100 μg of pCRB-LoxX-Amp-LoxX variant plasmids were digested with HindIII and SpeI to linearize the template. Digests were precipitated in isopropanol, washed in 70% ethanol, dried, and resuspended in Tris-EDTA buffer pH 8. Linearization was verified by gel electrophoresis. After quantifying template concentration via UV spectrophotometry, all templates were diluted to 450 ng/μl. 5 μg of purified template was then incubated in Cre reaction buffer with 20 U of Cre recombinase at 37°C replicated five times. Samples were removed at time point 0, 1 m, 2 m, 5 m, 10 m, 15 m, 30 m, 1 hr, and 2 hr combined with loading dye including proteinase K and frozen. Prior to gel electrophoresis, samples were treated with proteinase K for 2 hr at 37°C. After electrophoresis gel bands were recorded on a Typhoon imaging system and quantitated with ImageQuant TL v2003.02 software. Signal intensities were normalized to fragment length and were expressed +/− SEM. Five replicate samples were analyzed for each time point.

MMP5 is a BAC containing a region of the mouse HoxB complex (Beginning sequence of MMP-5 aagcttcacatcagccacggtaattctccatctctttttt End sequence of MMP-5 atgggctggtggctccaaagggcccccagaaagctt. UCSC Genome Brower Coordinates (as of 10/1/10): Chromosome 11: 96165471-96309155) isolated from a library prepared by partial Hind III digest of genomic DNA cloned into the pBelo backbone. DNA encoding the different H2B fluorescent proteins was inserted into the respective HoxB gene at the position of amino acid 35 in the predicted Hox protein sequence. In all cases, this would create a fusion protein containing the first 35 AA of the Hox protein and the H2B-FP domain. H2B-mcherry was inserted in Hoxb1, H2B-YFP (Venus) into Hoxb2, H2B-CFP (Cerulean) into Hoxb3, and H2B-EGFP into Hoxb4.

Primers common to all of the fluorochrome bearing selection cassettes were used for amplification. For targeting 50 bp of targeting homology was added to each of the base primers. Targeting fragments were generated by PCR using oligos (50 bp homology + amplification primer (primer list 2.12)) and Platinum Pfx polymerase PCR system (Invitrogen). Cycling parameters were 95°C × 10 min and then 30 cycles of 95°C × 30 sec, 64–68°C × 30 sec, and 68°C for 4 min followed by a final step of 68°C for 10 min. After amplification PCR fragments were digested with DpnI enzyme for 1 hr at 37°C and then either used directly or gel purified using Fermentas DNA extraction kit as per the manufacturer's directions.

BAC harboring recombineering strains of bacteria (EL350 or SW106) [4, 10] were grown overnight in 3 ml of LB broth at 32°C shaking at 200 rpm–250 rpm. One ml of this culture was used to inoculate 50 ml of LB and grown (32°C shaking 200–250 rpm) 2–3 hr until mid-log phase growth (ABS @600 nm of 0.4–0.6). Bacteria were then pelleted by centrifugation for 6 min at 5500 g, washed 2 times with 100 ml ice cold dH₂0, washed a final time in 2 ml cold dH₂0, and resuspended in 550 μl of cold dH₂0. 0.5–2 μl of purified targeting fragment was electroporated into 50 μl of electrocompetent bacteria (1.8 kV, 25 μF, 200 ohm). Electroporated bacteria were allowed to recover with 2–4 hr of growth in SOC medium at 32°C shaking 200–250 rpm. Bacteria were then plated with the appropriate antibiotic selection for 18–24 hr.

Recombined bacteria were first screened by colony PCR. Colonies were picked, stab cultured, and inoculated into a PCR reaction containing a primer inside and outside the targeting fragment. PCR reactions were done in 10 μl volumes. Cycling parameters were 95°C × 10 min and then 30 cycles of 95°C × 30 sec, 55°C × 30 sec, and 72°C for 40 sec followed by a final step of 72°C for 10 min. Colonies producing PCR bands of the appropriate size were selected for digest analysis.

