FEMS MicrobiologyLetters 97 (1992) 41-44 © 1992 Federation of European MicrobiologicalSocieties0378-1097/92/$05.00 Published by Elsevier
FEMSLE 05059
In vivo recombination and the production of hybrid genes S a b i n a C a l o g e r o , M a r c o E. Bianchi a n d A l e s s a n d r o Galizzi Dipartimento di Genetica e Microbiologia "Adriano Buzzati-Traverso'; Pavia, haly
Received 10 June 1992 Accepted 29 June 1992 Key words: Protein engineering; Recombination; recA
1. S U M M A R Y In vivo recombination between homologous genes is increasingly being favoured as a means of generating proteins with altered and novel specificities. The typical procedure requires the cloning of two related genes on a single replicative plasmid of Escherichia coli and the selection or screening of recombinants. Up to now the recombination process between the cloned genes was generally thought to involve the recA function and the availability of free ends in the D N A molecule to be recombined. Our results show that neither is necessary. Recombinants are obtained by simply growing the bacteria that host the plasmid carrying the two cloned genes.
2. I N T R O D U C T I O N The classical way to obtain variant gene products is to mutate cloned genes or to shuffle gene portions by splicing restriction fragments of reCorrespondence to: S. Calogero, Dipartimento di Genetica e Microbiologia "Adriano Buzzati-Traverso", Via Abbiategrasso 207, !-27100 Pavia, Italy.
lated genes. A different and less targeted approach is the generation of recombinant genes by means of in vivo recombination between partially related sequences. This approach can potentially create a large array of novel protein variants, since crossing overs can occur within regions of as little as two base pairs of homology. This is particularly advantageous when dealing with proteins with ill-defined active sites, or when one is looking for unpredictable new activities [1]. Weber and Weissmann [2] exploited in vivo recombination in Escherichia coli to obtain chimeric genes coding for alpha-interferon. They introduced in rec + E. coli cells linearized plasmid molecules bearing two different but related interferon genes at their termini. Recombination between the homologous genes restored the circularity and the transforming capacity of the plasraids. Rey et al. [3] generated hybrid alpha-amylase genes by the construction of a replicative plasmid in which the B. stearothermophilus amy promoter and two-thirds of the coding sequence were followed by the entire amy gene of Bacillus licheniformis lacking the promoter sequence. The plasmid was propagated in the rec-proficient E. coli strain 294, extracted and used to retransform E.
coli: after restriction of a site absent in the recombinant molecules, 25% of the rare transformants produced hybrid amylases. Both groups of investigators have assumed that recombinants would be produced via a recA-dependent pathway. In the former example linearization of the plasmid was intended to present double-strandod breaks to the recombination machinery of the E. coli, while in the latter it was conceived as a tool for selecting recombinants which had arisen spontaneously during propagation of the plasmid in the rec + E. coli host. Abastado et al. [4] supported the view that recombination proceeds from the termini of the D N A molecule, and calculated a stimulation factor of 30-300 in recombinant recovery when the D N A of a plasmid with internal homologous regions was restricted before transformation of E. coil at a site internal to a region of homology (no effect was reported following restriction outside of it). We have shown [5] that in vivo recombination between homologous insect-toxin genes of Bacillus thuringiensis cloned on a replicative plasmid in E. coli does not require the intervention of the recA protein, since a comparable number of recombinants were obtained from plasmid preparations derived from recA + and r e c A hosts. Pending the unsolved question of the role of double-stranded D N A ends, we have now constructed a new plasmid whose structure allows us to easily obtain quantitative data, and to discrimi-
nate between the two proposed models: recombination occurring during the normal plasmid propagation, or as a consequence of plasmid linearization.
3. M A T E R I A L S A N D M E T H O D S The bacterial strain used was E. coli D H 5 a [supE44 AlacU169 (dpSOlacZ AM15) hsdR17 fecAl endA1 gyrA96 thi-1 relAl]. The transformation procedure was carried out according to Hanahan [6].
Identification o f xyl- colonies LB plates containing 20 ~ g / m l kanamycin were sprayed with an aqueous solution of 0.5 M catechol and then incubated a few minutes at room temperature until colonies became intensely yellow.
