Characterization of Proteobacterial Plasmid Integron-Encoded qac Efflux Pump Sequence Diversity and Quaternary Ammonium Compound Antiseptic Selection in Escherichia coli Grown Planktonically and as Biofilms

ABSTRACT Qac efflux pumps from proteobacterial multidrug-resistant plasmids are integron encoded and confer resistance to quaternary ammonium compound (QAC) antiseptics; however, many are uncharacterized and misannotated. A survey of >2,000 plasmid-carried qac genes identified 37 unique qac sequences that correspond to one of five representative motifs: QacE, QacEΔ1, QacF/L, QacH/I, and QacG. Antimicrobial susceptibility testing of each cloned qac member in Escherichia coli highlighted distinctive antiseptic susceptibility patterns that were most prominent when cells grew as biofilms.

To characterize proteobacterial plasmid qac sequence diversity and homology, we collected 2,953 qac sequences encoded by proteobacterial plasmids deposited in the GenBank, INTEGRALL (14), UniProt (https://www.uniprot.org), and Comprehensive Antimicrobial Resistance Database (CARD) (15) databases. Plasmid sequences were retrieved using QacE (WP_000679427.1) as a query sequence by tBLASTn (16) analysis. After performing a multiple-sequence alignment of translated Qac sequences with the online server Clustal Omega (17) in Jalview (18), we identified a total of 37 unique Qac protein sequences for final analysis, all with highly variable and inconsistent annotations ( Fig. 1; see Tables S1 and S2 in the supplemental material). To accurately classify each Qac sequence, we performed a maximum likelihood phylogenetic analysis using PhyML v. 3.0 (19), with archetypical SMR members EmrE (NP_415057.1) and Gdx/SugE (NP_418572.4) as SMR family comparators (Fig. 2). This analysis reconfirmed that all Qac members were closely related to the SMR family member EmrE, in agreement with previous findings (3,20). It also revealed that Qac sequences grouped into one of three distinct clades: Qac annotated as (i) QacF/L/H/I, (ii) QacG or QacE, and (iii) QacE and QacED1 (Fig. 2). The alignment of the 37 unique Qac sequences revealed 5 sequence motifs in all 4 transmembrane helices distinguishing QacE from QacED1 and from QacG. The alignment also identified that QacF/L annotated sequences as well as QacH/I were in fact identical to each other (98 to 100% identity) (Fig. 1). Amino acid variations in each Qac motif occurred most often (61 to 82% frequency) at unconserved residue positions in the previously published SMR L K NWL F     Proteobacterial Plasmid-Transmitted qac Efflux Pumps Antimicrobial Agents and Chemotherapy motif (3, 4) (Fig. 1). Our attempts to identify a Qac sequence progenitor from bacterial genomes using tBLASTn were unsuccessful, as many qac genes are also transmitted on chromosomally integrated integrons as well as phages/prophages. Based on the high pairwise sequence identities between each Qac to either QacE or EmrE (Fig. 1), we propose that qac sequences have likely originated from a single qac progenitor incorporated into an integron that is rapidly diverging over time into these qac variants. This sequence analysis reconfirmed that qacED1 is the predominant qac representative (2,736 qacED1 genes/2,953 total qac genes [92.6%]) ( Fig. 2; see Fig. S1 and Table S1 in the supplemental material), given that qacED1 is part of the 39 conserved region of most class 1 integrons (9). The remaining qac genes were less frequently detected from plasmids (8% of 2,368 plasmids surveyed), where qacG was the second most predominant member (90/2,953 [3.0%]), followed by qacH/I (82/2,953 [2.8%]), qacF/L (29/2,953 [1.0%]), and qacE (16/2,953 [0.5%]) ( Fig. 2; Fig. S1 and Table S1). The majority of all qac sequences we identified were from class 1 integrons (93.0 to 100% of all plasmids), with a few qac genes detected at very low frequency (,4%) from class 2 or 3 integrons (Fig.  S1). This indicates that qac genes predominate in class 1 integrons, but caution should be taken when using these genes as genetic markers for class 1 integrons.

