Obstacles to the successful development of an efficacious T cell‐inducing HIV‐1 vaccine

Short abstract

Preclinical and clinical HIV‐1 cell vaccine development continues to be hampered by the lack of validated preclinical animal models: a discussion about way forward.

Abstract

An efficacious vaccine to HIV‐1 is direly needed to stem the global pandemic. Immunogens that elicit broadly cross‐neutralizing antibodies to HIV‐1 remain elusive, and thus, most HIV‐1 vaccine efforts are focusing on induction of T cells. The notion that T cells can mediate protection against HIV‐1 has been called into question by the failure of the STEP trial, which was designed to test this concept by the use of an E1‐deleted Ad vaccine carrier. Lack of efficacy of the STEP trial vaccine underscores our limited knowledge about correlates of immune protection against HIV‐1 and stresses the need for an enhanced commitment to basic research, including preclinical and clinical vaccine studies. In this review, we discuss known correlates of protection against HIV‐1 and different vaccine strategies that have been or are being explored to induce such correlates, focusing on T cell‐inducing vaccines and particularly on Ad vectors.

Abbreviations

AAV

adeno‐associated viral vector(s)

Ad

adenovirus

AdC

Ad vectors derived from chimpanzees

AdHu5

Ad vectors of the human serotype 5

AdHu35/26

Ad vector derived from other human serotypes

CAR

coxsackie and Ad receptor

EC

elite controller(s)

env

envelope protein

HAART

highly active antiretroviral therapy

HEK

human embryo kidney

HVTN

HIV Vaccine Trial Network

LTNP

long‐term nonprogressor(s)

MVA

modified vaccinia Ankara

NHP

nonhuman primate

RCA

replication‐competent Ad virus

SHIV

HIV/SIV

VLP

virus‐like particle

VRC

Vaccine Research Center

VSV

vesicular stomatitis virus

Introduction

The failure of Merck's HIV‐1 vaccine trial (called STEP trial), based on replication‐defective AdHu5, designed to induce T cell responses to conserved internal proteins of HIV‐1 [ ¹ ], raises questions about the concept of T cell vaccines to HIV‐1 [ ² ] and necessitates more basic research about correlates of protection and how such correlates can be achieved by vaccines. The STEP trial not only showed lack of vaccine efficacy but in contrast, showed a trend toward increased acquisition of HIV‐1 in vaccinees with moderate‐to‐high, pre‐existing neutralizing antibodies to the vaccine carrier prior to immunization [ ¹ ].

Most licensed vaccines induce protection through antibodies. The env of HIV‐1, the virus’ sole target for neutralizing antibodies, is extraordinarily variable, heavily glycosylated, and undergoes structural changes upon receptor binding [ ³ , ⁴ ]. It has thus been impossible so far to design env‐based immunogens, which can induce antibodies that reliably neutralize a broad spectrum of different HIV‐1 isolates. Recent HIV‐1 vaccine efforts have focused largely on induction of T cell responses, most notably CD8⁺ T cells [ ⁵ , ⁶ , ⁷ ], which have been implicated to protect against progression of HIV‐1 infection in humans and which have shown some efficacy in preclinical animal models based on infection of rhesus macaques with SIV or SHIV chimeras [ ⁸ , ⁹ ].

In this review, we will provide a general overview about immune responses to HIV‐1 and on different vaccine prototypes that have been developed and tested against this virus or its surrogate in animals and humans, with a focus on vaccines based on Ad vectors.

HIV‐1 AND AIDS

AIDS, caused by HIV‐1, is the leading cause of death in humans aged 15–59 worldwide. Between 1981 and 2003, an estimated 20 million individuals died of AIDS, and 40.3 million people were infected with HIV‐1 in 2005. It is estimated by UNAIDS that in 2007, 33.2 million humans were living with HIV‐1, an additional 2.1 million humans died of AIDS, and 2.5 million people became newly infected with HIV‐1. Sub‐Saharan Africa has by far the highest incidence rates of HIV‐1 infections, and in some Sub‐Saharan nations, such as in Botswana, a staggering ∼36% of the adult population is infected. HIV‐1 has started to spread in other regions, including highly populated countries of Asia, such as India and China. In the Americas, 65% of infected individuals are treated with antiviral drugs; in contrast, less than 10% of HIV‐1‐infected humans in Africa or Southeast Asia have access to antiretroviral treatment.

In most cases, HIV‐1 is sexually transmitted through the male or female genital tract or the rectal mucosa. Women are at a higher risk than men to become infected by heterosexual contact, and currently 60% of all HIV‐1‐infected individuals are women. Infected women in turn can transmit virus to their offspring, and it is estimated that more than 2.2 million infants have become infected from their mothers.

It was shown in NHPs, upon their low‐dose vaginal SIV infection, that the initial viral replication is limited to small foci within the cervical submucosa for the first few days [ ¹⁰ ]. After the initial local burst of viral replication, the virus disseminates to lymph nodes and then to various organs and lymphoid tissues [ ¹¹ ]. In the symptomatic phase, viral replication and CD4⁺ T cell loss occur predominantly in the GALT. Once the viral load has expanded, adaptive immune responses become activated, patients seroconvert, and numbers of HIV‐1 antigen‐specific T cells increase. At this stage, virus titers decrease to a plateau referred to as “viral load set point.” During the usually asymptomatic, latent phase following the acute infection, the patients show a slow but progressive loss of CD4⁺ T cells, which eventually reaches very low counts in the blood. At this time, the symptomatic phase of AIDS begins.

HAART to treat HIV‐1 infection was introduced in 1986. Although increasing quality of life and life expectancy, HAART is not an optimal solution to the HIV‐1 pandemic. HAART's costs, side‐effects, and complex regimens make it difficult for the patient to access or adhere to the treatment [ ¹² ]. The drugs are not available in the poorest countries, which have the highest prevalence rates of HIV‐1 infections. Last but not least, the drugs do not eliminate HIV‐1 but only reduce viral loads, which resurge rapidly once treatment is stopped [ ¹³ ]. An efficacious and safe vaccine to HIV‐1 would certainly be best to control the pandemic. Vaccine‐induced immune mechanisms, which can clear the virus or virus‐infected cells during the early stages of infection at the site of viral entry, when viral load is still limited and before the virus mutates extensively, would be best suited to prevent a systemic infection and its pathological consequences. Vaccines that only reduce viral loads may offer some temporary advantages, but one would expect that the highly mutating HIV‐1 virus, even if combated efficiently by a vaccine‐induced immune response, would eventually escape immunosurveillance through mutations.

CORRELATES OF PROTECTION

Although most current vaccines were developed without any knowledge of potential immune correlates of protection against the pathogen and in fact, helped to partially establish correlates of protection, such knowledge would clearly aid in the development of vaccines against complex pathogens such as HIV‐1.