4. R E S U L T S A N D D I S C U S S I O N
4.1. Plasmid construction Plasmid pK73X1 was constructed in several passages from plasmid p U G K [7], pES1 [8], pTG439 [9] and pJM109 (unpublished, kindly provided by Dr. M. Perego). A 1.5-kb B a m H I Sail fragment of pJM109 carrying the kanamycin resistance gene was first inserted in p U G K di-
BnmHI BnmHI Sma I / T a g I
HinIIId~4.~ K~j~ "HindIll Hindiil
~ $ m a l / TaqI ..... ///E,oa~ E0o.,\ \ //1 E©oRI
"o~. ~-~
A B Fig. !. Plasmids utilized for the construction of pK73X1. Thick black and white boxes represent respectivelycrilA(c) and crylA(a) genes (homologous regions are dashed). Dotted areas derive from B. subtilis (A) or B. licheniformis(B) chromosomes. The thin black arrow represents the xylE gene. Successivelysubcloned fragments (Xhol-Hindlll and BamHl-Bglll) are indicated inside the plasmids.
in Fig. IA (containing the 3'-truncated fragment of crylA(c)), and the 5.9-kb BamHI-Bglll fragment of pPKX shown in Fig. 1B (containing the xylE and the "crylA(a) genes). All restriction sites (Hindlll, X h o l , BamHl and Bglll) were blunte n d e d before ligation; yet the X h o l site was reformed by the filling-in and ligation o f the DNA ends. A B Fig. 2. Original pK73Xi plasmid(A) and derived recombinant (B). See Fig. I for symbols. gested with the same enzymes; plasmid p U G K K (Fig. IA) was obtained. Plasmid p P K X (Fig. 1B), derived from pSP64 (Riboprobe Gemini System, Promega) was constructed by cloning in two steps a 3.4-kb EcoRI fragment derived from a partial digestion of pES1 and carrying a 5'-truncated portion o f the crylA(a) gene, and a 2.5-kb Taql fragment derived from a partial digestion of pTG439, and carrying the xylE gene under the control of a promoter derived from B. licheniformis. These two fragments were inserted in the EcoRI and SmaI restriction sites of pSP64, respectively; the xylE gene was 5' to the 'cry gene. Plasmid pK73X1 (Fig. 2A) consists o f two portions derived from p U G K K and from pPKX: the 8.4-kb HindIII-XhoI fragment of p U G K K shown
4.2. In vivo recombination Plasmid pK73XI (Fig. 2A) carries two truncated genes of B. thuringiensis, containing sequences about 800 bp long and 63% homologous (hatched in Fig. 2), separated by the xylE gene. Recombination between the partially homologous sequences results in the generation o f xTl- plasraids (Fig. 2B), which would likely be undetectable due to the presence o f the xylE + unrecombined plasmids in the same cell, but can give rise to white colonies if extracted and reintroduced in E. coli by transformation. The plasmid contains a unique HpaI restriction site 500 bp upstream the 'crylA(a) gene, and a unique X h o l restriction site internal to the homologous region. A plasmid preparation o f pK73X1 was split in three parts, two o f which were restricted, and then used to transform E. coli DH5a. Results are shown in Table 1. The first line shows the data obtained with uncut DNA. A small number
Table 1 Detection of recombinantsby phenotypicscreening pK73X1 DNA
Number of coloniesobtained Relative transformation White Yellow a efficiency b
Fractionof true recombinants c
Recombinants//.Lg of DNA't
Uncut, 1 ~g Xhol digest, 4 p.g Hpal digest,4 p.g
6 40 10
0.21 (3/14) 0.~ (5/20) 0.55 (11/20)
1.3 3.2 3.4
150000 872 12200
1.0 0.79 0.79
E. coli strain DH5 was transformed with the indicated amounts of cleaved and uncleaved pK73XI, to which a small amount of supercoiled pBR322 DNA had been added. Transformed cells were selected on kanamycin-containingplates, and colonies were screened for the xTIE phenotype. One-twentieth of the transformed cells were plated on ampicinin-containingplates to measure the relative efficiencyof transformation. '~ YellowxylE+ coloniescan originate either from plasmid moleculesthat have not been cleavedby the restriction enzyme,or from cleaved plasmid moleculesthat have been recircularizedby end-to-end ligation inside the transformed cells. b Number of ampR colonies,normalizedto the uncut DNA control. c To obtain a reliable estimate of the fraction of true recombinants,we analysed white colonies pooled from several experiments (observed values are between brackets). o This number is normalized for the transformationefficiencyand was calculated as follows:(number of white colonies)×(fraction of true recombinants)/(amount of pK73XI DNA)×(transformation efficiency).
