SARS-CoV-2 specific sIgA in saliva increases after disease-related video stimulation

Secretory immunoglobulin A (sIgA) in saliva is the most important immunoglobulin fighting pathogens in the respiratory tract and may thus play a role in preventing SARS-CoV-2 infections. To gain a better understanding of the plasticity in the mucosal antibody, we investigated the proactive change in secretion of salivary SARS-CoV-2-specific sIgA in 45 vaccinated and/or previously infected, generally healthy persons (18 to 35 years, 22 women). Participants were exposed to a disease video displaying humans with several respiratory symptoms typical for COVID-19 in realistic situations of increased contagion risk. The disease video triggered an increase in spike-specific sIgA, which was absent after a similar control video with healthy people. The increase further correlated inversely with revulsion and aversive feelings while watching sick people. In contrast, the receptor binding domain-specific sIgA did not increase after the disease video. This may indicate differential roles of the two salivary antibodies in response to predictors of airborne contagion. The observed plasticity of spike-specific salivary antibody release after visual simulation of enhanced contagion risk suggests a role in immune exclusion.


Evaluation of video content
The disease and control video were rated with regard to (a) their disgust potential ("How strong was your feeling of disgust, antipathy and revulsion while watching the video?";Likert scale from 0="not at all" to 8="completely") and (b) the associated contagion risk ("During the video I had the feeling that I could get infected"; Likert scale from 0="completely disagree" to 7="strongly agree").Utilizing a Wilcoxon signed-rank test, we found that the disease video was on average rated as significantly more disgusting (z=-5.80,p<.001, η 2 =.75) and more contagious (z=-4.56,p<.001, η 2 =.46) than the control video.
In a more fine-grained rating of disgust we showed participants single screenshots of the situations shown in the video.Although this rating replicated the significant difference between the videos (z=5.84,p<.001, η 2 =.76), which we already documented for the broader disgust rating that referred to the whole video, there was also some unexpected interindividual variance in the screenshot disgust rating of the control video (see Figure S1), which led us to perform the exploratory analysis in 2.4.1.1   a This coefficient is set to zero because it is redundant.

Confirmatory analysis of total sIgA
In order to confirm that the present disease video had a comparable effect on total sIgA as the disease videos used in our previous study (Keller et al., 2022), we additionally analysed the saliva samples for the content of total sIgA (Please note, in this analysis we had to exclude an additional participant as he was an outlier with regard to the total secretion rate of all 6 saliva samples.).The confirmatory analysis was performed, because the videos from our previous study showed some differences from the presently employed stimulation.The first disease video of our previous only study showed sneezing or coughing people, who often visibly spread aerosols or droplets and never covered their nose or mouth, while the second disease video of that prior study used concealed contagion stimuli such as people lying sick in bed or sneezing into a tissue.Further, both previous disease videos were mute.The presently used disease video showed a mixture of content displaying openly and concealed contagious persons and also contained the sneeze and cough audios for a more realistic stimulation.The control video from our prior study also differed in some important aspects.distance and the video never focused on a certain person, whereas the present control video showed healthy people in everyday situations.
In the GLMM with total sIgA we found a significant main effect if Sample (F(2,258)=13.39,p<.001) as well a significant interaction between Sample and Video (F(2,258)=4.35,p=.014), but no significant main effect of Video (F(1,258)=1.61,p=.205, please find fixed coefficients in Table S7).In the post-hoc tests, we found a significant difference between the Baseline and the Post-Video 1 sample after watching the disease prime (z=-4.49,p<.001, η 2 =.46), as well as after watching the control prime (z=-2.22,p=.027, η 2 =.11).However, the increase (ΔsIgAtotal) after the disease video was significantly higher than the one after the control video (z=-1.75,p=.040, η 2 =.07, see Figure S2a).Furthermore, the samples Post-Video 1 and 2 differed significantly after disease video (z=-3.28,p=.001, η 2 =.25) but not after the control video (z=-.44,p=.657) (see also Figure S2b).a This coefficient is set to zero because it is redundant.

