According to phylogenetic analysis, H5N1 viruses can be divided into 10 clades (0–9). Since 1997, humans have mainly been infected by H5N1 viruses from clades 0, 1, and 2, although there have also been reports of infection by clade 7 virus 
. Clade 2 is the most complicated in its genetic evolution and has been further divided into five subclades (2.1 to 2.5). In the current study, we have selected HA antigens from one clade 0 virus (A/HongKong/156/97), one clade 1 virus (A/VietNam/1203/2004), two clade 2 viruses (A/Indonesia/5/2005, clade 2.1; and A/Anhui/1/2005, clade 2.3) 
While progress has been made in reducing the number of required immunizations during vaccination with inactivated H5N1 vaccines by incorporating various adjuvants into the vaccine formulations, a major next-step for H5N1 vaccine research is to determine to what degree the immunity elicited by one H5 avian influenza vaccine (currently, many candidate H5N1 vaccines were developed based on a clade 1 virus (A/VietNam/1203/2004)) can cross-protect against H5N1 viruses from other clades. Unlike human Type A influenza viruses (H1 or H3 serotypes), any potential pandemic caused by an H5N1 virus will be of avian origin and, in theory, any of the current known H5N1 avian viruses may jump to the human population leading to the next pandemic. Therefore, a systemic examination on the cross-protection among HA antigens from different clades is needed for strategic planning to determine whether more than one H5N1 vaccine is needed based on the analysis of protection profiles, and if so, what particular viral strains should be selected to provide the maximum breadth of protection.
DNA vaccination is an attractive strategy to provide relatively quick and straightforward production of vaccines against an influenza pandemic when the demand for such vaccines suddenly increases. However, a key issue surrounding the use of DNA vaccines is their low immunogenicity in humans. In recent years, the success of the prime-boost strategy has greatly enhanced the utility of DNA vaccination for future human applications 
. At the same time, optimization of the design of antigen inserts based on the uniqueness of each antigen (a process of “antigen engineering”) 
can also play a key role.
Results included in the current report indicated that the tPA leader sequence and the C-terminal transmembrane domain/cytoplasmic region of H5 HA both contribute to better functional antibody responses in H5.tPA DNA vaccines when compared to H5.wt and H5.dTM DNA vaccines. The above findings were different from our previous results on protective antibody responses induced by differently designed flu H1 and H3 HA DNA vaccines 
. In this previous study, only the full length H1.wt but not transmembrane truncated H1.dTM induced high level HI and MN responses against H1 virus while both the full length H3.wt and truncated H3.dTM induced similar HI and MN responses against H3 virus 
. These results provide a strong indication that the HA antigens from different influenza A subtypes (H1, H3 and H5) may have different preferences for antigen structure designs in order to generate optimal protective antibody responses.
Studies were conducted in this report to identify the mechanism responsible for better protective antibodies in rabbit immune sera elicited by the H5.tPA HA insert design but the exact mechanism is currently unclear. First, we tested whether a higher level of HA antigen expression was produced with the tPA-leader design. As shown in and , antigen expression levels between WT-leader design and tPA-leader design were similar, thus excluding this possibility. Next, we asked whether there is increased secretion of the HA antigen due to the use of the tPA leader. However, as shown in and , there is no major detectable level of secreted HA antigens in supernatant for either the WT-design or tPA-design. Furthermore, the dTM design did have a higher level of secretion due to the deletion of transmembrane and intracellular portion of HA protein but did not elicit better protective antibody responses. Finally, a study was conducted to examine the possible role of post-translational processing, such as a change in glycosylation, of HA the antigen, which may affect the antigen processing pathway, as we previously reported with a hepatitis B surface antigen DNA vaccine 
. However, Figure S1
showed that there is no major difference between WT-design and tPA-design after de-glycosylation treatment.
