Antibiotic-free Selection - Introduction

The cell wall is essential for every living bacterium as it ensures stability and structure, protection against osmotic pressure and regulation of molecule transport. The composition of the cell wall differs among differnt bacteria species, a feature which is commonly used to classify taxonomy. The most common classification is based on the Gram-staining into Gram-negative, for example Escherichia coli and Gram-positive bacteria like Bacillus subtillis. Gram-negative bacteria are characterized by an inner plasma membrane, a thin peptidoglycan layer, periplasmatic spaces and the outer membrane (Figure 1). In contrast Gram-positive bacteria generally lack the outer membrane but consist of a thicker peptidoglycan layer.
Hence, the peptidoglycan layer is an interesting target to control bacterial cell dvision. Peptidoglycan itself is a polymer consisting of a linear chain of polysaccharides and short peptides. The polysaccharides component is composed of alternating residues of beta-(1,4) linked N-acetylglucosamine and N-acetylmuramic acid. They are cross-linked in E. coli by a tetra-peptide of L-alanine, D-glutamic acid, meso-diaminopimelic acid and finally D-alanine. The cross-linkage is thereby realized by a transpeptide-linkage of meso-diaminopimelic acid and D-alanine (Cava et al., 2011).

Figure 1: Structure of the bacterial cell wall of a Gram-negative bacteria. The peptidoglycane layer consist of altering N-acetylglucosamine and N-acetylmuramic acid, which are crosslinked by a tetra-peptide. The crosslinkage is realized by D-alanine. Therefore, D-Alanin is an interesting target to control bacterial cell division.

Bacteria with a missing cross-linkage are not able to perform cell division. Cells would lysate instead because of the broken peptidoglycane layer. One possibility to prevent the crosslinkage is the use of a ß-lactam antibiotic, like penicilin. Penicillin inhibits the enzyme D-transpeptidase which is responsible for the cross linkage. In contrast to that another possibility is the prevention of the D-alanine-synthesis in the cell itself. Bacteria without D-alanine supplementation will lyse during cell division.
The synthesis of D-alanin in E. coli can be catalyzed by an alanine racemase (EC 5.1.1.1). This enzyme enables the reversible reaction from L-alanine into the enantiomer D-alanine. This reaction requires the cofactor pyridoxal-5'-phosphate (PLP) as shown in Figure 2. E. coli posseses two alanine racemases. One is encoded by alr and constitutively expressed and therefore responsible for the accumulation of D-alanine. The other one is encoded by dadX, under the control of the dad-operon and usually used in catabolism (Walsh, 1989).

Figure 2: The alanine racemase Alr from E. coli catalyses the reversible reaction from L-alanine to D-alanine, whereby Pyridoxal-5'-phosphate (PLP) is an essential cofactor for this reaction.

The deletion of the constitutive alanine racemase (alr) and the catabolic alanine racemase (dadX) in the genome will lead to a strict dependence on D-alanine. A bacterium with such a double deletion is only able to grow on media with D-alanine supplementation or with a complementation of the alanine racemase on a separate plasmid like BBa_K1465401. This approach can be used for an antibiotic-free selection system and even for molecular cloning without antibiotics.

Transformation efficiency

The double deletion of the constitutive alanine racemase (alr) and the catabolic alanine racemase (dadX) in the DH5α and KRX strain, resulting in the phenotypes KRX Δalr ΔdadX and DH5α Δalr ΔdadX respectively. This leads to a strict D-alanine auxotrophy, which can be complemented by a plasmid carrying one of the alanine racemases or D-alanine supplementation in the media.
The plasmid BBa_K1465401 which is used for the complementation carries the constitutive alanine racemase alr and the chloramphenicol acetyltransferase for the chloramphenicol-resistance (Cm). Besides the plasmid contains RFP BBa_J04450 as a reporter for the characterization of the complementation efficiency.
To investigate the antibiotic-free selection the plasmid was transformed in different concentrations into electrocompetent cells of DH5α Δalr ΔdadX. Due to the fact, that the cell can store D-alanine for a certain periode of time the incubation of the cells after the transformation was performed in normal SOC-media without supplementation and in SOC-media with supplementation of 3 mM D-alanine to maintain the cell viability of the transformants before they are able to express the alanine racemase which is essential for the accumulation of D-alanine and bacterial growth. After the incubation the cells were stread out either onto LB plates for the antibiotic-free selection via complementation of the D-alanine auxotrophy or onto LB plates containing 30 mg/L chlormaphenicol as control. The successful transformation can be verified by the red color of the reporter RFP, while false-positive colonies remain white.
An example for the transformation of 0.33 ng plasmid BBa_K1465401 into the E. coli strain DH5α Δalr ΔdadX is shown in Figure 3. The upper row shows different volumes from 50 µl, 100 µl and 200 µl were spread onto LB containing chloramphenicol and the lower row shows the same volumes were spread out onto normal LB for the selection without any antibiotic. It appears that the antibiotic-free selection is more efficient than the selection with chloramphenicol because more colonies were observed on each plate with equal volume plated.

