3 Jan 2021 Episode 97 Superspreading

Sun, 01/03/2021 - 20:21

Dear colleagues,

Did you spend relaxing holidays and are you now ready for the “final fight” against SARS-CoV-2?  I hope so!   I’m on your side (from a safe distance).

I’d like to pick up the story of the “superbugs" from UK and South Africa (SA): what is their significance and how can we explain their sudden appearance. I also look back to other (super)spreading stories from last year in Yokohama (Diamond Princess), Boston and Hunan, on which nice systematic studies have been published recently.

  1. The starting point is a short comment by Kai Kupferschmidt (Ep 97 1): he reiterates on the characteristics of both recent UK and SA variants.  The common mutation is N501Y in the spike, which is believed to increase the binding to the ACE2 receptor.  The SA variant has two other mutations in the RBD: K417N and E484K, which may further increase the binding to human cells and promote transmission.  The UK variant, on the other hand has a deletion 69-70del, of which it has already been shown by Gupta et al that it increases the infectiousness by a factor 2. The third mutation  P648H may alter the S1-S2 cleaving site, which is a crucial determinant of the higher infectiousness of SARS-CoV-1 as compared to SARS-CoV-1.  For sure, we will read about the effect of these single and combined mutations on “fitness” in engineered viruses soon.
  2. Kupferschmidt refers to two recent papers (Ep97 2 and 3), where various mutations occurred in heavily immune suppressed subjects with chronic recurrent SARS-CoV-2 infection and each treated with several courses of Remdesivir and either Regeneron (RBD-specific  monoclonal Ab cocktail) or convalescent plasma.
  • The first patient (Ep 97 2) only received the Regeneron therapy very late in the disease (about 10 days before he died), but nevertheless a large number of mutations in the spike and RBD already occurred earlier.  These mutations included N501Y and a cluster of mutations in the 480-495 positions.  Also, a few mutations were present in ORF1a (containing non-structural proteins), but NOT in the RNA-dependent RNA polymerase (RdRp - the molecular target of Remdesivir). Therefore, the mutations in spike could be rather interpreted as “fitness-promoting” in the hostile environment of a (weak) immune system and courses of Remdesivir.  Remarkably however, several of these mutations (484K, 493K, 494P) have been associated with reduced sensitivity to the Regeneron monoclonals (Ep 97 4 Table 2), which might explain why the Regeneron therapy did not save the patient(?).
  • In the second patient (Ep 97 3) Remdesivir treatment was also followed by a (temporal) emergence of N501Y and  a V157L mutation on RdRp.  After courses of convalescent plasma large, dynamic virus population shifts, with the emergence of a dominant viral strain bearing D796H in S2 and deletion H69/V70 in the S1 were observed.  In this case D796H mainly contributed to escape, while 69/70 deletion was compensatory, since on itself it increased viral replication. In this case, the sensitivity of the virus to RBD-specific monoclonals was not altered.


  1. I wondered whether this N501Y mutation has been described earlier and, sure enough, I found a nice Sept Science paper by Hongjing Gu et al, in which the authors succeeded to adapt SAR-CoV-2 to productively infect and even cause lung pathology as well as systemic inflammation in regular Balb/c mice after 6 passages.  Their purpose was to create a more convenient model for vaccination studies.  The “adapted” virus contained just 4 non-synonymous mutations: 2 in ORF1ab, 1 in Nuncleocapsid and 1 in Spike: N501Y ! These mice could be efficiently vaccinated with a RBD-Fc construct, which is somehow reassuring.


