Rickettsia parkeri hijacks tick hemocytes to manipulate cellular and humoral transcriptional responses

Introduction Blood-feeding arthropods rely on robust cellular and humoral immunity to control pathogen invasion and replication. Tick hemocytes produce factors that can facilitate or suppress microbial infection and pathogenesis. Despite the importance of hemocytes in regulating microbial infection, understanding of their basic biology and molecular mechanisms remains limited. Methods Here we combined histomorphology and functional analysis to identify five distinct phagocytic and non-phagocytic hemocyte populations circulating within the Gulf Coast tick Amblyomma maculatum. Results and discussion Depletion of phagocytic hemocytes using clodronate liposomes revealed their function in eliminating bacterial infection. We provide the first direct evidence that an intracellular tick-borne pathogen, Rickettsia parkeri, infects phagocytic hemocytes in Am. maculatum to modify tick cellular immune responses. A hemocyte-specific RNA-seq dataset generated from hemocytes isolated from uninfected and R. parkeri-infected partially blood-fed ticks generated ~40,000 differentially regulated transcripts, >11,000 of which were immune genes. Silencing two differentially regulated phagocytic immune marker genes (nimrod B2 and eater-two Drosophila homologs), significantly reduced hemocyte phagocytosis. Conclusion Together, these findings represent a significant step forward in understanding how hemocytes regulate microbial homeostasis and vector competence.


Introduction
Ticks are major vectors for bacterial, viral, and protozoan pathogens of human and veterinary importance. The Gulf Coast tick Amblyomma (Am.) maculatum is a competent vector of the spotted fever group Rickettsia parkeri, an obligate intracellular bacterium that causes an eschar-like lesions at the site of tick attachment (1,2). Vector competence is influenced by the ability of a pathogen to establish infection, replicate, and disseminate across vector tissues, processes counteracted by the arthropod's innate immune system. Similar to vertebrates, the tick immune system has both cellular and humoral arms. The cellular arm is represented by hemocytes, which are professional immune cells equivalent to vertebrate leukocytes. Following microbial infection, tick hemocytes execute cell-mediated responses including phagocytosis, encapsulation, and nodulation, clearing microbes from the system (3)(4)(5)(6). The humoral arm of the tick immune system produces soluble effector molecules that activate the complement pathway, prophenoloxidase pathway, and melanization cascade and produce reactive oxygen and nitrogen species (7)(8)(9)(10)(11). Hemocytes also secrete effectors that eventually activate humoral responses (9,12).
Tick hemocytes have historically been classified into prohemocytes, plasmatocytes, granulocytes, spherulocytes, and oenocytoids (13) based on their quantity, size, shape, nuclearcytoplasmic ratio, and presence of inclusion bodies. Recent ultrastructural studies have narrowed the classification to prohemocytes, granulocytes I, and granulocytes II (14,15), and functional characterization of tick hemocytes established the existence of core invertebrate hemocyte functions. Both hard and soft tick hemocytes use phagocytosis as the primary defense mechanism (3,(16)(17)(18), especially in the granulocyte and plasmatocyte subsets (8,16,19,20). Hemocytes can phagocytose tick-transmitted pathogens, such as hemocyte engulfment of Borrelia spirochetes, in a process described as coiling phagocytosis (21). Ixodes scapularis hemocytes were also reported to be infected with Anaplasma phagocytophilum, a requirement for subsequent salivary gland infection (22). The release of effector molecules complements hemocyte-mediated responses as part of the humoral defense response and include several pathogen recognition molecules such as lectins, antimicrobial peptides (AMPs), and thioester-containing proteins (23-25). However, the molecular mechanisms underlying tick hemocyte-pathogen interactions remain largely uncharacterized.
Here we used a conservative immunofluorescence and morphological approach to classify hemocyte subpopulations in Am. maculatum. Through in vivo phagocytosis of fluorescent beads, we functionally differentiate phagocytic from non-phagocytic hemocytes. For the first time, we demonstrate depletion of phagocytic hemocyte subsets using clodronate liposomes (CLD) in a tick species, thereby defining a role for phagocytic hemocytes in innate immune responses against bacterial challenge. Nextgeneration transcriptome analysis of uninfected and R. parkeriinfected hemolymph samples reveals molecular changes associated with pathogen recognition, immune pathway activation, and hemocyte production, amongst others. Using RNA interference, we characterize two previously unreported molecular markers of hemocyte phagocytosis in ticks. We also show that R. parkeri infects phagocytic hemocytes, thereby possibly playing a vital role in the systemic dissemination of R. parkeri to other tissues.

Tick maintenance and rearing
Am. maculatum ticks were maintained at the University of Southern Mississippi following previously established protocols for hard ticks (26). Established laboratory colonies of Rickettsia parkeriinfected ticks were generated from questing ticks. Unfed adult ticks were collected using the drag-cloth method during the summer months of 2019 from Mississippi Sandhill Crane, National Wildlife Refuge, Gautier, Mississippi (https://www.fws.gov/refuge/ mississippi_sandhill_crane/). R. parkeri-free Am. maculatum ticks were obtained from the tick rearing facility at Texas A&M (TAMU, College Station, TX, USA) tick-rearing facility. These ticks were fed on cattle, which clears R. parkeri from tick tissues across feeding stages.

Hemolymph perfusion and hemocyte quantification
Before hemolymph collection, unfed or partially fed ticks were cleaned in a 10% sodium hypochlorite solution for 5 minutes, followed by a 10-minute wash in 70% ethanol and cleaning using double distilled water. Hemolymph was collected from ticks as previously described (27) using a freshly prepared modified citrate-EDTA anticoagulant buffer (vol/vol 60% Schneider's Drosophila medium and 70% 5 mM EDTA in 1X PBS) (28) and kept on ice until needed. A modified perfusion method was used to collect hemolymph from unfed ticks. Briefly, ticks were placed with their ventral side facing downwards on double-sided tape mounted on a petri dish, which was then placed on ice for approximately 20 minutes to stimulate hemolymph flow within the tick. 0.5 µL of anticoagulant solution was injected into ticks between the basis capituli and scutum. Injected ticks were kept recovering at room temperature for 20 minutes before hemolymph perfusion. 1-2 incisions were made between the ridges of the festoon using a 33G removable needle (Hamilton Company, Franklin, MA, USA). Immediately, 5 µL of anticoagulant solution was injected from the basis capituli and the exiting hemolymph was collected in a 1.5 mL microcentrifuge tube. Perfused hemolymph was centrifuged at 500 rpm for 3 minutes at 4°C to pellet the hemocytes, and the supernatant was removed. Hemocyte pellets were resuspended in a fresh anticoagulant buffer and kept on ice until needed to allow for the collection of hemolymph from unfed ticks. Total hemocytes were quantified using the trypan blue exclusion method (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). Briefly, perfused hemolymph was mixed with 0.4% trypan blue, and 10 mL of the mixture was pipetted into a cell counting chamber. Hemocytes were quantified using a Countess automated cell counter (Invitrogen, Thermo Fisher Scientific Waltham, MA, USA). The number of hemocytes was estimated by placing perfused hemolymph onto the grooves of an improved hemocytometer chamber (Bright-Line, Hausser, Scientific Horsham, PA, USA), and hemocytes were differentiated morphologically as described previously (29,30). A total of five ticks was used for hemocyte quantification. Three separate count was carried out from individual ticks and the average represent the total hemocyte population from the tick.

