HSV-2 ΔgD elicits FcγR-effector antibodies that protect against clinical isolates

Immunization with low doses of ΔgD-2 is protective against lethal skin and intravaginal challenge.

C57BL/6 mice (5 mice per group) were s.c. primed and then boosted 21 days later with 5 × 10⁴, 5 × 10⁵, or 5 × 10⁶ PFU of ΔgD-2. Three weeks later, the mice were challenged either intravaginally (following pretreatment with medroxyprogesterone) or by skin scarification with a lethal dose that typically kills 90% of animals (LD90) (5 × 10⁵ PFU) of HSV-2(4674), a previously described clinical isolate (22). All HSV-2 ΔgD-2 immunized mice survived (5/5 per group), whereas all of the control-vaccinated mice (immunized with VD60 cell lysate) succumbed to disease (Figure 1, A and B). Although the mice vaccinated with the lowest dose (5 × 10⁴) showed mild epithelial disease (Supplemental Figure 1, A and B; supplemental material available online with this article; doi:10.1172/jci.insight.88529DS1), no signs of neurological disease were observed and all animals completely recovered by day 14 in both challenge models. HSV-specific Abs (measured by ELISA using infected cell lysate as the antigen) were detected in the serum of all vaccinated mice one week after boost, but not prime, and the Ab titer increased with administration of higher vaccine doses (Figure 1C).

Mice immunized with ΔgD-2 are protected from high viral challenges of virulent HSV-1 and HSV-2 clinical isolates.

To evaluate if the ΔgD-2 vaccine protects against diverse HSV-1 and HSV-2 strains, we obtained 5 HSV-1 (denoted B³ × 1.1 – B³ × 1.5) and 5 HSV-2 (denoted B³ × 2.1 – B³ × 2.5) clinical isolates from the Clinical Virology Lab at Montefiore (Bronx, New York, USA), as well as a previously described South African HSV-2 clinical isolate (SD90, ref. 17). The isolates were grown on Vero cells and were passaged no more than 3 times before sequencing and phenotyping. Illumina sequencing (Figure 2, A and B) showed that the strains exhibited substantial genetic diversity with pairwise distances as high as 6.3% between B³ × 1.5 and the other B³ × 1 isolates and 5.0% between B³ × 2.2 and the other B³ × 2 isolates. In vivo virulence of each clinical strain was compared with laboratory strains by challenging BALB/C mice using the skin scarification model with 1 × 10⁵ PFU of the HSV-1 strains or 5 × 10⁴ PFU of HSV-2 strains (Figure 2, C and D). The clinical isolates demonstrated a range of virulence with B³ × 1.1, B³ × 1.3, B³ × 2.3, and SD90 inducing more rapid disease, with the highest morbidity (Figure 2, C and D; Supplemental Figure 2, A and B). Similar results were observed in the vaginal challenge model, with the same 4 isolates exhibiting the most virulent disease (not shown). Interestingly, no significant differences between the isolates were observed by in vitro single and multistep growth curves on Vero cells (Supplemental Figure 3, A and B).

To assess whether ΔgD-2 protected against the different isolates, C57BL/6 or BALB/C mice were primed and boosted with 5 × 10⁶ PFU/mouse of ΔgD-2 (or VD60 lysate as the control immunogen) and then challenged with an LD90 dose of the 4 more virulent clinical isolates (Table 1) using the skin scarification model. All ΔgD-2–vaccinated mice survived challenge (n = 7 C57BL/6 mice per group, Figure 3, A and B; n = 5 BALB/C mice per group, Figure 3C). While some mice exhibited mild epithelial disease, which peaked on day 4, the majority of animals had fully recovered by day 8 after challenge (Supplemental Figure 4, A and B). No signs of neurological disease were detected in any of the mice at any time point.