Neutralizing antibodies are known to protect against many viruses, and the initial efforts to develop an HIV vaccine were concentrated on exploring humoral immune responses against HIV. In experimental models, neutralizing antibodies given by passive transfer protect against infections with pathogenic strains of SHIV chimeras [ ¹⁴ , ¹⁵ ], which carry the envelope of HIV‐1, and HIV‐1‐specific antibodies of the IgA isotype have been isolated from mucosal surfaces and blood of exposed, seronegative women, suggesting that they might prevent infection [ ¹⁶ , ¹⁷ ]. Although neutralizing antibodies would be expected to protect against HIV‐1 infection, with few exceptions, most neutralizing antibodies recognize only closely related isolates of HIV‐1 [ ¹⁸ ], and immunogens that elicit broadly cross‐neutralizing antibodies reliably remain elusive.

Although T cells are in general unable to prevent infection, they are crucial for its control, including an infection with HIV‐1, as suggested by several lines of evidence. Some humans are remarkably resistant to developing AIDS upon HIV‐1 infection, instead, controlling viral loads to low or even undetectable levels [ ¹⁹ , ²⁰ ]. These individuals are classified as LTNP or EC. LTNP show low but variable plasma viremia and occasionally develop AIDS after long asymptomatic phases [ ²¹ ]; EC are able to control HIV‐1 to undetectable levels. It was shown that certain HLA class I alleles are more common in ECs [ ⁹ ] and furthermore, that such individuals develop exceptionally strong CD8⁺ T cell responses to HIV gag upon infection, which can lead to mutations that impair the fitness of HIV‐1 [ ²² ]. Although these studies are far from conclusive, they suggest that HIV‐1‐specific CD8⁺ T cells can protect against the spread of an HIV‐1 infection. This notion is supported by other studies, which showed that after the acute phase of infection, the appearance of HIV‐1‐specific CD8⁺ T lymphocytes is associated temporally with a decrease in viral load [ ²³ ] or that resistance against progression is linked to the presence of polyfunctional CD8⁺ T cells in blood [ ⁸ ]. CD8⁺ T cell depletion studies in NHPs indicate that CD8⁺ T cells play a major role in controlling SIV/SHIV infection [ ²⁴ , ²⁵ ], but these results have to be interpreted with caution, as available antibodies to primary CD8⁺ T cells also deplete NK cells. In spite of the data, correlates of cell‐mediated protection against HIV‐1 remain ill‐defined and are likely to depend in part on route of transmission, host factors, and viral fitness.

Considering that most HIV infections start at the mucosal level, the anatomic distribution of vaccine‐induced immune responses is presumably critical. Vaccine‐induced immune responses present at the port of entry might be decisive to fight a primary infection before a viral reservoir is established within lymphoid tissues and the gut. CD4⁺ T cells in GALT are primary targets of HIV‐1 [ ²⁶ ]. This may relate to interactions between gp120 of the env and α4β7 integrins, the latter involved in leukocyte migration to and retention in the intestine. This interaction promotes formation of synapses that support cell‐to‐cell spreading of the virus [ ²⁷ ]. Studies in primates infected with SIV have shown that CD4⁺ T cell loss within the GALT and its associated lymphoid tissues disrupts the protective barrier of the intestine, allowing for bacterial antigens such as LPS to enter the bloodstream [ ²⁸ ]. This in turn results in a global activation of CD4⁺ T cells, which promotes progression of the virus by providing target cells to HIV‐1/SIV [ ²⁹ ]. Preventing the early CD4⁺ T cell loss within the gut and thus, preserving the mucosal barrier of the intestine may be key to protecting against high viral set‐point loads and progression to AIDS [ ³⁰ ].

ANIMAL MODELS FOR HIV‐1 VACCINES

A major obstacle in developing an efficacious HIV‐1 vaccine is the lack of a validated animal model. Chimpanzees are susceptible to HIV‐1 but cannot be used for ethical reasons. SIV, a lentivirus closely related to HIV‐1, causes chronic but benign infections in its natural hosts such as sooty mangabeys [ ³¹ ]. Macaques develop AIDS upon infection with SIV or SHIV, but the infections are more virulent than HIV‐1 infections in humans. It remains unknown to date if results obtained with SIV/SHIV infections in rhesus macaques or cynomolgus monkeys are relevant for human HIV‐1 infection. Nevertheless, most HIV‐1 vaccines undergo extensive preclinical testing in these primates to assess if they induce protection against challenge with SIV or SHIV. Table 1 shows some of the results obtained with different vaccines in SIV or SHIV challenge models.

Table 1.

NHP Challenge Experiments

Vaccine	Antigen	Challenge virus/challenge route	Result	References
DNA vaccine (IL‐15)	gag	SHIV89.6P, i.v.	Reduced viral loads	[ 32 ]
DNA vaccine (IL‐12) prime/VSV boost	gag, env	SHIV89.6P, i.v.	Reduced viral loads	[ 33 ]
DNA (IL‐12) prime/AdHu5 boost	multiple	SIVmac251, rectal	Reduced viral loads	[ 34 ]
DNA prime/Listeria monocytogenes boost	gag, env	SIVmac239, rectal	Reduced viral loads	[ 35 ]
DNA (IL‐2) prime/MVA boost	gag, pol, env	SHIV89.6P, rectal	Some protection against CD4+ T cell loss	[ 36 ]
DNA prime/MVA boost	multiple	SIVmac239, rectal	Reduced peak viral loads, no protection against disease	[ 37 ]
DNA prime/NYVAC boost	gag, pol, env	SIVmac251, rectal	Reduced viral loads	[ 38 ]
DNA (IL‐12/IL‐15) AdHu5/protein (gp120, nef)	multiple	SIVmac251, rectal	No protection	[ 39 ]
DNA prime/AdHu5 boost	gag	SIVmac239, rectal	Transient reduction in viral load	[ 40 ]
AdHu5 prime/AdHu5 boost	gag	SIVmac239, rectal	No protection	[ 40 ]
AdHu5 prime/AdHu5 boost	gag	SHIV89.6P, i.v.	Reduced viral loads, no CD4+ T cell loss	[ 7 ]
DNA prime/AdHu5 boost	gag/pol/env	SIVmac251, i.v.	Reduction in peak viral loads, no reduction in viral set point loads, prolonged survival	[ 41 ]
AdHu26 prime/AdHu5 boost	gag	SIVmac251, i.v.	Reduced viral loads	[ 42 ]
Replication competent AdHu5/protein (gp120)	gag, env, rev	SIVmac251, rectal	Reduced viral loads	[ 43 ]
Rabies virus vector	gag, env	SHIV89.6P i.v.	Reduced viral loads	[ 44 ]
RhCMV vector	gag, rev, tat, nef, env	SIVmac239, repeated low‐dose rectal	Increased resistance against infection and complete control of local infections in some NHPs	[ 45 ]

NYVAC, attenuated vaccinia virus; RhCMV, rhesus CMV.

