An effective HIV vaccine should induce powerful and durable immunity such as to prevent infection in healthy individuals and/or reduce viral replication and viral load in infected ones, slowing or halting disease transmission and progression. The first experimental immunization of human against the AIDS retrovirus HIV-1 was started in a series of HIV seronegative healthy volunteers in November 1986, using vaccinia virus recombinant (V25) expressing gp160 env determinants of HTLV III B at the surface of infected cells, four different protocols were used, for the first time the results show that an immune stage against HIV can be obtained in a man . Since then, more than 256 trials (concluded or ongoing mostly Phase I or II), involving over 44,000 healthy human volunteers have tested HIV vaccine candidates [38-43]. Of these trials, only six (VAX004, VAX003, Step, Phambili, RV114 and HVTN 505) have reached clinical efficacy (Phase IIb and III) (Table 1).
The first advances intended to induce neutralizing antibodies; however, difficulties were encountered with the immunogen. The focus of HIV vaccines then turned on to CD8+ T cell responses, which seemed to be a logical approach [38-43]. The scope was to induce a long-lived memory CD8+ T cell population, capable of rapid killing infected cells in secondary exposures to the virus. Instead, the results of the step trials were disappointing and generated new doubts on the strategy used. Hence, a combined approach, bNAbs neutralizing infecting viruses along with an efficient cellular response (capable of eliminating infected cells that bypassed humoral immunity), seems desirable but is difficult to achieve just by simple vaccination .
Many vaccine schemes have been attempted against HIV-1, and every step in their design from antigen, delivery system, adjuvant and booster strategy have been carefully analyzed. Live attenuated vaccines, inactivated virus vaccines, virus-like particles, subunit vaccines and DNA vaccines have been developed and tested. As a result, different immune responses have been elicited, which have been difficult to analyze. Several experiences have promoted new discoveries as follows: (1) new adjuvants were developed [44-46], (2) booster regimens were redesigned to enhance effector memory cell intended for peripheral tissues (homing for lymph nodes to control HIV replication) [47, 48] and to decrease refractory memory CD8+ T cells , (3) therapeutic vaccines, peptides plus dendritic cells (DC), were developed to enhance HIV immunity to eliminate the virus completely even after antiretroviral therapy discontinuation . However, the latent infected cell population, which would need to be eliminated by a robust immune cell vaccine-elicited response, might not be susceptible to cytolytic CD8+ T cell killing, and consequently, it may evade immune system control [50, 51].
As virus variability is a key important issue, two different approaches were proposed as follows: (1) to target constant epitopes, using constant immunogens and (2) to use mosaic antigens to broaden immune response against the virus [52, 53]. Engineered synthetic genes that only express the consensus, conserved, HIV envelope epitopes, elicited antibodies to a wide span of HIV pseudoviruses as compared to using wild-type Env proteins . In rhesus macaques, mosaic antigen vaccines expanded the span (diversity of recognized epitopes) and depth (diversity of recognized variants for a given epitope) of the specific immune response in CD8+ and CD4+ cells without affecting the neutralizing antibody response . It is has been shown to be an interesting tactic to improve vaccine efficacy.
Live attenuated and inactivated virus vaccines
Whole virus vaccines are processed and presented via MHC-I inducing a cellular response in a manner resembling natural infection. Both live attenuated and inactivated virus vaccines have proven to offer immunity in NHPs. Formalin inactivated simian immunodeficiency virus (SIV) induced protection in the majority of vaccinated macaques against intravenous pathogenic virus . Some attenuated nef-deleted SIV vaccines have shown success in inducing complete immunity in macaques . Macaques immunized with a nef and vpr-deleted SIV strain, SIVmac239Δ3, were protected against disease and reduced viral load when challenged intravenously with highly pathogenic SIV and SHIV strains . However, immune response is not constantly elicited .
