WO2009065800A1 - Recombinant human adenoviruses for eliciting mucosal immune responses - Google Patents

Abstract

The present invention relates to methods and means for obtaining and using adenoviral-based vaccines against viral infections. The invention in particular relates to recombinant replication-defective adenoviruses, preferably those based on human adenovirus serotype 26 (rAd26) and those based on human adenovirus serotype 5 (Ad5) or those based on human adenovirus serotype 5 carrying the HVR regions of the hexon protein of adenovirus serotype 48 (rAd5HVR48) in heterologous prime-boost regimens and in the induction of mucosal immune responses.

Description

TITLE

Recombinant human adenoviruses for eliciting mucosal immune responses

FIELD OF THE INVENTION

The invention relates to the field of medicine and in particular to the use of gene delivery vehicles such as low- neutralized recombinant adenoviruses in eliciting immune responses, preferably mucosal immune responses, in mammalian subjects .

BACKGROUND OF THE INVENTION

Replication-incompetent, recombinant adenovirus serotype 5 (rAd5) vectors are currently being developed as candidate vaccines for both HIV-I and other pathogens (Catanzaro et al. 2006; Shiver and Emini 2004; Thorner and Barouch 2007) . However, it has recently become clear that rAd5 vectors expressing HIV-I Gag, Pol, and Nef antigens failed to protect against HIV-I infection or to reduce viral replication following infection in a phase 2b efficacy study. Although the exact reasons for this failure have not yet been determined, it is clear that more potent or different phenotypes of immune responses will be required to afford protection against HIV-I, when using adenovirus-based vaccines .

A variety of recombinant adenoviruses based on serotypes other than Ad5 have been generated. These new vectors facilitate the development of more potent and beneficial heterologous rAd prime-boost regimens. Since many of these vectors are based on low-neutralized (or ^Λrare' ) serotypes, they also allow circumventing pre-existing anti- Ad5 immunity (WO 00/70071; WO 2004/037294; WO 2004/083418; US 2004/0191222 Al; WO 2006/040330; WO 2007/104792; Abbink et al. 2007; Barouch et al. 2004; Lemckert et al. 2006; Nanda et al . 2005; Reyes-Sandoval et al. 2004; Roberts et al. 2006; Vogels et al . 2003) . Although it is now known which serotypes encounter what percentages of neutralizing activity in what selected populations and how these percentages can be calculated based on neutralizing activity assays available in the art, the phenotype of cellular immune responses elicited by the low-neutralized rAd vectors and the immunogenicity of heterologous rAd prime-boost regimens are generally unknown: What exact immune responses, cellular and humoral, are elicited and to what extent? Since the different adenovirus serotypes have different tropisms and are each capable of infecting different ranges of host cells at different efficiencies, it is of great interest to know what precise qualitative immune responses (antibody/cellular) are actually induced by introduction of the different recombinant vectors in the host. Depending on the immune response, one would subsequently be able to select the right serotype for the required needs.

Replication-defective rAdll, rAd35, and rAd50 from adenovirus subgroup B, as well as replication-defective rAd26, rAd48, and rAd49 from adenovirus subgroup D have all been constructed to date (see WO 00/70071; Abbink et al. 2007) . These rAd vectors exhibit a low seroprevalence in human populations in sub-Saharan Africa, and many other selected populations around the world, and proved immunogenic in both mice and rhesus monkeys. In particular rAd26 proved highly immunogenic and was selected for further development into clinical trials as a carrier candidate in HIV-I vaccines. However, as it was known that an initial infection (priming vaccination or a common infection earlier in life with a wt virus) with a certain serotype would negatively influence a subsequent infection with that same serotype, the general thought in the art is that a heterologous prime-boost set-up in which an infection (or vaccination) with one serotype should be followed with a second (or boosting) infection with another (unrelated) adenovirus serotype to avoid the neutralizing effects of antibodies directed against the earlier serotype. In other words, an initial infection with an Ad26-based recombinant anti-HIV vaccine virus for treating an HIV-I infection should not be followed by the same Ad26-based recombinant virus, but should be followed by a virus from another serotype. It has been described in the art that heterologous rAd prime-boost regimens involving serologically distinct rAd vectors could be highly immunogenic in mice (Abbink et al. 2007; Lemckert et al. 2005; Pinto et al. 2003; Thorner et al. 2006) . Nevertheless, specific combinations with rAd26 in either the priming or boosting composition were still unknown. However, more specific and more beneficial prime- boost regimens could not be excluded and in view of the countless possibilities in such prime-boost set-ups, it would be beneficial to provide a combination that would outperform combinations already tested and envisioned.

Besides obtaining improved immune response phenotypes and improved prime-boost regimens, also obtaining knowledge about the route of vaccination is of great importance in the generation of new vaccines. It is generally thought that mucosal immunity is essential for medicaments such as HIV-I vaccines, not only because the genital and rectal mucosa represent the primary portals of virus entry but also because the gastrointestinal mucosa is the predominant site of destruction of memory CD4+ T-lymphocytes during acute infection (Veazey et al. 1998; Brenchley et al . 2004; Li et al. 2005; Mattapallil et al . 2005). Therefore, the ability to generate potent mucosal immune responses is also an important goal of vaccine development. A recent study in nonhuman primates has correlated vaccine efficacy with preservation of mucosal memory CD4+ T-lymphocytes after SIV challenge (Mattapallil et al. 2006). Studies in rhesus monkeys and humans have suggested that cellular immune responses will likely prove critical in controlling SIV and HIV-I infection, and thus vaccines that induce potent, durable and protective mucosal cellular immunity are of particular interest. However, the capacity of systemic vaccination to elicit mucosal cellular immune responses has not previously been well characterized.

Most current HIV-I vaccine candidates exclusively utilize systemic immunization strategies given the logistic and potential safety concerns associated with the delivery of recombinant vaccine vectors by mucosal routes. It has been widely believed, however, that potent mucosal immunity would be difficult to generate by systemic vaccination as a result of the anatomic and functional distinctness of the systemic and mucosal immune systems and the resultant compartmentalization of immune responses. Anatomic compartmentalization of cellular immune memory has been shown to be dependent on the initial site of antigen exposure in both humans and animal models of localized infections or malignancies. Similarly, anatomically biased recall responses have been observed in certain models of vaccination. In addition, dendritic cells isolated from specific anatomic sites have been reported to influence patterns of chemokine receptor and integrin expression and tissue homing specificities of primed CD8+ T-lymphocytes . In contrast, global CD8+ T-lymphocyte responses in both systemic and mucosal compartments have been observed in models of localized infection with rotavirus, Sendai virus and Listeria . However, the ability of systemic vaccination to overcome immune compartmentalization and to generate potent and protective mucosal cellular immune memory has not been elucidated in detail.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig 1 shows the immunogenicity of heterologous rAd prime- boost regimens in mice. (A) Priming with 10 vp rAd26-Gag. (B) Priming with 10⁹ vp rAd48-Gag. Arrows indicate days of immunization. Gag-specific cellular immune responses were assessed by D^b/AL11 tetramer binding assays. Mean responses with standard errors are shown.

Fig 2 shows the directionality of heterologous rAd prime- boost regimens in mice. (A) Naive mice were primed on day 0 and were boosted on day 28 with 10 vp rAd26-Gag, rAd5-Gag, or rAd5HVR48-Gag . (B) Idem, now with the mice being pre- immunized with two injections of 10¹⁰ vp rAd5-Empty (Ad5 neutralizing antibody titers of 8,192-16,384) . Fig 3 shows the immunogenicity of rAd vectors in rhesus monkeys. (A) CD4- or CD8-depleted IFN-γ ELISPOT assays in samples from adult animals immunized with a single injection of 10¹¹ vp rAd5-Gag, rAd26-Gag, rAd48-Gag, or rAd49-Gag. (B) Secretion of IFN-γ, TNF-α, and IL-2 in gated CD8+ T lymphocytes by intracellular cytokine staining (ICS) assays. (C) Idem, now gated for CD4+ T lymphocytes. Fig 4 shows the immune response phenotype of rAd5 and rAd26 vectors in rhesus monkeys. (A) Cytokine secretion phenotypes of CD8+ T lymphocytes elicited by rAd5-Gag. (B) Idem with CD4+ T lymphocytes. (C) Cytokine secretion phenotypes of CD8+ T lymphocytes elicited by rAd26-Gag. (D) Idem with CD4+ T lymphocytes. Values were determined by multiparameter flow cytometry. 8-color ICS assays were performed using methods and means described in the examples. Gag-specific absolute (bar graphs; left) and proportional (pie charts; right) IFN- γ, TNF-α, and IL-2 responses in all combinations are depicted in CD8+ and CD4+ central memory (CM) and effector memory (EM) T lymphocyte subpopulations .