Chemical depletion of phagocytic hemocytes
Phagocytic cells were depleted as previously described with slight modifications (31,32). To deplete circulating phagocytic hemocytes from the tick's hemolymph, clodronate liposomes (CLD) or control liposomes (LP) (Standard Macrophage Depletion Kit, Encapsula Nano Sciences LLC, Brentwood, TN, USA) were injected into the hemolymph using a 2.5 mL 600 series Hamilton microliter syringe connected to a 33G removable needle (Hamilton Company). The optimal concentration of CLD and LP necessary for depletion with no adverse effect on survival was initially determined by injecting CLD and LP (stock, 1:2, 1:5, and PBS) in 1X PBS into groups of 15 ticks. These ticks were then monitored for survival over eight days beginning 24 h post-injection. Subsequent depletion experiments were performed using 1:5 dilution of CLD and LP in 1X PBS.

Bacterial challenge and survival analysis
Escherichia coli (E. coli) strain DH5-alpha and Staphylococcus aureus (S. aureus) strain RN4220 were grown overnight at 37°C in LB and TSA media, respectively. Overnight cultures were carefully removed, centrifuged, and adjusted to a concentration of OD 600 = 0.5 for E. coli and OD 600 = 0.1 for S. aureus in 1X PBS. R. parkeri were cultured as previously described (33). Frozen stocks of R. parkeri were revived by infecting Vero cells. After the cells were revived and replicated, the concentration of rickettsiae was determined using the plaque assay. Rickettsiae were isolated from Vero cells by lysis using sonication (BioRuptor ™ Pico, Denville, NJ, USA) for 5 minutes in cycles of 30 seconds on and 30 seconds off at 4°C. After sonication, the suspension was centrifuged at 1000 x g for 5 minutes at 4°C to pellet cell debris, and the supernatant was passed through a 0.22 µm syringe filter (Fisher Scientific, Grand Island, NY, USA). R. parkeri was stored in SPG medium on ice until ready for use. Ticks were challenged with 0.2 µL of 10 7 rickettsiae 24 h post-CLD or LP injection. LPS and heatkilled S. aureus were used as positive controls for bacterial injection. The injection of sterile 1X PBS was used as a negative and injection control. Ticks were maintained and constantly monitored every 24 hours for signs of mortality.

In vivo phagocytosis assay
The phagocytic function of tick hemocytes was assessed by injecting fluorescent-conjugated carboxylated beads as previously reported (32, 34) with slight modifications to adapt for use in ticks. Briefly, ticks were injected with 2% (vol/vol) 0.2 µL of yellow-green Carboxylated-Modified Microspheres (Thermo Fisher Scientific, Waltham, MA, USA) diluted in anticoagulant buffer and allowed to recover in incubators for 4 hours at 22°C and 95% relative humidity (RH). Subsequently, hemolymph was perfused using an anticoagulant solution (vol/vol 60% Schneider's Drosophila medium and 70% 5 mM EDTA in 1X PBS). Perfused hemolymph was allowed to adhere on a glass microscope slide for 1 h at room temperature. Hemocytes were fixed in 4% paraformaldehyde (PFA) in 1X PBS for an additional hour. Fixed hemocytes were washed with 1X PBS. Hemocytes were incubated with 20 mM Hoechst 33342 (Thermo Fisher Scientific) diluted in 1X PBS and 75 mM Vybrant CM-Dil (Invitrogen, Carlsbad, CA, USA) for 1 hour at RT, after which slides were washed three times in 1X PBS and allowed to dry. Slides were mounted on a coverslip in 10 mL Fluoromount-G mounting medium (SouthernBiotech, Birmingham, AL, USA). Vybrant CM-Dil is a dye that stains the hemocyte cell membrane thus providing conclusive evidence for the presence of the injected beads within or outside of the hemocytes. A total of five ticks were used to quantify phagocytic hemocytes. For each tick, three separate fields were counted (200 hemocytes per field of view) and the proportion of phagocytic hemocytes out of the total number of hemocytes was estimated.

In-vivo EdU incorporation assay
We estimated hemocyte differentiation by visualizing and quantifying the synthesis of new DNA in vivo based on 5-ethynyl-2′-deoxyuridine (EdU) incorporation and subsequent detection using the Click-iT EdU Alexa Fluor 647 kit (Invitrogen, Grand Island, NY, USA) as previously described (35, 36), with the only exception that 0.5 µL of 20 mmol l−1 EdU in anticoagulant buffer was injected into ticks. Ticks were allowed to recover in incubators for 4 hour at 22°C and 95% RH. Following recovery, hemolymph was perfused from ticks and allowed to attach on a microscope glass slide for 1 hour at 4°C. Hemocytes were subsequently fixed with 4% paraformaldehyde diluted in 1X PBS for 30 minutes at RT, washed three times with 3% bovine serum albumin (BSA) in 1X PBS, permeabilized for 30 min with 0.5% Triton-X in PBS at RT, followed by another wash step with 3% BSA in 1X PBS. Hemocyte slides were subsequently incubated in the dark with the Click-iT reaction cocktail for 30 minutes at RT according to the manufacturer's instructions, followed by a wash step with 3% BSA in 1X PBS. Hemocyte slides were subsequently incubated with 20 mM Hoechst 33342 (Thermo Fisher Scientific) diluted in 1X PBS for 1 hour at RT, after which slides were washed three times in 1X PBS and allowed to dry before mounting on a microscope glass slide by adding 10 mL Fluoromount-G mounting medium (SouthernBiotech).

