Exploring the Role of Integral Membrane Proteins in ATP-Binding Cassette Transporters: Analysis of a Collection of MalG Insertion Mutants

Abstract

The maltose transport complex of Escherichia coli is a well-studied example of an ATP-binding cassette transporter. The complex, containing one copy each of the integral membrane proteins MalG and MalF and two copies of the peripheral cytoplasmic membrane protein MalK, interacts with the periplasmic maltose-binding protein to efficiently translocate maltose and maltodextrins across the bacterial cytoplasmic membrane. To investigate the role of MalG both in MalFGK₂ assembly interactions and in subsequent transport interactions, we isolated and characterized 18 different MalG mutants, each containing a 31-residue insertion in the protein. Eight insertions mapping to distinct hydrophilic regions of MalG permitted either assembly or both assembly and transport interactions to occur. In particular, we isolated two insertions mapping to extracytoplasmic (periplasmic) regions of MalG which preserved both assembly and transport abilities, suggesting that these are permissive sites in the protein. Another periplasmic insertion seems to affect only transport-specific interactions between MalG and maltose-binding protein, defining a novel class of MalG mutants. Finally, four MalG mutant proteins, although stably expressed, are unable to assemble into the MalFGK₂ complex. These mutants contain insertions in only two different hydrophilic regions of MalG, consistent with the notion that a restricted number of domains in this protein are critical complex assembly determinants. These MalG mutants will allow us to further explore the intermolecular interactions of this model transporter.

Integral membrane proteins play a central role in the ATP-binding cassette (ABC) transporter superfamily, whose prokaryotic and eukaryotic members traffic a variety of substrates such as ions, sugars, amino acids, peptides, and proteins (15). This large family of transporters is defined by a conserved cytoplasmic ATPase component and integral membrane domains which interact to carry out the specific transport process (4, 15). Among the eukaryotic members are such medically relevant proteins as the P-glycoprotein implicated in multidrug-resistant cancer cells, the cystic fibrosis transmembrane regulator protein, and the human peroxisomal adrenoleukodystrophy protein (2, 34, 35). Among the prokaryotic members of the ABC superfamily are the periplasmic binding protein-dependent transporters. These family members are characterized by a conserved region of the integral membrane component(s) in addition to the conserved cytoplasmic ATPase (4). One member of this prokaryotic subgroup, the maltose transport complex of Escherichia coli, presents a useful model for the integral membrane folding and assembly interactions required for ABC transporters. The maltose transport complex consists of the integral membrane proteins MalF and MalG and a peripheral cytoplasmic membrane ATPase, MalK (reviewed in reference 24). These three proteins copurify (11), forming a MalFGK₂ tetrameric complex which acts in concert with the periplasmic maltose-binding protein (MBP), the product of malE, to efficiently translocate maltose and maltodextrins across the bacterial cytoplasmic membrane.

MalF has been shown to have eight transmembrane (TM) domains (5), whereas MalG possesses six TM domains (6, 10). Following independent insertion of these proteins into the membrane (22a, 31), assembly of the MalFGK₂ complex is likely mediated by interactions among discrete domains of MalF, MalG, and MalK, resulting in tetramerization (20, 26).

Although the specifics of these interactions are unknown, a combination of biochemistry and genetics has allowed for a partial characterization of the complex. Shuman and colleagues isolated and characterized MalF and MalG mutants which enable the MalFGK₂ complex to transport maltose in the absence of MBP (7, 32). These analyses have pointed toward a direct interaction between MBP and periplasmic portions of MalG and MalF (16), between MalG and MalF themselves (7), and between MalK and both MalF and MalG (12). Davidson and Nikaido purified the MalFGK₂ complex and demonstrated extensive chemical cross-linking between MalG and MalF and among MalG, MalF, and MalK (11). Traxler and Beckwith observed that periplasmic loops of MalF become protease resistant only in the presence of MalG and MalK, also suggesting that specific interactions occur among the proteins in the context of an assembled complex (31). Finally, a potentially important MalG-MalK protein interaction signal has been identified in the hydrophilic cytoplasmic loop between the fourth and fifth TM domains of MalG (reference 9; Fig. 1). This motif is conserved in MalF and in other binding protein-dependent transporters of the ABC superfamily (9, 28) and has been hypothesized to mediate interactions with the conserved ATPase subunit of the complex (17, 22).

FIG. 1.

Topology model of MalG. Hydropathy plots and fusion protein analyses (6, 10) suggest that the N and C termini of the 296-residue protein are cytoplasmically localized. The shaded boxes represent putative TM domains, and the shaded amino acids are conserved in integral membrane proteins of periplasmic binding protein-dependent ABC transporters (9, 28). The location of each 31-residue insertion is shown by an arrowhead. The black arrowhead represents an insertion which did not significantly affect MalG transport function, the gray arrowhead depicts partial transport function, and the white arrowheads represent loss of transport ability for the corresponding insertion mutants. Each numbered disc shows the mutant classification of the adjacent insertion mutant (see Discussion for details).

