Computational Design of New and Recombinant Selenoproteins

Transcription

1 Computational Design of ew and Recombinant Selenoproteins Rolf Backofen and Friedrich-Schiller-University Jena Institute of Computer Science Chair for Bioinformatics 1 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

2 Outline 1. Biological Introduction 2. The Computational Problem 3. The Algorithm 4. Results 5. Conclusion and Future Work 2 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

3 Biological Introduction - 1 RA Sequence: RA Secondary Structure: string over {A,C,G,U}, e.g. AUG }{{} AGCCCGAUC Codon set of complementary and connected base pairs G A G U A C C C G A U C 3 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

4 Biological Introduction - 1 RA Sequence: RA Secondary Structure: string over {A,C,G,U}, e.g. AUG }{{} AGCCCGAUC Codon set of complementary and connected base pairs mra Protein: A U G A G C Met Ser = linear translation G A G U A C C C G A U C STOP-codon: UGA, UAA, UAG (symbolize end of translation) 3 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

5 Biological Introduction - 2 Selenocysteine: new amino acid, encoded by the STOP-codon UGA inserted, if UGA is followed by a SECIS-element AUGUGGUGAAUCUAGA Met Trp STOP SECIS-element: specific mra sequence forms hairpin-like structure Met Trp Sec C U A AUGUGGUGA Ile Selenoproteins: contain selenocysteine important to human health, component of major metabolic pathways greatly enhanced enzymatic activities compared to the cysteine homologues 4 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

6 Protein expression system: E.coli The Biological Problem Eukaryotes: E.coli (bacteria): SECIS-element outside the coding sequence SECIS-element immediately downstream the UGA-codon located inside the coding part of the protein 5 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

7 Protein expression system: E.coli The Biological Problem Eukaryotes: E.coli (bacteria): SECIS-element outside the coding sequence SECIS-element immediately downstream the UGA-codon located inside the coding part of the protein Problem: compromise between quality of the SECIS-element and changes in protein sequence 5 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

8 The Computational Problem given: G... typical SECIS secondary structure S = S 1...S 3n... nucleotide sequence (SECIS-consensus) A = A 1...A n... original amino acid sequence 6 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

9 The Computational Problem given: G... typical SECIS secondary structure S = S 1...S 3n... nucleotide sequence (SECIS-consensus) A = A 1...A n... original amino acid sequence wanted: = n... mra sequence that can adopt G has maximum similarity with S encodes amino acid sequence A with maximal similarity to A A = A 1... A i... A n A = A 1... A i... A n = { }}{ { }}{ 3i 2 3i 1 3i... { }}{ 3n 2 3n 1 3n S = S 1 S 2 S 3... S 3i 2 S 3i 1 S 3i... S 3n 2 S 3n 1 S 3n similarity function: F S 3i 2S 3i 1 S 3i A i ( 3i 2 3i 1 3i ) conditions at conserved pos. (nucleotide + amino acid) 6 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

10 Contrary Conditions Possible Problem: contrary conditions at nucleotide and amino acid level A i Met {}}{ A U G G G U S 3i 2 S 3i 1 S 3i insertions/deletions & optional bonds 7 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

11 The Computational Problem - Extentions - 1 (1) Insertions and Deletions (at amino acid level, represented by vector t with t i { 1,0,1}) A : A 1 A 2 A : A 1 A 2 A 3 {}}{{}}{{}}{ : S : S 1 S 2 S 3 }{{} S 4S 5 S 6 }{{} S 7S 8 S 9 }{{} t: SECpos insertion of A 2, t 2 = +1 A 2 has no counterpart in A similarity: IP + F S 4S 5 S 6 ( ) 8 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

12 The Computational Problem - Extentions - 1 (1) Insertions and Deletions (at amino acid level, represented by vector t with t i { 1,0,1}) A : A 1 A 2 A : A 1 A 2 A 3 {}}{{}}{{}}{ : S : S 1 S 2 S 3 }{{} S 4S 5 S 6 }{{} S 7S 8 S 9 }{{} t: SECpos A : A 1 A 2 A 3 A 4 A : A 1 A 2 A 3 {}}{{}}{{}}{ : S : S 1 S 2 S 3 }{{} S 4 S 5 S 6 }{{} S 7S 8 S 9 }{{} t: SECpos insertion of A 2, t 2 = +1 deletion of A 2, t 2 = 1 A 2 has no counterpart in A compare with A 3 similarity: IP + F S if amino acid cond. at 2 4S 5 S 6 ( ) similarity: DP + F S 4S 5 S 6 A 3 ( ) otherwise 8 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

