Pulmonary infection is a leading cause of morbidity and mortality after spinal cord injury (SCI). Although SCI causes atrophy and dysfunction in primary and secondary lymphoid tissues with a corresponding decrease in the number and function of circulating leukocytes, it is unknown whether this SCI-dependent systemic immune suppression also affects the unique tissue-specific antimicrobial defense mechanisms that protect the lung. In this study, we tested the hypothesis that SCI directly impairs pulmonary immunity and subsequently increases the risk for developing pneumonia. Using mouse models of severe high-level SCI, we find that recruitment of circulating leukocytes and transcriptional control of immune signaling in the lung is impaired after SCI, creating an environment that is permissive for infection. Specifically, we saw a sustained loss of pulmonary leukocytes, a loss of alveolar macrophages at chronic time points postinjury, and a decrease in immune modulatory genes, especially cytokines, needed to eliminate pulmonary infections. Importantly, this injury-dependent impairment of pulmonary antimicrobial defense is only partially overcome by boosting the recruitment of immune cells to the lung with the drug AMD3100, a Food and Drug Administration–approved drug that mobilizes leukocytes and hematopoietic stem cells from bone marrow. Collectively, these data indicate that the immune-suppressive effects of SCI extend to the lung, a unique site of mucosal immunity. Furthermore, preventing lung infection after SCI will likely require novel strategies, beyond the use of orthodox antibiotics, to reverse or block tissue-specific cellular and molecular determinants of pulmonary immune surveillance.

This article is featured in Top Reads, p.

Pulmonary complications, such as pneumonia, are common and contribute to high rates of mortality and impaired recovery of function after spinal cord injury (SCI) (15). Indeed, individuals with SCI are 37-fold more likely to die of pneumonia than non-SCI individuals (1). How or why SCI increases the risk of developing pneumonia is unknown, but rapid and prolonged suppression of systemic immune function is suspected.

SCI–induced immune deficiency syndrome (SCI-IDS), a phenomenon defined in part by depletion or dysfunction of immune cells, commonly arises after high-level SCI (6). Aberrant postinjury sympathetic reflex control of primary and secondary lymphoid tissues (e.g., bone marrow, spleen, adrenal gland, and lymph nodes) contributes to the onset of SCI-IDS and is associated with higher rates of infection (612). However, little is known about how SCI affects immune defense mechanisms in mucosal tissues such as the lung.

Because most pathogens, including bacteria, that cause pneumonia infect the host through the upper and lower respiratory tracts, gastrointestinal tract, or urogenital tract, mucosal immune responses are uniquely suited to respond to frequent, repeated exposure to pathogens (1315). The current study was designed to characterize cellular and molecular features of lung immunity at postinjury time periods coinciding with SCI-IDS onset and to determine whether lung immune responses are adversely affected by SCI. We hypothesized that critical pathways involved in lung immunity would become impaired after SCI, just as they are in primary and secondary lymphoid tissues, and that this will increase the risk of contracting bacterial pneumonia. New data indicate that cellular and molecular determinants of pulmonary immunity are impaired after SCI and that these deficits are exacerbated when SCI occurs at high spinal levels. Moreover, bacterial pneumonia develops coincident with disruption of pulmonary immune surveillance. Importantly, boosting immune cell recruitment to the lung with injections of the Food and Drug Administration–approved drug Mozobil (plerixafor, AMD3100) only partially protects SCI mice from lung infection. Subsequent unbiased analyses of lung gene transcription reveal that key interrelated signaling networks that control mechanisms of lung resistance and resilience are decreased throughout the acute postinjury period. Taken together, these data indicate that the immune-suppressive effects of SCI extend to the lung and that combinatorial therapies may be required to effectively prevent or treat bacterial pneumonia after SCI.

The Institutional Animal Care and Use Committee of the Office of Responsible Research Practices at The Ohio State University approved all animal protocols. All experiments followed the guidelines and regulations of The Ohio State University and outlines in the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health. All studies used female 8- to 10-wk-old C57BL/6J mice from The Jackson Laboratory (strain no. 000664). All animals received commercial food pellets and chlorinated reverse osmosis water ad libitum. Mice were housed at five mice per cage in ventilated microisolator cages containing corn cob bedding. Housing facilities had a 12-h light/12-h dark cycle at a constant temperature of 20 ± 2°C and humidity of 50 ± 20%.

Experimental injury mice received complete transections at the third thoracic vertebral level (T3Tx). After 5–7 d of acclimating mice to the laboratory environment, mice underwent surgery. On the day of surgery, anesthetized mice received ketamine (80 mg/kg) and xylazine (10 mg/kg) i.p. Of note, surgeries for 12 h postinjury (hpi) and 3 d postinjury (dpi) were typically completed on the same day and were performed in the evening, but before the beginning of the dark cycle to ensure that surgeries and tissue collection times were not during different phases of the light/dark cycle. Surgeries for 28 dpi groups were usually completed in the morning during the light cycle. Aseptic conditions were maintained during all surgical procedures. Body temperature was maintained by placing mice on a warming pad during surgery, and hair was shaved at the thoracic level of the spinal cord. A sequence of betadine–70% ethanol–betadine solutions were used to sterilize skin. To expose the thoracic vertebra, a small midline incision was performed, followed by a partial laminectomy. Lidocaine was topically applied to the spinal cord and then the periosteum and dura mater were opened on the dorsal surface of the spinal cord. Iridectomy scissors and gentle aspiration were used to completely transect the spinal cord by creating a separation between the rostral and caudal stumps of cut cord (7). Completeness of the transection was verified under a dissection scope by the surgeon. Muscle was subsequently sutured shut and the skin incision was closed with surgical staples. Animals were then given sterile saline (2 ml s.c.) and placed into warmed cages until they recovered from anesthesia. All SCI mice were kept on warmers at 35.5°C to maintain body temperature for the duration of the experiment. Mice were also supplemented with saline (1–2 ml s.c.) for 4 dpi, or longer if needed. Surgical staples were removed at 7 dpi. Bladders were manually voided twice daily for the duration of the experiments, and body weight/urinary pH were monitored weekly. Mice appearing dehydrated were supplemented with saline (1–2 ml s.c.), and preinjury weight was maintained with high-caloric nutrient pellets as needed.

Data in Figs. 3–5 were generated in mice subjected to a T1 spinal transection (Tx) injury or sham (laminectomy only) surgery. Surgical procedures were identical to those for T3Tx except that lidocaine was not applied to the spinal cord surface before transection, as we found that this increased mortality after first thoracic vertebral level transection (T1Tx). For T1 laminectomy sham controls, all surgical procedures were the same, but the spinal cord was not transected. Postoperative care was identical to that used in T3Tx mice except that T1Tx mice required additional hydration (supplemented with 3 ml of saline s.c.: 2 ml in the morning and 1 ml in the evening) for the duration of the experiment. Bladders were manually voided twice daily for the duration of the experiments, and preinjury weight was maintained with mash (made from a mixture of water and food pellets) and high-caloric nutrient pellets as needed.

Prophylactic antibiotics are often provided to SCI mice after surgery to prevent bladder infections. However, since antibiotics also could prevent the development of spontaneous pulmonary infections, mice were not given antibiotics at any time during the course of these studies. Omission of antibiotics is a University Laboratory Animal Resources–approved protocol in our laboratory.

After a lethal overdose (2× the surgical dose) of ketamine/xylazine, whole lungs were excised from mice at either 12 hpi, 3 dpi, or 28 dpi. Lungs were collected after bronchoalveolar lavage was performed using 3 ml of lavage fluid (1× HBSS, EDTA, protease inhibitor; lavage fluid was stored for later analysis not included in this study) and were then homogenized into a single-cell suspension in sterile 1× PBS by mashing lungs with the back of a 10-ml syringe plunger through a Falcon 100-μm sterile filter (BD Biosciences) and rinsed with 1× PBS. Lung homogenate was then centrifuged (7 min, 400 × g, 4°C), supernatant was discarded, and cells were resuspended in 1 ml of RBC lysis buffer (4.14 g of NH4Cl, 0.5 g of KHCO3, 0.019 g of EDTA in 500 ml of dH2O) and incubated for 2.5 min at room temperature to remove RBCs. Cells were then flooded with 10 ml of 1× PBS to stop lysis and retrieved via low-speed centrifugation (400 × g for 7 min at 4°C) and resuspended in 1× PBS and incubated with rat anti-CD16/32 (1:200) for 10 min at 4°C to block Fc receptors. After blocking, cells were stained with an Ab master mix of primary Abs necessary to identify lung immune cell populations (Cd45, Cd11b, CD11c, Ly6G, Ly6C, Cd4, and B220, all at 1:100; see Table I). After 30 min, cells were washed with 1× PBS, centrifuged (400 × g for 7 min at 4°C), and finally resuspended in 600 µl of cytometry blocking buffer (0.1 M PBS containing 1% BSA and 0.01% sodium azide). Samples were filtered with a 40–μm filter when there was visible debris before analysis. Fluorosphere counting beads (50 μl per sample; BD Biosciences liquid counting beads, NC0613836) were added to allow for the quantification of absolute cell numbers. Appropriate unstained and isotype controls were used to determine background autofluorescence along with single-stained beads as compensation controls. Compensation was applied to remove spectral overlap, and isotype controls were used to define the gating strategy (see Supplemental Fig. 1). Cells were analyzed using a BD LSRFortessa flow cytometer, FACSDiva, and FlowJo software. Forward scatter (FSC) area versus side scatter area was used to exclude debris, and doublets were excluded using FSC height versus FSC width plots. Dead cells were identified using the Zombie Green viability kit as per the manufacturer’s instructions (BioLegend, catalog no. 423112) and excluded from final analyses. The gating strategy used to identify immune cell populations in the lung is provided in Supplemental Fig. 1.

Table I.

