Protein:Protein interactions in the cytoplasmic membrane apparently influencing sugar transport and phosphorylation activities of the e. coli phosphotransferase system

The multicomponent phosphoenolpyruvate (PEP)-dependent sugar-transporting phosphotransferase system (PTS) in Escherichia coli takes up sugar substrates from the medium and concomitantly phosphorylates them, releasing sugar phosphates into the cytoplasm. We have recently provided evidence that many of the integral membrane PTS permeases interact with the fructose PTS (FruA/FruB) [1]. However, the biochemical and physiological significance of this finding was not known. We have carried out molecular genetic/biochemical/physiological studies that show that interactions of the fructose PTS often enhance, but sometimes inhibit the activities of other PTS transporters many fold, depending on the target PTS system under study. Thus, the glucose (Glc), mannose (Man), mannitol (Mtl) and N-acetylglucosamine (NAG) permeases exhibit enhanced in vivo sugar transport and sometimes in vitro PEP-dependent sugar phosphorylation activities while the galactitol (Gat) and trehalose (Tre) systems show inhibited activities. This is observed when the fructose system is induced to high levels and prevented when the fruA/fruB genes are deleted. Overexpression of the fruA and/or fruB genes in the absence of fructose induction during growth also enhances the rates of uptake of other hexoses. The β-galactosidase activities of man, mtl, and gat-lacZ transcriptional fusions and the sugar-specific transphosphorylation activities of these enzyme transporters were not affected either by frustose induction or by fruAB overexpression, showing that the rates of synthesis of the target PTS permeases were not altered. We thus suggest that specific protein-protein interactions within the cytoplasmic membrane regulate transport in vivo (and sometimes the PEP-dependent phosphorylation activities in vitro) of PTS permeases in a physiologically meaningful way that may help to provide a hierarchy of preferred PTS sugars. These observations appear to be applicable in principle to other types of transport systems as well.

The IIA and IIB proteins are cytoplasmic phosphoryl carrier proteins or protein domains while the IIC proteins/domains are integral membrane transporters that catalyze sugar phosphorylation concomitantly with sugar uptake into the cell [8,9]. In these coupled sugar transport/phosphorylation reactions, called "group translocation," the phosphoryl moiety of IIB-P is transferred to the incoming sugar to yield a cytoplasmic sugar phosphate (Sugar-P) [10]. The overall phosphoryl transfer reactions are shown in Fig 1. In this Figure, arrows indicate the pathways of phosphoryl transfer between the protein constituents of the PTS as defined in the introductory paragraph. Essentially, the same general phosphoryl transfer pathway is used for the phosphorylation of all PTS sugars such as glucose (Glc) and its non-metabolizable analogue, methyl α-glucoside (αMG), transported by the glucose PTS (IIA Glc /IIB Glc /IIC Glc   The figure shows the transfer of a phosphoryl group from phosphoenolpyruvate (PEP) to enzyme I (EI), then to HPr, then to the sugar-specific enzyme IIAs and then to the enzyme IIBs before transfer to the incoming sugar via the enzyme IIC, which catalyzes both transport and phosphoryl transfer in a coupled process. The PTS also catalyzes the group translocation of many other sugars. As indicated, the three extracellular sugars represented are mannitol (Mtl), fructose (Fru), and Nacetylglucosamine (NAG), and their phosphorylated derivatives are released into the cytoplasm. Arrows indicate the pathways of phosphoryl transfer. are transported and phosphorylated in a coupled transport/phosphorylation process known as group translocation (see Class 4 in TCDB).
In addition to these three D-sugars, mannitol (Mtl), fructose (Fru) and N-acetylglucosamine (NAG), the PTS transports and phosphorylates many other sugars, and each bacterial or archaeal species possessing the PTS has a different complement of PTS Enzyme II complexes. Sugars transported via the PTS in various organisms include aldo-and keto-hexoses, amino sugars and their N-acetylated derivatives, hexitols, pentoses, pentitols and a variety of disaccharides, oligosaccharides and glycosides [8]. In Escherichia coli, there are many Enzyme II (EII) complexes, some of which are still not functionally characterized [11]. The EIIs we will be concerned with in this report, in addition to the three illustrated in the scheme shown in Fig 1, are specific for sugars such as galactitol (Gat), glucose (Glc) and the non-metabolizable glucose analogue, methyl α-glucoside (αMG), trehalose (Tre), and mannose (Man) [this system also transports Glc, 2-deoxyglucose (2DG), glucosamine (Glm) and fructose]. These systems are tabulated in Table 1 with their protein abbreviations, protein domain orders, TC numbers (TC #s) in the Transporter Classification Database (TCDB; www.tcdb.org), and primary sugar substrates [12][13][14][15].
In addition to the transport reaction that can be studied in vivo, there are two in vitro sugar phosphorylation reactions, the PEP-dependent sugar phosphorylation reaction that depends on EI, HPr and the complete EII complex, and a sugar-P:sugar transphosphorylation (TP) reaction that depends only on IIB and IIC [16][17][18][19]. All of these reactions have been used to investigate the consequences of integral membrane protein interactions in the current study.
Recently, Babu et al published global interactome data for E. coli, including cytoplasmic proteins, integral inner and outer membrane proteins and periplasmic proteins [1]. These studies revealed that many Enzyme II components interact with other soluble and integral membrane proteins of the PTS as well as with other non-PTS proteins. A representative selection of these PTS protein interactions is reproduced in Table 2. Of all the Enzyme II constituents, the fructose PTS proteins, FruA and FruB, appeared to form the most extensive PTS interactome network.
The observation that the Fru PTS might serve to coordinate the activities of other PTS EII complexes correlated with the previously published suggestion that the fructose PTS was the primordial PEP-dependent PTS Enzyme II complex [20]. This suggestion resulted from several observations. (1) The fructose PTS, of all the sugar-specific Enzyme II complexes, is most widely distributed in the prokaryotic world. (2) The fructose system in E. coli has more fructose-like Enzyme II complex homologs than any other. (3) Fructose is the only sugar that feeds directly into glycolysis without interconversion to another sugar derivative. (4) Only fructose can feed into glycolysis via two distinct PTS-mediated pathways: Fru ! Fru-1-P ! Fru-1,6-bis-P (II Fru ), and Fru ! Fru 6-P ! Fru-1,6-bis P (II Man ). (5) Only the fructose PTS of  [21]. More recent studies revealed that there are at least three evolutionarily distinct superfamilies of PTS Enzyme IIC components [22]. First, the PTS Glucose-Glucoside (Glc) family, the Fructose-Mannitol (Fru) family, the Lactose-N, N 0 -Diacetylchitobiose (Lac) family, and (4) probably the Glucitol (Gut) family comprise one large sequence-divergent superfamily with all IIC constituents being homologous (TC# 4.A.1-4; see TCDB, www.tcdb.org) [12,13,23,24]. Second, the Galactitol (Gat) family (TC# 4.A.5) and the L-Ascorbate (L-Asc) family (TC# 4. A.7) comprise a distinct superfamily [11,24,25], and finally, the functionally diverse Mannose family (TC# 4.A.6) comprises the third PTS superfamily [26]. Each of these three superfamilies of IIC permeases are believed to have evolved independently of each other [22,27]. However, it should be noted that the IIA and IIB phospho-carrier constituents, which also probably evolved at least 3 times independently of each other, did not coevolve with the IIC permease constituents, suggesting that there has been extensive shuffling of the IIA, IIB and IIC constituents/domains during the evolutionary divergence of these protein complexes [22,[28][29][30].
Bacterial cytoplasmic membranes are crowded with integral membrane proteins that form complexes, inhibiting their lateral mobilities, constraining their flexibilities and probably altering their activities [31][32][33][34]. Our integral membrane interactome data suggested that many membrane proteins interact with many other membrane proteins, with a wide range of affinities [1]. Some of the high scoring interactions are likely to be permanent, high affinity interactions, while lower scoring interactions are more likely to be transient in nature [35,36]. Among the observed interactions were many involving Enzyme IIC constituents of the PTS, particularly the fructose IIC protein, FruA, which also bears two additional domains, IIB and IIB 0 , where IIB 0 is not involved in phosphoryl transfer, but plays a role in protein-protein interactions [21] (Table 2).
As noted above, the PTS offers technical advantages over other transport systems for its characterization. It can be assayed in vivo by measuring its sugar transport (uptake) activities, and it can be assayed in vitro by measuring its sugar phosphorylation reactions. These include its PEP-dependent sugar phosphorylating activities as well as its sugar-P-dependent  [1]. The higher the score, the stronger the interaction should be, although these scores are also dependent on the protein concentrations. transphosphorylating (TP) activities [19,37]. The latter reactions require only the IIB and IIC domains which are usually, but not always, fused together (see Table 1). Thus, while the PEP:  sugar activity requires protein:protein interactions involving EI, HPr, IIA and IIBC, the sugar-P:sugar TP reaction does not depend on these interactions (except for II Man ) since only the IIC  and IIB domains are directly involved. The results of the studies reported here suggest that in the cytoplasmic membrane of E. coli, the transport of various PTS sugars is influenced by the presence of other PTS permeases in general, and the degree of activation or inhibition depends on the specific systems under study as well as their concentrations in the membrane. Thus, high level expression of the fructose EII complex activates the mannitol, N-acetyl glucosamine, glucose and mannose systems, but it inhibits the galactitol and trehalose systems in wild type cells. Although these protein-protein interactions do not appreciably affect the sugar-P:sugar TP reactions or synthesis of the target EIIs, we did observe a highly specific activation of the in vitro PEP-dependent phosphorylation of mannitol and N-acetylglucosamine, dependent on both FruA and FruB. In this case, we propose that FruB activates MtlA, but that this activation depends on the presence of FruA which could anchor FruB to the membrane, adjacent to MtlA. The more general conclusion is that integral membrane transport proteins in the cytoplasmic membranes of E. coli are in close contact and influence each other's activities in the intact cell, either in one direction (activation) or the other (inhibition). In some cases, these effects appear to be highly specific, while in others they may be more general. This is, to the best of our knowledge, the first example where global proteomic analyses based on a complete interactome data set, have led to the discovery of regulating interactions influencing the enzymatic and transport activities of some of the interacting integral membrane proteins.

