Однако, в природе могут случиться ситуации, когда в одной клетке могут встретиться вектор и аденовирус (например, в лимфоците, зараженном аденовирусм (если у человека есть аденовирусная инфекция на момент прививки), который прибежал в место укола). И тогда уже возможны варианты, на мой взгляд, необычные.
SARS-CoV-2 vaccine ChAdOx1 nCoV-19 infection of human cell lines reveals low levels of viral backbone gene transcription alongside very high levels of SARS-CoV-2 S glycoprotein gene transcription
ChAdOx1 nCoV-19 is a recombinant adenovirus vaccine against SARS-CoV-2 that has passed phase III clinical trials and is now in use across the globe. Although replication-defective in normal cells, 28 kbp of adenovirus genes is delivered to the cell nucleus alongside the SARS-CoV-2 S glycoprotein gene.
The expected SARS-CoV-2 S coding transcript dominated in all cell lines. We also detected rare S transcripts with aberrant splice patterns or polyadenylation site usage. Adenovirus vector transcripts were almost absent in MRC-5 cells, but in A549 cells, there was a broader repertoire of adenoviral gene expression at very low levels. Proteomically, in addition to S glycoprotein, we detected multiple adenovirus proteins in A549 cells compared to just one in MRC5 cells.
Looking quantitatively, Table 2 shows that in A549 cells a large number of different ChAdOx1 vector backbone genes were expressed to one degree or another with the adenovirus protein DBP being dominant. For MRC-5 cells, transcripts that code for DBP and E4 ORFs 2 and 3 could be detected, but only at 48 hpi. However in A549 cells, in addition to DBP/pTP coding E2 transcripts, transcripts that code for all the E4 ORFs except E4 ORF6 could be detected, which could explain the detectable levels of major late transcript proteins in A549 cells. Notably, ChAdOx1 vector backbone gene expression declined over the time course analysed. Comparing the expression of ChAdOx1 gene expression in non-permissive vs permissive HEK293 cells illustrated how deletion of the E1 region affects ChAdOx1 gene expression in A549 cells where some vector backbone gene expression has occurred (Table 2). One key difference was the level of expression of the fibre (L5) transcript which was the dominant transcript in HEK293 cells (9.65% of transcripts mapped to ChAdOx1 nCoV19) after the S glycoprotein transcript but is expressed only at very low levels in A549 cells
Furthermore, transcripts that code for all the E4 region ORFs were readily detected, indicating that replacing the Y25 ORFs 4, 6 and 6/7 with their HuAd5 equivalents did not ablate the expression of the remaining E4 ORF transcripts
In both permissive and non-permissive cell lines, the most abundant ChAdOx1 nCoV-19 virus vector transcript, as expected, was the transcript for expression of the SARS-CoV-2 S glycoprotein. For MRC5 cells, the S transcript accounts for over 90% of transcripts from the ChAdOx1 nCoV19 genome; this is reduced to between 75 and 90% in A549 cells. However, in HEK293 cells (where there is productive viral replication), only 17% of transcripts code for the S glycoprotein relative to transcripts that code for adenovirus ORFs. This reflects the large numbers of adenovirus transcripts generated during productive replication. However, a number of transcripts were identified in all cell lines at low levels which appear to arise from aberrant splicing events. We also found evidence that occasionally some transcripts were spliced to other elements (notably pIX); these would still code for SARS-CoV-2 S glycoprotein but would be polycistronic
We also saw limited evidence of transcripts that extended beyond the pIX polyA signal and deep into the ChAdOx1 vector backbone in HEK293 cells when there was active viral replication. However, it is important to place these rare transcripts in context—the vast majority of transcripts starting at the transcription start site for the S glycoprotein message would generate mRNA that codes for S glycoprotein
Typically, in human adenoviruses, the late transcripts code for most of the structural viral proteins. The late transcripts typically originate at the major late promoter and include three exons (known as the tripartite leader) and optional i-leader exon and are grouped into five classes called L1 to L5 based on which of the five major polyadenylation sites they use [32, 33]. The transcript for preVI is usually regarded as part of the L3 group of transcripts which share a common polyadenylation site but utilise different splice acceptor/donor pairs to place one of three different ORFs (preVI, hexon or 23K) proximal to the 5′ cap for translation. In the case of ChAdOx1 replicating in HEK293 cells, although there were transcripts coding for preVI that would fit this pattern (e.g. utilising the same polyadenylation site as transcripts for hexon and 23K)the dominant transcript for preVI instead utilises an additional polyadenylation site upstream of the start codon for the hexon protein (Fig. 3a). This was also the case in A549 cells (Fig. 1b) where 12 transcripts utilised this additional polyadenylation site compared to just 4 using the classical L3 polyadenylation site (Additional file 5: Table S2). Examining the sequence of the viral genome in the region of this additional polyadenylation site reveals a GU-rich region preceded by a classical polyadenylation signal  (Fig. 3b)—this signal is not present in the equivalent region of HuAd5 (Fig. 3c) where no such preVI-specific polyadenylation site has been observed . However, this additional polyadenylation site is present in human adenovirus type 4 (HuAd4, Fig. 3c) which is also a member of the group E mammalian adenoviruses like chimpanzee adenovirus Y25 from which ChaAdOx1 was derived.
