A “mesmer”izing new approach to site-directed mutagenesis in large transformation-ready constructs

Abstract

Creating designer mutations in large genes is a challenge. Size limitations imposed by site-directed mutagenesis (SDM) coupled with the paucity of unique restriction enzyme sites make subsequent cloning of these constructs extremely difficult. “Mutagenesis via Serial Small Mismatch Recombineering” (MSSMR) combines sequential recombineering steps with SDM to create seamless, pre-specified mutations as small as a single base pair. We demonstrate the simultaneous cloning of wild-type and mutant constructs of a >30 kb gene directly into attB transformation vectors. No post-transformation manipulations are required, and because the technique relies on recombineering methods, addition of undesired mutations via PCR is minimized.

Introduction

The Drosophila model system is renowned for the plentiful techniques available for genetic manipulation. There are many established methods for generating mutations using various transposable elements, chemical and radiologic protocols.^1–3 Creating designer mutations is generally more difficult than these “random” methods, and incorporating specific small mutations in large genes (>20 kb) often extremely complicated. When working with these genes, established techniques are often hindered by (1) lack of unique restriction enzyme sites, (2) non-random and incomplete genomic coverage of P-element insertions^4,5 and (3) size limitations of PCR-based site directed mutagenesis (SDM) strategies,⁶ to name just a few. Mutants produced by cassette-based mutagenesis (e.g., homologous recombination) methods can answer many research questions, but can be inappropriate for some studies due to the addition of undesired genetic sequence or the requirement for extra manipulation following transformation to create mutant flies.^7–11 The MSSMR technique described here allows concurrent mutagenesis and cloning of wild-type and mutant DNA directly into a P-element-based transformation vector through a modified recombineering reaction.

The P[acman] system^12–14 combines several powerful techniques for genetic manipulation in Drosophila. Recombineering (reviewed in refs. ^15–17), P-element transformation,^18,19 attB targeting via phiC31 integrase,²⁰ and protein tagging¹³ have all been combined in a set of BAC-based vectors which allow cloning and germline transformation of large genes. We have adapted this system to allow the generation of designer mutations in target genes. The mutagenesis is performed using “small mismatch” regions nested within homologous recombination arms. Following recombination, the recovered colonies are a mixture of wild-type and mutant clones which can be identified by PCR, sequencing or restriction mapping. This MSSMR technique has the benefits of using no expression cassettes or post-transformation FLP/FRT steps.^7–11,21 It produces mutations with single base pair resolution, carries virtually no size restrictions on the target gene and allows wild-type and mutant clones to be recovered simultaneously. Because PCR is required only for production of small homology regions, the likelihood of PCR-induced mutations (other than those specifically intended) is minimized and easily monitored.

Results

Experimental design.

CG17150 (FBgn0035581) encodes a ∼34 kb dynein heavy chain (DHC) gene located at 64C1-2 (DHC64C). A ∼36 kb target segment of DHC64C genomic DNA contained in BAC RP98-17L24 (BACPAC, CHORI) was chosen for cloning. This segment includes the entire DHC64C gene prediction as well as 1,100 bp of upstream regulatory region and 1,300 bp of downstream sequence containing the consensus polyadenylation signal. Two homology arms each approximately 500 bp were amplified at the 5′ and 3′ extremes of the 36 kb target region (Fig. 2) using primers (Table 1) compatible with the P[acman] vector recombineering system.¹²

Figure 2.

Experimental steps in the MSSMR technique. BAC DNA (light blue box and thin arrows) containing the DHC64C target region and P[acman] vector DNA (thin black arrows) are shown in linear representation. Primers are indicated by thick arrows. Solid thick arrows indicate primers used for homology arm PCR. Gradient thick arrows indicate primers used for amplification of the mutagenesis segment and/or colony PCR screening. The various DNA segments are NOT shown to scale so that important aspects can be emphasized. (A) In the initial step, the left and right homology arms (solid blue and yellow boxes) and mutation segment (open red box) are amplified from the target DNA. The mutation segment is then used as the target for SDM to create the small mismatch mutation (asterisk). Nested PCR is used to amplify the mutation homology arm (solid red box). (B) The three homology arms are then assembled by subcloning each into the vector. The first recombineering reaction is performed using vector construct linearized at the MluI site. The product of this recombineering reaction, the Left Construct, is shown in (C) along with the primer pairs used to verify the insert by colony PCR. Following verification, the Left Construct is linearized at the NotI site and used for the second recombineering reaction. The Full-length Construct is diagrammed in (D) along with the primer pairs used to screen potential recombinants.