Figure 3: Colony forming units (cfu) for the selection with chloramphenicol (upper row) and antibiotic-free selection (lower row). 0.33 ng of the plasmid BBa_K1465401 was transformed into the E. coli strain DH5α Δalr ΔdadX, incubated in SOC-Media supplemented with 3 mM D-alanine and spread out in different volumes from 50 µl (left), 100 µl (middle) and 200 µl (right).

Figure 3 and 4 show the colony forming units (cfu) for the transformation of 0.52 ng plasmid BBa_K1465401 at one hour of incubation in either normal SOC-medium or SOC-medium supplemented with 3 mM D-alanine. First of all it is possible to see, that the incubation of the transformation can be performed without the supplementation of D-alanine (blue bars), even though there are not as many colonies compared to the incubation with 3 mM D-alanine (red bars). Besides, the graphics show that in the absent of D-alanine more false-positive phenotypic white colonies occur (cyan bars) compared to the samples with supplementation of D-alanine (purple bar). In some cases the absence of D-alanin also leads to huge contamination an higher amount of false-positive colonies than correct colonies. This effect can be clearly seen by spreading 200 µl of the transformants incubated in SOC-Media without D-alanine.
The most interesting aspect turns out by the comparison between the chloramphenicol selection and the antibiotic-free selection (AB-free). It can be clearly seen that the amount of positive red colonies is 2.88 ± 0.45 (3 mM D-alanine supplementation) and 5.91 ± 1.53 (without D-alanine) fold higher compared to the selection with the antibiotic chloramphenicol. This might be due to the active inhibition of chlormaphenicol, whereas the lack of D-alanine can be compensated by intracellular storage of D-alanine or eventually by temporary grow arrest, until the alanine racemase is expressed.

Figure 4: Colony forming units (cfu) for the selection with Chloramphenicol. 0.52 ng of the plasmid BBa_K1465401 were transformed into the E. coli strain DH5α Δalr ΔdadX and incubated in normal SOC-Media or SOC-Media supplemented with 3 mM D-alanine. After one hour different volumes were spread onto LB plates containing 30 mg/L chloramphenicol (n = 2 x 2).

Figure 5: Colony forming units (cfu) for the antibiotic-free selection with LB. 0.52 ng of the plasmid BBa_K1465401 was transformed into the E. coli strain DH5α Δalr ΔdadX and incubated in normal SOC-media or SOC-Media supplemented with 3 mM D-alanine. After one hour different volumes were spread onto normal LB plates (n = 2 x 2).

Another important aspect of an antibiotic-free selection approach is the quotient of false-positive and particularly the portion of revertants which could also reduce this quotient in an antibiotic-free selection. Therefore, the ratio of false-positive transformants (white color) to the positive transformants (red color) are shown in Figure 6. Besides, it should be mentioned that the false-positive transformants in the antibiotic-free selection can be also observed for chloramphenicol selection. These false-positive transformants could be bacteria with a spontaneous chloramphenicol resistance, contamination by bacteria with a chloramphenicol resistance or bacteria carrying a mutation within the expression of RFP (BBa_J04450) leading also to a white phenotype. Within the antibiotic-free selection an average portion of 2.83 % ± 0.09 false-positive transformants has been identified, while the antibiotic-selection with chloramphenicol only shows an average of 1.47 % ± 0.77 false-positive transformants. The slightly higher rate of false-positive transformants of the antibiotic-free selection might due to some revertants which are able to accumulate D-alanine without any alanine racemase by mutation and side reaction of another enzyme. Because of the higher transformation efficiency this effect is actually negligible.

Figure 6: Portion of false-positive colonies of antibiotic selection using chlormaphenicol (red bars) and the antibiotic-free approach (orange bars). The false positve colonies have been identified by their white phenotype. For the antibiotic selection an average of 1.47 % ± 0.77 false-positive colonies was investigated, whereas the portion amount for the antibiotic-free selection system is of to 2.83 % ± 0.09.