Hemocyte staining
Tick hemolymph was perfused into an anticoagulant buffer and allowed to adhere to a glass coverslip in a humid chamber at 4°C for 1 hour. Without washing, hemocytes were fixed by adding 4% PFA solution in 1X PBS for an additional 1 hour at RT. After fixation, cells were washed three times with 1X PBS and permeabilized with 0.1% Triton X-100 for 1 hour at RT. Without washing, hemocytes were blocked in 1% BSA solution in 0.1% Triton X-100 for an additional 1 hour at RT. Excess blocking solution was washed with 1X PBS.
Hemocytes were incubated with 1U phalloidin (Alexa Fluor ™ 488 Phalloidin, Molecular Probes, Thermo Fisher Scientific) and 20 mM Hoechst 33342 (Molecular Probes, Thermo Fisher Scientific) diluted in 1X PBS for 1 hour at RT, after which slides were washed in 1X PBS and allowed to dry before mounting on a microscope glass slide by adding 10 mL Fluoromount-G mounting medium (SouthernBiotech).
2.8 RNA extraction, cDNA synthesis, and qRT-PCR Hemolymph was collected from 10 individual ticks as described above and pooled, and an anticoagulant buffer was added to a total volume of 250 mL. RNA was extracted using the Trizol-chloroform separation and isopropanol precipitation method with slight modifications. Following the initial chloroform separation of RNA into the aqueous phase, a second separation was performed by adding a 1:1 volume of chloroform to the aqueous phase, centrifuging at maximum speed (19,980 RCF), and the transparent upper phase was used to proceed with isopropanol precipitation. Second, an ethanol wash of the RNA pellet was carried out twice to help completely remove the isopropanol carryover. The RNA pellet was air dried and resuspended in 30 mL of nuclease-free water, concentration and quality checked, and stored at -80°C until use. Complementary DNA synthesis and qRT-PCR were conducted as previously described (37). Sequences of gene-specific primers designed to amplify cDNA fragments are listed in Supplementary Table S1. Transcriptional gene expression was normalized against the Am. maculatum b-actin gene. The synthesized cDNA was used to measure mRNA levels by qRT-PCR using the CFX96 PCR Detection System (Bio-Rad Inc., Hercules, CA, USA) described previously (37-39).

Double-stranded RNA synthesis and delivery
Double-stranded RNA from the homologs of Drosophila nimrod B2 and eater transcripts was synthesized for gene silencing and microinjected into unfed adult female Am. maculatum ticks as previously described (37)(38)(39)(40)(41). Before injection, dsRNA targeting each gene was diluted to a working concentration of 1 mg/mL in nuclease-free water. Double-stranded RNA from the green fluorescent protein (Gfp) gene was synthesized and injected as an irrelevant control.

Illumina sequencing
RNA samples from uninfected and R. parkeri-infected Am. maculatum hemolymph were sent for sequencing by Novogene (China). Briefly, partially blood-fed (~50 mg, slow blood feeding phase and~200 mg, start of fast feeding phase) ticks were selected for RNA sequencing. Hemolymph was collected from 120 partially blood-fed ticks during the slow-feeding (~50 mg) and fast-feeding phases (~200 mg). Three biological replicates of hemolymph from the slow-feeding and fast-feeding phases were included in each sample of the R. parkeri-infected or uninfected group, i.e., a total of 12 samples. Hemolymph from ten partially-fed ticks was combined for each biological replicate. Hemolymph RNA was extracted as described above. RNA libraries were constructed from hemolymph RNA from six uninfected and six R. parkeri-infected ticks using the NEBNext Ultra ™ RNA library Prep Kit (New England Biolabs, Ipswich, MA, USA). RNA library preparation and sequencing were conducted by Novogene Co., Ltd. (Beijing, China).

Bioinformatics analysis
Raw reads were stripped of contaminating primers, and bases with qual values <20 were trimmed. Clean reads were assembled using the Abyss (42) and Trinity (43) assemblers. Resulting contigs were reassembled with a blastn and cap3 assembler (44) pipeline as described previously (45). Coding sequences were extracted based on blastx results derived from several database matches, including a subset of the non-redundant NCBI protein database containing tick and other invertebrate sequences, as well as the Swiss-Prot and Gene Ontology (GO) databases. All open reading frames larger than 200 nucleotides were extracted, and those matching known proteins or with a signal peptide were retained. The resulting peptide and coding sequences were mapped to a hyperlinked spreadsheet including blastp and rpsblast matches to several databases and an indication of the signal peptide (46), transmembrane domains (47), and O-galactosylation sites (48). edgeR was used in ancova mode to detect statistically significant differentially-expressed genes according to feeding or infection status (49). edgeR inputted the read matrix for genes with at least one library expressing an FPKM (fragments per thousand nucleotides per million reads) equal to or larger than 10. For heat map visualization of CDS temporal expression, Z scores of the FPKM values were used. All deduced coding sequences and their reads are available for browsing with hyperlinks to several databases (Supplementary Table S2).
2.12 RNA-seq and differential gene expression analysis As previously described, differentially expressed genes from edgeR analysis were analyzed using the iDEP (integrated Differential Expression and Pathway analysis) online tools (50). The expression matrix representing read counts of differentially-expressed genes and the gene IDs were uploaded to the iDEP user interface and used for data exploration.

Immunofluorescence of R. parkeri
Hemolymph from unfed or partially fed ticks was perfused onto a microscope coverslip, and hemocytes were allowed to adhere for 1 hour at RT. Hemocytes were fixed in 4% PFA (4% in PBS; J19943-K2, Thermo Fisher Scientific) for 30 minutes at RT. Coverslips containing hemocytes were permeabilized with 0.1% Triton X-100 for 30 minutes at RT. For non-permeabilized coverslips, 1X PBS was added to the hemocytes for 30 minutes at RT. This step was followed by three times washing with 1X PBS. Non-specific proteins were blocked with 1% BSA solution in PBS for 1 hour, followed by primary incubation with mouse anti-Rickettsia M14-13 (generously provided by T. Hackstadt, NIH/NIAID Rocky Mountain Laboratories (51,52), and rabbit anti-Sca2 (generous gift from Matthew D. Welch, UC Berkeley). A no primary antibody control sample (negative control) was prepared in parallel. R. parkeri was detected using goat anti-rabbit Alexa Fluor 568 and goat anti-mouse Alexa Fluor 568 (1:500 in 1% BSA; Invitrogen, Thermo Fisher Scientific). Samples were washed three times in 1X PBS to remove free antibodies. Hemocytes were incubated with 20 mM Hoechst 33342 (Molecular Probes, Thermo Fisher Scientific) diluted in 1X PBS for 1 hour at RT, after which slides were washed three times in 1X PBS and allowed to dry before mounting on a microscope glass slide by adding 10 mL Fluoromount-G mounting medium (SouthernBiotech). To test for lysosomal activity, hemocytes from R. parkeri infected ticks were stained with the acidotropic dye LysoTracker Red (Invitrogen) for 1 hour at room temperature, washed thrice in PBS and fixed with 4% paraformaldehyde for 30 minutes. Hemocytes were then processed for R. parkeri immunostaining as described previously

