NHP MGIA methods optimisation experiments

In order to minimise the variability associated with low-titre mycobacterial inocula, two stock

parameters were assessed: a) time from thawing to inoculation and b) de-clumping methods.

Mycobacteria were thawed and added to duplicate human PBMC co-cultures every hour for 5 hours after

resting on the bench at room temperature. Mycobacterial viability showed a progressive, albeit modest,

decrease for the first 3 hours, before beginning to recover at 4 hours (Figure 5A). Six methods of de-

clumping were compared using 6 replicate in-tube co-cultures containing cells from the same human

sample for each method: 1) vortexing for 5 minutes on the highest speed, 2) standing on the bench for 5

minutes to allow clumps to settle and then removing only the top fraction, 3) centrifuging at a low speed

to bring clumps down and then removing only the top fraction, 4) sonicating for 2 minutes, 5) vortexing

with 1mm borosilicate solid-glass beads for 2 minutes, and 6) syringing through a 5µM cellulose acetate

filter.

Mycobacterial recovery was highest using the glass beads method, while other methods (particularly

centrifuging and filtering) resulted in some loss of mycobacteria. BCG growth was significantly higher

(lower TTP) following vortexing with glass beads compared with centrifuging or filtering (p=0.0002, delta

mean TTP = 90 hours; and p=0.008, delta mean TTP = 83 hours respectively; Kruskal Wallis with Dunn's

multiple comparisons test, p=0.0002, Figure 5B). Reproducibility between replicates was greatest for

glass beads and filtering (coefficient of variation, CV = 2.2% and 1.2% respectively), and poorest for

vortexing (CV = 13%). Based on these findings, we recommend that mycobacterial stocks suffering from

clumping should be vortexed with sterile 1mm borosilicate glass beads (Sigma Aldrich, UK) for 2 minutes

prior to inoculation, and that inoculation should be conducted as soon after thawing as possible.

Due to limitations regarding the maximum blood volume permitted for collection from macaques, plasma

may be a more feasible alternative to serum. As specific antibodies are likely the main component of

serum contributing to control of mycobacterial growth in the MGIA, we compared levels of PPD-specific

IgM, IgG and IgA between serum and plasma from matched animals at baseline. In all cases there was a

strong correlation, although serum contained modestly but significantly higher levels of specific

antibodies at most time-points measured. We therefore compared the use of autologous serum vs. autologous

plasma in the direct NHP MGIA co-culture (n=7 animals), in which

other components such as complement factors may also contribute to functional control of mycobacterial

growth, and observed an intraclass correlation coefficient (ICC) of 0.58 (moderate agreement). As shown

by Bland-Altman analysis relating the difference between paired measurements to the mean of the pair,

there was minimal bias between the two methods (mean bias = 0.025). Furthermore, all samples were

within the 95% limits of agreement (the interval of 1.96 standard deviations of the measurement

differences either side of the mean difference), which extended from -0.20 (95% CI, -0.50 to -0.13) to 0.25

(95% CI, 0.18 to 0.55) log10 CFU (Figure 6). Although the sample size was small and there is some in

inherent intra-assay variability, this suggests that plasma may be substituted where serum is unavailable

or limited in volume, but we do not recommend using the two samples interchangeably within a single

experiment or direct comparison.

The effect of heat inactivating serum was assessed by measuring mycobacterial growth at the end of in-

tube n=6 human PBMC co-cultures. Mycobacterial growth was lower when co-cultures contained serum

that had been heat-inactivated compared with serum that had not been heat inactivated, but this was not

statistically significant by Wilcoxon (p=0.06, delta mean TTP = 24 hours; Wilcoxon, Figure 8A). It has been

reported that heat inactivation of serum decreases uptake of mycobacteria into monocytes due to the

destruction of complement. As monocytes provide the target host cell for mycobacterial survival and

replication, a decrease in monocyte invasion may lead to decreased mycobacterial growth. Finally, we

compared serum/plasma separated from blood collected in either serum clot-activator or

Ethylenediaminetetraacetic acid (EDTA) vacutainers. Adding plasma separated from an EDTA vacutainer

to the MGIA co-culture resulted in significant inhibition of mycobacterial growth (p=0.03, delta mean TTP =

68 hours; Wilcoxon, Figure 8B). EDTA has been shown to have anti-tubercular activity and has even been

suggested for potential use in treatment of drug-resistant TB. Based on these findings, we

recommend that autologous serum/plasma should be added to a final concentration of 20%, should not

be heat-inactivated and should not be collected in vacutainers containing EDTA.