Rapid localized spread and immunologic containment define Herpes simplex virus-2 reactivation in the human genital tract

6 figures, 9 videos and 6 tables

Figures

Figure 1 with 3 supplements
HSV-2 levels in the genital tract are stable over minutes, expand and decay markedly over hours, and fluctuate rapidly and unpredictably over days.

(A) Shedding quantity in a participant, who performed genital swabs every 5 min over 4 hr during a lesion, reveals low swab-to-swab variation in viral quantity. Using data from panel (A and B), …

https://doi.org/10.7554/eLife.00288.007
Figure 1—source data 1

Source data for Figure 1, Figure 1—figure supplement 1, Figure 1—figure supplement 2 and Figure 1—figure supplement 3.

https://doi.org/10.7554/eLife.00288.008
Figure 1—figure supplement 1
Dynamics of HSV-2 shedding over 5-min time intervals.

(A) and (B) Shedding quantity in two participants, who performed genital swabs every 5 min over 4 hr during a lesion, reveals low swab-to-swab variation in viral quantity. (C) Using data from panels …

https://doi.org/10.7554/eLife.00288.009
Figure 1—figure supplement 2
Dynamics of HSV-2 shedding with every 2-4 hr sampling.

Shedding quantity in four participants, who performed 10 genital swabs per day over 4–5 days during a lesion reveal a characteristic saw-tooth pattern; arrows denote re-expansion. Participants had …

https://doi.org/10.7554/eLife.00288.010
Figure 1—figure supplement 3
Dynamics of HSV-2 shedding with every 6-hr sampling over 8 days.

Shedding quantity in episodes detected in a participant who performed four genital swabs per day over 60 days demonstrates that viral re-expansion allows for shedding prolongation.

https://doi.org/10.7554/eLife.00288.011
Figure 2 with 3 supplements
HSV-2 replicates and is contained in widely dispersed microenvironments across the genital tract.

(A) HSV shedding quantity in a participant, who underwent daily swabs in 23 regions across the genital tract for 30 days; days without sampling are marked with an X; stars denote days with a lesion; …

https://doi.org/10.7554/eLife.00288.013
Figure 2—source data 1

Source data for Figure 2 and Figure 2—figure supplement 1.

https://doi.org/10.7554/eLife.00288.014
Figure 2—figure supplement 1
Spatial features of HSV genital tract shedding.

HSV shedding quantity in a study participant, who underwent daily swabs in 23 regions across the genital tract for 30 days; days without sampling are marked with an X. The participant had three …

https://doi.org/10.7554/eLife.00288.015
Figure 2—figure supplement 2
Spatial features of HSV-2 lesions.

A genital lesion consists of numerous round ulcers or vesicles (black dotted circle), clustered in space.

https://doi.org/10.7554/eLife.00288.016
Figure 2—figure supplement 3
Spatial features of CD8+ T-cell response in genital skin.

Immunofluorescent staining of a biopsy performed (A) at the edge, and (B) 1 cm away from an ulcer 3 days post-healing. CD8+ T-cells (red) and CD4+ T-cells (green) at the dermal epidermal junction …

https://doi.org/10.7554/eLife.00288.017
Figure 3 with 1 supplement
Mathematical model.

(A) Microregions are linked virally because cell-free HSV-2 can seed surrounding regions, and immunologically based on overlapping CD8+ T-cell densities between regions (not shown). (B) Schematic …

https://doi.org/10.7554/eLife.00288.019
Figure 3—figure supplement 1
Spatial mathematical model.

Viruses produced from neurons (green), cell-associated viruses from epidermal cells (yellow), and cell-free viruses (orange) that form after rupture of epidermal cells, are distinguished in the …

https://doi.org/10.7554/eLife.00288.020
Figure 4 with 1 supplement
The spatial model reproduces all shedding episode characteristics.

Colored bars represent results from (A) 14,685 genital swabs and (BG) 1020 shedding episodes from 531 study participants. The model simulation, represented with black bars in each panel, continued …

https://doi.org/10.7554/eLife.00288.021
Figure 4—figure supplement 1
Continuous sampling of spatial model output reveals more accurate measures of episode characteristics.

We subjected a 30-year simulation to daily and continuous sampling. (A) Median initiation to peak slope, and (B) peak to termination slopes increased substantially with continual sampling. (C) …

https://doi.org/10.7554/eLife.00288.023
Containment of infected cells within a single ulcer is extremely rapid, although secondary ulcers explain prolonged episodes.

