Phase Diagram of Purified CNS Myelin Reveals Continuous Transformation between Expanded and Compacted Lamellar States

Purified myelin membranes (PMMs) are the starting material for biochemical studies, from individual components up to the isolation of detergent-resistant membrane (DRM) fractions or detergent-insoluble glycosphingolipid (DIG) fractions, which are commonly believed to resemble physiological lipid rafts. The normal DIG isolation protocol involves the extraction of lipids under moderate cooling. The isolation of PMMs also involves the cooling of myelin as well as exposure to low ionic strength (IS). Here, we addressed the combined influence of cooling and IS on the structure of PMMs. The phase behaviour was investigated by small angle X-ray diffraction. Analysis of the diffraction peaks revealed the lamellar periodicity (d), the number of periodically correlated bilayers (N), and the relatives fractions of each phase. Departure from physiological conditions induced a phase separation in myelin. The effect of monovalent and divalent ions was also compared at equivalent IS, showing a differential effect, and phase diagrams for both ion types were established—Ca2+ induced the well-known over-compacted phase, but additionally we also found an expanded phase at low IS. Na+ promoted phase separation, and also induced over-compaction at sufficiently high IS. Finally, exploring the whole phase diagram, we found evidence for the direct isothermal transformation from the expanded to the compacted phase, suggesting that both phases could in fact originate from the identical primary lateral phase separation, whereas the apparent difference lies in the inter-bilayer interaction that is modulated by the ionic milieu.


Introduction
The various physical states of myelin have been described in the past, mainly based on the lamellar period (or "spacing") d as obtained by X-ray diffraction on the peripheral nervous system (PNS). Apart from the native period, myelin can also exhibit a so-called "expanded" and a so-called "(over)compacted" phase [1,2]. In our previous studies, we described the phase separation of purified myelin membranes (PMMs) induced by cooling [3] and its relationship to detergent-insoluble glycosphingolipid (DIGs) [4][5][6][7]. In the present work, we performed a systematic study on the influence of the ionic strength (IS) on PMMs and established temperature/IS phase diagrams revealing the identity of the phases involved in the phase separation. Previously, we have shown that the phase separation of PMMs from the central nervous system (CNS) leads to a coexistence of the native period (d = 7.4-8.0 nm) with another new segregated phase. This new phase can assume two different

Small-Angle X-Ray Diffraction in Suspension
Small angle X-ray diffraction was used to determine the lamellar period (or "spacing") d of the bilayers. Measurements were performed at the DO2A/SAXS2 beamline of the Laboratorio Nacional de Luz Sincrotron LNLS (Campinas, Brazil), with a beam energy of 8.3 keV corresponding to a wavelength λ = 1.488 Å. Lyophilized PMMs were suspended at 10-12 mg/mL and warmed up to 45 • C to ensure hydration. After three thawing and cooling cycles from 4 to 40 • C, the samples were stored at 4 • C until the measurements. Thus far, we have not observed any influence of the sample history on the behaviour of purified myelin.
A series of NaCl and CaCl 2 solutions with equivalent IS was prepared. The IS was calculated from Equation (1): where c i,0 is the concentration of ion species i at an infinite distance from the central charge, and z i is the valency of the respective ion, for instance z = +1 for Na + , z = +2 for Ca 2+ , and z = −1 for Cl − [10]. The selected IS were 0, 9, 36, 75, 150, 300, 600, and 1200 mM. In addition, three NaCl solutions (375, 450, and 525 mM) in the range relevant for the continuous phase transformation were used. The phase behaviour of PMMs was investigated for various temperatures (4-48 • C) below the myelin protein denaturation temperature. PMMs were injected into a liquid sample cell between two mica plates. The sample-to-detector distance was 1 m, calibrated with silver behenate. The exposure time for each experiment was 5 min. For radial integration of the Debye-Scherrer rings on the 2D detector (MARCCD 165), we employed the free software Fit2D V12.077 from Andy Hammersley at the European Synchrotron Radiation Facility [20].