To obtain more reliable data, the plasmid BBa_K1465401 was transformed in different concentrations (0.33 ng; 0.52 ng and 0.65 ng) and transformed into the D-alanine auxotrophic E. coli strain DH5α Δalr ΔdadX in various volumes (50 µl; 100 µl and 200 µl) to estimate a valid transformation efficiency by using equation (1):
Transformation efficiency [cfu/ng] = colony forming units / (Spread volume [µl]/total volume [µl])*Amount of Plasmid-DNA [ng]
The estimated transformation efficiency for the transformation of 0.52 ng of BBa_K1465401 for the distinct volumes is shown in Figure 7 and the average for the transformation of different amounts of DNA is shown in Figure 8. The result emphasizes, that the antibiotic-free selection is significantly more efficient than the selection with the antibiotic chloramphenicol due to the inhibition of E. coli by the antibiotic. Another important aspect is the reduced transformation efficiency without the supplementation of D-alanine due to the lytic effect when D-alanine is lacking and the cross-linkage of the peptidoglycane can not be performed by D-alanin lacking cells. Even though some cells show a growth.

Figure 7: Comparision of the transformation efficiency for the transformation of 0.52 ng BBa_K1465401. The incubation was performed either in normal SOC media or in SOC media supplemented with 3 mM D-alanine and the colony forming units were counted on LB_Cm as well as normal LB for the antibiotic-free selection (n = 2 x 2).

Figure 8: Comparision of the transformation efficiency for different amount of the plasmid BBa_K1465401. The incubation was performed either in normal SOC media or in SOC media supplemented with 3 mM D-alanine and the colony forming units were counted on LB_Cm as well as normal LB for the antibiotic-free selection (n = 2 x 2).

Transformation process

In order to further investigate the transformation process and to prove if the higher transformation efficiency is due to the temporary storage of D-alanine and to the lacking inhibition by an appropriate antibiotic the bacterial growth of the E. coli strain DH5α Δalr ΔdadX was analyzed during the incubation in the SOC media. Therefore, small volumes of 40 µl were continuously plated in a time interval of 15 min. The samples were spread in parallel onto normal LB-Agar for an antibiotic-free selection and onto LB containing 30 mg/L chloramphenicol for the classical selection with antibiotic.
As demonstrated in Figure 9 the higher transformation efficiency can be explained by the lacking inhibition of a corresponding antibiotic, because just after the transformation there are no colonies observable with antibiotic-selection. In contrast to that there are positive red colonies on the LB plates. This might be due to the quick over-inoculation directly after the transformation. The chloramphenicol acetyltransferase, which normally prevents the antibacterial effect of chloramphenicol to bind to the 50S ribosomale subunit, is not been translated and functionally expressed so far. Thus the chloramphenicol stops the bacterial growth. Whereas, cells spread onto LB for the antibiotic-free selection still accumulate some D-alanine from the SOC media, which enables a short-term growth without the complementation of the alanine racemase from the plasmid BBa_K1465401. A stop of bacterial cell division could be also possible until the alanine racemase is functionally expressed and the cell can start growing.

Figure 9: Colony forming during the transformation by using antibiotic-free selection (black) or antibiotic-selection with chlormaphenicol (red) for the positive red colonies carrying the plasmid BBa_K1465401. It can be demonstrated that the antibiotic-free selection is more efficient and requires less incubation time.

In addition to the higher transformation efficiency of the antibiotic-free selection system, it can be observed that the incubation time can be reduced by the antibiotic-free selection. A high transformation efficiency can be already observed shortly after the transformation. Besides the transformation efficiency does not increase with the incubation time, because the raising colony forming units in both curves are due to the bacterial growth in the SOC media. Another important aspect is the quotient of false-positives to identify the most suitable moment for a screening with the lowest possible portion of false colonies, phenotypically white colonies in this case. The quotient of false-positive white colonies was always about 2 %. An incubation from 30 to 60 min is advisable for the antibiotic-free selection, because afterwards the higher cfu is only due to the bacterial growth in SOC media and therefore due to the higher risk of contamination not advisable.