Recently, a transposon-mediated insertion mutagenesis technique was developed and used to characterize both permissive and nonpermissive regions of the integral membrane protein LacY (19), as well as the cytoplasmic MalK and LacI proteins (18, 23). These analyses not only identified tolerant hydrophilic regions of each protein but also defined several distinct mutant classes (18, 19, 23). In particular, the phenotypes attributable to the lacI insertion mutations that we isolated were strikingly similar to those of previously characterized amino acid substitutions mapping to the same sites in lacI. Here, we describe the results of this insertion mutagenesis on the MalG protein. This analysis provides a unique in vivo view of the requirements for proper MalG protein folding and of the interactions necessary for MalFGK₂ assembly and maltose transport.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The E. coli strains used in this study are described in Table 1, and the plasmids used are described in Table 2. The malG::i31 alleles were crossed from their pmalG plasmids onto λDBK261 (14) by homologous recombination between upstream promoter regions and between downstream bla sequences during infections of BT45 containing the various plasmids. Strain BT10 (lacI) was then lysogenized with these λmalG::i31-transducing phages to create strains BN30 to BN42, which constitutively express the malG alleles (Table 1).

TABLE 1.

Bacterial strains used in this study

Strain	Relevant genotype	Reference or source
CC118	Δ(ara-leu)7697 Δ(lac)X74 phoA20 galE galK thi rpsE rpoB argE(Am) recA1	19
CC191	CC118 F128lacIq Δ(lacZ)M15	19
NT205	araD139 ΔlacU169 rpsL thi ΔmalE444 malF500	32
BT60	F128lacIq/araD139 ΔlacU169 rpsL thi relA malT(Con)-1	Lab collection
BT10	Δ(lac pro)X111 araD rpsL rpoB malG(Am) malT(Con) zhe::Tn10 thi	31
BT45	BT10 F128lacIq	6
BN20	BT10 recA::cat	This study
BN27	BT45 (Tets) pcnB zad::Tn10	This study
BN30	BT10λ malG565::i31	This study
BN31	BT10λ malG566::i31	This study
BN32	BT10λ malG571::i31	This study
BN33	BT10λ malG572::i31	This study
BN34	BT10λ malG574::i31	This study
BN35	BT10λ malG575::i31	This study
BN36	BT10λ malG577::i31	This study
BN37	BT10λ malG578::i31	This study
BN38	BT10λ malG579::i31	This study
BN39	BT10λ malG580::i31	This study
BN40	BT10λ malG581::i31	This study
BN41	BT10λ malG582::i31	This study
BN42	BT10λ malG+	This study

TABLE 2.

Bacterial plasmids used in this study

Plasmid	Relevant genotype	Reference
pGAP1	ΔmalF derivative of pGA1 (6) which encodes the first 17 amino acids of MalG fused to alkaline phosphatase	This study
pTrc99A	Derivative of pKK233-2 with bla, lacIq, and trc promoter	1
pBDN4	pTrc99A with malG	This study
pNT7	pSC134 (carries ori region of low-copy-number plasmid pSC101) with malE and its promoter region	32
Plasmids containing in-frame malG fusionsa
pBGZ112/pmalG565	pBDN4 Φ(malG112′lacZ)/pBDN4 malG565::i31	This study
pBGP205/pmalG566	pBDN4 Φ(malG205′lacZ)/PBDN4 malG566::i31	This study
pBGP253/pmalG567	pBDN4 Φ(malG253′lacZ)/PBDN4 malG567::i31	This study
pBGZ271/pmalG568	pBDN4 Φ(malG271′lacZ)/PBDN4 malG568::i31	This study
pBGP277/pmalG569	pBDN4 Φ(malG277′lacZ)/PBDN4 malG569::i31	This study
pBGZ301/pmalG570	pBDN4 Φ(malG301′lacZ)/PBDN4 malG570::i31	This study
pBGZ361/pmalG571	pBDN4 Φ(malG361′lacZ)/PBDN4 malG571::i31	This study
pBGP445/pmalG572	pBDN4 Φ(malG445′lacZ)/PBDN4 malG572::i31	This study
pBGZ514/pmalG573	pBDN4 Φ(malG514′lacZ)/PBDN4 malG573::i31	This study
pBGZ580/pmalG574	pBDN4 Φ(malG580′lacZ)/PBDN4 malG574::i31	This study
pBGZ595/pmalG575	pBDN4 Φ(malG595′lacZ)/PBDN4 malG575::i31	This study
pBGZ661/pmalG576	pBDN4 Φ(malG661′lacZ)/PBDN4 malG576::i31	This study
pBGZ688/pmalG577	pBDN4 Φ(malG688′lacZ)/PBDN4 malG577::i31	This study
pBGZ733/pmalG578	pBDN4 Φ(malG733′lacZ)/PBDN4 malG578::i31	This study
pBGZ739/pmalG579	pBDN4 Φ(malG739′lacZ)/PBDN4 malG579::i31	This study
pBGP763/pmalG580	pBDN4 Φ(malG763′lacZ)/PBDN4 malG580::i31	This study
pBGZ871/pmalG581	pBDN4 Φ(malG871′lacZ)/PBDN4 malG581::i31	This study
pBGZ886/pmalG582	pBDN4 Φ(malG886′lacZ)/PBDN4 malG582::i31	This study