Flow cytometry Abs

AgHost, Amount per SampleRRIDVendor, Catalog No.
mFC Block (Cd16/32) Rat, 0.5 μl AB_394657 BD Biosciences, 553142 
mCD45-allophycocyanin Rat, 1 μl AB_398672 BD Biosciences, 559864 
mCD11b-PeCy7 Rat, 1 μl AB_394491 BD Biosciences, 552850 
mCD11c-BV711 Hamster, 1 μl AB_2734778 BD Biosciences 563048 
mLy6G-allophycocyanin Cy7 Rat, 1 μl AB_1727561 BD Biosciences, 560600 
mLy6C-BV421 Rat, 1 μl AB_2737748 BD Biosciences, 562727 
mB220-PE Rat, 1 μl AB_394619 BD Biosciences 553089 
mCD4-PerCp Rat, 1 μl AB_394587 BD Biosciences, 553052 
AgHost, Amount per SampleRRIDVendor, Catalog No.
mFC Block (Cd16/32) Rat, 0.5 μl AB_394657 BD Biosciences, 553142 
mCD45-allophycocyanin Rat, 1 μl AB_398672 BD Biosciences, 559864 
mCD11b-PeCy7 Rat, 1 μl AB_394491 BD Biosciences, 552850 
mCD11c-BV711 Hamster, 1 μl AB_2734778 BD Biosciences 563048 
mLy6G-allophycocyanin Cy7 Rat, 1 μl AB_1727561 BD Biosciences, 560600 
mLy6C-BV421 Rat, 1 μl AB_2737748 BD Biosciences, 562727 
mB220-PE Rat, 1 μl AB_394619 BD Biosciences 553089 
mCD4-PerCp Rat, 1 μl AB_394587 BD Biosciences, 553052 

m, mouse.

Lung spontaneous bacterial growth was assessed using methods outlined in previous studies (10, 16). Briefly, collected lung tissues were homogenized into a single-cell suspension in sterile 1× PBS by mashing lungs with the back of a 10-ml syringe through a 100-μm sterile filter. All tissue was collected, and homogenate was created under sterile conditions to prevent bacterial contamination from exterior sources. Lung homogenates (500 μl of the suspended solution) were then plated onto blood agar plates (Fisher Scientific, catalog no. R01201) under flame sterilization and incubated under aerobic conditions at 37°C for 24 h for blood agar plates and 48 h for CHROMagar orientation plates (Fisher Scientific, catalog no. B254102). Blood agar plates were used for total CFU counts, while CHROMagar plates were used to identify different types of bacteria. Bacteria were also incubated under anaerobic conditions in pilot experiments, but no difference was seen between bacterial growth in anaerobic and aerobic conditions (data not shown). After incubation, plates were imaged and bacterial colonies were manually counted. Serial dilutions of each sample were prepared and plated to determine optimal dilution for bacterial counts (ideally a dilution should result in 30–300 CFU per plate to achieve accurate counts). The ideal dilution was found to be 1:10 or 1:200 depending on the time point. Total CFU/ml were calculated based on the determined dilution factor.

After T1Tx SCI or T1 sham surgery, mice were injected s.c. with 5 mg/kg AMD3100 (plerixafor, Sigma-Aldrich) in 0.9% sterile saline, a dose previously shown to boost total numbers of mature immune cells in the blood in SCI mice (12). Control mice were injected with 0.9% sterile saline. The first dose was given 1 hpi, and then once daily until 3 dpi. Daily postsurgery saline s.c. injections and AMD3100 injections were given 2 h apart to prevent dilution of the drug. Mice were terminally anesthetized with ketamine and xylazine 1 h after the final dose of AMD3100. Lungs were processed for flow cytometry and spontaneous bacterial growth assays as described above. Blood was collected (∼100–200 µl) via cardiac puncture into the right ventricle of the heart and placed into EDTA-coated tubes until analysis on an automated analyzer system capable of analyzing whole blood with five-part WBC differential, platelets, and RBCs. Analysis was performed by the Comparative Pathology and Mouse Phenotyping Shared Resource Center at The Ohio State University.

The SCI acquired pneumonia model described in Brommer et al. (8) was used for the studies described below. Mice received at T3Tx injury, and treatment with AMD3100 began 1 h after injury and AMD3100 was then given daily until the end of the experiment, as described above. Pneumonia was induced at 3 dpi and tissue was collected 24 h after induction at 4 dpi.

Briefly, to induce pneumonia after SCI, Streptococcus pneumoniae was first grown in media to an OD620 of 0.4–0.6 and then diluted to deliver 500 CFU/mouse in sterile PBS. Mice were then anesthetized and suspended at a 60° angle by the two upper front teeth using a wire attached to a support. A 22G peripheral venous catheter in combination with a 0.5-mm optical fiber attached cold light source was used for illumination for the intubation procedure. The tongue was then pulled out with a small spatula and the mouse was intubated, allowing 30–50 μl of the prepared bacterial suspension (500 CFU) to be administered into the lung. Animals were placed in a heated cage for the duration of the experiment to maintain body temperature. To minimize spontaneous postinjury colonization of the lung by bacteria and ensure homogeneous infection in the lung by S. pneumoniae, mice received antibiotics (gentamicin, 5 mg/kg daily, i.p.) starting at the time of surgery, then daily for 2 dpi. Antibiotics then were terminated at 2 dpi to prevent them from interfering with pneumonia induction at 3 dpi. Blood was processed as described above. Lungs were homogenized 24 h postinoculation at 4 dpi and the number of CFU on sheep blood agar plates were counted (only colonies of S. pneumoniae were counted, not contaminating bacteria, which were minimal if any).

RNA was isolated from whole mouse lungs that were excised and then immediately homogenized in TRIzol (Thermo Fisher Scientific, catalog no.15596026) according to the manufacturer’s instructions. All RNA samples had an A260/A230 ratio >1.7, an A260/A280 ratio between 1.8 and 2.0, and a RNA integrity number >7. RNA from four mice per group was pooled (400 ng/group) and then cDNA was prepared using the RT2 first-strand kit as per the manufacturer’s directions (Qiagen, catalog no. 330404). The prepared cDNA mix (102 µl, as per the manufacturer’s instructions) was then loaded onto a 384-well plate preloaded with primers for mouse innate and adaptive immune response genes (RT2 Profiler PCR array mouse innate and adaptive immune responses, Qiagen, catalog no. 330231). PCR reactions were performed with RT2 SYBR Green ROX quantitative PCR (qPCR) mastermix (Qiagen, catalog no. 220521) as per the manufacturer’s instructions for the RT2 Profiler array kit and run on a 7900HT Applied Biosystems cycler (Applied Biosystems). Melting point analyses were performed for each reaction to confirm single amplified products. Initial analyses were performed using Qiagen’s GeneGlobe Data Analysis Center software to determine differentially expressed genes (DEGs). Data were normalized to the HKG panel included with the kit. Further analyses are as described below.

A heat map of gene expression patterns after hierarchical clustering analysis of lung homogenates at 12 hpi and 3 dpi was performed using Qiagen’s GeneGlobe Data Analysis software, and DEGs were determined using Qiagen’s GeneGlobe Data Analysis Center (genes with a fold regulation ≥2 were identified using a modified ΔΔCt method [17]). Specifically, fold regulation is the method used by Qiagen’s GeneGlobe Data Analysis Center to display data and provides the same information as fold change, but in a more interpretable format. Essentially, if the fold change is ≥1, the fold regulation is equivalent. If the fold change value is <1, the fold regulation is the negative inverse of the fold change (i.e., fold regulation = –1/fold change). Fold regulation values >1 indicate upregulated expression, and fold regulation less than −1 indicate downregulated expression (see RT2 Profiler PCR array analysis v3.5 handbook from Qiagen for additional details). The p values were not used for this initial analysis, as samples were pooled. Upregulated and downregulated gene lists from each identified cluster of the hierarchical cluster analysis map were then inputted into g:Profiler to identify the top five Gene Ontology (GO) biological processes associated with each cluster. To avoid selection bias, the list of 84 genes analyzed with the PCR array was used as a custom reference background instead of the whole murine genome. The top five terms were determined by selecting the immune- and lung-related terms associated with the highest gene ratio, which is the ratio of DEGs in the term to the total number of genes associated with that term. Additionally, the size of the functional category for terms was restricted to between 5 (minimum) and 350 (maximum), as outlined in Reimand et al. (18). DEGs at 12 hpi and 3 dpi were entered into g:Profiler (an online tool to perform enrichment analysis; https://biit.cs.ut.ee/gprofiler/orth), and a similar analysis to identify GO biological pathways increased and decreased at each time point compared with sham controls was conducted. Again, the list of 84 genes analyzed with the PCR array was used as a reference. Finally, a Venn diagram of the two time points showing the number and IDs of DEGs was then created to determine critical changes and overlap in gene expression at each time point.

A protein–protein interaction network was created by analyzing all DEGs at 12 h and 3 dpi with the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) network database (https://string-db.org/, v11) (19). A minimum required interaction score of 0.7 was used. The molecular interaction network was then visualized using Cytoscape (https://cytoscape.org, v3.8.2), an open-source bioinformatics platform. To asses hub genes, the Cytoscape plugin cytoHubba was used (20). cytoHubba provides 11 topological analysis methods to assess the interconnectedness of nodes in a network. The top 10 most interconnected genes in the network were assessed using each of the 11 methods, and then the top 10 hub genes overall were identified by determining which genes were consistently represented most of the 11 methods (see Supplemental Table I for results from each of the 11 assays).

To verify the hub gene and cytoHubba results, significant clusters of genes in the protein–protein interaction networks were identified using the Cytoscape plugin Molecular Complex Detection (MCODE) (21), and it was then determined whether the hub genes were identified in these subclusters. MCODE detects densely connected regions based on network topology and the following criteria: degree cutoff = 2, node cutoff = 0.2, K-core = −2, maximum depth = 100, and haircut = on to remove singly connected nodes from clusters. The top 10 significant GO enrichment and Kyoto Encyclopedia of Genes and Genomes pathway terms for the identified top cluster (the one that contained the most hub genes) were then identified by using the enrichGO function in the clusterProfiler package in R (22). The whole murine genome was used as a reference background.