The intensities are presented as a function of the magnitude of the scattering vector q (Equation (2)): where 2θ is the scattering angle with respect to the incident beam. The corresponding lamellar periodicities d then follow from the Bragg equation as Equation (3): where h = 1, 2, is the peak order. The number N of periodically correlated bilayers under various conditions was obtained by applying the Scherrer equation [21] to the Bragg peak full width at half maximum (w) of a Lorentzian curve (Equation (4)): where r is the instrumental resolution (here: r ≈ 0.04 nm −1 ). The diffracting power of native phase (P nat ) was defined by using the Patterson function [22] (Equation (5)): where hI nat (h)/d is the Lorentz-corrected intensity of the reflection of order h. The integrated I nat (h) was estimated from the areas under the associated Bragg peaks. The fraction of myelin in the native Cells 2020, 9, 670 4 of 12 phase is expressed as the relative diffracting power (RDP nat ), which is the ratio of the diffracting power P nat to the initial amount (P nat + P non−nat ) (Equation (6)): RDP nat = P nat P nat + P non−nat (6)

Results
According to the diffraction measurements, only lamellar phases were observed, as evidenced from the integer relationship between the positions of the peak maxima (some examples are shown in Figure 1) [10,23]. The period d of the lamellar phases were calculated by Equation (3). . In (A), the major Bragg peaks of myelin (at 0.8 and 1.6 nm −1 ) were the only peaks present at 37-48 °C (surrounded in a vertical rectangle), but under cooling a new phase appeared with a peak at lower (arrow pointing to the left), which coincided with higher -an expanded phase. In (B), the native peak appeared again (inside the rectangle), and additionally a new Bragg peak at higher appeared (≈1 nm −1 , arrow pointing to the right)-a compacted phase. The second order peaks (ℎ = 2) associated with the non-native phases are indicated with vertical arrows. Figure 2 shows the lamellar periods in PMMs as a function of the IS for various temperatures in NaCl-and CaCl2-based solutions. The native period (black symbols) is present at all temperatures in NaCl-based solution. At 26 °C, the obtained results are consistent with what is reported in the literature for room temperature conditions [13]. At 15 and 4 °C (DIG isolation temperature) the behaviour of purified CNS myelin is nevertheless very different. Upon addition of NaCl (from 0 to 0.3 M), first an increasingly expanded phase was observed (red symbols), which contracted in the 0.375-0.45 mM range and then (at NaCl > 0.5 M) collapsed to value characteristic for the (over)compacted phase, which was only marginally sensitive to a further increase in the NaCl concentration. Relatively low IS and temperature led to an expanded phase. Up to 3 mM CaCl2 (9 mM IS) led to a slightly expanded phase ( ≈ 8.3 nm) at low T (blue symbols), undistinguishable from the phase expanded by NaCl at the same IS. Nevertheless, as the IS increased, the behaviour of myelin in NaCl and CaCl2 solutions rapidly diverged. Previous works [12,25] had already shown that CaCl2 induces a cooperative compaction above 12 mM (36 mM IS, see Equation (1)) in nerve myelin, and a similar effect is seen here on PMMs. At extreme conditions, (high CaCl2 and low temperature) the native phase completely disappeared. On the other hand, in the presence of NaCl at 4 °C, the expanded phase persisted up to much higher salt concentrations (up to 450 mM IS), . In (A), the major Bragg peaks of myelin (at 0.8 and 1.6 nm −1 ) were the only peaks present at 37-48 • C (surrounded in a vertical rectangle), but under cooling a new phase appeared with a peak at lower q (arrow pointing to the left), which coincided with higher d-an expanded phase. In (B), the native peak appeared again (inside the rectangle), and additionally a new Bragg peak at higher q appeared (≈1 nm −1 , arrow pointing to the right)-a compacted phase. The second order peaks (h = 2) associated with the non-native phases are indicated with vertical arrows.

