What is the source of virus causing prolonged viremia?


Pregnant macaque 827577 has been viremic for >28 days as of this writing (compared to ~10 days in non-pregnant macaques). Her viral loads will be measured again today at day 35. Identifying the source of the virus leading to this prolonged viremia is an important question. There are two basic explanations we have considered:

  • 827577 is immunosuppressed due to pregnancy and does not mount an effective immune response against Zika virus. Thus, the virus is able to persist longer than in non-pregnant, immunocompetent macaques
  • During initial viremia of 827577, Zika virus infected the fetus. Zika virus is replicating in the fetus and being shed back into 827577's bloodstream

These results are still preliminary, but two lines of experimental evidence support the second explanation.

1. Sustained virus detection in 827577 is confined to the blood

In other Zika virus infected macaques, virus is detected in urine beginning a few days after it is detected in blood. Resolution of viruria happens at about the same time as virus can no longer be detected in blood. In 827577, urine is cleared with typical kinetics, even though plasma viremia is sustained. If 827577 was immunosuppressed, we would expect to see continued shedding in multiple fluids due to incomplete immunologic containment.

2. Viral variant compartmentalization

We recently obtained viral sequence from days 2, 7, and 21 post-infection in 827577 [note: we also have day 5 sequence but have reason to believe this sample may have been accidentally switched with another sample on the same sequencing run]

For those interested in the raw data, as well as the Zequencer output files:

Day 0 (challenge stock) R1_FASTQ | R2_FASTQ | BAM (BAI - FASTA) | VCF





High quality variants (>5% of reads; p<10e-60; strand-bias 10e-5 when exceeding 65% bias) were identified in each dataset relative to the challenge stock consensus sequence [Geneious | Genbank] using the Zequencer workflow in Geneious Pro 9.1.2.

There are 9 high quality variants at day 7. I tabulated the frequencies of these exact same variants at the other timpeoints like this:

daysTrack NameMinimumChangeVariant FrequencyAmino Acid ChangeProtein Effect
0Variants: 827577-inocula (without duplicates)3,219T -> C1.40%L -> SSubstitution
2Variants: 827577-Day2 (without duplicates)3,219T -> C1.60%L -> SSubstitution
7Variants: 827577-Day7 (without duplicates)3,219T -> C18.80%L -> SSubstitution
21Variants: 827577-Day21 (without duplicates)3,219T -> C0.04%L -> SSubstitution

 If a particular variant is not detected at all at a given timepoint, the variant frequency is set to 0.00%, like this:

daysTrack NameMinimumChangeVariant FrequencyAmino Acid ChangeProtein Effect
0Variants: 827577-inocula (without duplicates)4,502G -> C39.80%E -> QSubstitution
2Variants: 827577-Day2 (without duplicates)4,502G -> C32.30%E -> QSubstitution
7Variants: 827577-Day7 (without duplicates)4,502G -> C9.20%E -> QSubstitution
21 4,502 0.00% synthetic variant

 I then plotted the frequency of high-quality variants at day 7 throughout infection of 827577. Each colored line in the following plot represents a separate high quality variant. Most of the high quality variants detected at day 7 were also present at day 2. Strikingly, none are present at >0.3% of reads at day 21. So these variants, which were enriched during the first 7 days of replication in 827577 all essentially disappear in the virus sequenced at day 21.

I repeated the analysis using the high quality variants detected at day 21. The reciprocal occurs - variants observed at day 21 are almost entirely absent from the day 7 virus. If the day 21 viremia was due to a continuation of the infection from day 7, this would be highly unlikely (we are working with colleagues to put a number on exactly how unlikely). But this is exactly what you might expect if during early maternal viremia, a small number of viruses infect the fetus and replicate as a compartmentalized subpopulation.

This is still a preliminary analysis. There are a few important caveats:

  • We did not have complete genome coverage in all datasets. Therefore we limited our analysis to nucleotides 308-4597 where we have high quality coverage in all the datasets.
  • The day 5 dataset was excluded because it appears to have been switched with another sample on the same instrument run. Even though the results presented here make sense, we are wary of any sequencing experiment with known sample mix-ups
  • Each sample was only sequenced in a single replicate. We would like to perform duplicate or triplicate technical replicates, but sample availability is limiting
  • More timepoints will provide better variant resolution. Day 28 and 35 timepoints, as well as selected earlier timepoints are in the process of being sequenced

Nonetheless, the two complementary types of data both pointing to fetal infection and shedding into the mother are provocative.