As expected, we were able to detect a range of S glycoprotein peptides in MRC5 and A549 cells as well as one phosphorylation site (Ser 1292) derived from the C-terminal portion of the glycoprotein which is internal to the viral particle
For the vector backbone expressed proteins, we were only able to detect E4ORF3 in MRC5-infected cells whereas in A549 cells in addition to E4ORF3 we also detected DBP and hexon
The proteomic analysis supported the transcriptomic analysis suggesting that the ChAdOx1 nCoV-19 vaccine does not make additional unexpected proteins. Notably, in both A549 and MRC5 cells, we observed some of the highest fold increases over time for the S glycoprotein as expected
Despite their widespread development, this is the first study to directly and comprehensively survey the transcriptomic repertoire of a replication-defective adenovirus in a non-permissive host cell. A significant advantage of using simian-based adenoviruses like ChAdOx1 (a species E non-human adenovirus) to study the transcriptome of E1-deleted adenoviruses is the absence of replication-competent adenoviruses (RCAs) . RCAs can arise from homology between the HuAd5 (a species C human adenovirus) sequences present in the HEK293 cells used to produce the vaccine vector which extend beyond the E1 region into the pIX gene (as shown in Fig. 1e) and sequences present in the recombinant vector itself (e.g. pIX gene in the vector), which are frequently based on HuAd5. Whilst the emergence of an RCA could confound such an analysis in other adenoviral backgrounds, we have never detected RCA in any preparation of ChAdOx1-based vaccine vectors. This is presumably due to insufficient homology between the species E chimpanzee adenoviruses and the sequences in HEK293 cells
We were especially keen to determine that no additional unanticipated transcripts or proteins were being made. Any one of such proteins could be antigenic with unintended consequences (e.g. generating auto-immune responses for example).The two host cell lines chosen for this study have distinct properties despite both being derived from male human lungs.
In addition, NFκB has been implicated in enabling HuAd5-based vector backbone expression in non-permissive cell lines . However, no marked differences were observed between the two cell lines in the transcriptomic abundance data for NFKB1 in our datasets
A deeper understanding of the intracellular environments that allow the ChAdOx1 vector to overcome the lack of E1A, even in a limited fashion, could lead to better ways to prevent this which in turn may lead to improved transgene expression. The transcriptome and proteome of the two cell lines appeared to respond differently to the vaccine vector. However, whether these differences are driven by or a result of the distinct adenovirus vector expression patterns observed on the vector genome is yet to be determined.
That we observe quite distinct vaccine backbone expression in cell lines from the same tissue site is notable. However, care is needed not to overinterpret this study which was designed to investigate the potential repertoire of transcription from a respiratory virus-based vaccine vector in two different human respiratory cell lines. This vaccine vector is currently administered intramuscularly (other routes of administration including respiratory ones may also prove effective longer term), and so in further analysis, it might be useful to examine primary cells and/or cell lines derived from muscles or even biopsy material. In addition, there is clearly the potential for other cell types to become infected after intramuscular administration that may also be relevant to the adaptive immune response.