Spacings of the Different PMM Phases
Under physiological conditions (NaCl 150 mM/150 mM IS) PMMs normally show a series of major diffraction peaks (enclosed in vertical boxes in Figure 1) located at ≈0.8 and ≈1.6 nm −1 [24]; they correspond to a membrane period of d = 7.7 ± 0.2 nm. Consequently, these are the only peaks in Figure 1A at 37-48 • C. It is seen that the native period (vertical box) was present in almost all temperature-IS conditions.
Upon cooling to 15-4 • C (DIGs isolation temperature), the native peak split into two peaks, the new one shifted to lower q as the temperature decreased. This shift to lower q reflected the expansion of the lamellar period of the membranes. Figure 1B (CaCl 2 50 mM/150 mM IS) shows a permanent split of the peak into the native period and a non-native period, but the peak position of the latter shifted in the opposite direction (as compared to Figure 1A), to a higher q and consequently revealed a compaction of the membrane stack. Figure 1A,B are at equal IS and a differential effect of ions is evident. As a conclusion, Figure 1 shows two examples of measurements taken each at different temperatures for two particular cases: one for myelin experiencing expansion under cooling in physiological conditions ( Figure 1A) and other one experiencing (over)compaction ( Figure 1B) with high CaCl 2 concentration. The new major peaks can appear at low q (q < 0.8 nm −1 ), in which case the phase is called separated, expanded, or partially swelled, with periods ranging from 8.5 to 11 nm or more. Alternatively, the new peak can appear at q > 0.8 nm −1 , more precisely around 1.0 nm −1 , in which case the phase is called (over)compacted and this is clearly a lipid-enriched phase [25], with only d = 6.6 ± 0.4 nm, depleted of proteins [2]. These different non-native phases are well known and observations of them are found overall in the myelin literature [2,3].  [13]. At 15 and 4 • C (DIG isolation temperature) the behaviour of purified CNS myelin is nevertheless very different. Upon addition of NaCl (from 0 to 0.3 M), first an increasingly expanded phase was observed (red symbols), which contracted in the 0.375-0.45 mM range and then (at NaCl > 0.5 M) collapsed to d value characteristic for the (over)compacted phase, which was only marginally sensitive to a further increase in the NaCl concentration. Relatively low IS and temperature led to an expanded phase. Up to 3 mM CaCl 2 (9 mM IS) led to a slightly expanded phase (d ≈ 8.3 nm) at low T (blue symbols), undistinguishable from the phase expanded by NaCl at the same IS. Nevertheless, as the IS increased, the behaviour of myelin in NaCl and CaCl 2 solutions rapidly diverged. Previous works [12,25] had already shown that CaCl 2 induces a cooperative compaction above 12 mM (36 mM IS, see Equation (1)) in nerve myelin, and a similar effect is seen here on PMMs. At extreme conditions, (high CaCl 2 and low temperature) the native phase completely disappeared. On the other hand, in the presence of NaCl at 4 • C, the expanded phase persisted up to much higher salt concentrations (up to 450 mM IS), which indicated a specific role of Ca 2+ in the stabilization of the (over)compacted phase [25]. In other words, both salt types were able to promote the expanded phase as well as the (over)compacted phase, but at very different concentration thresholds. It should be noted that this difference between Na + and Ca 2+ persisted also when the IS was considered instead of the concentration. In other words, the effect must be considered ion-specific or at least dependent of the cation valence. As a general conclusion, the states of myelin under extreme conditions (i.e., at the lowest and highest salt concentrations) were qualitatively the same for CaCl 2 and NaCl. However, at intermediate (near-physiological) salt concentrations, the CaCl 2 shifted the balance more towards (over)compaction, whereas NaCl shifted it more towards expansion. Figure 3 shows the lamellar period of the PMMs' non-native phase as a function of temperature and IS for solutions based on NaCl ( Figure 3A) and on CaCl 2 ( Figure 3B). The white region designates conditions under which only the native phase is present. As can be seen, the expanded and (over)compacted phases were both present in CaCl 2 as well NaCl solutions but at different IS values. The most striking feature of the phase diagram is the continuous transformation of the non-native period from expanded to compacted, as indicated with vertical arrows in both panels. The transition can be induced isothermally, by variation of the NaCl. On the other hand, such a direct transformation from the expanded into the (over)compacted state