Although animal studies with whole virus vaccines show progress, viruses have not been tested in humans owing to safety concerns . As viruses stimulate a vigorous immune response, low-grade persistent infection with continuous antigenic exposure, more efforts have been directed to produce replication competent viral vectors for vaccine delivery [61, 62]. Hence, several replication competent viral vector vaccines have been developed for animal testing and human trials, including adenovirus, poxvirus, yellow fever virus, Venezuelan equine encephalitis virus and other promising immunogens [22, 40].
Protein subunit vaccines
In the light of the findings that an adequate neutralizing antibody response could offer protection in NHPs, early vaccine developments were designed with this goal [22, 40]. It seemed logical to use Env, the only HIV neutralizing antibody target site, with adequate adjuvants to elicit a protective humoral immune response. Vaccines from HIV-1 protein subunits gp160, gp120 and gp140 have been developed, including the first HIV vaccine to reach Phase I trial, a recombinant gp160 molecule produced in a baculovirus insect cell system . They produce neutralizing antibodies and activate CD4+ T cells; however, they do not produce a CD8+ CTL response , which is an important drawback.
In the 1990s, the AIDSVAX B/B' vaccine (VaxGen), employing HIV-1 clade B recombinant gp120 envelope proteins, and a primary clade B isolate (GNE8), with alum salts as adjuvant, showed good safety and immunogenicity in Phase I and II trials [61-63]. In 1998, the VAX004 trial became the first Phase III HIV vaccine candidate, testing the vaccine in North America and the Netherlands . A similar vaccine, AIDSVAX B/E, MN clade B gp120 with a bivalent clade E rgp120 protein from a primary isolate (A244) as the second antigen, passed Phase I/II trials in Thailand [64, 65]. Both trials recruited volunteers in population in high risk for HIV infection, VAX004 enrolling 5,417 men that have sex with men (MSM) and women at high heterosexual infection risk and VAX003 2546 intravenous drug users (IDUs) . Neither vaccine offered protection to infection, reduced viral load, nor slowed disease progression despite the presence, in some individuals, of neutralizing and non-neutralizing Env-specific antibodies [64-66]. ADCVI-mediating antibodies did inversely correlate with HIV infection probability; on the other hand, neutralizing antibodies did not . Thus, an effective vaccine would require mucosal antibody response along with a robust cellular immunity against the virus.
The interest in protein-based vaccines has been lost due to poor immunogenicity. New strategies have been developed to overcome this drawback. DC targeting has been shown to be a useful strategy. Homing the antigen to DC cells with anti-DEC205 antibody potentiated both humoral and CD4+, but not CD8+, Gag-specific immune responses in mice 
Live recombinant vaccines
A method of gene delivery to host tissues is the use of a viral vector containing the desired genes to be introduced. When these recombinant viruses enter host cells, with the foreign gene, they induce antigen expression (MHC class I and II), phagocytosis of apoptotic antigen-expressing cells and induction of CD4+ T cell responses . The most used viral vectors are poxviruses, including canarypox, fowlpox, but modified vaccinia virus ankara and the modified Copenhagen strain NYVAC [71-73]. These viruses, along with Ad26 and Ad35, predominate in clinical trials  as they are naturally or artificially replication-deficient in primates, but also less immunogenic than other structures. Given the fact that some of these vectors suppress immune responses, immunogenicity was enhanced by the deletion of interferon pathway-blocking genes in MVA or NYVAC . Also, pre-existing immunity to some vectors might negatively affect vaccine efficacy  while it does not affect others, fowlpox and canarypox, . Thus, the use of distinct, more effective, vector serotypes, such as Ad26 and Ad35, might enhance future developments in this field [74, 75].
The failure of the VAX003 and VAX004 trials, which elicited a B cell response, brought attention to T cell vaccines as an alternative to HIV protection. The step study (HVTN 502) was the first completed trial to evaluate whether a vaccine inducing cellular immunity to HIV-1 could protect against infection or reduce viral load [76-78]. The vaccine tested, developed by Merck Research Laboratories, was a recombinant live vector composed of replication-defective recombinant adenovirus 5 (rAd5) expressing HIV-1 clade B gag, pol and nef [76-78]. As env was not included, a neutralizing antibody response was not to be elicited. Cellular immunity was the scope of the vaccine. The study enrolled high-risk populations, mainly MSM, but also heterosexual males and females in North and South America, Australia and the Caribbean. The Phambili sister trial (HVTN 503) tested the same vaccine regimen in South African heterosexual males and females [76, 79].