Therefore, it is very likely that HA antigen expressed with the tPA leader may be more effective in eliciting conformational antibodies. This hypothesis is supported by two pieces of evidence. First, there was no difference in the levels of binding antibodies as measured by ELISA, which indicated that there was no difference in the general immunogenicity between WT and tPA leader designs; only a difference in the functional antibody was observed. Second, the HA.dTM DNA vaccine design also used the tPA leader but did not have better functional antibodies, proving that the proper folding of the HA antigen in the presence of a tPA leader is important and is dependent on the presence of an intact HA2 domain. It is possible that such an HA antigen conformation is part of a trimer structure of HA since the HA2 domain is involved in the formation of HA trimers. Only antibodies against the trimer structure are more functionally relevant to block the trimer form of HA spikes on viral particles.
By using the optimal H5.tPA HA insert design, studies in this report further demonstrated that there are good levels of cross protection by one H5 HA DNA vaccine against multiple H5N1 viruses from different clades. It is well documented that cross protection among H5N1 viruses can be detected 
. Clade 1 H5N1 vaccines (VN) cross protect against both clade 1 and clade 2 (Indonesia) viruses in ferrets with the use of a strong adjuvant 
. Using live attenuated cold adapted (ca) viruses expressing HA and NA from 1997, cross protection was observed against the late H5 virus from 1997 to 2005 
and ferrets 
and ferrets 
and mice 
and mice 
However, as shown in the current study, not all H5 HA vaccines can elicit the same levels of cross protective antibodies, and more significantly, maximum levels of protective antibodies were usually detected against the autologous viral isolates. Given the knowledge that protective anti-flu antibody responses in humans are much lower than in experimental animals, cross protection may not be very high in humans with one randomly selected H5 HA vaccine. In the current set of studies, it was encouraging to observe that the polyvalent H5 HA DNA vaccine was able to elicit high level protective antibody responses against multiple key H5N1 viruses. Such a polyvalent flu DNA vaccine can be used for stockpiling against a potential H5N1 pandemic even before information is available on which viral isolates may cause a human outbreak. At the same time, there are alternative approaches including the use of consensus HA antigen designs to achieve a broad coverage of various viral strains (personal communication with David Weiner). It will be interesting to compare the relative efficacy between polyvalent and consensus HA DNA vaccines in their abilities to elicit protective antibody responses. One unique advantage of the polyvalent formulation is its flexibility; one alternative HA antigen can replace or be added to the earlier polyvalent formulation in the event that a new strain of virus becomes a threat while it will be necessary to re-design the whole consensus HA insert in order to allow for broader coverage.
While HI titers against heterologous virus (cross-clade) were significantly reduced compared to HI titers against homologous virus in the current study, the heterologous virus titers were generally above 1
100, and in humans, a HI titer of 1
40 has been associated with protection 
, and so all of the constructs might be protective after multiple immunizations. At the same time, the titers observed in mice may not be the same as in humans. While HI titers have been associated with protection, H5 HA DNA vaccines described in the current report were not tested for protection against challenge and so there is the possibility that only some of these constructs may not protect against clinical disease or lethal infection.
No matter the design of DNA vaccines that may be used, recent studies have indicated that DNA priming immunization is effective as part of the prime –boost strategy for flu vaccine applications. In addition to DNA prime-inactivated flu vaccine boost 
, a study published in 2011 further demonstrated that DNA prime-live attenuated flu vaccine boost was equal to or more effective than twice immunization with the live attenuated flu vaccine against the H5N1 viruses based on antibody responses and viral clearance in immunized ferrets 
. Since live attenuated vaccines are considered the most immunogenic form of vaccines, it is impressive to observe that one time DNA prime was able to achieve the same priming effect as a live attenuated flu vaccine. In this particular study, the H5-VN.tPA DNA insert was used as part of the collaboration with the manufacturer of live attenuated H5N1 flu vaccine.
Based on the results published in the current report and other recent similar studies, H5N1 HA DNA vaccines evaluated in the current study should be included in the design of human studies to understand whether results reported here can be reproduced in humans when they are used as part of DNA prime, either individually or as part of the polyvalent HA DNA formulation. The finding from such studies will be very useful in the identification of simple yet powerful approaches to develop vaccines against major influenza pandemics.