[page 70↓]

5  Discussion

In the first part of this thesis, we characterized the transmembrane movement and transverse distribution of fluorescent phospholipid analogues across the inner membrane of E.coli. For this analysis, we used isolated IIMV and reconstituted proteoliposomes from detergent extracts of inner membrane vesicles (IIMV) of E.coli. To determine the transbilayer movement of phospholipids, a recently developed stopped-flow BSA back-exchange assay was established. We could show that this new approach is also applicable for IIMV and reconstituted systems.

We found that the transbilayer movement of the analyzed phospholipid analogues across IIMV membranes was head-group independent and very rapid. We observed that the rapid flip-flop of phospholipid analogues was restored in reconstituted vesicles from detergent extracts of IIMV. Moreover, a rapid transbilayer movement of fluorescent long-chain PE analogues was found. Our investigations revealed a strong requirement of proteins for the rapid transmembrane movement of phospholipid analogues.

In the second part of the thesis, we aimed at purifying the proteins responsible for the observed protein dependent flip-flop of fluorescent phospholipid analogues. To isolate the putative flippase(s) we used ion exchange chromatography. To our surprise, we were not able to enrich specific flippase activity in any of the analyzed fractions by this method, indicating that flippase activity is not mediated by one specific protein but rather by at least two distinct facilitators or by the presence of proteins within the membrane.


[page 71↓]

5.1  Transbilayer movement of short-chain, fluorescent phospholipid analogues in IIMV and reconstituted proteoliposomes derived from IIMV

To investigate the transmembrane movement of short-chain fluorescent phospholipid analogues across the inner membrane of E.coli, we combined the BSA back-exchange assay with a stopped-flow technique. This combination allowed us to directly record fluorescence changes that occurred during the extraction of fluorescent labeled phospholipid analogues from IIMV and proteoliposomes with high time resolution. Fluorescence kinetics were analyzed via the three-compartment model (Marx, et al., 2000), to deduce the rate constants (and respective half-times) of transbilayer movement, and the distribution of phospholipid analogues in the two leaflets of the bilayer.

To ensure that approaches used in this study for the determination of transmembrane movement of fluorescent phospholipid analogues were appropriate, a series of experiments were carried out to demonstrate that i) the analogues are equilibrated between both leaflets of IIMV at the beginning of the BSA back-exchange stopped-flow assay, and ii) BSA extracts all analogues in the time course of the assay. Otherwise, the estimation of half-times of transmembrane movement of analogues, and their transbilayer distribution would not have been possible. As shown in chapter 4.2 (Figure 8 and Figure 9), these preconditions were fulfilled by the chosen experimental set-up.

The applicability of the stopped-flow BSA back-exchange approach depends on the quantitative relation between the rate constants for the extraction of analogues by BSA, and the rate constants for the transbilayer movements of analogues. The latter can be measured by this approach only, if the extraction step is significantly faster compared to the transmembrane movement of the analogues. Apparent from the initial rapid fluorescence decline and the subsequent slower phase of fluorescence decrease, this assumption holds true in our system. For example, we found half-times of fluorescent analogues extraction of fluorescent PE analogues from protein-free liposomes by BSA of about eight seconds (see chapter 4.3). On the other hand, the half-times of transbilayer movement of fluorescent PE in IIMV membranes were about 1-3 min. Therefore, the rapid [page 72↓]initial fluorescence decline represents the extraction of analogues by BSA from the outer leaflet and the second slower phase of fluorescence decrease reflects the transbilayer movement of the fluorescent analogues. The observed half-time of about five seconds for the initial outer leaflet extraction of analogues from IIMV was similar to those previously described for microsomes from rat liver cells (Marx, et al., 2000).