Results
The E. coli integral membrane protein interactome, which includes thousands of putative interactions with other integral membrane proteins as well as other cellular proteins, has been published [1]. Table 2 summarizes the high scoring interactions observed between the constituents of different Enzyme II complexes of the PTS. Not surprisingly, the fructose-1-P-forming fructose-specific membrane constituent of the PTS, FruA (the enzyme IICBB' Fru ), interacts with its own FruB protein, the fructose-specific enzyme IIA-FPr protein, the immediate phosphoryl donor for fructose transport energization and phosphorylation, thus providing the energy for fructose uptake via FruA (Table 1; [21]). The interaction score is high (12.9), indicating that this interaction occurs with high affinity since the conditions used to measure these interactions involved growth in LB medium without induction of PTS gene expression [1]. In addition, FruA appeared to interact with several other Enzyme II constituents of the PTS with large scores, far more than for any other PTS Enzyme IIC (Table 2). These interactions include those with MtlA (the IICBA specific for mannitol; score of 11.4), GatC (The galactitol Enzyme IIC; score of 10.3), NagE (The N-acetylglucosamine Enzyme IICBA; score of 9.4), TreB (The trehalose Enzyme IIBC; score of 7.4), and MngA (the 2-0-α-mannosyl-D-glycerate Enzyme IIABC; score of 7.1) (Tables 1 and 2). Other Enzymes II appeared to interact with FruA with lower scores. MtlA additionally interacts with FruB (score, 8.4) and GatC (score, 9.7), while the broad specificity glucose/mannose/fructose/glucosamine Enzyme IIC, ManY, interacts with all three constituents of the galactitol Enzyme II complex, GatA (7.5), GatB (6.9) and GatC (7.4) as well as MngA (7.7) ( Table 2).
Many other integral membrane proteins, several of which proved to be transporters, also appeared to interact with FruA. These include the MDR pumps, AcrB (score of 12.9) and MdtF (17.9), the dicarboxylate transporter, DcuA (17.2), the iron porters, FeoB (11.4) and Fet (YbbM; 14.3), the FocA formate channel (9.5), the NupC nucleoside uptake porter (19.6), the MgtA Mg 2+ uptake ATPase (12.1) and others [1]. These results suggest that numerous integral membrane transport systems interact, possibly within the hydrophobic matrix of the cytoplasmic membrane, but their biochemical and physiological significance could not be determined from these data alone. Therefore, the transport studies and biochemical analyses reported here were conducted, specifically for the putative PTS protein interactions.