Many vaccines, including T cell‐inducing vaccines, were found to provide at least partial protection against SHIV89.6P, which carries a HIV‐1 clade B envelope [ ⁷ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² ]. SHIV89.6P differs in some crucial aspects from HIV‐1; e.g., it uses CXCR4 for cell entry during the initial infection, it causes AIDS much more rapidly, and its env is more neutralization‐sensitive [ ⁵³ ]. Therefore, SHIV89.6P may not be a valid challenge virus. This is further substantiated by findings such as the one in the STEP trial, where the vaccine used showed no efficacy in humans though had protected NHPs against challenge with SHIV89.6P. Subsequent studies showed that this or similar vaccines failed to protect against challenge with SIVmac239 [ ⁴⁰ ] or SIVmac251 [ ⁴² ]. Additional SHIVs, based on env from HIV‐1 virus clade B or C, have been developed that like HIV‐1, use CCR5 as a coreceptor [ ⁵⁴ , ⁵⁵ ]. Other studies assessed vaccine efficacy by challenging NHPs with SIV, such as SIVmac239 or SIVmac251, given systemically or mucosally [ ⁴¹ , ⁴³ , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ ]. Protection against SIV challenge can be achieved by prior infection of NHPs with an attenuated SIV. Most studies used nef‐deleted SIV [ ⁶¹ ], and others used SIVs attenuated through deletion of V1‐V2 sequences in env [ ⁶² ] or through modification of the transmembrane domain of env [ ⁶³ ]. In these challenge studies, attempts were made to correlate immune responses with level of protection. In most reports, immune responses measured by traditional assays failed to predict if the animals were to be protected against challenge. One study reported a correlation between frequencies of β‐chemokine‐producing CD8⁺ T cells and levels of protection [ ⁶⁴ ], and another study suggested a dominant role for neutralizing antibodies [ ⁶⁵ ].

It appears that protection might be achieved more readily against homologous challenge, i.e., challenge with a virulent SIV that carries the same or a closely related env as the immunogen than against heterologous challenge [ ⁵⁹ ], pointing toward an important role for antibodies that may act, not necessarily solely by neutralization but also by other mechanisms such as antibody‐dependent, cell‐mediated cytotoxicity. In humans, vaccines need to achieve protection against a heterologous challenge, and correlates of protection in preclinical models should thus be defined under conditions that afford this type of protection to NHPs. Other SIV challenge viruses such as SIVE660 have been developed and can be used to assess correlates of protection against a heterologous challenge in NHPs.

In early challenge experiments, the challenge viruses were given i.v.. Although HIV‐1 can be transmitted through this route, as attested by its spread in i.v. drug users, the most common route of transmission is through the mucosal sites of the genital tract or the rectum. For this reason, challenge models, based on mucosal challenge of NHPs, have been developed. Infection through mucosal routes can be achieved by a single administration of a high dose of virus or by repeated challenge with low doses of virus [ ⁶⁶ ]. Although the latter model is appealing, as only few virus particles penetrate the mucosal barrier, mimicking transmission in otherwise healthy humans, it may not be a stringent‐enough challenge model for humans with underlying sexually transmitted diseases, which are known to increase the risk for HIV‐1 acquisition.

Vaccine‐induced protection can be achieved more easily in so‐called EC NHPs, which have certain MHC class I haplotypes such as Mamu‐A*01 or Mamu‐B*17 [ ⁶⁷ ]. NHPs with these haplotypes develop exceptionally high CD8⁺ T cell responses to antigens of SIV that are commonly present in vaccines, again suggesting that a potent CD8⁺ T cell response can afford protection.

The field of HIV‐1 vaccinology would profit substantially if animal models were to be standardized with regard to the challenge virus as well as the route and dose of challenge. If challenge studies were performed using a consensus regimen, different vaccine strategies could be compared more readily. Unfortunately, as none of the different NHP challenge models have been validated by successful clinical trials so far, it is currently impossible to assess which model would be best suited to predict the outcome of vaccination in humans.

EXPERIMENTAL VACCINES TO HIV‐1

Most current HIV‐1 vaccine efforts are focused on inducing protection through T cells, mainly directed against conserved epitopes of the virus. A large number of different vaccine platforms are being investigated, such as adjuvanted protein vaccines [ ⁶⁸ ], VLPs [ ⁶⁹ ], plasmid vectors [ ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ ], or viral [ ⁴⁸ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ ] and bacterial recombinant vectors [ ⁸¹ , ⁸² ]. Viral recombinants based on poxviruses such as MVA [ ⁷⁶ ] or vaccinia virus [ ⁷⁸ ], AAV [ ⁸⁰ ], VSV [ ⁴⁸ ] or measles virus‐based vaccines [ ⁷⁷ ], or different serotypes of Ad vectors [ ⁷⁹ ] have undergone extensive preclinical and in some cases, clinical testing. Some of the different vaccine platforms are listed in Table 2. Most platforms use prime‐boost regimens to further increase the magnitude of the HIV‐1 antigen‐specific immune responses. Prime boost regimens are especially useful to reduce the dose of each individual vaccine, thus reducing dose‐related toxicity, which can be substantial for some of the viral vectors, such as those based on Ad viruses. Prime‐boost regimens commonly combine DNA vaccines or protein vaccines with other modalities [ ⁸¹ , ⁸² ] or heterologous types of a specific viral vector vaccine [ ⁸⁸ , ¹⁰³ ]. For examples, the VRC (Bethesda, MD, USA) has focused on developing a vaccine regimen based on DNA priming, followed by a booster immunization with a replication‐defective Ad of the AdHu5 serotype [ ¹⁰⁴ , ¹⁰⁵ ]; Robinson et al. [ ¹⁰⁶ ] have focused on DNA vaccine priming followed by a MVA boost; Barouch and co‐workers [ ¹⁰⁷ , ¹⁰⁸ ] are developing replication‐defective Ad vectors based on rare human serotypes; our group [ ⁸⁹ , ¹⁰⁹ ] has focused on replication‐defective AdC; Robert‐Guroff and co‐workers [ ¹¹⁰ , ¹¹¹ ] are exploring the use of RCA vectors of the human serotypes 4, 5, and 7; Rose and colleagues [ ⁴⁸ ] and Schnell and co‐workers [ ⁴⁴ ] are exploring rhabdovirus vectors such as VSV or attenuated rabies virus, respectively; Knipe and colleagues [ ⁵⁸ ] are testing herpes virus vectors; and Weiner and co‐workers [ ⁷⁵ , ¹¹² ] and others [ ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ ] are focusing on DNA vaccines.

Table 2.