without passing through a native phase intermediary cannot be observed in thermal scans at constant salt concentration (which would correspond to a horizontal path in the phase diagrams). This is likely the reason why the direct connection between the two non-native phases had not been identified in the classical literature [2]. The transition can be induced isothermally, by variation of the NaCl. On the other hand, such a direct transformation from the expanded into the (over)compacted state without passing through a native phase intermediary cannot be observed in thermal scans at constant salt concentration (which would correspond to a horizontal path in the phase diagrams). This is likely the reason why the direct connection between the two non-native phases had not been identified in the classical literature [2].  Figure 4 shows the relative amount for each phase as estimated from the relative diffracting power ( ) of each phase (see Equations (5) and (6)) for PMMs in aqueous salt solutions based on NaCl and CaCl2. Cooling shifted the equilibrium to the non-native phase for both salt types, as seen from the lower of the native phase. On the other hand, in NaCl solutions, heating promoted the increment of the native phase up to a point at which = 1.0, except for an extreme condition of 1.2 M IS where some (over)compacted phase persisted. In contrast to NaCl, CaCl2 quickly shifted the equilibrium in the direction of non-native phases, mainly towards the (over)compacted phase.  Figure 4 shows the relative amount for each phase as estimated from the relative diffracting power (RDP nat ) of each phase (see Equations (5) and (6)) for PMMs in aqueous salt solutions based on NaCl and CaCl 2 . Cooling shifted the equilibrium to the non-native phase for both salt types, as seen from the lower RDP nat of the native phase. On the other hand, in NaCl solutions, heating promoted the increment of the native phase up to a point at which RDP nat = 1.0, except for an extreme condition of 1.2 M IS where some (over)compacted phase persisted. In contrast to NaCl, CaCl 2 quickly shifted the equilibrium in the direction of non-native phases, mainly towards the (over)compacted phase.

Number ( ) of Periodically Correlated Bilayers
An interesting point apart from the relative amount of each phase is how well they are periodically arranged into stacks of membranes. The number of periodically correlated bilayers ( ) can be obtained from the width of the diffraction peaks (see Equation (4)). Figure 5 shows that the native period myelin was highly variable and grew up to a limiting value of 12-20 lamellae in the direction of heating as the non-native phases disappeared. Irrespective of the salt type, N decreased with increasing IS. It is known that CNS myelin persists in a packed state, even in pure water (IS = 0), which is in contrast to PNS myelin that swells and loses its lamellar correlations [13]. For the non-native periods, the data grouped into two distinct regimes that were not very responsive to temperature (encircled areas in Figure 5C,D). The values for the expanded period are about ≈ 9 ± 2 correlated bilayers, which agrees with the well-known transverse

Number (N) of Periodically Correlated Bilayers
An interesting point apart from the relative amount of each phase is how well they are periodically arranged into stacks of membranes. The number of periodically correlated bilayers (N) can be obtained from the width of the diffraction peaks (see Equation (4)). Figure 5 shows that the native period myelin was highly variable and grew up to a limiting value of 12-20 lamellae in the direction of heating as the non-native phases disappeared. Irrespective of the Cells 2020, 9, 670 8 of 12 salt type, N decreased with increasing IS. It is known that CNS myelin persists in a packed state, even in pure water (IS = 0), which is in contrast to PNS myelin that swells and loses its lamellar correlations [13]. For the non-native periods, the data grouped into two distinct regimes that were not very responsive to temperature (encircled areas in Figure 5C,D). The values for the expanded period are about N ≈ 9 ± 2 correlated bilayers, which agrees with the well-known transverse stacking of membrane domains in myelin membranes [13]. The (over)compacted phase is more highly correlated, with N = 10-20, as was reported earlier [25]. These observations led us to conclude that the non-native phases, no matter whether they are induced by NaCl or CaCl 2 , are largely non-responsive to environmental conditions once they are nucleated (encircled in Figure 5C,D), although the (over)compacted phase induced by CaCl 2 appeared to be somewhat more strongly correlated.

