Efflux pumps of the resistance-nodulation-division (RND) superfamily are common in Gram negative bacteria [7, 28] and are well studied for their role in antibiotic resistance and solvent tolerance in many Pseudomonas species [29, 30]. However, these may not be the native or dominant physiological functions of RND pumps in bacteria. Piddock  and Poole , among others, have suggested that RND pumps fulfill other crucial roles, including management of diverse physico-chemical and biochemical stresses, quorum sensing and virulence. One of the stress-responsive roles proposed for RND efflux pumps such as MexCD-OprJ in Pseudomonas aeruginosa [4, 7, 31] is the export of membrane constituents released by FA replacement due to natural turnover of membrane components during cell growth or resulting from membrane damage. Our results are consistent with that proposal: EmhABC appears to play a role in efflux of replaced membrane FA in response to temperature-induced membrane perturbation, in addition to its demonstrated function of transporting hydrophobic antibiotics, dyes and PAHs . Reciprocally, because RND efflux pumps are membrane-associated protein complexes, EmhABC activity may in turn be influenced by modulation of FA content in response to membrane stressors like temperature and hydrophobic compounds  that partition into lipid bilayers. As expected, our results indicate that temperature affects cell growth, FA content (measured as fatty acid methyl esters) and membrane integrity (measured as permeability index). In addition, we have now shown that temperature affects expression and activity of the EmhABC RND efflux pump (measured by using RT-qPCR, phenanthrene efflux and antibiotic MIC assays).
The FA content of cLP6a followed the expected trends at 10°C and at 35°C, shifting towards unsaturation and saturation respectively [11, 32]. The FA content of the membrane affected the partitioning of phenanthrene into the membrane, since cLP6a-1 cells grown at 35°C contained lower fractions of phenanthrene in the absence of active efflux compared to those grown at 28°C. This observation is consistent with the rationale that saturated FA pack closely, hindering partitioning of hydrophobic molecules like PAHs into the lipid bilayer  whereas angular cis-unsaturated FA pack more loosely, facilitating partitioning. The observed changes in FA with temperature are also consistent with results from the membrane integrity assay in which the permeability index increased with temperature.
Growth temperature also affected EmhABC activity in cLP6a, possibly indirectly through membrane perturbation including the modulation of FA. cLP6a cells having high unsaturated FA content (i.e., 72% in cells grown at 10°C) and greater membrane integrity had higher efflux activity than cells with lower proportions of unsaturated FA (i.e., 14% at 28°C or 4% at 35°C) and increased permeability. This observation suggests that increased unsaturated FA content may allow efficient or stable association of the three protein components of RND efflux pumps, which spans two membranes and the periplasm.
The enhanced phenanthrene efflux observed in cLP6a at 10°C is consistent with the additive effect of EmhABC with a postulated alternate efflux pump that is active at 10°C. The presence of an alternate pump in P. fluorescens is not unexpected, as multiple efflux pumps have been identified in other Pseudomonas species [2, 7] and additional efflux pumps were invoked by Hearn et al.  to explain anthracene and fluoranthene efflux in P. fluorescens strain cLP6a.
The induction of emhABC genes was observed in cLP6a cells exhibiting major changes in membrane FA composition due to sub-optimal growth conditions, namely at 10°C, 35°C and in the presence of tetracycline. Expression was also increased in logarithmic phase cells, which undergo rapid synthesis and turnover of FA, and in death phase cells that experience membrane deterioration. The relationship between induction of emhABC genes and membrane FA modulation indicates that the EmhABC efflux pump may be involved in the extrusion of replaced membrane FA as a result of membrane turnover. This possibility is further supported by the higher concentration of free FA in the medium of cLP6a cultures grown at 35°C concomitant with high membrane permeability and over-expression of emhABC genes. Comparable results were obtained recently by Stickland et al.  who reported that over-expression of mexCD-oprJ efflux genes in P. aeruginosa led to up-regulation of FA secretion and fitness impairment. Over-expression of emhABC genes in cLP6a cells grown at 35°C may be explained either as compensation for reduced activity of EmhABC (caused by the modulation of the FA content) or may be due to increased membrane permeability and membrane FA turnover. According to Denich et al. , damage to the membrane is still possible even with modulation of membrane FA quantity or composition to maintain fluidity and integrity. Our conclusion is supported by the observation of similarly high levels of emhABC over-expression in log phase cells. Such cells may have compromised cell membranes due to rapid phospholipid synthesis and turnover since membrane integrity is temporarily affected by physical cell wall reconstruction at the sites of cell division during the log phase of growth [33, 34]. It is unclear why there was differential expression of the three emhABC genes in log phase cells (emhA > B > C), although stability of the transcripts may differ as a result of rapid cell growth. The effect on membrane integrity was confirmed by the higher permeability index at 35°C. Similarly, the reduced cell yields and growth rates at 35°C compared to 10°C or 28°C, along with altered FA content, are consistent with compromised cell membranes at the higher temperature. The negative effects of the compromised membrane on growth are muted by the presence and activity of EmhABC, allowing cLP6a cells to out-grow cLP6a-1 at supra-optimal temperature.
The discovery that EmhABC activity influences growth of P. fluorescens cLP6a (and by extension wild type LP6a) at supra-optimal temperature suggests a role for efflux in temperature adaptation in the environment, and may apply to other Gram-negative species. For example, P. aeruginosa and Salmonella strains lacking RND efflux pumps are unable to colonize and infect their hosts [1, 35], which may in part result from an inability to adapt to host temperatures higher than the external environment. Temperature also may affect efflux-mediated antibiotic resistance although the effect on MIC was not pronounced in P. fluorescens cLP6a. It will also be interesting to examine whether temperature-sensitive efflux of antibiotics is a general phenomenon in other Gram-negative bacteria. Because bacterial cells are commonly exposed to temperature changes in the environment, we propose that RND efflux pumps in Gram-negative bacteria may play a major role in management of temperature-induced membrane damage.
Our study focussed on modifications to the FA portion of membrane lipids since phospholipid head group modification is typically less dynamic and critical in bacteria (reviewed by Denich et al. ), but it is possible that head group composition also changed in response to temperature, PAHs and/or antibiotics. Other indirect effects such as decreased proton motive force resulting from damaged membranes could also be factors. Such possibilities are incentives for clarifying the natural physiological roles of RND efflux pumps in Gram-negative bacteria in anticipation of devising new methods for combating antibiotic resistance or improving hydrocarbon transformation for bioremediation or biocatalytic processing of hydrophobic substrates.