Stab cultures of clones producing appropriate PCR bands were used to inoculate 5 ml LB overnight cultures. From these cultures, 800 μl was used to produce glycerol stocks of each clone, and 4 ml was used for BAC isolation via alkaline lysis [11]. Isolated BAC pellets were dissolved in 19 μl of digestion buffer and incubated at 55°C for 1 hr. One μl of NheI enzyme was then added to each isolate and digested for 1–4 hrs. Digests were then loaded on a 25 cm 0.8%–1% agarose gel and run at 40–50 V overnight. Banding patterns were then recorded by gel documentation systems or on a Typhoon imaging system.

Streak plates made from glycerol stocks of recombined BAC clones were incubated overnight at 32°C. Individual colonies were grown in LB + 0.1% L-arabinose for 1 hr and plated with chloramphenicol only selection. The resultant clones were analyzed by digestion analysis (Section 2.5), and positive clones were then subjected to targeted recombination [4].

Lox flanked Ampicillin cassettes generated in Section 2.1 were transferred into the pBelo BAC backbone in the EL350 strain by recombineering. Three clones were selected for each transfer and pBelo + cassette was purified from each clone and sequenced. Clones were then grown overnight at 32°C in LB + 0.1% L-arabinose and plated. Resultant clones were analyzed by colony PCR (Section 2.4) with multiplex PCR capable of detecting all reactants and products. One hundred forty-four clones were analyzed for each construct.

After two sequential rounds of targeted recombination using PCR generated targeting fragments containing two different antibiotic selection cassettes (completed as described in Section 2.2), resultant clones were subjected to simultaneous selection cassette removal (completed as described in Section 2.6).

Labeled fragments used for FCS experiments were generated by PCR using plasmids bearing the respective LoxX—Ampicillin—LoxX selection cassettes as templates. T7 primer and a (−) strand primer at the 5^' end Ampicillin (MEP-047) were used to produce a 360 bp amplicon which was subsequently reduced to 142 bp by digestion with NheI restriction enzyme. After column purification, complete digestion was verified by gel electrophoresis before utilization in FCS experiments.

Fluorescence correlation spectroscopy (FCS) was performed with a Carl Zeiss LSM 510 META microscope incorporating a Confocor 3 FCS attachment and a 40x C-Apochromat water objective. For all experiments the 488 nm laser power was 0.05% within the Confocor software, corresponding to approximately 6 μW at the sample, and the pinhole was set at one airy unit (70 μm). The temperature within the microscope chamber was maintained at 37 ± 1°C. Samples were pipetted directly onto #1.5 coverslips which were placed directly on the objective with a small amount of water. Intensity trajectories were collected in either 5 or 20 sec segments for a total acquisition time of 120 sec. Raw data was converted into photons per 10 μs time period and correlated via a fast Fourier transform using custom software written in Java. Diffusion times were obtained by fitting the data using Levenberg-Marquardt nonlinear least squares using the following formula: (1)G(τ)=G(0)(1+τ/τD)1+(ω02/z02)(τ/τD), where τ_D is the diffusion time, ω₀ is the radial waist of the confocal focus, and z₀ is the axial waist of the focus [12]. The z₀/ω₀ ratio was fixed to a value of 5 for this analysis according to the recommendations of Hess and Webb [13]. Diffusion times were converted into hydrodynamic volumes using the following relationships: (2)D=ω024τD,  R=kT6πηD,  V=43πR3, where D is the diffusion coefficient, k is Boltzmann's constant, T is the temperature in Kelvin's, η is the viscosity, R is the hydrodynamic radius, and V is the hydrodynamic volume. The hydrodynamic volume scales with the molecular weight of the diffusing complex as long as there are no shape changes in the complex. This may not be the case for the Cre-oligonucleotide complex. Nevertheless, our analysis only requires that the bound complex has a different hydrodynamic volume than the DNA alone.

