Resident Microbiome Disruption with Antibiotics Enhances Virulence of a Colonizing Pathogen

There is growing evidence that symbiotic microbes play key roles in host defense, but less is known about how symbiotic microbes mediate pathogen-induced damage to hosts. Here, we use a natural wildlife disease system, house finches and the conjunctival bacterial pathogen Mycoplasma gallisepticum (MG), to experimentally examine the impact of the ocular microbiome on host damage and pathogen virulence factors during infection. We disrupted the ocular bacterial community of healthy finches using an antibiotic that MG is intrinsically resistant to, then inoculated antibiotic- and sham-treated birds with MG. House finches with antibiotic-disrupted ocular microbiomes had more severe MG-induced conjunctival inflammation than birds with unaltered microbiomes, even after accounting for differences in conjunctival MG load. Furthermore, MG cultures from finches with disrupted microbiomes had increased sialidase enzyme and cytadherence activity, traits associated with enhanced virulence in Mycoplasmas, relative to isolates from sham-treated birds. Variation in sialidase activity and cytadherence among isolates was tightly linked with degree of tissue inflammation in hosts, supporting the consideration of these traits as virulence factors in this system. Overall, our results suggest that microbial dysbiosis can result in enhanced virulence of colonizing pathogens, with critical implications for the health of wildlife, domestic animals, and humans.


70
Because our qPCR assay is highly sensitive and subject to low levels of background 71 contamination (Leon et al, in review), we applied a conservative threshold of infection (1349 72 copies of pathogen; Adelman et al, 2015) to ensure results were not skewed by contamination. 73 One bird from the MG-inoculated, Long Antibiotics treatment was omitted from all analyses 74 because it was found to have pathogen load above this conservative threshold of infection one 75 day prior to MG inoculation, and thus we could not be confident that this bird was truly naïve to 76 MG at the time of inoculation. 77 Although some finches in the MG sham experimental groups (which never received MG 78 in our experiments) were seropositive, indicating potential prior exposure to MG in the wild, all 79 finches in these experimental groups were MG-negative via qPCR at the start of the experiment. 80 Two individuals in these groups had pathogen loads well below the conservative threshold Genomic DNA from select time-points that was used for qPCR of MG (pre-inoculation day -1 87 and post-inoculation day 8) was also used for 16S rRNA gene amplicon sequencing of the V4 88 region using primers 515F and barcoded 806R (Caporaso et al, 2012), and following methods 89 outlined in Thomason et al (2017). However, overall these samples had low DNA quantity (2.13 90 -17.7 ng/uL DNA concentrations) relative to other samples on the same plate, with samples 91 from antibiotic-treated birds being of particularly low quality. Doan et al. (2106) similarly found 92 that human conjunctival microbiomes are paucibacterial compared to buccal mucosa and facial 93 skin microbiomes, with quantitative PCR indicating the presence of less than 40 bacterial cells 94 per conjunctival swab. As a result of poor sequencing data, we did not complete a full analysis of 95 these samples. We produced an OTU table that was rarefied at 1100 reads, and only used a 96 subset of samples (No Antibiotics sham-inoculated controls on PID -1 only) to confirm that the 97 dominant bacterial ocular community member was Lactococcus, as was also the case in our prior 98 work (Thomason et al, 2017). Lactococcus in the present study represented 76% ± 0.02 of the 99 relative abundance of the resident ocular microbiome prior to MG inoculation, which was very 100 similar to our previous findings (Thomason et al, 2017).

149
Colonies isolated on BHI agar were divided into "morphotypes" but not positively identified.

150
Recovered isolates were preserved in 10% dimethylsulfoxide at -80°C for subsequent studies. events, control birds that did not receive antibiotics showed no statistically significant change in bacterial abundance (paired t-test: t = 1.24, df = 6, p = 0.26). There were also no differences in 163 culture success amongst the two groups of birds prior to antibiotic perturbation (pre-treatment: Antibiotic susceptibility testing (AST) for cefazolin was performed on mixed cultures from birds 191 in the in vivo sensitivity study (methods and result above). We used serial, two-fold broth 192 microdilution using cation-adjusted Mueller-Hinton broth (CAMHB, Difco, BD Sparks MD).

193
AST broth tubes were incubated at 37˚C for 24 h in ambient air. Minimum inhibitory 194 concentrations of cefazolin were measured for all mixed cultures, and they ranged from 0.5-2.5 195 ug/mL, all within the "susceptible" range for cefazolin.

208
To ensure that residual cefazolin did not influence the sialidase activity of M. gallisepticum 209 isolates, we performed replicate sialidase assays (using methods described in the main text) with 210 and without the addition of cefazolin (0.5 mg/mL in saline; 50µg per assay x 3 replicates / each). 211 We found no effect of cefazolin on sialidase activity of M. gallisepticum (cefazolin added (n=3):  Here, we report the full models, including house finches that received sham inoculations of 216 media only, that mirror the models reported in the main text.  Table S1. Post hoc comparisons for MG+ inflammation score, relative inflammation, and pathogen load models on each day post-inoculation (PID). Reported values are Tukey LSMeans ± SE with p-values corrected for multiple comparisons in parentheses. All estimates are in reference to the first treatment group listed in each pairwise comparison. Bolded values indicate statistically significant differences.