Results from the PCR array were confirmed by performing reverse transcriptase quantitative PCRs (RT-qPCR) for the 10 identified hub genes with the individual samples instead of pooling samples. Prior to cDNA creation, genomic DNA was eliminated using 1 μg/μl DNase I (Invitrogen, catalog no. 18068-015). One microgram of DNase-treated RNA was then primed with random hexamers (1 μM; Applied Biosystems) and then reverse transcribed into cDNA using SuperScript IV reverse transcription (Fisher Scientific, catalog no. 18090050) in a 20- or 50-μl reaction, as needed. Undiluted cDNA (1 μl) was loaded onto a 396-well plate and combined with primers (see Table II). SYBR Green master mix (Fisher Scientific, catalog no. 4385612) was used to detect amplified cDNA. Melting point curves were used to assess the quality of the reactions, and triplicated samples were run. Gene expression was normalized to an 18S control (from cDNA samples diluted 1:1000) for each sample using the ΔΔCt method (17). A standard dilution curve and no-template control were run for each primer as well. SCI data were then expressed as a fold regulation, as calculated in the PCR array data.

Table II.

RT-qPCR primers for identified hub genes

Gene ProductForward Primer (5′ to 3′)Reverse Primer (5′ to 3′)
CCL5 GTGTGCCAACCCAGAGAAGAA GGGAAGCGTATACAGGGTCAG 
CXCL10 TTGAAATCATCCCTGCGAGC CGTGGCTTCACTCCAGTTAAGG 
Foxp3 GTCTGGAATGGGTGTCCAGG ATGATCTGCTTGGCAGTGCT 
Ifng CACGGCACAGTCATTGAAAG GCTGATGGCCTGATTGTCTT 
IL-4 ATGGATGTGCCAAACGTCCT TGCAGCTCCATGAGAACACT 
IL-6 GCCTTCTTGGGACTGATGCT AGTCTCCTCTCCGGACTTGTG 
IL-10 CAGCCGGGAAGACAATAACTG CCGCAGCTCTAGGAGCATGT 
MyD88 AGGCGATGAAGAAGGACTTTCC TCAGTCTCATCTTCCCCTCTGC 
Stat1 CAGCCCAAGGATGTCACAGT CGAGACATCATAGGCAGCGT 
TNF GTGATCGGTCCCCAAAGG GGTCTGGGCCATAGAACTGATG 
18S TTCGGAACTGAGGCCATGAT TTTCGCTCTGGTCCGTCTTG 
Gene ProductForward Primer (5′ to 3′)Reverse Primer (5′ to 3′)
CCL5 GTGTGCCAACCCAGAGAAGAA GGGAAGCGTATACAGGGTCAG 
CXCL10 TTGAAATCATCCCTGCGAGC CGTGGCTTCACTCCAGTTAAGG 
Foxp3 GTCTGGAATGGGTGTCCAGG ATGATCTGCTTGGCAGTGCT 
Ifng CACGGCACAGTCATTGAAAG GCTGATGGCCTGATTGTCTT 
IL-4 ATGGATGTGCCAAACGTCCT TGCAGCTCCATGAGAACACT 
IL-6 GCCTTCTTGGGACTGATGCT AGTCTCCTCTCCGGACTTGTG 
IL-10 CAGCCGGGAAGACAATAACTG CCGCAGCTCTAGGAGCATGT 
MyD88 AGGCGATGAAGAAGGACTTTCC TCAGTCTCATCTTCCCCTCTGC 
Stat1 CAGCCCAAGGATGTCACAGT CGAGACATCATAGGCAGCGT 
TNF GTGATCGGTCCCCAAAGG GGTCTGGGCCATAGAACTGATG 
18S TTCGGAACTGAGGCCATGAT TTTCGCTCTGGTCCGTCTTG 

Data analysis and data visualization were performed using Prism version 9.1.1 software by GraphPad Software. All data are presented as mean ± SD. Appropriate sample sizes were determined with a priori and post hoc power analysis, with the level of power set at 0.8 and α = 0.05, to detect a 1.8-fold change using G*Power software. Statistical outliers were detected in a subset of data (see Figs. 1 and 5) using the ROUT method in GraphPad Prism (23). These data were removed from analysis. This affected a subset of data in Figs. 1 and 5. The total numbers of cells were analyzed using Mann–Whitney U tests. Data from three separate replicate experiments (n = 4 mice per group per experiment) were combined for the final analysis for 12 hpi, 3 dpi, and 28 dpi time points of T3Tx experiments. Similar analyses were performed for T1Tx flow cytometry experiments at 3 dpi (n = 3–7 per group per experiment). Spontaneous bacterial growth data were compared using a Mann–Whitney U test. For bacterial analysis, data from at least two separate replicate experiments at each time point were combined (n = 4–6 mice per group per experiment) for both T1Tx and T3Tx experiments. For T1Tx experiments with bacterial growth on CHROMagar plates, all bacteria were counted for each identified group and a percentage of the total bacteria number was calculated for each type. For immune cell counts in the blood after AMD3100 treatment for spontaneous bacterial growth and induced pneumonia groups, cells were represented as 1000 cells/μl, and Kruskal–Wallis analyses with a Dunn’s multiple comparisons test were performed as appropriate. All analyses for gene array data are described in the sections above. Group differences were considered statistically significant at p < 0.05. All data are presented as mean ± SD.

To determine whether the lung exhibits signs of immune suppression after SCI, pulmonary leukocyte subsets were quantified in mouse lung homogenates at 12 hpi and 3 dpi. These times correspond with the onset of spontaneous pneumonia and systemic immune suppression and bone marrow failure after SCI (810, 12, 16, 24).

At 12 hpi, lung neutrophils increased significantly (194,712 ± 123,420 [sham] versus 631,289 ± 326,538 [SCI], p = 0.0002; (Fig. 1B). Other leukocyte subsets, including CD4+ T lymphocytes (Fig. 1F), B220+ B lymphocytes (Fig. 1J) and alveolar macrophages (Fig. 1N), were unchanged relative to sham-injured controls. However, by 3 dpi, neutrophil numbers returned to sham levels (Fig. 1C), but CD4+ T lymphocytes (400,746 ± 275,892 [sham] versus 127,851 ± 71,595 [SCI], p = 0.0003; (Fig. 1G) and B220+ B lymphocytes (862,419 ± 523,907 [sham] versus 274,441 ± 153,648 [SCI], p = 0.0006; (Fig. 1K) decreased >60% relative to sham-injured controls (Fig. 1E, 1G, 1I, 1K). Alveolar macrophage numbers were unchanged at 3 dpi (Fig. 1O).

FIGURE 1.

SCI causes acute pulmonary neutrophilia with delayed and persistent lymphopenia and a reduction in alveolar macrophages. (A) Representative flow cytometry plots for Ly6G+ neutrophils from the 12 h time point for sham and SCI mice. (B) At 12 hpi, neutrophil numbers were increased in SCI mice, but this increase was gone by 3 (C) and 28 dpi (D). (E) Representative flow cytometry plots for CD4+ T lymphocytes from the 3 d time point for sham and SCI mice. (F) There was no change in CD4+ T lymphocyte number at 12 h, but there was decrease in total cell number at both 3 (G) and 28 dpi (H). (I) Representative flow cytometry plots for B220+ B lymphocytes from the 3 d time point for sham and SCI mice. A similar pattern of results was seen over time with total B220+ B lymphocyte number with no difference in total cells at 12 hpi (J), followed by a decrease in total cell number at 3 (K) and 28 dpi (L) in SCI mice. (M) Representative flow cytometry plots for alveolar macrophages from the 28 dpi time point for sham and SCI mice. (N and O) No difference in alveolar macrophage total cell number was seen at either 12 hpi (N) or 3 dpi (O) between groups. (P) A significant reduction in alveolar macrophage number was seen at 28 dpi in SCI mice. For 12 h data experiments, n = 8–12 mice per group combined from three replicate experiments, for 3 d experiments, n = 12 mice per group combined from three replicate experiments, and for 28 d experiments, n = 12 mice per group combined from three replicate experiments. Statistical outliers detected using the ROUT method were removed from analysis (n = 1 sham mouse in B; n = 1 each sham and SCI in D; n = 1 sham in G, and n = 1 sham in K) prior to Mann–Whitney U analysis, *p < 0.05. Bars represent mean ± SD.

FIGURE 1.

SCI causes acute pulmonary neutrophilia with delayed and persistent lymphopenia and a reduction in alveolar macrophages. (A) Representative flow cytometry plots for Ly6G+ neutrophils from the 12 h time point for sham and SCI mice. (B) At 12 hpi, neutrophil numbers were increased in SCI mice, but this increase was gone by 3 (C) and 28 dpi (D). (E) Representative flow cytometry plots for CD4+ T lymphocytes from the 3 d time point for sham and SCI mice. (F) There was no change in CD4+ T lymphocyte number at 12 h, but there was decrease in total cell number at both 3 (G) and 28 dpi (H). (I) Representative flow cytometry plots for B220+ B lymphocytes from the 3 d time point for sham and SCI mice. A similar pattern of results was seen over time with total B220+ B lymphocyte number with no difference in total cells at 12 hpi (J), followed by a decrease in total cell number at 3 (K) and 28 dpi (L) in SCI mice. (M) Representative flow cytometry plots for alveolar macrophages from the 28 dpi time point for sham and SCI mice. (N and O) No difference in alveolar macrophage total cell number was seen at either 12 hpi (N) or 3 dpi (O) between groups. (P) A significant reduction in alveolar macrophage number was seen at 28 dpi in SCI mice. For 12 h data experiments, n = 8–12 mice per group combined from three replicate experiments, for 3 d experiments, n = 12 mice per group combined from three replicate experiments, and for 28 d experiments, n = 12 mice per group combined from three replicate experiments. Statistical outliers detected using the ROUT method were removed from analysis (n = 1 sham mouse in B; n = 1 each sham and SCI in D; n = 1 sham in G, and n = 1 sham in K) prior to Mann–Whitney U analysis, *p < 0.05. Bars represent mean ± SD.

