Effector Overlap between the lac and mel Operons of Escherichia coli: Induction of the mel Operon with β-Galactosides

ABSTRACT

The lac (lactose) operon (which processes β-galactosides) and the mel (melibiose) operon (which processes α-galactosides) of Escherichia coli have a close historical connection. A number of shared substrates and effectors of the permeases and regulatory proteins have been reported over the years. Until now, β-thiogalactosides like TMG (methyl-β-d-thiogalactopyranoside) and IPTG (isopropyl-β-d-thiogalactopyranoside) have not generally been considered to be inducers of the mel operon. The same is true for β-galactosides such as lactose [β-d-galactopyranosyl-(1→4)-d-glucose], which is a substrate but is not itself an inducer of the lac operon. This report shows that all three sugars can induce the mel operon significantly when they are accumulated in the cell by Lac permease. Strong induction by β-thiogalactosides is observed in the presence of Lac permease, and strong induction by lactose (more than 200-fold) is observed in the absence of β-galactosidase. This finding calls for reevaluation of TMG uptake experiments as assays for Lac permease that were performed with mel⁺ strains.

IMPORTANCE The typical textbook picture of bacterial operons is that of stand-alone units of genetic information that perform, in a regulated manner, well-defined cellular functions. Less attention is given to the extensive interactions that can be found between operons. Well-described examples of such interactions are the effector molecules shared by the lac and mel operons. Here, we show that this set has to be extended to include β-galactosides, which have been, until now, considered not to effect the expression of the mel operon. That they can be inducers of the mel operon as well as the lac operon has not been noted in decades of research because of the Escherichia coli genetic background used in previous studies.

INTRODUCTION

Operons and regulons are commonly regarded as more or less independent functional units of bacterial metabolism. Relatively few publications deal with regulatory overlaps between different operons of Escherichia coli (e.g., the gal and mgl operons [1]). Another pair of operons with well-known interactions are the lac (2, 3) and mel (4, 5) operons. The mel operon consists of the genes melA (which encodes Mel α-galactosidase) and melB (which encodes Mel permease) (6). Their expression is controlled by the AraC-type activator protein MelR, whose gene lies in opposite orientation immediately upstream of the mel promoter (7). Today, the mel operon is mainly known as a model system for permease function (8, 9) and promoter regulation (10). Interest in the mel operon arose initially from the facts that its permease shares the substrate TMG (methyl-β-d-thiogalactopyranoside) with the lactose (Lac) permease (11) and that the inducer and substrate of the mel α-galactosidase, melibiose (6-α-d-galactopyranosyl–d-glucose), is an inducer of the lac operon, as are several other α-galactosides (12, 13), and a substrate of Lac permease (but is not cleaved by β-galactosidase) (12). This and the fact that the Mel permease of E. coli K-12 is temperature sensitive (11, 14) have actually been used to screen for lacY-positive revertants by selecting for growth on melibiose minimal medium at 42°C (15, 16).

Like the lac operon (17), the mel operon is known to be induced by galactose in a galactokinase-negative background (where galactose accumulates in the cell). Schmitt found more than 50% maximal α-galactosidase activity in a galK strain induced with 2.8 mM galactose (5). However, the mel operon is not thought to be induced by β-galactosides at all (5, 17, 18). On the other hand, weak activation of mel expression in the K-12 wild-type (WT) strain MG1655 (19) by IPTG (isopropyl-β-d-thiogalactopyranoside) has been reported by Blattner and coworkers since the 1990s. They found moderate (3- to 9-fold) increases in mel operon mRNA upon induction with IPTG in cDNA hybridization experiments (20–22), but they point out that there are no prior published reports that support their observation with an independent method (21).

This report investigates this discrepancy by determining the inducibility of the mel operon by IPTG and two other β-galactosides with sensitive α-galactosidase assays in different genetic backgrounds. The results are discussed with respect to the interaction of the MelR apo-activator and its inducers (strictly speaking, coactivators).

RESULTS

Table 1 shows that in the absence of Lac permease, β-d-thiogalactosides IPTG and TMG induce the mel operon only by a factor of about two. However, the response of the mel operon is very different when Lac permease accumulates the thiogalactosides in the cytoplasm; IPTG induces the mel operon by about 60-fold and TMG induces it by about 170-fold. The control measurements of lac operon expression confirm the well-known fact that, in the absence of Lac permease, 1 mM IPTG induces the lac operon better than 1 mM TMG (2). All cultures were grown in M9 minimal medium (23) with 0.4% glycerol as the carbon source.