BAC DNA was purified from 500 ml of overnight culture by either Quiagen or NucleoBond maxi purification kits with the following modifications. Alkaline lysis volumes were increased from 10 ml to 50 ml, and elution from the column was done with elution buffer heated to 70°C.

Transgenic mice carrying modified BAC DNA constructs with the fluorescent proteins were generated using well-established methods [14–16]. Basically, BAC DNA was purified using a Qiagen Maxi-prep kit. The DNA was resuspended at 1 ng/μl and used for pronuclear injection into fertilized eggs from a cross of C57B6/JxCBA-F1 mice. Injected embryos were cultured overnight to the 2-cell stage and then transferred into the uterine horns bilaterally (10–20 embryos per side) of anesthetized pseudopregnant recipient hosts. Embryos were harvested from the hosts at 9.5 days postcoitum (dpc), and transgenic founder embryos (F₀) were identified by genotyping for GFP.

To determine whether the BAC manipulations generated the expected fluorescent proteins with appropriate activity in transgenic embryos, F₀ transgenic embryos were examined using fluorescent microscopy and established methods for multispectral detection [17–22]. We successfully detected the four fluorescent proteins, H2B-EGFP, H2B-CFP, H2B-YFP, and H2B-mCherry using a Leica UV Fluo 206 Stereoscope with appropriate filters. Global expression patterns of the reporters were scored in whole embryos, and hindbrain tissues were also dissected from the embryos and placed in flat-mount preparations to examine the specific restricted patterns of reporter expression in hindbrain segments. In both cases, the patterns of reporter activity observed correlated with previously described expression of the endogenous HoxB genes [23].

(see Tables 3, 4, and 5).

Initially to assess the possibility of utilizing multiple Lox sequences simultaneously, we evaluated the reaction rates of a series of individual Lox variants to verify that they were similar. Figure 1 illustrates the LoxP, Lox5171, Lox2722, and Loxm2 spacer variants and the Lox71 (L), Lox66 (R), and Lox72 inverted repeat variants [7, 9] which we evaluated. A linear template containing various Lox site pairs of was subjected to Cre-mediated recombination which produced a shorter linear fragment and an excised circle (Figure 2(a)). The progress of the reaction was measured by determining the ratio of the linear product/template as measured by gel electrophoresis and band quantitation. Samples were taken at various time points from t = 0 to t = 120 min. LoxP (WT) sites showed measurable recombination even at 1 min incubation with Cre at 37°C (Figure 2(b)). This proceeded over the 120 min time course to produce a ratio of 1.5. Lox5171 sites exhibited an even higher initial reaction rate (Figure 2(b)); however, at the end time-point the Lox5171 ratio was only 50.7% of the LoxP ratio (Table 1). Lox2722 and Loxm2 lagged behind LoxP and Lox5171 in the accumulation of Cre products (Figure 2(b)) however; at t = 120 min. their product ratios were 41 and 34% of the LoxP ratio (Table 1). Cre reactions for each of the Lox spacer variants produced substantial amounts of products over the time course, opening the opportunity for the use of multiple sites to generate products produced by each Lox spacer variant.

While the functionality of individual spacer variants has been previously demonstrated [9], the functionality and reaction rates of Lox spacer variants combined with the inverted repeat variants Lox71/66 (LR) have not been analyzed. We therefore generated linear DNA templates containing the spacer variants combined with LR-Lox variants to test the functionality and reaction rates of these Lox site combinations. The relative order of reaction rates compared with the spacer variants alone remained the same, LR-LoxP > LR-Lox5171 > LR-Lox2722 > LR-Loxm2 (Figure 2(c)). The Lox5171 spacer again had an initial reaction rate that was higher than LoxP but was not sustained. While LR-LoxP and LR-Lox5171 produced similar amounts of reaction products at t = 120 to LoxP without LR inverted repeat mutations, the disparity between LoxP and Lox2722/Loxm2 rates was enhanced when combined with LR Lox variants. LR-Lox2722 and LR-Loxm2 reaction progress ratios at t = 120 were only 5.1 and 2.3% of the LoxP ratio at the same time point (Table 2). In fact, inclusion of LR variations of the Lox site reduced the reaction rates for all of the spacer variants (Figures 2(c) and 2(d)). LR-Lox5171 was the least affected and the rates of Lox2722 and Loxm2 were reduced almost 10 folds compared to the respective Lox spacer variant alone (Table 2).