Confirmation of the interactome results using a bacterial two hybrid system
As reported by Babu et al. [1], FruA interacts with other membrane proteins as well as its cognate cytoplasmic partner, FruB ( Table 2). A bacterial adenyl cyclase two hybrid (BACTH) system was used to confirm several of these interactions, conducted according to the manufacturers' instructions (Euromedex bacterial two-hybrid system, Cat # EUK001). Interactions of FruA were demonstrated for the GatC, NagE, TreB and MtlA membrane proteins as well as the cytoplasmic FruB protein. Also, interactions of FruB were observed for MtlA and NagE (S1

Transport studies involving PTS group translocators
Because the FruA protein seemed to interact with many other Enzyme II protein complexes with high scores, more than for any of the other PTS proteins, we decided to examine the effect of growth with and without fructose to induce synthesis of the fructose-specific Enzyme II complex, FruA/FruB, on the rates of uptake of other PTS sugars in wild type (WT) E. coli cells. The results are summarized in Fig 2A, which gives the relative apparent rates of sugar transport by the fructose-induced cells relative to uninduced cells. The raw data for this Figure are presented in S1, S2, S3, S3B, S4, S4B, S5 and S5B Tables. S3B, S4B and S5B Tables are controls for S3, S4 and S5 Tables,  respectively. For example, in S3 and S4 Tables, fructose was present during growth, but in S3B and S4B Tables, fructose was absent. Clearly, induction by fructose was necessary for regulation; values recorded represent actual values within experimental error, as indicated by the standard deviation values. As an additional control, the same experiments were conducted with the isogenic strain deleted for the fruBKA operon (ΔfruBKA; triple mutant (TM)), Fig 2B. The results presented in Fig 2A show that in a wild type E. coli strain, growth in the presence of fructose enhanced the amounts of uptake of some radioactive sugars while decreasing the amounts of uptake of others. Thus, the apparent mannitol (Mtl) transport rate increased 7-fold, the apparent N-acetylglucosamine (NAG) transport rate increased 6.4-fold, 2-deoxyglucose (2DG) uptake increased 8.7-fold, and methyl α-glucoside (αMG) uptake increased 2.8-fold, due to the presence of fructose in the growth medium. Interestingly, the apparent uptake of trehalose (Tre) and galactitol (Gat) decreased while galactose (Gal, a non-PTS sugar) uptake did not change appreciably.
The fructose operon was completely deleted, and this ΔfruBKA deletion strain (triple mutant; TM) was used in parallel experiments (Fig 2B). With the fruBKA operon absent, the apparent fructose transport rate was reduced to 5% of the wild type rate. None of the other apparent PTS sugar uptake rates, except for that of 2-deoxyglucose, taken up by the mannose system, was affected by the inclusion of fructose in the growth medium. This last fact may have resulted because fructose is a (poor) substrate of the mannose EII complex, and therefore may also be an inducer of this system. When the TM was examined relative to the wild type (WT) strain, with only LB in the growth medium, none of the apparent sugar uptake rates was affected (Fig 2C), but when fructose was present during growth, the apparent transport of those sugars that had been stimulated by fructose induction in the WT showed low ratios of uptake, while those that were inhibited in the WT showed increased ratios (Fig 2D), in agreement with those presented in Fig1A. When either fruA or fruB was deleted, all regulatory effects were abolished (Fig 2E and  2G), as was true for the ΔfruBKA strain ( Fig 2C). However, the apparent transport ratios, ΔfruA/WT and ΔfruB/WT, showed depressed ratios for those sugars whose uptake was (C&D) Relative apparent sugar uptake by the triple mutant (TM) relative to that of the wild type E. coli BW25113 strain when grown in LB or LB plus 0.2% fructose, respectively, as indicated. (E&F) Relative apparent sugar uptake by a fruA mutant relative to that of the wild type E. coli BW25113 strain when grown in LB and LB plus 0.2% fructose, respectively. (G&H) Relative apparent sugar uptake by a fruB mutant relative to that of the wild type E. coli BW25113 strain when grown in LB or LB plus 0.2% fructose, respectively. 1, Fructose; 2, Mannitol; 3, N-acetylglucosamine; 4, Methyl alpha glucoside; 5, 2-Deoxyglucose; 6, Trehalose; 7, Galactitol and 8, Galactose. The raw data for these plots are shown in supplementary materials (S1, S2, S3, S3B, S4, S4B, S5 and S5B Tables). Note: In this Figure and elsewhere, αMG and 2DG uptake values represent accumulation levels, while for other sugars, a combination of uptake + metabolic rates were estimated. Galactose is taken up via the GalP secondary carrier, not by the PTS. increased by fru operon induction in the WT (Mtl, NAG, αMG and 2DG), and increased values for those sugars whose apparent transport was depressed by fru operon induction (Tre and Gat) (compare Fig 2A with Fig 2D, 2F and 2H).
In summary, it appears that high level expression of the fruBKA operon has effects on the uptake of other PTS sugars, with apparent Mtl, NAG, αMG and 2DG transport consistently showing activation while those for Gat and Tre show inhibition. In ΔfruA and ΔfruB mutants, no regulatory effects were observed. These results suggested that either FruA, FruB, or more likely, both, serve(s) regulatory roles, influencing the apparent rates of transport of other PTS sugars, positively or negatively, depending on the sugar-specific transport system assayed. Fig 3 presents corresponding results for strains in which the fruA, fruB, or fruA and fruB genes were overexpressed. Overexpression of fruA in the TM genetic background increased apparent Mtl and NAG transport rates 4-5-fold, but increased apparent αMG and 2DG transport rates 7 and 11-fold, respectively, compared to the control strain bearing the empty pMAL plasmid (Fig 3A). When the same experiment was conducted in the wild type genetic background, qualitatively similar results were obtained, but enhancement of the apparent transport rates were diminished except for the mannose/2DG uptake system, which was increased ( Fig  3B). Only in the WT background did overexpression of fruA or fruB increase the apparent 2DG uptake rate. Surprisingly, in both cases, apparent Tre and Gat transport rates were not appreciably depressed, possibly due to the non-native conditions used. Thus, when fruB was overexpressed in either the triple mutant (TM; Fig 3C) or the WT genetic background (WT; Fig 3D), there was a significant increase in activity only when 2DG uptake was measured in the WT background. fruB overexpression did seem to stimulate 2DG uptake. Finally, when both fruA and fruB were simultaneously overexpressed using two different compatible plasmids, in either the TM or WT genetic background, the apparent stimulatory effects on Mtl, NAG, αMG and 2DG transport were similar to those observed when only fruA was overexpressed. Again, in this experiment, overexpression of fruA, fruB, or fruA and fruB did not cause apparent Tre and Gat transport to decrease.
To explain these results, we must consider that the overexpression of fruA and fruB caused by fructose induction was dissimilar from that obtained using the two compatible plasmids, pMAL and pZA31-PtetM2, which have different copy numbers. Thus, balanced and equimolar expression of fruA and fruB was achieved in the former case but not in the latter case. Additionally, both FruA and FruB interact with many proteins (see Introduction), which may influence the consequences of their high levels of expression. Nevertheless, the results confirm the suggestion that FruA is primarily responsible for the stimulation of Mtl, NAG, 2DG and αMG uptake, although FruB may play a lesser role.

