Transcript 3D Structure - Canadian Bioinformatics Workshops

Protein Expression,
Structural Proteomics &
Bioinformatics
David Wishart
University of Alberta
Edmonton, AB
[email protected]
Lecture 3.0
1
Expression Questions
•
•
•
•
•
•
•
•
Which host cell system?
Which expression vector?
Which cloning/expression protocols?
Is it membrane or water soluble?
Is it single domain or multi-domain?
How soluble and how stable?
Where will this protein be found?
How to purify & how to identify?
Lecture 3.0
2
Host Cell System?
•
•
•
•
•
•
•
•
•
Escherichia coli
Other bacteria
Pichia pastoris
Other yeast
Baculovirus
Animal cell culture
Plants
Sheep/cows/humans
Cell free
Lecture 3.0
Polyhedra
3
Host Cell System?
• Choice depends on size and character of
protein
– Large proteins (>100 kD)? Choose eukaryote
– Small proteins (<30 kD)? Choose prokaryote
– Glycosylation essential? Choose baculovirus
or mammalian cell culture
– Isotopic labelling esential? Choose E. coli
– Post-translational modifications essential?
Choose yeast, baculovirus or other eukaryote
Lecture 3.0
4
Host Cell System?
• Try different hosts when optimizing
expression (protease negative, strains
with enhanced expression of rare tRNAs)
• Expression levels can vary by a factor of
10 or more depending on strain choice
• Example E. coli strains
– MC1061, UT580, GM48, JM101, DH5, MG1065,
NM522, MC4100, TOP10F’, BL21(DE3) BL21CodonPlus (DE3)
Lecture 3.0
5
Codon Bias
http://www.kazusa.or.jp/codon/
Lecture 3.0
6
Arginine Codon Bias
E. coli
AGA 2.7
AGG 1.6
Eubacteria
(rare)
Lecture 3.0
M. jannaschii
AGA 27.5
AGG 9.9
Archaebacteria
(abundant)
H. sapiens
AGA 11.2
AGG 11.1
Eukaryote
(normal)
7
Host Cell System?
• American Type Culture Collection
– http://www.atcc.org
• Clontech Cell Lines
– http://www.clontech.com
• Stratagene Cells (BL21)
– http://stratagene.com
• Invitrogen Cell Lines (Pichia)
– http://www.invitrogen.com
Lecture 3.0
8
Fermentor or Shake Flask?
Lecture 3.0
9
Media Optimization
• Still using L-broth? Try using T-broth
– Tryptone - 12 g, Yeast Extract - 24 g, glycerol - 4
ml, KH2PO4 - 2.3g, K2HPO4 - 12.5g
• Extra Spicy Media
– More ATP: 10 ml/L glycerol + 10g glucose/L
– More AA: Add 10g casamino acids + 10mg L-Trp
• Add more media (30%) when you induce
• Add more antibiotic when you induce
– prevents overgrowth by cells that lost plasmid
Lecture 3.0
10
Expression Questions
•
•
•
•
•
•
•
•
Which host cell system?
Which expression vector?
Which cloning/expression protocols?
Is it membrane or water soluble?
Is it single domain or multi-domain?
How soluble and how stable?
Where will this protein be found?
How to purify & how to identify?
Lecture 3.0
11
Which Vector?
• Must be compatible with host cell system
(prokaryotic vectors for prokaryotic cells,
eukaryotic vectors for eukaryotic cells)
• Needs a good combination of
–
–
–
–
–
strong promoters
ribosome binding sites
termination sequences
affinity tag or solubilization sequences
multi-enzyme restriction site
Lecture 3.0
12
Which Vector?
• Promoters
– arabinose systems (pBAD), phage T7 (pET),
Trc/Tac promoters, phage lambda PL or PR
• Tags
–
–
–
–
–
–
–
His6 for metal affinity chromatography (Ni)
FLAG epitope tage DYKDDDDK
CBP-calmodulin binding peptide (26 residues)
E-coil/K-coil tags (poly E35 or poly K35)
c-myc epitope tag EQKLISEEDL
Glutathione-S-transferase (GST) tags
Cellulose binding domain (CBD) tags
Lecture 3.0
13
Which Vector?
