Differences in eGFP expression according to variations of genetic modules
In this study we tested the influence of a modification of the
transcription initiation region, the presence of three genetic modules
on the expression vectors and combinations thereof. The
L-rhamnose-inducible vector pJOE4056.1 had already included the first
19 base pairs (bp) upstream of the AUG start codon from the highly
translated bacteriophage T7 gene 10. This sequence included a SD sequence perfectly matching the E. coli canonical sequence but lacked the stem-loop structure. In the first step, the original transcription initiation region of rhaPBAD on pJOE4056.2 was replaced by a modified T7 gene 10 untranslated leader sequence which included the stem-loop structure. The difference to the original T7 gene 10 upstream sequence 
is a deletion of an 8 bp sequence between the stem-loop and SD which
was replaced by the recognition site of restriction endonuclease AflII. We have named the resulting promotor rhaPBAD-T7sl
(stem-loop) and the corresponding plasmid pJOE5058.1, which was the
source of the subsequent modifications. In the next step, a ccdAB cassette was inserted (pWA19), one of the two cer sites was excised (pWA21) and the rop site was deleted (pWA23). Plasmids with all possible combinations of these modifications were constructed (Tab. 1 and Fig. 1).
Figure 1. Schematical overview of the vectors used in this study.
A physical map for relevant restriction endonucleases is given for the
plasmid pJOE4056.1 and the location and orientation of the rhaPBAD promotor, the genes encoding eGFP, ampicillin resistance (bla), rop and addiction modules ccdA and ccdB are indicated by triangle and arrows. The transcription terminator sequence (ter) is derived from the E. coli rrnB operon.
Table 1. Plasmids used in this study
To compare the performance of the modified plasmids, the amount of eGFP produced after induction in E. coli JM109 transformed with the individual plasmids was measured at intervals of 60 minutes (Fig. 2). The most significant enhancement was achieved by the conversion of rhaPBAD to rhaPBAD-T7sl, whereas the insertion of ccdAB had
only slight effects comparing pWA19 with pJOE5058.1. The fluorescence
received with pJOE5058.1 is threefold that of pJOE4056.2 referring to
100 μl of a cell suspension of 0.1 OD600 and it is quadrupled referring to the same volume of culture. In the case of pWA22, pWA24 and pWA29 ccdAB even slightly decreased the expression of eGFP. The reduction of the two cer sites
to one increased the measured fluorescence value, as recorded for
pWA21, pWA22, pWA28 and pWA29, compared to the corresponding plasmids
with a second cer. Eventually, the removal of rop again increased the production of eGFP, especially if combined with only one cer site,
as it is the case for pWA28 and pWA29. This effect is more apparent if
the values relating to fluorescence per ml culture are compared, which
is due to a slightly increased optical density of these cultures (data
not shown). The enhanced fluorescence achieved with the Δrop plasmids correlates with an increased quantity of DNA extracted from the corresponding cultures (Fig. 3). This observation is according to expectations, as rop is
a regulator of copy number, and a deprivation of Rop leads to an
elevated copy number, in some cases concomitant with higher yields of
plasmid-encoded protein. The amount of plasmid DNA extracted from
strains with pWA21 and pWA22, which in contrast to pJOE5058.1 and pWA19
only contain one cer site, is slightly shortened. This might be due to the fact that cer not only is responsible for the resolution of multimers, which incidentally does not occur in recA strains like JM109, but also the Rcd hosted in cer can slow down growth of cells bearing multimers. Therefore, plasmids with two cer sites possibly mimic multimers, decelerate cell division and facilitate accumulation of plasmid DNA.
Figure 2. Fluorescence intensity of eGFP in E. coli JM109 expressed from the indicated plasmids.
The cells were induced with 0.2% (w/v) L-rhamose and the fluorescence
intensity was measured. Intensity after 2, 4, 6, 8 and 24 h are shown
in (A) fluorescence per 0.01 OD600 and in (B) fluorescence
per ml culture. Values shown are the averages of three independent
experiments, for the measuring points at 24 h the standard deviation is
Figure 3. Comparison of plasmid-DNA amounts in the E. coli strain JM109.