Fig 5 shows the immunogenicity of heterologous rAd prime- boost regimens in rhesus monkeys over time. (A) Comparison between rAd5/rAd5, rAd26/rAd5, rAd48/rAd5 and rAd49/rAd5. (B) Comparison between rAd26/rAd5, rAd5/rAd26, rAd35/rAd5 and rAd5/Ad35. The prime rAd26 - boost rAd5 is the same in both panels. Fig 6 shows the breadth of Gag-specific cellular immune responses elicited by the heterologous rAd26/rAd5 regimen in rhesus monkeys. (A) IFN-γ ELISPOT responses at 4 weeks upon prime and (B) responses at 4 weeks upon boost. Fig 7 shows the gating and phenotype of lymphocytes isolated from systemic and mucosal anatomic compartments. (A) Representative analysis of ALIl tetramer-positive CD8+ T- lymphocyte responses in mice spleen on day 14 following i.m. immunization with rAd5-Gag. The memory phenotype of CD8+ T- lymphocytes was determined by expression of CD44, CD62L and CD127. (B) Lymphocytes from naive, unvaccinated mice were isolated from selected anatomic sites and stained with mAbs against CD3, CD4 and CD8. (C) The memory phenotype of total CD8+ T-lymphocytes from each anatomic site in naive mice determined by CD44 and CD62L expression using the gating algorithm shown in (A) .

Fig 8 shows urine systemic and mucosal cellular immune responses after i.m. immunization with rAd5-Gag. (A) Mice were immunized i.m. with rAd5-Gag, and CD8+ T-lymphocyte responses were followed over a 24-week time course in multiple anatomic compartments. (B) Memory or effector phenotype of CD8+ T-lymphocytes elicited by i.m. vaccination with rAd5-Gag. Effector (black bars), effector memory (white bars) and central memory (gray bars) phenotypes were assessed in CD8+ T-lymphocytes at week 2, 12 and 24 after vaccination using the gating algorithm shown in Fig 7. (C) Mice were immunized with rAd5-Gag, and the production of IFN-γ (top panel) and IL-2 (bottom panel) in response to pooled overlapping 15 aa peptides spanning the SIV Gag protein (black bars) or the individual ALIl epitope peptide (white bars) were assessed on day 14 after vaccination by ICS assays. Grey bars represent control stimulations without peptide.

Fig 9 shows the systemic and mucosal cellular immune responses to rAd viruses administered alone or in heterologous prime-boost regimens. (A) CD8+ T-lymphocyte responses to various viruses were examined in systemic and mucosal anatomic compartments in mice at week 2 following i.m. immunization with rAd5-Gag (black bars), rAd5HVR48-Gag (white bars), rAd26-Gag (dark gray bars) or rAd35-Gag (light gray bars). (B) Idem, at week 12. (C) CD8+ T-lymphocyte responses to rAd5HVR48-Gag were evaluated at week 2 to compare primary and recall responses after i.m. immunization in naive mice (white bars) or in mice previously primed 6 weeks earlier with rAd26-Gag (black bars) . (D) Idem, at week 12. (E) Mice were primed i.m. at week 0 with rAd26-Gag and boosted at week 8 with either rAd26-Gag (homologous vector; dashed lines) or rAd5HVR48-Gag (heterologous vector; solid lines). CD8+ T-lymphocyte responses were assessed at multiple time points following the priming and boosting immunizations. ILN = inguinal lymph nodes; MLN = mesenteric lymph nodes; PP = Peyer's patches; SB = small bowel; LB = large bowel; IEL = intraepithelial lymphocytes; LPL = lamina propria lymphocytes; VT = vaginal tract. Fig 10 shows the systemic CD8+ T-lymphocytes migration to multiple mucosal compartments after adoptive transfer. (A) Mice primed at t=0 i.m. with rAd26-Gag and boosted at week 6 with rAd5HVR48-Gag. CD8+ T-lymphocytes were purified from splenocytes by negative immunomagnetic selection on day 10 following the boost; 2xlO⁷ purified lymphocytes were injected intravenously (i.v.) into naive recipient mice. Tissue distribution of transferred CD8+ T-lymphocytes was determined at week 2 after adoptive transfer. (B) The effector and memory phenotypes were determined for CD8+ T- lymphocytes both prior to transfer and at week 2 post- transfer. (C) Idem, with respect to the pattern of mucosal homing marker expression. (D) Congenic adoptive transfer studies involved injection of 10⁷ purified CD8+ T- lymphocytes isolated from the spleens of naive Ly5.1+ donors into naive congenic Ly 5.2+ recipients. The tissue migration pattern of un-stimulated transferred CD8+ T-lymphocytes (expressed as a ratio of transferred/native CD8+ T- lymphocytes in each tissue) was determined on day 14 post- transfer. (E) Recipient mice were vaccinated i.m. with rAd5- Gag on day 2 post-transfer and the ratio of transferred/native responding ALll-specific CD8+ T- lymphocytes was determined at each anatomic site on day 14. Fig 11 shows the mucosal protection of mice from intranasal (i.n.) vaccinia challenge by i.m. heterologous rAd prime- boost vaccination. Mice were vaccinated with the rAd26/Ad5HVR48 prime-boost regimen described in Fig 9E . At week 8 following the boost, vaccinated mice and control mice were challenged i.n. with 10⁸ PFU rVaccinia-Gag . (A) Control mice were monitored for weight loss. (B) Idem, for vaccinated mice. Animals were sacrificed after losing >20% of their initial body weight (asterisks) . (C) Pre-challenge and post-challenge CD8+ T-lymphocyte responses in blood for control and vaccinated mice. (D) CD8+ T-lymphocyte responses in mucosal and systemic compartments were compared pre- challenge (white bars) and on day 12 post-challenge (black bars) in vaccinated animals. Fig 12 shows the mucosal cellular immune memory in rhesus monkeys after i.m. rAd vaccination. Two rhesus monkeys expressing the MHC class I allele Mamu-A*01 (animals 184-03, black bars, and 210-03, grey bars) and one monkey negative for this allele (153-03, white bars) were vaccinated i.m. with rAd5HVR48-Gag. (A) Gag pile-specific CD8+ T-lymphocyte responses were evaluated in systemic and mucosal compartments at week 4 following vaccination. The memory phenotype of the responding lymphocytes was determined by CD28 and CD95 expression. (B) Idem, at week 32. (C) Idem, at week 52.

SUMMARY OF THE INVENTION

The present invention relates to a kit of parts comprising a priming vaccine composition comprising a recombinant replication-defective adenovirus of serotype 26 (rAd26) comprising a nucleic acid of interest; and a boosting composition comprising a recombinant replication- defective adenovirus of serotype 5 (Ad5), or a recombinant replication-defective adenovirus of serotype 5 comprising hexon proteins wherein the HVR regions of the serotype 5 adenovirus have been replaced by the corresponding HVR regions of adenovirus 48 (rAd5HVR48) .

The invention also relates to a method for inducing mucosal immune response against an antigen in a mammalian subject, comprising the steps of obtaining a recombinant replication-defective adenovirus comprising a nucleic acid of interest encoding said antigen; and administering said adenovirus systemically into said mammalian subject, preferably followed by the step of administering a second recombinant replication-defective adenovirus comprising said nucleic acid of interest, wherein said first adenovirus is a recombinant replication-defective adenovirus serotype 26 (rAd26) .

The invention further relates to a method of inducing the migration of activated CD8+ T-lymphocytes from systemic to mucosol immune compartments in a mammalian subject, comprising the steps of obtaining a recombinant replication- defective adenovirus comprising a nucleic acid of interest encoding an antigen; and administering said adenovirus systemically into said mammalian subject, wherein said adenovirus is a recombinant replication- defective adenovirus serotype 26 (rAd26) .

DETAILED DESCRIPTION

The phenotype of cellular immune responses elicited by rAd5 and rAd26 vectors were compared and the immunogenicity of various heterologous rAd prime-boost regimens in rhesus monkeys was assessed. It was observed that cellular immune responses elicited by rAd5-Gag and rAd26-Gag differed not only in magnitude but also in phenotype. Quite unexpectedly, rAd26 vectors induced more balanced CD8+ and CD4+ T lymphocyte responses and more polyfunctional cytokine secretion responses as compared with rAd5 vectors. Cellular immune responses primed by rAd26-Gag were boosted remarkably effectively by rAd5-Gag in both mice and rhesus monkeys, suggesting that qualitative aspects of T lymphocyte responses may prove critical in determining the overall potency and directionality of heterologous rAd prime-boost regimens .