Close modal

To determine whether acute SCI-induced changes in lung immunity persist into chronic stages of recovery when systemic immune function remains impaired and the risk of infection is high (7, 9, 24), the above analyses were repeated at 28 dpi using lungs isolated from a separate cohort of mice. Similar to the 3 dpi cohort, numbers of T (369,130 ± 249,660 [sham] versus 167,709 ± 124,795 [SCI], p = 0.0100; (Fig. 1H) and B lymphocytes (837,150 ± 424,365 [sham] versus 412,254 ± 290,168 [SCI], p = 0.0045; (Fig. 1L) were reduced ∼50% relative to sham-injured, age-matched controls. The number of alveolar macrophages also decreased ∼30% (272,813 ± 62,796 [sham] versus 179,827 ± 54,324 [SCI], p = 0.0005; (Fig. 1P). Neutrophils were unchanged at 28 dpi (Fig. 1D). Taken together, these data indicate that SCI disrupts pulmonary immune homeostasis, culminating with a chronic imbalance in the normal composition of lung immune cells.

Lung “bacterial load” (number of CFU) is the diagnostic standard for evaluating SCI-acquired pneumonia and serves as a biomarker for the severity of immune suppression (25). To determine whether SCI-induced changes in pulmonary immune cells (Fig. 1) are associated with corresponding changes in bacterial colonization of the lung, lung homogenates obtained from sham and SCI mice at 12 hpi, 3 dpi, and 28 dpi were plated on blood agar plates (Fig. 2). Counting the number of CFU revealed minimal or no bacterial colonization in sham-injured mice at any time point (Fig. 2A, 2B). Conversely, in SCI mice, bacterial CFU increased significantly at 12 hpi, but by 3 dpi few bacteria were present (p = 0.0020, 12 hpi [4146 ± 6284] versus 3 dpi [177 ± 206], p = 0.0449, Kruskal–Wallis test with Dunn’s multiple comparisons test; (Fig. 2C, 2D). Interestingly, lung bacteria increased again at 28 dpi, reaching CFU levels equivalent to those at 12 hpi (3 dpi [177.1 ± 206.1] versus 28 dpi [3579 ± 2961], p = 0.0014, Kruskal–Wallis test with Dunn’s multiple comparisons test; (Fig. 2C, 2D).

FIGURE 2.

Spontaneous bacterial infection develops in the lung after SCI. (A) Lung homogenates plated on blood agar plates from 12 hpi (1:10 dilution), 3 dpi (1:10 dilution), and 28 dpi (1:200 dilution) from sham injured mice and cultured for 24 h. White circles highlight areas of bacterial growth. (B) No change in spontaneous bacterial growth from sham mice lungs at any time point. (C) Representative blood agar plates from 12 hpi, 3 dpi, and 28 dpi T3Tx SCI mice. White circles highlight areas of bacterial growth. (D) There was a significant increase in spontaneous bacterial growth at both 12 hpi and 3 dpi compared with 3 dpi in SCI mice lungs. n = 7 for 12 hpi and 3dpi T3Tx and 3 dpi sham mice combined from two replicate experiments, n = 12–14 for 28 dpi sham mice, n = 12–14 for 28 dpi SCI mice combined from three replicate experiments. Kruskal–Wallis test U analysis with Dunn’s multiple comparisons test, *p < 0.05. Bars represent mean ± SD.

FIGURE 2.

Spontaneous bacterial infection develops in the lung after SCI. (A) Lung homogenates plated on blood agar plates from 12 hpi (1:10 dilution), 3 dpi (1:10 dilution), and 28 dpi (1:200 dilution) from sham injured mice and cultured for 24 h. White circles highlight areas of bacterial growth. (B) No change in spontaneous bacterial growth from sham mice lungs at any time point. (C) Representative blood agar plates from 12 hpi, 3 dpi, and 28 dpi T3Tx SCI mice. White circles highlight areas of bacterial growth. (D) There was a significant increase in spontaneous bacterial growth at both 12 hpi and 3 dpi compared with 3 dpi in SCI mice lungs. n = 7 for 12 hpi and 3dpi T3Tx and 3 dpi sham mice combined from two replicate experiments, n = 12–14 for 28 dpi sham mice, n = 12–14 for 28 dpi SCI mice combined from three replicate experiments. Kruskal–Wallis test U analysis with Dunn’s multiple comparisons test, *p < 0.05. Bars represent mean ± SD.

Close modal

The absence of pulmonary infection at 3 dpi is consistent with independently published data (16, 26, 27) and could indicate resolution of acute infection due to the increase in pulmonary neutrophils at 12 hpi (see (Figs. 1B and 2D). However, these data conflict with other independent studies showing that spontaneous pulmonary infection persists at 3 dpi (10). It is possible that differences in pulmonary infection vary as a function of spinal injury level (8).

The complete T3 SCI model used in the current report causes profound systemic and pulmonary immune suppression (see data above and also Refs. 7, 8, 12). An identical lesion located two spinal segments away at the T1 spinal level causes systemic immune suppression (see Ref. 10) and abolishes sympathetic innervation of the lung and intercostal muscles via the pulmonary plexus. Thus, the combination of both impaired immune function and lung compliance could cause pneumonia to persist for longer postinjury periods after T1 SCI. To test this hypothesis, the analyses in (Figs. 1 and 2 were repeated. This time, when comparing T1 SCI mice with sham-injured mice, significant colonization of lung with potentially pathogenic bacteria was detected at 3 dpi (30.62 ± 52.58 [sham] versus 3270 ± 9757 [SCI], p = 0.0003; (Fig. 3A–C). Notably, T1 SCI caused an increase in the relative proportion of Staphylococcus saprophyticus (a cause of urinary tract infections) and Pseudomonas aeruginosa (known to cause pneumonia and bacteremia) in SCI lung (28, 29). Coincident with lung infection was a reduction in total numbers of pulmonary innate and adaptive immune cells. This injury-dependent decrease in pulmonary leukocytes was more profound after T1 SCI than after T3 SCI (compare (Figs. 1 and 3). Specifically, 3 d after T1 SCI, all lung leukocyte subsets were reduced relative to sham controls, especially lung neutrophils, T lymphocytes (45,128 ± 14,535 [sham] versus 8,417 ± 1,053 [SCI], p = 0.0006; (Fig. 3E), B lymphocytes (167,373 ± 17,060 [sham] versus 68,405 ± 28,746 [SCI], p = 0.0006; (Fig. 3F), and alveolar macrophages (51,245 ± 8,649 [sham] versus 37,937 ± 6,528 [SCI], p = 0.0262; (Fig. 3G).

FIGURE 3.

Spontaneous lung bacterial infections and leukopenia persist after T1 SCI. (A) Lung homogenates (1:200 dilution) plated on blood agar plates from 3 dpi sham-injured or T1 transection SCI mice and cultured for 24 h. Spontaneous bacterial growth from SCI mice lungs increased at 3 dpi. (B) Representative blood agar plates from sham-injured or T1 SCI mice. (C) Example CHROMagar plates from 3 dpi SCI mice chosen to demonstrate the different bacteria identified in sham mice (top row) and T1 SCI mice (bottom row). (D) Bar graphs show the percentage of each bacteria type in cultured lung homogenate for sham and T1 SCI mice at 3 dpi. Flow cytometry analysis on T1 sham and T1 SCI lungs at 3 dpi show the development of immune suppression in the lung. (EH) While no change in neutrophils (E) was seen, a significant loss of T cells (F), B cells (G), and alveolar macrophages (H) occurred. n = 18 for T1Tx, n = 13 for T1 sham mice for bacterial assays from three replicate experiments, n = 7 per group for flow cytometry analysis from one experiment. Mann–Whitney U analysis, *p < 0.05. Bars represent mean ± SD.

FIGURE 3.

Spontaneous lung bacterial infections and leukopenia persist after T1 SCI. (A) Lung homogenates (1:200 dilution) plated on blood agar plates from 3 dpi sham-injured or T1 transection SCI mice and cultured for 24 h. Spontaneous bacterial growth from SCI mice lungs increased at 3 dpi. (B) Representative blood agar plates from sham-injured or T1 SCI mice. (C) Example CHROMagar plates from 3 dpi SCI mice chosen to demonstrate the different bacteria identified in sham mice (top row) and T1 SCI mice (bottom row). (D) Bar graphs show the percentage of each bacteria type in cultured lung homogenate for sham and T1 SCI mice at 3 dpi. Flow cytometry analysis on T1 sham and T1 SCI lungs at 3 dpi show the development of immune suppression in the lung. (EH) While no change in neutrophils (E) was seen, a significant loss of T cells (F), B cells (G), and alveolar macrophages (H) occurred. n = 18 for T1Tx, n = 13 for T1 sham mice for bacterial assays from three replicate experiments, n = 7 per group for flow cytometry analysis from one experiment. Mann–Whitney U analysis, *p < 0.05. Bars represent mean ± SD.

Close modal

Data above indicate that susceptibility to lung infection after SCI varies as a function of spinal injury level and that more severe immune suppression occurs after T1 SCI. To test whether postinjury lung infection can be blocked by boosting immune cell recruitment to the lung, T1 SCI mice were injected (s.c.) with plerixafor (AMD3100), a CXCR4 antagonist. Specifically, AMD3100 was injected at 1 hpi after T1 SCI, then again daily for 3 d. Previously we found that AMD3100 augments the release and trafficking of immune cells from bone marrow into secondary lymphoid tissues after T3 SCI (12). Similar results were found in this study in mice after T1 SCI. A complete blood count differential count of circulating blood obtained from all mice at 3 dpi revealed that AMD3100 increased the total number of circulating leukocytes (Fig. 4A), lymphocytes (Fig. 4B), monocytes (Fig. 4C), neutrophils (sham [0.6200 ± 0.1301] versus AMD3100 [1.676 ± 0.9560], p = 0.0018, Kruskal–Wallis test with Dunn’s multiple comparisons test; (Fig. 4D) and eosinophils (sham [0.01000 ± 0.007559] versus AMD3100 [0.05500 ± 0.04403], p = 0.0057, Kruskal–Wallis test with Dunn’s multiple comparisons test; (Fig. 4E).

FIGURE 4.