of white colonies (6 out of 150000 in our experiments) originated from the uncut sample of pK73X1. A restriction analysis conducted on white colonies obtained from seve~a! experiments reveated that 21% were true recombinants (see col. 5); the rest contained rearranged or mutated plasmids. Restriction of pK73XI prior to transformation (second and third lanes) produced a considerable increase in the percentage of white colonies (up to three orders of magnitude), but only a moderate increase in their absolute numbers (see col. 2 and 3). To b e t t e r calculate the absolute n u m b e r of true recombinants obtained from cut or uncut DNA, we spiked a small amount of supercoiled pBR322 in each transformation tube and plated 1 / 2 0 of the transformed cells on ampicillin-containing plates, in order to measure the relative efficiency of transformation. In addition, we analysed by restriction and in some case by sequencing the plasmids extracted from white colonies obtained from different transformations (observed values were comparable in all transformations). By multiplying the n u m b e r of white colonies (col. 2) by the fraction of true recombinants (col. 5), and dividing by the relative transformation efficiency (col. 4) and by the amount of D N A (col. 1), we obtained the data reported in col. 6. In all three cases, the n u m b e r of reeombinants p e r / ~ g of transforming D N A was comparable. The increase of recombinants due to restriction, if any, is mild indeed, and not d e p e n d e n t on the position of the restriction site relatively to the homology region. We therefore conclude the following: (i) T h e recombination is i n d e p e n d e n t from the recA function, and proceeds presumably via non-conservative pathwa~'~. From a practical point of view, it is advisable to use r e c A - hosts, since these produce a lower n u m b e r of complex rearrangements [5]. We have not ruled out that recombination of sequences with a high level of homology may proceed via recA-mediated mechanisms. However, we stress that for most protein engineering purposes, recombination between
genes with a low level of homology is required, and this proceeds readily through recA-independent pathways. (ii) Recombinants are formed during growth of the host, possibly as a consequence of plasmid replication. (iii) The stimulation of recombination by double-stranded fragments, unlike in yeast [10] and in mammalian cells [ll], is mild at best, and is not affected by the position of the fragment. (iv) Restriction enzyme digestions are quite effective as a selection tool.
ACKNOWLEDGEMENTS This work was partially supported by a grant from Ministero dell'Universit~ e della Ricerca Scientifica, Rome.
REFERENCES [1] Pompon, D. and Nicolas, A. (1989) Gene 83, 15-24. [2] Weber, H. and Weissmann, C. (1983) Nucleic Acids Res. 11, 5661-5669. [3] Rey, M.W., Requadt, C., Mainzer, S.E., Lamsa, M.H., Ferrari, E., Lad, P.J. and Gray, G.L. (1986) In: Bacillus Molecular Genetics and Biotechnology Applications (Ganesan, A.T. and Hoch, J.A., Eds.), pp. 229-239. Academic Press, Orlando, FL. [4] Abastado, J.-P., Darche, S., Godeau, F., Cami, B. and Kourilsky, P. (1987) Proc. Natl. Acad. Sci. USA, 84, 6496-6500. [5] Caramori, T., Albertini., A.M. and Galizzi, A. (1991) Gene 98, 37-44. [6] Hanahan, D. (1985) In: DNA Ctoning: A Practical Approach (Glover, D.M., Ed.), Vol. I, pp. 229-239. IRL Press, Oxford. [7] Calogero, S., AIbertini, A.M., Fogher, C., Marzari, R. and Galizzi, A. (1989) Appl. Environ. Microbiol. 55, 446-453. [8] Schnepf, H.E. and Whiteley, H.R. (1981) Proc. Natl. Acad. Sci. USA 78, 2893-2897. [9] Zukowski, M.M., Gaffney, D.F., Speck, D., Kauffman, M., Findeli, A., Wisecup, A. and Lecocq, J.-P. (1983) Proc. Natl. Acad. Sci. USA 80, 1101-1105. [10] Symington, L.S. (1991) EMBO J. 10, 987-996. [11] Lin, F.L, Sperle, K. and Sternberg, N. (1984) Mol. Cell Biol. 4, 1020-1034.