Here, it is disclosed that rAd5-Gag and rAd26-Gag elicit qualitatively quite distinct T lymphocyte responses in mammalian subjects. In particular, rAd5 vectors elicit skewed CD8+ > CD4+ T lymphocyte responses that are characterized primarily by IFN-γ+, TNF-α+, and IFN-γ+/TNF-α+ cells. In contrast, rAd26 vectors elicit lower CD8+ T lymphocyte responses but significantly more balanced CD8+ and CD4+ responses that included a higher proportion of IL- 2+ and polyfunctional IFN-γ+/TNF-α+/IL-2+ cells. rAd5 and rAd26 vectors utilize different cellular receptors. Now, it appears that not only the magnitude but also the phenotype of cellular immune responses generated by adenoviral-based vaccine vectors is important in determining their practical utility. In particular, qualitative differences among various rAd serotypes most likely underlie the observed potency and directionality of heterologous rAd prime-boost regimens. Probably, the efficiency of rAd26-Gag priming reflects the ability of this vector to elicit balanced, polyfunctional CD8+ and CD4+ T lymphocyte responses (Fig 4A-B) . Similarly, the potency of rAd5-Gag boosting likely reflects the capacity of this vector to drive potent IFN-γ+ CD8+ T lymphocyte responses (Fig 4C-D) . Consistent with this, cellular immune responses elicited by the optimal heterologous rAd26/rAd5 prime-boost regimen were remarkably high in magnitude (Fig 5) and exhibited substantial breadth (Fig 6) .

Prior studies have shown that heterologous rAd prime- boost regimens elicited potent cellular immune responses in mice. The data disclosed herein confirm and extend such prior observations by assessing the detailed phenotype of CD8+ and CD4+ T lymphocyte responses elicited by rAd5 and rAd26 vectors and by comparing the immunogenicity of various heterologous rAd prime-boost regimens in primates. Surprisingly, two regimens were found to provide unexpected good results: The rAd26 prime - rAd5 boost regimen proved optimal in the absence of pre-existing immunity, while the rAd26 prime - rAd5HVR48 boost regimen turned out to be the optimal choice where pre-existing immunity against Ad5 was present .

Polyfunctional T lymphocyte responses elicited by gene delivery vehicles such as recombinant adenoviruses comprising an adenoviral genome comprising a gene of interest, seem to be highly relevant. Betts et al. (2006) have shown that polyfunctional HIV-1-specific CD8+ T lymphocyte responses could be correlated with control of viral replication in HIV-1-infected patients and with clinical non-progression. Moreover, highly effective vaccines such as vaccinia virus have been demonstrated to induce polyfunctional CD8+ T lymphocyte responses in humans (Precopio et al. 2007). In addition, polyfunctional CD4+ T lymphocyte responses have been shown to be required for optimal protection against Leishmania major challenges in mice (Darrah et al. 2007) . It is therefore concluded that the generation of polyfunctional CD8+ and CD4+ T lymphocyte responses in an adenovirus-based vaccine, preferably in an HIV-I vaccine, may be highly desirable. Hence, a suitable recombinant vector, or combination of vectors in a heterologous prime-boost regimen that may fulfill such needs would be therefore be highly beneficial.

It is known in the art that there is a critical importance of CD4+ T cell help in determining the overall functionality of CD8+ T lymphocyte responses. In particular, "helped" CD8+ T lymphocytes primed in the presence of adequate CD4+ T cell help are able to expand much more efficiently following subsequent boost immunizations as compared with "helpless" CD8+ T lymphocytes primed without CD4+ T cell help (Janssen et al. 2003; Shedlock and Shen 2003; Sun and Bevan 2003) . Consistent with this model, the efficient priming by rAd26 vectors is likely related to the balanced CD8+ and CD4+ T lymphocyte responses elicited by this vector, although the precise degree of T cell help required for optimal CD8+ T lymphocyte function remains yet to be determined.

The present studies demonstrate that priming with one rAd vector and boosting with a serologically distinct rAd vector can elicit remarkably potent and broad cellular immune responses in primates, as long as the priming and boosting compositions are carefully selected and in concert with one another. It is further concluded that not only the magnitude but also the phenotype of cellular immune responses is highly important in determining the overall potency and directionality of heterologous rAd prime-boost regimens . As disclosed herein, also the magnitude, kinetics, phenotype and durability of mucosal cellular immune responses in mice and rhesus monkeys after systemic immunization with rAd viruses were studied. Patterns of lymphocyte trafficking and mucosal homing marker expression by antigen-specific CD8+ T-lymphocytes following vaccination, were assessed. It is shown that activated CD8+ T-lymphocytes, but not quiescent CD8+ T-lymphocytes, rapidly migrate from systemic to mucosal immune compartments following vaccination to generate potent and durable mucosal immune responses. In addition, systemic vaccination with heterologous rAd prime-boost regimens expressing SIV Gag effectively protected mice from lethal mucosal challenge with recombinant vaccinia virus expressing the same antigen. These results have broad implications for vaccination strategies aimed at inducing mucosal immunity.

Circulating systemic T-lymphocytes typically exhibit a limited capacity to traffic to mucosal sites, and such anatomic compartmentalization has been considered a major barrier in generating effective mucosal cellular immune responses by systemic vaccination. Herein, it is demonstrated that intramuscular vaccination with rAd vectors expressing SIV Gag elicited potent, durable and protective CD8+ T-lymphocyte memory at multiple mucosal sites in mammals, such as mice and primates. Following rAd vaccination, systemic CD8+ T-lymphocytes trafficked rapidly to mucosal sites, up-regulated mucosal homing integrins and chemokine receptors, and adopted memory phenotypes characteristic of resident mucosal T-lymphocytes. These data indicate that systemic CD8+ T-lymphocytes elicited by intramuscular vaccination exhibit substantial plasticity and have the capacity to overcome immune compartmentalization to induce global systemic and mucosal cellular immune responses in multiple tissues.

These findings are particularly significant for the development of anti-viral vaccine candidates, preferably those directed against HIV-I infections. After mucosal transmission, HIV-I replicates locally within the genital or rectal mucosa and associated lymphoid tissues before disseminating. Potent mucosal immune responses at the portal of virus entry might therefore be able to alter dynamics between the host and the virus. Moreover, peak viral replication in acute HIV-I infection is accompanied by a massive, irreversible destruction of memory CD4+ T- lymphocytes, particularly in gastrointestinal mucosal tissues. HIV-I vaccination strategies that drive potent mucosal cellular immune responses may therefore be critical. The recent failure of a rAd5-based HIV-I vaccine in clinical trials suggests that substantially more potent mucosal immune responses may be required. It is now demonstrated that certain heterologous rAd prime-boost regimens generate potent secondary cellular immune responses in multiple mucosal tissues that are significantly superior to those induced by homologous rAd regimens, which are substantially limited by anti-vector neutralizing antibodies induced by the priming immunization. These findings highlight a novel and potentially important benefit of heterologous rAd prime- boost regimens for protecting against mucosal pathogens, as exemplified here utilizing a mucosal vaccinia challenge model . After i.m. rAd vaccination, antigen-specific CD8+ T- lymphocytes acquired the capacity to traffic from systemic to mucosal compartments, whereas resting CD8+ T-lymphocytes lacked this capacity. In addition, vaccine-elicited systemic CD8+ T-lymphocytes that migrated to the gastrointestinal tract acquired an effector/effector memory phenotype typical of resident mucosal CD8+ T-lymphocytes and expressed integrins and chemokine receptors critical for mucosal homing. Thus, systemic CD8+ T-lymphocytes appear to have remarkable migratory and phenotypic plasticity following systemic rAd vaccination, providing a likely mechanism for the potent mucosal cellular immune responses observed. It is unlikely that vector dissemination and direct priming of anatomically isolated T-lymphocyte populations in mucosal tissues played a significant role, since bio-distribution studies with rAd vectors showed no detectable vector dissemination to mucosal sites (data not shown) and since adoptively transferred splenocytes from vaccinated animals exhibited broad migratory properties in naive hosts in the absence of further antigenic stimulation. The phenotypic plasticity observed here extends previous observations of the differentiation of memory CD8+ T-lymphocytes in systemic and mucosal microenvironments (Masopust et al. 2006; Marzo et al. 2007) and demonstrates for the first time the functional relevance of this phenomenon for vaccine-elicited mucosal immunity.

Multiple pre-clinical and clinical studies of localized infection and vaccination using different vaccine modalities have shown anatomic skewing of primary and recall responses to the initial site of antigen exposure. In contrast, clear evidence is provided herein that systemic vaccination with rAd vectors can elicit rapid, potent and durable mucosal cellular immune responses in both mice and primates with magnitudes that are comparable with systemic cellular immune responses. These data show that CD8+ T-lymphocytes may acquire either local or global patterns of tissue migration as a result of specific properties of the innate and early adaptive immune responses. Prior reports have demonstrated anatomically widespread cellular immune responses in murine models of rotavirus, Sendai virus and Listeria infection (Masopust et al. 2001 and 2004), indicating that the findings provided herein are presumably not limited to rAd vaccine vectors.

These data have major implications for vaccine development against mucosal pathogens. It is clear that systemic immunization with rAd vectors, and in particular heterologous prime-boost regimens involving rAd viruses, effectively overcomes immune compartmentalization and generates potent, durable and protective mucosal cellular immunity. These findings imply that mucosal routing of vaccines may not necessarily be required for protecting against mucosal pathogens.