13 The Computational Problem - Extentions - 2 Similarity Function: f (L i,a i,t i ), where L i... codon corresponding to 3i 2 3i 1 3i t i { 1 (deletion),0 (subst.),+1 (insertion)} f i (L i,a i,0) = F S 3i 2S 3i 1 S 3i A i ai ( 3i 2 3i 1 3i ) a i = i j=1 t j f i (L i,a i,+1) = IP + F S 3i 2S 3i 1 S 3i ( 3i 2 3i 1 3i ) { if amino acid cond. at i (a i t f i (L i,a i, 1) = i ) DP + f i (L i,a i,0) otherwise 9 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

14 The Computational Problem - Extentions - 2 Similarity Function: f (L i,a i,t i ), where L i... codon corresponding to 3i 2 3i 1 3i t i { 1 (deletion),0 (subst.),+1 (insertion)} f i (L i,a i,0) = F S 3i 2S 3i 1 S 3i A i ai ( 3i 2 3i 1 3i ) a i = i j=1 t j f i (L i,a i,+1) = IP + F S 3i 2S 3i 1 S 3i ( 3i 2 3i 1 3i ) { if amino acid cond. at i (a i t f i (L i,a i, 1) = i ) DP + f i (L i,a i,0) otherwise (2) Optional Bonds bond... mandatory bond optional bond... not fixed, but of advantage if formed 9 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

15 The Algorithm - Input structure represented by a graph: outer-planar edgelabeled graph G = (V,E,Lab), with V = {v 1,...,v 3n } Lab(v k,v l ) Condition Meaning {v k,v l } E(G) : +1 k l C bond +2 k {A,C,G,U} optional bond -1 k l C prohibited bond -2 k {A,C,G,U} opt. prohib. bond 10 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

16 The Algorithm - Input structure represented by a graph: outer-planar edgelabeled graph G = (V,E,Lab), with V = {v 1,...,v 3n } Lab(v k,v l ) Condition Meaning {v k,v l } E(G) : +1 k l C bond +2 k {A,C,G,U} optional bond -1 k l C prohibited bond -2 k {A,C,G,U} opt. prohib. bond amino acid sequence A = A 1 A 2...A n amino acid + nucleotide conditions n similarity functions f 1,..., f n, where f i is associated with {v 3i 2,v 3i 1,v 3i } 10 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

17 The Algorithm - Output Vectors with: = ( 1,..., 3n ) {A,C,G,U} 3n and t = (t 1,...,t n ) { 1,0,1} n, k assigned to v k t... succession of insertions and deletions all complementarity conditions hold (given by the labeling) n i=1 f i (L i,a i,t i ) is maximized (L i correponds to 3i 2 3i 1 3i ) 11 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

18 The Algorithm - Preparing Definitions - 1 Definition 1 Given a structure graph G, we define the implied graph G impl as a graph on the vertices V (G impl ) = {u 1,...,u n } with: E(G impl {u i,u j } r {3i 2,3i 1,3i} : ) = s {3 j 2,3 j 1,3 j} : (v r,v s ) E(G) v v v v v v v v v v v v u u 1 u 3 u 4 ucleotide level (G) AA level (G impl ) (independent of index: u V (G impl ), v V (G)) 12 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

19 The Algorithm - Preparing Definitions - 2 Compatibility: E(G) (u i,l i ) E(G) (u j,l j ) = true, if {u i,u j } / E(G impl ) true, false, if {u i,u j } E(G impl ) and the corresp. nucleotides of L i and L j satisfy the complementarity cond. of G otherwise 13 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

20 The Algorithm - Preparing Definitions - 2 Compatibility: E(G) (u i,l i ) E(G) (u j,l j ) = true, if {u i,u j } / E(G impl ) true, false, if {u i,u j } E(G impl ) and the corresp. nucleotides of L i and L j satisfy the complementarity cond. of G otherwise Central Function: [ l... insertions - deletions left s... insertions - deletions inside interval ] w 1 n(l 1,L n,m,l,s) = max L 2...L n 1 t(s) 1< j n f j (L j,l + j i t g,t j i ) g=1 to L 1...L n corresp. nucl. satisfy E(G), m real. opt. bonds at L 1...L n 13 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

21 The Algorithm - 1 Value of Interest: max L 1,L n,m l 1, s n 1 {w 1 n(l 1,L n,m,l,s) + f 1 (L 1,l,l)} solved by dynamic programming 14 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

22 The Algorithm - 1 Value of Interest: max L 1,L n,m l 1, s n 1 {w 1 n(l 1,L n,m,l,s) + f 1 (L 1,l,l)} solved by dynamic programming Subproblems: split the problem on {1,..., n} into two parts, solve subproblems optimally vertex to split: next(i, j)= r i + 1 if u r is the farthest vertex in {u i,...,u j 1 } adjacent to u i otherwise u u i u next(i,j) j 14 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