Treatment with AMD3100 increases circulating leukocytes and boosts pulmonary immune responses after SCI but without affecting spontaneous bacterial infections in the lung. (AE) Quantification of 3 dpi lung flow cytometry data of mice with T1Tx SCI and treated daily with either AMD3100 (AMD) or saline (Vehicle). Overall treatment with AMD3100 boosted levels total levels of WBCs (A), lymphocytes (B), and monocytes (C), and levels of neutrophils (D) and eosinophils (E) were significantly elevated with treatment. (FJ) AMD3100 also boosted total immune cells (F), B lymphocytes (G), and T lymphocytes (H), but neutrophils and alveolar macrophages were unchanged (I and J). (K and L) Treatment with AMD3100 did not alter total CFU in the lung after injury, as seen in the quantification (K) and in the representative blood agar plates (1:200 dilution) (L). (M) AMD3100 treatment also did not alter the type of bacteria seen on CHROMagar plates (n = 10 vehicle and n = 13 AMD3100 mice for flow analysis pooled from three replicate experiments, n = 23 vehicle and n = 27 AMD3100 mice for bacterial analysis combined from four replicate experiments, and n = 8 sham, n = 9 vehicle mice, and n = 10 AMD3100 mice for blood count analysis from one experiment. Kruskal-Wallis test with Dunn’s multiple comparisons test (A–E) or Mann–Whitney U analysis (F–K), *p < 0.05. Bars represent mean ± SD.

FIGURE 4.

Treatment with AMD3100 increases circulating leukocytes and boosts pulmonary immune responses after SCI but without affecting spontaneous bacterial infections in the lung. (AE) Quantification of 3 dpi lung flow cytometry data of mice with T1Tx SCI and treated daily with either AMD3100 (AMD) or saline (Vehicle). Overall treatment with AMD3100 boosted levels total levels of WBCs (A), lymphocytes (B), and monocytes (C), and levels of neutrophils (D) and eosinophils (E) were significantly elevated with treatment. (FJ) AMD3100 also boosted total immune cells (F), B lymphocytes (G), and T lymphocytes (H), but neutrophils and alveolar macrophages were unchanged (I and J). (K and L) Treatment with AMD3100 did not alter total CFU in the lung after injury, as seen in the quantification (K) and in the representative blood agar plates (1:200 dilution) (L). (M) AMD3100 treatment also did not alter the type of bacteria seen on CHROMagar plates (n = 10 vehicle and n = 13 AMD3100 mice for flow analysis pooled from three replicate experiments, n = 23 vehicle and n = 27 AMD3100 mice for bacterial analysis combined from four replicate experiments, and n = 8 sham, n = 9 vehicle mice, and n = 10 AMD3100 mice for blood count analysis from one experiment. Kruskal-Wallis test with Dunn’s multiple comparisons test (A–E) or Mann–Whitney U analysis (F–K), *p < 0.05. Bars represent mean ± SD.

Close modal

We next tested whether the enhanced mobilization of immune cells into the blood would in turn boost pulmonary immune surveillance (as measured by flow cytometry) and reduce bacterial burden. In a separate cohort of T1 SCI mice, AMD3100 boosted total numbers of lung leukocytes (Fig. 4F). This effect was due mostly to increases in pulmonary B lymphocytes (Fig. 4G) and T lymphocytes (Fig. 4H). AMD3100 did not consistently increase lung neutrophils or alveolar macrophages (Fig. 4I, 4J). Even though AMD3100 boosted recruitment of immune cells to the lung, the drug failed to reduce the density or alter the composition of lung bacteria (Fig. 4K–M). Specifically, we found that 16 out of 23 mice (69.6%) in the vehicle-treated group and 25 out of 27 of mice (92.3%) in the AMD treatment group had a high bacterial load (>31 CFU, a cutoff based on mean T1 sham bacterial load in (Fig. 3A) (Fig. 5G). Longer treatment durations with AMD3100 until 7 dpi also failed to reduced lung bacterial burden (Supplemental Fig. 2A).

FIGURE 5.

AMD3100 treatment improves pulmonary immune surveillance and reduces bacterial colonization in SCI mice with induced pneumonia. (AD) Quantification of 3 dpi whole-blood automated cell count data of mice with T3 transection SCI and induced pneumonia who were treated daily with either AMD3100 (AMD) or saline (Vehicle). Overall treatment with AMD3100 significantly increased total WBC counts (A), neutrophils (B), lymphocytes (C), and monocyte numbers (D). (E and F) Blood eosinophil (E) and basophil (F) counts were not changed with treatment. (G) Quantification of 4 dpi (and 24 h after pneumonia inoculation) injury-induced pneumonia bacterial growth (pneumonia-specific counts) in the lungs of mice with T3 SCI and treated with either AMD or vehicle. Although there was no significant difference between treatment groups, there was a strong trend toward a reduction in total CFU counts in AMD3100-treated mice, as four out of nine mice (44.4%) in the vehicle-treated group had a high bacterial compared with one out of eight mice (12.5%) in the AMD treatment group. n = 11 vehicle and n = 11 AMD3100 mice from one experiment. Statistical outliers detected using the ROUT method were removed for analysis (n = 1 mouse in the AMD group in E; n = 1 mouse each in AMD and vehicle in F, and n = 2 mice in the AMD group in G). Mann–Whitney U analysis (A–G), * p > 0.05. Bars represent mean ± SD.

FIGURE 5.

AMD3100 treatment improves pulmonary immune surveillance and reduces bacterial colonization in SCI mice with induced pneumonia. (AD) Quantification of 3 dpi whole-blood automated cell count data of mice with T3 transection SCI and induced pneumonia who were treated daily with either AMD3100 (AMD) or saline (Vehicle). Overall treatment with AMD3100 significantly increased total WBC counts (A), neutrophils (B), lymphocytes (C), and monocyte numbers (D). (E and F) Blood eosinophil (E) and basophil (F) counts were not changed with treatment. (G) Quantification of 4 dpi (and 24 h after pneumonia inoculation) injury-induced pneumonia bacterial growth (pneumonia-specific counts) in the lungs of mice with T3 SCI and treated with either AMD or vehicle. Although there was no significant difference between treatment groups, there was a strong trend toward a reduction in total CFU counts in AMD3100-treated mice, as four out of nine mice (44.4%) in the vehicle-treated group had a high bacterial compared with one out of eight mice (12.5%) in the AMD treatment group. n = 11 vehicle and n = 11 AMD3100 mice from one experiment. Statistical outliers detected using the ROUT method were removed for analysis (n = 1 mouse in the AMD group in E; n = 1 mouse each in AMD and vehicle in F, and n = 2 mice in the AMD group in G). Mann–Whitney U analysis (A–G), * p > 0.05. Bars represent mean ± SD.

Close modal

It is possible that a therapeutic effect of AMD3100 is not evident because the spontaneous infection caused by SCI in mice does not produce enough lung CFU to detect an effect (i.e., a “floor effect”). In preclinical models of pneumonia in mice with intact nervous systems (i.e., no SCI), direct inoculation of animals with 105–108 CFU is used to model disease and test therapies (3032). In contrast, our spontaneous colonization of lung by environmental or commensal bacteria, regardless of spinal injury level or time postinjury, rarely exceeded 104 CFU (see (Figs. 24). Thus, we retested the effects of AMD3100 in SCI mice after intentionally infecting them with S. pneumoniae via an intratracheal infusion. Using this well-characterized model of lung infection (8), ≥105 CFU of bacteria colonize the lung of SCI mice within 24 h postinoculation. To avoid the high risk of respiratory complications associated with T1 SCI, which would be exacerbated by intratracheal infusion of bacteria, mice in this experiment received a T3 SCI. Beginning at 1 hpi, mice received daily postinjury injections of vehicle or AMD3100, and then at 3 dpi mice were inoculated with S. pneumoniae. One day later, blood and lungs were collected to quantify circulating leukocytes and bacterial CFU, respectively. Again, circulating leukocytes were increased significantly by AMD3100, although cell-specific effects were slightly different from after T1 SCI without intentional induction of pneumonia (Figs. 4A–F and 5A–F). Specifically, numbers of circulating leukocytes increased (1.642 ± 0.5727 [sham] versus 2.309 ± 0.5141 [SCI], p = 0.0101; (Fig. 5A). Notably, neutrophils (0.3555 ± 0.1116 [sham] versus 0.5664 ± 0.2051 [SCI], p = 0.0088; (Fig. 5B), lymphocytes, (1.135 ± 0.5153 [sham] versus 1.506 ± 0.3520 [SCI], p = 0.0222; (Fig. 5C), and monocytes (0.1109 ± 0.05412 [sham] versus 0.1791 ± 0.05665 [SCI], p = 0.0091; (Fig. 5D) increased significantly, but eosinophils and basophils were not affected (Fig. 5E, 5F). Importantly, despite high variance across the relatively small experimental and control cohorts, we found that AMD3100 reduced bacterial burden; n = 1 out of 8 of mice (12.5%) in the AMD treatment group had a high bacterial load (>105 CFU, a cutoff based on bacterial load data in Prass et al. [25, 33] and Brommer et al. [8]) compared with n = 4 out of 9 mice (44.4%) in the vehicle-treated group (Fig. 5G).

Data in (Fig. 5 indicated that AMD3100 treatment partially reduced lung bacterial burden, but the effect of treatment varied, indicating that enhancing leukocyte recruitment to the lung alone may not be sufficient to prevent or reverse lung infections after SCI. This could indicate that mechanisms of lung resistance (i.e., innate and adaptive immune responses) and lung resilience (i.e., mechanisms that allow the lung to withstand or tolerate tissue-specific stress), both of which are required for optimal defense against infection (13), become dysfunctional after SCI.

To reveal mechanisms of lung resistance and/or resilience affected by SCI, an analysis of the lung transcriptome was performed on lung tissue isolated from sham-injured control and T3 SCI mice at 12 hpi and 3 dpi. Hierarchical clustering analysis of PCR array data revealed three major clusters of DEGs. Of these, clusters 2 and 3 could each be subdivided into two groups (2a/2b and 3a/3b). For each gene cluster, the top associated GO terms for biological process were assessed (Fig. 6B).

FIGURE 6.

SCI impairs the coordinated expression of genes controlling immune resistance to pathogens and tissue resilience in the lung. (A) Heat map of gene expression patterns after hierarchical clustering analysis of lung homogenates at 12 hpi and 3 dpi. Red and blue represent high and low gene expression, respectively. (B) Top Gene Ontology (GO) biological process terms associated with PCR array clusters, color coded to match the diagram in (A). Count indicates the number of DEGs represented in the associated GO term. n = 4 per group pooled for PCR array analysis from one experiment.