Non-Monotonic Dependence of Non-Native Phase Lamellar Period
An important finding in this work was the direct and continuous transition between the expanded and compacted phases of PMMs, which was salt-induced and best observed at 4 °C. Figure 6 shows an isothermal cut at 4 °C for different NaCl concentrations (same data as Figure 2A at 4 °C) in a double-logarithmic representation. As can be seen from the distinct linear regimes in this representation, different power laws seemed to characterize the initial expansion (at low NaCl concentrations), the compaction (at intermediate NaCl concentrations), and finally a plateau at high NaCl concentrations, when no further compaction was possible due to short-range repulsive interactions. In contrast to this complex behaviour, the native period of myelin was rather insensitive to the salt concentration because its period was dictated by the protein components [2]. This is the very first report about this behaviour in myelin. Previous studies have revealed a swollen or expanded period as well a native period at room temperature and 0.1-0.2 NaCl IS, consistent with what was found at 26 °C [13]. We expanded such a work in a systematic way and showed that the expanded myelin period contracted under increased IS. At 15 and 4 °C, the expanded phase further expanded up to concentrations of 0.3 M NaCl. This region of the phase diagram very clearly showed continuity or direct transformation between the expanded and compacted phase.

Non-Monotonic Dependence of Non-Native Phase Lamellar Period
An important finding in this work was the direct and continuous transition between the expanded and compacted phases of PMMs, which was salt-induced and best observed at 4 • C. Figure 6 shows an isothermal cut at 4 • C for different NaCl concentrations (same data as Figure 2A at 4 • C) in a double-logarithmic representation. As can be seen from the distinct linear regimes in this representation, different power laws seemed to characterize the initial expansion (at low NaCl concentrations), the compaction (at intermediate NaCl concentrations), and finally a plateau at high NaCl concentrations, when no further compaction was possible due to short-range repulsive interactions. In contrast to this complex behaviour, the native period of myelin was rather insensitive to the salt concentration because its period was dictated by the protein components [2].

Discussion
A major result of this work was the observation of a direct continuous transition between the two non-native phases of PMMs without passing through the native phase. The transition was continuous in the sense that intermediate values between the (over)compacted and the expanded spacings could be found (Figure 2). The same applied to the evolution of the RDP shown in Figure 4. Regarding these two observables, the transition clearly was not an "all or nothing" process. However, regarding the number of correlated membranes shown in Figure 5, the picture was somewhat different, because the transition there rather appeared as a step-like change, as was previously shown for the transition from native to (over)compacted myelin in the old literature [25]. This transition did not occur at physiological or higher temperature, but was relevant for DIG extraction at the low temperature. Importantly, the continuous transition was controlled by the IS along an isothermal and could not be induced by temperature variation and constant IS. It is also interesting to note that close to 300 mM [NaCl] (maximum in Figure 6), there was a change in behaviour from expanding to compacting regime as a function of [NaCl]. This was evident at 4-15 °C, but an increase of temperature towards the physiological range weakened this effect (Figure 2 at 26 °C). Further increase in the temperature eventually transformed the non-native phases into the native phase.
Another remarkable fact is that during the transformation from expanded to compacted spacing at 4-15 °C, the of the native phase ( Figure 4) did not cross through a minimum but it displayed a monotonic step-like behaviour. This meant that the expanded phase did not transform first into the native (otherwise should have been 1) and then into the (over)compacted phase, but rather it transformed directly from the expanded into the (over)compacted phase. A similar monotonic behaviour in line with this picture could also be observed with the number of correlated bilayers of the native state.