The half-times of movement of fluorescent PE and PC analogues to the periplasmic leaflet were about 3-4 times slower than those observed for the movement to the cytoplasmic layer. About 23% of the PE analogues and 18% of the PC analogues were located in the periplasmic leaflet of the vesicle membranes. However, it is unlikely that this asymmetric distribution of the analogues corresponds to the in vivo distribution across the inner membrane of E.coli cells. This is particularly true for the PE analogues. PE is not only the major phospholipid of the inner membrane, it also has to be conveyed to the outer membrane to populate the periplasmic leaflet of the outer membrane. Furthermore, PE is utilized for modifications on periplasmic oligosaccharides and protein modification (Huijbregts, et al., 1996;Sankaran and Wu, 1994). Huijbregts et al.speculated that the consumption of PE by intracellular processes for the above mentioned modifications or by integration of PE into the outer membrane provides a naturally sink for phospholipid transport processes in vivo (Huijbregts, et al., 1996). However, isolated IIMV from E.coli lack these processes and therefore, lack the need for substantial amounts of PE in the periplasmic leaflet, which could partially explain our findings. Possibly, the lack of such a sink resulted in the loss of an additional trigger for phospholipid synthesis and subsequently, in a difference of the protein mediated transmembrane equilibration of phospholipids in vivo and in vitro. Nevertheless, Huijbregts et al. showed that newly synthesized radioactive labeled PE preferentially redistributes to the periplasmic leaflet (65%) (Huijbregts, et al., 1998). The discrepancy to our findings might be related to the chemical structure of the used analogues, causing them to behave differently from their endogenous counterparts in in vitro systems. Therefore, a specific transbilayer distribution of phospholipids may occur, which was different in whole cells compared to isolated IIMV. Moreover, it is not known how the isolation procedure of IIMV affects the [page 73↓]transbilayer distribution of endogenous phospholipids. Additionally, the inverted structure of the isolated inner membrane vesicles and the relative abrasive isolation method possibly led to the loss or shielding of protein components in IIMV, which are important for the equilibration of phospholipids across the inner membrane in intact cells.

The transbilayer movement of short-chain, fluorescent PC (M-C6-NBD-PC) analogues was similar to that of the PE analogues. Interestingly, although PC is not present in the envelope of E.coli., we found that a fluorescent analogue of PC moved rapidly across the IIMV bilayer. This is consistent with previous findings that this process is not head-group specific (Bishop and Bell, 1985;Herrmann, et al., 1990;Huijbregts, et al., 1996).

The transbilayer distribution of the short-chain fluorescent PG analogues (M-C6-NBD-PG) was ~40% at the periplasmic leaflet and ~60% at the cytoplasmic leaflet of the vesicle membranes and therefore, close to a symmetrical distribution across the membrane of IIMV. The almost symmetrical transmembrane distribution indicates that PG has a different transverse membrane distribution in E.coli compared to PE. The transmembrane movement of the used PG analogues is also different from that of PS and PC analogues. While the half-times for outward movement of the PG analogues (~65 s) were similar to those found for fluorescent PE and PC analogues (~53 s and 40 s, respectively) the inward movement was faster as found for PE, PC and PS analogues. Additionally, inward movement of PG was 1.5 times slower than outward movement of PG compared to PC and PE analogues. This also indicates that PG phospholipid analogues redistribute different across the IIMV membranes, possibly facilitated by a different protein.

The rapid transmembrane movement of fluorescent analogues, which we found in this study, is in agreement with previously reported results (Huijbregts, et al., 1996). Using the dithionite assay, Huijbregts et al.(Huijbregts, et al., 1996) observed that the velocity of transmembrane movement of short-chain, fluorescent labeled phospholipid analogues across IIMV membranes at 37°C was in the same order of magnitude (about seven minutes). The same group reported half-times of redistribution of endogenously synthesized radioactive labeled PE analogues of about one minute in both IIMV and right-side out vesicles from E.coli [page 74↓] (Huijbregts, et al., 1998). However, the authors reported a conversion of the PG analogue to a fluorescent cardiolipin (CL) (Huijbregts, et al., 1996). The transmembrane movement of this CL analogue was found to be 2.4 times slower than that of the fluorescent short-chain PG analogue. Huijbregts et al. concluded that structural differences like head-group charges and fatty acid composition possibly led to different transbilayer movement (Huijbregts, et al., 1996). The data presented here do not support this hypothesis. As evident from TLC analysis of IIMV labeled with fluorescent short-chain PG, CL was not present in the IIMV membrane (data not shown), indicating that no conversion of PG to CL took place in the time course of experiment.