Both studies were terminated early due to higher infection rates in the vaccinated group, despite finding, for the first time, IFN-γ secreting HIV-specific T cells by ELISPOT assays [79, 80]. HIV-specific CD4+ and CD8+ T cells were identified in 41% and 73% of the individuals (32% for both) . However, cell activations do not differentiate between HIV cases and non-cases and/or cytokine profiles . Vaccination did not affect the course of infection at a 2-year follow-up; time to initiate antiretroviral therapy, HIV RNA levels, CD4+ T cell counts, and AIDS-free survival . Thus, IFN-γ response was not associated with protection or lower infection rates.
Post hoc analysis showed that the higher infection rate in the step trials was restricted to MSM who were uncircumcised and/or had high basal neutralizing antibody titres to Ad5 (2.3-fold increase in the latter) [77, 78]. Two hypotheses might explain this phenomenon as follows: (1) rAd5 administration expands specific memory cell population, thus providing HIV more cellular targets to infect; (2) the rAd5 vaccine forms immune complexes with Ad5 neutralizing antibodies, promoting CD4+ T cell infection by HIV. The first hypothesis is largely refuted by the fact that basal Ad5 neutralizing antibody titres do not correlate with prior CD4+ or CD8+ T cell immunity, thus to memory cell population expansion and circulating CCR5+ T cell levels were not greater between patients with high and low Ad5 antibody titres . Nevertheless, rAd5 vaccine administration did not result in different rAd5-specific CD4+ T cell trafficking to mucosal sites between rhesus macaques with high and low basal anti-Ad5 antibody titres . The second hypothesis, on the other hand, was strengthened by the finding that rAd5-antibody complexes activate Langerhans cells, more than serum or Ad5 alone and render CD4+ cells vulnerable to HIV infection [73, 74]. The higher infection susceptibility decreased over the 18- and 36-month follow-up until it equaled the placebo group [78-80].
The Phambili trial was interrupted with <10% of the vaccinated participants due to step results. Analysis showed no statistical difference in HIV acquisition risk or viral set point between placebo and vaccine groups . The trial used the same vaccine from step protocol; a clade B vaccine in a clade C area. The trial was justified on the basis of cross-clade CTL responses [79, 80].
These results have led to propose a combined vaccine: to prime with a live vector vaccine and boost with a protein subunit . This strategy induces both a cellular and humoral responses and has yielded some of the most effective results to date, including a Phase III trial RV144.
The Phase III HIV-1 vaccine trial RV144 was a test-of-concept trial conducted by the Thai Ministry of Public Health and sponsored by the US Army Surgeon General, managed by the US Military HIV Research Program (MHRP) [81, 82]. The vaccine tested was based on non-replicating recombinant canarypox virus (ALVAC) encoding clade B gag/pol and CRF01_AE env antigens (ALVAC-HIV vCP1521 from SanofiPasteur). It was administered in 2 primer doses, followed by 2 booster immunizations with the same ALVAC recombinant along with subtype B and CRF_AE rgp120s, (the VaxGen AIDSVAX immunogen) . Clades B and E were chosen for the vaccine due to their local prevalence. Alum, instead of MF59, was used. The rationale was that using a live vector vaccine followed by a protein boost would induce both cellular and humoral immunity. The trial was launched in Thailand in 2003 [82, 83], it ended in 2009 and involved 16,402 volunteers (the largest vaccine trial to date) at risk of infection. In a modified intention-to-treat analysis (excluding patients infected at the first visit), the vaccine showed 31.2% efficacy (95% CI, 1.1–52.1; P = 0.04) at preventing HIV infection , which is encouraging for future HIV vaccine development. The protection rate was higher among low and middle infection risk populations, compared with individuals with high infection risk . The vaccine did not, however, reduced the viral load or CD4+ count on individuals who were subsequently infected after vaccination . The most common response to the vaccine was a robust CD4+ T cell response to Env antigens [82, 83]. Only 24% of the individuals showed a CD8+ T cell response, <20% of the group showed an IFN-γ ELISPOT response, and, in general, low titres of neutralizing antibodies were recorded. After Env priming, CD4+ intracellular cytokines were significantly higher in the vaccine as compared to the placebo group (34 versus 3.6%; P < 0.001) . Serum virion capture antibodies were elicited in a fraction of individuals, including antibodies able to cross-react with transmitter/founder strains . As with other rgp120-based vaccines, most volunteers developed ADCC activity . It is unknown, however, if these antibodies are present in the mucosa. In summary, the trial showed a protection efficacy of 61% 1 year after vaccination, but it decreased to about half at the end of the trial.