FEMSLE 05059
In vivo recombination and the production of hybrid genes S a b i n a C a l o g e r o , M a r c o E. Bianchi a n d A l e s s a n d r o Galizzi Dipartimento di Genetica e Microbiologia "Adriano Buzzati-Traverso'; Pavia, haly
Received 10 June 1992 Accepted 29 June 1992 Key words: Protein engineering; Recombination; recA
1. S U M M A R Y In vivo recombination between homologous genes is increasingly being favoured as a means of generating proteins with altered and novel specificities. The typical procedure requires the cloning of two related genes on a single replicative plasmid of Escherichia coli and the selection or screening of recombinants. Up to now the recombination process between the cloned genes was generally thought to involve the recA function and the availability of free ends in the D N A molecule to be recombined. Our results show that neither is necessary. Recombinants are obtained by simply growing the bacteria that host the plasmid carrying the two cloned genes.
2. I N T R O D U C T I O N The classical way to obtain variant gene products is to mutate cloned genes or to shuffle gene portions by splicing restriction fragments of reCorrespondence to: S. Calogero, Dipartimento di Genetica e Microbiologia "Adriano Buzzati-Traverso", Via Abbiategrasso 207, !-27100 Pavia, Italy.
lated genes. A different and less targeted approach is the generation of recombinant genes by means of in vivo recombination between partially related sequences. This approach can potentially create a large array of novel protein variants, since crossing overs can occur within regions of as little as two base pairs of homology. This is particularly advantageous when dealing with proteins with ill-defined active sites, or when one is looking for unpredictable new activities [1]. Weber and Weissmann [2] exploited in vivo recombination in Escherichia coli to obtain chimeric genes coding for alpha-interferon. They introduced in rec + E. coli cells linearized plasmid molecules bearing two different but related interferon genes at their termini. Recombination between the homologous genes restored the circularity and the transforming capacity of the plasraids. Rey et al. [3] generated hybrid alpha-amylase genes by the construction of a replicative plasmid in which the B. stearothermophilus amy promoter and two-thirds of the coding sequence were followed by the entire amy gene of Bacillus licheniformis lacking the promoter sequence. The plasmid was propagated in the rec-proficient E. coli strain 294, extracted and used to retransform E.
coli: after restriction of a site absent in the recombinant molecules, 25% of the rare transformants produced hybrid amylases. Both groups of investigators have assumed that recombinants would be produced via a recA-dependent pathway. In the former example linearization of the plasmid was intended to present double-strandod breaks to the recombination machinery of the E. coli, while in the latter it was conceived as a tool for selecting recombinants which had arisen spontaneously during propagation of the plasmid in the rec + E. coli host. Abastado et al. [4] supported the view that recombination proceeds from the termini of the D N A molecule, and calculated a stimulation factor of 30-300 in recombinant recovery when the D N A of a plasmid with internal homologous regions was restricted before transformation of E. coil at a site internal to a region of homology (no effect was reported following restriction outside of it). We have shown [5] that in vivo recombination between homologous insect-toxin genes of Bacillus thuringiensis cloned on a replicative plasmid in E. coli does not require the intervention of the recA protein, since a comparable number of recombinants were obtained from plasmid preparations derived from recA + and r e c A hosts. Pending the unsolved question of the role of double-stranded D N A ends, we have now constructed a new plasmid whose structure allows us to easily obtain quantitative data, and to discrimi-
nate between the two proposed models: recombination occurring during the normal plasmid propagation, or as a consequence of plasmid linearization.
3. M A T E R I A L S A N D M E T H O D S The bacterial strain used was E. coli D H 5 a [supE44 AlacU169 (dpSOlacZ AM15) hsdR17 fecAl endA1 gyrA96 thi-1 relAl]. The transformation procedure was carried out according to Hanahan [6].
Identification o f xyl- colonies LB plates containing 20 ~ g / m l kanamycin were sprayed with an aqueous solution of 0.5 M catechol and then incubated a few minutes at room temperature until colonies became intensely yellow.
