, indicating that the proteins are intact in the crystal and no part of them has been cleaved off during the crystallization process

Crystals were tested at an in-house X-ray imageplate detector facility for diffraction quality and optimization of cryo-condition Cryo-conditions with glycerol concentration less than 20% always resulted in the formation of ice-rings in diffraction images Suitable cryo-protecting conditions were identified for crystals obtained from both the conditions Essentially, the chosen cryo-conditions had 5% additional PEG in addition to 20-25% glycerol

Crystals from the vapor batch droplets were carefully picked-up with loops and immediately transferred to the respective cryo-protectant Crystals were either directly tested on the in-house system or flash-cooled in liquid nitrogen for later experiments The native crystals from the condition containing PEG 6000 and t-butanol as precipitants diffracted up to 3 Å whereas those from the condition containing PEG 4000 and isopropanol diffracted only up to 4 Å The diffraction quality of all selenomethionylated crystals was poor But a few of these crystals from the condition containing PEG 6000 and t-butanol diffracted up to about 4 Å

cryo-4.3.2 Data collection and analysis

Data collection was done at the X25 beamline, National Synchrotron Light Source, Brookhaven National Laboratory (Upton, NY, USA) with a Q315 charge-coupled device detector (Area Detector Systems Corporation) Prior to data collection,

a fluorescence scan was carried out to identify the peak, remote and inflection

wavelengths for MAD dataset based on selenium absorption spectrum 360 frames of the native and 360 frames for each of the 3 MAD wavelengths were collected with an oscillation one degree for the respective crystals The crystal parameters and data

collection statistics are given in Table 1 below

Table 1 Crystal parameters and data-collection statistics for wt

AtCyP38 Values in parentheses are for the highest resolution shell

(Native: 2.59-2.50 Å and Se-Met: 3.63-3.50 Å)

Completeness (%) 97.8 (99.8) 98.4 (93.3) 98.2 (91.9) 98.3 (92.3) Redundancy 6.7 (6.7) 6.3 (6.0) 6.2 (5.7) 6.1 (5.6)

1Rsym 0.071 (0.29) 0.059 (0.09) 0.065 (0.10) 0.050 (0.07)

〈I/σ(I)〉 21.3 (5.4) 30.2 (18.3) 30.7 (17.5) 20.5 (12.1)

1Rsym = ∑hkl∑i [|I i (hkl) - <I(hkl)>| / ∑ hkl∑i I i (hkl)]

All data sets were processed and scaled using the HKL2000 (Otwinowski and

Minor, 1997) program package The Matthew’s coefficient (Matthews, 1968) of the crystal indicated a solvent content of 57% and the asymmetric unit contained one molecule The structure was determined by the three-wavelength (Multi-wavelength) Anomalous Dispersion (MAD) method using the selenomethionine data Even though mass-spectrometric analysis indicated that 9 out of the 10 methionine residues of the

protein had been replaced by selenomethionine, the BnP program (Weeks et al., 2002),

which was used to solve the seleniumsubstructure, found only seven selenium sites The sites were confirmedby comparison with the anomalous difference Patterson map

Automated model building with the use of the ARP/wARP program (Perrakis et al.,

1999) was not very successful, except for tracing a few short stretches in the

C-terminal region

The first electron density map that was calculated from the MADphases was clean and well interpretable The overall structure could very well be seen, wherein the N-terminus had a bundle of 5 helices and the C-terminus had the typical β-barrel cyclophilin domain, as expected from bioinformatics predictions However, the early maps did not permit rapid model building Certain loop regions connecting the five helices of the N-terminus were missing The first three helices of the N-terminus had

no long or bulky aromatic side-chain residues that could help in identifying the region and the missing density in the connecting loop regions made it all the more difficult Also, the extreme N-terminus was not quite clear in the density which was later found

to form part of the C-terminal cyclophilin β-barrel domain In spite of these difficulties, an initial model was built with all available maps in the graphics suite O

(Jones et al., 1991) This partial model was refined using the CNS program (Brunger

et al., 1998) But the R-factor did not change from the high 40’s, even after repeated

attempts of refinement This clearly indicated some serious error in the model which could not be easily identified and rectified due to the low resolution of the maps

