**4. Discussion**

Despite the exploration and implementation of numerous HIV-1 prevention strategies, 1.8 million new HIV infections occurred in 2017 [44]. There is an urgent need for the development of an effective, safe, and affordable HIV-1 vaccine. We have constructed an rBCG vaccine candidate expressing the HIVACAT T-cell immunogen, HTI, using the integrative antibiotic-resistance free *E. coli*-mycobacterial shuttle vector, p2auxo.int. The HTI immunogen was designed to target T-cell responses to the most beneficial T-cell targets and the most vulnerable sites of HIV-1. When delivered using DNA and MVA vectors, it has been shown to be capable of inducing broadly and evenly distributed immune responses of high magnitude in mice and monkeys [7]. We have previously demonstrated the in vitro and in vivo stability of the integrative plasmid p2auxo.int for expressing HIV-1 immunogens in BCG [33,43]. Here, we produced the BCG.HTI2auxo.int vaccine candidates under Good Laboratory Practice compatible conditions, and characterized them genotypically and phenotypically, confirming the presence of the HTI gene and protein in the lysates of working vaccine stocks. Furthermore, we demonstrate that the BCG.HTI2auxo.int vaccine in combination with ChAdOx1.HTI induced TB and HIV-1-specific IFN-γ-producing T-cell responses in adult BALB/c mice. The vaccination regimen was well tolerated during the follow-up period, although a longer safety assessment will be necessary, as symptoms related to a lack of attenuation of mycobacteria could take at least 50 days to emerge and impact body mass in the mouse model [45].

BCG is a remarkable live vaccine vehicle due to its capability of delivering antigens to APCs enabling the development of antigen-specific cell-mediated immune responses [24]. Mycobacterial antigens have been shown to be presented to T-cells by non-classical antigen presentation molecules such as CD1a/b/c, Major histocompatibility complex class I-related gene protein (MR1), and Human leukocyte antigen-E (HLA-E). The latter is of specific interest in the context of HIV-1 vaccine development due to the resistance of HLA-E to downregulation by the HIV-1 Nef during infection [46], as well as displaying a low level of allelic variation, with only forty three existing variants as opposed to the thousands of classical HLA class I molecules [47]. Furthermore, BCG-immunized humans elicited HLA-E restricted CD8<sup>+</sup> T-cell responses to *Mtb* peptides, which display cytotoxic as well as immunoregulatory activities [48]. One of the most successful HIV-related vaccine trials in animal models has been a Cytomegalovirus (CMV) vectored vaccine against SIV. This vaccine was able to establish persistent, SIV-specific effector memory T-cell responses in rhesus macaques and control pathogenic SIV infection following mucosal challenge [49]. CMV infection is known to upregulate HLA-E expression in humans, and the vaccine regimen in the monkeys induced strong Mamu-E restricted T-cell responses [50]. It is still unknown if BCG can elicit a similar immune response to a heterologous antigen when used as a vaccine vector in humans. However, BCG administered as an oral adjuvant along with inactivated simian immunodeficiency virus (SIV)mac239 particles in Chinese macaques was shown to confer protection to a high-dose SIVmac239 challenge [51]. The protection was attributed to non-cytolytic Major histocompatibility complex (MHC) Ib/E-restricted CD8<sup>+</sup> T-regulatory cells that suppressed the activation of SIV-positive CD4<sup>+</sup> T-lymphocytes.

As a vector, rBCG shares several traits with plasmid DNA. Both are often used as priming agents in combination with a virally vectored boost [8,24]. A rBCG–DNA prime-boost regimen showed less immunogenicity when explored using the HIVA immunogen, as compared to boosting with viral vectors [28]. On the other hand, a combination regimen of rBCG and DNA expressing HIVA was shown to confer protection in a pathogenic vaccinia–HIVA surrogate challenge model [39]. However, little is known about the advantages for specific immunity to heterologous immunogens upon combining rBCG, DNA, and viral vectors.