Imaging acquisition
Confocal images were acquired with a Leica STELLARIS STED (Leica Microsystems, Wetzlar, Germany) confocal microscope using either a 40X, 63X, or 100X objective (zoom factor 3-5; numerical aperture of 1). Images were obtained using both sequential acquisition and variable z-stacks. The 405 UV laser was used to acquire the DAPI channel, while the tunable white light laser (WLL) was used to capture the Alexa-Fluor channel. A z-stack of the images consisting of 150-250 slices was compiled for all images captured, and the proprietary Leica built-in post-processing plugin was used for deconvolution and to carry out lightning processing. All images were exported as acquired and compiled in PowerPoint software.

Discrimination of hemocyte types
Light microscopic examination of direct hemolymph smears revealed a heterogeneous hemocyte population. There were two distinct small and large cell populations, the latter comprising cells with varying cytoplasmic contents, nuclear shape, and cytoplasm size (Supplementary Figures S1A, B). The position of the nucleus was variable: certain hemocytes possessed large, centrally placed nuclei occupying most of the cytoplasmic space, while in some cells the nucleus was peripheral and binucleated. Variable granulation was also observed in the cytoplasm of some hemocyte types.
However, the resolution of light microscopy limited our discrimination of hemocyte subsets. We therefore assessed whether we could further classify hemocytes using commonly used fluorescent markers. Perfused hemolymph was stained with wheat germ agglutinin (WGA), Vybrant CM-Dil (a lipophilic cell membrane stain), and Hoechst 33342 (a nuclear stain). WGA discriminated hemocyte populations with varying degrees of binding intensity, indicating a potential difference in hemocyte function based on their lectin binding activity. By contrast, all hemocytes were positive for CM-Dil (Supplementary Figure S1C). To further differentiate between hemocyte subtypes, we co-stained with phalloidin (an actin stain) and DAPI (a nuclear stain). Five distinct hemocyte types were identified based on shape, actin projections, and nuclear-cytoplasmic ratio: (i) granulocytes were relatively large and had multiple actin projections; (ii) plasmatocytes were pyriform with a centrally placed nucleus; (iii) spherulocytes possessed a peripherally placed nucleus; (iv) prohemocytes were characterized by a relatively high nuclear to cytoplasmic ratio; and (v) oenocytoids had a smaller nuclear to cytoplasmic ratio ( Figure 1A).
The total hemocyte population differed significantly between partially fed and unfed female ticks, with hemocyte numbers increasing after the blood meal (Supplementary Figure S1D). Feeding also affected the hemocyte subtype distribution in male and female ticks: feeding significantly increased granulocyte numbers in females (Supplementary Figure S1E) and significantly decreased the spherulocyte population in both male and female ticks (Supplementary Figure S1F). Plasmatocyte numbers were lower in blood-fed females than unfed females but increased in males on feeding (Supplementary Figure S1G). There was a trend to oenocytoids numbers increasing following feeding in both males and females (Supplementary Figure S1G), but prohemocytes were absent in both male and female ticks following feeding (Supplementary Figure S1I).
Hemocyte populations were histomorphologically similar in male and female ticks, so we next examined functional differences between male and female hemocytes by assessing their phagocytic abilities. The proportion of phagocytic hemocytes were higher in female ticks than in male ticks in an in vivo phagocytosis assay of yellow-green FluoSpheres ( Figures 1B, C). Taken together, these data demonstrate that Am. maculatum hemocyte heterogeneity might influence hemocyte function.

Clodronate liposomes deplete and impair phagocytic hemocyte functions
In the absence of definitive molecular markers of hemocyte subtypes in ticks, there is a need for alternative tools to study the role of hemocytes in cellular immunity and vector competence. To this end, clodronate liposomes (CLD), a pharmacological agent that specifically targets and depletes professional phagocytes via apoptosis (31,32,53,54), was used to deplete phagocytic hemocytes in tick hemolymph. CLD exclusively targets cells with phagocytic abilities (55). Upon phagocytosis, the liposome is degraded by the lysosome, which releases toxic clodronate to promote apoptosis ( Figure 2A).
We first tested different clodronate and control liposome (LP) concentrations to determine their impact on tick survival ( Figure 2B). Injection of CLD or LP at a 1:5 dilution had no adverse impact on tick survival ( Figure 2C) but significantly reduced total hemocyte populations ( Figure 2D) due to reduced numbers of phagocytic hemocytes ( Figure 2E). To determine the effect of the blood meal on phagocytic hemocyte depletion, we injected unfed ticks with a 1:5 dilution of CLD or LP before a blood meal and quantified the phagocytic hemocyte population following partial feeding. The blood meal did not interfere with the ability of CLD to deplete phagocytic hemocytes ( Figure 2F).
To further assess the effect of CLD depletion on phagocyte function, ticks injected with CLD or LP were fed and in vivo hemocyte phagocytosis assayed with yellow-green FluoSpheres. Hemolymph of LP-injected ticks contained more hemocytes that engulfed one or two FluoSpheres than hemocytes from clodronatedepleted ticks ( Figure 3A). Hemocyte populations have previously been shown to increase following a blood meal due to cellular division (36), so we determined the impact of clodronate depletion on hemocyte DNA replication. Partially blood-fed CLD or LP-injected ticks were injected with EdU, and their hemocytes were assayed for EdU incorporation into hemocyte DNA. Significantly more EdU was incorporated into LP-injected tick hemocytes compared with CLDinjected ticks ( Figure 3B). Together, these data confirm effective chemical depletion of the phagocytic hemocyte population and show that CLD interferes with the abilities of phagocytic hemocytes to undergo replication.