TABLE 1.

Induction of the mel and lac operons with β-thiogalactosides

Inducer	α-Galactosidase units (mel operon) for genotype:		β-Galactosidase units (lac operon) for genotype:
	lacY	WT	lacY	WT
None	0.320 ± 0.015	0.414 ± 0.021	2.229 ± 0.051	2.16 ± 0.15
1 mM IPTG	0.571 ± 0.035	27.0 ± 3.4	4,740 ± 110	3,930 ± 340
1 mM TMG	0.733 ± 0.042	70.6 ± 4.5	1,635 ± 22	3,170 ± 170

Since TMG induces the mel operon quite well in the presence of Lac permease but not in its absence, although Mel permease alone is able to accumulate TMG efficiently (5), we tested whether a preinduced mel operon might be able to maintain its on state in the presence of TMG (as the lac operon does in the presence of low concentrations of TMG [24, 25]). Preinduced cells (grown on melibiose) and uninduced cells (grown on glycerol) were transferred to minimal medium with glycerol and 1 mM TMG and grown for about 10 generations before specific α-galactosidase activities were determined. Table 2 shows that the result is negative. In both cases, cells exhibited the same low activities; the preinduced cells did not maintain the induction of the mel operon in the presence of TMG.

TABLE 2.

Deinduction of the mel operon in the presence of TMG

Carbon source (0.4%)	α-Galactosidase units after 10 generations (lacY genotype)
	mel operon uninduced	mel operon preinduced
Glycerol	0.492 ± 0.051	0.377 ± 0.065
Glycerol + 1 mM TMG	1.070 ± 0.075	0.99 ± 0.10
Melibiose	371 ± 22	316.4 ± 8.5

It has long been known that galactose, a weak inducer of the lac operon (17), can also induce the mel operon—however, apparently only in a galactokinase-negative background, where galactose can accumulate in the cytoplasm (5). Employing a sensitive α-galactosidase assay on concentrated cell extracts (see Materials and Methods), moderate (7- to 8-fold) but significant upregulation was also seen in strains with the WT gal operon, where galactose is metabolized. Interestingly, a similar upregulation of the mel operon happens in WT bacteria growing on lactose (Table 3).

TABLE 3.

Induction of the mel and lac operons by galactose, lactose, and melibiose

Carbon source (0.4%)	α-Galactosidase units (mel operon) for genotype:		β-Galactosidase units (lac operon) for genotype:
	lacY	WT	lacY	WT
Glycerol	0.468 ± 0.039	0.550 ± 0.014	3.280 ± 0.080	3.413 ± 0.020
Galactose	3.30 ± 0.24	4.70 ± 0.35	2.717 ± 0.038	3.81 ± 0.12
Lactose	No growth	3.11 ± 0.39	No growth	1,404 ± 52
Melibiose	394 ± 45	431 ± 67	2,052 ± 98	2,050 ± 120

To determine whether it is lactose itself that acts as an inducer of the mel operon or a downstream product (where galactose would be the obvious candidate), the effect of lactose was investigated in cells lacking β-galactosidase, the only enzyme in E. coli that can hydrolyze lactose into glucose and galactose. All cultures were grown in minimal medium with 0.4% glycerol (Table 4). The addition of 0.4% lactose to the medium induced the mel operon in a lacZ-negative strain only weakly. This is expected, as lactose itself does not induce the lac operon but rather the inducer is the transgalactosylation product allolactose [β-d-galactopyranosyl-(1→6)-d-glucose], which is formed through the action of β-galactosidase (26). However, without the help of Lac permease, lactose crosses the cell membrane only very slowly (27). The expression of Lac permease was therefore induced with plasmid-mediated operator titration (28). A lac operator on a multicopy plasmid will sequester Lac repressor molecules, thus reducing the number of repressor molecules available to repress expression of the chromosomal lac operon. As a consequence, the expression of the lac operon will increase. Plasmid pBR322 does not contain a lac operator and will thus not induce the lac operon. Multicopy plasmid pBR322-Oid carries an ideal lac operator (28, 29), and the relaxed multicopy plasmid pUC19-Δα carries both the first and the third operators of the lac operon. For details, see Materials and Methods. β-Galactosidase activities of WT strain MG1655 carrying these plasmids were measured (Table 5) to determine the degree to which the plasmids induce the lac operon and thus expression of Lac permease. The resulting α-galactosidase activities are given in Table 4. The data show that lactose itself acted as an inducer of the mel operon. In the presence of basal-level Lac permease, the induction was less than 10-fold, but when accumulated through the action of highly expressed Lac permease, lactose induced the mel operon to a degree similar to accumulated IPTG and TMG (compare Tables 1 and 4).