To further evaluate the robustness of the immune response, we increased the challenge dose in the C57BL/6 mice to 10 times and 100 times the LD90 doses of SD90 and 10 times the LD90 of B³ × 1.1. All of the ΔgD-2–vaccinated mice survived (Figure 3D) with no signs of neurological disease (Supplemental Figure 4C). The ΔgD-2–vaccinated mice had significantly less infectious virus detected in skin biopsies by plaque assay on day 5 after challenge (Figure 4A, n = 3 mice per group). Histopathology of skin biopsies revealed ulceration and necrosis covering 75%–95% of the epithelium in control-vaccinated mice compared with < 10% epithelial necrosis and ulceration in the ΔgD-2–vaccinated mice; the histopathological findings in ΔgD-2–vaccinated mice resembled those of the unvaccinated, uninfected controls that were treated with depilation and skin scarification (Supplemental Figure 5). Moreover, there was no replicating or latent virus detected by plaque assay (Figure 4B) or quantitative PCR (qPCR) (Figure 4C), respectively, in DRGs isolated on day 14 after challenge in the ΔgD-2–vaccinated mice (n = 5 mice per each challenge dose and strain). Similarly, we did not detect any reactivating virus when DRGs from ΔgD-2–vaccinated mice (isolated day 5 after challenge with LD90 of SD90) were cocultured for 3 weeks with Vero cells. In contrast, viral DNA and reactivatable virus were recovered from all control-vaccinated (VD60 cell lysate–vaccinated) mice (DRG isolated at time of euthanasia; Figure 4D, n = 5 mice per group).

To determine if viral gene expression is needed to elicit complete immunity against HSV clinical isolates, we compared the vaccine efficacy of live or inactivated ΔgD-2 stocks. Virus was inactivated by brief exposure to UV light, which allows virus to bind and enter cells but not express viral gene products (Figure 5A) (23) or heat inactivation (HI), which results in virus that does not enter cells. Mice were primed and boosted with 5 × 10⁶ PFU of live ΔgD-2, or equivalent particles of UV or HI ΔgD-2 or VD60 lysate, and subsequently challenged with an LD90 dose of HSV-2(SD90) on the skin. HI virus failed to protect against lethal disease with only 1 of 5 mice surviving (Figure 5, B and C) and did not prevent the establishment of latency (Figure 5D). UVΔgD-2–immunized mice survived lethal skin challenge (n = 5 per group, Figure 5B); however, disease scores were higher compared with mice that received live viral vaccine (Figure 5C), and UVΔgD-2 did not prevent establishment of latency (Figure 5D). HSV genomes were detected in the DRG from all of the mice that received HI- or UV-inactivated virus or VD60 control lysates (Figure 5C). In contrast, a low level of HSV DNA was detected in only 1 of 33 mice immunized with live ΔgD-2 (Figures 4, C and D, and Figure 5C).

HSV-2 ΔgD-2 elicits FcγR effector antibodies that are rapidly recruited into the skin following challenge.

Serum from ΔgD-2–vaccinated mice obtained 1 week after boost induced antibody-dependent cellular phagocytosis (ADCP) of HSV-coated beads by THP-1 cells with the release of higher levels of IFN-γ compared with serum from VD60 lysate (control) immunized mice, indicative of FcγR effector activity (Figure 6A). Consistent with this, ΔgD-2 serum obtained 1 week after boost and 1 week after challenge activated murine FcγRIV and FcγRIII, whereas control serum showed no mFcγRIV and 23- to 47-fold less mFcγRIII activation at 1 week after challenge (Figure 6B). Furthermore, ΔgD-2 serum was able to activate mFcγRIV when the serum was incubated with cells infected with 5 different clinical isolates (Figure 6C).

To further explore the differences in efficacy observed in response to inactivated ΔgD-2, immune serum obtained 1 week after boost from mice vaccinated with live, UV, or HI ΔgD-2 were assayed for total HSV-specific Abs by ELISA and for mFcγRIV activation. UV or HI ΔgD-2 elicited lower HSV-specific Abs (1:50,000 titer) compared with live viral vaccine (1:250,000; Figure 6D). Moreover, there was a decrease in mFcγRIV activation by immune serum obtained from mice vaccinated with inactivated compared with live ΔgD-2 with HI virus eliciting little FcγR activation (P < 0.01 compared with live ΔgD-2), and UV virus eliciting an intermediate response (Figure 6E). These findings parallel the differences in efficacy (Figure 5).