Recombinant BCG delivered on its own induces weak transgene-specific immune responses that are difficult to measure. Thus, we have assessed the enhancement of HIV-1 specific cellular immune responses by a prime vaccination with BCG.HTI2auxo.int when delivered with ChAdOx1.HTI in BALB/c mice. The magnitude of the total T-cell response was significantly higher in BCG.HTI2auxo.int primed mice as compared to mice receiving ChAdOx1.HTI alone. A similar magnitude was observed in BCGwt primed mice (Figure 4). The IFN-γ secretion in response to the individual HTI peptide pools was higher in all the assessed pools, although the differences did not reach statistical significance between mice receiving BCG prime and those not. The evident priming effect, even by BCGwt, is in line with the ability of BCG derivatives to act as potent adjuvants for subsequent boosting vaccines. It is known that immunization with BCG not expressing any transgene often can lead to higher vaccine-specific responses when delivered as a prime/adjuvant in combination with a virally vectored vaccine [39,52]. Components of BCG have also been used as an adjuvant for an HIV-1 DNA vaccine [53].

Interestingly, both BCG.HTI2auxo.int primed mice and mice receiving ChAdOx1.HTI alone responded to an average of seven to eight peptide pools, whereas mice primed with BCGwt alone only responded to an average of 4.5 peptide pools. This together suggests that priming with BCG.HTI2auxo.int enhances the HTI-specific immune response when delivered with ChAdOx1.HTI, while maintaining the breadth of the response. Priming with BCGwt appeared to boost the overall magnitude of the response, but ultimately directing responses to fewer peptide pools. The differences in the breadth of immune responses between the BCG.HTI2auxo.int and BCGwt could possibly be related to the strong adjuvant properties of BCG. A possible explanation is its capability of inducing trained immunity. This involves enhanced monocyte function and Natural killer (NK) cell function as reviewed by van der Meer et al. [54]. Macrophages, monocytes, and natural killer cells display enhanced responsiveness following a second encounter with a pathogen. Enhanced monocyte function has been demonstrated in humans three months after BCG vaccination along with increased cytokine production, as well as CD11b and toll-like receptor 4 (TLR-4) expression [17]. This effect could be observed up to one

year following immunization [55]. The unspecific immune activation could perhaps be involved in mechanisms enhancing the production of cytokines such as IFN-γ in T-cells, which leads to higher responses in the epitopes, which are more dominant. It is notable that the HTI was designed to avoid useless immunodominant epitopes in humans, but nevertheless, it is inevitable that there will be a hierarchy of epitopes in mice in which cellular immune responses are elicited against. Thus, the higher breadth observed following priming with the BCG.HTI2auxo.int could be related to the priming of responses that are specifically related to the transgene expressed (HTI), whereas the general increase in IFN-γ induced by BCGwt could be related to unspecific or adjuvanticity-related effects on the immune system rather than the specific priming of immune responses directed toward HTI. Previously, C57BL/6 mice immunized with DNA.HTI alone were shown to respond to two to six HTI peptide pools, and following an immunization schedule consisting of three DNA.HTI prime immunizations and one MVA.HTI immunization delivered at three-week intervals, this number was increased to six to 11 peptide pools [7]. Delivering a combination of BCG.HTI2auxo.int, DNA, and a viral vector expressing HTI could present a strategy for increasing the number of recognized pools further.

Correlates of protection for HIV-1 vaccines have been discussed following the RV144 trial, such as the IgG Ab response to the variable regions 1 and 2 (V1V2) loops being associated with a reduction in HIV-1 acquisition [56]. However, the translatability of data obtained in small animal models is limited, and pre-clinical design and the evaluation of HIV-1 vaccines remains a challenge. In humans, T-cell responses have also been associated with protection or decreased viral loads following infection. For instance, in the RV144 trial, CD4<sup>+</sup> T-cells secreting IL-2, TNF-α, IFN-γ, IL-4, and CD154 in response to HIV-1 envelope peptides were associated with lower infection rates in vaccine recipients [57].