Table 1.

Primers described in the experiments

Primer name	Use	Sequence (5′-3′)
Left Arm Forward (LAF)	HA PCR	[GGC GCG CC] CGC TC A AAA CAC
Left Arm Reverse (LAR)	HA PCR	[GGA TCC] TAG TCC AGA GCG CAC GAA G
Mutation Arm Forward (MAF)	HA PCR	[GGA TCC] [ACG CGT] CTT TC A GGC TAT ATG C
Mutation Arm Reverse (MAR)	HA PCR	[GCG GCC GC] GGT ATT GAA AGT CAC G
Right Arm Forward (RAF)	HA PCR	[GCG GCC GC] TC A ATT GGA GC
Right Arm Reverse (RAR)	HA PCR	[TTA ATT AA] GAT GAT TAG CGC
MCS -F*	Colony screening	TTT AAA CCT CGA GCG GTC CGT TAT C
MCS -R*	Colony screening	CTA AAG GGA ACA AAA GCT GGG TAC
5′ Check Reverse (5CR)	Colony screening	GTG TGC GCG TGA TCC TTC CGC
3′ Check Forward (3CF)	Colony screening	GGA TCC GCG CAC GCT GGA ATC
M Segment Forward (MSF)	PCR SDM target	GCA ACC CC A CC G ACA CTG CC
M Segment Reverse (MSR)	PCR SDM target	CGA TTA TCC CGG TGC TGC TG
Mutation 1 Forward	SDM	CTG AAG GAC CT G CC G AAA CT G GCA AGA CAG A
Mutation 1 Reverse	SDM	TCT GTC TTG CCA GTT TCG GCA GGT CCT TCA G
Mutation 2 Forward	SDM	CTC CTG AAG GAC CTC CCG AAA CTG GCA AGA CAG
Mutation 2 Reverse	SDM	CTG TCT TGC CAG TTT CGG GAG GTC CTT CAG GAG
Mutation 3 Forward	SDM	TCT CTG TCT TGC CAG ATT CGG GAG GTC CTT CAG GAG C
Mutation 3 Reverse	SDM	GCT CCT GAA GGA CCT CCC GAA TCT GGC AAG ACA GAG A

Abbreviations included with primer names are as shown in Figure 2. Note that the homology arm primers include compatible restriction enzyme sequences (indicated with brackets and boldface) for subcloning into the vector (PacI and AscI), joining adjacent homology arms (BamHI and NotI), and/or linearizing the construct (MluI and NotI).

^*Venken et al.¹²

A third homology region, the mutagenesis (M) segment, was also amplified by PCR. This segment contains the gene region chosen for mutation (Fig. 2). For DHC64C, this region is in exon 24, approximately 15.5 kb downstream of the left homology arm. The M segment was directly cloned into the pCR2.1 TOPO vector (Invitrogen), and subsequently used as the template for PCR-based SDM. Three mutations (M1, M2 and M3) were designed containing one, two or three base pair substitutions, respectively, in a 3-codon stretch (Table 2). The individual “M” products were amplified (QuikChange II kit, Stratagene), cloned and sequenced to confirm the integrity of each mutation. The central 725 bp region of each M segment was then amplified with nested “mutation arm” primers (Table 1 and Fig. 2) containing restriction enzyme sites compatible with subcloning into the P[acman] vector¹² between the left and right homology arms (Fig. 1). That is, each homology arm was separated from the next by a restriction enzyme site unique for the initial P[acman] construct. Additionally, at least one of the 2 enzyme sites separating the left and right homology arms from the mutation arm was unique for the product of the first recombineering reaction (Fig. 2). This allowed the product of the first reaction to be linearized for the second recombineering step.

Table 2.

