23 Dec Episode 96: Questions, threats and innovation in COVID vaccines

Wed, 12/23/2020 - 17:41

Dear colleagues,


We have now two mRNA vaccines , which seem very efficacious after two doses in the phase 3 trials, but, as emphasized by some of you, a number of questions remain:  

  • Will efficacy be equally impressive in the frail elderly (who were not or hardly represented in the trials) ?  This will show from the evolution of the epidemic after vaccination, of course.
  • What about the efficacy and safety profile in population groups not represented at all in the trials, such as children and people with immune deficiency?  Very difficult to answer today.  
  • Will roll-out of the vaccine be able to block transmission at some point?   Wait and see….
  • Are two doses really needed for protection, as in both trials there was some indication of protection already after the first dose?
  • What is the chance that other innovative vaccine concepts would complement (or even replace) the present forerunner generation of vaccines?
  • What will be the effect of the recently discovered mutants in UK, South-Africa and elsewhere on vaccine efficacy?


In this episode, I want to share some information and thoughts with regards to the latter 3 questions:


  1. On the mutants, I add two documents:
  • A report from WHO on UK variant, which is spreading rapidly within and already outside the UK (closing the borders will not really help). As you can read it contains 14 significant mutations:   
    • Two mutations N501Y and P681H alter key amino-acids in the receptor binding domain (RBD).  Clearly, the neutralizing capacity of sera from vaccinated subjects (from the phase 3 trials) can and will be used to evaluate whether these “spontaneous” mutations could “escape” induced immunity.  I hope Pfizer and Moderna will communicate about that next week or so…?
    • A deletion may impact on the diagnostic tests, based on PCR of Spike protein


  • The CAPRISA slide series (kindly provided by Chris Kenyon) shows that in South-Africa a similar (but independent) variant spreads rapidly. It also carries the  N501Y + 2 other mutations in RBD and should also rapidly be evaluated for sensitivity to vaccine-induced neutralizing antibodies.


  1. With regard to novel vaccine concepts, I will focus on:
  1. Replication competent, but attenuated viruses and vectors, as opposed to the  replication-incompetent Adenoviral vectors (Chimp-Ad AstraZeneca; Ad26 J§J,  Ad5-Ad26 Sputnik V) presently in trial or already being rolled out.
  2. Self-amplyfying RNA, as opposed to mRNA (Pfizer, Moderna).
  3. Nanoparticle protein vaccine, as opposed to “simple protein S” (Novavax)    


  1. Attenuated vaccines:


    1.  A groundbreaking paper by the group of Kao Dallmeier and Johan Neyts proposes a recombinant Yellow Fever 17 vaccine (Ep 96.1), which has incorporated either the native S1-S2, an uncleavable SO or S1 alone (slide 2). 

This S0 or “prefusion stabilized S” has a deletion of furin cleavage site at S1/S2 boundary and proline substitutions at 986 and 987 and is (to my understanding) similar/same as the constructs used in the ‘forerunner” mRNA and Adenovirus vaccines.

In slide 3, it is evident that two doses of the S0 construct is slightly superior to S1-S2, while S1 alone is inactive in inducing neutralizing antibodies and protecting Syrian hamsters from a vigorous challenge (slide 3).

Slide 4 shows that this vaccine in 2 doses also functions well in cynomolgous macaques.  

From slide 5, it becomes evident that a single dose works as well to induce neutralizing antibodies and protection against challenge in the hamster model.  