L A T S I I F E V I A T S A L K S S E G F T R L V P S F I V V AG Y A A A F Y F L S L T L K S I P VG I A Y
To determine the substrate selectivity of the five representative Qac sequence motifs, we gene synthesized and cloned qacE (NP_044260.1), qacED1 (YP_003264406.1), qacF/L (YP_006961976.1), qacH/I (L0FU64), and qacG (YP_006965429.1) in the isopropyl-b-D-1-thiogalactopyranoside (IPTG)-inducible expression vector pMS119EH (21) (Tables 1 and 2), using the same cloning, plasmid expression, and AST methods described in a recent study of gdx/sugE (22). We chemically transformed each plasmid into Escherichia coli K-12 BW25113 (wild type) (23), as well as strain KAM32 (24), which lacks a competing dominant efflux pump gene, acrB, and an additional efflux pump gene, mdtK, improving qac substrate selectivity determination by AST. To determine differences in antiseptic resistance that may be attributed to different cell growth physiologies, as noted in our previous study (21), we performed three different AST culturing techniques in Luria-Bertani (LB) medium with 100 mg/ml ampicillin selection and 0.05 mM IPTG addition to determine the MICs for each cloned vector transformant. We measured planktonic growth using 96-well broth microdilution plating techniques and cell colony growth on agar spot plating as described by Slipski et al. (22). We also determined the minimal biofilm eradication concentration (MBEC) for transformants grown as biofilms using the MBEC device (Innovotech, Inc., Canada) as described in reference 22. All AST involved a library of 13 antimicrobials commonly tested in previous SMR studies (as reviewed in reference 4) (Tables 1 and 2). For AST, we applied a significance threshold of 4-fold or greater when determining differences in MIC and MBEC values to account for potential 2-fold-value errors. Nearly all BW25113/pQac transformants we examined using broth or agar AST methods had MIC values that were statistically insignificant (within a 2-fold MIC difference or identical) to the vector control pMS119EH, with the exception of the pEmrE transformant (Table 1). BW25113/pEmrE transformant agar colony AST results showed higher QAC resistance (MIC values of $4-fold) to ethidium bromide (ET) and acriflavine  (AC) than to the control vector pMS119EH (Table 1). This is in agreement with previous agar spot plate AST findings for emrE transformants exposed to these intercalating dyes (25). In broth, only BW25113/pEmrE conferred significant resistance (.4-fold MIC) to methyl viologen (MV) compared to all other transformants and controls (Table 1); MV is one of the original substrates initially identified for EmrE (26). All BW25113 qac gene transformants grown as biofilms demonstrated a significant increase in resistance ($4-fold change) to at least one QAC, intercalating dye, and/or antibiotic for each SMR transformant compared to the parental vector control ( Table 1). The biofilm AST results shown in Tables 1 and 2 were repeated in duplicate based on 6 transformed biological replicates (n = 6). The biofilm AST profile of recognized antimicrobial compounds was unique for each pQac transformant we tested, reflecting their sequence motif differences (Fig. 1). BW25113/pQacH transformant biofilms conferred resistance to the broadest range of antimicrobials (6 QACs plus tobramycin [TOB]), with pQacE, -F/L, or -G transformants resistant to 3 to 4 antimicrobials, and pQacED1 expectedly conferring resistance to the fewest substrates (cetrimide bromide [CET] and MV in Table 1). Therefore, in the wild-type efflux pump BW25113 strain, pQacE, pQacF, pQacH, and pQacG transformants grown as biofilms confer unique antimicrobial resistance profiles to a limited range of QACs compared to pEmrE (Table 1).
To improve substrate selection identification conferred by each representative qac, we repeated AST with KAM32 DacrB DmdtK/pQac transformants ( Table 2). As previously reported (22,27), KAM32 has slower growth and higher drug susceptibility than BW25113, resulting in lower MIC and MBEC values for all antimicrobials we tested compared to BW25113/pMS119EH (Table 2). Broth and agar spot plate AST results for KAM32 transformed with pEmrE or pQac vectors (including pQacED1) demonstrated a significant increase ($4-fold) in MIC values for one or more QACs compared to pMS119EH (Table 2) or compared to the same AST results from BW25113 transformants (Table 1). These findings show that AST in strains lacking competing efflux pumps helped identify a broader range of QACs selected for by each qac gene when grown planktonically or as colonies. The KAM32 agar AST findings are in agreement with previous qac studies, as we identified increased resistance to similar QAC substrates (ET, cetyltrimethylammonium bromide [CTAB], and benzalkonium chloride [BZK]) (Tables 1  and 2), as reported for previous agar plate AST studies of qacE and qacED1 (9, 12, 13), as well as QacF (10,11). However, KAM32/pQac transformant biofilm AST unexpectedly identified fewer antimicrobials that significantly increased MBEC values compared to the control vector (Table 2) or compared to BW25113 biofilm results (Table 1). KAM32 transformant biofilm MBEC results identified enhanced antimicrobial susceptibility (#4-fold reduction in MBEC values) for pQacED1 and pQacF exposed to QACs, CET, BZK, didecyldimethylammonium bromide (DDAB), and CTAB (Table 2). Enhanced susceptibility was also observed for biofilm BW25113/pEmrE and -pQacE transformants for CTAB (Table 1). This suggests that overexpression of these qac efflux pumps works against the cell under these biofilm growth conditions, making cells more susceptible to the aforementioned QACs. The ability of SMR members to confer enhanced antimicrobial susceptibility in the presence of different antimicrobials has been reported in previous studies (28,29) and may be due to amino acid variations that switch these pumps from exporters to importers for these particular drugs.
In conclusion, our findings reveal that many proteobacterial plasmid-carried qac genes are misannotated in sequencing databases, and the comprehensive Qac motif comparison herein can improve annotation of qac variants. We observed that qacH/I variants had the broadest antimicrobial recognition profile when grown as biofilms, whereas qacED1 transformants conferred significant QAC resistance to the smallest number of QACs (CET and MV), indicating that even this relatively inactive qac variant can still confer limited QAC resistance. Our analysis also importantly shows that qac efflux pumps are most effective when E. coli cells grow as a biofilm and least effective when cells grow as planktonic cultures, which is concerning when considering that biofilm prevention and eradication strategies frequently rely on the use of QAC disinfectants (30). Altogether, this information provides more context to ongoing antimicrobial resistance genetic surveillance studies by providing qac-specific antimicrobial phenotypes to uncharacterized qac genes, clear and improved annotations, and identification of optimal growth physiologies influencing their conferred phenotypes.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.1 MB. SUPPLEMENTAL FILE 2, XLSX file, 0.22 MB.