The present invention relates to a kit of parts comprising: a priming vaccine composition comprising a recombinant replication-defective adenovirus of serotype 26 (rAd26) comprising a nucleic acid of interest; and a boosting composition comprising a recombinant replication- defective adenovirus of serotype 5 (Ad5) , or a recombinant replication-defective adenovirus of serotype 5 comprising hexon proteins wherein the HVR regions of the serotype 5 adenovirus have been replaced by the 7 corresponding HVR regions of adenovirus 48 (rAd5HVR48) (as disclosed in WO 2006/040330) . Preferably, said compositions in said kit further comprise a pharmaceutically acceptable excipient. Pharmaceutically acceptable carrier or excipients are well known in the art. In a preferred embodiment, said nucleic acid of interest encodes at least one antigen, or an immunogenic part of said antigen, of a simian or human immunodeficiency virus.

The invention also relates to a method for inducing mucosal immune response against an antigen in a mammalian subject, comprising the steps of: obtaining a recombinant replication-defective adenovirus comprising a nucleic acid of interest encoding said antigen; and administering said adenovirus systemically into said mammalian subject, wherein said adenovirus is a recombinant replication-defective adenovirus serotype 26 (rAd26) . Preferably, said mammalian subject is a monkey or a human. In a preferred embodiment, said antigen is a simian or human immunodeficiency virus antigen, or an immunogenic part thereof. In yet another preferred embodiment of the methods of the present invention, said adenovirus is administered intramuscularly (i.m.) . In a preferred embodiment of the method of the present invention, said step of administering said adenovirus is followed by the step of: administering a second recombinant replication-defective adenovirus comprising said nucleic acid of interest, wherein said second adenovirus is heterologous to the earlier (first) administered adenovirus serotype. Preferably, said second adenovirus is rAd5 or rAd5HVR48. The present invention also relates to a method of inducing the migration of activated CD8+ T-lymphocytes from systemic to mucosol immune compartments in a mammalian subject, comprising the steps of: obtaining a recombinant replication-defective adenovirus comprising a nucleic acid of interest encoding an antigen; and administering said adenovirus systemically into said mammalian subject, wherein said adenovirus is a recombinant replication-defective adenovirus serotype 26 (rAd26) . Preferably, said mammalian subject is a monkey or a human. Also preferably, said antigen is a simian or human immunodeficiency virus antigen, or an immunogenic part thereof. In a preferred embodiment, said step of administering said adenovirus is followed by the step of: administering a second recombinant replication- defective adenovirus comprising said nucleic acid of interest, wherein said second adenovirus is heterologous to the earlier (first) administered adenovirus serotype. Even more preferred is a method according to the present invention, wherein said second adenovirus is rAd5 or rAd5HVR48.

EXAMPLES

Example 1. Methods and assays

Replication-incompetent, El/E3-deleted rAd5, rAd35, rAd26, rAd48, and rAd49 vectors expressing SIVmac239 Gag were prepared as previously described (Abbink et al. 2007; Lemckert et al. 2006; Vogels et al. 2003) . The rAd5HVR48-Gag was made and produced essentially as described in WO 2006/040330.

C57BL/6 and B6. SJL-Ptprc^a Pepc^b/BoyJ mice were obtained from Jackson Laboratories. Six to eight week-old C57BL/6 mice were injected intramuscularly (i.m.) with 10 vp replication-incompetent rAd vectors expressing SIVmac239 Gag in 100 μl sterile PBS in both quadriceps muscles. To induce anti-Ad5 immunity, mice were pre-immunized twice separated by a 4-week interval i.m. with 10¹⁰ vp rAd5-Empty. Ad5 neutralizing antibody (NAb) titers were determined by luciferase-based virus neutralization assays (Sprangers et al. 2003; WO 2004/037294) . Adult rhesus monkeys (Macaca mulatta) that did not express the MHC class I allele Mamu- A*01 (determined by PCR and sequencing; Kuroda et al. 1999) were immunized i.m. with 10¹¹ vp replication-incompetent rAd viruses expressing SIVmac239 Gag in 1 ml sterile PBS containing 5% sucrose in both quadriceps muscles. All animal studies were approved by the Institutional Animal Care and Use Committees (IACUC) .

Tetramer binding assays were essentially performed as described (Altman et al. 1996; Barouch et al. 2004), wherein tetrameric H-2D^b complexes folded around the immunodominant SIV Gag ALIl epitope were prepared and utilized to stain peptide-specific CD8+ T lymphocytes. CD8+ T lymphocytes from naϊve mice were utilized as negative controls and exhibited <0.1% tetramer staining at all anatomic sites. Monoclonal antibodies used in multiparameter flow cytometry were purchased from BD Biosciences (CD44-FITC (IM7), TCRγδ-FITC (GL3), β7 integrin-FITC (M293), CD4-PE or Pacific Blue (L3T4), CD8α-PerCP-Cy5.5 (53-6.7) and CD3-APC (145-2ClIl)) and eBioscience (CD127-PE-Cy7 (A7R34), CD62L-APC- AlexaFluor750 (Mel-14), CD45.1-PE-Cy7 (A20), CD45.2 APC- AlexaFluor750 (104), CD3-AlexaFluor 700 (17A2), CD103-FITC (2E7) and CCR9-FITC (eBioCW-1.2 ) ) . LIVE/DEAD Fixable Violet was used for vital dye exclusion in flow cytometric assays according to the manufacturer's instructions (Invitrogen) . For rhesus monkey tetramer binding assays, Mamu-A*01 tetramers labeled with phycoerythrin and folded around the immunodominant SIV Gag epitope pile were used in conjunction with monoclonal antibodies against CD3-Alexa700 (SP34), CD8- APC-Cy7 (SKl), CD28-PerCP-Cy5.5 (L293) and CD95-PE (DX2) (BD Biosciences) to stain CD8+ T lymphocytes from peripheral blood and extracted from tissue biopsy specimens.

Gag-specific cellular immune responses in vaccinated mice or rhesus monkeys were assessed by IFN-γ ELISPOT assays essentially as described (Barouch et al. 2004; Nanda et al. 2005; WO 2007/104792) .

Murine intracellular cytokine staining (ICS) assays were performed as previously described (Liu et al. 2006) . Briefly, lymphocytes isolated from various anatomic sites were stimulated at 37°C in 200 μl media containing 4 μg/ml ALIl peptide or pooled overlapping SIV Gag peptides. After 2h, 50 μl media containing 100 μg/ml GolgiStop (BD Biosciences) was added, and the cells were cultured for an additional 4h at 37°C. Cells were stained with fluorescently-conjugated anti-CD3α, CD4, CD8, CD44, CD62L and CD127 monoclonal antibodies and then fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences) . Permeablized cells were incubated with PE-conjugated anti- interleukin-2 (JES6-5H4; BD Biosciences) and APC-conjugated anti-IFN-γ (XMGl.2; BD Biosciences) antibodies and were washed and resuspended in PBS containing 1.5% formaldehyde. Samples were analyzed using an LSRII flow cytometer and FIoJo software.

The magnitude and phenotype of Gag-specific cellular immune responses in vaccinated rhesus monkeys were assessed by multiparameter ICS assays. 3x10 PBMC were incubated for 6 h at 37°C with RPMI 1640 containing 10% FBS alone as the negative control, the SIVmac239 Gag peptide pool consisting of 2 μg/ml of each peptide, or 10 pg/ml phorbol myristate acetate (PMA) and 1 μg/ml ionomycin (Sigma-Aldrich) as the positive control. Cultures contained monensin (GolgiStop; BD Biosciences, San Diego, US) and 1 μg/ml anti-CD49d mAb (BD Biosciences) . Cells were then stained with pre-titered amounts of anti-CD3-Alexa700 (SP34; Alexa Fluor 700), anti- CD4-AmCyan (L200), anti-CD8-APC-Cy7 (SKl; allophycocyanin Cy7), anti-CD28-PerCP-Cy5.5 (L293; peridinium chlorophyll protein Cy5.5), and anti-CD95-PE (DX2; phycoerythrin) mAbs and fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences) . Cells were then stained intracellularly with anti-IFN-γ-PE-Cy7 (B27; phycoerythrin Cy7), anti-TNF-α-FITC (Mabll; fluorescein isothiocyanate) , and anti-IL-2-APC (MQl- 17H12; allophycocyanin) mAbs and fixed with 1% paraformaldehyde. Samples were analyzed with an LSR II (BD Biosciences) and analyzed using FlowJo software (TreeStar, Ashland, OR, US). Approximately 500,000 to 1,000,000 events were collected per sample. Central memory (CM) and effector memory (EM) CD8+ and CD4+ T lymphocyte subpopulations were defined by CD28 and CD95 expression as previously described (Picker et al . 2006; Pitcher et al. 2002). Background levels of cytokine expression were typically <0.02% of CD4+ or CD8+ T lymphocytes.