^aEach is named according to the malG base pair directly preceding the malG-lacZ or malG-phoA fusion junction. GZ, malG-lacZ fusion; GP, malG-phoA fusion. Following the removal of most of the IS element, which yields a 31-codon insertion in malG, each plasmid is designated by the malG allele. Based on DNA sequencing, one insertion (after malG bp 514) was independently recovered twice, two insertions (after malG bp 271 and 595) were recovered three times, and one insertion (after malG bp 733) was recovered five times. 

To create plasmid pBDN4, the mal region of pHS2 (25) was restricted at the HpaI site upstream of malG and at the downstream SnaBI site. This malG fragment was then ligated into the SmaI site of a plasmid containing the M13 mp10 and mp11 multiple cloning sites (5) to form plasmid pBDN1. Following BamHI site removal from the polylinker, the malG fragment was restricted and ligated into the pTrc99A polylinker (1) by using the upstream EcoRI and downstream XbaI sites of each multiple cloning site. The expression of malG on pBDN4 and of its insertion derivatives are under control of the trc promoter and can be induced with isopropyl-β-d-thiogalactopyranoside (IPTG) (Table 2).

Media and chemicals.

The minimal (M63), rich (LB), and MacConkey media used were described previously (19, 21). The following medium supplements were used at the indicated concentrations: sucrose, 5% (wt/vol); maltose, 0.2 or 1% for minimal or MacConkey medium, respectively; glycerol, 0.2% for minimal medium; maltodextrin (from Pfanstiehl), 0.2% for minimal medium (2 mM [final concentration] maltodextrins between maltotetrose and maltoheptose); chloramphenicol, 30 μg/ml; kanamycin, 30 μg/ml; ampicillin, 100, 75, or 25 μg/ml for high-, low-, or single-copy-number conditions, respectively; tetracycline, 15 or 10 μg/ml for high- or low-copy-number conditions, respectively; 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal), 40 μg/ml; 5-bromo-4-chloro-3-indolylphosphate, p-toluidine salt (X-P), 40 μg/ml; and IPTG, 1 mM except where noted otherwise.

DNA techniques.

Standard DNA preparations and manipulations were used as previously described (27). To locate the insertion site of a transposon element, plasmid DNA was sequenced by the dideoxynucleotide termination method with Sequenase (United States Biochemical), using double-stranded DNA templates and the TnlacZ-II oligonucleotide primer (19).

Transposon mutagenesis.

E. coli strains CC118 and CC191 containing malG on plasmid pBDN4 were infected with a replication-deficient lambda phage carrying the TnlacZ/in or TnphoA/in transposon, using the method developed by Manoil and Bailey (19; see references 18 and 23 for adaptations). Transposition of the ISlacZ/in or ISphoA/in element into pBDN4 was selected on ampicillin- and chloramphenicol-containing medium (with resistances specified by the plasmid and insertion sequence [IS] element, respectively). In-frame transposition events into malG were screened by addition of the color indicator X-Gal or X-P to the selective medium. All but 93 nucleotides from each inserted IS element were then removed by BamHI digestion to create the in-frame 31-codon insertion, shown here in the one-letter amino acid designation: (S, P, A, or T)DSYTQVASWTEPFPFSIQGDPRSDQET(G, A, V, E, or D)XX. The amino acids denoted as X are specified by the two codons directly 5′ of each insertion event; these codons are duplicated as part of the transposition event.

Maltose transport assays.

Mal phenotypes of the MalG insertion mutants were initially assayed in strain BN20 transformed by plasmids expressing the various malG alleles (Table 2). Colony morphologies of the resulting strains were examined on maltose-MacConkey and maltose minimal medium in the absence of IPTG. In addition, the phenotypes of selected MalG mutants produced by strains BN30 to BN42 were examined in a similar fashion.

[¹⁴C]maltose uptake was quantified for insertion mutants MalG566 and MalG578 in strains BN31 and BN37, respectively, using BT10 as a negative control and BN42 as a positive control. All strains were grown at 37°C in LB or LB-low ampicillin to an A₆₀₀ of 0.5. Following two washes and a resuspension in M63 salts, the cell density was normalized to a final A₆₀₀ of 0.4. Fifty-microliter volumes of these cell resuspensions were assayed with [¹⁴C]maltose (specific activity, 591 mCi/mmol; Amersham Corp.) and total maltose at final concentrations of 0.5 and 5.0 μM, respectively. At time points ranging from 15 s to 10 min, these mixtures were collected on 0.45-μm-pore-size HA Millipore filters. Filters were then washed with a total volume of 7 ml of 0.5 M LiCl, dried, and counted in scintillation vials with Aquasol (Du Pont-NEN). Assays were performed in duplicate at 37°C and corrected for no-label background counts. Results are presented as a relative percentage of wild-type (BN42) uptake, using the average uptake per minute in linear-range assays (BN42, 2,495 cpm; BT10, 29 cpm, or 1.2% of the wild-type level; standard error of the mean ≤ 15% of total cpm for each assay).

