Ethics statement
Rhesus macaques were housed in a biosafety level 4 containment facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. Experimental procedures were approved by the National Institute of Allergy and Infectious Diseases (NIAID), Division of Clinical Research (DCR), Animal Care and Use Committee and were in compliance with the Animal Welfare Act regulations, Public Health Service policy, and the Guide for the Care and Use of Laboratory Animals recommendations.
Virus preparation
For aerosol inhalation, MERS-CoV-Hu/Jordan-N3/2012 strain (GenBank accession no. KC776174.1) [9] was grown in Eagle’s Minimum Essential Medium (Lonza, MD, USA) on Vero E6 cells.
Aerosol challenge
Prior to aerosol challenge, four rhesus macaques, two males and two females, 3–5 years old, weighing 3–5 kg each, were anesthetized by intramuscular ketamine (10–15 mg/kg) injection. Head-out plethysmography (Buxco-Data Sciences International, MN, USA) was used to calculate an average respiratory minute volume (mL/min) by multiplying the respiration rate by the tidal volume. Aerosol concentrations derived from a SKC biosampler (SKC Inc., PA, USA) were used to calculate the presented dose [10]. Within a negative-pressure (−24.9 Pa), head-only aerosol exposure chamber, macaques were exposed to a small-particle (0.5–3 μm aerodynamic diameter targeting lung alveoli) aerosol challenge (inhaled dose = log10 – 4.64 plaque-forming units).
Data acquisition
Imaging data were acquired with Gemini PET/CT clinical scanner (Philips Healthcare, Andover, MA, USA) [5, 11]. With an axial field-of-view (FOV) of 180 mm of the PET scanner, the entire NHP thorax is imaged in a single bed position. Use of the scanner’s brain protocol resulted in a transverse field of view of 256 mm and led to cubic 2-mm-wide voxels in the reconstructed images. Low-dose CT images of the thorax for PET attenuation purposes were acquired at 120 kVp, 3-mm slice thickness, and 1.5-mm spacing. No contrast was given, and the subjects were freely breathing during the scan. PET image acquisition was initiated immediately after the low-dose CT scans and 1 min prior to intravenous injection of [18F]-FDG (9–10 MBq/kg) into the saphenous vein and continued for up to 60 min (3600 s). Nine imaging sessions per animal were conducted on pre-inoculation days −14 or −13 and −11 or −10 and post-inoculation days +1 or +2, +3 or +4, +7 or +8, +9 or +10, +15 or +16, +21 or +22, and +28 or +29 with MERS-CoV.
Image reconstruction
SUV PET images were reconstructed iteratively using the manufacturer supplied 3D line-of-response (LOR)-based row-action maximum-likelihood algorithm [12]. Methods for scatter, decay, random, and attenuation correction were applied during the image reconstruction process. Both scatter and attenuation corrections [13] were based on the low-dose CT images acquired prior to the PET scans.
The list mode data were sorted into 46 dynamic frames during creation of the histograms. To extract the early tracer dynamic distribution in the arterial blood, the initial data set (up to 720 s or 12 min) was comprised of 39 frames with the following time sequence: 15 frames × 2 s, 6 frames × 5 s, 5 frames × 10 s, 5 frames × 20 s, 4 frames × 40 s, and 4 frames × 120 s. This sequence was followed by 3 frames × 240 s and 4 frames × 480 s to capture the late slow phase of dynamic distribution of the tracer in both the blood and the tissues. PET images were reconstructed iteratively using 3D ordered-subset expectation-maximization algorithm with two iterations and nine subsets followed by 18 iterations of maximum a posteriori reconstruction [14]. Maximum a posteriori parameters were adjusted to provide a uniform spatial resolution of 4.8 mm (full-width half-maximum = 4.8 mm) in all three directions. Methods for scatter, decay, random, and attenuation correction were applied during the process of PET image reconstruction.
Volume of interest definition
Reconstructed SUV PET images were analyzed without any post-reconstruction smoothing using PMOD version 3.5 (PMOD Technologies, Zurich, CH). To extract an image-derived input function (IDIF), VOI (2-mm spheres) were placed on the left ventricles and arch of the aorta using frames over the first 6 min (360 s) after [18F]-FDG injection (Fig. 1a, b). Averaged data from two VOIs were used to generate the IDIF (Fig. 1c, d). Two-ml spheres were placed on axillary and mediastinal LNs and lumbar spine bone marrow as described previously [6], and 5-mm spheres were placed on right and left sides of the lungs to obtain the tissue time activity curves (TACs). The last time point of the TACs was used to generate the SUV data.
Kinetic modeling
Using the standard two-tissue compartment kinetic model with irreversible tracer metabolism (k
4 = 0, Fig. 1e), the tracer influx constant, K
i
, was computed for lymphoid and lung tissues [15]. The blood volume fraction (V
b
) was included in the modeling. Patlak linear regression method was applied for parameter estimation utilizing the IDIF, [18F]-FDG tissue TACs [16], and PMOD version 3.5 (PMOD Technologies). Tissue TACs were fitted to the models by use of the nonlinear least-squares method with the Levenberg-Marquardt algorithm, which minimizes the weighted sum of squared errors between PET measurement and model solutions. A plot of the ratio C
tis(t)/C
bl(t) against the ratio of cumulative to instantaneous blood activity concentration (“normalized time”) became linear in the late phase after the tracer injection when the concentration of free (i.e., unmetabolized) [18F]-FDG in the blood had equilibrated with that of free tracer in extravascular volume of distribution. This linear part of the plot was fitted by Eq. (1) to identify the K
i
as a slope of a regression line:
$$ \frac{C_{\mathrm{tis}}}{C_{\mathrm{bl}}(t)}={K}_i\frac{{\displaystyle \underset{t=0}{\overset{t}{\int }}{C}_{\mathrm{bl}}(t)}}{C_{\mathrm{bl}}(t)}+{V}_{\mathrm{dist}} $$
(1)
in which C
tis(t) and C
bl(t) represent the radioactivity concentration in the region of interest and the arterial blood assessed from the PET images at different time points after an [18F]-FDG injection, respectively, and V
dist is an initial distribution volume. A criterion for maximum error was set to 5 % to derive the model parameter values. For the [18F]-FDG model described in Fig. 1, the slope equals K
1 × k
3 ÷ (k
2 + k
3).
Hematology and clinical observations
Complete blood cell counts were determined on PET-scan days [5]. Body temperature or body weight were monitored once daily or once every other day, respectively.
Statistical analysis
Two-way repeated measures analysis of variance (ANOVA) with post hoc Bonferroni multiple comparison test used K
i
obtained pre-inoculation and post-inoculation with MERS-CoV and VOI location as independent variables to characterize the host immune response. For two-way repeated measures ANOVA, we used K
i
at different time points pre-inoculation with MERS-CoV and VOI location as within and between two factors, respectively.
The correlations between K
i
values in mediastinal and axillary LNs and bone marrow and between monocyte fraction in the blood and mediastinal and axillary LNs were calculated using the Pearson product moment correlation coefficient (r). The D’Agostino and Pearson test [17] was applied to confirm that the data followed a Gaussian distribution. GraphPad Prizm 6.01 (GraphPad Software Inc., La Jolla, CA, USA) was used for all statistical analyses.