Given these results, changes in vaccination scheme regimen and adjuvant are underway [39, 2]. It has been reported that a high anti-Env V1/V2 loop-specific and non-neutralizing, but ADCC enhancing antibody response was associated with a 71% decreased infection risk comparing low to medium responders; however, high anti-Env IgA titres increased in the infection risk group [2, 87, 88]. These high specific IgA titres, possibly monomeric IgA, could compete with protective IgG .
DNA vaccines' mechanism of action relies on the administration of non-living, non-replicating, non-transmissible plasmids, which that are taken up by host cells (but very rarely integrated into their DNA). It encode proteins expressed by host tissues (predominantly skin and muscle), which induce class I and class II antigenic presentation. Cytoplasmic DNA sensing machinery may recognize the transfected plasmids and its metabolites and induce the production of type I IFNs . However, these vaccines have been shown to induce low immunogenicity, due a low transfection efficiency, low DNA uptake by non-APCs and consequently, low antigen transcription rate. In vivo plasmid electroporation and cytokine coadministration have been effective increasing CD4+ and CD8+ responses .
One of the most difficult tasks is to enhance antigen presentation. Targeting DC surface molecules to induce cell homing and to facilitate antigen uptake has been attempted. A PD1-p24 DNA vaccine was able to enhance CD8+ (cytokine secretion, cytotoxicity), CD4+ T cell function and antibody titres in mice superiorly in comparison with an anti-DEC205 approach  suggesting that PD-L1 homing antigens are effective vaccine enhancers.
HIV DNA vaccines work better with primers in a heterogonous prime-boost or co-immunization strategy (administration of the same antigen in different vehicles) . Mice immunized with a gag DNA, primed and boosted, recombinant Tiantan vaccinia virus were shown to exhibit potent Gag-specific T cell responses, independently of the doses used . Likewise, DNA bound to mannosylated polyethyleneimine enhanced Gag-specific cellular and humoral responses in a murine DNA/Ad prime-boost protocol . Co-immunizing mice and macaques with DNA and protein, rather than prime and booster schemes, provoked marked Env-specific humoral responses without jeopardizing cellular immunity .
In the latest clinical efficacy trial, HVTN 505 tested in circumcised MSM without pre-existing antivector antibody titres, a DNA prime-Ad boost regimen was used . The DNA prime consisted of 3 doses of a plasmid encoding env from HIV clades A, B and C, plus gag, pol and nef from clade B. The Ad5 boost encoded env from 3 clades and clade B gag and pol. The trial, which started in 2009, was scheduled to end in 2015, but was prematurely stopped in April 2013 due to lack efficacy [1, 95].
Lipoid vesicular biocarrier systems for drug delivery have good bioavailability and pharmacokinetic properties [69, 96, 97]. These systems are suitable for encapsulation of both hydrophilic and hydrophobic drugs, enhance drug stability, increase therapeutic index, and delay elimination [69, 96, 97].
