Design and Characterization of MPER Display Mutants
To focus the antibody response to MPER in its native conformation, large deletion mutants of gp41 expression plasmids were generated encoding only 18, 24, 27, 32, 37, or 42 of the most membrane proximal amino acids of the ectodomain of gp41 (). For efficient display of MPER on the cell membrane and on the surface of virus-like particles, the MPER sequences were flanked by a leader peptide at the N-terminus and a C-terminal transmembrane domain of HIV-gp41 fused to the intracytoplasmic domain of VSV-G. To further mimic the natural conformation and to facilitate detection of the different MPER mutants, a trimerization domain and a peptide tag (Ollas-Tag) were also included. The MPER sequences were separated from the N-terminal heterologous sequences by a flexible linker to facilitate folding of MPER into its native conformation.
Map of domains and amino acid sequences of the MPER display mutants.
To determine expression and accessibility of MPER, HEK293T cells were transfected with the different MPER display mutants and stained with 4E10 and 2F5 antibodies for flow cytometric analysis (). The mean fluorescence intensities for cells transfected with the MPER display mutants were generally higher than those observed after transfection of the parental gp41 expression plasmid and similar in magnitude to the full-length Env (). As expected from the deletion of the 2F5 core epitope, the MPER18 mutant was only detected by the 4E10 antibody, while expression of ΔEnv, which lacks the entire MPER could neither be detected by 4E10 nor by 2F5. The trimerization domain did not affect expression and accessibility, since transfection of MPER18, MPER24 and MPER42 mutants lacking the trimerization domain resulted in similar staining intensities as observed for the parental MPER display mutants (data not shown).
Surface expression of the MPER display mutants.
Incorporation of MPER Display Mutants into VLPs
Since we aimed to analyze the immunogenicity of MPER display mutants by immunization with VLPs, the incorporation of the different MPER mutants into particles was analyzed. VLPs were partially purified from the supernatants of HEK293T cells cotransfected with HIV-1 gag-pol expression plasmid and the different MPER display mutants and MPER content was determined by Western blot analysis using 2F5, 4E10 and anti-Ollas antibodies (-C).
Incorporation of MPER display mutants into VLPs.
The predicted molecular weight as calculated by Vector NTI program (Invitrogen) ranged between ~18.5 kD for MPER18 and ~21.3 kD for MPER42. None of the Western blot bands observed confirmed these predictions. For each of the MPER mutant three bands of higher electrophoretic mobility were observed (-C). The apparent molecular weights of these three bands gradually increased with the numbers of MPER amino acids encoded by the MPER18 mutant up to the MPER37 mutant (-C). For the MPER42 mutant, however, the electrophoretic mobility is substantially higher than the one observed for the MPER37 mutant. The results obtained are consistent for the three antibodies 2F5, 4E10, and anti-Ollas used for the detection of the MPER display mutants. Under non-reducing conditions, a larger band appears for all the mutants with approximately three times the apparent molecular weight of the smaller band obtained for each MPER mutant under stringent denaturing conditions (-C). These results suggest substantial differences in electrophoretic mobility due to conformational differences associated with the high content of alpha-helical structures within MPER.
To obtain further evidence for incorporation of MPER mutants into VLPs rather than exosome like vesicles, the MPER42 mutant was cotransfected with a gag-gfp expression plasmid. The supernatant of the transfected cells was then incubated with fluorescently-labeled 4E10 antibody conjugated to Alexa Fluor 647. VLPs were pelleted through a 20% sucrose cushion and imaged by confocal microscopy. Co-localization of 70.1% of the Alexa Fluor 647 positive spots with GFP positive particles indicated incorporation of the MPER42 mutant into VLPs (). For control VLPs lacking MPER a co-localization was observed in less than 17.5% of the Alexa Fluor 647 positive spots representing the background staining (). The MPER42 mutant was incorporated with higher frequency than gp41 (40.5%) (), and with comparable frequency as the wild type Env (73.3%) ().
Co-localization of MPER42 with VLPs.
Immunogenicity of MPER Display Mutants After DNA Prime-VLP Boost Immunization
Although the different MPER mutants did not differ substantially in accessibility to 4E10 and 2F5 (with the exception of MPER18), the Western blot analyses suggested different conformations. We therefore explored in a pilot immunization experiment in mice whether the MPER18, MPER24 and MPER42 mutants would induce different levels of MPER-specific antibodies. In addition, the MPER24 and MPER42 mutants lacking the trimerization domain (MPER24ΔTRIM, MPER42ΔTRIM) were included to explore a potential influence of the trimerization domain on the immunogenicity.
Mice were first immunized three times by intramuscular electroporation of DNA vaccines encoding the different MPER mutants (). Although we had previously observed that two intramuscular electroporations of DNA vaccines encoding HA of influenza were sufficient to induce readily detectable levels of HA-specific antibodies 
, MPER-specific antibody responses remained undetectable even after the third DNA immunization. Therefore, the mice were further boosted three times by VLPs containing the same MPER display mutants used for the DNA immunizations (). Three weeks after the last VLP immunization, MPER-specific antibody responses could be detected in all animals immunized with MPER display mutants containing the trimerization domain. In contrast, MPER specific antibody responses after immunization with MPER24ΔTRIM and MPER42ΔTRIM were only seen in a single mouse ().
Antibody response after VLP boost in DNA immunized mice.
To confirm the induction of MPER-specific antibody responses and to determine the contribution of the DNA immunizations on the antibody responses seen after the VLP booster immunizations, mice were vaccinated with MPER24, MPER42, gp41 and full length Env gp160 containing VLPs with or without prior DNA priming immunizations ().
Influence of DNA priming on the MPER-specific antibody response after the VLP boost.
As observed previously, DNA immunization alone did not induce detectable MPER-specific Ab response (data not shown and ). After the third VLP immunization, MPER-specific Ab responses were detectable in all animals that had been primed by DNA vaccination (). VLP immunization alone also induced MPER-specific antibodies, but the levels were generally lower and some of the mice did not respond. As observed in the previous immunization experiment (), the MPER42 display mutant tended to induce the highest MPER-specific antibody response (). In two of the mice of the MPER42 group, the MPER-specific antibody response was more than 10-fold higher than the one observed following immunization with full-length gp160. For this group, MPER-specific antibodies were also analyzed longitudinally demonstrating a substantial increase in MPER-specific antibody levels after the second VLP immunization in two of the immunized mice ().
Characterization of the MPER-specific Antibody Response
To determine whether the MPER-specific antibodies induced by MPER42 DNA-VLP immunization could compete for binding with 4E10 and 2F5, sera were tested in the MPER peptide ELISA in the presence or absence of an excess of 4E10 and 2F5. The monoclonal antibody 3D6 binding to an epitope of gp41, which does not overlap with the MPER peptide used in the ELISA, was used as a negative control. Analyzing the sera with the highest MPER-specific antibody response, 2F5, but not 3D6 or 4E10 (data not shown) competed for binding to the MPER peptide () suggesting that antibodies were induced that either recognize or overlap the 2F5 epitope.
Characterization of the MPER-specific antibody response.
The same sera were also analyzed for neutralizing activity using the TZM-bl pseudotype assay 
and compared to the neutralizing activity of sera from mice responding to DNA prime VLP boost immunization with gp41 or gp160 vaccines. No evidence of neutralization could be observed with the sera from MPER42 or gp41 immunized mice (). Weak neutralization activity with a 50% neutralization titer of 1/25 was observed with the serum from the mouse immunized with gp160 vaccines.