Cells were grown at 37°C in LB with ampicillin for 16 h, plasmid DNA
was isolated by a boiling preparation  and the supernatants were
analyzed by electrophoresis on a 0.5% agarose-gel and visualized by
EtBr staining. Lanes 1 and 11: 1 Kb DNA Ladder (Invitrogen).
Comparison of plasmid stability
The usability of a plasmid is not only determined by the amount of
protein produced in a specific combination of strain, gene and plasmid,
but also by the stability of the expression vector. Especially if the
induction is carried out on a large scale and during a longer period of
process, plasmid loss can have a tremendous effect on total yield. In
order to test the influence of the variations of the genetic modules,
we determined how many cells lost their plasmids during a prolonged
induction. If the cultures were induced at 30°C for 48 or 96 hours,
which is roughly equivalent to 50 generations or 100 generations,
respectively, no plasmid loss could be observed in none of the cultures
(data not shown). But if the induction was carried out at 37°C for 48
hours, major differences between the plasmids could be detected. The
originating pJOE4056.2 and its two direct derivatives pJOE5058.1 and
pWA19 were perfectly stable (Fig. 4). These results show, that the exchange of the original transcription initiation region by that of φ10 promoter from T7 and the insertion of the ccdAB locus in addition had no negative effect on plasmid maintenance. The elimination of one cer site
producing pWA21 and pWA22 has led to a negligible raise of plasmid loss
of 5–8% in approximately 50 generations, a faint disadvantage which is
acceptable in regard to the enhanced expression. In contrast, the
deletion of rop added more instability to the vectors, for
instance 25% of the cells bearing pWA23 had lost their plasmids.
Furthermore, the combination of Δrop and ccdAB increased
the percentage of plasmid free cells to 32% (pWA24). This observation
is not according to expectations, because the addiction module should
mediate programmed cell death if the gene coding for the unstable
antidote is lost. Moreover, the combination of one cer site and Δrop had
a remarkably negative impact on plasmid persistence, as it resulted in
a plasmid withdrawal of about 68%. Again, the addition of ccdAB adversely affected the plasmid stability. Apparently, combination of the specific genetic modules cer, rop and ccdAB,
which are supposed to have a stabilizing impact on plasmids, does not
simply lead to summable effects, but can contrarily destabilize
vectors. In general, the influence on plasmid maintenance of
plasmid-borne elements and their interaction with the particular host
strain has to be examined in individual cases.
Figure 4. Plasmid stability after 48 h in liquid culture without antibiotic selection.
Cells were grown at 37°C in LB supplemented with 0.2% (w/v) L-rhamnose
for 24 h, starting from this culture fresh medium was inoculated, and
again incubated for 24 h under the same conditions, which roughly
matches 50 generations. The percentage of cells without plasmid or loss
of fluorescence are shown, the values are the averages of three
Variations in cell length after induction of genes in dependency of E. coli strains
When induction of eGFP was carried out in the E. coli strains BW3110 and BL21 Rha- additionally to the experiments in JM109 mentioned above, lower values of fluorescence were observed (referring to 100 μl
of a cell suspension of 0.1 OD600, data not shown). However, the
relations between the fluorescence obtained with the individual vectors
were comparable. But we noticed that in BW3110 and BL21 Rha- the
optical density of the cultures increased significantly, though these
strains are not able to use L-rhamnose as a carbon source, due to the
lack of rhaB. Examination by microscopy showed, that the
reason for this raise was not a higher cell count, but an increased
cell length (Fig. 5). As these differences were most significant in BL21 Rha-, this strain was used to examine the effect on morphology in more detail (Fig. 6).
Figure 5. Fluorescent light micrographs of L-rhamnose induced E. coli BL21 Rha- without plasmid (A) and BL21 Rha- with pWA21 (B). The bars represent 1 μm length.
Figure 6. Effect of plasmids, carbohydrates and eGFP expression on cell length of different strains.