To investigate the transbilayer movement of fluorescent phospholipid analogues in more detail, we established a reconstitution assay. After detergent solubilization of IIMV, fluorescent phospholipid analogues (M-C6-NBD-PE or M-C6-NBD-PG) were added, and the detergent was removed resulting in equably labeled proteoliposomes. In parallel, we created liposomes without bacterial proteins but in the presence of detergent during reconstitution. While the rapid initial fluorescence decay was similar between protein containing and protein-free vesicles, only protein containing proteoliposomes showed a second slower phase of fluorescence decay similar to that of IIMV. So far, we were able to reconstitute the flip-flop without significant impairment.

Our investigations of the transbilayer movement and transverse distribution of fluorescent PE analogues across reconstituted proteoliposomes, confirmed the results that we obtained from BSA back-exchange kinetics with IIMV. Nevertheless, the reconstitution experiments revealed that the fluorescent PG analogues behaved different compared to the respective PE analogues. Almost nothing is known about the transbilayer distribution in bacterial membranes. As already outlined above, PG possibly has a different transbilayer distribution compared to PE in E.coli cells, and phospholipids in general are substrates for many metabolic enzymes. For example, in one study it has been demonstrated that pss was regulated by the presence of PG (Saha, et al., 1996). Using artificial transmembrane peptides, Kol et al.reported a coherence of the presence of PG and PE in order to regulate the transmembrane transport of phospholipids across [page 75↓]vesicle membranes (Kol, et al., 2003). Additionally, anionic lipids in the E.coli inner membrane were found to localize positively charged membrane protein segments to the cytoplasmic side of the membrane (van Klompenburg, et al., 1997). In turn, this mechanism possibly regulates the orientation of fluorescence PG analogues. Therefore, it is possible that PG has distinct lateral and transverse distribution at the inner membrane of E.coli, which is different from that of PE. Thus, transbilayer distribution of PG might be regulated by different (additional) protein dependent mechanisms. However, both phospholipid analogues underwent a rapid transbilayer movement across reconstituted proteoliposomes, which was protein dependent.

In reconstituted proteoliposomes as well as IIMV, we found a faster outward movement of analogues compared to the inward movement. Assuming that ATP-independent flippases work bi-directional, the rate constants in reconstituted vesicles should be similar for inward and outward movement. In particular, since the specific transbilayer orientation of membrane proteins in IIMV was not preserved in reconstituted proteoliposomes. The reason for the observed differences is unclear. A possible explanation for our findings is that the protein dependent flip-flop of phospholipids is facilitated by more than one protein, one of which was not accurately incorporated into the membrane or partially damaged. Supportive of this explanation are findings by Menon et al .. They observed that at least two different proteins were able to facilitate phospholipid transmembrane movement (Menon, et al., 2000). It also has to be taken into account that the experimental set-up, with large excess of BSA on the outside of the vesicles, somehow biased the results to give a faster outward movement. We can also not preclude that the amount of analogues in the outer leaflet of proteoliposomes was overestimated, since a small residual pool of fluorescent analogues (<10%) could not be extracted by BSA. Presumably, these non-extractable analogues stacked on the inner leaflet of vesicles lacking a rapid flip-flop activity.

To prove our findings of the BSA back-exchange assay, we performed a dithionite assay (4.4.2). Using this assay, the half-time of the outward movement of fluorescent labeled PE analogues in reconstituted proteoliposomes derived [page 76↓]from IIMV (~40 s) were very similar compared to those calculated from the BSA back exchange assay (~30 s). The calculated half-times of outward movement were slower than those found by BSA back-extraction due to slow dithionite penetration. The permeation of dithionite could not completely prevented. Thus, the half-times were slightly overestimated by our three-compartment model, since the slow penetration of dithionite affected the fluorescence reduction kinetics. However, the kinetic data from the reduction assay were coherent with the findings of the BSA back-extraction assay.