TABLE 4.

Induction of the mel operon by lactose

Genotype + plasmid	Medium supplement (0.4%)	α-Galactosidase units
lacZ + pBR322	None	0.1743 ± 0.0023
lacZ + pBR322	Melibiose	369 ± 12
lacZ + pBR322	Lactose	1.47 ± 0.20
lacZ + pBR322-Oid	Lactose	48.17 ± 0.56
lacZ + pUC19-Δα	Lactose	55.7 ± 2.9
WT + pBR322	Lactose	2.483 ± 0.084

TABLE 5.

Induction of the lac operon by operator titration

Genotype + plasmid	Medium supplement	β-Galactosidase units	% induction
WT + pBR322	None	1.79 ± 0.20	0
WT + pBR322	1 mM IPTG	3,827 ± 78	100
WT + pBR322	0.4% lactose	1,358 ± 37	35
WT + pBR322-Oid	None	302 ± 20	8
WT + pUC19-Δα	None	3,737 ± 20	98

DISCUSSION

It is generally true for catabolic operons that permeases have a relatively broad specificity for binding and/or transport of substrates while transcriptional repressors and activators have a higher effector specificity and the catabolic enzymes have the highest substrate selectivity. In general, a good inducer of one system will be a less-efficient inducer of another. This work investigated the role of β-galactosides as inducers of the mel operon.

It should be pointed out that the interpretation of the data in terms of direct apo-activator/coactivator (for simplicity, inducer) interaction of the melibiose operon regulator MelR with β-galactosides is, at this point, still speculative. While it is the simplest and most plausible explanation, it will have to be confirmed by in vitro biochemical experiments.

The mel operon literature contains an unresolved inconsistency. Assays of α-galactosidase activities seem to demonstrate clearly that IPTG (like all other β-galactosides) does not induce the mel operon (13–15), while cDNA hybridization indicates a small but significant (less than 10-fold) upregulation (20–22). On the other hand, the inducibility of the mel genes by galactose (17) has been known for many decades and is supported by biochemical evidence (30). The results presented here confirm and quantify the inducibility by IPTG and demonstrate mel operon induction by other β-galactosides. This inducibility was missed by previous studies for two reasons, relative insensitivity of the α-galactosidase assays and the use of Lac permease-negative bacteria. α-Galactosidase was initially only assayed in intact cells, which is possible because the rate-limiting step in hydrolysis of the chromogenic substrate α-PNPG (p-nitrophenyl-α-d-galactopyranoside) is α-galactosidase activity and not uptake by the cells (31). Our more sensitive α-galactosidase assay on extracts of concentrated cells enabled us to accurately determine the uninduced background expression of the mel operon and, consequently, the weak induction by β-galactosides in the absence of Lac permease. The presence of Lac permease leads to the accumulation of β-galactosides in the cell, just as a galactokinase-negative background leads to the accumulation of galactose. Consequently, induction dramatically increases up to more than 100-fold. Blattner and coworkers (20–22) used the same lacY⁺ WT strain, MG1655, that was employed in the present study. Thus, they were able to detect the upregulation of mel operon expression by IPTG, albeit to a much lower degree than reported here. The reason for this is most likely the well-known ratio compression effect observed in microarray experiments (32). Indeed, their experiments quantified the induction of the lac operon at only 60- to 80-fold (21), while it is well established to be more than 1,000-fold (Tables 1 and 5).

Studies on Mel permease generally use lacY-negative E. coli strains to avoid the effects of the functional overlap of the two permeases. However, the reverse is not true. Rather, in studies on Lac permease activity, a separation of the two uptake systems is mostly sought by specifically inducing the Lac permease with IPTG or TMG, assuming that the mel operon will remain uninduced so that the contribution by Mel permease is negligible (11). It is shown here that this discriminative induction does not work well. Cells with WT Lac permease induced with IPTG or TMG will also exhibit Mel permease activity. This will even be true for E. coli K-12 cells grown at up to 37°C. Only at higher temperatures does the temperature-sensitive Mel permease activity decline strongly (14). Therefore, qualitative and quantitative data on in vivo Lac permease activity obtained under such conditions should be interpreted with caution.