Efficiency of the MSS MR recombination scheme

Step 1							Step 2
Construct (∼14.5 kb)	Target	Size (kb)	# colonies screened	% FLb	% WT	% Mut	Construct (∼28.5 kb)	Target	Size (kb)	# colonies screened	% FLb	% WT	% Mut
WT-Ca	BAC	163.7	50	4	100	NA	NA	NA	NA	NA	NA	NA	NA
M1	BAC	163.7	111	4	0	100	M1-L	BAC	163.7	89	38	17	83
M2	WT-Pa	∼49	18	11	45	55	M2-L	WT-Pa	∼49	9	89	44	56
M3	WT-Pa	∼49	20	20	0	100	M3-L	WT-Pa	∼49	18	100	44	56

The MSSMR reactions were performed using either BAC DNA (for isolation of DHC64C^wt and M1 clones) or DHC64C^wt in P[acman] (for M2 and M3) as the target. Using the smaller target was 2–5x more efficient than using the BAC target, but mutants were isolated at each step regardless of the target used.

^aThe wild-type recombineering construct (WT-C) consisted of the M1 construct double cut with MluI and NotI to linearize, thereby removing the “M” segment. This reaction required only one recombineering step to obtain the full length product; the designation NA (not applicable) is therefore used for Step 1, % Mutant and all Step 2 fields. This experiment was done only as a control for comparison of the efficiency of one-step vs. two-step recombineering. WT-C (the recombineering construct) should not be confused with the recombineering target WT-P (DHC 64C^wt in attB-Pacman).

^bBecause different numbers of colonies were screened in the various experiments, efficiency is shown as % FL (full length), where FL indicates that the clone passed all screening tests (PCR confirmation with all “check” primer pairs and restriction digest of the full length plasmid). The percentage of wild-type and mutant clones is calculated from full length clones only (e.g., %WT, number of full length clones with wild-type sequence in M segment/number of full length clones).

Figure 1.

Schematic of the assembled MSSMR vector. Three homology arms (yellow, red and blue boxes) are shown inserted in the multiple cloning site of the attB-P[acman]-AmpR vector.¹² The desired mutation (indicated by an asterisk) has already been introduced in the mutation arm via site directed mutagenesis. The restriction enzymes sites separating the mutation arm from the left and right arms must fullfill two criteria: (1) both are unique for the vector construct so that the vector can be linearized for each recombineering reaction, and (2) one of these enzymes must also be unique for the product of the first recombineering reaction (Left Construct), so it can be linearized for the second recombineering event. Note that the P[acman] vector includes P-element transposase recognition sites so that the recombineered insert can be mobilized from transformant flies if desired. For the purposes of this experiment, the salient features are the attB targetting sequence, the BAC backbone capable of stably incorporating large inserts, and the w⁺ marker to screen potential transformants.

An attB-P[acman]¹² construct was assembled for each desired mutation (Figs. 1 and 2). Recovery of the full-length mutated construct proceeded through 2 steps: For the first reaction, the vector construct was linearized between the left and M arms using MluI, allowing recombineering of the ∼15 kb portion of the target 5′ to the mutation arm (Left Construct, Fig. 2C). Verification of each end of the target insert generally proceeded as outlined by Venken et al.¹² via colony PCR. Specifically, a PCR reaction was performed to check the integrity of each end of the recombineered product: one primer was specific for the P[acman] vector (MCS-F or MCS-R,¹² Fig. 2) and the other specific for a portion of target segment (5′ Check Reverse (5CR), Table 1 and Fig. 2) or the right homology arm (RAF, Table 1 and Fig. 2). When amplification from both ends indicated a complete Left Construct, an additional colony PCR reaction was added to check the identity of the M segment. For this amplification, the primers consisted of one in the 3′ end of the left target segment paired with the reverse primer for the M arm (Mutation Segment Forward (MSF) and Mutation Arm Reverse (MAR), respectively, Table 1 and Fig. 2). Two techniques—direct sequencing and restriction enzyme digestion of the mutation segment PCR products—were then used to confirm the identity of the M segment in each clone (Fig. 3). Additionally, restriction digests were performed on the Left Constructs to verify that fragment sizes were consistent with the original BAC template.

Figure 3.