Depletion of phagocytic hemocytes impairs survival against bacterial challenge
Hemocytes are a vital defense mechanism against invading microbes in ticks, mosquitoes, and Drosophila (32, 33). We therefore attempted to determine how immunocompromised ticks would survive challenge with both Gram-positive and Gram-negative bacteria ( Figure 4A). Ticks were unaffected by injection with PBS ( Figure 4B). Phagocyte depletion impaired tick survival against Gram-negative E. coli ( Figure 4C), but Gram-positive S. aureus (live and heat-inactivated) significantly affected tick survival in both CLD and LP-treated groups ( Figure 4D). Similarly, phagocyte depletion significantly impaired survival against Am. maculatum-transmitted R. parkeri ( Figure 4E). Since phagocytic granulocytes act as scavengers of invading microbes (6,31,32), these data further support the hypothesis that phagocytic hemocytes are critical components of the immune response in ticks and maintain tick microbial homeostasis via cell-mediated immunity and interactions with pathogenic microbes (6, 56).

R. parkeri can infect circulating hemocytes
The tick hemolymph contains a heterogeneous population of circulating hemocytes, as shown by ourselves ( Figure 1A) and others (4,7,10,15,30,(57)(58)(59). R. parkeri acquired during a blood meal must circumvent both cellular and tissue barriers in the midgut to access hemolymph for systemic dissemination and subsequent transmission to a mammalian host. In the hemolymph, R. parkeri must either avoid, evade, or suppress hemocyte-mediated immune responses to successfully disseminate. R. parkeri is closely related to Anaplasma (A.) phagocytophilum, both existing as obligate intracellular pathogens and belonging to the same Rickettsiales order. Dissemination of A. phagocytophilum to the salivary gland in its tick vector is facilitated by direct hemocyte infection following midgut colonization by the bacteria (22). Since our understanding of how R. parkeri disseminates through the hemolymph from the midgut to other tissues is still limited, we asked whether R. parkeri can infect circulating hemocytes. Hemolymph from infected and unfed female Am. maculatum were incubated with primary antibodies targeting the outer membrane of R. parkeri, and R. parkeri was detected in the cytoplasm of hemocytes ( Figure 5A). In addition, intracytoplasmic infection was detected in the hemocytes of uninfected ticks previously injected or capillary-fed with GFP-expressing R. parkeri (Figures 5B, C). While positive Sca2 staining suggested the presence of R. parkeri in hemocytes, this could also have occurred through binding of R. parkeri to the surface of hemocytes arising from the immune response. However, detection of R. parkeri Sca2 signal in both permeabilized and unpermeabilized hemocytes further confirmed active entry into hemocytes. Lysosomal imaging of R. parkeri infected hemocytes showed very few bacteria associated with the lysosomal compartments (Supplementary Figure S3). Intracellular, Clodronate liposomes deplete hemocyte functions. Clodronate liposomes impaired hemocyte phagocytosis (A) and interfered with EdU incorporation into hemocytes DNA (B). Scale bar = 20 mm. Confocal microscopy images of phagocytic hemocytes infected with Rickettsia parkeri. Representative confocal images of immunofluorescence staining for R. parkeri proteins showing hemocytes from natural and artificially infected ticks. (A) Immunolocalization of R. parkeri in hemocytes of naturally infected Am. maculatum. Hemocytes were incubated with primary antibodies targeting R. parkeri outer membrane protein (red) and Hoechst 33342 (blue). Infection of hemocytes with R. parkeri following (B) capillary feeding and (C) microinjection of GFP-expressing R. parkeri into uninfected Am. maculatum. Hemocytes were incubated with R. parkeri Sca2 antibody (red) and Hoechst 33342 (blue). Hemocytes were fixed, permeabilized and blocked prior to incubating with antibody. Scale bar = 5 mm. mosquito hemocytes, respectively (61), further corroborating our observations.

RNA-seq reveals changes in hemocyte gene expression associated with R. parkeri infection
Identifying hemocyte-specific transcripts is crucial for the discovery of immune factors participating in hemocyte-mediated immune responses. While tick hemocyte transcripts have previously been characterized (62), the effect of pathogen infection on tick hemocyte gene expression is unknown. Here we generated and compared hemocyte transcripts from R. parkeri-infected (n = 6) and uninfected (n = 6) Am. maculatum. Assembly of the 305,276,990.3 reads from 12 libraries allowed us to identify 37,430 CDS, and database searching matched the CDS into 28 categories (Table 1). Secreted proteins accounted for 18.7% and 38.8% of the coding sequences and mapped reads, respectively, and included enzymes, protease inhibitors, lipocalins, and immune-related genes. Coding sequences classified with immune functions represented 0.86% and 2.45% of the total CDS and reads, respectively. More than 30% of coding sequences and 3.1% of total reads represented sequences of unknown function. Sorting of significantly expressed coding sequences with their respective reads identified 2,859 differentially expressed CDS between R. parkeri-infected and uninfected hemocytes (Supplementary Table S3). Of these, 938 belonged to the secreted class, 46 to the immunity class, 568 were unknown, and 48 were associated with cytoskeletal functions. Forty bacterial-derived coding sequences were also significantly differentially expressed between R. parkeri-infected and uninfected hemocytes. In addition, we identified 39 coding sequences with functions as regulators of hematopoiesis, hemocyte differentiation, and immune functions. Of the 39, 14 (30.8%) were significantly differentially regulated between R. parkeri-infected and uninfected hemocytes (Table 2), and these genes include transglutaminases, astakines, hemocytin, prokineticin, thymosin and the two Drosophila homologs eater, nimrod B2, Runt-related transcription factor (Runx) and GATA-binding factor (GATA). We also identified coding sequences in the Toll and immune defi ciency (IMD) immune pathways.  Figure S4A). Seven of the nine GATA transcription factors and all the Runt domaincontaining sequences were significantly upregulated in R. parkeri-infected hemocytes. GATA factors and Runt proteins have previously been shown to be critical in maintaining pluripotent hemocyte precursors in the hematopoietic organ (23, 63).