FIGURE 6.

SCI impairs the coordinated expression of genes controlling immune resistance to pathogens and tissue resilience in the lung. (A) Heat map of gene expression patterns after hierarchical clustering analysis of lung homogenates at 12 hpi and 3 dpi. Red and blue represent high and low gene expression, respectively. (B) Top Gene Ontology (GO) biological process terms associated with PCR array clusters, color coded to match the diagram in (A). Count indicates the number of DEGs represented in the associated GO term. n = 4 per group pooled for PCR array analysis from one experiment.

Close modal

Generally, cluster 2 genes were upregulated at 12 hpi in SCI mice and were primarily associated with mechanisms of lung resistance. Cluster 2a genes (yellow) are associated with coordinating cellular immune responses to bacteria and immune cell recruitment (Fig 6B). Cluster 2b genes (orange) are also genes that coordinate cellular response to bacterial stimuli, activating the complement system and recruiting innate immune cells such as neutrophils (Fig. 6B). Cluster 1 genes (navy) were expressed at higher levels at 3 dpi and are genes associated with regulation of defense responses to bacteria and NF-κB transcription (Fig. 6B). Interestingly, one of the GO biological processes that was identified was the negative regulation of NF-κB transcription pathways, indicating a suppression of lung resistance pathways in 3 dpi SCI mice. In contrast to genes in clusters 1 and 2, cluster 3 genes were downregulated in SCI mice, at both 12 hpi and 3 dpi. Cluster 3a (green) genes are those needed to produce key inflammatory cytokines (e.g., IL-17) and recruit/regulate leukocytes, and cluster 3b genes (aqua) are needed to activate innate immune cells and regulate differentiation of myeloid cells, leukocytes, and lymphocytes (Fig. 6B). All cluster 3 genes are key genes regulating lung resistance pathways. Thus, transcriptional control of lung resistance pathways increased and decreased within the first 3 d after SCI. An analysis of the individual genes that were differentially expressed (≥2-fold, upregulated or downregulated) between sham and SCI lungs at 12 hpi revealed that approximately equal numbers of genes were upregulated (30 genes) and downregulated (24 genes) (Fig. 7A, 7B). However, by 3dpi most DEGs were decreased relative to lungs from sham-injured control mice (21 downregulated versus 7 upregulated) (Fig. 7A, 7B).

FIGURE 7.

SCI disrupts molecular pathways regulating immune resistance and tissue resilience in a time-dependent fashion. (A) Table of genes differentially expressed (genes with ≥2-fold regulation) in SCI lungs relative to sham lungs at each time point. (B) Venn diagram showing the number and names of DEGs (fold change ≥2) shared between each time point. Blue text indicates downregulated genes in SCI lungs, whereas red indicates upregulation when compared with corresponding shams at 12 hpi (C) and 3 dpi (D). Purple text indicates genes that were upregulated and downregulated in SCI lungs at different time points. (C) Top Gene Ontology (GO) biological process terms associated with DEGs in SCI lungs relative to sham at each time point. Count indicates the number of DEGs represented in the associated GO term. n = 4 per group pooled for PCR array analysis from one experiment.

FIGURE 7.

SCI disrupts molecular pathways regulating immune resistance and tissue resilience in a time-dependent fashion. (A) Table of genes differentially expressed (genes with ≥2-fold regulation) in SCI lungs relative to sham lungs at each time point. (B) Venn diagram showing the number and names of DEGs (fold change ≥2) shared between each time point. Blue text indicates downregulated genes in SCI lungs, whereas red indicates upregulation when compared with corresponding shams at 12 hpi (C) and 3 dpi (D). Purple text indicates genes that were upregulated and downregulated in SCI lungs at different time points. (C) Top Gene Ontology (GO) biological process terms associated with DEGs in SCI lungs relative to sham at each time point. Count indicates the number of DEGs represented in the associated GO term. n = 4 per group pooled for PCR array analysis from one experiment.

Close modal

Analyzing the GO terms associated with increased genes (e.g., Tlr1, Ccl12, Nlrp3, Apcs, Cd80, Cd86) at 12 dpi suggested that early after SCI, lung innate defense mechanisms are poised to survey and detect pathogens (i.e., confer lung resistance) (Fig. 7C). However, genes controlling T cell differentiation (e.g., Il6, Stat6, Stat3, Il23a, Foxp3, Tbx21) are decreased, indicating that despite the detection of pathogens, the recruitment of immune cells needed to confer lung resistance is impaired (Fig. 7B, 7C). Similar disruptions in lung resistance pathways were evident at 3 dpi. Specifically, genes associated with decreasing leukocyte proliferation (e.g., Ccl12, Foxp3) and inhibiting NF-κB transcription and TLR pathway signaling (e.g., Irf7, NFĸBIA, Ticam) were increased. These changes were coupled with a decrease in genes regulating T cell lineage commitment (e.g., Stat6, Stat3, Tbx21, Il4, Il6) (Fig. 7B, 7C).

Importantly, at 3 dpi when mechanisms controlling lung resistance are dysfunctional, there was also evidence of impaired lung resilience. Proper lung resilience is needed to control and resolve the inflammatory response initiated by pathogen detection in the lung (13, 34, 35), but several genes that confer lung resilience (e.g., IL-10, Stat3) were downregulated after SCI (Fig. 7A–C). Overall, these data indicate that SCI impairs lung-intrinsic mechanisms of resistance and resilience against infection.

To better understand the functional implications of DEGs in lung after SCI, a network diagram was created using the STRING (https://string-db.org) database and Cytoscape (see Materials and Methods). This network was comprised of all DEGs at 12 hpi and 3 dpi and consisted of 59 nodes and 329 edges. The network was color-coded to reflect DEGs that were upregulated (red) or downregulated (blue) or that were differentially regulated at both time points (purple; (Fig. 8A). cytoHubba was then used to analyze network features of each gene and that gene’s relative importance within the network (see Materials and Methods). This analysis revealed Myd88, Ccl5, Cxcl10, Foxp3, Stat1, Ifng, Tnf, Il6, Il10, and Il4 as key “hub” genes within the network (Fig. 8A). Importantly, eight hub genes were consistently downregulated after SCI (Tnf, Il6, Il10, Il4, Ifng, Stat1, Cxcl10, and Myd88). It is notable that Myd88, Infg, and TNF are downstream intermediates of NFĸB-mediated resistance pathways and Il10 and Il6 are key components of lung resilience (13, 34, 35). Subsequent analysis of the network using MCODE revealed four key gene clusters, of which cluster 1 included 9 out of 10 of the identified hub genes. The remaining hub gene, Myd88, was found in cluster 4 (Fig. 8B). The high network connectivity within cluster 1 likely implicates these genes as master regulators of pulmonary immune function (21). Key GO terms indicate that the main molecular function of the hub genes relate to cytokine and chemokine activity (Fig. 8C), and that these genes regulate lymphocyte and leukocyte proliferation (Fig. 8C).

FIGURE 8.

Protein–protein interaction and hub gene analyses of DEGs identify disrupted cytokine pathways after SCI. (A) DEGs (fold regulation ≥2) from both 12 hpi and 3 dpi were inputted into the STRING network database and used to create a protein–protein interaction (PPI) network from and then visualized with Cytoscape. A total of 59 nodes and 329 edges were identified in the network. Genes upregulated at both time points are in red, and those downregulated at both time points are in blue. Genes that were upregulated or downregulated differently at the two time points are in purple. The top 10 network hub genes (outlined in yellow) were identified using CytoHubba in Cytoscape. (B) Densely interconnected subregions in the PPI network were identified using the MCODE algorithm. Four major clusters were found. All identified hub genes were represented in the clusters, with most the hub genes (9 out of 10, identified as those in yellow) being found in cluster 1, the most interconnected cluster. (C) As a result, the genes represented in cluster 1 were analyzed for their related GO terms using the enrichGO function in the clusterProfiler package in R. The top 10 molecular function and biological process GO terms were identified. The top 10 terms were determined by selecting those with the smallest adjusted p value after correction for multiple comparison with the Benjamini–Hochberg Procedure. The gene ratio represents the number of DEGs in the terms compared with the total number of genes, and the count is the number of DEGs genes found in the respective GO term. (D) RT-PCR analysis results of 10 identified hub genes at each time point in individuals (n = 4 per group from one experiment) to validate PCR array results.

FIGURE 8.

Protein–protein interaction and hub gene analyses of DEGs identify disrupted cytokine pathways after SCI. (A) DEGs (fold regulation ≥2) from both 12 hpi and 3 dpi were inputted into the STRING network database and used to create a protein–protein interaction (PPI) network from and then visualized with Cytoscape. A total of 59 nodes and 329 edges were identified in the network. Genes upregulated at both time points are in red, and those downregulated at both time points are in blue. Genes that were upregulated or downregulated differently at the two time points are in purple. The top 10 network hub genes (outlined in yellow) were identified using CytoHubba in Cytoscape. (B) Densely interconnected subregions in the PPI network were identified using the MCODE algorithm. Four major clusters were found. All identified hub genes were represented in the clusters, with most the hub genes (9 out of 10, identified as those in yellow) being found in cluster 1, the most interconnected cluster. (C) As a result, the genes represented in cluster 1 were analyzed for their related GO terms using the enrichGO function in the clusterProfiler package in R. The top 10 molecular function and biological process GO terms were identified. The top 10 terms were determined by selecting those with the smallest adjusted p value after correction for multiple comparison with the Benjamini–Hochberg Procedure. The gene ratio represents the number of DEGs in the terms compared with the total number of genes, and the count is the number of DEGs genes found in the respective GO term. (D) RT-PCR analysis results of 10 identified hub genes at each time point in individuals (n = 4 per group from one experiment) to validate PCR array results.

Close modal

SCI-induced changes in hub genes detected by PCR array were validated using RT-qPCR (Fig. 8D). Consistent with data in (Figs. 6 and 7, Cxcl10, Myd88, Stat1, Tnf, Il6, and Infg were decreased (≥2 fold) at 12 hpi. Foxp3 also was decreased after SCI, but the magnitude of change was less than predicted by PCR array. The increase in Ccl5 at 12 hpi detected by PCR array was not validated with qPCR (Fig. 8D). At 3 pdi, Il4 was decreased ≥2 fold as predicted by PCR array. Other hub genes also decreased as expected, but to a lesser extent than was predicted from the PCR array (Fig. 8D). Taken together, these analyses indicate that SCI impairs key gene regulatory networks needed to confer both lung resistance and resilience to infection.