The presence of salt typically leads to a screening of electrostatic repulsion when the membrane surfaces carry a non-negligible surface charge density. In this case, one expects a monotonic decrease of the lamellar period with increasing salt concentration. However, favourable interactions of certain ions with the surfaces can result in an effectively enhanced charge density and, in turn, stronger repulsion. The strength of this effect in general depends on the ion type and on the surface chemistry and can strongly deviate from the usual Hofmeister series [26]. For example, an increase in the lamellar period of neutral phosphatidylcholine (PC) lipid membranes by the addition of salt is well documented and has been studied best for calcium ions, which exhibit preferential interactions with PC [27][28][29][30][31]. Increasing the CaCl2 concentration not only increases the surface charge density but also the screening effect [31], so that also the repulsion is reduced. In this case, a non-monotonic behaviour, as in our data in Figure 6, is expected because the surface charge density will saturate at some point, whereas the screening strength further increases as the square root of the concentration. This is the very first report about this behaviour in myelin. Previous studies have revealed a swollen or expanded period as well a native period at room temperature and 0.1-0.2 NaCl IS, consistent with what was found at 26 • C [13]. We expanded such a work in a systematic way and showed that the expanded myelin period contracted under increased IS. At 15 and 4 • C, the expanded phase further expanded up to concentrations of 0.3 M NaCl. This region of the phase diagram very clearly showed continuity or direct transformation between the expanded and compacted phase.

Discussion
A major result of this work was the observation of a direct continuous transition between the two non-native phases of PMMs without passing through the native phase. The transition was continuous in the sense that intermediate values between the (over)compacted and the expanded spacings could be found (Figure 2). The same applied to the evolution of the RDP shown in Figure 4. Regarding these two observables, the transition clearly was not an "all or nothing" process. However, regarding the number of correlated membranes shown in Figure 5, the picture was somewhat different, because the transition there rather appeared as a step-like change, as was previously shown for the transition from native to (over)compacted myelin in the old literature [25]. This transition did not occur at physiological or higher temperature, but was relevant for DIG extraction at the low temperature. Importantly, the continuous transition was controlled by the IS along an isothermal and could not be induced by temperature variation and constant IS. It is also interesting to note that close to 300 mM [NaCl] (maximum in Figure 6), there was a change in behaviour from expanding to compacting regime as a function of [NaCl]. This was evident at 4-15 • C, but an increase of temperature towards the physiological range weakened this effect (Figure 2 at 26 • C). Further increase in the temperature eventually transformed the non-native phases into the native phase.
Another remarkable fact is that during the transformation from expanded to compacted spacing at 4-15 • C, the RDP nat of the native phase ( Figure 4) did not cross through a minimum but it displayed a monotonic step-like behaviour. This meant that the expanded phase did not transform first into the native (otherwise RDP nat should have been 1) and then into the (over)compacted phase, but rather it transformed directly from the expanded into the (over)compacted phase. A similar monotonic behaviour in line with this picture could also be observed with the number of correlated bilayers of the native state.