Mutation segment verification by sequencing and restriction digest. Full-length inserts underwent further verification to determine whether the mutation segment or wild-type target segment was included in each clone. (A) Pherograms from each mutant in the M segment, indicating incorporation of the desired mutation. Wild-type DHC 64C contains an MspI restriction sequence at the site of the mutation (blue bracket). In each mutation (M1, 2 or 3), this site is lost due to the G to A base pair substitution. Base pair mutations are indicated by asterisks. (B) PCR using the M segment primers, followed by MspI digestion of the amplified products, illustrates the loss of the restriction site. In the DHC 64C^wt clone, the largest restriction fragment is ∼600 bp (solid black arrow), while all three mutants present a larger (720 bp, open arrow) band.

Following verification of the Left Construct for each mutation, a second recombineering reaction was performed to add the right target segment, and thus recover the full-length products. For this reaction, the ∼28.5 kb Left Construct (Fig. 2C) was linearized with NotI between the mutation and right homology arms. Following this second recombineering reaction, colony PCR was once again used to screen individual colonies for the presence of left and right segments, as well as the integrity of the mutation, within the ∼49 kb full-length construct (Fig. 2D). Final confirmation of intact constructs included restriction digests to eliminate the possibility of undesired rearrangements, and limited sequencing within the target segments to confirm that the “joints” between the arms were recombineered seamlessly (i.e., no terminal restriction enzyme recognition sequence remained from the homology arms).

Efficiency of MSSMR vs. traditional recombineering.

To test the efficiency of the technique using different target sizes, the MSSMR reaction was performed using either full-length BAC DNA (∼164 kb) or a previously recombineered wild-type DHC64C P[acman] construct (DHC64C^wt; ∼49 kb) as the target (Table 2). To generate DHC64C^wt, the M1 construct was assembled in attB-P[acman]-Amp¹² then linearized with both MluI and NotI, thus removing the mutation arm and creating a “traditional” one step recombineering construct (Fig. 1). Recombineering from BAC DNA produced hundreds of potential DHC64C^wt clones. Fifty of these were screened by colony PCR for full-length recombinant products and 4% contained full-length clones. This level of efficiency is similar to that calculated by our lab for recovery of other large recombinant constructs from BAC DNA (Jacobs J and Eberl D, unpublished results). Most “background” clones which failed colony PCR screening with “check” primer sets were determined by PCR with MCS-F and MCS-R to be intact (uncut) recombineering constructs (data not shown) which failed to recombine. Using colony PCR to eliminate these non-recombinants was an extremely efficient method for selecting clones—allowing us to easily test ∼100 clones in several hours.

In a side-by-side reaction using the M1 construct linearized with only MluI, 111 potential M1Left colonies recombineered from BAC DNA were screened, of which five (∼4%) contained complete left segments (Table 2). Sequencing showed that all five of the colonies were the desired M1 Left Construct, rather than a mixture of wild-type and mutant clones. This indicates that an M arm of ∼700 bp is long enough to reasonably assure inclusion of the desired mutation in the product.

In the second reaction, M1Left was linearized with NotI for recombineering from BAC DNA (Fig. 2). Of the 89 colonies produced from the reaction, 38% contained a full-length product. Six of the full-length clones were analyzed by PCR and restriction digest. Of these, five (83%) were DHC64C^M1 and only one (17%) was a DHC64C^wt clone. Therefore, there may be a bias toward retention of the construct homology arm over the target homologous segment. In fact, all “M” reactions yielded as many or more full-length mutant clones than full-length wild-type colonies (summarized in Table 2).

For isolation of M2 and M3 full-length products, the homology arms were assembled in attB-P[acman]-Chl^R.¹² This allowed us to test the efficiency of performing the recombineering reaction using DHC64C^wt in attB-P[acman]-Amp^R generated above as the recombineering target. Results are summarized in Table 2. In every two-step “M” experiment, both wild-type and mutant full-length constructs were recovered. Interestingly, in two of the three “M” experiments, wild-type clones were only isolated in the second recombineering step. It should be noted, however, that screening stopped when a full-length mutant was recovered—therefore, not all colonies were screened. The efficiency of recovery of full-length products was much higher in the second recombineering step, likely due to the 13 kb homology region shared in this step by the target and recombineering constructs (Table 2). Therefore, the limiting factor in the recovery of desired clones appears to be the initial recombineering reaction.