We have previously analysed the proteome of purified adenovirus particles, both wild type HuAd5- and HuAd5-based recombinant vaccine vectors, including a sample of adenovirus manufactured to clinical grade. We were able to detect non-structural proteins DBP, 100K and E4 14.7K protein in purified virus particles in addition to the expected viral structural proteins . If any cells in a vaccinee did express the full range of adenovirus proteins over time similar to the A549 transcriptomic profile, then T-cell responses to the virus vector could both derive from the incoming viral proteins and from subsequent low-level expression of any of the remaining virus vector backbone genes. That some cells allow low-level expression of ChAdOx1 vector backbone transcripts would be consistent with data from human adenovirus vector studies where additional deletions in E2 and E4 or helper-dependent adenovirus vectors (where essentially all the vector backbone genes are removed) were shown to afford longer transgene expression in vivo [13, 15, 42,43,44]. The continuing low-level expression of vector backbone genes is therefore likely to be the main driver of immune-mediated clearance of cells infected with E1/E3-deleted adenoviruses. The repercussions in a vaccinee of broad low-level expression as observed in A549 cells is difficult to evaluate at this stage, but in principle, it suggests that repeated re-administration of a particular vaccine vector beyond the usual prime-boost vaccination regimen should be examined in further detail. In addition, low-level expression of vector backbone genes in A549 cells may be related to our finding that A549 cells express lower levels of S glycoprotein mRNA as a proportion of total mRNA than MRC5 cells
Our proteomics analysis is consistent with the transcriptomics data in that the S glycoprotein is readily detected by MS/MS and that we detect a slightly wider array of ChAdOx1 proteins in A549 cells compared to MRC-5 cells. We also detect evidence of phosphorylation of the S glycoprotein as we have noted previously in our work on SARS-CoV-2 ; the significance of this phosphorylation is unclear at this time. We do not detect any evidence of additional unexpected proteins being expressed by the ChAdOx1 nCoV19 vaccine despite a broad search of proteins coded for by all transcripts that map to the ChAdOx1 nCoV-19 genome.
We previously analysed the transcriptomic repertoire of wild type HuAd5 in MRC-5 cells showing that during a productive infection there is significant low-level heterogeneity in the usage of both splice sites and polyadenylation signals . Here, we show that as with the human virus there is similar low-level expression of transcripts with aberrant splice site and polyadenylation signal usage in cell lines infected with the recombinant ChAdOx1 nCoV-19 virus in replication permissive and non-permissive cell lines. Critically, the overwhelming majority of S glycoprotein transcripts have the expected structure with only a small minority of transcripts originating from the S glycoprotein transcription start site being unable to express the S glycoprotein. The most common issue seems to be failure to use the polyadenylation signal immediately after the S glycoprotein ORF. This leads to, for example, usage of the pIX polyadenylation site and the opportunity for splicing events that may disrupt S glycoprotein expression. Other S transcripts with aberrant splice patterns or polyadenylation site usage were detected, but these were minor compared to the dominant transcript and we did not observe peptides that could arise from such aberrant transcripts. In future vector designs, this approach should highlight such issues potentially allowing changes to vector design to minimise issues identified.
Notably, in our analysis of wild type HuAd5 transcripts, we proposed that a key aspect of HuAd5 evolution would be the exploration of different splice acceptor/donor sites and initiating codon combinations. Here, analysis of expression from the ChAdOx1 backbone identified alternative usage of polyadenylation signals.
Examination of the HEK293 cell data revealed, as expected, expression of HuAd5 transcripts corresponding to E1 region genes dominated by E1b19K, E1a12S and E1a13S as we saw for wild type HuAd5 gene expression at 16hpi infection . As noted in the results, in HEK293 cells, we observed a small number of HuAd5 pIX transcripts, approximately 100-fold fewer than ChadOx1 pIX transcripts. However, with 240 copies of protein IX per virus capsid [46,47,48], it is possible that the ChAdOx1 virus particles grown in HEK293 cells (or derivatives) contain one or two copies of HuAd5-pIX per viral particle assuming the two proteins are equivalently interchangeable on the virus particle.