The presence of salt typically leads to a screening of electrostatic repulsion when the membrane surfaces carry a non-negligible surface charge density. In this case, one expects a monotonic decrease of the lamellar period with increasing salt concentration. However, favourable interactions of certain ions with the surfaces can result in an effectively enhanced charge density and, in turn, stronger repulsion. The strength of this effect in general depends on the ion type and on the surface chemistry and can strongly deviate from the usual Hofmeister series [26]. For example, an increase in the lamellar period of neutral phosphatidylcholine (PC) lipid membranes by the addition of salt is well documented and has been studied best for calcium ions, which exhibit preferential interactions with PC [27][28][29][30][31]. Increasing the CaCl 2 concentration not only increases the surface charge density but also the screening effect [31], so that also the repulsion is reduced. In this case, a non-monotonic behaviour, as in our data in Figure 6, is expected because the surface charge density will saturate at some point, whereas the screening strength further increases as the square root of the concentration. The concentration at the maximum repulsion as well as the precise repulsion strength depend on the magnitude of the ion affinity and details of the interfacial force balance. In this light, and considering additional factors such as a potential influence of the ions on the area per lipid and on the bending rigidity, we refrained from modelling the experimental data on a quantitative level. Regarding the affinity of ions for the membrane surfaces, we note that neither Na + nor Cl − had significant affinities for commonly studied PC lipids. Myelin membranes, however, display various chemical motifs including glycolipids, which can exhibit preferential interactions with monovalent and divalent ions, as was recently shown [32].
As just discussed, the crossover from expansion to compaction upon IS variation (at ≈300 mM) at low temperature points leads to a near-critical state of myelin under physiological conditions regarding inter-membrane interactions. The change in the interlamellar water amount upon variation of the lamellar period by ∆d can be roughly estimated when assuming a constant membrane thickness. In this case, ∆d directly translated into a change in the equivalent number ∆n l = ∆d/l w of "water layers", where l w = (v w ) 1/3 = 0.31 nm is the linear dimension of a water molecule of partial molecular volume, v w = 0.030 nm 3 . This procedure yielded ∆n l = 10 between the lamellae in the more extreme calculation; this was considering full expansion in NaCl at 4 • C and taking the most compacted state (CaCl 2 , at the same temperature). More conservative calculations (taking NaCl also for the most (over)compacted state) rendered ∆n l = 8 for the highest hydration.
In the present work, we showed that in CNS these non-native phases related to DIGs [3] can in fact be interconverted directly without passing through the native state. This result suggests that the expanded and compacted phases are based on the same lateral phase separation that can be observed also in Langmuir monolayers [33]. This phase separation is mainly due to the insolubility of many transmembrane proteins on cholesterol enriched phases, which is further increased upon cooling, and is related to the cholesterol-enriched phase present in PMM Langmuir monolayers [33]. In multilayers, the inter-membrane interactions then additionally induce changes in the lamellar period according to the temperature and IS conditions (see vertical arrows in Figure 3). It is well known that the compacted phase is a lipid-enriched, protein-free lateral portion of the membrane; this phase can be so compact only because of the absence of proteins that dictate the period of native myelin [2]. The same lipid-enriched phase has already been observed in Langmuir monolayers [34,35]. It might thus be that the expanded phase represents the lipid-enriched phase, but in an environment that favours repulsion. In fact, it has also been shown that protein-depleted phases of myelin can exhibit more expanded periods under certain conditions [12]. In the literature, it was suggested that both expanded and compacted phases share the same electron density profile, at least at low resolution [36]. Furthermore, our group has previously shown that the DIG fraction from CNS PMMs can be compacted (at 20 mM CaCl 2 ) or expanded (at 150 mM NaCl), mimicking the non-native phase of whole myelin under the corresponding environmental conditions [3].
This combined evidence lets us postulate that a single DIGs phase is on the basis of both the expanded phase and the compacted phase.

Conclusions
Our main result is that the compacted and expanded phases are closely related to each other. In fact, a continuous transformation from one phase to the other can be clearly observed at low temperature in the phase diagram. That is, this transformation took place without any detectable discontinuity in spacing and directly from the expanded to the compacted phase without passing through the native phase. This isothermal transformation followed a power law in the inter-membrane distance as a function of the IS. Thus, one diffraction peak was continuously shifted into the peak position classically assigned to the other phase. The sensitivity of the inter-membrane interaction to the ionic environment produced differential stacking of the non-native phase(s), and this fact gave origin to the differential nomenclature commonly used for the non-native phase(s) of myelin adopted when this interconnection was unknown. This shift can only be observed below room temperature in combined phase diagrams such as the ones presented here.