    1. Ep 96.2 describes a traditionally prepared “cold-adapted attenuated” strain (at 22 °C). It acts as a single dose vaccine in ACE-2 transgenic mice: full protects against non-attenuated strain 3 weeks after intranasal immunization (slide 6).  It induces strong cross-neutralizing antibodies against the non-attenuated wild-type virus in serum, but also specific IgA in various tissues and Th1 T cells.  Table 1 shows the non-synonymous mutations that were induced by the cold adaptations: 7 in ORF1a, 1 in ORF1b and 3 in S (slide 7).  Since there are many mutations, one may be confident that the attenuated virus will not easily revert to wild-type… 


    1. Another interesting question is whether measles vaccine vector could be used as a live-attenuated vector for Spike (Ep 96.3)).  As can be seen in the Fig on slide 8, two different constructs were made with the genetic code for the full, unmodified S protein after the P (phosphoprotein) or H (hemagglutinin), with the latter showing best growth.  A prime-boost vaccination (day 0 and 28) was done with both live attenuated viruses or S protein + Alum as a control  in IFNAR−/−-CD46Ge mice (since they are the prime small animal model for analysis of MeV derived vaccines). Slide 9 shows that S-binding antibodies were induced and some SARS-CoV-2 neutralization by the H-construct (but not in all animals), while the P construct and protein immunization S + Alum were ineffective to induce neutralizing antibodies.  Also, some Th1 and CTL activity was induced in mice. More importantly, in the Syrian hamster model, the vaccine induced a partial protection against SARS-CoV-2 challenge (slide 10).

Clearly, this vaccine, although conceptually appealing should be improved.  One of the possibilities is to use a “prefusion stabilized S” (deletion of furin cleavage site at S1/S2 boundary and proline substitutions at 986 and 987).    


    1. In the paper Ep 96.4 a Rabies vector is used. In this case the expression of SARS-CoV-2 S1 is localized between the RABV N and P gene, and codon optimization for human cells has been shown to result in high expression levels (slide 11).  The reason for choosing S1 is that a similar vaccine, using MERS-S was protective in animal models. Immunization studies were done in Balb/c mice with 2 doses of either (1) the live S1 recombinant rabies alone, (2) the inactivated S1 recombinant rabies in absence or (3) presence of a Th1 stimulating adjuvant.  The latter induces the highest titers of antibodies with Th1 profile (IgG2a >> IgG1) and with strong in vitro  neutralizing activity (slide 12 and 13). Could a prefusion stabilized S also be used here?    


    1. The more advanced technology to create attenuated viruses is by codon or codon pair de-optimization.  The principle is explained in a review of 2018 (Ep 96.5). You can change individual codons or codon pairs in a virus by taking advantage of “codon degeneracy”: you replace the codons which are “optimal” (often used in humans), by codons for the same amino acids, but which are infrequently used in humans, resulting in lower efficiency.  Doing so, the same virus will be made by human cells, but much (strongly) attenuated and potentially useful as a vaccine.  A recent paper in Sci Reports analyzes which part of SARS-CoV-2 could be most suited for this strategy (Ep 96.6). They conclude that S is a good target because it uses codon pairs, frequently used by humans as well as N, ORF6 and ORF7b, because they do not use very rare codon pairs.  Hence for all these proteins the “gain” towards attenuation is high by replacing the native codon pairs by homologous pairs that are rarely used in humans.


Clearly, this strategy for “deoptimization” at first view is paradoxical, as in many other novel vaccine strategies (e.g. plasmid DNA, mRNA, recombinant vectors), the developers often claim that they have “optimized” codon usage.  The reason is that in the latter case, only one (S) or a few (S + N) proteins of SARS-CoV-2 are being produced and this should be done as efficiently as possible, while in the “de-optimization” strategy the whole infectious virus is used.  In this case, one want to limit the growth of the virus as much as possible, but keep the relevant proteins antigenically intact and this can be obtained by de-optimizing the codon (pairs) without changing the amino acids….


  1. Self-amplyfying (sa) RNA vaccines


    1. This type of vaccines is, to some extent, a variant on the mRNA.  The principles are explained in the review Ep 96.7.   They are based on alpha-viruses (+RNA) with a “complex cycle”: they contain the code for the replicase, which is immediately translated by the cell and then catalyzes the production of “antigenomic (-) RNA”.  The replicase then transcribes the vaccinal “gene of interest” as “subgenomic mRNA” from the antigenomic (-) RNA (see slide 17) and this mRNA then functions just like the ‘simple” mRNA vaccines .  So clearly, the structural proteins of the alphavirus are replaced by the vaccine antigen and no infectious virus can be produced: only the replicase to make more copies of the RNA genome and the vaccine antigen, ensuring longer  production of the vaccinal antigen than is “simple” mRNA was used as the vaccine.  The various delivery systems are illustrated on slide 18 and are in fact similar as mRNA.  The major advantage is that the expression of the “vaccine antigen” is prolonged as compared to simple mRNA . Immunization against the alphavirus replicase is a byproduct.


