In good agreement with the literature, the present set of experiments confirms that the LLP domains present in the TM portion of the Env protein of HIV-1 form amphipathic α-helical structures in the presence of a 10% POPG: 90% POPC lipid environment. Each of these peptides were able to bind to and disrupt membranes of this composition, despite a lack of amino acid sequence identity. The presence of an NBD tag did not appear to affect the biochemical characteristics of the peptides to which it was attached. These experiments represent the first direct comparison of all three LLP domains' interactions from the same HIV-1 virus – HXB2 from clade B – with identical lipid membranes. Additionally, it is the first example of a direct comparison of structure and function of an entire LLP-1 domain from the laboratory adapted HXB2 strain of HIV-1 (i.e., LLP-1) with a natural sequence variant from a clade D HIV-1 virus (i.e., LLP-1D) under identical conditions.
Based on similarities to other amphipathic α-helices, such as magainin-2, it has been hypothesized that the LLP domains could insert into bilayers and form a pore with their hydrophobic faces oriented towards the lipid bilayer and the hydrophilic faces oriented towards the lumen of the newly formed pore [8
]. The results presented here are consistent with this hypothesis. LLP-1, -2, and -3 domains partition into membranes as an α-helix and disrupt the membrane. The methodologies used here are not able to distinguish between membrane insertion and interactions at the membrane-water interface in which the peptides lie on the cell surface to cause a generalized disruption of the membrane. However, our observation of nearly complete leakage from vesicles at P:L ratios exceeding 1:1000 for LLP-1 and -2 supports the idea of a membrane-spanning pore. Such high activity has not been observed for surface-active membrane-spanning pore. Such high activity has not been observed for surface-active pore-forming peptides which generally cause 100% leakage at P:L around 1:50 [33
], whereas barrel-stave peptide pores with the observed level of potency have been described in the literature [34
Previous studies have sought to define the size of the pore created by LLP-1 peptides alone. Miller et al., 1993 measured the amounts of 45
C-sucrose, and 14
C-inulin that were able to enter LLP-1 treated CEM cell cultures. 45
Ca (MW 45 Da) and 14
C-sucrose (MW 342.3 Da), but not 14
C-inulin (MW ~5000 Da) were able to pass through LLP-1 treated membranes, but not untreated CEM cells [21
]. In good agreement, membrane perturbation studies utilizing whole virus show that hygromycin b (MW 527 Da) was able to enter cells after infection with HIV-1, while the similar sized G418 (MW 496 Da) was not able to enter [35
]. This suggests that the pore created by the LLP domains has a cutoff around MW = 500 Da.
LLP-3 forms an amphipathic α-helix in the presence of lipids and binds to lipids and disrupts lipid vesicles, but lacks the overall positive charge of the LLP-1 and -2 domains. Kliger et al., 1997 originally identified a leucine zipper-like sequence on its hydrophilic face [14
]. The authors proposed that this type of domain is likely useful in oligomerization of the cytoplasmic tails. This is analogous to an amphipathic α-helical/leucine zipper-like sequence in the TM ectodomain already proposed to play a role in Env oligomerization [36
]. Whether this LLP-3 mediated oligomerization takes place through spanning the membrane, on the inner surface of the membrane, or not at all is unknown and will require characterization of the domain in the context of the protein. LLP-3 is additionally suspected to contain at least one region that interacts with the matrix protein of the virus, both in the virion and in the infected cell [38
It is possible that the membrane lipid composition could affect results in the types of studies presented here. There is recent precedent for this in the virus literature, including the HIV-1 literature [39
]. The presence of sphingomyelin in LUV's exposed to 15-mer peptides overlapping the LLP domains increased membrane disruption as well as lipid mixing and fusion activities [15
]. An attempt was made to perform the above experiments in a different vesicle composition (18%PE : 65%PC : 10%PI : 2%PS : 5%SM and cholesterol/PL (mol/mol) of 0.5). This composition incorporates SM and reflects the basic lipid composition of Xenopus laevis
oocytes and would have allowed for a more direct comparison to physiological experiments to be performed with the same peptides in that system [40
]. Unfortunately, LUV's of this composition were inherently unstable and unusable (data not shown). Therefore, the simpler vesicles composed of 10% POPG: 90% POPC were utilized as a reasonable mimic of the thickness, fluidity and electrostatic surface potential of a biological membrane. It has been previously shown that the positive charge of the LLP peptides are important for its ability to interact with negatively charged lipid membranes [19
]. Thus the use of the negatively charged POPG was appropriate for these studies defining the structure of these domains while binding to an anionic membrane surface.