4. R E S U L T S A N D D I S C U S S I O N
4.1. Plasmid construction Plasmid pK73X1 was constructed in several passages from plasmid p U G K [7], pES1 [8], pTG439 [9] and pJM109 (unpublished, kindly provided by Dr. M. Perego). A 1.5-kb B a m H I Sail fragment of pJM109 carrying the kanamycin resistance gene was first inserted in p U G K di-
BnmHI BnmHI Sma I / T a g I
HinIIId~4.~ K~j~ "HindIll Hindiil
~ $ m a l / TaqI ..... ///E,oa~ E0o.,\ \ //1 E©oRI
"o~. ~-~
A B Fig. !. Plasmids utilized for the construction of pK73X1. Thick black and white boxes represent respectivelycrilA(c) and crylA(a) genes (homologous regions are dashed). Dotted areas derive from B. subtilis (A) or B. licheniformis(B) chromosomes. The thin black arrow represents the xylE gene. Successivelysubcloned fragments (Xhol-Hindlll and BamHl-Bglll) are indicated inside the plasmids.
in Fig. IA (containing the 3'-truncated fragment of crylA(c)), and the 5.9-kb BamHI-Bglll fragment of pPKX shown in Fig. 1B (containing the xylE and the "crylA(a) genes). All restriction sites (Hindlll, X h o l , BamHl and Bglll) were blunte n d e d before ligation; yet the X h o l site was reformed by the filling-in and ligation o f the DNA ends. A B Fig. 2. Original pK73Xi plasmid(A) and derived recombinant (B). See Fig. I for symbols. gested with the same enzymes; plasmid p U G K K (Fig. IA) was obtained. Plasmid p P K X (Fig. 1B), derived from pSP64 (Riboprobe Gemini System, Promega) was constructed by cloning in two steps a 3.4-kb EcoRI fragment derived from a partial digestion of pES1 and carrying a 5'-truncated portion o f the crylA(a) gene, and a 2.5-kb Taql fragment derived from a partial digestion of pTG439, and carrying the xylE gene under the control of a promoter derived from B. licheniformis. These two fragments were inserted in the EcoRI and SmaI restriction sites of pSP64, respectively; the xylE gene was 5' to the 'cry gene. Plasmid pK73X1 (Fig. 2A) consists o f two portions derived from p U G K K and from pPKX: the 8.4-kb HindIII-XhoI fragment of p U G K K shown
4.2. In vivo recombination Plasmid pK73XI (Fig. 2A) carries two truncated genes of B. thuringiensis, containing sequences about 800 bp long and 63% homologous (hatched in Fig. 2), separated by the xylE gene. Recombination between the partially homologous sequences results in the generation o f xTl- plasraids (Fig. 2B), which would likely be undetectable due to the presence o f the xylE + unrecombined plasmids in the same cell, but can give rise to white colonies if extracted and reintroduced in E. coli by transformation. The plasmid contains a unique HpaI restriction site 500 bp upstream the 'crylA(a) gene, and a unique X h o l restriction site internal to the homologous region. A plasmid preparation o f pK73X1 was split in three parts, two o f which were restricted, and then used to transform E. coli DH5a. Results are shown in Table 1. The first line shows the data obtained with uncut DNA. A small number
Table 1 Detection of recombinantsby phenotypicscreening pK73X1 DNA
Number of coloniesobtained Relative transformation White Yellow a efficiency b
Fractionof true recombinants c
Recombinants//.Lg of DNA't
Uncut, 1 ~g Xhol digest, 4 p.g Hpal digest,4 p.g
6 40 10
0.21 (3/14) 0.~ (5/20) 0.55 (11/20)
1.3 3.2 3.4
150000 872 12200
1.0 0.79 0.79
E. coli strain DH5 was transformed with the indicated amounts of cleaved and uncleaved pK73XI, to which a small amount of supercoiled pBR322 DNA had been added. Transformed cells were selected on kanamycin-containingplates, and colonies were screened for the xTIE phenotype. One-twentieth of the transformed cells were plated on ampicinin-containingplates to measure the relative efficiencyof transformation. '~ YellowxylE+ coloniescan originate either from plasmid moleculesthat have not been cleavedby the restriction enzyme,or from cleaved plasmid moleculesthat have been recircularizedby end-to-end ligation inside the transformed cells. b Number of ampR colonies,normalizedto the uncut DNA control. c To obtain a reliable estimate of the fraction of true recombinants,we analysed white colonies pooled from several experiments (observed values are between brackets). o This number is normalized for the transformationefficiencyand was calculated as follows:(number of white colonies)×(fraction of true recombinants)/(amount of pK73XI DNA)×(transformation efficiency).