Laminin receptors and CLIP-domain serine proteases
Laminin receptors are a group of proteins with diverse biological functions, including cellular differentiation. They serve as binding partners with different homeostasis-associated proteins to maintain hemocyte homeostasis (67). The CLIP-domain serine protease (CLIPsp) is abundantly present in the hemolymph of insects and arthropods. Four transcripts with a Clip or disulfide knot domain (AmHem-164283, Amac-hemSigP-457530, AmHem-462558, and AmHem-462556) were differentially expressed in our dataset (Supplementary Figure S4A): R. parkeri led to six-fold upregulation of AmHem-AmHem-164283, 14-fold upregulation of Amac-hemSigP-457539, and two-fold upregulation of AmHem-462558 and AmHem-462556. In invertebrates, these proteins play dual roles in innate immune responses and hematopoiesis, acting as binding partners of toll-like receptor Spaetzle, leading to downstream transcriptional activation of antimicrobial peptides. Similarly, they activate the prophenoloxidase (PPO) cascade necessary for melanization (68). A direct role has been described for CLIPsp-induced PPO maintenance of hematopoiesis (64).

Nimrod B2 and eater mediate hemocyte phagocytosis
Hemocytes participate in humoral and cellular defenses in response to microbial infections. Their specific roles are defined by their expressed cell surface receptors. Nimrod B2 and eater, which mediate microbial phagocytosis upon infection (81, 82), were downregulated in hemocytes from R. parkeri-infected ticks. Their role in hemocyte phagocytosis and as markers of phagocytic hemocytes have been described in hematophagous and nonhematophagous organisms (81)(82)(83)(84). We therefore further defined the role of nimrod B2 (AmHem-305744; 40% homology to Drosophila nimrod B2) and eater (AmHem-270031; 32% homology     to Drosophila eater) homologs in hemocyte phagocytosis using a combination of RNAi and in vivo phagocytosis approaches ( Figure 7A). Quantitative PCR validation of their expression supported the RNA-seq result ( Figure 7B). Their transcripts were significantly depleted in dsRNA-injected tick hemocytes ( Figure 7C), and the in vivo bead phagocytosis assay also revealed depletion in hemocyte phagocytosis of yellow-green FluoSpheres (beads) ( Figure 7D). The Drosophila homologs, nimrod B2 and eater were both significantly downregulated in CLD-depleted hemocytes in our phagocytosis assay ( Figures 7E, F), confirming their specificity to phagocytic hemocytes. Together, these data demonstrate that nimrod B2 and eater are two potential candidate marker genes regulating hemocyte phagocytosis.

Discussion
Here we report morphological and functional heterogeneity in Am. maculatum hemocytes. We define a role for phagocytic hemocytes in the immune response against bacterial infections and identify potential molecular markers of hemocyte phagocytosis. We report for the first time direct evidence of R. parkeri infection of phagocytic hemocytes, which might play a role in the dissemination of these organisms throughout the tick body.
Previous studies have identified different hemocyte subtypes in hard (19,29,30) and soft (3,59) tick species, which vary depending on the developmental stage, infection status, and sex. Here we i d e n t i fi e d fi v e u n i q u e h e m o c y t e s u b t y p e s b a s e d o n histomorphological analysis. Female adult ticks pose the most threat to human and animal hosts through hematophagy and pathogen transmission, but we did not detect significant differences between hemocyte populations in unfed male and female ticks. Nevertheless, there were functional differences between male and female ticks, with hemocytes from female ticks displaying more phagocytic hemocytes than those from males. The functional distinction between male and female ticks could be attributed to the presence of more phagocytic plasmatocytes in female than male ticks. More phagocytic hemocytes may also be needed in female hemolymph due to their large body size relative to males, with the prolonged feeding time on the host of females increasing the chance of microbial growth within the tick, thus necessitating a more robust and primed immune system. We however noted that some hemocytes did not stain with Vybrant-CM Dil which could be attributed to the loss of membrane structure that occur during cell death. Immune priming in invertebrates mimics the vertebrate adaptive immune response  (85)(86)(87), only one study has described the presence of active immune priming in ticks (88). We also observed that blood feeding increased granulocytes in female and plasmatocytes in male ticks. Although granulocytes are professional phagocytic hemocytes, plasmatocytes have also been shown to display phagocytic functions. The decline in the prohemocyte population in blood fed ticks suggests their differentiation to mature form of hemocytes. Prohemocytes are immature hemocytes that can differentiate into other hemocyte types, and their absence would indicate that new hemocytes are produced during feeding, with prohemocytes the source of those new hemocytes. Together, these data for the first time show the heterogeneous nature of the Am. maculatum hemocytes and functional differences in male and female hemocytes. Developing specific functional or molecular markers will now be important to confirm our morphological classification.
Chemical inhibitors have been widely used to characterize mammalian immune cells. In the absence of molecular markers to characterize tick hemocytes, hemocytes can be studied and characterized by inhibiting their functions, similar to the widely used reverse genetic approach for gene characterizations. Clodronate liposomes (CLD) are widely used to deplete phagocytic macrophages in mammalian systems (53,54,89). CLD mediates phagocyte killing by releasing toxic clodronate upon macrophage phagocytosis, which subsequently induces apoptosis (53). Phagocytic hemocytes have been depleted with CLD in mosquitoes and Drosophila (31,32). We also showed that CLD can successfully reduce the phagocytic hemocyte population in Am. maculatum, as evidence by a reduced proportion of phagocytic hemocytes and subsequent loss in functional phagocytosis confirmed by immunofluorescence on CLD treatment. The EdU incorporation assay employed in this study further shows that CLD could also interfere with the replication of hemocyte, however the mechanism behind this process requires further investigation. These findings were further supported by a decrease in transcript levels of nimrod B2 and eater homolgs in CLD-treated ticks, both of which are well characterized phagocyte markers in mosquitoes (32, 82) and Drosophila (80,81,83). In the tick system, we showed that nimrod B2 and eater homolog knockdown significantly reduced the function of phagocytic hemocytes, further suggesting that these two genes may be useful candidate markers of phagocytic hemocytes in Am. maculatum. By showing successful depletion of phagocytic hemocytes using CLD, these experiments provide the means to functionally characterize phagocytic hemocytes in Am. maculatum and other tick species. Our experiments also serve as proof of concept for using CLD in the functional study of phagocytic hemocytes in a non-model organism such as ticks. Due to the lack of specific antibodies targeting these two genes, further characterization of their functions is limited.
As the first component of the cellular immune response, phagocytic hemocytes are primarily responsible for scavenging and removing invading pathogens, mainly via phagocytosis. The reduced survival of CLD-depleted ticks following bacterial challenge further highlights the critical role of hemocytes in maintaining immune responses in ticks, consistent with immune impairments following CLD depletion seen in mosquitoes and Drosophila (31,32). We show that survival is reduced in CLD-depleted ticks following the R. parkeri challenge. Although the pathogenicity of R. parkeri to its tick vector is unclear, this result suggests that phagocytic hemocytes have a role in controlling R. parkeri infection. While ours is the first functional validation of phagocytic hemocytes by CLD depletion in ticks, two studies used alternative approaches to inhibit hemocyte function and demonstrated reduced survival in ticks following bacterial challenge (6,11). The outcome of our experiments and similar reports in other organisms highlight the functional conservation of these immune cells in invertebrates.