HIV Vaccine Strategies

Vaccine platform	Immune response/development	References
Plasmid vectors	CD4+ and CD8+ T cells
Naked DNA	Humans—phase I	[ 83 ]
Adjuvanted DNA	Humans—phase I	[ 84 ]
Subunit	Humoral responses
Protein	Humans—phase III	[ 85 ]
Adjuvanted protein	Humans—phase I	[ 68 ]
VLP	Humans—phase II	[ 86 ]
Viral vectors	Cellular and humoral immune responses
AAV vector	NHP	[ 87 ]
AdC	NHP	[ 88 , 89 ]
AdHu5	Humans—phase IIb	[ 1 , 40 ]
AdHu35, AdHu26	Humans—phase I	[ 42 , 90 ]
Canarypox virus (ALVAC)	Humans—phase II	[ 91 ]
Fowlpox virus (FPV)	Humans—phase I	[ 92 ]
MVA	Humans—phase I	[ 81 ]
Vaccinia virus	Humans—phase I	[ 93 ]
Sendai virus	NHP	[ 94 ]
Measles virus	Mice	[ 95 ]
Poliovirus	NHP	[ 56 ]
Rabies virus	NHP	[ 44 ]
VSV	NHP	[ 48 ]
Semliki forest virus (SFV)	NHP	[ 96 ]
Venezuelan equine encephalomyelitis virus (VEE)	Humans—phase I	[ 97 ]
HSV‐1	NHP	[ 58 ]
CMV	NHP	[ 45 ]
Bacterial vectors	Varied immune responses
Clostridium perfringens	Mice	[ 98 ]
Mycobacterium bovis	Guinea pigs	[ 99 ]
Mycobacterium smegmatis	Mice	[ 100 ]
L. monocytogenes	NHP	[ 101 ]
Salmonella enterica serovar Typhimurium	Humans—phase I	[ 102 ]

DNA vaccines, which are being used by many investigators, are immunogenic in small animal models, inducing potent cellular and humoral immune responses to a variety of pathogens. However, a major obstacle has been transferring their immunogenicity/efficacy to larger animal models and humans. This may be achieved with novel delivery methods such as electroporation, which increases uptake of DNA by cells close to the injection site [ ¹¹³ ]. DNA vaccines given alone have so far failed to induce potent immune responses in clinical trials [ ⁸³ ] and thus regimens using DNA vaccines in combination with viral vectors have been studied in several early stage trials. DNA priming followed by MVA boost resulted in detectable cellular immune responses in one trial [ ¹¹⁴ ] but performed poorly in another [ ¹¹⁵ ]. DNA vaccine priming followed by a boost with env [ ¹¹⁶ ] induced antibody and CD4⁺ T cell responses but failed to elicit high titers of neutralizing antibodies or CD8⁺ T cells. DNA priming followed by a boost with an AdHu5 vector was evaluated by Merck and VRC in phase I trials. This combination was immunogenic but did not result in responses that were superior to those induced with the Merck AdHu5 vector alone.

Ad VECTOR VACCINES

From all of the vaccines tested preclinically, those based on Ad vectors have shown superior immunogenicity with regard to induction of CD8⁺ T cells. Ads can be derived from bacterial molecular clones, which allows for easy modifications. The most widely used manipulation of Ad vectors consists of a deletion of the E1 domain, which encodes polypeptides that initiate transcription of other viral genes. E1‐deleted Ad vectors are thus replication‐defective and presumably safer than replication‐competent vectors [ ¹¹⁷ ]. The gene products of the E1 domain are essential for viral replication; E1‐deleted Ad vectors need to be grown on packaging cell lines that provide E1 in trans. Several different packaging cell lines are available. HEK 293 cells contain not only the E1 genes of the AdHu5 virus but also additional sequences flanking the E1 domain [ ¹¹⁸ ]. E1‐deleted AdHu5 vectors grown on HEK 293 cells can undergo homologous recombination with the E1 genes present in HEK 293 cells and thus, revert to RCA. PerC6 cells, developed more recently, only contain the E1 genes of the AdHu5 virus, thus minimizing the risk for homologous recombination and outgrowth of RCAs [ ¹¹⁹ ]. Additional serotypes of primate Ad viruses have been vectored as gene carriers. These viruses were shown to grow in the presence of the AdHu5 E1 gene products [ ⁷⁹ ], or they were modified by replacing early genes [ ¹²⁰ ], or more precisely, the open reading frame 6 of E4 [ ¹²¹ ] with that of the AdHu5 virus or another permissive virus to allow for their propagation in HEK 293 or PerC6 cells.

Another commonly used deletion is that of the E3 domain, which encodes polypeptides that subvert immune response [ ¹²² ]. Gene products of E3 are not essential for viral replication, and E3‐deleted Ad vectors are replication‐competent. Ad vectors deleted in E1 and E3 have also been used. In our hands, E1‐only deleted Ad vectors, and E1‐ and E3‐deleted vectors performed similarly [ ¹²³ ]. This presumably reflects that the E3 genes are poorly transcribed in E1‐deleted vectors and that levels of E3 polypeptides are below the threshold, where they affect a detectable impairment of antigen presentation or cytokine pathways. Deletion of the E3 genes has the advantage to generate more space for transgenes. It also offers the possibility to generate dual expression vectors, in which one transgene and its regulatory elements are placed in the E1 domain, and the other is placed into the E3 domain. VRC has used Ad vectors deleted in E1 and E4, in part to prevent outgrowth of RCAs in HEK 293 cells [ ¹²⁴ ]. In our hands, Ad vectors deleted in E1, E3, and E4 showed lower expression of the transgene product compared with vectors deleted in E1 or E1 and E3, and consequently, they induced overall lower immune responses [ ¹²³ ].

Immunogenicity testing in mice, NHPs, and humans showed that Ad vectors not only induce potent transgene product‐specific CD8⁺ T cells but also that responses are exceptionally sustained [ ⁵³ , ¹²⁵ ], which most likely relates to Ad vectors’ ability to persist, albeit at low levels [ ¹²⁶ ]. During persistence, which occurs mainly in activated T cells, the Ad vectors remain transcriptionally active. This in turn, results in continuous activation of T cells preventing the pronounced contraction of vector‐induced T cells typically seen at the end of the effector phase. Lack of contraction of the Ad vector‐induced immune responses may be advantageous by maintaining high frequencies of effector/effector memory cells. On the other hand, the prolonged presence of fully activated CD4⁺ T cells may pose risks by providing targets for HIV‐1. It may also prevent the timely development of central memory cells that proliferate more effectively upon re‐encounter of antigen, compared with more activated effector or effector memory CD8⁺ T cells.