(A) Episode duration was a function of the number of ulcers before episode termination during 500 simulated episodes. (B)–(E) A 10-day simulated episode consisting of 24 ulcers: (B) Total cell-free …

https://doi.org/10.7554/eLife.00288.026
Figure 6 with 2 supplements
Random spatial dispersion of viral particles from neurons reproduced the full diversity of episode characteristics if particles were released continuously, daily, or weekly from neurons.

White circles represent results from (A) 14,685 genital swabs and (BG) 1020 shedding episodes from 531 study participants (Figure 4). The model simulations represented with colored bars in each …

https://doi.org/10.7554/eLife.00288.036
Figure 6—figure supplement 1
Random spatial dispersion of viral particles from neurons reproduced the full diversity of episode characteristics during simulations in which particles were released into only a minority of the 300 modeled regions.

White circles represent results from (A) 14,685 genital swabs and (BG) 1020 shedding episodes from 531 study participants (Figure 4). The model simulations represented with colored bars in each …

https://doi.org/10.7554/eLife.00288.037
Figure 6—figure supplement 2
Dispersion of viral particles from neurons reproduced the full diversity of episode characteristics during simulations in which particles were released into a minority of modeled regions, provided that dispersion was random rather than clustered within a single geographic region.

White circles represent results from (A) 14,685 genital swabs and (BG) 1020 shedding episodes from 531 study participants (Figure 4). The model simulations represented with colored bars in each …

https://doi.org/10.7554/eLife.00288.038

Videos

Video 1
Spatiotemporal demonstration of a 10-day episode.

The left panel represents total cell-free HSV DNA copies per milliliter present over time. The right panel represents spatial spread of virus during the episode, each hexagon contains one region of …

https://doi.org/10.7554/eLife.00288.025
Video 2
Spatiotemporal demonstration of a 14-day episode according to viral production within each single region.

The left panel represents cell-free HSV DNA measured over time with each region's production demonstrated with a different color. The right panel represents spatial spread of virus during the …

https://doi.org/10.7554/eLife.00288.027
Video 3
Spatiotemporal demonstration of infected cell and viral spread during a 6-day episode.

The upper left panel represents spatial spread of cell-free virus during the episode; the upper right panel represents spatial spread of cell-associated virus during the episode; the bottom left …

https://doi.org/10.7554/eLife.00288.028
Video 4
Spatiotemporal demonstration of 365 days of simulated shedding.

The left panel represents total cell-free HSV DNA copies per milliliter present over time. The right panel represents spatial spread of virus during the episode; each hexagon contains one region of …

https://doi.org/10.7554/eLife.00288.029
Video 5
Spatiotemporal demonstration of 365 days of simulated shedding with noninfectious cell-free particles.

The left panel represents total cell-free HSV DNA copies per milliliter present over time. The right panel represents spatial spread of virus during the episode; each hexagon contains one region of …

https://doi.org/10.7554/eLife.00288.030
Video 6
Spatiotemporal demonstration of immune response to a pair of 2-day episodes.

The upper left panel represents total cell-free HSV DNA copies per milliliter present over time. The upper right panel represents spatial spread of cell-free virus during the episode. The lower left …

https://doi.org/10.7554/eLife.00288.032
Video 7
Spatiotemporal demonstration of immune response to a 7-day episode.

The upper left panel represents total cell-free HSV DNA copies per milliliter present over time. The upper right panel represents spatial spread of cell-free virus during the episode. The lower left …

https://doi.org/10.7554/eLife.00288.033
Video 8
Spatiotemporal demonstration of immune response to a 14-day episode.

The upper left panel represents total cell-free HSV DNA copies per milliliter present over time. The upper right panel represents spatial spread of cell-free virus during the episode. The lower left …

https://doi.org/10.7554/eLife.00288.034
Video 9
Spatiotemporal demonstration of immune response over 20 years.

