Random dissection to select for protein split sites and its application in protein fragment complementation

Abstract

To identify protein split sites quickly, a selection procedure by using chloramphenicol acetyl transferase (CAT) as reporter was introduced to search for folded protein fragments from libraries generated by random digestion and reassembly of the target gene, which yielded an abundant amount of DNA fragments with controllable lengths. Experimental results of tryptophan synthase alpha subunit (TSalpha) and TEM-1 beta-lactamase agreed well with what the literature has reported. The solubility of these fragments correlated roughly with the minimum inhibitory concentrations of the CAT fusions. The application of this dissection protocol to protein fragment complementation assay (PCA) was evaluated using aminoglycoside-3'-phosphotransferase I (APH(3')-I) as a model protein. Three nearly bisectional sites and a number of possible split points were identified, and guided by this result, four novel pairs of fragments were tested for complementation. Three out of four pairs partially restored the APH activity with the help of leucine zippers, and a truncated but active APH(3')-I (Delta1-25) was also found. Finally, the weakly active APH(3')-I-(1-253)NZ/CZ (254-271) containing a short 18 residue tag was further improved by error-prone PCR, and a best mutant was obtained showing a fourfold improvement after just one round of evolution. These results demonstrate that protein random dissection based on the CAT selection can provide an efficient search for protein breakage points and guide the design of fragments for protein complementation assay. Furthermore, more active fragment pairs can be achieved with the classical directed evolution approach.

Figure 1

Vectors constructed in this study. Vectors pCAT-1, pCAT-2, and pCAT-2d used for selection of soluble fragments are derivatives of pET30a(+), linker sequences (boxed) are placed upstream of the CAT gene, and an internal BamH I site for pCAT-1 and EcoR I site for pCAT-2 are used for insertion of gene fragments, respectively. The pCAT-2d vector was created by deleting the ATG in the cat gene of pCAT-2. The pCY-T7, a derivative of pTWin 1, contains two T7 promoters which control the expression of the N-terminal and C-terminal fragments, respectively. Linker sequences between zippers and inserted fragments are shown (boxed). T7, T7 promoter; RBS, ribosome binding site; T7-ter, T7-terminator.

Figure 2

Experimental procedure for protein random dissection and selection to search for soluble fragments of a target protein. A target gene was PCR amplified and digested with DNase I, and the gene segment libraries were generated by reassembly of the smaller digestion products. Folded protein fragments were selected by using CAT as a reporter. DNA was analyzed by 1.2% agarose gel. Lane M, DNA ladder.

Figure 3

Coverage plots of non-redundant sets of fragments identified by the protein random dissection and selection method, arrayed against the parental sequences (lower arrow lines). (A) Fragments of TSα deduced from the sequences of inserts contained in fragment-CAT fusion clones. The three dot lines represent previously reported breakage points at residues A73, G170, and A189. A substitution L127P was found in fragment TSα-6. (B) Fragments obtained from TEM-1 β-lactamase dissection. Domain boundaries of TEM-1 β-lactamase at residues F66 and K215 are shown in dot lines. A substitution G220R was found in fragment BLA-7. (C) Identified fragments from APH(3′)-I random dissection. The presumed domain boundary is marked by 9 dot line. (D) The score curve of APH(3′)-I generated by the STAR predictor. Solid lines, sites obtained in this work.

Figure 4

Histograms for the size distribution frequency (N) of fragments determined by colony PCR with flanking primers from chloramphenicol resistant cells harboring fragment-CAT fusions. Black bars: TSα fragments; hatched bars: TEM-1 β-lactamase fragments; white bars: APH(3′)-I fragments.