Proteolysis assay of MalFGK₂ complex assembly.

The ability of each MalG insertion mutant to assemble into a stable MalFGK₂ complex was tested by assaying trypsin sensitivity of MalF in each mutant strain as described previously (18, 31). E. coli BN27 (pcnB) was transformed with the different malG insertion mutant plasmids, pTrc99A or pGAP1 as a negative control and pBDN4 as a positive control. The resulting strains were grown in M63 medium (with glycerol, ampicillin [75 μg/ml], and all amino acids except cysteine and methionine) at 37°C. At an A₆₀₀ of between 0.2 and 0.3, expression of malG was induced in each strain by the addition of 1 mM IPTG for 2 h, whereupon the cells were converted to spheroplasts (31). Strains BN30 to BN42 were examined in a similar fashion, except that 25 μg of ampicillin per ml was used in the M63 medium and no IPTG was added prior to spheroplast conversion. MalF protein in the cell extracts was detected after sodium dodecyl sulfate-polyacrylamide gel electrophoresis by Western blot analysis.

Western blot analysis.

Western blot analysis was performed as previously described (23, 31). Protein levels were normalized prior to gel loading by use of the Bio-Rad protein assay, and the MalG mutant proteins were detected with a 1:1,000 dilution of polyclonal antibody specific for the 31-codon insert (19). For Western blot analyses following proteolysis assays, protein levels were normalized according to final A₆₀₀ (BN27-transformed strains) or according to Bio-Rad protein assays (strains BN30 to BN42). MalF protein was detected by a polyclonal antibody specific for MalF (31).

RESULTS

Isolation of malG insertion mutations.

We wished to investigate the role of various regions of the MalG protein in maltose transport function and in maltose transport complex assembly. As one approach to characterizing MalG, we utilized the TnlacZ/in and TnphoA/in transposon mutagenesis technique as described by Manoil and Bailey (19; see also references 18 and 23) to generate 31-residue insertion mutations in malG. The first step of this transposon mutagenesis involves the recovery of translational malG-lacZ or malG-phoA fusions created from the transposition of the ISlacZ/in or ISphoA/in transposon element, respectively. From approximately 140,000 colonies possessing plasmid-linked ISlacZ/in insertions, we identified about 250 malG-lacZ fusion candidates based on blue colony phenotypes on X-Gal-containing medium. Likewise, from approximately 36,000 colonies possessing plasmid-linked ISphoA/in insertions, we identified about 50 malG-phoA fusion candidates on X-P-containing medium. After analyzing these fusions by restriction enzyme digestion, we sequenced 28 malG-lacZ candidates and 11 malG-phoA candidates and recovered 18 different in-frame fusions to malG. The second step of the transposon mutagenesis, a BamHI-mediated removal of all but 93 nucleotides of the IS element, yielded 18 different 31-codon insertion mutations in malG. Each of these mutations was given an allele number (Table 2); the positions of the 18 insertions as they relate to MalG membrane topology are shown in Fig. 1.

Transport activity of MalG insertion mutants.

The transport phenotypes of the 18 insertion mutants were initially characterized on maltose-containing media. When examined in strain BN20, only periplasmic insertion mutant MalG566 retained transport activity similar to that of wild-type MalG. We additionally tested 12 of the MalG insertion mutants at single-copy-number (strains BN30 to BN41, with BN42 as a positive control), since constitutive malG expression in a lacI strain may allow the identification of weakly active mutants. Consistent with this idea, we observed an increase of approximately 1.5- to 2-fold in relative MalG mutant protein levels in the six constitutive malG expression strains tested compared to their uninduced BN20/pmalG counterparts (data not shown).

Under these constitutive malG expression conditions, MalG566 in strain BN31 retained both maltose (Mal⁺ phenotype) and maltodextrin (Dex⁺ phenotype) transport activity indistinguishable from that of wild-type MalG and approximately 43% of wild-type activity when measured by maltose uptake assays at 37°C. Additionally, insertion mutant MalG578 in strain BN37 retained a small but significant level of maltose transport activity on MacConkey-maltose agar and about 3% of wild-type uptake activity at 37°C. MalG578, however, displayed a Mal⁻ phenotype on MacConkey-maltose agar at 30 and 42°C, and all other MalG insertion mutants lacked observable maltose transport function at 30, 37, and 42°C (Fig. 1). In agreement with its maltose transport phenotype, MalG578 also grew, albeit poorly, on maltodextrin minimal medium at 37°C. Surprisingly, Mal⁻ mutants MalG581 (BN40) and MalG582 (BN41) also displayed a weak but reproducible growth phenotype on maltodextrin minimal medium.