of white colonies (6 out of 150000 in our experiments) originated from the uncut sample of pK73X1. A restriction analysis conducted on white colonies obtained from seve~a! experiments reveated that 21% were true recombinants (see col. 5); the rest contained rearranged or mutated plasmids. Restriction of pK73XI prior to transformation (second and third lanes) produced a considerable increase in the percentage of white colonies (up to three orders of magnitude), but only a moderate increase in their absolute numbers (see col. 2 and 3). To b e t t e r calculate the absolute n u m b e r of true recombinants obtained from cut or uncut DNA, we spiked a small amount of supercoiled pBR322 in each transformation tube and plated 1 / 2 0 of the transformed cells on ampicillin-containing plates, in order to measure the relative efficiency of transformation. In addition, we analysed by restriction and in some case by sequencing the plasmids extracted from white colonies obtained from different transformations (observed values were comparable in all transformations). By multiplying the n u m b e r of white colonies (col. 2) by the fraction of true recombinants (col. 5), and dividing by the relative transformation efficiency (col. 4) and by the amount of D N A (col. 1), we obtained the data reported in col. 6. In all three cases, the n u m b e r of reeombinants p e r / ~ g of transforming D N A was comparable. The increase of recombinants due to restriction, if any, is mild indeed, and not d e p e n d e n t on the position of the restriction site relatively to the homology region. We therefore conclude the following: (i) T h e recombination is i n d e p e n d e n t from the recA function, and proceeds presumably via non-conservative pathwa~'~. From a practical point of view, it is advisable to use r e c A - hosts, since these produce a lower n u m b e r of complex rearrangements [5]. We have not ruled out that recombination of sequences with a high level of homology may proceed via recA-mediated mechanisms. However, we stress that for most protein engineering purposes, recombination between
genes with a low level of homology is required, and this proceeds readily through recA-independent pathways. (ii) Recombinants are formed during growth of the host, possibly as a consequence of plasmid replication. (iii) The stimulation of recombination by double-stranded fragments, unlike in yeast [10] and in mammalian cells [ll], is mild at best, and is not affected by the position of the fragment. (iv) Restriction enzyme digestions are quite effective as a selection tool.
ACKNOWLEDGEMENTS This work was partially supported by a grant from Ministero dell'Universit~ e della Ricerca Scientifica, Rome.
REFERENCES [1] Pompon, D. and Nicolas, A. (1989) Gene 83, 15-24. [2] Weber, H. and Weissmann, C. (1983) Nucleic Acids Res. 11, 5661-5669. [3] Rey, M.W., Requadt, C., Mainzer, S.E., Lamsa, M.H., Ferrari, E., Lad, P.J. and Gray, G.L. (1986) In: Bacillus Molecular Genetics and Biotechnology Applications (Ganesan, A.T. and Hoch, J.A., Eds.), pp. 229-239. Academic Press, Orlando, FL. [4] Abastado, J.-P., Darche, S., Godeau, F., Cami, B. and Kourilsky, P. (1987) Proc. Natl. Acad. Sci. USA, 84, 6496-6500. [5] Caramori, T., Albertini., A.M. and Galizzi, A. (1991) Gene 98, 37-44. [6] Hanahan, D. (1985) In: DNA Ctoning: A Practical Approach (Glover, D.M., Ed.), Vol. I, pp. 229-239. IRL Press, Oxford. [7] Calogero, S., AIbertini, A.M., Fogher, C., Marzari, R. and Galizzi, A. (1989) Appl. Environ. Microbiol. 55, 446-453. [8] Schnepf, H.E. and Whiteley, H.R. (1981) Proc. Natl. Acad. Sci. USA 78, 2893-2897. [9] Zukowski, M.M., Gaffney, D.F., Speck, D., Kauffman, M., Findeli, A., Wisecup, A. and Lecocq, J.-P. (1983) Proc. Natl. Acad. Sci. USA 80, 1101-1105. [10] Symington, L.S. (1991) EMBO J. 10, 987-996. [11] Lin, F.L, Sperle, K. and Sternberg, N. (1984) Mol. Cell Biol. 4, 1020-1034.