• VectorDB
– http://vectordb.atcg.com
• Invitrogen Vectors
– http://www.invitrogen.com/vectors.html
• Qiagen Vectors
– http://www.qiagen.com/literature/vectors.asp
• Stratagene Vectors
– http://stratagene.com/vectors/vectors.htm
Lecture 3.0
14
Cloning Software
•
•
•
•
•
•
•
MacVector (Accelrys)
SimVector (Premier BioSoft)
GeneTool (BioTools)
Vector NTI (Informax/Invitrogen)
DNAStrider
LaserGene (DNAStar)
PlasMapper (Bioinformatics Help Desk)
Lecture 3.0
15
PlasMapper
http://wishart.biology.ualberta.ca/PlasMapper/
Lecture 3.0
16
How to Clone?
Echo Cloning
Lecture 3.0
17
How to Clone?
Yeast Cells
Lecture 3.0
18
How to Clone?
Mammalian Cells
Lecture 3.0
19
Gateway System (Invitrogen)
• No need to design, construct or ID unique
restriction sites
• Uses lambda phage site-specific
recombination for gene/plasmid integration
• No need for restriction enzyme digestions
• No need for gel fragment separation and
purification
• Ideal for high throughput proteomics efforts
Lecture 3.0
20
Gateway System (Invitrogen)
+
Entry
Vector
Entry
Clone
PCR product
X
Desired
Clone
Destination
Vector
Lecture 3.0
21
Gateway System (Invitrogen)
Gene
attR1
attL2
attL1
Entry
Clone
Kmr
Gene
Desired
Clone
Ampr
Lecture 3.0
-ve selector (anti-gyrase)
attR2
Destination
Vector
+
Int
IHF
Xis
Ampr
-ve selector (anti-gyrase)
Dead-end
Clone
Kmr
22
Gateway Protocol
• Mix and incubate
for 60’ @ 25 oC
Clonase reaction buffer 4 mL • Add proteinase K
and incubate for 10’
Destination Vector
300 ng
at 37 oC
Entry Clone
100 ng
• Transfer to E. coli
Clonase Enzyme mix
4 mL
(competent) DH5
Total volume
20 mL
cells
• Express for 60’ and
plate on LB-Amp
Ingredients
•
•
•
•
•
Lecture 3.0
23
Expression/Cloning -Which Protocols?
• Molecular Cloning 3rd Edition (Sambrook and
Maniatis / Russell)
– http://www.molecularcloning.com
• Molecular Biology Protocols
– http://micro.nwfsc.noaa.gov/protocols/
• Molecular Biology Shortcuts
– http://highveld.com/f/fprotocols.html
• Protocols Online
– http://www.protocol-online.org/
Lecture 3.0
24
Expression Questions
•
•
•
•
•
•
•
•
Which host cell system?
Which expression vector?
Which cloning/expression protocols?
Is it membrane or water soluble?
Is it single domain or multi-domain?
How soluble and how stable?
Where will this protein be found?
How to purify & how to identify?
Lecture 3.0
25
Membrane or Water Soluble?
Lecture 3.0
26
Membrane or Water Soluble?
• Most protein scientists prefer to work with
water soluble proteins or domains
• Membrane proteins are very difficult to
clone, express and purify and special
techniques must be used
• Potential problems can be avoided by
knowing whether the protein contains one
or more membrane spanning helices and
where these helices are located (cleaved?)
Lecture 3.0
27
Predicting via
Hydrophobicity
Bacteriorhodopsin
4
2
OmpA
3
1.5
2
1
0.5
1
0
0
-0.5
1
-1
-1
-2
-1.5
-3
-2
Bacteriorhodoposin
Lecture 3.0
OmpA
28
Membrane Helix Prediction
• Neural Network and HMM methods now claim
>80% accuracy
• PredictProtein (PHDhtm)
– http://cubic.bioc.columbia.edu/predictprotein/
• TMpred
– http://www.ch.embnet.org/software/TMPRED_form
.html
• TMHMM
– http://www.cbs.dtu.dk/services/TMHMM-2.0/
Lecture 3.0
29
TMPred (Principles)
Table 6
Protein Family
Cytokine/growth factor receptors
(EGF, interleukin, Insulin receptors)
G-coupled receptors
(rhodopsin, bacteriorhodopsin etc.)