Table 4 shows that lactose accumulated at less than 10% maximal lac operon expression induced the mel operon nearly as well as when the lac operon was fully induced. A similar observation has been made with TMG; cells that express the lac operon at a fractional level accumulate TMG to a steady-state concentration close to that of cells with a fully induced lac operon (33).

Another interesting observation is that, although the mel operon was significantly induced by TMG accumulated by the Lac permease, the induction was not maintained by TMG in the absence of Lac permease, despite the fact that Mel permease also transports TMG (11). The expression of permease will increase as a function of the intracellular inducer concentration, and the intracellular inducer concentration will rise when more permease is present. A level x of permease will be stably maintained when it gives rise to exactly the concentration of intracellular inducer that induces the permease gene to express that level. This is the maintenance concentration of the inducer. If the permease produces less intracellular inducer, its level will decrease, and if it produces more, its level will increase. TMG is an efficient inducer of the lac operon but is relatively inefficient in inducing the mel operon. When TMG accumulates through the action of Lac permease, the expression of the mel operon only reaches about 20% of the maximal level (Tables 1 and 3). This is not sufficient to maintain the high intracellular TMG concentration, and as a consequence, the mel operon becomes deinduced.

Figure 1 shows structural sketches of the inducers used in the experiments. The basic unit of recognition by MelR appears to be the galactose moiety (boxed). It binds with a low affinity to MelR and is an efficient inducer only when it is accumulated in the cell by permeases. The binding to MelR triggers an allosteric change to the activating conformation of the protein. Additional contacts provided by an α-galactosidic aglycone, as in the case of melibiose, appear to be required for high-affinity binding. The presented data suggest that β-galactosidic aglycones do not help with binding but also do not seem to strongly sterically interfere, even in the case of lactose. Whether allolactose is also an inducer of the mel operon remains unclear for now, as it is produced in the described experiments only in the presence of lactose, which was itself shown to induce the mel operon. In conclusion, it appears true for the mel operon as for the lac operon that “inductivity is dependent on an intact galactoside radical, either free or with an α or β linkage” (34).

FIG 1.

Structures of the lac and the mel operon inducers used in the presented experiments. It is unknown if allolactose induces the mel operon. All other sugars are inducers of the mel operon, and all but lactose are inducers of the lac operon.

MATERIALS AND METHODS

Bacterial strains.

Wild-type (with respect to the lac and mel operons) Escherichia coli K-12 strain MG1655 (F⁻ λ⁻ ilvG rfb-50 rph-1) (19) was obtained from the Coli Genetic Stock Center at Yale University. Isogenic lacY (strain StADa1) and lacZ (strain StADb1) mutants were generated with UV mutagenesis of MG1655 and selected with the penicillin method (23).

MG1655 cells were UV irradiated to >99.9% lethality in 100 mM MgSO₄ and then grown overnight at 37°C in 1× A minimal medium (23) with 0.4% glucose. A 1:100 dilution of these cultures in minimal medium lacking any carbon source was first incubated at 37°C with shaking for about 1 h. Then, lactose was added to 0.4% and ampicillin was added to 50 μg/ml. Incubation was continued for 4 h to counterselect lac⁺ cells. The bacteria were then pelleted, resuspended in 1× A with glucose, and again grown overnight. The ampicillin selection was repeated once before dilutions were plated on MacConkey agar (Difco, USA), on which lac mutant colonies are identified by their white color (35). These colonies were restreaked on 1× A minimal agar plates with 0.4% glycerol, 0.1 mM IPTG, and 0.004% X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) to distinguish lacZ mutant (white) from lacY mutant (blue) colonies. Tentative lacZ and lacY mutant isolates were grown in minimal medium with glycerol without IPTG, as well as in the presence of 1 mM IPTG. Cells were harvested and resuspended in Z buffer (23). β-Galactosidase assays (23) and ONPG (ortho-nitrophenyl-β-d-galactopyranoside) uptake assays (36) were performed to identify clones that were fully inducible and exhibited WT LacZ activity while lacking LacY activity (no ONPG uptake) and clones completely lacking LacZ activity. Finally, the isolates were transformed with plasmid pWB310, which expresses β-galactosidase, and with plasmid pT7-5(lacY) (37), which expresses Lac permease. Transformants were streaked onto 1× A minimal agar plates with 0.4% lactose as the sole carbon source. All tentative lacZ mutant isolates were complemented with pWB310 to Lac⁺, and all tentative lacY mutant isolates were complemented with pT7-5(lacY) to Lac⁺, confirming the location of the respective lac mutations.