Pulmonary complications, such as pneumonia, increase mortality and impair neurologic recovery after SCI (15). How or why the risk to contract pneumonia increases after SCI is poorly understood, but the onset of systemic immune suppression postinjury has been implicated as a causative factor (7, 8, 1012, 36, 37). New data in the present study indicate that, in addition to causing systemic immune suppression, SCI also disrupts lung-intrinsic host-defense mechanisms. Notably, dysfunction of pulmonary bacterial defense is exacerbated after T1 SCI when compared with T3 SCI, and, although it is possible to boost leukocyte recruitment to the lung, this does not prevent bacterial pneumonia.

By 12 h after SCI (T3 level), we noted an increase in lung bacterial burden with a corresponding increase in lung neutrophils. By 3 dpi, lung neutrophils returned to baseline and lung bacteria were no longer detected. This rapid resolution of bacterial pneumonia could indicate that the unique neutrophil-dependent host defense niche that exists in the lung capillary beds remains intact, at least temporarily, after T3 SCI (38). In contrast, T1 SCI mice develop acute pulmonary leukopenia and spontaneous bacterial pneumonia that persists until at least 3 dpi. We and others have previously implicated aberrant sympathetic nervous system control over the adrenal glands and primary and secondary lymphoid organs as mechanisms underlying spinal level–dependent changes in systemic immune suppression and bacterial pneumonia (7, 8, 10, 12, 37). A similar neurogenic mechanism may contribute to observed differences in lung immune surveillance after T1 and T3 SCI.

The lung is innervated by parasympathetic (vagus) and sympathetic nerve fibers, which collaborate to control airway compliance, glandular secretions, pulmonary blood flow, and resistance to pathogens (3943). The vagus nerve is a cranial nerve with neuronal cell bodies originating in the brainstem. Given its anatomical location, the vagus nerve is not injured by traumatic SCI. Conversely, sympathetic neurons, which are located in the thoracic and upper lumbar spinal cord, are either directly damaged or become isolated from executive control by supraspinal axons (e.g., from brain and brainstem) after SCI. The higher the level of SCI, the greater the disruption to spinal sympathetic neurons and associated spinal reflexes. As such, when SCI occurs at the T1 spinal level, all or most sympathetic control over the trachea, lung, and intercostal respiratory muscles is lost, whereas pulmonary sympathetic tone is partially preserved after T3 SCI. Partial or complete loss of sympathetic nervous system tone in the lung would proportionally increase the relative influence that the vagus nerve has on the lungs, airways, and respiratory muscles, at least early after SCI before intraspinal plasticity and structural remodeling of spinal sympathetic reflexes develop (see below).

At the cellular and molecular levels, the vagus nerve negatively regulates immune signaling via the release of acetylcholine with subsequent binding to α7 nicotinic cholinergic receptors (α7nAchRs) on immune cells (4450). Lung-resident cells, including alveolar epithelial cells (AECs), alveolar macrophages, and type 2 innate lymphoid cells (ILCs) also express α7nAchRs (45, 5153). In fact, AECs express ∼100-fold more α7nAchRs than do circulating myeloid cells, suggesting that the parasympathetic nervous system exerts profound control over AEC function in health and disease (44). In the context of infection, acetylcholine binding to α7nAchRs on AECs reduces inflammatory cytokine release, indicating that enhanced vagal tone could suppress lung-resident host-defense functions (44). Peptidergic, TRPV1+ vagal sensory neurons also exert tonic immunoregulatory effects on lung neutrophils and γδ T cells (54).

Approximately 20% of nerve fibers innervating the lung and airways are sympathetic nerves, and both alveolar macrophages and type 2 ILCs express high levels of adrenergic receptors (45, 55). In normal (CNS intact) mice, norepinephrine or specific agonists of the β2-adrenergic receptor inhibit inflammatory signaling in alveolar macrophages and ILCs, and ablation of lung sympathetic nerves exacerbates inflammatory signaling and cellular responses to lung infection (45, 55). Thus, the sympathetic nervous system, similar to the parasympathetic nervous system, negatively regulates pulmonary inflammation. After SCI, because the vagus nerve is not injured and spinal shock temporarily impairs residual sympathetic reflexes, an acute increase in pulmonary vagal tone is expected with a corresponding decrease in tissue-specific bacterial defense mechanisms. A similar phenomenon occurs in stroke patients, that is, parasympathetic dysfunction manifests early after stroke (within 1–3 d), and an imbalance in autonomic tone is associated with worse outcome (25, 44, 56). At increasingly longer times after SCI, these effects will be exacerbated due to plasticity and structural remodeling of spinal sympathetic networks, which subserve the formation of exaggerated sympathetic reflexes that suppress systemic immune function (7, 9, 24). The delayed formation of these aberrant spinal sympathetic reflexes could restrict leukocyte recruitment or exacerbate killing of lung macrophages and lymphocytes as occurs in other immune organs after SCI (36, 37, 57). These aberrant sympathetic reflexes could also explain why there is a delayed but persistent reduction in innate and adaptive immune cells in the lung with a corresponding increase in bacterial pneumonia as late as 28 dpi (see (Figs. 1 and 2). Thus, postinjury pulmonary dysautonomia likely causes early and lasting impairment of lung-resident host defense mechanisms and may explain why simply boosting immune cell recruitment to the lung, whether by systemic injection of AMD3100 (this study) or immune-potentiating cytokines such as GM-CSF or IFN-γ (58, 59), has de minimis effects on reversing bacterial pneumonia.

The loss of appropriate autonomic control of lung and airways may also cause the rapid onset of SCI-dependent changes in the transcriptional control of genes that regulate lung resistance and resilience. Resistance relates to the lungs ability to eliminate pulmonary pathogens, whereas resilience refers to the lung’s capacity to withstand damage caused by the pathogen and associated immune responses (13). In general, the coordinated activity of lung resistance functions requires activation of NF-κB, while the activation of Stat3-dependent signaling pathways regulate lung resilience (34). Our data indicate that Nfkb1 gene transcription is downregulated early (12 hpi) postinjury with delayed upregulation (at 3 dpi) of NF-κBia, a potent inhibitor of NF-κB activation (60). The inhibition of NF-κB gene transcription is accompanied by downregulation of genes that are required for optimal pathogen detection and activation of innate immune cells (e.g., Tlr9, Il6, Ifng, Myd88). Impaired innate immunity would subsequently weaken adaptive immune responses. With bacterial pneumonia, inflammatory cytokines and chemokines produced by lung innate immune cells (e.g., neutrophils, alveolar macrophages) coordinate the recruitment of T lymphocytes and their subsequent differentiation into Th1/Th17 cells (61). An upregulation of Tbx21 and Stat4, as well as Ifng and Il12, are needed to ensure Th1/Th17 differentiation (61, 62). After SCI, Tbx21, Stat4, and Ifng are decreased in lung, and Ifng may be a master regulator of lung resistance after SCI (Figs. 7, 8). It is tempting to speculate that aberrant autonomic reflex function in the lung contributes to or causes these changes in gene transcription. For example, activating α7nAchR in AECs suppresses Il6 and TNF-α (44). In lungs infected with bacteria, activating CGRP, TRPV1+ vagal sensory neurons inhibits production of TNF-α, Il6, and Cxcl1 (54). Similarly, activation of the sympathetic nervous system blocks production of TNF-α and Ccl2 by alveolar macrophages and Il5 and Il13 by ILCs (46, 6365). Spinal cord injury also perturbs transcriptional control of prominent genes linked to lung resilience (e.g., downregulation of Il10 and Stat33), although it is not clear how or whether aberrant autonomic tone in the lung contributes to these injury-dependent changes in gene transcription. Indeed, activation of both nicotinic cholinergic and adrenergic receptors increases Il10- and Stat3-dependent signaling in immune cells (47, 6670). Future studies are needed to determine cell-specific changes in key resistance and resilience genes and whether changes in the expression of these genes are modulated by cholinergic and adrenergic signaling.

Because we designed this as an exploratory study to evaluate the effects of SCI on cellular and molecular mechanisms of lung host defense, most data in the current report are correlative. Our efforts to enhance immune cell recruitment to the lung and prevent or reverse bacterial pneumonia after SCI were largely unsuccessful, indicating that combination therapies that target multiple cell types, receptors, and/or signaling pathways may be more effective approaches. Indeed, a recent publication identified gene expression profiles, functions, and the anatomical locations of 58 different cell populations in normal human lung (71). A similarly comprehensive cellular and molecular mapping study of the lung after experimental SCI would reveal how various cells and intercellular interactions change, both over time and as a function of injury level. Additionally, this study does not address the potential source of bacteria that colonize the lung after SCI. Although it is likely that some environmental bacteria are inhaled, there is also evidence that pathogenic bacteria may migrate to the lung from the gut. Evidence for bacterial translocation to the lung has been described after stroke in both humans and preclinical animal models (7274). A similar phenomenon may occur after SCI. Indeed, gut barrier integrity is disrupted after SCI with a corresponding increase in bacterial colonization occurring in the lung and other organs after SCI (16). However, further research is needed to determine the source and composition of lung bacteria after SCI.

In conclusion, novel data in this study show that SCI impairs the lung’s ability to mount a rapid and coordinated defense against bacteria and that the development of combinatorial strategies that target nonneurogenic and neurogenic mechanisms of immune modulation may be more efficacious for preventing or limiting the effects of bacterial pneumonia.