[page 77↓]

5.2  Effect of proteins on the transport of fluorescent phospholipid analogues

E.coli is a rapidly growing organism, dividing once every half hour. This requires rapid synthesis of new membrane, and consequently, a rapid transbilayer movement of newly synthesized phospholipids. For a bilayer membrane, expansion of one monolayer with respect to the other causes curvatures and extrusion, and eventually vesiculation, a process which can be easily rationalized in the frame of the bilayer couple model (Sheetz and Singer, 1974). Therefore, to preserve the stability of the inner membrane of E.coli, rapid redistribution of phospholipids is of importance. In vesicles composed only of lipids from the inner membrane of E.coli transbilayer movement of phospholipids was slow (data not shown) and as previously shown for pure lipid membranes (Kornberg and McConnell, 1971). Thus, proteins acting as a flippase have been assumed to mediate efficient phospholipid flip-flop. According to available data, flippase activities are typical for phospholipid synthesizing membranes of bacteria (Hrafnsdottir and Menon, 2000;Huijbregts, et al., 1996;Huijbregts, et al., 1998) and eukaryotic cells (Buton, et al., 1996;Herrmann, et al., 1990;Marx, et al., 2000;Menon, et al., 2000;Nicolson and Mayinger, 2000). Indeed, in the ER of eukaryotic cells such as rat liver cells (Marx, et al., 2000) as well as yeast cells (Marx, U. and Herrmann, A., unpublished observation) a rapid protein-dependent transbilayer movement has been unequivocally demonstrated. As it was shown previously by the stopped-flow approach, the half-times of the flip-flop of short-chain, fluorescent PC and PE in microsomes (Marx, et al., 2000) were in the same order as found here for IIMV and proteoliposomes.

To elucidate the role of proteins in the transbilayer movement of phospholipid analogues, the reconstitution assay was used. A number of vesicles with different protein to phospholipid ratios were created and the resulting proteoliposomes were studied by the stopped-flow BSA back-extraction assay. With increasing amounts of protein (up to 100µg/ml), the fluorescence decline became stronger (Figure 13). In proteoliposomes containing more than ~100µg/ml protein, we observed no substantial increase of flippase activity compared to proteoliposomes containing less than 100µg/ml protein. As evident from Figure 13 B and Figure 14, we found that vesicles with a high protein content (>100µg/ml) showed an activity maximum of the putative flippase. Thus, in a vesicle population prepared with ~100 µg/ml protein, all vesicles were equipped with a putative flippase. As deduced from the rate constants (see legend to Figure 13), the kinetics of transbilayer movement of fluorescent PE analogues in proteoliposomes containing proteins in the range 0-100 µg/ml (Figure 13 B, Figure 14) was essentially independent of protein content. This suggests that in this range of protein content the proteoliposome population included some vesicles with at least one flippase and some vesicles lacking transport mediating protein(s).

Experiments carried out with proteoliposomes containing different amounts of protein and probed with dithionite (Figure 16) confirmed our findings from the BSA back-exchange assay (Figure 13 and Figure 14): (i) the transmembrane movement of fluorescent PE analogues was strongly protein-dependent, and (ii) at a distinct protein quantity every vesicle must have contained at least one functional facilitator of transbilayer flip-flop of PE analogues. We observed that beyond the inflection point (~80 µg/ml) of the dose response plot shown in Figure 16, no enhancement in transport activity occurred. This inflection point was very similar to the inflection point (~100 µg/ml) we had determined by the BSA back-exchange assay.

The dose response plot in Figure 14 can be used to calculate the abundance of flippase among solubilized and reconstituted IIMV proteins. At the inflection point of the dose response plot with a protein to phospholipid ratio of ~35 µg/µmol (corresponds to 100µg/ml, see Figure 14), on average each proteoliposome is expect to contain a single flippase. For samples prepared at >35 µg/µmol, vesicles contain one or more flippases per vesicle. Using this figure and assuming that the mean molecular mass of IIMV membrane proteins is ~50kDa, each vesicle prepared at 35 µg/µmol contains 300 protein molecules on average. This implies that the abundance of functional flippases is about 1 to 300 or 0.33% by weight of the reconstituted IIMV membrane proteins. This estimate is similar to the previously described 0.2% (Menon, et al., 2000) and 0.15% (Gummadi and Menon, 2002), which were deduced from experiments with rat [page 79↓]liver ER membrane proteins, as well as the estimate of 0.6% obtained in an earlier study on rat liver microsome membranes (Backer and Dawidowicz, 1987).