DNA modification enzymes.

Restriction enzymes, Klenow fragments, and ligases with their respective buffers were purchased from Minotech Biotechnology (Greece), New England BioLabs (USA) or Thermo Fisher Scientific (USA).

β-Galactosides.

Lactose (analytical grade) and IPTG (purity, ≥99%) were purchased from SRL, India. TMG (purity, 99.5%), product number M8146, was obtained from Sigma-Aldrich, USA.

Plasmid construction.

The repressor titrating plasmid pBR322-Oid was created by cloning the ideal lac operator (28) from pWB310 (38) as an XbaI fragment into the compatible NheI site of multicopy plasmid pBR322 (39).

The repressor titrating plasmid pUC19-Δα was constructed by cutting the relaxed multicopy plasmid pUC19 (40) with restriction enzymes NdeI and EcoRI, filling in with Klenow fragment, and ligating the blunt ends. This deleted the coding sequence for the α-fragment of β-galactosidase but retained the first and third lac operators (41).

Galactosidase assays.

Cells were grown in M9 medium (23) supplemented with different carbon sources at 0.4% (wt/vol) each and harvested in the exponential growth phase. Where nothing else is specified, the carbon source was glycerol. Because of the temperature sensitivity of the mel operon, all cultures were grown at 30°C. Where required, IPTG and TMG were added to 1 mM. When plasmid pBR322, pBR322-Oid, or pUC19-Δα had to be maintained, the medium was supplemented with ampicillin to a concentration of 200 μg/ml. Spectrophotometric measurements were performed with a BioSpectrometer basic (Eppendorf, Germany) or a NanoDrop 2000c (Thermo Fisher Scientific, USA).

β-Galactosidase assays were performed according to the method of Miller (23), and specific activities are given in Miller units. They are calculated as follows: (OD₄₂₀ × 1,000)/(fractional sample volume × OD₆₀₀ of cell suspension in assay buffer × reaction time in minutes), where OD₄₂₀ and OD₆₀₀ are the optical densities at 420 and 600 nm, respectively.

The protocol for determining α-galactosidase activities is a combination of the conditions reported by Schmitt and Rotman (4) and by Burstein and Kepes (18) with the procedure of Miller (23). The specific enzyme activities are reported in Burstein units, named for an author of the paper that first described the requirement for NAD⁺ for MelA activity in vitro (31). Cells were harvested during exponential growth (at OD₆₀₀ values of between 0.5 and 1), pelleted, and resuspended in assay buffer (50 mM Tris-HCl [pH 7.5], 5 mM MnCl₂, 0.25 mM NAD⁺, 120 mM 2-mercaptoethanol, and 100 μg/ml chloramphenicol). Cells were concentrated in this step 5- to 10-fold for determining low specific activities below 10 Burstein units. Cell suspensions of between 0.6 and 1 ml were sonicated for 65 s in an ice-water bath using a Q700 (Qsonica, USA) equipped with a microtip and set to an output of 5 W. Samples of 100 μl were equilibrated for 5 min at 35°C, and the reactions were started by adding 20 μl of prewarmed 18 mM α-PNPG in assay buffer. After sufficient color development, reactions were terminated with 50 μl of stop solution (136 mM EDTA, 728 mM Na₂CO₃). Cell debris was removed by centrifugation at a relative centrifugal force (RCF) of 12,000 for 5 min, and OD₄₁₀ values were determined spectrophotometrically. Burstein units, like Miller units, are proportional to the release of nitrophenol per unit of OD₆₀₀ per minute. Specific activities were calculated in analogy to Miller units as follows: (OD₄₁₀ × 1,000)/(fractional sample volume × OD₆₀₀ of cell suspension in assay buffer × reaction time in minutes) = Burstein units.

All Miller and Burstein units are the averages from three biological replicates and are reported with the standard errors of the means. Under our conditions, cells grown in minimal glycerol medium and fully induced with 1 mM IPTG exhibited a specific β-galactosidase activity of about 4,000 to 5,000 Miller units, and cells grown in minimal medium and induced with 0.4% melibiose exhibited a specific α-galactosidase activity of about 400 to 500 Burstein units.

Software.

Figure 1 was prepared with ChemSketch 14.01 (ACD/Labs, Canada).