FIGURE 6
Representative signaling pathways and immune-related genes. Reconstruction of the immune-signaling genes derived from R. parkeri-infected and uninfected hemocytes showing the components of the (A) Toll and (B) IMD signaling pathways. The Toll and IMD signaling pathways are highly conserved in ticks. Transcripts highlighted in red were not identified in this study.
Hemolymph is a well-defended niche containing hemocytes and several soluble effector molecules that directly inactivate or kill invading microbes (90,91). Pathogens acquired via an infected host's blood must leave the blood bolus and infect the midgut and resident tissues. The process of R. parkeri dissemination from the point of midgut infection to other tick tissues is not fully understood. It has been proposed that rickettsiae can cross the midgut barrier to infect hemocytes during blood feeding (92), and others have also demonstrated Rickettsia infection of the tracheal system (93). The constant bathing of tissues with hemolymph and the persistence of the tracheal system through developmental stages make them viable routes for the systemic maintenance and dissemination of rickettsiae throughout the tick body. In the current study, we detected rodshaped R. parkeri in the hemocytes of naturally infected Am. maculatum. Likewise, capillary feeding and microinjection of GFPexpressing R. parkeri also led to the observation of Rickettsia organisms in hemocytes, suggesting active infection of hemocytes by the pathogen. Dissemination of R. parkeri to the Am. maculatum midgut, salivary gland, and ovarian tissues has previously been described following capillary feeding (94); however, dissemination into the hemolymph and infection of hemocytes is a novel observation that argues for a role for circulating hemocytes in rickettsial trafficking throughout the tick body. Our detection of R.
parkeri in the hemocytes of infected ticks and tissue dissemination of R. parkeri in Am. maculatum salivary glands, midguts, and ovaries reported by Harris et al. (94) and (95) strongly suggest that the pathogen survives long enough in the hostile hemolymph environment to find its way to other tick organs. Similarly, the detection of Rickettsia Sca2 proteins in unpermeabilized and permeabilized hemocytes likely suggest a bacteria mediated entry into the hemocytes or at the very least a direct interaction between the bacteria and the immune cells. Sca2 is a formin-like protein that is used for adherence, cellular entry and actin-based motility in mammalian host cells, however it is not required for R. parkeri infection of the tick vector (94). Infection of tick hemocytes was also shown to enable the migration of A. phagocytophilum from the gut to salivary glands of Ix. scapularis (22). R. parkeri and other vector-borne diseases have evolved a complex process enabling them to colonize and disseminate throughout the arthropod host through transovarial and transstadial transmission (38,39,95,96). Hemocytes, especially phagocytic hemocytes are also required for the dissemination of arboviruses in mosquitoes. Leite at al (61). demonstrated that blocking of phagocytosis in A. aegypti prior to feeding on Zika or Dengue virus infected blood lead to a significant decrease in the prevalence of infection. The evidence presented here and previous report by Liu  Nimrod B2 and eater homologs as functional markers of hemocyte phagocytosis. The role of nimrod B2 and eater silencing on in vivo phagocytosis was evaluated in Am. maculatum hemocytes (A). qPCR validation of bulk RNA expression profiles of nimrod B2 and eater in uninfected and R. parkeriinfected Am. maculatum hemocytes (B). dsRNA was injected into Am. maculatum female ticks to disrupt the expression of nimrod B2 and eater genes and confirmed by qPCR (C). The proportion of phagocytic hemocytes was compared with dsGFP-injected ticks (D). Additional validation of phagocyte depletion showing a significant reduction in nimrod B2 (E) and eater (F) transcript in CLD-injected ticks. Gene expression was normalized to Am. maculatum actin. Data were analyzed using unpaired t-tests in GraphPad Prism v8.4.1. *P < 0.05, ****P < 0.0001.
argues for a wide array of mechanisms by which R. parkeri potentially disseminate through the tick vector. In addition, tracheal infection, and subsequent dissemination from the midgut to the salivary glands have been demonstrated for other rickettsiae as shown with R. monacensis in Ix. scapularis (93). However, further experiment is required to show direct evidence of R. parkeri infection of tick hemocytes and its role in dissemination across tick tissues. Similar to this, salivary gland colonization during feeding is a crucial step for infecting the mammalian host (97,98). The presence of sessile hemocytes (tissue-associated hemocytes) has been described in mosquitoes and Drosophila (99, 100), and these hemocytes are found in those regions of the body that interact most with invading pathogens such as the periosteal region and abdominal walls (99). However, their role in transovarial and transstadial maintenance of pathogens remains unknown. Whether hemocytes play a role in the trafficking of pathogenic bacteria to tick tissues as observed for mosquito phagocytes and viral infection still needs to be established (61). The mechanisms by which R. parkeri enter and survive inside hemocytes also require further examination.
Molecular studies of tick hemocyte biology and hemocytemediated immune responses have been limited due to, in part, the technical challenges surrounding hemolymph collection from different tick stages, the lack of a hemocyte-like cell line (as with Drosophila and mosquitoes), and a lack of hemocyte-specific markers. Current tick cell lines have been utilized extensively to characterize humoral immune responses which are reflective of changes occurring in the tissue in-vivo (101, 102). To identify the molecular responses in Am. maculatum hemolymph during infection, hemocyte transcriptomes from R. parkeri-infected and uninfected ticks were isolated and analyzed. We identified an entire repertoire of transcripts differentially expressed in hemocytes on R. parkeri infection (Supplementary Table S2). We identified several genes that mediate the cellular immune response, especially genes encoding hematopoietic functions and hemocyte differentiation. The anatomical structure and molecular basis of hemocyte production in ticks are still unknown. An earlier study on hemolymph circulation proposed the presence of a lymph-like organ as the site of hemolymph production in ticks (13), but the active production of hemocytes from the tick lymph gland or a specialized hematopoietic organ has yet to be demonstrated. The presence of transcriptional and humoral regulators of hematopoiesis genes suggests that the regulation of hematopoiesis is conserved in arthropod and insect species. For instance, the transcription factors GATA and Runt were detected in our data, which