Certain products have been used in vaccination strategies as antigen delivery systems and adjuvants [69, 96, 97]. CAF01, a liposomal adjuvant system, is composed of cationic surfactant dimethyldioctadecylammonium and the immunomodulating glycolipid trehalose dibehenate, rendering a strong Th1 response and a humoral response [96, 97]. Charged cationic vesicles facilitate antigen adsorption at the site of administration, promoting APC presentation and stimulating strong Th1 and Th17 responses . Moreover, virosomes induced mucosal antibodies protecting non-human primates against vaginal SHIV challenges .
A variety of liposomes have been studied for vaccine development against pathogens. Virosomes, vaccine candidates for HIV-1, influenza and hepatitis C virus vaccines, are reconstituted viral envelope lipid bilayers that function as antigen carriers and adjuvants. They are able to stimulate good antibody production and cellular immunity [96-98]. Virosomes may elicit either a CD4+ or CD8+ response as desired. Antigens attached to the virosome outer membrane will undergo APC endocytosis and thus MHC-II presentation. Antigens embedded inside the virosome are transported directly to the cytosol, and then processed, expressed via MHC-I, to elicit CD8+ responses. Thus both helper T cells and CTLs may be primed . In contrast to other viral vector vaccines, pre-existing antibodies aid virosome delivery to APCs without inhibiting immune response .
Lipopeptides have been also used in vaccine strategies to activate DCs through TLR-2 [46, 97]. The LIPO-4 vaccine contains four palmitoyl-lysine associated Gag, Pol and Nef peptides covalently liked to tetanus toxin peptide (French National Agency for AIDS and Hepatitis Research VAC16 trial). The vaccine was able to induce HIV-specific CD4+ and CD8+ responses independently of the administration route, intramuscular or intradermal . However, the intradermal route induced a weaker CD4+ response .
In summary, lipogenic biocarriers represent an interesting approach for future vaccines, not only in HIV, but with several other pathogens.
Inducing effective long-lasting immunity in the mucosa would prevent virus entry and systemic infection. Considering that only some HIV quasi-species are infective through the mucosa, and that 80% of heterosexuals are infected by a single ‘founder/transmitter’ virus [23, 101], a mucosal antibody response could be effective to this reduced infecting virus repertoire .
The site of vaccine application need not be the site of immunity induction, and it might alter cellular and humoral responses. In rhesus macaques, peroral administration of a SIV vaccine regimen was as effective as vaginal administration in producing vaginal IgAs and more effective than nasal, intestinal or vaginal routes in producing rectal IgAs . In this study, immunization in the oral cavity elicited the most significant and diverse T cell responses as compared to multiple mucosal sites . Nasal immunization may elicit widely disseminated mucosal IgA responses and greater IgG systemic responses than the peroral, rectal or vaginal routes. In an additional study, systemic immunization with SIV Gag, either through a DNA/NYVAC or recombinant BCG/NYVAC prime-boost strategy elicited potent and durable GI mucosal CD8+ responses. These cells were recruited from the systemic circulating pool early after vaccination and were mostly of the effector memory phenotype . Thus, strong cellular immunity may be achieved at the mucosa through systemic vaccination. Cellular mucosal immunity is impaired by IL-13. Recently, a ‘cytokine trap’ mucosal vaccine has shown to enhance the mucosal CD8+ response by incorporating soluble and membrane bound IL-13 receptor α2 to blocking the cytokine . Nonetheless, vaccines capable of inducing immunity at the mucosal sites may not trigger systemic immunity and vice versa.
Recently, a virosome harbouring surface HIV-1 lipidated gp41 P1 peptide (MYM-V101) was shown in a Phase I trial to be able to elicit, after intramuscular priming and intranasal booster doses, vaginal IgAs in 63% and rectal IgAs in 29% of women; while these antibodies were not found to be neutralizing, they were shown to be able to inhibit virus transcytosis . This effort could be the beginning of a new class of vaccines for several pathogens.