  1. Then I wondered, would one of the other mutations also occur in human infections?  And, yes, I ended up in Yokohama, on board of the Diamond Princess, where Tsuyoshi Sekizuka (Ep 97 6) found that all isolates contained the G11083T transversion, which implies a substitution of Leu to Phenylalanine at position 37 of the non-structural protein 6 .  Clearly, this finding suggests that the fast spreading on board of this cruise ship really constitutes a superspreading event.  As we know, the epidemic was contained by progressively stricter quarantine  and finally 712 were infected and 13 died out of 3600 passengers and crew.  This case fatality rate of 2 % is not very high in view of the fact that the passengers were rather older (most between 60 and 80 Ep 97 7), which seems rather favorable.  In this study, the proportion of asymptomatics was estimated between 20 and 40 %. .
  2. A further in depth analysis of the Diamond Princess (DP) cohort revealed that 74 % (CI 70-78) of all infections proceeded asymptomatically and that (consequently?) asymptomatic individuals were also the source for 69 % (with a wide CI 20-85) of all infections (Ep 97 8).  This study was a modeling exercise. I could not find formal serology studies in the entire DP cohort.  Nevertheless, there are two reports on a subset of the DP passengers, who were repatriated to Hong Kong( n=215, of which 9 infected)  and Israel (n = 6), wherein it was shown that asymptomatic patients indeed were shedding virus and infected their close contacts.  All virus(+) subjects also seroconverted, some asymptomatics or paucisymptomatic patients had radiological evidence of pneumonia  (Ep 97 9-10)   
  3. I also found a paper about the function of nsp6 and the possible significance of the L37F (G11083T) mutation (Ep 96 11) It appears that the function of nsp6 is to undermine the capability of autophagosomes to transport viral components to lysosomes for degradation, hence indirectly favoring viral replication.  The L37F mutation results in destabilization of nsp6, hence potentially reduced virulence. Analyzing over 20,000 sequences over time, some with clinical info, they conclude that this SNP (single nucleotide polymorphism) is rather associated with asymptomatic infection(which may explain the large proportion of asymptomatics on board of DP), but also that it’s prevalence decays, suggesting reduced transmission(which at first view doesn’t seem consistent with the high “attack rate” at DP.  


  1. The story about the 2 Boston superspreading events has also formally been published (Ep 97 12) in Science. The two events are quite different:
  •  The most important superspreading took place in a conference with a strain labeled C2416T (referring to polymorphism in ORF1a) and  a sublineage which occurred during or shortly after the conference C2416T/G26233T (with an additional polymorphism in the E protein).  These polymorphisms are now present in over 33 % of all genomes in the Boston area and in a total of 245,000 cases in the US, from where they have also spread worldwide. 
  • A second event took place in a nursing home, where 128 residents and staff were infected.  75/83 sequences done were in a single G3892T  (in ORF1a) cluster with 59 identical sequences.  24 of these residents died!!  Nevertheless this cluster spread not outside this particular nursing home.

So clearly, the superspreading virus from the conference seems to be very infectious, while that from the nursing home seems rather highly pathogenic (at least in elderly).  I have been looking for a pathogenic explanation for these events, but I haven’t found it.


  1. Finally, a very nice and rich analysis of the early epidemic in Hunan (Ep 94 13).  Some (of many) interesting observations before and during a lockdown period.  
  • Most infected subjects failed to transmit, 15 % of all infected subjects account for 80 % of transmissions.  This is very visually shown in fig 1 p. 10.
  • Children (0-12) were significantly less and older adults (> 65) significantly more susceptible than young and middle aged adults (24-64).
  • Chances of transmission were highest in households > extended family > social and community contacts > health care contacts.   
  • Presymptomatic infections accounted for a majority of transmissions.
  • Case isolation and quarantine successfully reduced transmission, but needs to be complemented with physical distancing (and face masks).
  • Testing and tracing, aided with localization technology is key to ease lockdown.


All very logical, but it has been very cumbersome to apply in the West !!!


Some preliminary conclusions:  


  1. N501Y mutation can arise in the context of chronic infection of immuuncompromised people and by repeated passaging in originally non-suscpeptible Balb/c mice: increased affinity to both human and mouse ACE-2?


Other associated mutations include

  • in S1: 69/70 deletion (increased “fitness”), K417N, mutations in the range 480-495 (escape from antibodies).
  • Cleaving site S1-S2: P648H
  • S2: D796H (escape from antibodies)


It remains unclear whether these will influence vaccination.


  1. Superspreading in Diamond Princess: linked to G11083T (Leu37Phen): this mutation cripples viral nsp6.  This may explain high proportion of asymptomatic infections on board DP and low mortality in elderly population?  However, this variant seems to be competed out in the global epidemic.  


  1. Two superspreading events in Boston:
  • C2416T linked to conference contributes massively to, Boston, US and even worldwide epidemic.  Hence very infectious.
  •  G3892T  linked to nursing how.  Very deadly in elderly, but failed to spread much outside.  Poorly infective, but highly pathogenic?


  1.  Epidemiology Hunan: confirms that
  • Superspreading is important
  • Households are important site of transmission (especially during lockdown)
  • Young children less and elderly more susceptible to infection.



Best wishes,