Mucosal lymphocyte isolation was performed as follows . Murine small and large bowel, vaginal tract and respiratory tract were dissected free of associated connective tissue and cut into small pieces using straight scissors. Bowel specimens were washed extensively with HBSS and incubated with HBSS supplemented with 0.1 mm EDTA and 10% FBS at 37 °C for 30 min with vigorous shaking. The specimens were then vortexed and washed again with a fresh aliquot of the above media. Pooled supernatants from these two steps contained the IEL population. Cells were resuspended in 40% Percoll (Sigma Chemical) and layered over 67% Percoll; samples were centrifuged at lOOOg for 25 min. The interface between the two Percoll layers contained the lymphocyte population. After IEL removal, bowel specimens were washed three times with RPMI 1640 supplemented with 5% FBS to remove all traces of EDTA. Mucosal tissues were digested by two serial 30 min incubations at 37 °C in RPMI 1640 containing 5%FBS supplemented with type IV collagenase (Sigma Chemical) at 300 U/ml with vigorous shaking to isolate small and large bowel LPL as well as vaginal tract and respiratory tract lymphocytes. Pooled supernatants from serial incubations were purified with a Percoll gradient as above to isolate the lymphocyte population. Lymphocytes were isolated from monkey mucosal biopsies by incubating samples in RPMI 1640 supplemented with 10% FBS, type IV collagenase at 300 U/ml and DNasel (Sigma Chemical) at 30 U/ml at 37°C for 45 min with vigorous shaking. Cells were washed once with RPMI 1640 containing DNasel at 30 U/ml, and lymphocytes were purified with a Percoll gradient as above.

Adoptive transfer studies were performed as follows.

Spleens from donor animals were isolated and perfused gently with cold HBSS containing 4% FBS using 1 cc insulin syringes to release splenocytes. CD8+ T-lymphocytes were purified from splenocytes by negative selection using immunomagnetic beads (CD8+ T-cell isolation kit; Miltenyi Biotec) , according to the manufacturer's instructions. CD8+ T- lymphocytes were >90% pure by flow cytometric analysis. Purified cells were washed twice with PBS and resuspended at 2.5-5.0 x 10⁶ cells/10 μl in cold sterile PBS. Cells were transferred to recipient mice by tail vein injection.

Vaccinia challenge studies were performed as follows. Naive or immunized C57BL/6 mice were challenged with either 10⁸ or 10⁶ PFU of replication-competent, recombinant vaccinia virus expressing SIVmac239 Gag (Therion Biologies) administered intranasally in 10 μl of PBS (5 μl/nostril) .

Mice were assessed daily after infection for clinical status and weight loss and sacrificed after losing >20% of their initial body weight. To assess vaccinia virus titers, ovaries were harvested on day 6 after low-dose challenge and homogenized by three freeze-thaw cycles and vigorous vortexing. The homogenate was treated with trypsin for 30 min at 37°C. COS-7 cells were plated in 6-we11 plates at a density of 5xlO⁵ cells/well and incubated overnight. Cell monolayers were then infected with log dilutions of the homogenate in medium. After 2 days, vaccinia virus plaques were visualized by staining with 0.1% crystal violet and 20% ethanol.

Statistical analyses were performed with GraphPad Prism version 4.01 (GraphPad Software, Inc., 2004) . Immune responses among groups are presented as means with standard errors. Comparisons of mean immune responses were performed by t-tests for two groups or by analyses of variance (ANOVA) for more than two groups. Comparisons of mortality in challenge studies were performed using two-sided Fisher's exact tests, p-values of less than 0.05 were considered significant.

Example 2. Immunogen±city of heterologous rAd prime-boost regimens in mice

Groups of C57BL/6 mice (N=4/group) were primed on day 0 with 10 vp rAd26-Gag or rAd48-Gag and were boosted on day 28 with 10⁹ vp rAd26-Gag, rAd48-Gag, rAd5HVR48-Gag, or rAd5- Gag . The rAd5HVR48 vector is a hexon-chimeric rAd5 vector in which the 7 hexon hypervariable regions (HVRs) have been exchanged with the corresponding regions from Ad48 as previously described (described in detail in WO 2006/040330; Roberts et al. 2006). Gag-specific CD8+ T lymphocyte responses were assessed by D^b/AL11 tetramer binding assays at multiple time points following immunization.

As shown in Fig IA, CD8+ T lymphocyte responses primed by rAd26-Gag were boosted efficiently and similarly by rAd5- Gag and rAd5HVR48-Gag. Responses primed by rAd26-Gag were not boosted effectively by a second injection of rAd26-Gag, presumably as a result of anti-vector immunity generated by the priming immunization. Similarly, as depicted in Fig IB, responses primed by rAd48-Gag were boosted efficiently by the heterologous vectors rAd5-Gag and rAd26-Gag. Responses primed by rAd48-Gag were boosted less well by rAd5HVR48-Gag, consistent with the common idea that dominant vector- specific neutralizing antibodies are directed primarily against the hexon HVRs . As expected, responses primed by rAd48-Gag were not boosted effectively by a second injection of rAd48-Gag. These data demonstrate that optimal heterologous rAd prime-boost regimens require two vectors that are serologically distinct and specifically that lack hexon HVR cross-reactivity.

Further, the directionality (prime or boost, depending on the efficacy) of heterologous rAd prime-boost regimens in mice was studied. C57BL/6 mice (N=4/group) were primed on day 0 and boosted on day 28 with rAd26-Gag, rAd5-Gag, or rAd5HVR48-Gag as shown in Fig 2A. CD8+ T lymphocyte responses primed by rAd26-Gag were boosted efficiently and comparably by rAd5-Gag and rAd5HVR48-Gag (closed symbols) .

Interestingly, responses primed by rAd5-Gag or rAd5HVR48-Gag were boosted less effectively by rAd26-Gag (open symbols; p<0.01, ANOVA). These data indicate that the order of administering rAd vectors is important in determining the overall potency of heterologous rAd prime-boost regimens, and that viruses may preferably be used in either a priming infection or a boosting infection. Similarly, it was observed that rAd35/rAd5 regimens were more potent than rAd5/rAd35 regimens in mice (data not shown) . The impact of anti-Ad5 immunity on the optimal rAd26/rAd5 and rAd26/rAd5HVR48 regimens in this model was also studied. In mice with pre-existing anti-Ad5 immunity, the rAd5-Gag vector was not effective as a boosting vector, while the rAd26/rAd5HVR48 regimen proved optimal (Fig 2B; p<0.01, t- test) . Functional IFN-γ ELISPOT assays revealed similar results (data not shown) .

Example 3. CD8+ v CD4+ T cell responses in rhesus monkeys The possible existence of important qualitative differences in the phenotype of cellular immune responses elicited by various serotype rAd viruses was explored in rhesus monkeys by assessing the magnitude and phenotype of CD8+ and CD4+ T lymphocyte responses elicited by different rAd vectors. Un- fractionated IFN-γ ELISPOT responses in rhesus monkeys (N=3/group) following immunization with 10¹¹ vp rAd5-Gag, rAd26-Gag, rAd48-Gag, or rAd49-Gag have been described (Abbink et al . 2007) . Those studies were now expanded by evaluating fractionated CD8+ and CD4+ T lymphocyte responses in these animals at week 4 following immunization. As shown in Fig 3A, CD4-depleted and CD8-depleted IFN-γ ELISPOT assays demonstrated that rAd5-Gag elicited higher magnitude CD8+ T lymphocyte responses as compared with rAd26-Gag (white bars) but that both vectors elicited comparable magnitude CD4+ T lymphocyte responses (black bars) . Interestingly, rAd5-Gag elicited rather skewed CD8+ > CD4+ T lymphocyte responses, whereas rAd26-Gag elicited a much more balanced CD8+ - CD4+ T lymphocyte response. Also the rAd48- Gag and rAd49-Gag vectors induced quite a balanced CD8+ - CD4+ T lymphocyte response, albeit at lower magnitudes.

Next, intracellular cytokine staining (ICS) assays were performed to evaluate the IFN-γ, TNF-α, and IL-2 secretion from CD8+ and CD4+ T lymphocytes. As depicted in Fig 3B-C, a single immunization with rAd5-Gag elicited mean CD8+ T lymphocyte responses of 0.38% for IFN-γ, 0.25% for TNF-α, and 0.15% for IL-2 and mean CD4+ T lymphocyte responses of 0.09%-0.14% for each of these cytokines. In contrast, a single immunization with rAd26-Gag induced CD8+ and CD4+ T lymphocytes responses of 0.08-0.15% for IFN-γ, TNF-α, and IL-2, which again, is more balanced than the immune response found with the Ad5-based virus. These data confirm the ELISPOT results and indicate that the increased CD8+ T lymphocyte responses elicited by rAd5-Gag consisted primarily of IFN-γ and TNF-α producing cells, while IL-2 producing cells are under-represented.