Operon induction properties using lacZ fusions
The three transport systems showing largest responses to fructose induction in wild type cells were the systems encoded by the mannitol (mtl), galactitol (gat) and mannose (man) operons (see Fig 2). Therefore, in order to determine if these effects reflected changes in operon transcription or EII activities, transcriptional lacZ fusions were constructed to operons encoding these three PTS systems (mtlA-lacZ; gatY-lacZ and manXYZ-lacZ). These were used in studies to determine the effects of fructose in the growth medium prior to transport rate determinations. The results are presented in Table 3. It can be seen that the presence of fructose in the growth medium, or the overexpression of specific fru operon genes, had no apparent effect on the induction of mtl, gat, and man operon expression. These results imply that the effects observed in Figs 2 and 3 reflect the enzyme II activities and not their syntheses. This conclusion is substantiated by the results obtained when the transphosphorylation reactions were studied (see Table 4 below).
Summarizing, growth with or without fructose, or after overexpressing fruA and/or fruB had essentially no effect on the levels of the β-galactosidase activities of the mtlA-, gatY-and manXYZ-lacZ bearing strains. This clearly implied that the mtl, gat and man operons were not repressed or induced by the presence of fructose in a wild type or ΔfruBKA strain or by single or dual overexpression of fruA and fruB in the triple mutant, ΔfruBKA, genetic background.

PEP-dependent phosphorylation of PTS sugars in vitro
Fig 4A presents the consequences of growth in the presence of fructose on the PEP-dependent phosphorylation of various PTS sugars. As expected, fructose phosphorylation activity increased dramatically, but mannitol and N-acetylglucosamine phosphorylation activities increased as well. Interestingly, galactitol phosphorylation decreased dramatically although phosphorylation of other PTS sugars were apparently not affected. Upon overexpression of the complete fruBKA operon (Fig 4B), or of the fruA and fruB genes co-expressed on compatible plasmids, similar results were obtained with increased phosphorylation activities of Fru > Mtl > NAG (Fig 3C and 3D). FruA or FruB overproduction alone had little or no effect, and the phosphorylation of other PTS sugars also showed little effect. Mannitol phosphorylation increased 4 to 6-fold, while phosphorylation of N-acetylglucosamine increased 2 to 3-fold, and phosphorylation of galactitol decreased to about one eighth. Inclusion of fructose, fructose 1-P, fructose 6-P or fructose 1,6-bisphosphate during the in vitro assay had essentially no effect on mannitol, N-acetylglucosamine or galactitol phosphorylation.
The mtlA gene was inactivated with a kanamycin resistance gene insertion or was deleted (ΔmtlA), and the crude extracts were compared with the wild type. Crude extracts were assayed for PEP-dependent [ 14 C]mannitol phosphorylation after growth with or without fructose, and the activity was reduced to a few percent of the wild type level in the insertion or deletion mutant (S22 Table), showing that mannitol cannot be phosphorylated at an appreciable level by other PTS Enzyme II complexes under these conditions.
To further examine the effects of the FruA and FruB proteins on activation of other PTS Enzyme II complexes, recombinant FruB was purified to near homogeneity (S2 Fig), and it was added to crude extracts of the recombinant ΔfruBKA mutant ± overexpression of fruA. The results, presented in Table 5, show that purified FruB had a FruA-dependent activating Table 3. Ratios of the responses of lacZ fusion-bearing strains to various conditions and genetic backgrounds (S13-S15 Tables). WT = wild type, BW25113; OE = overexpression; ΔfruBKA = deletion of the entire fruBKA operon, also called triple mutant, TM. All strains were grown in LB medium to which 0.2% fructose was added, only for the first two entries as indicated. The remaining entries reveal the consequences of the overexpression of specific fru genes or gene combinations on lacZ-fusion gene expression. In no case were the changes significant, suggesting that the presence of fructose in the growth medium or the overexpression (OE) of specific fru genes, or the entire fru operon, did not influence expression of the mtl, gat, or man operons.
ΔfruBKA, fruA OE/ΔfruBKA (S14 Table) 1.1 1.1 1.0 ΔfruBKA, fruB OE/ΔfruBKA (S14 Table) 1.1 1.1 1.0 ΔfruBKA, fruA/fruB OE/ΔfruBKA (S15 Table) 1.0 0.9 1. Interacting PTS integral membrane proteins effect on [ 14 C]mannitol phosphorylation in vitro when PEP was the phosphoryl donor, regardless of whether membrane pellets were used from the WT or ΔfruBKA strain, or whether a crude extract was used from the mutant lacking the fru operon. Purified HPr did not stimulate mannitol phosphorylation. Moreover, while mannitol phosphorylation responded dramatically to the inclusion of FruB in the assay mixture, a lesser activation was observed when N-acetylglucosamine was the  sugar substrate (Fig 5). Activation by FruB but not HPr showed that FruB activation is not due to the activity of the FPr domain of FruB but is dependent on the presence of FruA (Table 5). The crude extract of E. coli strain BW25113-mtlA:kn or BW25113ΔmtlA grown in LB+0.2% fructose was prepared and used to demonstrate the absence of mannitol phosphorylation by FruA (S22 Table). The results eliminate the possibility that stimulation of mannitol phosphorylation was due to cross substrate specificity causing mannitol phosphorylation by EII Fru .

Effect of a high-speed supernatant (HSS), derived from E. coli cells with fruBKA overexpressed, on the in vitro PEP-dependent phosphorylation activities of membrane pellet preparations with overexpressed fruA or galP
In view of the in vitro phosphorylation results described above and presented in Table 5 and Fig 5, a high-speed supernatant [(HSS), supplying all of the soluble proteins of the PTS (Enzyme I, HPr and FruB)] from a strain overexpressing the fruBKA operon, was examined for the effects on phosphorylation activities of membrane pellet preparations overexpressing either fruA or galP (Fig 6 and S24 Table).
Increased expression of fruB in a high speed supernatant (HSS) from cells overexpressing fruBKA, gave a large increase (about 56-fold) of fructose phosphorylation as expected, but mannitol phosphorylation increased 9.1-fold, while N-acetylglucosamine phosphorylation increased 2.7-fold for the overexpressed fruA membrane pellets (Fig 6A). Overexpressed galP membrane pellets did not show activation (Fig 6B; S24 Table). Interestingly, inhibition of [ 14 C]galactitol phosphorylation observed in fruA overexpressed membrane pellets was similarly observed in galP overexpressed membrane pellets. These results suggest that while the activation of mannitol (and to a lesser extent, N-acetylglucosamine) phosphorylation in vitro is  Table. https://doi.org/10.1371/journal.pone.0219332.g005 Interacting PTS integral membrane proteins specific to the fructose PTS enzymes, FruA and FruB, the inhibition of galactitol phosphorylation is not as specific.