The presence of HSV-specific Abs was examined 21 days after boost and 2 or 5 days after challenge in skin homogenates by ELISA using either an HSV-2(4674)– or HSV-1(17)–infected cell lysate as the antigen. The ΔgD-2–immunized mice had low levels of HSV-specific Abs in the skin after boost, which rapidly increased as early as day 2 after challenge (Figure 7A). B³ × 1.1 elicited higher titer Ab response compared with SD90. In contrast, no HSV-specific Abs were detected in the skin of control-vaccinated mice on day 2 or day 5 after challenge (Figure 7, A and B). The Abs recovered from the skin were predominantly IgG (1:24,000 titer; Figure 7B) with no anti–HSV IgA or IgM detected (data not shown), enriched in IgG2 (equal proportions of IgG2a and IgG2b, Figure 7C), and induced mFcγRIV and mFcγRIII activation (data shown for day 2 after challenge, Figure 7D).

HSV-2 ΔgD-2–immunized mice have minimal inflammation following skin HSV challenge.

Biopsies were obtained 2 or 5 days after challenge and processed for histology and/or homogenized. They were then evaluated for presence of cytokines/chemokines and neutrophils. Consistent with the histopathology (Supplemental Figure 5), local Ab response (Figure 7), and rapid clearance of virus (Figure 4), there was a decrease in inflammatory cytokines/chemokines detected in skin homogenates in the ΔgD-2–immunized compared with control immunized mice on day 5 after challenge (Figure 8). HSV-1– and HSV-2–infected mice had higher levels of TNFα (Figure 8A), IL-1β (Figure 8B), and IL-6 (Figure 8C) compared with mock-infected mice on day 2 independent of immunization. However, the levels decreased by day 5 in ΔgD-2 but not the control-immunized mice. Similar results were obtained for the chemokines CXCL9 (Figure 8D) and CXCL10 (Figure 8E). Interestingly, IL-33 levels were consistently higher in ΔgD-2–immunized compared with control-immunized mice at both time points (Figure 8F). A decrease in Ly6G⁺ neutrophils (Figure 8G) in skin biopsies from ΔgD-2–immunized was observed compared with control-immunized mice.

Cells and viruses.

Vero (African green monkey kidney cell line; CCL-81; American Type Culture Collection [ATCC]) cells, VD60 cells (Vero cells encoding gD-1 under endogenous promoter, ref. 36), and CaSki (human cervical epithelial cell line; CRL-1550; ATCC) were passaged in DMEM supplemented with 10% FBS (Gemini Bio-Products). THP-1 cells (human monocyte cell line; TIB-202; ATCC) were passaged in RPMI-1640 (Invitrogen) supplemented with 10% FBS. Laboratory strains HSV-2(G) (37), HSV-2 (333-ZAG) (38), HSV-1(17) (39), and HSV-1(F) (37) were propagated on Vero cells. Previously described clinical isolates HSV-2(4674) (38) and HSV-2(SD90) (South African isolate provided by David Knipe, Harvard Medical School, Boston, Massachusetts, USA) (17) were propagated on CaSki and Vero cells, respectively. Ten additional deidentified clinical isolates (five HSV-2 isolates designated B³ × 2.1 through B³ × 2.5 and five HSV-1 isolates designated B³ × 1.1 through B³ × 1.5) were provided by the Montefiore Clinical Virology Lab (Amy Fox, Montefiore Medical Center, Bronx, New York, USA) and passaged 3 times on Vero cells for a low-passage working stock. Construction of the single-cycle vaccine candidate strain, HSV-2(G) ΔgD-2, and its propagation on VD60 cells has been previously described (18, 19). Inactivation of virus was performed by exposing ΔgD-2 stocks to UV light for 10 minutes (10 cm distance from light source, ref. 23) or by heating to 60°C for 5 minutes. The impact of UV or HI on viral gene expression and infectivity was assessed by isolating RNA from Vero cells infected with 5 PFU/cell 2 and 6 hours after infection and performing reverse transcription PCR (RT-PCR) for infected cell protein-0 (ICP0) or thymidine kinase (TK) by using primers from Applied Biosystems (Carlsbad, California, USA) and conducting plaque assays on VD60 cells, respectively.

Murine immunization and viral challenge studies.

Female C57BL/6 and BALB/c mice were purchased from the Jackson Laboratory (JAX) at 4–6 weeks of age. Mice were primed and boosted 3 weeks later with 5 × 10⁴ to 5 × 10⁶ PFU of ΔgD-2, 5 × 10⁶ PFU UV-inactivated ΔgD-2, 5 × 10⁶ PFU HI ΔgD-2, or an equal amount of uninfected VD60 cell lysates (control) s.c. (medial to the hind limb and pelvis) at 100 μl/mouse.