4.4 SELECTIVE MUTATION OF RESIDUES TO AID PHASING

The first three helices of the N-terminus did not have any methionine residues, which, if substituted by selenomethionine, can help in correct model building A strategy of selective mutation of a few residues to methionine by site-directed mutagenesis and subsequent selenomethionylation was planned for The choice of residues to be mutated to methionine was mainly based on the studies by Gassner and Matthews (1999) In their studies, leucine was found to be the optimal amino acid to

be substituted by methionine Furthermore, this is in complete agreement with the

ranking suggested by the Dayhoff mutation probability (Dayhoff et al., 1978), i.e by

the frequency of amino-acid substitutions in the sequences of related proteins Amongst closely related proteins, leucine is the most frequent amino acid that is replaced by methionine, without any disruption of local structure and change in the

overall fold (Jones et al., 1992; Leahy et al., 1994) The methionine side chain can, to

a great extent, adopt a conformation so as to occupy the space vacated by leucine Leucine has a side chain volume of 76 Å3, which is the same for methionine It has been shown that up to 10 such mutations to methionine are tolerated without any

change in overall structure or fold for T4 lysozyme (Gassner et al., 2003)

As mentioned earlier, while refining the AtCyP38 structure, assignment of correct sequence was difficult for the first three helices of the N-terminus AtCyP38 has several leucines in this region We decided to mutate five selected leucines in this region to methionine Using the currently available model and secondary structure predictions for the N-terminal helical region, adequate care was taken in the design

that the replacement would not lead to any structural disturbance and hence would not interfere with the growth or quality of crystals The selected leucines were likely not

to occupy internal sites within the protein; thereby disrupting a zipper formation (if any) Leucines 107, 111, 125, 140 and 154 of AtCyP38 (83-437) were chosen to be mutated to methionine Individual replacement mutant genes as well as a mutant with all the 5 leucines mutated to methionine were produced

The mutant genes were inserted between the XbaI/XhoI sites of the pGEX-KG vector The sequence of AtCyP38 (83-437) showing the 10 inherent methionine residues as well as the five mutated residues is given in Fig 28

083 VANPVIPDVS VLISGPPIKD PEALMRYAMP 112

113 IDNKAIREVQ KPMEDITDSL KIAGVKAMDS 142

143 VERNVRQASR TMQQGKSIIV AGFAESKKDH 172

173 GNEMIEKLEA GMQDMLKIVE DRKRDAVAPK 202

203 QKEILKYVGG IEEDMVDGFP YEVPEEYRNM 232

233 PLLKGRASVD MKVKIKDNPN IEDCVFRIVL 262

263 DGYNAPVTAG NFVDLVERHF YDGMEIQRSD 292

293 GFVVQTGDPE GPAEGFIDPS TEKTRTVPLE 322

323 IMVTGEKTPF YGSTLEELGL YKAQVVIPFN 352

353 AFGTMAMARE EFENDSGSSQ VFWLLKESEL 382

383 TPSNSNILDG RYAVFGYVTD NEDFLADLKV 412

413 GDVIESIQVV SGLENLANPS YKIAG 437

Figure 28 The sequence of AtCyP38 (83-437) The native methionine

residues are highlighted in grey and the leucine residues that were

mutated to methionine are highlighted in yellow

4.5 CRYSTALLOGRAPHY OF ATCYP38 MUTANTS

4.5.1 Expression, purification and crystallization

The mutants and selenomethionine derivatives were expressed and purified exactly as wild type AtCyP38 and its selenomethionine derivative, respectively Optimal crystallization conditions varied slightly for the mutant proteins Suitable

cryo-conditions were used and the crystals were tested at the in-house facility for diffraction quality The mutant crystals were found to be of better diffraction quality compared to the wild type On an average, these crystals diffracted to about 2.8 Å at the in-house facility and MAD data collection at the NSLS synchrotron was planned for

MALDI-TOF mass spectrometric analysis of the crystals gave a mass of 40470.24 + 1.06 for the native multiple mutant and 41102.15 + 1.16 for its selenomethionine derivative The mass difference indicated the incorporation of at least 13 selenium atoms The same mass was obtained for the respective proteins before crystallization as well This indicated that the whole protein is present in the crystal and no part of it has got cleaved off during the crystallization process