Multiparameter flow cytometry was used to evaluate the phenotype of these cytokine secretion responses in greater detail. 8-color ICS assays were performed utilizing the following monoclonal antibodies: CD3-Alexa700, CD4-AmCyan, CD8-APC-Cy7, CD28-PerCP-Cy5.5, CD95-PE, IFN-γ-PE-Cy7 , TNF-α- FITC, and IL-2-APC. Central memory (CM) and effector memory (EM) T lymphocytes were defined, respectively, as CD28+CD95+ and CD28-CD95+ cells, as known in the art (Picker et al. 2006; Pitcher et al. 2002) . As shown in Fig 4A-B, rAd5-Gag elicited primarily IFN-γ+ (red) , TNF-α+ (orange) , and IFN- γ+/TNF-α+ (blue) CD8+ T lymphocyte responses but only low levels of IL-2+ (yellow) and polyfunctional IFN-γ+/TNF- α+/IL-2+ (black) cells. In contrast, as depicted in Fig 4C- D, rAd26-Gag elicited lower levels of IFN-γ+ (red), TNF-α+ (orange) , and IFN-γ+/TNF-α+ (blue) CD8+ T lymphocyte responses, but substantially higher levels of IL-2+ (yellow) and polyfunctional IFN-γ+/TNF-α+/IL-2+ (black) cells. Specifically, a mean of 27% of CD8+ CM T lymphocytes and 33% of CD8+ EM T lymphocytes elicited by rAd26-Gag secreted all three cytokines, whereas only 10% of CD8+ CM T lymphocytes and 7% of CD8+ EM T lymphocytes elicited by rAd5-Gag were polyfunctional. Similarly, a substantially greater proportion of CD4+ CM and EM T lymphocytes induced by rAd26- Gag secreted all three cytokines as compared with CD4+ CM and EM T lymphocytes induced by rAd5-Gag. These data indicate that rAd5-Gag and rAd26-Gag elicit qualitatively and striking different phenotypes of cellular immune responses in primates, such as rhesus monkeys. Specifically, rAd26 induced larger proportions as well as greater absolute numbers of IL-2+ and polyfunctional CD8+ and CD4+ T lymphocytes, whereas rAd5 induced greater numbers of IFN-γ+ and TNF-α+ CD8+ T lymphocytes.

Example 4. T cell responses upon heterologous rAd injection using rAd26 as a priming and boosting agent

To explore whether priming with rAd26-Gag would be beneficial and highly effective in subsequent boosting with rAd5-Gag, and how it would relate to other prime/boost regimens, groups of rhesus monkeys (N=3/group) were immunized with various candidate vaccine regimens: rAd5/rAd5, rAd26/rAd5, rAd48/rAd5, rAd49/rAd5, rAd5/rAd26, rAd35/rAd5, and rAd5/rAd35. All vectors expressed the same SIV Gag insert, and the boost immunization was performed 24 weeks after the priming immunization. Mean IFN-γ ELISPOT responses in the 16 weeks following the boost immunization are shown in Fig 5. Animals primed with rAd5-Gag were not efficiently boosted by a second injection of rAd5-Gag, presumably due to anti-Ad5 immunity generated by the priming immunization. In contrast, monkeys primed with rAd26-Gag were boosted remarkably well by rAd5-Gag to mean ELISPOT responses of 2,553 SFC/10⁶ PBMC against the single SIV Gag antigen at week 2 following the boost immunization (Fig 5A and B) . These responses were >8-fold higher than those generated by the homologous rAd5/rAd5 regimen, and clear differences between groups of monkeys persisted for >16 weeks. Monkeys primed with rAd48-Gag or rAd49-Gag were also boosted by rAd5-Gag, although to a lesser extent. Although the limited numbers of animals in this study precluded a formal statistical analysis, it is clear that heterologous rAd prime-boost regimens were substantially more immunogenic than the homologous rAd5 regimen in primates. Moreover, as shown in Fig 5B, a directionality of heterologous rAd prime- boost regimens in primates was also observed: consistent with the mouse studies, the rAd26/rAd5 regimen proved more potent than the rAd5/rAd26 regimen, and the rAd35/rAd5 regimen was more potent than the rAd5/rAd35 regimen. The ability of rAd26-Gag to prime responses for a subsequent efficient boost by rAd5-Gag strongly supports the functional relevance of the balanced-, polyfunctional responses elicited by rAd26 viruses.

The breadth of cellular immune responses in rhesus monkeys that received the optimal rAd26/rAd5 regimen was further investigated by assessing IFN-γ ELISPOT responses against all 125 individual SIV Gag peptides. As shown in Fig 6A, at least 2 epitope-specific responses were observed in each animal at week 4 following the rAd26-Gag priming immunization. As depicted in Fig 6B, also a marked 10-fold increase in the magnitude of these epitope-specific responses was observed at week 4 following the rAd5-Gag boost. Importantly, the emergence of numerous additional epitope-specific responses following the boost immunization was also observed. These data indicate that heterologous rAd boosting augmented not only the magnitude but also the breadth of Gag-specific cellular immune responses in primates, indicating that the order in which different recombinant adenoviruses are administered in prime/boost regimens may heavily influence the immune response of the host .

Example 5. Mucosal immune response upon immunization with rAd5-Gag

The ability of systemically administered rAd5-Gag virus to stimulate mucosal cellular immune responses was studied. C57BL/6 mice were immunized i.m. with 10 vp rAd5-Gag. Gag- specific CD8+ T-lymphocyte responses specific for the dominant Derestricted epitope ALIl were assessed by D^b/AL11 tetramer binding assays. The magnitude and kinetics of CD8+ T-lymphocyte responses in multiple systemic and mucosal compartments was assessed, including those in peripheral blood, spleen, inguinal and mesenteric lymph nodes, Peyer's patches, vaginal mucosa, and the intraepithelial and lamina propria lymphocyte populations (IEL and LPL) of both the small and large intestines. To determine the effector or memory phenotype of ALll-specific CD8+ T-lymphocytes, multiparameter flow cytometry was utilized to assess CD44, CD62L and CD127 expression (Fig 7A) . These markers are well known in the art and indicate the phenotypical status of the lymphocytes. Lymphocytes from both systemic and mucosal compartments had comparable viability as determined by vital dye exclusion and their ability to produce IFN-γ and IL-2 following stimulation with phytohemagglutinin and ionomycin (data not shown) . Lymphocytes isolated from various anatomic sites also exhibited different phenotypic characteristics, indicating the anatomic distinctness and purity of the isolated cell populations. For example, splenocytes and gastrointestinal LPL of naive mice comprised a mixture of CD3- B-lymphocytes, CD4+ T-lymphocytes, and CD8+ T- lymphocytes (Fig 7B) . In contrast, gastrointestinal IEL were predominantly CD8+ T-lymphocytes . Furthermore, CD8+ T- lymphocytes in different anatomic compartments had distinct effector and memory phenotypes. While blood, spleen, lymph nodes and Peyer's patches contained a mixture of naive, central memory and effector/effector memory CD8+ T- lymphocytes, gastrointestinal and vaginal effector sites contained almost exclusively effector/effector memory CD8+ T-lymphocytes (Fig 7C) . Effector/effector memory cells are available for immediate action upon an encounter with a pathogen; central memory cells are long-term cells that can expand upon antigen stimulation. The terms are well known in the field of immunology. The gastrointestinal IEL and LPL compartments, but not systemic ones, also contained substantial numbers of γδ-TCR+ T-lymphocytes (data not shown) .

After a single i.m. immunization with rAd5-Gag, CD8+ T- lymphocyte responses in multiple anatomic compartments were assessed by D^b/AL11 tetramer binding assays over a 24-week time course. The rate of rise and decay (kinetics) of the CD8 responses was similar at all anatomic sites evaluated (Fig 8A) . At weeks 2-4 following vaccination, peak ALIl- specific CD8+ T-lymphocyte responses in blood were 7-8% of total CD8+ T-lymphocytes, while peak responses in spleen were 4%. Peak responses in both systemic and mucosal lymphoid inductive sites (inguinal lymph nodes, mesenteric lymph nodes and Peyer's patches) were several-fold lower at 1-2%. Surprisingly, peak responses in small and large bowel lamina propria (4-6%) were similar in magnitude to those seen in blood and spleen, although peak responses in the small and large bowel IEL compartment were several-fold lower (1-2%) . In the vaginal tract, peak responses were 30- 40% of CD8+ T-lymphocytes, although the absolute numbers of lymphocytes in the vaginal mucosa remained low. In all anatomic compartments, ALll-specific CD8+ T-lymphocyte responses exhibited considerable durability for at least 24 weeks. At week 2, ALll-specific CD8+ T-lymphocytes from all anatomic sites exhibited predominantly an effector (CD62L- CD127 low) phenotype. However, by weeks 12 and 24, there was a transition of the ALll-specific CD8+ T-lymphocytes into the memory phase that was characterized by the progressive accumulation effector memory lymphocytes (CD62L- CD127 high) at all anatomic sites as well as central memory lymphocytes (CD62L+ CD127 high) in mucosal and systemic lymphoid inductive sites (Fig 8B) .