MalG mutant protein production.

We determined the relative protein abundance of the 18 MalG insertion mutants by Western blot analysis using antiserum directed against the IS (Fig. 2). In addition to the MalG insertion mutant band migrating at about 29 kDa, a background band migrating at about 33.3 kDa was detected. Of the 18 MalG insertion mutants, 12, including 2 mutants active in transport and 10 deficient in transport, accumulated protein to significant levels. These results indicate that the loss of transport activity by 10 MalG insertion mutants is not attributable to an absence of MalG mutant protein in the cells. Observed mobility differences among several of the mutant MalG proteins are not attributable to transposon-mediated rearrangements in malG since restriction digest analysis confirmed the expected size and location of each malG insertion (data not shown). Mobility differences have also been observed among similar 31-residue insertion mutants of MalK and LacI (18, 23).

FIG. 2.

Western immunoblot analysis of MalG mutant proteins. Proteins from BN20-derived cell extracts were detected by incubation with affinity-purified antibody directed against the 31-residue insert (19). MalG⁺ (from BN20/pBDN4) served as the negative control since it lacks the insert. The background band of approximately 33 kDa appeared in all lanes. The locations of the size standards, in kilodaltons, are shown on the left.

The remaining six insertional mutant proteins (MalG567, MalG568, MalG569, MalG570, MalG573, and MalG576) were undetectable or present at significantly lower levels in cells, as seen by the Western blot analysis (Fig. 2). These mutants likely are unstable polypeptides, as their insertions occur in or near putative TM domains of MalG. In agreement with these results, earlier studies of lac permease have suggested that 31-residue insertions in TM domain sequences may generally lead to decreased protein recovery (19).

MalG mutant assembly competence.

We wished to ascertain whether individual MalG mutants were compromised in the ability to assemble the MalFGK₂ complex. To this end, we used the MalF protease sensitivity assay as described by Traxler and Beckwith (31). Briefly, periplasmic loops of MalF are cleaved by exogenous proteases such as trypsin in the absence of proper MalFGK₂ complex formation. Upon complex formation, however, MalF becomes resistant to these same proteases. Therefore, by assaying MalF protease susceptibility in the presence of individual MalG mutants, we can gauge the ability of these mutants to form a stable MalFGK₂ complex.

Previously, we analyzed several MalG-PhoA fusion proteins initially characterized by Boyd et al. (6) to determine whether C-terminally truncated MalG derivatives could associate with MalF and MalK, leading to protease-resistant MalF. We observed assembled, protease-resistant MalF only with the MalG-PhoA fusion 7, in which PhoA is fused after the final MalG residue (data not shown). We concluded that analysis of truncated MalG proteins provides only limited information about assembly interactions; therefore, our continued studies focused on full-length MalG disrupted by the 31-residue insertions.

We assayed the assembly competence of the MalG insertion mutants in either the low-plasmid-copy number strain BN27 (data not shown) or at single copy number (Fig. 3). Whereas the MalG⁺ control largely protected MalF from proteolytic degradation, lack of MalG resulted in essentially complete trypsin degradation of MalF (compare MalG⁺ and MalG⁻ control lanes in Fig. 3). For the 12 stable mutants possessing insertions in hydrophilic regions of MalG, several different MalFGK₂ assembly competence phenotypes were observed. Insertion mutants MalG566, MalG571, MalG580, MalG581, and MalG582 protected MalF from trypsin proteolysis at or near MalG⁺ levels (Fig. 3 and data not shown). Insertion mutant MalG565 provided intermediate levels of protection to MalF, and MalG572 and MalG578 reproducibly provided lower yet substantial protection. Insertion mutants MalG574, MalG575, MalG577, and MalG579, however, failed to protect MalF from trypsin proteolysis (Fig. 3), indicating that they are unable to assemble properly with MalF and MalK.

FIG. 3.

MalFGK₂ assembly competence of selected MalG mutants. The Western blots show the abilities of different MalG insertion mutants to assemble into a MalFGK₂ complex in which MalF is protected from trypsin proteolysis. The blots shown are representative of at least three proteolysis assays for each insertion mutant; 10 μg of protein from each of strains BN30 to BN39, BN41, BN42 (malG⁺), and BT10 [malG(Am)] was loaded per lane. The position of full-length MalF is indicated. The two prominent bands directly below full-length MalF are high-molecular-weight tryptic MalF peptides (31). Numbers above the blots refer to the different malG alleles tested. −, no-protease controls; +, 25 μg of trypsin added. A faint band corresponding to intact MalF is observed in the proteolyzed BT10 sample; this low-level protection may be due to incomplete spheroplasting or read-through expression of the chromosomal malG(Am) allele in this background. MalG581 in strain BN40, which is not shown, protected MalF from proteolysis at levels similar to those for MalG582 in BN41.