Humans are infected frequently with Ad viruses throughout their life, and infections with the AdHu5 virus are common. In the United States, depending on the age of the study population and the sensitivity of the assay, 40–60% of humans carry readily detectable neutralizing antibodies to AdHu5 virus [ ¹²⁷ ]. Seroprevalence rates to AdHu5 virus are markedly higher in human populations from undeveloped countries. Pre‐existent neutralizing antibodies to the vaccine carrier can prevent its uptake by cells and thus, transcription and translation of the vaccine antigen, thus reducing the levels of antigen presented to the immune system; consequently, adaptive immune responses to the vaccine antigen are reduced in the presence of vector‐specific neutralizing antibodies [ ⁸⁹ ]. To circumvent impairment of vaccine efficacy by pre‐existing neutralizing antibodies to the vaccine carrier, our group developed vaccine vectors based on AdC [ ¹²⁸ ]. Vectors were derived from five different viruses: AdC1, AdC5, AdC6, AdC7, and AdC68. These represent three different serotypes [ ⁷⁹ , ⁸⁸ , ⁸⁹ , ¹²⁰ ]. AdC7 and AdC5 belong to the same serotype and are cross‐neutralized completely by antibodies. AdC68 and AdC7 are closely related and are cross‐neutralized partially by high‐titer neutralizing antibodies. AdC6 belongs to a distinct serotype. All of these Ad viruses (except for C1) have close sequence homology to the Ad virus of the human serotype 4 and thus, belong to subgroup E of Ad viruses. AdC1 belongs to subfamily B. Ad viruses have been isolated from a multitude of different species; we developed vectors from Ad viruses that naturally infect chimpanzees under the assumption that these viruses would closely resemble human Ad viruses. Indeed, AdC vectors are grouped phylogenetically within human serotype Ads [ ¹²⁹ ]. In addition, sequencing of the AdC vectors showed that their genetic organization is identical to that of human Ads. Further studies showed that their basic characteristics with regard to molecular organization, structure, growth, receptor use, and interactions with cells of the immune system are similar to those of human Ad viruses [ ⁷⁹ , ¹³⁰ ].

E1‐deleted AdC5, AdC6, AdC7, and AdC68 vectors can be grown on HEK 293 packaging cells that carry the E1 of AdHu5 virus. The use of packaging cell lines that carry a heterologous E1 for growth of E1‐deleted AdC vectors has advantages: U.S. Food and Drug Administration‐approved cell lines, such as HEK 293 cells or PerC6 cells, have been generated. PerC‐6 cells carry only the E1 of AdHu5; HEK 293 cells carry additional Ad sequences. The E1‐flanking sequences of the AdC viruses show limited sequence homology with the AdHu5 virus, which virtually abolishes the chance of recombination and reversion to replication competence in HEK 293 cells. Prevalence rates of neutralizing antibodies to AdC viruses are low in the United States or Asia; they are higher in sub‐Saharan Africa, presumably as a result of closer contacts with chimpanzees [ ¹²⁷ ]. In mice and NHPs, the AdC vectors induce high CD8⁺ T cell responses that can be increased by the sequential use of two serologically distinct AdC vectors or by combining AdC vectors with DNA vaccines and poxvirus vectors [ ⁸⁸ , ⁸⁹ , ¹⁰⁹ , ¹³¹ ]. AdC vectors induce significantly stronger immune responses compared with AdHu5 vectors in animals with pre‐existing, neutralizing antibodies to AdHu5.

To avoid the negative effects of pre‐existing neutralizing antibodies to Ad viruses, investigators from Harvard Medical School together with Crucell developed and tested Ad vectors derived from so‐called rare serotypes, such as AdHu26, AdHu48, AdHu35, and AdHu11 [ ¹⁰⁸ , ¹³² ]. Prevalence rates of neutralizing antibodies to these rare serotypes are low in humans residing in the United States and Europe, being higher in African countries [ ¹³³ ]. Immune responses to these Ad vectors were compared with those induced by AdHu5 vectors, which induced higher transgene product‐specific CD8⁺ T cell responses than Ad vectors based on the other human serotypes; however, in AdHu5‐preimmune animals, the AdHu35/AdHu11 combination used in either order outperformed combinations that included the AdHu5 vector [ ¹³² ]. AdHu35 and AdHu11 use CD46 rather than the CAR (used by AdHu5, AdHu4, or most of the AdC vectors) as their cell‐attachment receptor. CD46, unlike CAR, is ubiquitously expressed on cells of NHPs and humans, and in our experience, CD46‐binding vectors, such as the AdC1 vector, are poorly immunogenic compared with Ad vectors that bind to CAR [ ¹²⁰ ]. The Harvard/Crucell group developed a chimeric AdHu35 vector, in which the fiber knob of AdHu35 was replaced with that of AdHu5. This vector induced better transgene product‐specific CD8⁺ T cell responses than the original AdHu35 vector [ ⁹⁰ ]. Nevertheless, the caveat that fiber is a target for Ad virus‐specific neutralizing antibodies should be noted. The Harvard/Crucell group also explored the use of an AdHu5 hexon chimera, in which the variable loops of hexon of AdHu5, which bind most of the virus‐neutralizing antibodies, were replaced with those of AdHu48. The chimeric AdHu5/AdHu48 vector was highly immunogenic, even in animals with pre‐existing immunity to the AdHu5 virus [ ¹³⁴ ]. Nevertheless, despite the low prevalence in the United States, ∼20% of human sera from sub‐Saharan Africa carry neutralizing antibodies against the AdHu48 serotype, precluding the use of AdHu48 vectors in these regions that are most direly in need of a HIV‐1 vaccine. This group also tested AdHu26 vectors and showed impressive results from NHP vaccinated with an AdHu26/AdHu5 prime‐boost regimen [ ⁴² ].

CLINICAL EXPERIENCE WITH VACCINES TO HIV‐1

A number of HIV‐1 vaccines have undergone clinical testing. Early vaccines focused on those expressing the env of HIV‐1 for the induction of neutralizing antibodies. One of those, a monomeric HIV‐1 gp120 vaccine, underwent large‐scale efficacy trials. The vaccine did not prevent infection nor reduce the viral load in infected subjects [ ¹³⁵ ]. Humoral responses were detected, but they failed to neutralize field isolates. Overall, vaccines designed to induce neutralizing antibodies universally failed to induce the desired responses.

T cell‐inducing vaccines such as DNA vaccines, MVA or AAVs, which induced promising results in NHPs, were found to lack immunogenicity in early clinical trials [ ¹¹⁵ , ¹³⁶ , ¹³⁷ ]. Novel delivery methods for DNA vaccines through electroporation of the vector or by combining antigen‐expressing plasmids with plasmids expressing cytokines have been shown to increase the immunogenicity of this vaccine platform in animal models [ ¹³⁸ ], and clinical trials are planned to test if this also translates into better immune responses in human subjects. The initial AAV vectors that were tested clinically were based on serotype 2, which is comparatively poorly immunogenic than other serotypes that have a higher tropism for muscle cells. Discussions are under way to test alternative serotypes such as AAV‐1 in human volunteers. The caveat should be pointed out that in preclinical studies in mice, single‐stranded AAV vectors of different serotypes were found to induce functionally impaired CD8⁺ T cells that were defective in their proliferative capacity and showed a pronounced up‐regulation of immunoinhibitory molecules [ ¹³⁹ ]. Induction of such responses in humans could potentially be harmful. MVA vectors were found to be safe in humans but only elicited barely detectable cellular immune responses to the vaccine antigens in a fraction of the volunteers. Current efforts thus focus on combining MVA with other vaccine modalities such as DNA vaccines. Several of the current and past vaccine trials, focusing on large‐scale trials, are listed in Table 3.