Extracellular activated gated channels
(Glutamate, GABA, ACh sensitive)
Intracellular activated gated channels
5) photosynthetic proteins (H chain)
6) photosynthetic proteins (L chain)
7) porins
8) microsomal cytochrome p450
9) cytochrome b
10) Fo ATPases
Lecture 3.0
No. of Membrane Segments
1 (helix)
7 (helices)
4 (helices)
6 (helices)
5 (helices)
5 (helices)
17 (b-strands)
1 (helix)
1 (helix)
4 (helices)
30
TMHMM
Lecture 3.0
31
PredictProtein
Lecture 3.0
32
Expression Questions
•
•
•
•
•
•
•
•
Which host cell system?
Which expression vector?
Which cloning/expression protocols?
Is it membrane or water soluble?
Is it single domain or multi-domain?
How soluble and how stable?
Where will this protein be found?
How to purify & how to identify?
Lecture 3.0
33
Single Domain or
MultiDomain?
Lecture 3.0
34
Modular Protein Domains
BH
PDZ
FYVE
PH
DED
DEATH
SH3
1433
WW
Lecture
3.0
FHA
PTB
SH2
35
Single Domain or
MultiDomain?
• Many eukaryotic proteins are multi-domain
• Size is a good indicator (roughly 1 domain
for every 15 kD)
• Small domains behave better (Xray & NMR)
• Limited proteolysis allows experimental
identification of domains prior to structure
determination by NMR or X-ray
– digestion followed by HPLC or MS analysis to
detect fragments > 10 kD
Lecture 3.0
36
Domain Prediction
• Domain Prediction (PredictProtein-GLOBE)
– http://cubic.bioc.columbia.edu/predictprotein
• BLAST alignments can be used to detect or
predict the presence of domains by
sequence homology
• Protein domains can also be predicted
using CDD (Conserved Domain Database)
at http://www.ncbi.nlm.nih.gov/Structure/cdd/cdd.shtml
Lecture 3.0
37
Lecture 3.0
38
Lecture 3.0
39
Expression Questions
•
•
•
•
•
•
•
•
Which host cell system?
Which expression vector?
Which cloning/expression protocols?
Is it membrane or water soluble?
Is it single domain or multi-domain?
How soluble and how stable?
Where will this protein be found?
How to purify & how to identify?
Lecture 3.0
40
Predicting Solubility
• Even if a protein is identified to be a nonmembrane protein this does not
necessarily indicate it will be soluble
• Solubility depends on many factors
–
–
–
–
–
size (smaller ones are more soluble)
hydrophobicity (average and local hphob)
3D structure and ligand interactions
overall charge, predicted accessibility
distribution and frequency of amino acids
Lecture 3.0
41
Predicting Solubility
• Solvent accessibility prediction
– PredictProtein (PHDacc)
– http://cubic.bioc.columbia.edu/predictprotein/
• Protein property/scale prediction
– EXPASY ProtScale
– http://www.expasy.ch/cgi-bin/protscale.pl
• PepTool
– www.biotools.com
Lecture 3.0
42
Accessible Surface Area
Reentrant Surface
Solvent Probe
Accessible Surface
Van der Waals Surface
Lecture 3.0
43
Score
Predicted Accessibility
3
32
1
20
-1
1
-2
-3
0
-4
1
Lecture 3.0
51
101
151
201
251
301
44
Buried Surface Area (BASA) &
Fractional Burial (FB)
• For an average protein
• ASA (NP) = 0.35 x BASA
• ASA (P) = 0.61 x BASA
• ASA (+/-) = 0.04 x BASA
• BASA can be estimated
from a protein’s amino
acid composition
BASA = S AAi x FBi
Lecture 3.0
Table 9
Amino Acid Residue Fraction Buried
Residue frac. bur. Residue frac. bur.