To inestigate the functionality of vaccine-elicited mucosal CD8+ T-lymphocyte responses, ICS assays to assess IFN-γ and IL-2 production at week 2 following i.m. vaccination were also performed (Fig 8C) . Potent IFN-γ responses were observed in all anatomic compartments following stimulation with either pooled Gag peptides or with the single ALIl epitope peptide, and the anatomic distribution of these responses was concordant with the tetramer binding assays. IL-2 responses were of lower magnitude as compared with IFN-γ responses, consistent with studies of rAd5 vectors in rhesus monkeys (data not shown) , but they were comparable in frequency between spleen and small bowel lamina propria (Fig 8C) . The majority of IL-2 secreting cells also produced IFN-γ (data not shown) .

Example 6. Mucosal immune response upon immunization in a heterologous prime-boost regimen Despite the results from the previous example, the utility of rAd5 as a stand-alone vaccine modality is substantially limited by the inability of homologous vector re- administration to boost responses. The possibility of generating secondary anamnestic responses in mucosal tissues by using rare and hexon-chimeric rAd viruses alone or in heterologous prime-boost combinations has not previously been investigated. Because the anatomic distribution of recall responses is biased by the site of initial antigen exposure (Mora and Von Andrian 2006; Neutra and Koslowski 2006), it was investigated whether systemic administration of heterologous rAd prime-boost regimens using serologically distinct rare serotype and hexon-chimeric rAd vectors would boost mucosal responses or alternatively would bias recall responses away from mucosal surfaces.

First, mucosal cellular immune responses elicited by different rAd viruses after a single systemic immunization were compared. C57BL/6 mice were immunized i.m. with 10 vp rAd5-Gag, rAd5HVR48-Gag, rAd26-Gag or rAd35-Gag. CD8+ T- lymphocyte responses were analysed at week 2 at the peak of the response and at week 12 during the memory phase using D^b/AL11 tetramer binding assays. At week 2, all vectors induced comparable frequencies of Gag-specific CD8+ T- lymphocytes in blood and spleen, as expected for this dose. Importantly, rAd26-Gag and rAd5HVR48-Gag induced high frequency CD8+ T-lymphocyte responses in multiple mucosal compartments (Fig 9A) . At week 12, rAd5-Gag and rAd5HVR48- Gag induced similarly high and significant frequencies of mucosal ALll-specific memory CD8+ T-lymphocytes, whereas rAd26-Gag and rAd35-Gag induced somewhat lower frequencies (Fig 9B) . These data show that rare serotype and hexon- chimeric rAd vectors elicited mucosal CD8+ T-lymphocyte responses that persisted after a single systemic immunization . Next, it was assessed whether heterologous boosting with a serologically distinct rAd virus would result in enhanced secondary mucosal CD8+ T-lymphocyte responses. Naive C57BL/6 mice or mice previously primed with 10⁹ vp rAd26-Gag were immunized i.m. with 10 vp rAd5HVR48-Gag, and CD8+ T-lymphocyte responses were examined in multiple anatomic compartments. As compared with naive mice, mice previously primed with rAd26-Gag exhibited substantially higher peak frequencies of CD8+ T-lymphocytes following rAd5HVR48-Gag immunization in both systemic and mucosal compartments (Fig 9C) . The magnitude of the boost effect was comparable at systemic and mucosal sites. Peak frequencies of CD8+ T-lymphocytes approached 20% in the small bowel lamina propria and exceeded 60% in the vaginal tract, and these responses persisted for over 12 weeks (Fig 9D) . These data demonstrate that systemic heterologous rAd prime-boost regimens induced potent secondary recall responses in multiple mucosal compartments.

Then, the magnitude and kinetics of mucosal CD8+ T- lymphocyte responses elicited by systemic heterologous versus homologous rAd prime-boost regimens were compared. C57BL/6 mice were primed i.m. at week 0 with rAd26-Gag and were boosted i.m. at week 8 with the homologous rAd26-Gag vector or the heterologous rAd5HVR48-Gag vector. Systemic and mucosal ALll-specific CD8+ T-lymphocyte responses were assessed for a total of 20 weeks. Homologous administration of rAd26-Gag resulted in little to no boosting of CD8+ T- lymphocyte responses as expected (Fig 9E) . In contrast, heterologous administration of rAd5HVR48-Gag generated significantly enhanced peak and memory CD8+ T-lymphocyte responses in both systemic and mucosal compartments (p=0.005 for spleen, p=0.001 for small bowel LPL, p=0.005 for large bowel LPL, and p=0.02 for vaginal tract lymphocytes comparing heterologous versus homologous responses at week 10 using two-tailed t-tests) . These data once again demonstrate that heterologous rAd prime-boost regimens are significantly superior to homologous rAd regimens for generating potent and durable cellular immune memory in the gastrointestinal and vaginal tracts, and that the specific rAd26-rAd5HVR48 prime-boost is a preferred regimen.

Example 7. Systemic CD8+ T-lymphocyte trafficking to mucosal surfaces after i.m. rAd vaccination

The comparable magnitude and kinetics of CD8+ T-lymphocyte responses in systemic and mucosal compartments following i.m. rAd vaccination suggested a coordinated immune response that bridged anatomic sites, contrasting with the anatomically skewed responses reported in prior studies using several different vaccine modalities (Gallichan and Rosenthal 1996; Belyakov et al. 1998 and 2001; Kantele et al. 1999; Santosuosso et al . 2005). Importantly, rAd bio- distribution studies in rabbits showed no evidence of direct vector trafficking to mucosal lymphoid inductive sites following i.m. immunization (data not shown), suggesting that T-lymphocyte priming in this context is largely restricted to systemic inductive sites. It was therefore hypothesized that systemic CD8+ T-lymphocytes may have acquired the capacity to migrate to gastrointestinal and vaginal mucosa and to persist at those sites following i.m. rAd vaccination. To test this, adoptive transfer studies were performed to evaluate the trafficking of systemic CD8+ T-lymphocytes activated by rAd immunization. In the first set of experiments, C57BL/6 mice were immunized i.m. with the heterologous rAd26-Gag/rAd5HVR48-Gag prime-boost regimen described supra to generate high frequencies of ALIl- specific CD8+ T-lymphocytes . On day 10 after the boost immunization, systemic CD8+ T-lymphocytes were purified from splenocytes by negative selection using immunomagnetic beads. 2x10 CD8+ T-lymphocytes were then transferred intravenously (i.v.) to naive recipient mice, and the anatomic distribution and phenotype of the transferred CD8+ T-lymphocytes were determined 12 days later. Tetramer- positive CD8+ T-lymphocytes rapidly migrated from the blood to all anatomic sites examined and established a tissue distribution pattern that recapitulated the pattern seen after direct immunization (Figs 9 and 10A) . Moreover, the anatomic distribution of effector and memory phenotypes of the transferred tetramer-positive CD8+ T-lymphocytes proved comparable with that seen after active immunization, with central memory cells accumulating at systemic and mucosal inductive sites but largely excluded from mucosal effector surfaces (Figs 8B and 10B) . Importantly, transferred tetramer-positive CD8+ T-lymphocytes that trafficked to the gastrointestinal tract also showed markedly increased expression of integrins and chemokine receptors critical for intestinal homing. β7 integrin, CCR9 and CD103 (integrin αlEL) were dramatically up-regulated on ALll-specific CD8+ T-lymphocytes migrating to gastrointestinal LPL and IEL compartments despite being expressed at very low levels on donor lymphocytes prior to adoptive transfer (Fig IOC) . These findings suggest that vaccine-activated, systemic CD8+ T-lymphocytes exhibited substantial phenotypic plasticity as well as the capacity to traffic widely to mucosal tissues. To quantitate more precisely the relative contribution of systemic CD8+ T-lymphocytes to the mucosal immune responses induced by i.m. rAd vaccination, adoptive transfer experiments were performed utilizing congenic mice. CD8+ T- lymphocytes were purified from splenocytes of naive Ly5.1+ mice (B6.SJL), and 10⁷ cells were transferred i.v. to naive Ly5.2-congenic recipients (C57BL/6). As expected, transferred naive CD8+ T-lymphocytes migrated rapidly to the spleen and lymph nodes in recipient mice (Fig 10D) . In the absence of immunization, trafficking of adoptively transferred CD8+ T-lymphocytes to mucosal effector sites was highly restricted at day 14 post-transfer, as demonstrated by the low ratio of Ly5.1/Ly5.2+ T-lymphocytes in the gastrointestinal IEL and LPL compartments (Fig 10D) . In contrast, i.m. immunization of the recipient mice with 10⁹ VP rAd5-Gag on day 2 post-transfer generated Ly5.1+ ALIl- specific CD8+ T-lymphocytes that efficiently migrated to gastrointestinal mucosa by day 14, as shown by comparable numbers of Ly5.1+ and Ly5.2+ ALll-specific CD8+ T- lymphocytes at these anatomic sites (Fig 10E) . These findings demonstrate that vaccine-activated, but not quiescent, peripheral CD8+ T-lymphocytes have the capacity to migrate rapidly and extensively to multiple mucosal tissues. Moreover, these data suggest that systemic lymphocytes migrating to mucosal effector sites accounted for the vast majority of the responding cells observed at those sites. Thus, trafficking of systemic lymphocytes, rather than direct local priming at mucosal sites, appeared to be the predominant mechanism of generating potent and widespread mucosal immunity following systemic vaccination with rAd viruses in vaccine settings.