Further supportive of the protein dependence of the fast transbilayer movement of fluorescent analogues were our findings regarding proteinase K treatment of proteoliposomes. We observed that the transport activity in proteoliposomes was eliminated after 60 min of proteinase K treatment. Proteinase K hydrolyzes peptide bonds exposed to the aqueous milieu and leaves the transmembrane segments untouched. Since the rapid flip-flop of phospholipid analogues was almost eliminated, it is likely that the putative flippase(s) must contain ectodomains, which are important for the phospholipid transmembrane redistribution. Moreover, our data suggest that transmembrane segments alone are not sufficient for the facilitated flip-flop of phospholipids across the vesicle membrane (for further discussion see 5.4).


[page 80↓]

5.3  Ion exchange chromatography - Attempts to enrich flippase activity

In this thesis, we showed that proteins were involved in rapid flip-flop of fluorescent phospholipid analogues across IIMV derived membranes (4.3; 4.4). The question whether specific protein(s) were responsible for the rapid transbilayer movement of phospholipids or the rapid flip-flop is a consequence of the presence of membrane proteins remained unclear.

To investigate this question in more detail, we separated transmembrane proteins by their distinct properties, in particular, by their charges. We utilized anion exchange chromatography (AEC) to separate distinct fractions from a total IIMV protein extract of E.coli. After AEC, protein mixtures were yielded, which were less complex compared to the total protein content. After separation, fractionated proteins were reconstituted into proteoliposomes and we analyzed the ability of selected protein fraction to mediate rapid flip-flop of phospholipids across the membranes. If specific proteins are required for the rapid phospholipid flip-flop, it should be possible to enrich this transport activity by protein fractionation.

The transbilayer movement of fluorescent PE analogues across reconstituted proteoliposome membranes derived from these fractions was investigated. A prerequisite for this analysis is the generation of proteoliposomes with approximately similar protein content to allow comparison of flippase activities between different fraction samples (see chapter 4.4).

We applied TE from IIMV to a strong anion exchange resin (Q-Sepharose). The resulting flow-through (QF) and eluate (QE) were reconstituted into proteoliposomes and analyzed by the dithionite assay ( 3.6 ). Both reconstituted protein fractions, pQF and pQE showed flippase activity compared to control liposomes. However, as revealed from the fluorescence traces, pQF and pQE exhibited only minimum differences in fluorescence reduction. This indicates that AEC was not efficient to enrich the putative flippase. It is also possible that more than one putative flippase exists. If the distinct flippases contain different biochemical and structural properties, they could be hardly separated by AEC under the conditions used in our approach.


[page 81↓]

However, if only one type of a putative flippase exists, the experimental conditions we used were possibly not sufficient to separate this protein in one fraction. To exclude that the protein quantity impaired our assay, we carried out experiments with higher protein concentrations in the proteoliposomes (Figure 21) but in the linear range of the dose-response plot (see Figure 13 and Figure 14). We observed no differences in specific activities compared to lower protein content reconstituted into proteoliposomes. The phospholipid flippase activity of reconstituted fractions was lower compared to control proteoliposomes prepared from the total TE of IIMV at similar protein concentration (Figure 20 and Figure 21).

To improve the efficiency of protein fractionation, several conditions were tested. For example, successive elution, i.e. bound proteins were discontinuously eluted with buffers of increasing ionic strength. This resulted in only slightly enriched phospholipid flippase (specific) activity in pQF (Figure 22 and Table 6). Apparently, the stepwise elution diluted the transport activity originally found in the eluate QE into two fractions (QE4 and QE). Consequently, since the fluorescence kinetics and calculations on the specific activities revealed no enrichment of flippase activity in a specific anion exchange column fraction, we found no indications whether specific flippases were responsible for flip-flop of fluorescent PE. Nevertheless, nothing is known about the structure of the putative bacterial flippase. It can be a simple protein spanning the membrane only once or a complex protein with several transmembrane domains and/or various subunits. Hence, it cannot be ruled out that subunits or cofactors necessary for the efficient rapid phospholipid flip-flop were inactivated and/or dispersed into different fractions by the strong anion exchange chromatography procedure, but were present in proteoliposomes generated from total TE of IIMV.