A
C
D
E
F
G
H
I
K
L
0.38
0.45
0.15
0.18
0.5
0.36
0.17
0.6
0.03
0.45
M
N
P
Q
R
S
T
V
W
Y
0.4
0.12
0.18
0.07
0.01
0.22
0.23
0.54
0.27
0.15
45
ProtScale
Lecture 3.0
46
ProtScale
Lecture 3.0
47
Solubility (PepTool)
• Average Hydrophobicity
AH = S AAi x Hi
• Hydrophobic Ratio
RH = S H(-)/S H(+)
• Hydrophobic % Ratio
RHP = %philic/%phobic
• Linear Charge Density
LIND=(K+R+D+E+H+2)/#
• Solubility
SOL=RH + LIND 0.05AH
Lecture 3.0
• Average AH = 2.5 +/- 2.5
Insol > 0.1 Unstrc < -6
• Average RH = 1.2 +/- 0.4
Insol < 0.8 Unstrc > 1.9
• Average RHP = 0.9 +/- 0.2
Insol < 0.7 Unstrc > 1.4
• Average LIND = 0.25
Insol < 0.2 Unstrc > 0.4
• Average SOL = 1.6 +/- 0.5
Insol < 1.1 Unstrc > 2.5
48
Structural Proteomics and
Solubility Prediction
• Global efforts have led to the cloning and
attempted expression of more than 5000
water soluble proteins
• Data contained on databases such as
TargetDB allow correlations to be
developed between sequence and
expression levels and solubility
• Excellent opportunity to used data mining
to find “rules” to predict protein solubility
Lecture 3.0
49
Lecture 3.0
50
Binary Decision Trees
• Used to partition or classify data that is
not linearly separable
• Unknown objects are classified by
“traversing” the tree
• Traversing is accomplished by performing
tests at each node, direction of traversal
determined by results of the test
• Decision trees can be trained (test
threshold cutoff, test order, architecture)
Lecture 3.0
51
Binary Decision Trees
# not forming
crystals
Lecture 3.0
# forming
crystals
52
Predicting Protein Solubility
1) Residue frequency [ACDEFGHIKLMNPQRSTVWY]
2) Grouped residue frequency {[KR],[NR],[DE],[ST]
[LIM],[FWY],[HKR],[AVILM],[DENQ],[GAVL],[SCTM]}
3) Predicted % secondary structure [a,b,c]
4) Presence of signal sequence
5) Length of polypeptide
6) Number of residues in low complexity region (L,S)
7) Normalized low complexity value (SEG/Len)
8) Maximum hydrophobicity value
9) Length of maximum hydrophobic region
Lecture 3.0
53
Solubility Decision Tree
Size of black oval =
% that are soluble
Lecture 3.0
54
Binary Decision Trees
• Have been used to predict protein
solubility and protein crystallization
• Somewhat similar to self-organizing
feature maps (SOFM)
• Bertone P, Kluger Y, Lan N, Zheng D,
Christendat D, Yee A, Edwards AM,
Arrowsmith CH, Montelione GT, Gerstein
M. Nucleic Acids Res 2001 1;29(13):288498
Lecture 3.0
55
Predicting Stability
• Even if a protein expresses and remains
soluble it may turn out to be quite
unstable (easily proteolyzed)
• Proteins that are rich in Proline (P),
Glutamic acid (E), Serine (S) and
Threonine (T) or which have regions that
are rich in these amino acids (PEST
sequences) tend to have half lives of less
than 2 hours
Lecture 3.0
56
PEST Finder
http://www.at.embnet.org/embnet/tools/bio/PESTfind/
Lecture 3.0
57
Expression Questions
•
•
•
•
•
•
•
•
Which host cell system?
Which expression vector?
Which cloning/expression protocols?
Is it membrane or water soluble?
Is it single domain or multi-domain?
How soluble and how stable?
Where will this protein be found?
How to purify & how to identify?
Lecture 3.0
58
Protein Localization
• Is it exported? Does it go to the nucleus?
Does it go through the ER? Does it
localize to mitochondria? Chloroplasts?
Does it go to the membrane? How do you
tell?