Effects of fruA overexpression and purified FruB on sugar-P:sugar transphosphorylation reactions
In contrast to the PEP-dependent phosphorylation reactions catalyzed by the PTS, the sugarphosphate:sugar transphosphorylation reactions depend only on the Enzyme IIB-IIC complex (as well as IID for the mannose system), not on Enzyme I, HPr or the Enzyme IIA proteins. Therefore, these reactions were studied as controls. The phosphorylation of several sugars (mannitol, N-acetylglucosamine, trehalose, methyl α-glucoside and 2-deoxyglucose) were assayed in the presence of either FruA (due to the overexpression of the fruA gene) alone or FruA and variable quantities of purified FruB (0.24, 0.37, 1.48 and 1.85 μg per reaction mixture). As shown in Table 4, in no case was there a significant effect (>50%) on the transphosphorylation reactions examined (see S25 & S26 Tables for the raw data).
These transphosphorylation reactions do not depend on protein:protein interactions, except for the mannose system, in which the IIAB man protein must interact with the IIC man -IID man complex, and in this case, it appears that all constituents of the system are associated in the membrane. In all other cases examined, the IIB and IIC domains are present in a single polypeptide chain (see Table 1). Thus, it appears likely that the effects of varying the expression  Table. https://doi.org/10.1371/journal.pone.0219332.g006 Interacting PTS integral membrane proteins level(s) of one or more PTS IIC/IIB component(s) on the in vivo transport and in vitro phosphorylation activities of other Enzyme II complexes of different specificities is due to facilitation or inhibition of protein-protein interactions within and between the Enzyme II complexes of the PTS. Thus, the results confirm the results obtained using the lacZ fusions and suggest that turnover of the EIIs does not influence the in vivo transport or in vitro PEP-dependent sugar phosphorylation results. FruA and FruB do not promote or inhibit synthesis of other PTS Enzymes. The effects appear to be on the transport and PEP-dependent phosphorylation activities of these enzyme transporters, not on their syntheses or sugar-P:sugar transphosphorylation activities.

Discussion
As noted in the introductory section of this paper, the PTS, found only in prokaryotes, both bacteria and archaea, but not in eukaryotes so far examined, is complex, both in structure and function [2][3][4][5][6][7][8][9][10][11]. It consists of energy-coupling phosphoryl transfer proteins (Enzyme I, HPr, and the sugar-specific IIA and IIB proteins or protein domains) as well as the PTS sugar transporters, the IIC constituents of the systems that catalyze transfer of the phosphoryl moieties of the IIB~P proteins to the incoming sugars [3,4,6] (see Introduction and Table 1). However, it has also been shown that the PTS serves as a chemoreception system [38][39][40][41] and regulates carbon, nitrogen and energy metabolism [42][43][44]. Prior to the present study, the enzyme IIC proteins had not been shown to function as parts of regulatory networks involving direct proteinprotein interactions within the membrane.
In a global interactome study involving both soluble and integral membrane proteins in E. coli K12 [1], we reported that many integral membrane proteins interact with each other, and particularly prominent, were those involving the fructose-specific Enzyme IIC proteins of the PTS (see Table 2). Because these PTS enzyme-transporters can be assayed in vitro as well as in vivo, we investigated the consequences of these interactions. Most striking was the observation that many of the PTS permeases interact with the fructose PTS (FruA/FruB), more than with any other PTS Enzyme II complex (Table 2). However, the biochemical and physiological significance of these interactions was not known. As a result, we carried out the molecular genetic, biochemical and physiological studies reported here, revealing that the activities of the Enzyme IICs of several different PTS sugar-specific transporters are enhanced by their interactions with FruA (the fructose Enzyme IIBC), but that others are inhibited by the corresponding interactions. These effects were large, as great as 20-fold, and far in excess of the standard errors of the measurements. The results also showed that regulation of the uptake of PTS sugars in vivo does not correlate quantitatively with regulation of sugar phosphorylation in vitro. This confirms conclusions of previous studies suggesting that although the two processes, transport and phosphorylation, are coupled under normal conditions in vivo, the rate limiting steps for phosphorylation in vitro are different from those for sugar uptake in vivo [45][46][47]. This can be explained, for example, if cell lysis partially disrupts the relevant protein-protein interactions.
The level of the fructose Enzyme II complex in the membrane proved to determine the extent of activation or inhibition of the target Enzyme II systems. Moreover, the magnitude and direction of regulation depended on the specific target PTS transport system under study. Thus, the glucose, mannose, mannitol and N-acetylglucosamine permeases exhibited enhanced in vivo sugar transport activities, and sometimes enhanced in vitro PEP-dependent sugar phosphorylation activities (but to differing extents), while the galactitol and trehalose systems showed inhibited activities (also to differing extents) when the fructose system was induced to a high level in wild type cells. Most of these effects were shown to occur either due to fruBKA operon induction by the presence of fructose during growth of the wild type cells used for the assay, or to fruA and/or fruB gene overexpression in the absence of fructose in the growth medium. We found that the presence of fructose during assay in vivo or in vitro was unimportant for this regulation, except for induction of the regulating system as noted above. In fact, exhaustive studies failed to reveal a significant role for the presence of the sugar substrate or sugar-phosphate product of the activating or inhibiting Enzyme II complex on the extent of regulation, either during assay or during growth, except for induction of the regulating system.
As controls, first, the sugar-specific transphosphorylation activities of these PTS enzyme/ transporters (dependent only on the presence of the IIBC domains, not on the presence of Enzyme I, HPr or the Enzymes IIA) [16][17][18][19]48], and second, the rates of synthesis of the target PTS permeases, as measured using lacZ transcriptional fusions, were shown not to be affected. These observations clearly suggested that the activating or inhibiting effects observed were due to the consequences of direct protein-protein interactions within the membrane, only on the activities, not the syntheses, of the target enzymes II. The fact that the transphosphorylation reactions of these EIIs were not decreased by high level expression of II Fru additionally suggested that these EIIs did not undergo degradation or turnover.
It seemed reasonable, but was not proven, that regulation depended on interactions of the target PTS permeases with the energy coupling proteins of the PTS as well as the regulating Enzyme II. It should be noted, however, that the interactions might also influence the stabilities of the target Enzymes II, although this seems less likely due to the remarkable stability of the membrane-embedded EIIs as well as the uninfluenced transphosphorylation activities. We thus suggest that specific protein-protein interactions (PPIs) within the cytoplasmic membrane regulate the activities of PTS permeases in a physiologically meaningful way that may contribute to and enhance the importance of the hierarchy of preferred PTS sugars as previously demonstrated when studying catabolite repression [2,49,50]. Since these effects could be demonstrated in the absence of a functional fructose-transporting system, we suggest that it is the direct PPIs, and not the activities of the regulating proteins, that determine the regulatory effects. Thus, it is the levels of the regulating system that seem to determine the degrees of activation or inhibition of the target Enzyme II complexes.
While these conclusions appear valid for the in vivo transport results, the same seems to be true of the PEP-dependent sugar phosphorylation results, although this is less certain as the regulatory effects were substantially smaller. In this regard, both the lack of any regulatory effect on the IIBC-dependent in vitro transphosphorylation reactions and target PTS gene transcription, as well as the lesser regulatory effects in vitro compared to in vivo, strengthen the suggestion that complex formation involving PTS proteins plays an important role, and that cell disruption may have dissociated some of these interactions.
It is interesting to note that in a previous publication, we suggested that the fructose Enzyme IIABC complex was the primordial system, and that all other PTS Enzymes II arose later during evolution of the more complex PTS [20,51]. The original evidence for this suggestion has been summarized in the Introduction. It is relevant to the present studies that only the fructose Enzyme II has a duplicated IIB domain (IIB') that functions in protein-protein interactions and not in phosphoryl transfer [21]. It is possible that this "extra" domain plays a role in the regulatory interactions documented in this report. Moreover, we have demonstrated the biochemical significance of the large number of protein-protein interactions for the fructose PTS, as compared with those for any other sugar-specific Enzyme II complex, again consistent with the suggestion that the fructose PTS was the primordial system.
The observations cited above, explaining the functions of the interactions of the fructose PTS with other Enzyme II complexes, appear to be applicable in principle to other Enzymes II as well as to other (non-PTS) types of transport systems. Thus, in preliminary studies, we found that several Enzymes II, in addition to the fructose Enzyme II, interact with other Enzymes II (see Table 2), having activating or inhibiting effects that depend both on the level of the inhibiting Enzyme II complex and apparently, on the hierarchical position of the target systems under study. Moreover, we found that high level expression of genes encoding the Enzymes II of the PTS influence the activities of certain non-PTS integral membrane transporters in vivo, and vice versa. These observations lead to the suggestion that interactions of many integral membrane transport proteins (and possibly a variety of integral membrane enzymes) occur in the plane of the membrane, influencing their activities, possibly in a physiologically significant and relevant way. The experiments reported here therefore serve as guides to direct further experimentation aimed at defining the consequences of intra-membrane protein-protein interactions. Future studies will be required to determine the extent to which these interactions occur and prove to be physiologically important.