4.5.2 Data collection and analysis

The native and selenomethionylated crystal data sets for all the five single mutants and the multiple mutant were collected at the X12B beamline, National Synchrotron Light Source, Brookhaven National Laboratory (Upton, NY, USA) with

a Q210 charge-coupled device detector (Brandies) However, the dataset from the multiple mutant alone was used to solve the structure The crystal parameters, data collection and initial refinement statistics are given in Table 2

All data sets were processed and scaled using the HKL2000 program package

The Matthew’s coefficient of the crystal indicated a solvent content of 57% and the asymmetric unit contained only one molecule The structure of the five LÆM mutant protein was attempted to be solved by the three-wavelength method, making use of the selenomethionine data

Table 2 Crystal parameters and data-collection statistics for the

multiple L→M mutant Values in parentheses are for the highest

resolution shell (Native: 2.73-2.64 Å and Se-Met: 2.57-2.46 Å)

1Rsym 0.069 (0.29) 0.059 (0.30) 0.065 (0.25) 0.067 (0.23)

〈I/σ(I)〉 21.40 (5.4) 20.2 (5.3) 20.7 (6.5) 19.21 (5.1)

1Rsym = ∑hkl∑i [|I i (hkl) - <I(hkl)>| / ∑ hkl∑i I i (hkl)]

4.6 STRUCTURE DETERMINATION OF MUTANT ATCYP38

Mass-spectrometric analysis had predicted replacement of 13 out of 15 methionine residues of this mutant protein by selenomethionine However, the BnP program found only twelve selenium sites These sites were manually confirmedby

refinement by density modification and solvent flattening was performed by using the RESOLVE (Terwilliger, 2000) program The 5 refined selenium sites of the mutations L107M, L111M, L125M, L140M and L154M were significantly helpful in model building

Due to the higher resolution limits, the electron density map calculated from the combinedphases was much better than that of the wild type protein The native data for the mutant was of lower resolution (2.64 Å) compared to that of the derivative (2.46 Å) and hence was not used at all The remote data from the derivative was used for all subsequent map calculations and refinement steps The maps were very clear at the loop regions that connect the N-terminal helices Also, the extreme N-terminus (residues 83-99) was clearly modeled and found to form part of the C-terminal cyclophilin β-barrel domain However, density for two long loop regions within the C-terminal cyclophilin domain was not clearly observed

The initial R-factor was 0.43 However, ARP/wARP could build only a few short stretches and did not dock any sequence Manual model building was performed using the program O Crystallographic refinement was carried out with the program

CNS as well as Refmac5 of the CCP4 suite (Murshudov et al., 1997) Iteratively, 2Fo

-Fc and Fo-Fc maps were computed to check any misfit in the model Temperature factors were refined isotropically in the later stages of refinement and water was picked up using the Water-Pick program from CNS based on the Fo-Fc map at the 3.0

σ level and was checked with the 2Fo-Fc map at the 2.0 σ level

The final R-factor and Rfree (with 10% of reflections that were not included for refinement) were 0.22 and 0.28, respectively using reflections with |F| > 3.0 σ(|F|) The geometry of the final model was checked with the program PROCHECK

(Laskowski et al., 1993) All parameters were within acceptable ranges, except for a

few residues in the two long loop regions which were found to be in the disallowed region of Ramachandran plot These long loops were quite flexible, with high B-factors All figures were prepared using the PYMOL (DeLano, 2002) program A summary of the refinement statistics is shown in Table 3

• Protein (with two loops removed) 54.86 (2186 atoms)

1R-factor = ∑ hkl ||F o (hkl)| – |F c (hkl)|| / ∑ hkl |F o (hkl)|

4.7 THREE-DIMENSIONAL STRUCTURE OF ATCYP38 (83-437)

The secondary structural organization of AtCyP38 is illustrated in Fig 28

AtCyP38 (83-437) has two distinct domains as seen from its crystal structure The

N-terminus is a helical domain made up of 5 helices (α1 – α5) of varying lengths This

domain is followed by a typical cyclophilin domain, made up of an eight-stranded

β-barrel that is capped by one α-helix at each end There are differences in the

cyclophilin domain of AtCyP38 and other known cyclophilin structures (section

4.9.2) The helical domain and the cyclophilin domain of the structure are connected

Figure 29 Secondary structural organization of AtCyP38 β-strands

are shown as yellow arrows and α-helices are shown as red cylinders

The two β-strands from the N-terminus which form part of the β-barrel