Spike-specific sIgA
As an explorative analysis, we ran a second GLMM on spike-specific sIgA with the covariate Video Order.This was done, because the first test day, and the type of video one had watched, may have had an influence on spike-specific sIgA secretion on the second test day.We found that the two-way interaction between Video and Sample was still significant (F(2,258)=6.60,p=.002), while the main effects of Video (F(1,258)=.70,p=.403) and Sample (F(2,258)=.90, p=.409) were not significant anymore.The additional factor Video Order (F(1,258)=.23,p=.636) and its two two-way interactions with Video* Video Order (F(1,258)=.33,p=.565) and Sample* Video Order (F(2,258)=.701,p=.497) were also not significant.However, the three-way interaction of Video*Sample* Video Order (F(2,258)=7.33,p<.001) was significant (please find fixed coefficients in Table S9).
When data was split according to Video Order, post-hoc tests on ΔsIgA showed that the increase between Baseline and Post-Video 1 was only significantly higher after the disease video compared to the control video, when participants watched the disease video first (z=-2.71,p=.007, η 2 =.16), but not when they watched the control video on day one (z=1.43,p=.153).
Within the participants, who watched the disease video first, we found similar results as in the analysis across the whole group: Accordingly, spike-specific sIgA at Post-Video 1 was significantly higher when having watched the disease video (z=-2.58,p=.010, η 2 =.30).There was also a significant difference between Baseline and Post-Video 1 after the disease video (z=-3.00,p=.003, η 2 =.41).In participants, who watched the Control video first, we found a significant difference between the Baselines (z=-2.80,p=.005, η 2 =.34) and the Post-Video 1 sample (z=-2.01,p=.045, η 2 =.18) between the two days.As well as a significant difference between and Post-Video 2 Sample and Baseline (z=-2.46,p=.014, η 2 =.26) as well as Post-Video 1 Sample (z=-1.98,p=.048, η 2 =.17) on the second day, when participants watched the disease video (also see Tabel S8 and Figure S3).specific sIgA, we excluded these participants, who rated more than two of the screenshots from the control video with a score of 4 or higher (this coincided with participants who had an average score above one in the screenshot rating).This left an exploratory sample of 36 participants.In the respective 2 x 3 GLM we found no significant main effect of Video (F(1,35)=3.81,p=.059), or Sample (F(2,70)=2.91,p=.061), but replicated the interaction from the total sample (F(2,70)=4,34, p=.026, ηp 2 =.110).
In the post-hoc tests, we found a significant difference between Baseline and the Post-Video 1 sample after watching the disease video (z=-1.98,p=.024, η 2 =.11), but not after watching the control video (z=-.24, p=.648).This was also the case for the difference between the samples taken at Post-Video 1 and Post-Video 2 (Disease: z=-2.75, p=.006, η 2 =.21; Control: z=-.58, p=.561).Furthermore, we found that the sample collected at Post-Video 1 differed significantly between the videos (z=-3.39,p<.001, η 2 =.32), while the Baseline (z=-1.07,p=.285) and Post-Video 2 sample (z=-1.30, p=.192) did not (see Figure S5a).Finally, the comparison of the rise in spike-specific sIgA from Baseline to Post-Video 1 (ΔsIgAspike) showed that the ΔsIgAspike of the disease video was significantly higher than the one from the control video (z=-1.95,p=.026, η 2 =.11) (see Figure S5b).represents the median and the 25th and 75th percentiles, whiskers the smallest and the largest value or no further than 1.5*IQR (inter-quartile range).Significant difference is marked with an asterisk (*p<.05).