The left panel represents CD8+ T-cell density within each region. The right panel indicates reproductive number within each region. Quantities are displayed according to a heat map adjacent to each …

https://doi.org/10.7554/eLife.00288.035

Tables

Table 1

Five cohorts of HSV-2 genital tract shedding

https://doi.org/10.7554/eLife.00288.003
CohortSubjectsTotal swabsSwabbing frequencyTotal episodesSwabbing durationAnatomic swabbing regionPurpose
A396Every 5 min34 hr when lesion presentTotal genital tractSwab-to-swab sampling/assay variability
B520010 times/day (every 2 hr during the days and 4 hr overnight)54–5 days when lesion presentTotal genital tractEpisode expansion, clearance and re-expansion kinetics
C2547064 times/day10930–60 days without or with a lesionTotal genital tractAccurate estimates for expansion/decay slopes for clinical and subclinical episodes
D2216Daily430 days without or with a lesion23 separate regionsSpatial dispersion of HSV
E53114,685Daily1020>30 days with or without a lesionTotal genital tractModel fit
Table 2

Parameter ranges that result in accurate reproduction of model outcomes

https://doi.org/10.7554/eLife.00288.024
ParameterUnitsSymbolBest fit valueGood fitAverage fit
Lower limitUpper limitLower limitUpper limit
Cell-associated HSV infectivityDNA copy days/cell (viruses needed per day to infect one adjacent cell)βi5.4e−8 (111)4.86e−8 (123)7.83e−8 (76)3.78e−8 (158)1.32e−7 (45)
Cell-free HSV infectivityDNA copy days/cell (viruses needed per day to initiate one new ulcer)βe2.65e−11 (2.26e5)1.73e−11 (3.46e5)2.78e−11 (2.15e5)3.98e−12 (1.50e6)5.04e−11 (1.19e5)
Epidermal HSV replication rateHSV DNA copies per cell per dayp1.03e57.21e41.7e55.15e41.96e5
Neuronal release rateHSV DNA copies per day per genital tractϕ82459041123
Free-viral decay ratePer day (half-life, hours)c8.8 (1.9)7.0 (2.4)9.7 (1.7)6.2 (2.7)12.3 (1.4)
Maximal CD8+ T-cell expansion ratePer dayθ2.841.853.271.855.25
CD8+ T-cell decay ratePer day (half-life, days)δ1.47e−3 (471)1.12e−3 (619)1.69e−3 (409)6.64e−4 (1040)2.21e−3 (314)
CD8+ T-cell local recognitionInfected cells at which θ is half maximalr423047474
CD8+ regional codependence0 = no codependence, 1 =full codependenceρ0.690.590.860.380.86
Viral production lagDaysε0.960.531.10.341.1
Table 3

Predictive model parameters for key model outcomes

https://doi.org/10.7554/eLife.00288.031
Single episode featuresLong-term shedding features
Peak viral loadDurationShedding rateEpisode rate
CD8+ T-cell density at reactivation site−0.56−0.47NANA
Cell-associated HSV infectivity0.120.13
Cell-free HSV infectivity
Epidermal cell replicate rate0.130.14−0.25−0.31
Neuronal release rate0.430.55
Free-viral decay rate−0.2
Maximal CD8+ T-cell expansion rate−0.090.370.51
CD8+ T-cell decay rate0.09−0.16
CD8+ T-cell local recognition
CD8+ regional co-dependence0.320.34
Viral production lag0.240.23
  1. Partial correlation coefficients are listed only for parameters that are found to improve predictive effect on outcomes using Akaike information criteria models. Episode features are from 500 single episode simulations. Long-term shedding outcomes were measured over 10-years during 500 simulations.

Table 4

Spatial model simulations that varied only according to duration of sampling (30 days, 60 days, 365 days, and 10 years)

https://doi.org/10.7554/eLife.00288.039
Simulation durationPercent of time with HSV DNA > 150 copies per mLPercent of time with lesions* presentEpisodes per yearLesions per year
30 dayMean13.77.6113.2
Median3.40120
Range0–82.80–58.80–36.50–24.3
60 dayMean19.09.813.44.4
Median18.63.4126
Range0–54.20–46.90–42.60–18
365 dayMean19.610.114.34.6
Median19.99.9144
Range7.1–36.82.3–228–241–9
10 yearMean17.59.114.54.3
Median17.69.014.54.2
Range14.8–19.81.9–12.811.5–17.11.4–6
  1. *

    Lesions were defined as > 1 mm diameter ulcers.

  2. Sixty simulations were performed at each of the sampling durations. Within shorter sampling duration simulations, lesion, and shedding frequency varied significantly, while ranges narrowed with prolonged sampling.