are critical for the proliferation of hematopoietic stem cells and differentiation of myeloid stem cells, respectively, in invertebrates (103). Lozenge, a Runt homolog, has been reported to mediate crystal cell maturation in Drosophila (103, 104), while in crayfish it mediates differentiation from hematopoietic stem cells to mature granular and semigranular cells (64). Kwon et al. also demonstrated a reduction in oenocytoid differentiation following silencing of Lozenge in An. gambiae mosquitoes (55) as well as a decrease in the expression of PPO3/8, which are molecular markers of oenocytoid cells (55). Serpent is a Drosophila GATA factor that regulates hematopoiesis and is an ortholog of the vertebrate GATA family (105, 106). Serpent is expressed in hemocyte precursors but, unlike lozenge, it is still expressed in mature prohemocytes and crystal cells, suggesting roles beyond hematopoiesis (107). No studies have yet identified Runt or its homolog in any tick species; however, a GATA factor has been described in Haemaphysalis longicornis that activates Vitellogenin, which is essential for reproduction (108-110). The presence of transcripts that encode humoral regulators of h e m a t o p o i e s i s , s u c h a s A s t a k i n e s , b -t h y m o s i n , and transglutaminases, was an unexpected finding. Astakines regulate hematopoiesis via interaction with transglutaminases or via direct interactions with hematopoietic cells to promote structural rearrangements (64). Proteins containing thymosin domains have been described in Drosophila, where Ciboulot, a protein with three thymosin domains, regulates axon growth during brain metamorphosis (111). Thypedin, a protein with 27 thymosin domains, regulates foot regeneration in Hydra (112). Ovarian expression of b-thymosin was reported in the dog tick, Dermacentor variabilis (113). In invertebrates, TGases were initially described as coagulation factors and later shown to regulate hematopoiesis in crustaceans (114,115) by maintaining hematopoietic cells in an undifferentiated state, thus suppressing hematopoiesis (116,117). It is currently unclear how these genes regulate the production and differentiation of hemocytes in ticks. However, the presence of some of these genes, such as Astakines, in the genomes of crustaceans, ticks, scorpions, spiders, and other arthropods but not Drosophila or mosquitoes suggests evolutionary divergence in hematopoietic processes (118).
Several PGRP-encoding transcripts were detected, consistent with a previous hemocyte transcriptome study in Ixodes scapularis, where twenty-six PGRP-encoding sequences were detected (62). Genes encoding the Toll and IMD pathways have previously been reported in ticks, although some components of these pathways have yet to be identified (88,(119)(120)(121)(122)(123). While many genes encoding Toll and IMD pathway components were identified in our data, it is unsurprising that some Toll and IMD pathway components were not identified in this study. For instance, the absence of GNBP, Tube, and Pelle transcripts -components of the Toll pathway -and the IMD components IMD, FADD, and DREDD is supported by the absence of these transcripts from the Ix. scapularis and Rhipicephalus microplus genomes (119,124). The differences observed in the components of the immune pathways from our study when compared to other tick species may be a result of the unique evolutionary relationship that exist between the tick vector and pathogenic microbes they vector or that driven by their resident commensal microbial communities. Shaw and colleagues described plasticity in the immune pathways between insects and arthropods, demonstrating activation of the Ix. scapularis IMD pathway by two infection-derived glycerol following pathogen infection (88). Our knowledge of tick immunity has hitherto relied on model organisms like Drosophila and mosquitoes, but our data and that from recent studies on tick immunity argue that immunity in hematophagous arthropods such as ticks is very different (88,119,124).
Cellular immunity relies on hemocytes directly killing invading microbes via phagocytosis, melanization, or the production of reactive oxygen species. Although there have been many studies of cellular immunity in ticks, the roles of specific immune cell types, their molecular signatures, and their contribution to the hemocytemediated immune response are still not fully understood. Our study identifies a repertoire of potential candidate genes that regulate hemocyte functions. An unexpected observation was the identification of nimrod B2 and eater transcripts, two Drosophila homologs which were significantly expressed in uninfected ticks but downregulated on R. parkeri infection. Nimrod B2 and eater are both transmembrane receptors expressed on hemocyte membranes shown to be responsible for the phagocytosis of Gram-positive and Gramnegative bacteria in Drosophila (81,82,84) and mosquitoes (32,82,125). Although the precise roles of nimrod B2 and eater in tick immunity are unknown, our RNAi experiments suggest that they may play a direct role in hemocyte phagocytosis, however the precise mechanisms involved are yet to be clarified. Further studies are now required to determine the cellular localization of these two genes and whether their expression is unique to phagocytic hemocytes from other tick species. Our study also confirms the presence of an active complement component system with several TEPs. The tick complement system is essential for killing tick-transmitted pathogens, as demonstrated by Urbanováand colleagues (4), who demonstrated phagocytosis of the spirochete Borrelia afzelli in Ix. ricinus. A similar role for tick complement-like proteins in the phagocytosis of non-pathogenic microbes was also demonstrated in Ix. ricinus ticks (4, 126, 127).

Conclusion
Here we describe morphological and functional differences in Am. maculatum hemocytes and characterize transcriptional changes in cellular and humoral responses in R. parkeri infected and uninfected tick. Our results reveal heterogenous hemocyte populations in Am. maculatum showing variable phagocytic capacity. We also describe an integral role for phagocytes in responses to microbial pathogens. We for the first-time observed R. parkeri in phagocytic hemocytes. Our transcriptome analysis of R. parkeri-infected and uninfected hemocytes allowed us to explore differentially expressed immunity and hematopoietic genes on R. parkeri infection. This "big" transcriptome dataset will be important for identifying potential biomarkers of hemocyte subtype, function, and production. Our findings also raise important questions about the role of each hemocyte subtype in immune responses and vector competence.

Ethics statement
The animal study was reviewed and approved by Protocols for tick blood-feeding were approved by the University of Southern Mississippi's Institutional Animal Care and Use Committee (USMIACUC protocols #15101501.3 and 17101206.2).