Table 3.

Clinical Trials

Development	Platform	Vaccine/antigen	Immunogenicity/efficacy	References
Phase III, closed	Protein	Protein; gp120	Seroconversion in all participants, no efficacy against HIV‐1 acquisition or viral loads	[ 135 ]
Phase III, enrollment closed	ALVAC/protein	gp120, gag and protease	Study results are not yet available	[ 140 ]
Phase IIB, stopped and unblinded	AdHu5	gag, pol, nef	T cell responses in most participants, no efficacy against HIV‐1 acquisition or viral loads, a trend for increased acquisition in a subgroup	[ 141 ]
Phase II closed	Lipopeptide	Epitopes of gag, pol, nef	T and B cell responses	[ 142 , 143 ]
Phase II closed	Canarypox/protein	env, gag, parts of pol and nef	Neutralizing antibodies to laboratory strains, no T cell responses	[ 144 ]
Phase II, ongoing but recruitment stopped	DNA/AdHu5	env, gag, pol	Cellular immune responses	[ 124 ]
Phase II closed	AAV	gag/pol	Cellular immune responses in a low percentage of participants	No publication available
Phase I closed	DNA/MVA	gag epitopes	T cell responses were infrequent and transient	[ 137 ]
Phase I closed	DNA/MVA	gp160, gag, RT	T cell responses in 95% of trial participants	[ 114 ]
Phase I closed	DNA/protein	gp120 multiple clades, gag	T cell responses and neutralizing antibodies	[ 116 ]
Phase I closed	DNA/NYVAC	env, gag, nef, pol	Priming effect for T and B cell responses by the DNA vaccine	[ 145 ]
Phase I closed	S. enterica	gag	Low level of specific cellular immune responses	[ 112 ]

In humans, Ad vector vaccines induced robust immune responses in phase I/II trials [ ¹²⁴ , ¹⁴⁶ ] and were thus tested in two‐phase IIb clinical efficacy trials, called STEP [ ¹ ] and Phambili trials, in men and women at high risk for HIV‐1 acquisition. Both trials enrolled individuals with low or absent baseline titers of neutralizing antibodies to AdHu5 virus and individuals with moderate to high titers of such antibodies. In early trials, the AdHu5 vector induced detectable T cell responses, even in individuals with pre‐existing, neutralizing antibodies to the vaccine carrier; the STEP and Phambili trials thus did not exclude such individuals. The trials were powered to allow for testing of vaccine efficacy in reducing HIV‐1 acquisition and viral set‐point loads. The STEP trial, which had planned to enroll 3000 subjects, was stopped and unblinded before enrollment was completed, as an interim analysis showed that the vaccine lacked efficacy [ ¹⁴⁷ ]. The Phambili trial was stopped and unblinded shortly thereafter [ ¹⁴⁸ ]. In the STEP trials infections were mostly in men who have sex with men; only one woman in the placebo group had become infected with HIV‐1 at the time of unblinding. As of spring of 2008, in the placebo group, 33/922 men became infected, and in the vaccine group, 49/914 men became infected, suggesting a trend for higher HIV‐1 acquisition in vaccinated individuals. Viral set points initially showed no significant difference between the placebo and the vaccine groups. Further analyses showed that the trend for increased acquisition was seen mainly in men with pre‐existing, neutralizing antibodies to AdHu5 virus, i.e., in men with titers <18; 20/382 in the vaccine group and 20/394 in the placebo group became infected, and in the male group with titers ≥18, 21/392 in the vaccine group and 9/386 in the placebo group became infected. Further differences were seen when data were analyzed according to the circumcision status of men, and those who were vaccinated and not circumcised were at higher risk for acquisition than those who were vaccinated and circumcised. Prevaccination‐neutralizing antibodies to AdHu5 were more common in uncircumcised men [ ¹ ].

The Phambili trial was stopped after enrollment of 801 participants; at the time of unblinding, 11 cases of HIV‐1 infection had been confirmed, with four in the placebo group and seven in the vaccine group. In individuals with neutralizing antibody titers to AdHu5 virus >18, six infections were observed in vaccinated and three infections in unvaccinated individuals. Ten of 11 infections were seen in women.

A lack of efficacy of AdHu5 vector vaccines in humans with pre‐existing, neutralizing antibodies to AdHu5 virus was potentially foreseeable. Lack of efficacy in humans without pre‐existing neutralizing antibodies to AdHu5 virus, in spite of the demonstrated ability of the vaccine to induce a robust transgene product‐specific T cell response [ ¹⁴⁹ ], was unexpected and raised the question of whether T cells can indeed protect against HIV‐1 infections or control viral load in individuals that become infected.

We would argue that the failure of the trials reflects a product rather than a concept failure, i.e., that T cells can protect but that the vaccine failed to induce a T cell response that was of sufficient potency or quality to afford resistance. The vaccine regimen used in the two trials was based on repeated vaccination with an AdHu5 vector. Vaccination with a viral vector not only induces immune responses to the transgene product but also to the vector, and vector‐specific neutralizing antibodies inhibit vector uptake and transgene product expression required for induction of immune responses to the vaccine antigen. Individuals who were seronegative at the time of enrollment produced vector‐specific neutralizing antibodies after the first immunization and thus, developed immunity to subsequent doses of the same vaccine. Homologous prime‐boost regimens with viral vectors using the same vector repeatedly are thus suboptimal compared with heterologous prime‐boost regimens based on the sequential use of different viral vectors. Additional results from the HVTN showed that Ad seronegative‐vaccinated individuals of the STEP trial with MHC genotypes commonly found in elite controllers were able to reduce viral loads significantly better than unvaccinated individuals with similar genotypes. HVTN also showed that in vaccinated AdHu5‐seronegative individuals with these genotypes, known to develop exceptionally potent T cell responses to gag of HIV‐1, viral loads were inversely correlated with the magnitude of the T cell responses to HIV‐1, as tested for by IFN‐γ ELISpot assays [ ¹⁴⁷ ]. These results are based on small numbers of infected individuals, and the trial had not been designed to assess responses in such subgroups. Even with these caveats, the results are suggestive that vaccine‐induced T cells might be able to control an infection with HIV‐1 in individuals that can mount potent CD8⁺ T cell responses to the vaccine antigen.

Additional data are being gathered from the STEP trial. Upon unblinding, data about trends of HIV‐1 acquisition will no longer be interpretable.