Constructions of bacterial two-hybrid plasmids
The Euromedex bacterial two-hybrid system (Euromedex, Cat # EUK001), consisting of two compatible plasmids, pUT18 and pKNT25, were designed to detect protein-protein interactions in vivo. pUT18 encodes Ap resistance and carries the ColE1 origin and the Plac promoter driving T18 domain of cAMP synthase of Bordetella pertussis, while pKNT25 encodes Kn resistance and carries the p15A origin and the Plac promoter driving T25 domain of cAMP synthase. To determine any possible interactions between two target proteins, one gene was fused to the N-terminus of T18 in pUT18 while the other was fused to the N-terminus of T25 of pKNT25. These two plasmids were co-transformed into an E. coli strain lacking its own cAMP synthase gene cyaA. In the case in which the two target proteins closely interacted, thereby bringing together the T18 and T25 domains, the E. coli cells synthesized cAMP, thereby exhibiting unique phenotypes that can be assessed.
Similarly, fruA, fruB, mtlA and nagE were individually fused to the N-terminus of the T25 domain in pKNT25. Each of these resultant plasmids carries a single hybrid gene encoding the target protein at the N-terminus and the T25 domain at the C-terminus. These plasmids are denoted pKNT25-fruA, pKNT25-fruB, pKNT25-mtlA and pKNT25-nagE, respectively (S27 Table).
To test the possible interaction between two proteins, a recombinant pUT18 plasmid and a recombinant pKENT25 plasmid were co-transformed by electroporation into the E. coli BTH101 strain deleted for cyaA. The transformants were inoculated onto MacConkey agar plates supplemented with maltose, kanamycin (Kn), ampicillin (Ap) and IPTG. The plates were incubated at 30˚C for up to 48 h. The appearance of red colonies indicated positive interactions between the target proteins. The intensity of red color was proportionally correlated to the level of protein-protein interactions.

Constructions of fruA, fruB, fruBKA and mtlA deletion mutants
Strains (and plasmids) and DNA oligonucleotides used in this study are described in Supplementary S27 and S28 Tables, respectively. The deletion mutants of fruA, fruB, fruBKA and mtlA were generated from the parental strain (E. coli K-12 strain BW25113) using a standard method as described in [52]. Briefly, to construct each mutant, a kanamycin resistance gene (kn), flanked by the FLP recognition site (FRT), amplified from the template plasmid pKD4 using a pair of specific mutation oligos (S28 Table), was first substituted for the target gene or operon. Where indicated, the kn gene was subsequently eliminated (leaving an 85-bp FRT sequence) using plasmid pCP20 that bears the FLP recombinase. Replacements of the target genes/operons by the FRT-flanking kn gene and subsequent removal of the kn gene were confirmed by colony PCR and DNA sequencing, yielding deletion mutants ΔfruA, ΔfruB, ΔfruBKA and ΔmtlA, respectively (S27 Table).

Construction of the Ptet driven fruBKA strain
Using plasmid pKDT:Ptet [53] as template, the DNA region containing the kn gene and the Ptet promoter was PCR amplified using a specific primer pair, PtetfruB-P1/PtetfruB-P2 (S28 Table). The PCR products were integrated into the BW25113 chromosome to replace the fruBKA promoter (P fruBKA between the 123 th nucleotide and the 1 st nucleotide relative to the translational start point of fruB). This chromosomal integration led to Ptet-driven fruBKA expression (strain BW_Ptet-fruBKA) (S27 Table).

Construction of overexpression plasmids
pMAL-p2X [54] was used to overexpress various cytoplasmic and transport proteins in E. coli. This plasmid carries a ColE1 origin, lacIq and a strong IPTG inducible promoter Ptac, driving malE gene expression (useful in making N-terminal fusions for protein purification purposes). To make a control plasmid (carrying Ptac but not malE), pMAL-p2X was digested with BglII (+435 to +440 relative to the start site of malE) and SalI (located downstream of malE in the multiple cloning site region) and then re-ligated, yielding pMAL-empty which is the same as pMAL-p2X except that most of malE has been removed.
The structural regions of fruA (encoding the fructose uptake transporter), fruB (encoding the fructose-specific PTS multiphosphoryl transfer protein) and galP (encoding the galactose: H + symporter), were PCR amplified from BW25113 genomic DNA (using specific pairs of primers as indicated in S28 Table), digested with NdeI and BamHI, and then ligated into the same sites of pMAL-p2X individually. In each resultant recombinant plasmid, the target structural gene (no promoter and no 5' UTR) is substituted for malE in pMAL-p2X, and its expression is exclusively under the control of the IPTG inducible promoter Ptac. These plasmids were referred to as pMAL-fruA, pMAL-fruB and pMAL-galP, respectively (S27 Table).