Table 5

Mathematical models of HSV-2 pathogenesis

https://doi.org/10.7554/eLife.00288.040
ModelEquations (additions to previous model are denoted in bold)Variables (model fitting variable)New features
1 (Schiffer et al., 2009; Schiffer et al., 2010)ΔS=[λ(βi×S×V)(βi×S×Vneu)]ΔtΔI=[(βi×S×Vi)+(βi×S×Vneu)(a×I)(f×I×E)]ΔtΔE=[(F(I)×θ×E)(δ×E)]ΔtΔVi=[(p×I)(c×Vi)(βi×S×Vi)]ΔtΔVneu=[φ(c×Vneu)(βi×S×Vneu)]ΔtF(I)=I/(I+r)Vtot=Vi+Vneuλ=d(SS0)S, I, E, Vi Vneu,
2ΔS=[λ(βi×S×V)(βi×S×Vneu)]ΔtΔI=[(βi×S×Vi)+(βi×S×Vneu)(a×I)(f×I×E)]ΔtΔE=[(F(I)×θ×E)(δ×E)]ΔtΔVneu=[φ(c×Vneu)(βi×S×Vneu)]ΔtΔVi=[(p×I)(a×Vi)(βi×S×Vi)]ΔtΔVe=[(a×Vi)(c×Ve)]ΔtF(I)=I/(I+r)Vtot=Vi+Ve+Vneuλ=d(SS0)S, I, E, Vi, Vneu, (Ve)Cell-free and cell-associated particles
3ΔS(i300)==[λ(βi×S×Vi)(βi×S×Vneu)(βe×S×Ve)]ΔtΔI(i300)=[βi×S×Vi)+(βi×S×Vneu)(βe×S×Vetot)(a×l)(f×l×E]ΔtΔS(i300)=[(F(I)×θ×E)(δ×E)]ΔtΔVneu(i300)=[φ(c×Vneu)(βi×S×Vneu)]ΔtΔVi(i300)=[(p×I)(a×Vi)(βi×S×Vi)ΔtΔVe(i300) = [(a×Vi)(c×Ve)]ΔtF(I) =I/(I+r)Itot=I1+I2++Ie300Vetot=Ve1+Ve2++Ve300Vitot=Vi1+Vi2++Vi300λ=d(SS0)S, I, E, Vneu, Vitot, (Vetot)Concurrent plaques from cell-free particles
4*S0=1.67 e5 per regionΔS(i300)=[λ(βi×S×Vi)(βi×S×Vneu)(βe×S×Ve)]ΔtΔI(i300)=[(βi×S×Vi)+(βi×S×Vneu)+(βe×S×Veadj)(a×I)(f×I×E)]ΔtΔE(i300)=[(F(I)×θ×E)(δ×E)]ΔtΔVneu(i300)=[φ(c×Vneu)(βi×S×Vneu)]ΔtΔVi(i300)=[(p×I)(a×Vi)(βi×S×Vi)]ΔtΔVe(i300)=[(a×Vi)(c×Ve)]Δtλ=d(SS0)F(I)=I/(I+r)Veadj=Vefrom 6 adjacent regionsItot=I1+I2++I300Vetot=Ve1+Ve2++Ve300Vitot=Vi1+Vi2++Vi300S, I, E, Vneu, Vitot, (Vetot)Spatial model
  1. *Model 4 has a parameter of regional CD8+ co-dependence (ρ) within each plaque-forming region. At episode onset within a region, the CD8+ density is adjusted to infer the spatial co-dependence of CD8 density from surrounding regions: Ei (time + 0.001) = (Ei × (1 − ρ)) + (Eavg × ρ) where Eavg is average of E from the 6 surrounding regions.

  2. Models are described in the ‘Methods’. Units and values of each parameter in the optimized model are listed in Table 2. Variables include: S (susceptible skin cells), I (infected skin cells), E (CD8+ T cells), Vi (cell-associated HSV DNA particles), and Ve (cell-free HSV DNA particles).

Table 6

Model fit to Cohort E

https://doi.org/10.7554/eLife.00288.041
Summary measureSummed criteria scoresBest fitting scoreAIC
Shedding frequencyEpisode durationFirst positive swabLast positive swabPeak positive swabMedian expansionMedian decayEpisode frequency
Model 12.135.590.050.080.240.070.430.018.61−50
Model 21.268.330.380.340.220.040.090.0310.69−41
Model 30.253.750.310.220.610.050.500.225.92−62
Model 4 solved for 10 parameters0.250.130.290.150.090.010.010.030.96−139
Model 4 solved for 5 parameters0.390.320.440.210.090.0400.181.67−125
  1. Summed criteria scores measure the degree of fit for each model according the eight individual shedding episode features using a weighted sum of squares. Model 4 is the spatial model. Models 1–3 are described in the ‘Methods’. Best fitting score is a sum of all summed criteria scores for a particular model with lower scores indicating better fit. AIC: Akaike information criteria with lower scores indicating better fit.

Download links