The introduction of the 31-codon insertion and its novel trypsin cleavage site into periplasmic domains of MalG did not detectably alter the ability of the MalF protease sensitivity assay to monitor complex assembly. The MalG566 and MalG580 mutant proteins were trypsin sensitive in the proteolysis assays described above (data not shown) yet still enabled MalF to acquire its protease-resistant conformation. Therefore, the cleavage of MalG during this assay does not lead to destabilization of the complex and proteolytic cleavage of MalF.

The expression of the six low-abundance, TM domain-associated mutants (MalG567, MalG568, MalG569, MalG570, MalG573, and MalG576) in BN27 did not enable MalF to assemble into its protease-resistant conformation (data not shown). Since their lack of assembly was most likely due to the instability of the MalG mutant proteins themselves, they were not further characterized.

Mutant MalG-MBP interactions.

Shuman and colleagues have identified mutations in malF and malG which allow maltose transport in E. coli cells in the absence of MBP (7, 32). One of these mutant strains, NT205 (ΔmalE malF500 malG⁺ malK⁺), grows well on maltose minimal medium but is Mal⁻ when transformed with a plasmid producing wild-type MBP (32). This transport block is thought to be due to a nonproductive association between the MBP-independent MalF500GK₂ complex and wild-type MBP (32, 33). We reasoned that if this nonproductive interaction prevents maltose transport, a class of MalG mutant that disrupts the MBP interaction but which retains MalFGK₂ assembly competence and is otherwise active in transport might be detected.

To test this possibility, we assayed our periplasmic insertion mutants MalG565, MalG566, MalG572, MalG578, and MalG580, plus the cytoplasmic insertion mutant MalG582 as a negative control. We transformed the respective malG plasmids into strain NT205 containing pNT7 (a compatible low-copy-number plasmid which produces MBP) and noted each resulting strain phenotype on maltose-MacConkey agar. Insertion mutant MalG580, whose insertion junction maps to the third periplasmic loop of MalG, reproducibly allowed a partial restoration of transport function, observed as a dark pink “fisheye” colony morphology of strain NT205/pNT7/pmalG580. This result suggests that malG580 is partially dominant to wild-type malG in this background and that low-level transport is restored by interference of the MBP interaction with the MalF500G580K₂ complex. The other MalG derivatives each failed to restore any detectable maltose transport to NT205/pNT7.

MalG mutant dominance assays.

To test the MalG insertion mutants for potential transdominant negative phenotypes, we initially transformed the Mal⁺ strain BT60 with each of the pmalG plasmids and observed the colony phenotypes on MacConkey-maltose agar. All of the mutant malG alleles were recessive to malG⁺, even when their expression was induced with 10 μM IPTG. Further induction of malG allele expression from the high-copy-number plasmids severely impaired colony formation on the agar plates. We additionally examined MalG mutant transdominance in strain NT205, which possesses a 3-log reduction in external maltose-binding affinity compared to a wild-type transport system (32). We reasoned that the decreased transport activity of NT205 might allow us to detect partial dominance effects of MalG mutants not observable in the wild-type background. We transformed NT205 with each of the 12 pmalG plasmids expressing stable MalG protein, plus pmalG567 (low-abundance protein) and pBDN4 (wild-type MalG) as negative controls, and again observed the colony morphologies on maltose-MacConkey medium. In the absence of IPTG, transdominance over malG⁺ was observed only for allele malG571, resulting in a Mal⁻ phenotype. In the presence of 10 μM IPTG, however, we observed transdominance over malG⁺ for malG571, malG572, malG577, malG578, malG579, malG581, and malG582 (summarized in Table 3).

TABLE 3.

Summary of MalG insertion mutants

Allele	Insertion site		Mutant class	Maltose transport activityc	Relative protein levele	MalFGK2 assembly competencef	Transport activity in NT205g
	bpa	aab
malG565	112	Ile37	II	−/−	++	+	+
malG566	205	Glu68	I	++/++	++	++	+
malG567	253	Trp84	IV	−/NT	−+	−	+
malG568	271	Lys90	IV	−/NT	−+	−	NT
malG569	277	Ala92	IV	−/NT	−+	−	NT
malG570	301	Val100	IV	−/NT	−+	−	NT
malG571	361	Thr120	II	−/−	+	++	−+
malG572	445	Gly148	II	−/−	+	+−	−h
malG573	514	Ala171	IV	−/NT	−	−	NT
malG574	580	Ala193	III	−/−	++	−	+−
malG575	595	Thr198	III	−/−	++	−	+−
malG576	661	Ile220	IV	−/NT	+−	−	NT
malG577	688	Glu229	III	−/−	+	−	−
malG578	733	Thr244	I	−/+−d	+	+−	−
malG579	739	Ala246	III	−/−	++	−	−
malG580	763	Asn254	II	−/−	+	++	+−
malG581	871	Thr290	II	−/−	++	++	−
malG582	886	Lys295	II	−/−	++	++	−
malG+	NA	NA	NA	++/++	NT	++	+

^aThe base pair in the malG sequence directly preceding each 31-codon insertion; NA, not applicable. 