Hrafnsdottir and Menon showed that chromatographic fractionation of a detergent extract from Bacillus subtilis cell membranes resulted in populations of vesicles showing different specific activities (Hrafnsdottir and Menon, 2000). Furthermore, they found by glycerol gradient analysis an enriched transport activity in a distinct fraction, which sedimented at ~4S, correlating with the presence of specific proteins involved in transmembrane movement of phospholipids (Hrafnsdottir and Menon, 2000).


[page 82↓]

Partial purification and characterization of putative flippase proteins in the ER of rat liver microsomes were reported by Gummadi et al. and Menon et al.(Gummadi and Menon, 2002;Menon, et al., 2000). Menon et al. clearly demonstrated that the transport of a synthetic PC across the ER membrane was facilitated by specific proteins. After glycerol gradient fractionation and IEC they presented data, consistent with the idea that two transporter are capable of affecting the bi-directional translocation of phospholipids (Menon, et al., 2000). Moreover, Gummadi and Menon were able to enrich the ER flippase activity up to 15-fold, after TE samples from rat liver microsomes, which were passed sequentially over an weak anion and over a weak cation exchange resin (Gummadi and Menon, 2002). This implies that, very likely, specific proteins are involved in the rapid, energy independent transbilayer transport of phospholipids. However, our data indicates that different strategies are needed to purify flippase proteins from E.coli inner membranes.


[page 83↓]

5.4  Are specific proteins required for phospholipid flip-flop?

The molecular nature of the flippase remains to be determined. We showed that the phospholipid transbilayer movement was independent of the head-group of phospholipids, as tested with fluorescent analogues of PE, PG, PS and PC. This was also observed in the ER membrane of rat liver cells (Buton, et al., 1996;Herrmann, et al., 1990;Marx, et al., 2000).

It is evident that these proteins must possess domains that enable a rapid redistribution of phospholipids between the two leaflets of a membrane bilayer. The model of Kol et al. postulates that the accumulated weak effects of many transmembrane helices were sufficient to allow phospholipid flip-flop in biogenic membranes (Kol, et al., 2001;Kol, et al., 2003). They found that phospholipid flop mediated by transmembrane peptides in model membranes, was modulated by the lipid composition (Kol, et al., 2003). Furthermore, they observed that the transmembrane movement of phospholipids was head-group dependent and argued this was due to the distinct charge pattern of the lipid species. It is unlikely that transmembrane domains per se are sufficient to mediate a fast flip-flop. Otherwise, a fast transbilayer movement of phospholipids would be a typical feature of all biological membranes. This is not the case. Our findings substantiated observations that appreciable levels of phospholipid flip-flop occurred in biogenic membranes. Furthermore, protease treatment led to a striking decrease of phospholipid transmembrane movement across reconstituted IIMV derived proteoliposome membranes. Since, proteinase K treatment leaves transmembrane domains intact, this indicates that transmembrane domains are not the only determinant of transbilayer movement of phospholipids.

Assuming that rapid phospholipid flip-flop is a general, essential property of biogenic membranes and that each of those membranes has a characteristic lipid assembly depending on the organism, it is unlikely that newly synthesized (phospho)lipids were transported due to their charges. A more reasonable explanation is that specific protein(s) facilitate transbilayer movement, possibly, under control of specific co-factors like negatively charged (e.g. PG) or non-bilayer forming lipids (e.g. PE). For example, as mentioned above, such a regulatory mechanism is known for the pss, a major enzyme in the PE synthesis in [page 84↓] E.coli: This enzyme is up-regulated by the presence of PG (Saha, et al., 1996), leading to a specific transverse and/or lateral membrane distribution. Kol et al. found that flop of short-chain fluorescent labeled PG was inhibited by increasing PE concentrations and stimulated by increasing the fraction of PG (Kol, et al., 2003). The hypothesis made by Kol et al. that this could be an autoregulatory mechanism (Kol, et al., 2003) is supported by our observations, that PG exhibited a different transmembrane distribution in reconstituted vesicles compared to PE.

It is obvious that transmembrane distribution is a very important process and needs to be regulated. In consequence, it is very likely that a protein mediated tuning of lipid distribution is necessary for every cellular membrane in particular for biogenic membranes with regards to the specific function of the individual membranes.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 3.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
02.08.2004