Example 8. Heterologous rAd prime-boost vaccination followed by a lethal recombinant vaccinia-Gag challenge To determine the functional significance of mucosal CD8+ T- lymphocyte responses generated by i.m. rAd immunization, the protective efficacy of the optimal systemic heterologous rAd26-rAd5HVR48 prime-boost regimen against a lethal mucosal challenge with recombinant vaccinia virus expressing SIV Gag (rVac-Gag) was assessed. C57BL/6 mice were immunized with the heterologous rAd26-Gag prime, rAd5HVR48-Gag boost regimen described supra. At week 8 following the boost, groups of vaccinated and control mice were challenged intranasally with 10⁸ PFU rVac-Gag. Control animals rapidly lost weight and were sacrificed between days 5-7 post- challenge after losing >20% of their body weight (Fig HA) . In contrast, vaccinated animals remained clinically healthy with relatively stable body weights (Fig HB) . By day 6 post-challenge, vaccinated mice exhibited robust, anamnestic ALll-specific CD8+ T-lymphocyte responses in blood, whereas responses in control animals were undetectable (Fig HC) . To assess the tissue distribution of antigen-specific CD8+ T- lymphocyte responses, vaccinated mice were sacrificed on day 12 post-challenge, and responses in multiple mucosal compartments were compared with those in vaccinated but uninfected mice. Broadly distributed anamnestic responses were observed in respiratory mucosa, gastrointestinal mucosa and the periphery following mucosal challenge (Fig HD) . In a concurrent sub-lethal challenge experiment using 10 PFU rVac-Gag, vaccinia virus titers were detected in the ovaries of all control mice but in none of the vaccinated animals on day 6 post-challenge (data not shown) . These data indicate that the widely distributed cellular immune responses generated by the systemic heterologous rAd prime-boost regimen afforded robust protection against this mucosal virus challenge. Example 9. Mucosal CD8+ T-lymphocyte memory in rhesus monkeys upon rAd immunization

Three rhesus monkeys (2 that expressed the MHC class I allele Mamu-A*01 and 1 that did not express this allele) were immunized i.m. with 10¹¹ vp rAd5HVR48-Gag . Mamu-A*01- restricted CD8+ T-lymphocyte responses to the dominant Gag epitope pile (Kuroda et al. 1998) were assessed by multiparameter tetramer binding assays for up to 1 year following immunization in multiple systemic and mucosal compartments. CD8+ T-lymphocyte responses were observed in duodenal mucosa as well as in blood and lymph nodes in the Mamu-A*01-positive animals at weeks 4 and 32 after vaccination, and the magnitude of mucosal responses proved comparable with the magnitude of systemic responses (Fig 12A and 12B) . Moreover, pile-specific memory CD8+ T-lymphocytes persisted for at least 52 weeks following vaccination. At this late time point, pile-specific CD8+ T-lymphocytes were detected in duodenal mucosa, colorectal mucosa, bronchoalveolar lavage and vaginal mucosa (Fig 12C) . Importantly, the magnitude of these long-term mucosal responses proved comparable with those found in blood and lymph nodes, except for responses in vaginal mucosa that were approximately 5-fold higher in magnitude than responses in blood, consistent with our mouse studies (Figs 8A and

9A) . At all anatomic sites, pile-specific CD8+ T-lymphocytes were predominantly of a CD28+CD95+ memory phenotype (Pitcher et al. 2002) . These data demonstrate that a single i.m. rAd vaccination generated potent and durable CD8+ T-lymphocyte memory that persisted for over one year in multiple mucosal tissues in primates. REFERENCES

Abbink P et al. (2007) J Virol 81:4654-63 Altman JD et al. (1996) Science 274:94-6 Barouch DH et al. (2004) J Immunol 172:6290-7

Belyakov IM et al. (1998) Proc Natl Acad Sci USA 95:1709-14 Belyakov IM et al . (2001) Nature Med 7:1320-1326 Betts MR et al. (2006) Blood 107:4781-9 Brenchley JM et al. (2004) J Exp Med 200:749-759 Catanzaro AT et al . (2006) J Infect Dis 194:1638-49 Darrah PA et al. (2007) Nat Med 13:843-50

Gallichan WS & Rosenthal KL (1996) J Exp Med 184:1879-90 Janssen EM et al . (2003) Nature 421:852-6 Kantele A et al. (1999) J Immunol 162:5173-7 Kuroda MJ et al. (1999) J Virol 73:1573-1579 Lemckert AA et al . (2005) J Virol 79:9694-701 Lemckert AA et al . (2006) J Gen Virol 87:2891-9 Li Q et al. (2005) Nature 434:1148-1152 Liu J et al. (2006) J Virol 80: 11991-7 Marzo AL et al. (2007) J Immunol 179:36-40

Masopust D et al. (2001) Science 291:2413-2417 Masopust D et al. (2004) J Immunol 172:4875-4882 Masopust D et al. (2006) J Immunol 176:2079-2083 Mattapallil JJ et al. (2005) Nature 434:1093-1097 Mattapallil JJ et al. (2006) J Exp Med 203:1533-1541

Mora JR & von Andrian UH (2006) Trends Immunology 27:235-43 Nanda A et al . (2005) J Virol 79:14161-8

Neutra MR & Kozlowski PA (2006) Nature reviews 6:148-158 Picker LJ et al. (2006) J Clin Invest 116:1514-24 Pinto AR et al. (2003) J Immunol 171:6774-9 Pitcher CJ et al . (2002) J Immunol 168:29-43 Precopio ML et al . (2007) J Exp Med 204:1405-16 Reyes-Sandoval A et al. (2004) J Virol 78:7392-9 Roberts DM et al . (2006) Nature 441:239-43 Santosuosso M et al. (2005) J Immunol 174:7986-94 Shedlock DJ and H Shen (2003) Science 300:337-9 Shiver JW and EA Emini (2004) Annu Rev Med 55:355-72 Sun JC and MJ Bevan (2003) Science 300:339-42 Thorner AR et al. (2006) J Virol 80:12009-16

Thorner AR and DH Barouch (2007) Curr Infect Dis Rep 9:71-5 Veazey RS et al. (1998) Science 280:427-431 Vogels R et al. (2003) J Virol 77:8263-71

Claims

1. A kit of parts comprising: a priming vaccine composition comprising a recombinant replication-defective adenovirus of serotype 26 (rAd26) comprising a nucleic acid of interest; and a boosting composition comprising a recombinant replication-defective adenovirus of serotype 5 (Ad5) , or a recombinant replication-defective adenovirus of serotype 5 comprising hexon proteins wherein the HVR regions of the serotype 5 adenovirus have been replaced by the corresponding HVR regions of adenovirus 48 (rAd5HVR48) .

2. The kit of parts of claim 1, wherein said compositions further comprise a pharmaceutically acceptable excipient.

3. The kit of parts of claim 1, wherein said nucleic acid of interest encodes at least one antigen, or an immunogenic part of said antigen, of a simian or human immunodeficiency virus .

4. A method for inducing mucosal immune response against an antigen in a mammalian subject, comprising the steps of: - obtaining a recombinant replication-defective adenovirus comprising a nucleic acid of interest encoding said antigen; and administering said adenovirus systemically into said mammalian subject, wherein said adenovirus is a recombinant replication- defective adenovirus serotype 26 (rAd26).

5. The method of claim 4, wherein said mammalian subject is a monkey or a human.

6. The method of claim 4, wherein said antigen is a simian or human immunodeficiency virus antigen, or an immunogenic part thereof.

7. The method of claim 4, wherein said adenovirus is administered intramuscularly.

8. The method of claim 4, wherein said step of administering said adenovirus is followed by the step of: administering a second recombinant replication- defective adenovirus comprising said nucleic acid of interest, wherein said second adenovirus is heterologous to the earlier administered (first) adenovirus serotype.

9. The method of claim 8, wherein said second adenovirus is rAd5 or rAd5HVR48.

10. A method of inducing the migration of activated CD8+ T- lymphocytes from systemic to mucosol immune compartments in a mammalian subject, comprising the steps of: obtaining a recombinant replication-defective adenovirus comprising a nucleic acid of interest encoding an antigen; and administering said adenovirus systemically into said mammalian subject, wherein said adenovirus is a recombinant replication- defective adenovirus serotype 26 (rAd26) .

11. The method of claim 10, wherein said mammalian subject is a monkey or a human.

12. The method of claim 10, wherein said antigen is a simian or human immunodeficiency virus antigen, or an immunogenic part thereof.

13. The method of claim 10, wherein said adenovirus is administered intramuscularly.

14. The method of claim 10, wherein said step of administering said adenovirus is followed by the step of: administering a second recombinant replication- defective adenovirus comprising said nucleic acid of interest, wherein said second adenovirus is heterologous to the earlier (first) administered adenovirus serotype.

15. The method of claim 14, wherein said second adenovirus is rAd5 or rAd5HVR48.