It has been shown that Cre recombinase binds as a dimer to each Lox site cooperatively with an effective affinity of 7.4 × 10⁻¹⁰ M and induces an asymmetric bend in Lox site containing DNA [24, 25]. If Cre had an altered affinity for the different Lox sites or was less effective at bending the DNA, this might result in altered reaction rates for Lox site variants. To investigate this possibility, we generated DNA fragments containing a single Lox site labeled at one end with Alexa Fluor 488 (Figure 3(a)). Fluorescence correlation spectroscopy (FCS) was used to measure the diffusion coefficients for Lox site bearing fragments in the presence or absence of Cre recombinase (Figure 3(b)). Labeled DNA fragments alone had a diffusion rate of 360 μsec. DNA fragments with Lox spacer variants LoxP, 5171, 2722, and m2 all had similar diffusion rates ranging between 453–488 μsec in the presence of Cre. In contrast, all Lox site variants that included the L-Lox inverted repeat variant had an increased diffusion rate, suggesting a reduced binding of Cre and/or inhibition of Cre-mediated bending (Figure 3(b)). Examination of the concentration dependence of this effect on LoxP with or without the L-Lox inverted repeat variation showed a concentration dependence of the L-Lox effect on the diffusion rate in the presence of Cre (Figure 3(c)). This indicates that the binding of Cre to L-Lox is impaired rather than the ability of Cre to induce a bend in the Lox containing DNA, since bending should independent of Cre concentration.

Exploiting multiple Lox variants opens the possibility of multiple manipulations of the same BAC or the simultaneous manipulation of multiple loci. To test the feasibility of serial modifications, we have used a mouse BAC, MMP5, which contains a portion of the mouse HoxB cluster as a substrate (Figure 4(a)). The objective was to sequentially insert each of the four different fluorescent protein reporters (H2B-EGFP (GFP), H2B-Cerulean (CFP), H2B-Venus (YFP), and H2B-mCherry) into four adjacent Hoxb genes and assay for predicted functional activity in transgenic mice to evaluate success of the approach (Figure 4(a)).

We have used a PCR strategy to generate the series of targeting fragments that encode four different fluorochrome reporter proteins. Figure 4(b) illustrates the strategy for generating and applying a fragment with specific homology arms to Hoxb4 and containing green fluorescent protein (eGFP). In a similar manner, specific homology arms were used for targeting H2B-mcherry to Hoxb1, H2B-YFP to Hoxb2, and H2B-CFP to Hoxb3. Each insertion was designed to create a fusion protein with the first 35 amino acids of the respective Hox protein whose expression would be under control of the endogenous promoter.

Each fluorochrome was inserted into a locus by homologous recombination stimulated by temperature regulated recombination machinery in the recombineering strains EL350 or SW106 [4, 10]. Positive clones from each step were selected for using Kanamycin resistance conferred by the selection cassette at the 3^' end of each fluorochrome targeting fragment (Figure 4(b)). Correct targeting events were identified by colony PCR, and potential positives were subjected to a secondary validation via a NheI digest of purified BAC DNA. Figure 4(c) shows the results for targeting H2B-Venus into the Hoxb2 gene and removal of the selection cassette, in a version of MMP5 which already contained H2B-mCherry in the Hoxb1 locus. By this method, the presence of a single, correctly targeted insertion could be verified, and the integrity of the remainder of the BAC could be confirmed by comparing the predicted banding patterns of the product BAC with that of the initial BAC used for recombination.