Long-term stability of the antiobiotic-free selection

To verify if the plasmid stability of the antibiotic-free selection system is comparable to an appropriate antibiotic selection system, E. coli carrying the plasmid BBa_K1465401 was cultivated for 40 h in LB media, containing normal LB, LB supplemented with 5 mM D-alanine and LB containing 30 mg/L chloramphenicol. After reaching the stationary phase of the cultivation the culture was diluted to an OD₆₀₀ of 0.0001 and then diluted again 1:1000. Volumes of 10 µl, 20 µl and 50 µl of the dilution were spread onto the corresponding media to estimate the ratio between red colonies (harboring the plasmid) and white colonies (loss of the plasmid). The cultivation with 5 mM D-alanine should probably lead to the loss of the plasmid, because it is not necessary for bacterial growth in case of D-alanine supplementation. As shown in Figure 10, white cells occur after 17 h of cultivation. After an incubation of about 12 h on LB plates, the first phenotypically identification which intends a loss of the plasmid BBa_K1465401 took place after 29 h. In contrast to that, there was no formatioin of white colonies within the antibiotic-free selection system. In comparison to the selection with chloramphenicol, we can suggest that the plasmid stability might be the same for 52 h. A more accurate quantification would be achieved with a long-term cultivation to investigate plasmid loss in one of the selection media. Usually E. coli is not cultivated for such a long time, which means that the data should be sufficient enough. Yet, the investigation of the plasmid stability using the antibiotic-free selection system can be improved by the fluorescence measurement of a reporter protein to quantify the plasmid stability more precisely.

Figure 10: Long-term stability of the plasmid BBa_K1465401 by calculation of the ratio between white colonies (loss of the plasmid) and red colonies (harboring the plasmid).

Remaining Challenges

The E. coli strains KRX Δalr ΔdadX and DH5α Δalr ΔdadX respectively showed a strict dependence of D-alanine, but as mentioned before the ratio of false-positive transformants was slightly higher compared to the selection ratio on the antibiotic selection with chlormaphenicol. Even on the negative plate some colony forming untis were observable, while there were no colonies on the LB plate containing 30 mg/L Chloramphenicol. This effect migth be due to some revertants to the D-alanine auxotrophy and to the corresponding selection pressure.
Therefore, the revertants were analyzed by spreading an overnight culture of the strains DH5α Δalr ΔdadX and DH5α Δalr kan:dadX on normal LB and several dilution onto LB medium containing 5 mM D-alanine. The same procedure was performed with the transformation approach. In both cases nearly the same revertants rate of 3.4 10^-7 (overnight culture) and 3.11 10^-7 ± 2.29 10^-7 (transformation) was investigated. There was no significant difference between the ratio of revertants of the strain DH5α Δalr ΔdadX (3.27 10^-7 ± 2.27 10^-7) and DH5α Δalr kan:dadX (2.95 10^-7 ± 2.65 10^-7), which indicates that an effect by contamination can be excluded and additional colonies are probably revertants which are able to accumulate D-alanine in some way.
A possible explanation might be a point mutation in the coding sequence of the methionine repressor metJ, resulting in a similar mutation rate of 7 x 10^-7 (Kang et al., 2011). Under normal circumstances the MetJ represses all essential genes for the biosynthesis of L-methionin like metA, metB, metC, MetF, metE and metK as well as the genes of the metD operon by using S-adenosylmethionine (SAM) as cofactor (Figure 11).

Figure 11: Supprression of the Methionine biopsynthesis by MetJ. A point mutation within the repressor leads to a higher expression of MetC, which is also able to catalyze the conversion from L-alanine into D-alanine in E. coli.

It was demonstrated that in the presence of L-methionine all affected genes are suppressed, no revertants are observable, and the reversion could not be quantified in its absence. This suggests that there is another methionine-repressible enzyme, which is able to accumulate D-alanine in E. coli. The revertants formed in the absence of L-methionine showed a higher expression of the Cystathionine β-lyase and point mutations in the MetJ repressor like R42C. It was demonstrated that a strict D-alanine auxotrophy can be restored by an expression of the natural metJ repressor from a plasmid or the additional deletion of metC (Kang et al., 2011).
Until now the antibiotic-free selection was only demonstrated for small plasmids like BBa_K1465401 as a backbone (3163 bp) and BBa_J04450 (RFP, 1069 bp) as an insert, resulting in a plasmid-size of 4232 bp. For greater plasmids or more complicated cloning approaches the transformation and selection might be problematic if none or very few positive colonies are formed. In such a case the ratio of false-positive transformants might be higher which necessitates the addition of L-methionine or the deletion of metC to obtain an effective selection. For easy cloning approaches an effective selection might be already possible without L-methionine or an additional deletion of metC.