For intravaginal HSV infections, mice were treated with 2.5 mg of medoxyprogesterone acetate (MPA; Sicor Pharmaceuticals) s.c. 5 days prior to challenge. Mice were then inoculated intravaginally with an LD90 (5 × 10⁵ PFU/mouse) of HSV-2(4674) at 30 μl/mouse and scored for disease and monitored for survival for 14 days as previously described (19).

For HSV skin infections, mice were depilated on the right flank with Nair and allowed to rest for 24 hours. Depilated mice were anesthetized with isoflurane (Isothesia, Henry-Schein); they were then abraded on the exposed skin with a disposable emory board for 20–25 strokes and subsequently challenged with 1 × 10⁵ PFU HSV-1 or 5 × 10⁴ PFU HSV-2 strains for in vivo virulence studies or challenged with an LD90, 10 times LD90, or 100 times LD90 of select HSV strains (see Table 1). Mice were monitored for 14 days and scored as follows: 1, primary lesion or erythema; 2, distant site zosteriform lesions, mild edema/erythema; 3, severe ulceration and edema, increased epidermal spread; 4, hind-limb paresis/paralysis; and 5, death. Mice that were euthanized at a score of 4 were assigned a value of 5 on all subsequent days for statistical analyses.

In vitro growth curves.

For single-step growth of each virus, Vero cells were infected with virus at an MOI of 5 PFU/cell, and supernatants and cells were collected every 4, 8, 16, and 24 hours after infection and stored at −80°C. For multistep growth of each virus, Vero cells were infected at a MOI of 0.01 PFU/cell, and supernatants and cells were harvested every 12 hours after infection up to 72 hours. Infectious virus was measured by performing plaque assays with supernatants and lysed cells as previously described (38).

Viral DNA isolation and sequencing of clinical isolates.

HSV DNA was prepared by infecting confluent Vero cells in a T150 flask with different clinical isolates or HSV-2(G) at an MOI of 10. Cells were harvested 16 hours after infection and washed twice with PBS. DNA was extracted using DNeasy Blood and Tissue (Qiagen) following the manufacturer’s recommendations. DNA was quantitated by Qubit dsDNA HS assay (Invitrogen). Paired-end libraries were prepared by the Nextera XT DNA library preparation kit (Illumina) following the manufacturer’s instructions. Libraries were sequenced on a Illumina MiSeq Desktop Sequencer. Viral genome sequences were assembled with the VirAmp pipeline (40) following removal of host sequence by alignment to the Macaca mulatta genome as a substitute for the incomplete Chlorocebus sabaeus (source of Vero cells) genome. HSV-1 and HSV-2 genomes were annotated with Genome Annotation Transfer Utility on ViPR by comparison to HSV-1(17) (GenBank accession no. JN555585.1) and HSV-2(HG52) (JN561323) prior to submission to GenBank. Whole genome alignments including the previously sequenced HSV-2(SD90e) (KF781518), HSV-2(333) (KP192856), ChHV 105640 (NC_023677.1), and HSV-1(F) (GU734771.1) were performed using ClustalW (41), and phylogenetic trees were constructed using the unweighted pair group method with arithmetic mean (UPGMA) with 1,000 bootstrap replicates in Molecular Evolutionary Genetics Analysis v.6 (MEGA6) (42). All positions containing gaps or missing data were eliminated. GenBank numbers for newly sequenced genomes are as follows: HSV-2(G) (KU310668), HSV-2(4674) (KU310667), B³ × 1.1 (KU310657), B³ × 1.2 (KU310658), B³ × 1.3 (KU310659), B³ × 1.4 (KU310660), B³ × 1.5 (KU310661), B³ × 2.1 (KU310662), B³ × 2.2 (KU310663), B³ × 2.3 (KU310664), B³ × 2.4 (KU310665), and B³ × 2.5 (KU310666).

Virus detection in tissue.