Chromosomal PmtlA-lacZ, Pman-lacZ and PgatY-lacZ Transcriptional Fusions
The mtlA promoter region (-386 bp to +57 bp relative to the mtlA translational start point), the man promoter region (-234 bp to +54 bp relative to the manX translational start point) and the gatY promoter region (-204 bp to +42 bp relative to the gatY translational start point), each plus a stop codon at the 3' end, were amplified from BW25113 genomic DNA. These DNA fragments, referred to as promoter regions (each containing a promoter region, the first 14 to 19 codons plus a stop codon at the end) were then inserted between the XhoI and BamHI sites of the plasmid pKDT [53], yielding the plasmids pKDT_PmtlA, pKDT_Pman and pKDT_PgatY, respectively. In each of these newly made plasmids, an rrnB terminator (rrnBT) is present between the kn gene and the downstream cloned promoter. The DNA fragments containing the kn gene, rrnBT and those of the promoter regions (plus the first 14 to 19 codons followed by a stop codon) were PCR amplified from the above plasmids and individually integrated into the chromosome of MG1655 carrying the seamless lacY deletion [53] to replace lacI and PlacZ but not the 5' UTR of lacZ. All of these chromosomal integrations were confirmed by colony PCR and subsequent DNA sequencing analyses. These promoter-lacZ fusions were individually transferred to BW25113 by P1 transduction, yielding strains BW_PmtlA-lacZ, BW_Pman-lacZ and BW_PgatY-lacZ, respectively. Similarly, these reporters were transferred to ΔfruBKA, yielding strains ΔfruBKA_PmtlA-lacZ, ΔfruBKA_Pman-lacZ and ΔfruBKA_PgatY-lacZ, respectively (S27 Table). In each of these reporter strains, the promoter of interest drives the first 14 to 19 codons of the target gene followed by the lacZ' 5' UTR and the structural gene.

β-Galactosidase assays
The E. coli reporter strains were grown in 5 ml of media contained in 18 mm diameter glass test tubes under the same conditions as for the uptake experiments. During incubation, samples were removed for measurements of OD 600nm and β-galactosidase activities after being appropriately diluted.
To measure β-galactosidase activities, 0.8 ml of Z-buffer containing β-mercaptoethanol (2.7 μl/ml) and SDS (0.005%) were mixed with 0.2 ml of sample and 25 μl of CHCl 3 in test tubes [55]. The tubes were vortexed twice (each time for 10 seconds) at a constant speed and incubated in a 37˚C water bath until equilibration. A 0.2 ml aliquot of ONPG substrate (4 mg/ ml) was then added to each test tube. When yellow color developed, the reaction was stopped by adding 0.5 ml 1 M Na 2 CO 3 followed by vortexing. After that, the reaction mixtures were centrifuged, and the absorbance values of the supernatants were measured at 420 nm and 550 nm. A control tube was run in parallel using diluted or undiluted LB broth instead of the test sample. β-galactosidase activity was expressed in Miller units = [(OD420-1.75XOD550)/[(sample volume in ml) X time in min X OD 600nm ] X 1000 X dilution factor [56].

Culture conditions for uptake of radioactive substrates
A fresh culture (100 μl of the test strain) was used to inoculate 5 ml of LB (in the case of wild type or mutant strains) or LB plus either 100 μg/ml ampicillin for pMAL plasmid harboring strains, or 100 μg/ml ampicillin plus 25 μg/ml chloramphenicol for strains harboring the pMAL/pZA31P tet M2-GFM plasmids. The 18 mm diameter tubes were incubated in a shaking water bath at 250 rpm and 37˚C for about 6 h. Aliquots of 100 μl of the cultures were used to inoculate 50 ml of LB/5 mM MgSO 4 with or without 0.2% fructose (in the cases of wild type or mutant strains, and strains containing the P tet promoter in their chromosomes), or 50 ml of LB plus either 100 μg/ml ampicillin or 100 μg/ml ampicillin plus 25 μg/ml chloramphenicol (in the cases of the dual plasmid-harboring strains mentioned above) contained in 250 ml conical flasks, which were incubated in a shaking water bath at 250 rpm and 37˚C for 8 h (in the cases of wild type and mutant strains and strains containing the P tet promoter in their chromosomes) or 6 h, followed by a 2 h induction period with IPTG at a final concentration of 0.2 mM (recombinant plasmid harboring strains). During the IPTG induction period, fructose (0.2%) was included in certain experiments, and the shaking rate and the incubation temperature were lowered to 200 rpm and 30-32˚C, respectively. The cells were harvested by centrifugation in a Sorvall centrifuge at 4˚C and 10,000 rpm for 20 min. The cell pellets were washed 3x (in the absence of added sugar in the growth medium) or 4x (when a sugar was present in the growth medium). Each wash was with 35 ml of 50 mM Trizma-maleate buffer, pH 7, containing 5 mM MgCl 2 . The pellets were resuspended in the same buffer to OD 600nm values of 0.5, 0.25 or 0.125 for radioactive substrate uptake experiments.

Radioactive substrate uptake
For the uptake assays, the stock solution of each radioactive substrate was used at 5 μCi/μmole for 14 C while radioactive tritium was used at 30 μCi/μmole, and in all cases, the substrate concentration of the stock solution was 1 mM. Whenever stated, uptake assays were performed in the presence and absence of certain added non-radioactive sugars. For all uptake assays, the prepared bacterial suspensions of the test strains were used within a time period not exceeding 12 h. The assay mixtures contained 900 μl of a bacterial cell suspension of specified OD 600nm , 50 μl of 1 M arginine (pH 7), and 20 μl of the 1 mM stock radioactive substrate. The volume was brought to 1 ml with a cold sugar solution and/or 50 mM Trizma-maleate buffer, pH 7, containing 5 mM MgCl 2 .
The final concentration of the radioactive sugar substrate in the assay mixture was 20 μM for uptake assays and 10 μM for phosphorylation assays unless otherwise stated. The uptake assays were carried out in a shaking water bath at 37˚C for 5-10 min followed by immediate filtration of withdrawn samples (100-250 μl) through 0.45 micron membrane filters under vacuum. The filters containing cells were washed 3x with cold 50 mM Trizma-maleate buffer, pH 7, containing 5 mM MgCl 2 before being dried under infra-red lamps. Each dried filter was mixed with 10 ml Biosafe NA solution in scintillation vials, and the radioactivity, expressed as counts per min (CPM), was measured in a Beckman scintillation counter. For normalization of values among samples for different test strains, the radioactivity was expressed as CPM/0.1 OD 600nm /0.1 ml/min.