^bThe amino acid (aa) of MalG directly preceding each 31-residue insertion site. 

^cAssayed on maltose-MacConkey medium. ++, wild-type activity; +, intermediate activity; +−, low activity; −, no activity; NT, not tested. The first phenotype corresponds to that of uninduced pmalG plasmids in BN20 [malG(Am) recA::cat]. The second phenotype corresponds to that of the BT10λ malG lysogens which constitutively express the malG allele (strains BN30 to BN41). 

^dInsertion mutant MalG578 retained maltose transport activity only at 37°C in strain BN37. 

^eTaken from Western blot analysis (Fig. 2) and expressed as a range from ++ (highest level of protein) to − (no detectable protein). 

^fAssayed by using the MalF protease sensitivity assay in strain BN27 (pcnB) transformed with the various pmalG plasmids, pTrc99A (negative control), and pBDN4 (positive control). In addition, strains BN30 to BN41 were assayed with BN42 as a positive control and BT10 as a negative control (see Results for details). ++, highest protection of MalF; +, good protection of MalF; +−, partial protection of MalF; −, no protection of MalF. 

^gmalG alleles were expressed from pmalG plasmids in strain NT205; maltose transport phenotype was observed on maltose-MacConkey medium–10 μM IPTG; see footnote c for phenotype designations. 

^hNT205/pmalG572 colonies also displayed a transport-defective phenotype (−) in the absence of IPTG induction. 

DISCUSSION

As one step toward a more complete view of the participation of the MalG protein in MalFGK₂ complex assembly and in maltose transport, we have used a transposon-mediated insertion mutagenesis strategy to generate 18 insertion mutations of 31 codons each in the malG gene. A summary of the resulting mutant protein phenotypes is shown in Table 3. We have identified two distinct periplasmic regions of MalG which are mutationally tolerant to various degrees, suggesting that the immediately surrounding regions do not play a vital role in assembly of the MalFGK₂ complex or in the transport process. In addition, we have identified several hydrophilic regions of MalG which are not critical for complex assembly but which are necessary for transport function; this distinction will enable a further examination of transport-specific protein interactions mediated by these regions. In particular, the identification of a MalG periplasmic region which is likely involved in interactions with MBP will allow a more extensive characterization of MalG-MBP interactions. Our insertion mutagenesis results also support the idea that a discrete set of hydrophilic domains of MalG, including portions of the third cytoplasmic and third periplasmic domains, are important for assembly-specific interactions. To facilitate the following discussion, we have grouped the MalG insertion mutants into four classes based on similar phenotypes (Table 3).

Class I mutants define tolerant regions of MalG.

Class I mutants MalG566 and MalG578 retain both assembly competence and significant maltose transport function, suggesting that the periplasmic regions of MalG affected by these insertions are surface exposed within the context of the functional complex. The region around residue Glu68 of periplasmic loop 1, where the insertion junction of MalG566 maps, seems to be particularly tolerant of large insertions, as MalG566 retained about 43% of wild-type MalG transport activity at 37°C. MalG578 maps to the third periplasmic loop of MalG after residue Thr244 and possesses partial assembly competence, a low but significant level of transport function (3% of wild-type MalG activity at 37°C), and a temperature-sensitive transport phenotype observed at 30 and 42°C. The temperature-sensitive phenotype seems to be a result of altered transport-specific interactions, since the partial assembly competence of MalG578 (measured by MalF protease protection) was the same at 30, 37, and 42°C in strain BN37 (data not shown). This partial assembly competence is also supported by the observation that malG578 is transdominant to malG⁺ in strain NT205. One possible explanation for these results is that MalG578 is partially defective for transport-specific interactions with MalF.

Class II mutants are impaired in transport-specific interactions.

Class II mutants MalG565, MalG571, MalG572, MalG580, MalG581, and MalG582 all possess a Mal⁻ phenotype but produce significant levels of protein and protect MalF from proteolysis to various degrees. Therefore, these mutants are specifically defective for maltose transport but retain the ability to assemble the MalFGK₂ complex. MalG571, MalG581, and MalG582, whose insertions map to cytoplasmic regions of MalG, may disrupt transport-specific contacts with MalK or with cytoplasmic regions of MalF. In contrast, insertion mutants MalG565, MalG572, and MalG580 may disrupt transport-specific interactions with periplasmic regions of MalF or with MBP.

The assembly competence of MalG565 is consistent with the observations of Dassa (8), who isolated several partially active linker insertion mutants mapping to this region of MalG. Hence, the first portion of the first periplasmic loop of MalG is apparently dispensable for MalFGK₂ assembly. The assembly competence of MalG572, on the other hand, is supported by the strong dominance of allele malG572 over malG⁺ in strain NT205, resulting in a Mal⁻ phenotype.