Skin biopsies (~5–10 mm in diameter by mechanical excision) and DRG were obtained from ΔgD-2 or VD60 lysates immunized mice at day 2, 5, 7, 14, or 21 after viral skin challenge. The tissue was weighed and homogenized in RNase/DNase free Lysing Matrix A tubes (MP Biomedicals) with serum-free DMEM at 6.0 m/sec for three 30-second cycles in the FastPrep-24 5G (MP Biomedicals). Samples were spun at 2,500 × g for 10 minutes at 4^oC, and the resulting supernatant was overlaid on confluent Vero cell monolayers (2 × 10⁵ cells/well in a 48-well plate) for 1 hour to measure replicating virus via plaque assay. Wells were washed with PBS and then with 199 medium (Gibco) containing 1% heat-inactivated FBS, overlaid with 0.5% methylcellulose and incubated at 37^oC for 48 hours. Cells were fixed with 2% paraformaldehyde and stained with a crystal violet solution, and the number of PFU was quantified. Neuronal ex vivo coculture assays were performed at day 5 after challenge as previously described (19).

HSV qPCR.

DNA was extracted from weighed tissue samples using DNeasy Blood and Tissue (Qiagen) following the manufacturer’s recommendations. Extracted DNA was then normalized to 10 ng of DNA per reaction and viral DNA quantified using qPCR using ABsolute qPCR ROX Mix (Thermo Scientific). Primers for HSV polymerase (U_L30) were purchased from Integrated DNA Technologies (forward primer sequence 5′-GGCCAGGCGCTTGTTGGTGTA-3′, reverse primer sequence 5′-ATCACCGACCCGGAGAGGGA-3′, probe sequence 5′-CCGCCGAACTGAGCAGACACCCGC-3 ′) and used to detect viral genomic DNA. Isolated HSV-2 viral DNA was calibrated for absolute copy amounts using QuantStudio 3D Digital PCR (dPCR, ThermoFisher Scientific) and subsequently used as a standard curve to determine HSV viral genome copies. Samples that read 4 or fewer copy numbers were considered negative. Data are presented as log₁₀ HSV genomes per gram of DRG tissue.

Histopathology and immunofluorescence of skin tissue.

Mice were euthanized on day 5 after challenge, and the skin at the infection site was excised and either frozen in optimal cutting temperature (OCT) media, cut into 5-μm sections and stored at –80^oC for immunofluorescence, or formalin fixed for 48 hours at room temperature (RT), paraffin-embedded, and sectioned for histopathology. Slides from paraffin-embedded sections were stained with H&E and evaluated by a board-certified veterinary pathologist who was blinded to the identity of the samples. The slides for immunofluorescent studies were fixed in –20^oC acetone for 15 minutes, washed with wash buffer (0.05% Tween 20 in PBS), blocked for 2 hours with blocking buffer (2% BSA, 5% heat inactivated goat serum in PBS) at RT, and washed twice with wash buffer (WB). Slides were then incubated with anti-Ly6G (clone 1A8, 1:500, BD Biosciences) in blocking buffer for 1 hour at RT, washed, and incubated with a goat anti-rat secondary antibody conjugated with Alexa Flour 555 (1:500, ThermoFisher Scientific) for 30 minutes at RT. Slides were washed and mounted with media containing DAPI (ProLong Diamond Antifade Mountant with DAPI, ThermoFisher Scientific). Slides were imaged using a Nikon Eclipse Ti-U inverted light microscope at 20× magnification from apical layer (epidermal) to basal layer (striated muscle) at 2 different locations per sample.

Detection of antibodies and cytokines in skin biopsies and serum.

Skin biopsies were obtained from HSV-2 ΔgD-2– or VD60 lysate–immunized (control-immunized) mice (~5–10 mm in diameter by mechanical excision) day 21 after boost or day 2 and 5 after viral skin challenge. The tissue was weighed and homogenized as described above and was evaluated for presence of anti-HSV antibodies, cytokines, and chemokines. Serum was also analyzed for HSV-2 antibodies from HSV-2 ΔgD-2– or control-immunized mice collected before immunization, 1 week after prime, or 1 week after boost. Anti-HSV antibodies were detected by ELISA as previously described using uninfected, HSV-1(17)–infected, or HSV-2(4674)–infected Vero cell lysates as the coating antigen (19). Biotin anti–mouse Igκ (clone 187.1) or biotin anti-mouse IgA (clone C10-1), IgM (clone I1/41), IgG1 (clone A85-1), IgG2a (clone R19-15), IgG2b (clone R12-3), or IgG3 (clone R40-82) at 1 μg/ml (all from BD Biosciences) were used as secondary detection antibodies. Wells were read on a SpectraMax (M5 series) ELISA plate reader at an absorbance of 450 nm. The resulting absorbance was determined by subtracting values obtained for uninfected cell lysates to values obtained with infected cell lysates. Total anti–HSV Ig is reported as the optical density (OD) at 450nm normalized to relative tissue weight at a 1:1,000 dilution of tissue homogenate. Anti–HSV IgG, IgA, IgM, or IgG1-3 are reported as the OD at 450 nm at all dilutions except IgG1-3, which is reported only at a 1:100 dilution of skin homogenate.