    1. Three recent papers, applying this technology for SARS-CoV-2 are briefly presented here (Ep 96.8-10).  They all use the backbone with the replicase of Venezuelan Equine Encephalitis virus and a full length SARS-CoV-2 S protein (in the Mc Kay paper Ep 96.8, it is the optimal pre-fusion stabilized SARS-CoV-2 spike protein;  for the two other papers, it is also codon optimized, but not immediately clear if it is also the pre-fusion stabilized form).  In the three cases different types of lipo-nanoparticles are used as formulation and adjuvant.   In all cases, strong immune responses with neutralizing antibodies and a Th1 cellular response are induced in mice.  The Erasmus paper also shows that good titers of neutralizing antibodies are induced in macaques with either one of two doses of the sa-RNA (slide 19) and in the de Alwis paper protection against infection is shown in hACE-2 mice after a single dose (slide 20).


  1. Nanoparticle Protein Vaccines Based on the Receptor Binding Domain (RBD) and Heptad Repeat (HR) (Ep. 96.11): in this paper ferritin from H pylori is used as a self-assembling 24-mer to couple both the RBD and the HR (in S2). This construct induces strong neutralizing antibodies and Th1 responses, much better than RBD-HR monomers.  After a prime-boost regimen hACE2 mice are protected from a deadly challenge. Interestingly, there is also cross-neutralizing activity in vitro against other beta-Coronaviruses, based on the conservation of the HR part.  Could this type of construct evolve into a pan-beta Corona virus vaccine?  



Preliminary conclusions:


  1. The threat of escape mutations is already emerging, independent from, but only days after the first approved vaccine has been given to people.  Clearly, this aspect will have to be carefully monitored and anticipated in follow-up vaccine development.


  1. The cumbersome logistics of the first mRNA vaccines (stringent cold chain, two doses, high pricing) will have to be addressed if we want to create herd immunity over the globe.
  • Cheaper and less complicated vaccines (inactivated virus; adjuvanted trimeric S protein) are being tested, but those will presumably still require two doses.  
  • Janssen is already comparing one and two doses of Ad26 in its ongoing phase 3 trials


  1. From the various live attenuated virus candidates discussed in this episode, the S0-recombinant YF17 construct seems the most promising, but other candidates, including the MeV, (inactivated) rabies or even stringently attenuated SARS-CoV-2 itself (e.g. combination of cold adaptation and codon pair deoptimization) could be further improved.  This type of vaccines might be easier in terms of cold chain, more likely to be long-term efficacious, even after one dose and (in case of YF, MeV and rabies) constitute a “double vaccine” (i.e. also protecting against the wild type “vector” virus). Clearly, safety issues will have to be addressed carefully, especially in people with underlying immune-deficiency, immune-senescence and other vulnerable populations (e.g. pregnant women). 


  1. The role of “self-amplifying” RNA is presently not clear.  The most important potential advantage could be that 1 dose may suffice, but this has yet to be shown.  For the rest, there are no clear advantages, as sa-RNA vaccines will have to be formulated in similar “lipo-nanoparticles” as the “simple” mRNA vaccines, with the same risk on side effects (including anaphylactic shock). The inclusion of an additional protein (the alpha-viral replicase) could have unexpected drawbacks as well.


  1. Similarly, the  described ferritin nanoparticle protein vaccine and other “virus-like particles” could certainly be more potent than “simple” S proteins.  They may be preferred, if they would require only one dose and the additional proteins and adjuvants do not create novel disadvantages.


To be continued in 2021 (in life and well-being)


Keep safe and COVID-free, if possible.


Best wishes for X-mas