Class II mutant MalG580, whose insertion junction maps to the third periplasmic loop of MalG, defines a region of MalG likely involved in transport-specific interactions with MBP. Interestingly, the MalG580 insertion maps adjacent to a previously isolated linker insertion (following Pro255) whose resulting Mal^+/− but Dex⁻ mutant phenotype suggested the presence of either a substrate specificity site or an MBP interaction domain in the region (8). MalG580 in the presence of MalF500, MalG⁺, MalK⁺, and MBP⁺ restores a low level of maltose transport to the complex, strongly suggesting that a defective interaction exists between MalG580 and MBP. Since the inhibition of MalF500GK₂-mediated maltose transport by MBP is likely due to a nonproductive association between them (32, 33), MalG580 may partially restore transport ability by decreasing the affinity of MalF500G580K₂ for MBP and thus preventing the initial docking of MBP to the complex. We propose that part of the third periplasmic loop of MalG is normally involved in the initial contact between MBP and MalFGK₂, which is thought to precede MBP-mediated activation of MalK ATPase and substrate translocation (7, 30). We will continue to focus on this periplasmic region of MalG in an effort to further define its involvement in maltose transport and MBP-specific interactions.

Class II mutants MalG581 and MalG582 both have insertion junctions which map to the C-terminal cytoplasmic domain of MalG. The unusual Mal⁻ but Dex^+/− phenotype attributed to these mutants is not readily explainable; this phenotype, however, was previously identified among point mutants mapping to the sixth TM domain of MalF (13), leading the authors to speculate on the presence of multiple contact sites for the substrate within the transport channel itself. The role of the C-terminal cytoplasmic domain of MalG in this model is unclear, but the ability of MalG581 and MalG582 to transport maltodextrins and the transdominance of malG581 and malG582 to malG⁺ in strain NT205 confirms the MalFGK₂ assembly competence of these insertion mutants.

Class III mutants are MalFGK₂ assembly deficient.

Class III mutants MalG577 and MalG579, mapping to the third periplasmic loop of MalG (Fig. 1), fail to transport maltose and are defective in complex assembly, as demonstrated by the MalF protease sensitivity assay (Fig. 3). We additionally observed that malG577 and malG579 alleles are transdominant to malG⁺ in strain NT205. These mutant phenotypes may be explained by oligomerization of the mutant MalG proteins with MalK but not with MalF500 (or MalF), thus forming a “dead-end” partial complex and preventing transport. A similar albeit stronger transdominant negative mutant blocking MalFGK₂ assembly was identified previously in MalK (18). We therefore propose that the third periplasmic domain of MalG defined here by insertion mutants MalG577 and MalG579 is a previously unidentified assembly motif, specifying interactions with MalF.

In contrast, class III mutants MalG574 and MalG575 map to the conserved region of MalG which is thought to directly interact with MalK (Fig. 1; references 9, 17, and 28), and both mutant alleles are recessive to malG⁺ in strain NT205 (Table 3). The importance of this conserved hydrophilic MalG loop is supported by the recent isolation of substitutions at conserved residues in this region and in that of MalF. Several of these mutations abolished transport function, which was subsequently restored by suppressor mutations isolated in malK (22). Site-directed mutageneses of other ABC transporters have similarly established the importance of the conserved hydrophilic domain for the activity of the prokaryotic FhuB transporter and for the yeast peroxisomal Pxa1p transporter (3, 29).

Class IV mutants are likely defective in protein stability.

Class IV mutants MalG567, MalG568, MalG569, MalG570, MalG573, and MalG576 displayed a Mal⁻ phenotype, produced protein at low or undetectable levels in strain BN20 (Fig. 2), and failed to protect MalF from trypsin proteolysis in strain BN27. As the insertion junctions of these mutants map in or near putative TM domains, their primary defect is likely one of improper intramolecular folding, leading to protein instability and degradation by endogenous cellular proteases (19).

Although the specifics of MalFGK₂ assembly and maltose transport remain largely unknown, we have characterized a collection of MalG insertion mutants which have provided new clues. In particular, the identification of a largely tolerant periplasmic region of MalG and of other regions which are important for transport function but not for MalFGK₂ assembly has helped to further define the roles of various regions of MalG in each process. In addition, our analysis has pointed toward a region of MalG which may be important for transport-specific interactions with MBP. Finally, our results suggest the presence of assembly determinants in the third periplasmic loop of MalG in addition to those in the conserved cytoplasmic domain of MalG, as previously postulated (9, 22, 28). The identification of only two insertion-permissive sites in MalG is strikingly different from a similar analysis of LacY in which 10 of 12 insertions in hydrophilic regions retained some transport activity (19). This difference suggests that proteins like MalG, whose activities are dependent on multiple intermolecular contacts, may be less tolerant of insertions than are monomeric proteins like LacY. The further characterization of MalG should thus provide new insights not only into the nature of integral membrane protein folding and oligomerization but also into those interactions which are unique to the ABC transporter superfamily.