Confirmatory analysis with total sIgA
Finally, we ran a confirmatory analysis of total sIgA, which had significantly increased in response to disease-related video content in our previous study 16 .In the 2 x 3 GLM of total sIgA we found no significant main effect of Video (F(1,43)=.17,p=.680), but the main effect of Sample was significant (F(2,86)=10.62,p<.001, ηp 2 =.190), as well as the interaction between the factors (F(2,86)=6.31,p=.003, ηp 2 =.128).In the post-hoc tests, we found a significant difference between the Baseline and the Post-Video 1 sample after watching the disease prime (z=-4.49,p<.001, η 2 =.46), as well as after watching the control prime (z=-2.22,p=.027, η 2 =.11).
The ΔsIgAtotal was further positively correlated with ΔsIgAspike for the disease video (rho=.593,p<.001).

RBD-specific sIgA
In a second step, we analyzed the RBD-specific sIgA for changes induced by the disease video.

Figure S1 :
Figure S1: Screenshot disgust rating of control (grey) and disease (white) video.Box represents the median and the 25th and 75th percentiles, whiskers the smallest and the largest value or no further than 1.5*IQR (inter-quartile range).Significant differences are marked with asterisks (***p<.001).

Figure
Figure S2: a) Change in total sIgA for disease and control video intervention.Mean, standard errors and individual data points of the secretion rate at Baseline, directly after the video (Post-Video 1), and several minutes after the

Figure S3 :
Figure S3: Bar plots with mean, standard errors and individual data points of the spike-specific sIgA secretion rate at Baseline, directly after the video (Post-Video 1), and several minutes after the video (Post-Video 2).For (a) participants (n=22) who watched the disease video first and (b) participants (n=23) who watched the control video first.Significant changes are marked with asterisks (*p<.05;**p<.01),based on Wilcoxon signed rank test.

Figure S5 :
Figure S5: a) Exploratory analysis of spike-specific sIgA in a subgroup of 36 participants.Mean and standard errors of the secretion rate at Baseline, directly after the video (Post-Video 1), and several minutes after the video (Post-Video 2).Significant changes are marked with asterisks (*p<.05;***p<.001).b) Exploratory analysis of

Table S2 :
Content of the disease video.Sequence (Seq.), length and description of the video content shown.

Table S3 :
Content of the control video.Sequence (Seq.), length and description of the video content shown.
9Woman, side view, entire body visible, typing on the computer, smiling 10 Man, top half of body visible, typing on his phone, scratching his nose 11 Woman, facing camera, looking and smiling at camera 12 Man, facing camera, looking around, scratching his cheek 13 Woman, side view, entire body visible, outside, leaning against a railing, drinking out of a cup 14 Man, side view, entire body visible, sitting on a bed, typing on computer 15 Woman, facing camera, looking at the camera 16 Man, facing camera, looking and smiling at camera 17 Woman, side view, upper half of body visible, sitting on a bed, typing on computer 18 Man, facing camera, lying in bed with closed eyes, moving around 19 Woman, side view, entire body visible, sitting on an armchair, reading a book 20 Man, front view, entire body visible, sitting on a bed, typing on computer

Table S4 :
Kupfer et al. (2021)onnaire that follows the Post-Video 1 sample (modified and adapted questionnaire ofKupfer et al. (2021); two subscales were formed by averaging corresponding questions)

Table S5 :
Fixed Coefficient of spike-specific sIgA secretion model with the Fixed Factors Video and Sample and their Interaction.With Coefficient (β), Standard Error of Coefficient (SEβ), t-value, p-value and lower as well as upper Confidence Intervals (CI).
a This coefficient is set to zero because it is redundant.

Table S6 :
Fixed Coefficient of RBD-specific sIgA secretion model with the Fixed Factors Video and Sample and their Interaction.With Coefficient (β), Standard Error of Coefficient (SEβ), t-value, p-value and lower as well as upper Confidence Intervals (CI).

Table S7 :
Fixed Coefficient of total sIgA secretion model with the Fixed Factors Video and Sample and their Interaction.With Coefficient (β), Standard Error of Coefficient (SEβ), tvalue, p-value and lower as well as upper Confidence Intervals (CI).

Table S8 :
Post-hoc results of Wilcoxon signed rank test wit z values and p-values in parenthesis, comparing spike-specific sIgA at Baseline, Post-Video 1 (PV1) and Post-Video 2 (PV2).