Skin homogenate supernatants were assayed for IL-6, IL-1β, IL-33, TNFα, monokine induced by IFN-γ (MIG, CXCL9), and IFN-inducible cytokine (IP-10, CXCL10) using a Milliplex mouse cytokine/chemokine immunoassay (Millipore) and a Luminex Magpix system and analyzed with Milliplex Analyst (Version 3.5.5.0; VigeneTech Inc.).

ADCP assay.

To determine HSV-specific ADCP, a protocol modified from Ackerman, et al. (43), was used. Briefly, 2 × 10⁸ 1 μm Neutravadin-red fluorescent beads (Invitrogen, F-8775) were coated with 0.3 mg of biotinylated HSV-2–infected or uninfected (control) Vero cells overnight at 4^oC in 500 μl of BlockAid (ThermoFisher Scientific, B-10710). Beads were washed twice with 1% BSA in PBS, and then 1 × 10⁶ beads/well were added to a 96-well round-bottom plate. Of a 1:5 dilution (diluted in serum-free RPMI), 50 μl of 1-week postboost immune serum that had been heat-inactivated at 56^oC for 30 minutes were added to the wells and incubated for 2 hours at 37^oC. THP-1 cells (2 × 10⁴ cells/well) were then added to each well at a final volume of 200 μl/well and incubated for 8 hours at 37^oC at 5% CO₂. Subsequently, 100 μl of supernatant was removed and stored at –20^oC and then resuspended with 100 μl 4% paraformaldehyde. Samples were read on 5-laser LSRII flow cytometer (BD Biosciences) at the Einstein Flow Cytometry Core Facility. Phagocytic score is reported by gating on events representing THP-1 cells then applying the following equation: [(% of cells bead positive × MFI of cells positive for beads)/10⁶] using FlowJo software (version 10, Tree Star Inc.). IFN-γ secretion from activated THP-1 cells via antibody phagocytosis was determined in parallel by analyzing stored cultured supernatants using a Milliplex human custom immunoassay (Millipore) and a Luminex Magpix system as previously described.

FcγR activation reporter assay.

To determine HSV-specific FcγR activation of Abs, we utilized Promega’s Fc effector murine and bioassay kits (a gift from Mei Cong, Promega Corporation, Madison, Wisconsin, USA). The Ab samples included serial dilutions of mouse serum or skin homogenate obtained 1 week after boost or 2–7 days after skin challenge diluted in serum-free RPMI. The Ab samples (25 μl) were incubated with target cells (1.25 × 10⁴cells/well) in a 96-well plate for 30 minutes. The targets were Vero cells that had been infected for 16 hours with different strains of HSV-1 or HSV-2 (MOI 1 PFU/cell) or uninfected control cells. Jurkat cells stably expressing either murine FCγRIII ADCC reporter bioassay (mFcγRIII), mFcγRIV, and a downstream luciferase reporter gene based on NFAT activation acted as the effector source at a 6:1 effector/target ratio. The effector and target cells were incubated for 6 hours at 37^oC, and luminescence was detected after the addition of 75 μl of Bio-Glo Reagent (Promega) and read on an SpectraMax M5 plate reader. Fold FcγR induction was calculated as luminescence of [(Ab sample with effector and target cells – luminescence of background)/(luminescence of spontaneous activation of effector cells by target cells alone – luminescence of background)].

Statistics.

Results were compared by 2-way ANOVA with multiple comparisons or 2-tailed unpaired Student’s t tests using GraphPad Prism version 6. Mantel-Cox survival curves were compared by log rank tests. P values < 0.05 (*), < 0.01 (**), < 0.001 (***) were considered significant.

Study approval.

Murine animal experiments were performed with approval from Albert Einstein College of Medicine Institutional Animal Care and Use Committee, Protocol 20130913 and 20150805.