Integrating the current biochemical and physiological data gathered using the lentiviral lytic peptides, a hypothetical model of their action in the membrane is proposed in Figure . The LLP regions of TM are α-helical in a lipid environment, partition into lipid bilayers, and disrupt lipid membranes. Since Env is known to associate in trimers on the cell surface and in virions [42
], it is easy to speculate that the LLP regions of the Env trimers could associate with each other, forming a pore or channel in the area between them.
Figure 5 Proposed models of the C-terminus of TM. Proposed models of LLP domains in the context of TM and in a lipid membrane. a) A nine pass transmembrane configuration and b) Association of the LLP domains with the inner leaflet of the lipid membrane allowing (more ...)
Figure depicts one possible configuration of a pore formed by the cytoplasmic tail and LLP regions of gp41 (TM). Further support for this transmembrane configuration of the cytoplasmic tail of TM has come from the detection and characterization of neutralizing antibodies to several regions of the Kennedy peptide, a very hydrophilic region spanning approximately residues 731–752 of the cytoplasmic tail of TM (between TMD2 and TMD3 in Figure ) [44
]. Cleveland et al, 2000 suggest that the major TM domain of gp41 actually span the membrane twice (labeled as TM and TMD2 in Figure ). This could allow the TMD3 and TMD4 to be LLP-2 and LLP-1 respectively, placing LLP-3 on the inner leaflet of the plasma membrane to interact with the matrix protein. However, direct evidence for this model is currently lacking, leaving open the possibility of an as of yet unidentified membrane spanning region that would constitute TMD2.
The presence of the hydrophilic region, or Kennedy peptide, outside the membrane suggests that there would need to be at least one additional membrane spanning domain to bring the rest of the cytoplasmic tail back into the interior of the cell. This could make the environment more favorable to additional membrane spanning regions, such as LLP-1 or even LLP-3.
Based on its average W-W hydropathy score, LLP-3 may lie on the surface of the inner leaflet of the plasma membrane. LLP-3 domains in this case may interact with each other through the leucine zipper-like motifs formed from the peptide's α-helical secondary structure and/or the LLP-3 domain could then be free to interact with the matrix protein of HIV [38
]. This could have the effect of stabilizing the Env trimers and/or the resulting transmembrane pore that could then be formed by the LLP-1 and -2 domains. The location of the LLP-3 domain on the inner leaflet of the cell membrane could also serve to nonspecifically destabilize the lipid bilayer to increase viroporin function as has been observed with Simliki forest virus domains [50
Viral ion channels, or viroporins, are present in many lytic animal viruses. Increased membrane permeability caused by viroporins, glycoproteins, and proteases is a typical feature of animal virus infections [51
]. Viroporins are virally encoded, small (generally ≤ 120 amino acid residues) membrane proteins that form selective channels in lipid membranes. Features common to viroporins include: promoting the release of virus, altering cellular vesicular and glycoprotein trafficking, and increasing membrane permeability. Amphipathic α-helical domains of viroporins generally oligomerize to form the channel by inserting into lipid membranes with the hydrophobic residues oriented towards the lipid bilayer and the hydrophilic residues facing in towards the lumen of the channel. Though viroporins are not essential for virus replication, they may be necessary for full pathogenesis in vivo
as they are known to enhance virion production and release [52
]. Mounting evidence, including data presented here, suggests that the intracellular tail of gp41 constitutes a viroporin and deserves further investigation as such to determine its exact role in the viral replication cycle.
That HIV may code a viroporin in its major surface glycoprotein would ensure that the membrane perturbation, ion fluxes, volume changes, and resulting "loosening" of the plasma membrane and cytoskeleton always occur when and where it is needed for budding, syncytial formation, and/or single cell balloon degeneration. Concentrating HIV glycoproteins in lipid rafts could allow for localized unstable membrane regions at the exact points where it is needed by HIV. While it seems possible that Vpu could also act at these stages to accomplish the same goals, given that it also causes membrane leakage, it is more difficult to envision how it could accomplish the task as Vpu has been shown to be excluded from the plasma membrane and HIV virions [53
Prior observations that LLP-1 can bind to intracellular signalling molecules, such as calmodulin to ultimately induce apoptosis and/or necrosis [16
] suggest that the LLP domains may be configured in certain situations to be associated exclusively with the inner leaflet of the lipid membrane where they are able to interact with these intracellular molecules (see Figure ). Flip-flopping between lipid bilayers of amphipathic pore forming peptides has been documented with melittin [57
]. Based on reported similarities between melittin and LLP peptides, it is reasonable to hypothesize that the LLP domains may be flip-flopping between a transmembrane state and parallel association with the inner leaflet of the lipid bilayer. On the other hand, the LLP domains may possess different activities in the different cell types that it infects, or there may be some as of yet undefined temporal control that allows these two alternate functions to take place at appropriate times during infection.