The AdHu5 vaccine used by Merck and in combination with DNA vaccine priming by VRC were the first HIV‐1 vaccine regimens used in humans that elicited T cells of sufficient magnitude to allow for an in‐depth analysis of some of the T cells’ pertinent characteristics. In the STEP trial, the vaccine induced T cell responses detectable by ELISpot in ∼80% of vaccinees [ ¹⁴⁶ ]. More than 40% of the subjects with measurable ELISpot responses developed CD4⁺ T cell responses to HIV antigens, and ∼60% developed CD8⁺ T cell responses [ ¹⁴⁹ ]. HIV‐specific CD8⁺ T cells induced by the vaccine produce IFN‐γ or IFN‐γ in combination with TNF‐α. The magnitude and overall composition of T cell responses were similar between vaccinated subjects who acquired HIV infection and those who did not. In additional trials conducted by Merck and VRC, conversion rates and magnitude of CD8⁺ and CD4⁺ T responses were slightly lower in individuals with pre‐existing, neutralizing antibodies to AdHu5. Curiously, this was more pronounced with the VRC than the Merck vaccine [ ¹⁵⁰ ]. The VRC AdHu5 vaccine has a deletion in the E4 domain of Ad, and it has been suggested (but not formally proven) that this deletion would reduce the residual transcription of Ad genes in E1‐deleted vectors and thus, the impact of pre‐existing cellular immunity to AdHu5.

One worrisome aspect of AdHu5 vaccines used by Merck or in combination with DNA vaccine priming by VRC is that both failed to induce a broad T cell response to different epitopes of HIV‐1 in humans. Both vaccines expressed several antigens of HIV‐1, but they only induced, on average, T cells to two or three epitopes [ ¹⁴⁸ , ¹⁵⁰ ]. Such a narrow response would obviously not provide sufficient coverage to the multitude of HIV‐1 variants and would allow for escape through viral mutations. The underlying cause for this narrow response is not yet fully understood. Recent results in NHPs, comparing the breadth of T cell responses with different Ad vector‐based vaccine regimens, showed that the AdHu5 vector, given repeatedly, induced a narrow T cell response, as observed in the STEP trial, and a heterologous prime‐boost regimen, in which an Ad vector derived from a different serotype was used for priming, followed by a booster immunization with an AdHu5 vector, induced a much broader response [ ¹⁰⁸ ]. These results are promising, as they indicate that some prime‐boost regimens may elicit a broad T cell response. Clearly, these results thus far obtained in NHPs need to be confirmed in clinical trials.

The underlying cause for the observed trend toward increased acquisition in individuals with moderate‐to‐high neutralizing antibodies to AdHu5 vectors also remains poorly understood. It has been suggested that the vaccine recalled Ad‐specific CD4⁺ T cells, which provided targets for HIV‐1. Indeed, a similar pathway had been shown to result in enhanced SIV replication in NHPs [ ¹⁵¹ ]. Studies in HIV‐1‐infected patients showed that vaccination with a pneumococcal vaccine or an influenza virus vaccine could cause a transient increase in viral load, suggesting that activation of CD4⁺ T cells might have sponsored replication of HIV‐1 [ ¹⁵² , ¹⁵³ ]. It should be noted though that other studies failed to confirm long‐term effects of influenza virus vaccination on HIV‐1‐infected individuals [ ¹⁵⁴ ]. Increased acquisition as a result of vaccine‐induced activation of Ad‐specific CD4⁺ T cells, which cross‐react between different serotypes of Ad [ ¹⁵⁵ ] and are thus not linked to neutralizing antibody titers to a specific serotype, should have been observed in all vaccines and not just in those with pre‐existing antibodies to AdHu5.

One could postulate that pre‐existing, neutralizing antibodies to the vaccine carrier affected antigen presentation cause a shift of the ratio of vaccine‐induced Th1 versus Th2 cells and in turn, reduced induction of “protective” CD8⁺ T cells. Neutralizing antibodies to Ad are serotype‐specific, and non‐neutralizing‐binding antibodies cross‐react between different serotypes. When we tested human sera for neutralizing and binding antibodies to AdHu5, we were unable to find any correlation in titers of these two types of antibodies. In fact, all adults had binding antibodies to AdHu5 (unpublished), which had presumably been induced in part by other serotypes. Changes in antigen uptake as a result of pre‐existing antibodies to AdHu5 are thus unlikely to explain the trend toward increased HIV acquisition in the STEP trial, as such changes should have been caused by all binding antibodies and not just those that are virus‐neutralizing.

The argument has been made that neutralizing antibodies to AdHu5 were not causative for the trend of increased HIV‐1 acquisition but rather, reflected a yet‐to‐be‐identified condition that caused a heightened status of immune activation, which in turn promoted HIV‐1 acquisition. Again, it is unclear why this should have affected vaccinated individuals primarily. In fact, placebo recipients with pre‐existing, neutralizing antibodies to AdHu5 showed a slight trend toward decreased acquisition of HIV‐1.

HIV‐1 differs from other pathogens in preferentially infecting activated CD4⁺ T cells. Ad vectors, unlike other vaccine carriers that have been tested in advanced trials for HIV‐1, were shown to persist at low levels in a transcriptionally active form in T cells of experimental animals [ ¹²⁶ ], and one would assume that they may also persist in humans. Continued low‐level production of antigen in lymphatic tissues in turn maintains high frequencies of activated T cells and is at least part of the reason why responses to Ad vectors are typically sustained. Continued presence of CD4⁺ T cells activated by the vaccine within the genital or rectal mucosa could provide a risk factor by increasing the number of cells that supports acquisition and replication of HIV‐1. Clinical trials have thus far not assessed the effect of vaccination on mucosal T cell responses, and even in experimental animals, such knowledge remains uncertain.

AdHu5 virus and other human serotypes of Ad virus only replicate in humans (and potentially chimpanzees), which become infected during their lives with a variety of different serotypes. It may thus not be possible to closely mimic the conditions that lead to the trend toward increased HIV‐1 acquisition in the STEP trial in a rhesus macaque model. An AdHu5 virus that can replicate in rhesus macaques has been developed [ ¹⁵⁶ ], but as this virus shows a high degree of attenuation in rhesus compared with wild‐type AdHu5 virus in humans, results obtained by pre‐exposure of macaques with this virus to assess the effects of pre‐existing immunity to AdHu5 in a preclinical model would need to be interpreted with caution.

The overall lesson from the STEP trial is that AdHu5 vector vaccines should not be used in humans with pre‐existing, neutralizing antibodies to the vaccine vector, and as in some regions of Africa, up to 90% of adults have such antibodies, AdHu5 vectors will not be suitable for mass vaccination against HIV‐1.

Several questions remain.

Will it be safe to use other AdHu5 vectors as vaccines for HIV‐1? The argument has been made that the problems encountered in the STEP trial could reflect peculiarities of the Merck vaccine and that other AdHu5 vaccines, such as the VRC vaccine, which has an additional deletion in E4, might not cause an increased risk in HIV‐1 acquisition. It would be unethical to substantiate or refute this argument by clinical trials in individuals at high risk for HIV‐1 acquisition.