The four constructs were injected separately (Model System Genomics, Duke Univ.) into embryos containing an attP2 site at 68A4,^20,22 and nanos-phiC31 integrase on X.²³ Stable transformant lines independently carrying insertions of DHC64C^wt, DHC64C^M1, DHC64C^M2 or DHC64C^M3 on chromosome 3 were successfully isolated from each injection (average 1 transformant from 150 injected). Transformants from each injection mapped exclusively to the third chromosome and every insert had an eye color indistinguishable from the other transformants; therefore, we concluded that the insert targeted correctly to the identical attP location in each line. To confirm correct targeting, PCR across the phiC31 target sites in all transformant lines confirmed loss of attP and attB products, and generation of attR and attL sites. Subsequent experiments have shown that all the transformant lines are capable of rescuing the mutant phenotype at permissive temperatures (data not shown).

Discussion

Our goal was to generate Drosophila transformant lines carrying specific small mutations of DHC64C. The mutations (Table 3) were designed in predicted P-loop sequences homologous to those previously shown to generate temperature sensitive (ts) alleles in C. elegans dhc-1.²⁴ Minimally each mutation required generation of a single base pair change which would result in a glycine (G) to glutamic acid (E) substitution in the fourth amino acid of the AAA3 domain of the dynein (DHC64C^M1). However, because the worm and Drosophila sequences vary in the “x” positions at two other sites in this P-loop, two additional mutations were generated to more closely replicate the worm sequence, which produced the ts allele. DHC64C^M2 includes a third residue A to P mutation in addition to the G to E at amino acid 4, and DHC64C^M3 expands again on DHC64C^M2 by adding a T to S mutation at residue 5 of the P-loop motif (Table 3).

Table 3.

Comparison of the AAA3 P-loop sequence in C. elegans and D. melanogaster

Generic	Worm	Drosophila
GxxGxGKT/S	GPPGSGKT (wild-type)	GPAGTGKT (wild-type)
ts mutants	GPPESGKT (Schmidt, et al.24)	GPAET GKT (M1)
		GPPET GKT (M2)
		GPPES GKT (M3)

Mutations 1–3 were designed to replicate the sequence of the ts mutation published in worm. The amino acid changes produced by the mutations are underlined.

To incorporate the mutations into a P-element construct for transformation, a combined SDM and two-step P[acman] technique was developed. Serial recombineering techniques for cloning by gap repair have been described previously (reviewed in refs. ^{12, 25} and ²⁶) for use in situations when target genes span two BAC clones. The MSSMR technique provided here expands on that concept by showing that serial recombineering can be effectively and efficiently used to produce precise mutations in large target genes.

One obstacle to applying MSSMR to a gene of interest is identifying appropriate restriction enzyme sites for the cloning and recombineering steps. Based on published sequences for the attB-P[acman] vectors,¹² the multiple cloning site (MCS) contains five restriction sites with non-redundant recognition sequences (AsiSI, PacI, SwaI, NotI and AscI), one of which (SwaI) produces a blunt cut.²⁷ In addition, there are at least nine commercially available restriction enzymes that are predicted not to cut the attB-P[acman] vectors (AatII, AvrII, BamHI, BbvCI, BmtI, BsiWI, MluI, NheI and ZraI).²⁷ This list does not include homing endonucleases, which could also be used with additional testing. The MSSMR experiment must be planned such that three criteria are met: (1) the three homology arms can be assembled in the attB-P[acman] MCS; (2) after the three homology arms are assembled in the attB-P[acman] vector, the resulting recombineering construct can be linearized at both the left/mutation and mutation/right arm joints; and (3) it is possible to linearize the product of the first recombineering reaction at EITHER the left/mutation or mutation/right arm joint. As with any technique that relies on a restriction enzyme step, it is possible that an appropriate combination may be impossible for a specific gene. However, the need for only two non-cutting enzymes to simultaneously clone and mutate a gene over 30 kb makes the MSSMR technique applicable to a plethora of targets.

Mutation potential of MSSMR.