Utilizing this approach, we were able to successfully generate a HoxB BAC containing the four different fluorescent protein reporters at specific sites by serial rounds of targeted homologous recombination, each followed by Cre-mediated excision of the selection cassette (Figure 4). The intermediate stages and final BAC were verified by sequencing across homology regions and coding regions, as occasionally errors occur due to error rates in the generation of long primers or as a result of the induction of recombination machinery.

We next tested the functional activity of the four targeted fluorescent protein reporters by generating transgenic mice carrying the modified MMP5 BAC. A 9.5 dpc F₀ mouse founder embryo carrying the MMP5 BAC transgene is shown in Figure 4(a). The bright field view shows the entire embryo. The fluorescent illuminations display of the same embryos shows nested and spatially restricted patterns of expression. This functionally demonstrates that each of the serially targeted reporter fragments inserted into the proper location and displayed an expression pattern appropriate for its site in the HoxB cluster. Furthermore, flat-mount preparations of the dissected hindbrain from this embryo display reporter activity in rhombomeric segments that also correlate with those expected for the endogenous genes (Figure 4(a)) [23].

In the serial targeting experiments, while we exploited the expected preference against utilization of Lox72 (the result of L-Lox × R-Lox recombination) sites in recombination, such reactions did occur. During the fourth round of Cre-mediated selection cassette excision in which the attempted reaction (L-LoxP × R-LoxP) was challenged with two additional Lox72 sites and a Lox72-5171 site present within the BAC, many of the BAC products resulting from this reaction exhibited recombination between Lox72 and the L-LoxP or R-LoxP sites. Two of 17 clones displayed the desired banding pattern while 10 of 17 had NheI digest patterns consistent with utilization of the Lox72 site in the recombination (data not shown).

In order for multiple Lox variants to be used simultaneously, Cre-mediated recombination must be highly specific for recombination between homologous spacer variants. The specificity of Lox5171 and Lox2722 has been partially compared in vitro [9] however, a detailed comparison of the four sites against each other has not been done. Therefore, we induced Cre expression in vivo in a single-copy template system challenging variants against each other in the presence of a wt LoxP site (Figure 5(a)). These experiments were done in a single-copy format because Cre reactions done in vitro or in multicopy plasmids resulted in complex inter-molecular reaction products. As each template is challenged with Cre three products are possible and multiplex PCR with four primers distinguishes among these events (Figure 5(b)). After overnight L-arabinose Cre induction, clones were plated and colony PCR was performed to identify the products present in each clone (Figure 5(c)).

While some promiscuity was observed between Lox72 and Lox66/71 inverted repeat Lox site variants, Cre-mediated recombination between nonhomologous spacers was rare. All possible combinations of Lox spacer variants were tested in this manner (Table 2). No events of nonhomologous recombination were observed for Lox sites 2722 and 5171. One instance of a LoxP × Loxm2 recombination, however, was recorded (Table 2). Given that all spacer variant pairs tested against each other were done in the presence of a wt LoxP site, the resultant rate of LoxP × Loxm2 hetero-specific reactions was rare at a frequency of 1/566.

Given the selectivity of Lox spacer variants for homo-specific Cre-mediated recombination, if multiple selection cassettes are available, simultaneous Cre excision of such cassettes should be possible allowing more rapid generation of multiply targeted constructs. To verify the feasibility of this technique, we inserted two different PCR generated fluorochrome bearing targeting cassettes to Hoxb1 and Hoxb2 in the BAC MMP5 (Figure 6(a)). A H2B-EGFP fusion protein was targeted into the Hoxb1 locus using an L-LoxP Kanamycin R-LoxP, and H2B-DsRed was targeted into the Hoxb2 locus using a L-Lox-5171 Ampicillin R-Lox-5171 resistance cassette. Subsequently, selection cassette removal was achieved by L-arabinose induced Cre expression, and resultant clones were analyzed by their NheI banding pattern. Of the 28 clones analyzed, 23 had recombined in the expected manner while 5 remained in an unrecombined state (Figure 6(b)). No clones exhibited bands indicative of a LoxP × Lox5171 reaction. Lox spacer variants can thus direct very efficient and specific simultaneous recombination.