Culture conditions for phosphorylation assays
A fresh culture of the test strain was used as an inoculum which was prepared by inoculating 5 ml contained in an 18 mm diameter test tube or 50 ml contained in 250 ml conical flasks of LB medium without or with the appropriate sugar (e.g., fructose at 0.2%) and antibiotic(s) (100 μg/ml ampicillin (Ap) for pMAL plasmid harboring strains, or 100 μg/ml ampicillin plus 25 μg/ml chloramphenicol (Cm) for strains harboring the pMAL/pZA31P tet M2-GFM plasmids). The flasks were incubated at 37˚C, either in a gyratory shaking water bath at 250 RPM for the 250 ml flasks, or in a shaking incubator at 275 RPM for the 2 l flasks for 8 h (wild type, mutant strains and strains containing the chromosomal P tet promoter), or for 6 h plus a 2 h induction period (recombinant strains). Induction was carried out by adding IPTG to a final concentration of 0.2 mM at 32˚C and 200 RPM. The cells were harvested by centrifugation at 4˚C, washed 3x with cold modified M63, and then re-suspended in about 7 ml (pellets from 50 ml cultures) or 30 ml (pellets from 1 l cultures) of modified M63 containing 5 mM DTT. The prepared cell suspensions were disintegrated by three passages through a French press at 12,000 PSI. The resultant cell lysates were centrifuged at 10,000 RPM for 10 min at 4˚C in a SORVALL centrifuge, and the supernatants produced, termed crude extracts (CE), were either used directly for PEP-dependent phosphorylation assays of PTS enzymes II or subjected to ultracentrifugation in a Beckman centrifuge, Ti-70 titanium rotor, for 2 h at 40,000 RPM. The pellets produced were re-suspended in an appropriate volume of modified M63/5 mM DTT to give the membrane pellets (MP) that were used for PEP-dependent and transphosphorylation assays of PTS enzymes. The 2 h high speed supernatants (HSS) produced were kept overnight on ice at 4˚C and then subjected to one or more 2 h centrifugation cycles under the previously mentioned conditions. This was occurred once, twice or thrice to get 4, 6 or 8 h HSSs, which were used as sources of soluble PTS enzymes for PEP-dependent phosphorylation assays. The repeated centrifugations were necessary to remove the last traces of the membranous Enzymes II [16,54].

PTS phosphorylation assays
These assays were performed as previously described by Aboulwafa and Saier, 2002 [57]. For the PEP-dependent reactions, Enzyme II preparations were either membrane pellets (MP), or crude extracts (CE), and the following assay mixtures were used: 50 mM potassium phosphate buffer (pH 7.4), 10 μM [ 14 C]sugar, or [ 3 H]sugar (in the case of galactitol), 5 mM phosphoenolpyruvate (PEP), 12.5 mM MgCl 2 , 25 mM KF and 2.5 mM dithiothreitol (DTT). For the sugar-P-dependent reactions (transphosphorylation), the same was applied except that membrane pellets only were used as the enzyme II source, PEP was replaced by a sugar-P (usually at a 1000-fold higher concentration than that of the radioactive sugar unless otherwise stated), and the pH of the potassium phosphate buffer was 6.0. The sugar phosphates used were: glucose-6-phosphate with [ 50-100 mesh (Sigma-Aldrich). After 4x washing of columns containing resin with deionized water, the 1 M lithium chloride eluate (9 ml), containing the radioactive sugar-P, was mixed with 10 ml of Biosafe II solution for radioactivity measurements in a Beckman scintillation counter. Cold sugar, sugar phosphate, the soluble purified FruB enzyme or the purified HPr protein was incorporated in the reaction mixtures at the specified concentrations.

Protein purification
The His-tagged FruB from an E. coli strain BW25113-fruBKA:kn-pMAL-fruB culture, overexpressing the fruB gene or the equivalent strain overexpressing the ptsH (HPr) gene, was purified by nickel affinity chromatography. The cells were grown in LB and induced for 2 h with 0.2 mM IPTG as stated before for the preparation of cells for the transport assays. The cells were harvested by centrifugation, washed 3x with modified M63 and subjected to 3 cycles of freezing at -20˚C and thawing before being lysed with B-PER Complete Bacterial Protein Extraction Reagent (from Thermo Scientific, Cat # 89821) following the manufacturer's instructions. His-FruB or His-HPr in the cell extract was purified using HisPur TM -Ni-NTA Superflow Agarose (from Thermo Scientific, Cat # 25214) following the manufacturer's instructions. The eluates, produced from the agarose columns, containing purified His-FruB or His-HPr, were desalted by passage through preequilibrated (with modified M63/5 mM DTT) Zeba TM Spin Desalting columns (Thermo Scientific, Cat # 89894). Each Zeba column contained 10 ml of size exclusion resin MWCO 7000 Da. The desalted preparations were concentrated using Amicon ultra centrifugal filter devices, MWCO 5000 Da, Cat # UFC 900502. The purity of the resultant preparation was checked by SDS-PAGE as described by Aboulwafa and Saier, 2011 [54]. Phenylmethylsulfonyl fluoride (PMSF) was incorporated into the purified preparation at 1 mM before being aliquoted and stored at -80˚C.

Determination of protein concentrations
The protein concentrations were determined using the Biorad colorimetric protein assay (Cat. #500-0006) with bovine serum albumin as the standard protein.

Materials
All radioactive sugars were purchased from New England Nuclear (NEN) Corp. or American Radiolabeled Chemicals (ARC). Nonradioactive compounds were from commercial sources, usually from the Sigma Chem. Corp. unless otherwise noted, and were of the highest purity available commercially.
Supporting information S1   Table. Oligonucleotides used in this study. (DOCX) S1 Fig. Results from the use of a bacterial two hybrid system for testing interactions of FruA and FruB with other membrane proteins. pKNT25-fruA (expression in low copy plasmid pKNT25) against pUT18-gatC, pUT18-nagE, pUT18-treB, pUT18-mtlA, and pUT18-fruB (expression in high copy plasmid pUT18); and pUT18-fruB against pKNT25-mtlA and pKNT25-nagE are presented. As reported by Babu et al., 2018, FruA interacts with other membrane proteins. In this study, the interactions of FruA could be confirmed for GatC, NagE, TreB and MtlA, and the interactions of FruB could be detected and reproduced for MtlA and NagE. The interactions of FruB, the soluble partner, with FruA can be considered as a positive control. Note that the interactions of FruB with MtlA and NagE appear to be substantially weaker than those with FruA.