The extent and variety of mutations that can be tolerated in the MSSMR technique remains to be determined. As shown above, when nested within a ∼700 bp homology arm, the nine amino acid sequence GCC GGA ACT exhibited sufficient surrounding homology with CCC GAA TCT to allow efficient recombination and mutation of three consecutive codons. Larger numbers of base substitutions, as well as the addition or deletion of base pairs through size-mismatched homology arms is an intriguing, but untested, possibility. It is also likely that the length of the M homology arm can be substantially reduced, based on previously published recombineering data (reviewed in refs. ^{26, 28} and ²⁹). MSSMR as described here demonstrates a two-step experiment for the mutation of a motif in the middle of a large gene. However, given the efficiency of mutation at each individual recombineering step, it seems likely that single-step reactions in which the mismatch is contained in either (or both) homology arm of a traditional two-arm recombineering reaction is also feasible.

There may be several factors working against the incorporation of mutations in recombinant products. Costantino and Court³⁰ present information on the efficiency of methyl-directed mismatch repair (MMR) system in E. coli which functions to inhibit incorporation of mutations during recombination. This system is functional in the SW102 strain, and the three base mismatches induced in our M3 construct are among those reportedly efficiently repaired by the MMR system.^30,31 Our ability to efficiently overcome the repair system may lie in the length of the mutation homology arm as well as the location of the mutation in the approximate middle of the arm, rather than at the end. Similarly, the decreased efficiency of recombination discussed by Muyrers et al.³² when using linear fragments containing base pair mismatches in homology arms was not observed to the extent that it precluded generation of our intended product.

Advantages of MSSMR in Drosophila.

The many tools available for, and benefits of, recombineering in the Drosophila model system are summarized elsewhere (reviewed in refs. ¹² and ²³). Using P-element vectors and attP fly lines compatible with the attB/phiC31 integrase targeting system²⁰ was particularly beneficial for this study and for MSSMR in general. Our wild-type and mutant rescue constructs were stably inserted³³ in identical genomic locations in transgenic lines, thereby avoiding position effects of random P-element insertions that could confound our interpretation of mutant phenotypes relative to each other and to the wild-type control.

There have been many recombineering techniques published for generating and screening potential mutations in BACs, which are applicable to the Drosophila model system. Muyrers et al.³² induced point mutations via insertion and subsequent removal of a Sac/neo cassette. Swaminathan et al. introduced MAMA-PCR screening for recombinants and Yang and Sharan²⁸ have published a “hit and fix” method for creating single base mutations, deletions and insertions in BACs using oligonucleotides. Sopher and LaSpada's²⁶ seamless “hybrid recombineering” protocol was the closest approximation to our needs, but relied on SacB selection and RecA function encoded by cassettes expressed in the recombineering vector, while it lacked the P-element vector and targeting capabilities of the P[acman] system. MSSMR increases the efficiency of all these techniques by combining the mutagenesis step with direct transformation vector cloning, rather than mutating a BAC prior to cloning.

Another advantage of creating mutations in vitro through recombineering vs. in vivo via homologous recombination is that it avoids manipulations in fly strains with reduced viability. That is, post-transformation fly crosses and in vivo recombinase steps required by Drosophila Cre/lox-,³⁵ FLP/FRT-,³⁶ and phiC31-based³⁷ recombinase-mediated cassette exchange (RMCE) methods are not needed. Finally, the genomic “side effects” of tandem duplications or partial deletions produced by homologous recombination techniques are avoided.⁷

It should be noted that while we chose the P[acman] system due to the size of DHC64C and our familiarity with it, other vectors and targets could theoretically be used. P[acman] vectors are based on a BAC backbone containing an inducible promotor, allowing them to incorporate and maintain large insertions at low copy number, while still permitting conditional growth at high copy number.¹² The recombineering events described above for MSSMR are dependent upon homologous recombination proteins conditionally expressed by the SW102 bacterial chromosome (for a description of the defective prophage system, reviewed in ref. 15) and many vector/target combinations can be used. Therefore, while MSSMR is extremely efficient for precise mutagenesis of large genes, it could be used with any gene of interest for which cloning and/or mutagenesis is incompatible with other techniques. Recently, Drosophila germline transformation vectors compatible with Gateway cloning technology, incorporating N- and C-terminal protein tags and with³⁸ or without³⁹ phiC31 integrase targeting elements have been published. While not specifically designed for recombineering, the added features of these vectors could make them powerful adjuncts.