Will it be safe to use AdHu5 vectors as vaccine carriers for other pathogens? One would expect that the trend toward increased HIV‐1 acquisition in vaccinated seropositive individuals was caused by the vaccine carrier rather than by the vaccine antigens. AdHu5‐based vaccines expressing antigens of other pathogens and used in a population at risk for HIV‐1 acquisition would thus be expected to cause the same problem as the vaccine used in the STEP trial.

Will it be safe to use other Ad vectors based on common human serotypes? One could make the argument that AdHu5 vectors, which have consistently been the most immunogenic of all human serotype Ad vectors, are fundamentally different. As mentioned previously, without animal data, for which suitable models are not available, it is currently impossible to address this argument, and one should thus view all Ad vectors based on common human serotypes as potentially too risky.

Will it be safe to use a RCA Ad vector based on a common human serotype? Data obtained in preclinical models with replication‐competent human serotype Ad vectors such as AdHu4 have been encouraging. Neutralizing antibodies to AdHu4 can be found in a significant percentage of the U.S. population (>30%), and again, it is currently not possible to assess the potential risks of a RCA vectors based on a common human serotype.

Will it be safe to use Ad vectors based on rare human serotypes or on serotypes from other species? The AdHu5 vectors of the STEP trial did not increase HIV‐1 acquisition rates in AdHu5‐seronegative individuals, and one would thus assume that Ad vectors to which humans have no or only low titers of pre‐existing neutralizing antibodies would be safe.

Will it be safe to use Ad vectors that contain parts of the AdHu5 virus? Ad vectors have been developed in which the hexon loops of the AdHu5 virus, the main targets for neutralizing antibodies, were replaced with those of a rare serotype, i.e., AdHu48. Preclinical data showed that the immunogenicity of this vector was not affected by pre‐existing, neutralizing antibodies to AdHu5 virus. Until it is known to what degree Ad‐specific, neutralizing antibodies induced by experimental exposure in NHPs resemble those induced by natural infections of humans, the safety of such chimeric vectors remains uncertain.

FUTURE OF HIV‐1 VACCINES

In 1983, when HIV‐1 was first shown to be causative for AIDS, the United States Health and Human Services Secretary Margaret Heckler announced that “We hope to have a vaccine [against AIDS] ready for testing in about two years.” In March of 2008, at a summit in Washington, D.C., organized by the National Institutes of Health to assess future HIV‐1 vaccine funding in the wake of the STEP trial results, some activists as well as HIV‐1 scientists asked for an overall stop on all HIV‐1 vaccine research, as in their minds, an HIV‐1 vaccine could never be made, and further attempts were nothing but a waste of taxpayers’ money. So far, many HIV‐1 vaccines have been tested in phase I trials for safety in a limited number of volunteers; few of those were then tested in larger trials for immunogenicity/efficacy. Protein‐based env vaccines failed to induce protection [ ¹³⁵ ]; this was not surprising, as such vaccines failed to induce neutralizing antibodies to field isolates. Results from another phase III trial, which combines a poxvirus vector with a protein boost, are not yet available. AdHu5 vectors, which by many were viewed as the most promising vaccine platform for HIV‐1 prevention, failed. A large trial, called PAVE‐100, initially designed to test a prime‐boost regimen using a DNA vaccine prime, followed by an AdHu5 boost in individuals at high risk for HIV‐1 acquisition in several countries, was reduced to a smaller trial to be conducted in AdHu5‐seronegative individuals in the United States only [ ¹⁵⁷ ]. Although the initial trial was powered to assess effects of vaccination on viral loads and HIV‐1 acquisition rates, the scaled‐down trial would only enroll a sufficiently high number of individuals to allow an assessment of vaccination on viral loads. In early stage clinical trials, the PAVE‐100 vaccine composition induced T cell responses that were comparable with those induced by the STEP trial vaccine. Nevertheless, as correlates of protection remain undefined, it is felt that testing of the vaccine in AdHu5‐seronegative individuals is justified, as it might yield information that could aid future vaccine development. It is not expected that the trial will lead to a clinical product, as the use of AdHu5 vectors for HIV‐1 prevention is not justifiable in light of the high prevalence rates of AdHu5‐neutralizing antibodies in humans. Other vaccines are in early‐stage clinical testing or still in preclinical development, so currently, the prospects for an efficacious HIV‐1 vaccine in the near future are rather dim.

Before dismissing future research about HIV‐1 vaccines, one should examine other infectious diseases that take a large global toll on human lives. It took longer, by far, than 30 years to develop efficacious vaccines for measles or pertussis, and after 40 years of research, there is still no vaccine for malaria, a vector‐borne disease caused by protozoan parasites. Each year, there are ∼515 million cases of malaria causing 1–3 million deaths. It was shown as early as 1967 that immunization with live, radiation‐attenuated sporozoites provides protection to rodents, suggesting that this disease could be prevented by vaccination [ ¹⁵⁸ ]. Nevertheless, in spite of numerous clinical trials, a commercial vaccine to malaria is not currently available. Yet no one is suggesting that attempts to generate such a vaccine should be abandoned.

Although many have called the STEP trial a failure, as it did not show vaccine efficacy, it was in all fairness a success in its execution and the scientific knowledge that could be gained from it. Reduction in viral load in the small group of elite controllers that became infected in the STEP trial suggests, with the caveat that the number of such individuals was small, that CD8⁺ T cell vaccines, provided that they induce responses of sufficient potency, might be able to induce protection against high viral loads. Although factors other than magnitude of the vaccine‐induced CD8⁺ T cell response, such as CD8⁺ T cell functionality, differentiation status, migration patterns, proliferative potential, etc., may affect the efficacy of a T cell‐inducing HIV‐1 vaccine, optimizing vaccine‐induced CD8⁺ T cell frequencies is likely to be crucial. In HIV‐1 vaccine development, most investigators have restricted their efforts to a single vaccine carrier or to related vaccine carriers, such as different poxvirus vectors or Ad vectors derived from different serotypes, which in some instances, are being combined with DNA vaccines. More potent responses can be achieved, at least in animals, through a combination of different platforms, such as poxvirus vectors with Ad vectors or even DNA vaccines with poxvirus vectors and Ad vectors. Combining different platforms, although technically straightforward, poses logistic challenges, as it not only requires intense collaborations of scientists but also acceptance by the commercial entities that own the different vaccine platforms.

SUMMARY

Not too long ago, someone jokingly told us that American settlers, when under attack, formed a wagon circle and shot out of this circle—that's how the West was won; scientists, when under attack, also form a circle but then shoot each other. Clearly, this is not a time to divide and conquer but rather, to join forces and continue our efforts to develop a safe and effective vaccine against HIV‐1.