23 The Algorithm - 2 Recurrence Theorem: w i i+k (L i,l i+k,m,l,s) equals if (u i,l i ) E(G) (u i+k,l i+k ) max L p splits (m 1,m 2 ) of m splits (s 1,s 2 ) of s ( w i p(l i,l p,m 1,l,s 1 ) + w p i+k(l p,l i+k,m 2,l + s 1,s 2 ) if (u i,l i ) E(G) (u i+k,l i+k ) ) where p = next(i,i + k) and m = m 1 + m 2 + opt L i L i+k with opt L i L i+k = number of realized optional bonds only between codons L i and L i+k. 15 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

24 Results - 1 SECIS-element of E.coli: A) Standard B) Additional Codon C) Lacking First Codon G U G A C G/C U/A U G/C G/C UGA C U G C/G A/U C/G C/G UGA G U G A C G/C U/A U G/C G/C C U G C/G A/U C/G C/G G/C U/A G/C G/C UGA G U G A C U C U G C/G A/U C/G C/G from Liu et al., AR Computational Design of ew and Recombinant Selenoproteins July 06, 2004

25 Results - 2 Mammalian methionine sulfoxide reductase B (MsrB): (also analyzed by Bar-oy et al., 2002) A) mouse MsrB... U I F S S S L K F V P K G K E... Sim = 65 new MsrB... U I F S S L P G L V P K G K E... Sim = 43 MsrB[Bar-oy et al.]... U I F S T V A G L H P K G K E... Sim = 35 new mra... UGAAUUUUCUCUUCGCUACCAGGUCUGGUGCCAAAAGGAAAAGAA new struc......(((((.((.((.((((...)))))).)).)))))... opt = 8 (9) B) mouse MsrB... U I F S S S L K - F V P K G K E... new MsrB... U I F S S S L P G L V P K G K E... Sim = 57 + IP new mra... UGAAUAUUUUCCUCUUCGCUACCAGGUCUGGUGCCAAAAGGAAAAGAA new struc......(.(((((.((.((.((((...)))))).)).))))))... opt = 9 (9) C) mouse MsrB... U I F S S S L K F V P K G K E... new MsrB... U I F S L P - G L V P K G K E... Sim = 41 + DP new mra... UGAAUCUUUUCGCUACCAGGUCUGGUGCCAAAAGGUAAAGAA new struc......(((((((.((.((((...)))))).)))))))... opt = 6 (7) 17 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

26 Results for Human Selenoproteins A) GPX2... G T T T R D F T Q L E L Q... GPX2... G T T T L A G L Q L E L Q... Sel... G S G R T L R E T V L E S S P... Sel... G - G R T L P G L V L E S S P... SelP S Y S L R Y I L L K K S L E... SelP S Y S L L A G L L R K S L E... B) Sel... G S G R T L R E T V L E S S P... Sel... G S G R T L P G L V L E S S P... SelP Y L C I I E A S K L E D L R V... SelP Y L C I Y V A G L L E D L R V... SelP Q C K E L P S L C S C Q G L... SelP Q C K E L P G L V A C Q G L... C) IOD2... P P F T S Q L P A F R K L... IOD2... P P F L A G L Q A F R K L... IOD3... P P F M A R M S A F Q R L... IOD3... P P F L A G L Q A F Q R L... GPX1... G T T V R D Y T Q M E L... GPX1... G T T V A G L L Q M E L... GPX3... G L T G Q Y I E L A L Q... GPX3... G L T L P G L E L A L Q... SelT... G Y R R V - F E E Y M R V... SelT... G Y R L P G L E E Y M R V... SelV... S Y S L R Y I L L K K S L... SelV... S Y S L A G L L L K K S L... SelW... G Y K S K Y L Q L K K K L... SelW... G Y R L A G L Q L K K K L... SPS2... G C K V P Q E A L L K L L... SPS2... G C R V P G L V L L K L L Computational Design of ew and Recombinant Selenoproteins July 06, 2004

27 Conclusion and Future Work Conclusion: inserted bacterial SECIS-element in mra of eukaryotic selenoproteins only few amino acids changed 19 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

28 Conclusion and Future Work Conclusion: inserted bacterial SECIS-element in mra of eukaryotic selenoproteins only few amino acids changed Future Work: consider folding probability additionally (contrary) increase probability: local search methods keep a certain similarity 19 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

29 Thanks Thanks to my supervisor Rolf Backofen. 20 Computational Design of ew and Recombinant Selenoproteins July 06, 2004

30 Thanks Thanks to my supervisor Rolf Backofen. I thank you for your attention. 20 Computational Design of ew and Recombinant Selenoproteins July 06, 2004