Drosophila fosmid⁴⁰ and P[acman]¹⁴ libraries provide exciting possibilities for MSSMR. These preparations contain 20–80 kB genomic insertions, have published protocols for the addition of tagging cassettes, and contain attB sequences for targeted insertions. Given the much higher efficiency of MSSMR with smaller targets (Table 2), these libraries are a great alternative to (generally larger) BACs. In addition, these libraries could allow tags to be initially added to the wild-type clone, followed by MSSMR to recover tagged mutants simply by creating homology arms based on the vector sequence either in combination with or instead of, the insert region.

Beyond Drosophila.

The MSSMR technique is also applicable outside the Drosophila model system. For instance, Rozwadowski et al.⁴⁷ combined plant transformation vectors, the Gateway system, and recombineering techniques to tag and express the Arabidopsis gene PAP85. Recombineering is routinely used to facilitate the generation of transgenic mice. Generally, murine BACs are modified via introduction and removal of selection cassettes using Cre/loxP or Flp/FRT systems.^15,29,42,43 MSSMR could streamline mutagenesis and construct assembly in both of these transgenic model systems, especially when several closely related mutant alleles are desired.

Tools for the creation of transgenics are becoming available in an increasing number of species and models. The utility of phiC31 integrase to localize constructs to site-specific targets has now been demonstrated not only in Drosophila, but also in mouse ES cells,⁴⁴ human ES cell lines,⁴⁵ and zebrafish embryos.⁴⁶ As this method of gene targetting becomes accessible in more model systems, the ability to analyze the effects of very subtle mutations may be possible in ways never before envisioned. MSSMR can be a valuable tool for efficiently creating mutations in even the largest of genes when a target BAC is available.

In the last 10 years many powerful and elegant recombination-based strategies for mutagenesis have been published. MSSMR provides another tool when pinpoint disruption of genes, especially large genes, is the goal. The combination of high resolution mutagenesis and low disruption of genetic background creates an excellent platform for the study of the function of motifs, domains or binding sites in a way not previously possible.

Materials and Methods

Reagents.

Primers used for the amplification of homology arms, SDM, colony PCR and sequencing are listed in Table 1. The SW102 E. coli strain was provided by the NCI-Frederick (http://web.ncifcrf.gov/research/brb/recombineeringInformation.aspx). attB-P[acman]-Amp^R and Chl^R vectors (Accession numbers EF106980 and EF106979, respectively) were acquired from the Drosophila Genomic Resource Center (https://dgrc.cgb.indiana.edu). The target BAC containing the CG17150 genomic sequence (BACR17L24; accession number AC108877) was obtained from BACPAC (BACPAC Resources Center at Children's Hospital Oakland Research Institute, Oakland, California; http://bacpac.chori.org).

Flies.

Injections were performed by Model System Genomics (Duke University) into embryos of genotype y[1] sc[1] v[1] P{y[+t7.7] = nos-phiC31\int.NLS}X; P{y[+t7.7] = CaryP}attP2 (Insertion site 68A4; FlyBase ID FBst0025710; Bloomington Stock number 25710).

Gene model.

The primer and mutation designs for CG17150 were based on the published sequence at accession number AE003482.4 (GI:55380499).

Site-directed mutatgenesis.

The QuikChange II Site-Directed Mutagenesis Kit (#200523; Stratagene) was used per protocol to generate targeted mutations within the Mutation arm. Primer sequences are listed in Table 1.

Recombineering.

Recombineering steps specific to the P[acman] system were performed as outlined in Venken et al.¹² Preparation of competent cells proceeded as outlined in “Recombineering Protocol 1” available from the NCI Frederick recombineering website: http://recombineering.ncifcrf.gov/Protocol.asp. A complete MSSMR protocol is provided as Supplementary Protocol 1.

Abbreviations

DHC

dynein heavy chain

ts

temperature sensitive

BAC

bacterial artificial chromosome

SDM

site directed mutagenesis

MSSMR

Mutagenesis via Serial Small Mismatch Recombineering

MMR

methyl-directed mismatch repair

ES

embryonic stem