(A) The strains were grown for 24 h at 30°C in LB supplemented with
0.2% (w/v) L-rhamnose (+), L-arabinose (A) or D-glucose (G) as
indicated in the row 'induction'. For each culture the lengths of 100
cells were determined, the average value and the standard deviation are
shown. The optical density of the cultures is indicated by a black
diamond above the bar. The number of cells per milliliter of culture
were determined by counting them in a Thoma-chamber.
Arabinose and glucose supplement heightened the optical
density but had no effect on cell length. Moreover, no changes could be
observed in cells bearing the plasmids without eGFP (pWA124.1
and pWA125.1) under conditions of induction, or in cells carrying the
plasmid pWA21 under non-induced conditions. These experiments showed,
that the increased length is not caused by the vector itself.
Furthermore, the elongation was less distinct if eGFP was expressed from pJOE4056.2, which also achieved lower fluorescence compared to pWA21 (Fig. 2). It is quite evident, that the cell-length effect in fact is dependent on the amount of eGFP protein produced therein.
All strains showed an increase of cell length and optical density of
the cultures under conditions of induction compared to cells without
plasmid. The standard deviation of length in cultures with prolonged
cells was larger than in the control batches, which are more
homogenous. In cultures with rhaB-positive cells the optical density was elevated more than in rhaB-negative strains, an effect that is revealed in the comparison of W3110 and BW3110 or BL21 and BL21 Rha-. Obviously, these differences were a result of metabolization of L-rhamnose.
In the E. coli K12 derivatives JM109, W3110 and BW3110 the
number of cells raised concomitantly to the optical densitiy if they
were induced. In BL21 and BL21 Rha-, strains that belong to the E. coli B-type
group, in contrast the number of cells declined if L-rhamnose was
added. This effect probably is in conjunction with the stronger impact
on cell length, since cell division problems might occur.
Influence of adjusted CRP-binding site on expression of eGFP
As CRP is a key regulator for catabolite repression, the
corresponding binding sites can be found in or near many promotors
involved in carbohydrate catabolism. For instance, the E. coli lac promoter DNA site for CRP differs from the consensus DNA site at 7 of 22 positions. It has been shown, that CRP in vitro exhibits a 450-fold higher affinity for the consensus site than for the natural binding site .
According to these observation, derivatives of pWA21 with modified
CRP-binding sites were constructed. In pWA73.1 four nucleotides out of
eight differing from the consensus have been adjusted, whereas in
pWA64.1 one nucleotide in the most conserved region accidentally was
deleted (Tab. 2). To test the influence of the mutations, the amount of eGFP produced by E. coli JM09
transformed with the individual plasmids, grown in liquid medium
supplemented with different concentrations of D-glucose and induced
with L-rhamnose, was measured (Fig. 7).
The amounts of eGFP produced with pWA73.1 were comparable to those
produced with pWA21. Even under conditions of D-glucose addition no
differences in production were detectable, neither in the total values
reached at the end of the process nor in the production rates.
Obviously, the adjustment implemented in pWA73.1 did not weaken the
sensitivity to catabolite repression mediated by CRP. However, the
missing nucleotide in the CRP-binding site of pWA64.1 had a tremendous
effect. The fluorescence was shortened to about 10% of the reference
magnitude achieved by pWA21, but still it was repressed by D-glucose at
a concentration of 0.2%. These results suggested that this poorly
inducible promotor still is dependent on the binding of CRP, but may
have a lower affinity.
Figure 7. The influence of D-glucose on the eGFP expression in E. coli JM109 with the plasmids pWA21, pWA73.1 and pWA64.1.
The cells were grown for 2 h at 37°C in LB supplemented with D-glucose
as indicated, then shifted to 30°C and induced with 0.2% (w/v)
L-rhamnose and the fluorescence intensity was measured. Intensity after
2, 4, 6, 8 and 24 h are shown in (A) fluorescence per 0.01 OD600 and
in (B) fluorescence per 1 ml culture. Values shown are the averages of
three independent experiments, for the measuring points at 24 h the
standard deviation is indicated.
Table 2. Sequences of the varied CRP-binding sites. The most conserved nucleotides are underlined.