• Eukaryotic signal sequences are usually
incompatible with prokaryotic signal
sequences so expressing eukaryotic
proteins in bacteria can lead to problems
Lecture 3.0
59
Location Prediction
http://psort.nibb.ac.jp
Lecture 3.0
60
Proteome Analyst
http://www.cs.ualberta.ca/~bioinfo/PA/Sub/
Lecture 3.0
61
PSORT-B (bacteria)
http://www.psort.org/psortb/index.html
Lecture 3.0
62
Location Prediction
http://www.cbs.dtu.dk/services/TargetP/#submission
Lecture 3.0
63
Other Sites or Modifications?
• Phosphorylation
– NetPhos http://cbs.dtu.dk/services/NetPhos/
• O-Glycosylation
– NetOGlyc http:/cbs.dtu.dk/services/NetOGlyc/
• Coil-Coil Dimerization domains
– www.ch.embnet.org/software/COILS_form.html
• Tyrosine Sulfation
– http://ca.expasy.org/tools/sulfinator/
Lecture 3.0
64
NetPhos 2.0
Lecture 3.0
65
Expression Questions
•
•
•
•
•
•
•
•
Which host cell system?
Which expression vector?
Which cloning/expression protocols?
Is it membrane or water soluble?
Is it single domain or multi-domain?
How soluble and how stable?
Where will this protein be found?
How to purify & how to identify?
Lecture 3.0
66
Finding and Identifying Your
Protein
Lecture 3.0
67
Isoelectric Point
• The pH at which protein has charge=0
•
Q = S Ni/(1 + 10pH-pKi)
pKa Values for Ionizable Amno Acids
Residue
C
D
E
Lecture 3.0
pKa
10.28
3.65
4.25
Residue
H
K
R
pKa
6
10.53
12.43
68
Isoelectric Point & MW
Calculation
Lecture 3.0
69
More Help?
•
•
•
•
•
•
•
http://www.abrf.org
http://www.abrf.org.JBT/JBTindex.html
http://www.BioTechniques.com
http://expasy.ch/alinks.html
http://www.neehow.org/wonderful/protocols
http://research.newfsc.noaa.gov/protocols.html
http://www.horizonpress.com/gateway/protocol
s.html
Lecture 3.0
70
Bioinformatics & Structural
Proteomics
• Key to identifying targets
• Key to reducing time and material wastage
in protein expression/purification steps
• Key to tracking and communicating target
progression (multi-lab LIMS)
• Key to reducing redundancy and
duplication by other X-ray or NMR
structure labs (TargetDB, SPINE)
Lecture 3.0
71
TargetDB
Lecture 3.0
http://targetdb.pdb.org/
72
Structural Proteomics - Status
•
•
•
•
•
•
•
•
20 registered centres (~30 organisms)
82700 targets have been selected
52705 targets have been cloned
29855 targets have been expressed
12311 targets are soluble
1493 X-ray structures determined
502 NMR structures determined
1743 Structures deposited in PDB
Lecture 3.0
73
Structural Proteomics - Status
•
•
•
•
•
•
•
543 structures deposited by Riken
265 structures deposited by Mid-West
187 structures deposited by North-East
179 structures deposited by New York
178 structures deposited by JCSG (UCSD)
52 structures deposited by Berkeley
31 structures deposited by Montreal/Kingston
Lecture 3.0
74
Protein Expression in E. coli
good
promising
unfolded
poor
precipitated
Lecture 3.0
75
Proc. Natl. Acad. Sci. USA, Vol. 99,1825-1830, 2002
Protein Expression in E. coli
Cloned (517 total)
expressed (85%)
soluble (68%)
M. th.= Methanobacter thermoautotrophicum
E. coli = Escherichia coli
S. ce. = Saccharomyces cerevisae
Myx. = Myxoma virus
T. ma. = Thermotoga maritima
Lecture 3.0
76
X-ray vs. NMR Results for
Methanobacter
Lecture 3.0
77
Conclusions
• The success of proteomics (structural,
functional, expressional) hinges almost
entirely on successful protein production
and expression
• Bioinformatics (web databases, servers,
data mining tools, NN’s, HMM’s) can and
does play an increasingly important role in
optimizing or improving protein
expression and coordinating large scale
proteomics efforts
Lecture 3.0
78