This work was supported by the Craig H. Neilsen Foundation (Grants 457267 to F.H.B., 647110 to K.A.M., 596764 to J.M.S., 457328 to A.R.F.), The Ohio State University Presidential Postdoctoral Fellowship (to K.A.M.), Wings for Life (to F.H.B., P.G.P., J.M.S., and A.R.F.), National Institute of Neurological Disorders and Stroke/National Institutes of Health Grants R01NS099532 (to P.G.P.), R01NS083942 (to P.G.P.), R35NS111582 (to P.G.P.), and R01NS118200 (to J.M.S.), the National Institute of Disability, Independent Living and Rehabilitation Research (Grant 90SI5020 to J.M.S.), the Ray W. Poppleton Endowment (to P.G.P.), the European Union (EU Era Net–Neuron Program, SILENCE Grant 01EW170A to J.M.S.), and by the William E. Hunt and Charlotte M. Curtis Endowment (to J.M.S.). J.M.S. is also a Discovery Theme Initiative Scholar (Chronic Brain Injury) of The Ohio State University.

The online version of this article contains supplemental material.

Abbreviations used in this article:

AEC

alveolar epithelial cell

DEG

differentially expressed gene

dpi

day postinjury

GO

Gene Ontology

hpi

hour postinjury

IDS

immune deficiency syndrome

ILC

innate lymphoid cell

MCODE

Molecular Complex Detection

α7nAchR

α7 nicotinic cholinergic receptor

qPCR

quantitative PCR

RT-qPCR

reverse transcriptase–qPCR

SCI

spinal cord injury

SCI-IDS

SCI–induced IDS

STRING

Search Tool for the Retrieval of Interacting Genes/Proteins

T1Tx

first thoracic vertebral level transection

T3Tx

third thoracic vertebral level transection

1.
DeVivo
M. J.
,
K. J.
Black
,
S. L.
Stover
.
1993
.
Causes of death during the first 12 years after spinal cord injury.
Arch. Phys. Med. Rehabil.
74
:
248
254
.
2.
Strauss
D. J.
,
M. J.
Devivo
,
D. R.
Paculdo
,
R. M.
Shavelle
.
2006
.
Trends in life expectancy after spinal cord injury.
Arch. Phys. Med. Rehabil.
87
:
1079
1085
.
3.
Failli
V.
,
M. A.
Kopp
,
C.
Gericke
,
P.
Martus
,
S.
Klingbeil
,
B.
Brommer
,
I.
Laginha
,
Y.
Chen
,
M. J.
DeVivo
,
U.
Dirnagl
,
J. M.
Schwab
.
2012
.
Functional neurological recovery after spinal cord injury is impaired in patients with infections.
Brain
135
:
3238
3250
.
4.
Kopp
M. A.
,
R.
Watzlawick
,
P.
Martus
,
V.
Failli
,
F. W.
Finkenstaedt
,
Y.
Chen
,
M. J.
DeVivo
,
U.
Dirnagl
,
J. M.
Schwab
.
2017
.
Long-term functional outcome in patients with acquired infections after acute spinal cord injury.
Neurology
88
:
892
900
.
5.
Savic
G.
,
M. J.
DeVivo
,
H. L.
Frankel
,
M. A.
Jamous
,
B. M.
Soni
,
S.
Charlifue
.
2017
.
Long-term survival after traumatic spinal cord injury: a 70-year British study.
Spinal Cord
55
:
651
658
.
6.
Riegger
T.
,
S.
Conrad
,
K.
Liu
,
H. J.
Schluesener
,
M.
Adibzahdeh
,
J. M.
Schwab
.
2007
.
Spinal cord injury-induced immune depression syndrome (SCI-IDS).
Eur. J. Neurosci.
25
:
1743
1747
.
7.
Zhang
Y.
,
Z.
Guan
,
B.
Reader
,
T.
Shawler
,
S.
Mandrekar-Colucci
,
K.
Huang
,
Z.
Weil
,
A.
Bratasz
,
J.
Wells
,
N. D.
Powell
, et al
2013
.
Autonomic dysreflexia causes chronic immune suppression after spinal cord injury.
J. Neurosci.
33
:
12970
12981
.
8.
Brommer
B.
,
O.
Engel
,
M. A.
Kopp
,
R.
Watzlawick
,
S.
Müller
,
H.
Prüss
,
Y.
Chen
,
M. J.
DeVivo
,
F. W.
Finkenstaedt
,
U.
Dirnagl
, et al
2016
.
Spinal cord injury-induced immune deficiency syndrome enhances infection susceptibility dependent on lesion level.
Brain
139
:
692
707
.
9.
Ueno
M.
,
Y.
Ueno-Nakamura
,
J.
Niehaus
,
P. G.
Popovich
,
Y.
Yoshida
.
2016
.
Silencing spinal interneurons inhibits immune suppressive autonomic reflexes caused by spinal cord injury.
Nat. Neurosci.
19
:
784
787
.
10.
Prüss
H.
,
A.
Tedeschi
,
A.
Thiriot
,
L.
Lynch
,
S. M.
Loughhead
,
S.
Stutte
,
I. B.
Mazo
,
M. A.
Kopp
,
B.
Brommer
,
C.
Blex
, et al
2017
.
Spinal cord injury-induced immunodeficiency is mediated by a sympathetic-neuroendocrine adrenal reflex.
Nat. Neurosci.
20
:
1549
1559
.
11.
Mironets
E.
,
P.
Osei-Owusu
,
V.
Bracchi-Ricard
,
R.
Fischer
,
E. A.
Owens
,
J.
Ricard
,
D.
Wu
,
T.
Saltos
,
E.
Collyer
,
S.
Hou
, et al
2018
.
Soluble TNFα signaling within the spinal cord contributes to the development of autonomic dysreflexia and ensuing vascular and immune dysfunction after spinal cord injury.
J. Neurosci.
38
:
4146
4162
.
12.
Carpenter
R. S.
,
J. M.
Marbourg
,
F. H.
Brennan
,
K. A.
Mifflin
,
J. C. E.
Hall
,
R. R.
Jiang
,
X. M.
Mo
,
M.
Karunasiri
,
M. H.
Burke
,
A. M.
Dorrance
,
P. G.
Popovich
.
2020
.
Spinal cord injury causes chronic bone marrow failure.
Nat. Commun.
11
:
3702
.
13.
Quinton
L. J.
,
J. P.
Mizgerd
.
2015
.
Dynamics of lung defense in pneumonia: resistance, resilience, and remodeling.
Annu. Rev. Physiol.
77
:
407
430
.
14.
Iwasaki
A.
,
E. F.
Foxman
,
R. D.
Molony
.
2017
.
Early local immune defences in the respiratory tract.
Nat. Rev. Immunol.
17
:
7
20
.
15.
Hartl
D.
,
R.
Tirouvanziam
,
J.
Laval
,
C. M.
Greene
,
D.
Habiel
,
L.
Sharma
,
A. Ö.
Yildirim
,
C. S.
Dela Cruz
,
C. M.
Hogaboam
.
2018
.
Innate immunity of the lung: from basic mechanisms to translational medicine.
J. Innate Immun.
10
:
487
501
.
16.
Kigerl
K. A.
,
J. C. E.
Hall
,
L.
Wang
,
X.
Mo
,
Z.
Yu
,
P. G.
Popovich
.
2016
.
Gut dysbiosis impairs recovery after spinal cord injury.
J. Exp. Med.
213
:
2603
2620
.
17.
Livak
K. J.
,
T. D.
Schmittgen
.
2001
.
Analysis of relative gene expression data using real-time quantitative PCR and the 2−ΔΔC(T) method.
Methods
25
:
402
408
.
18.
Reimand
J.
,
R.
Isserlin
,
V.
Voisin
,
M.
Kucera
,
C.
Tannus-Lopes
,
A.
Rostamianfar
,
L.
Wadi
,
M.
Meyer
,
J.
Wong
,
C.
Xu
, et al
2019
.
Pathway enrichment analysis and visualization of omics data using g:Profiler, GSEA, Cytoscape and EnrichmentMap.
Nat. Protoc.
14
:
482
517
.
19.
Szklarczyk
D.
,
A. L.
Gable
,
D.
Lyon
,
A.
Junge
,
S.
Wyder
,
J.
Huerta-Cepas
,
M.
Simonovic
,
N. T.
Doncheva
,
J. H.
Morris
,
P.
Bork
, et al
2019
.
STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets.
Nucleic Acids Res.
47
(
D1
):
D607
D613
.
20.
Chin
C. H.
,
S. H.
Chen
,
H. H.
Wu
,
C. W.
Ho
,
M. T.
Ko
,
C. Y.
Lin
.
2014
.
cytoHubba: identifying hub objects and sub-networks from complex interactome.
BMC Syst. Biol.
8
(
Suppl 4
):
S11
.
21.
Bader
G. D.
,
C. W. V.
Hogue
.
2003
.
An automated method for finding molecular complexes in large protein interaction networks.
BMC Bioinformatics
4
:
2
.
22.
Yu
G.
,
L.
Wang
,
Y.
Han
,
Q.
He
.
2012
.
clusterProfiler: an R package for comparing biological themes among gene clusters.
OMICS
16
:
284
287
.
23.
Motulsky
H. J.
,
R. E.
Brown
.
2006
.
Detecting outliers when fitting data with nonlinear regression—a new method based on robust nonlinear regression and the false discovery rate.
BMC Bioinformatics
7
:
123
.
24.
Mironets
E.
,
R.
Fischer
,
V.
Bracchi-Ricard
,
T. M.
Saltos
,
T. S.
Truglio
,
M. L.
O’Reilly
,
K. A.
Swanson
,
J. R.
Bethea
,
V. J.
Tom
.
2020
.
Attenuating neurogenic sympathetic hyperreflexia robustly improves antibacterial immunity after chronic spinal cord injury.
J. Neurosci.
40
:
478
492
.
25.
Prass
K.
,
C.
Meisel
,
C.
Höflich
,
J.
Braun
,
E.
Halle
,
T.
Wolf
,
K.
Ruscher
,
I. V.
Victorov
,
J.
Priller
,
U.
Dirnagl
, et al
2003
.
Stroke-induced immunodeficiency promotes spontaneous bacterial infections and is mediated by sympathetic activation reversal by poststroke T helper cell type 1-like immunostimulation.
J. Exp. Med.
198
:
725
736
.
26.
Diaz
D.
,
E.
Lopez-Dolado
,
S.
Haro
,
J.
Monserrat
,
C.
Martinez-Alonso
,
D.
Balomeros
,
A.
Albillos
,
M.
Alvarez-Mon
.
2021
.
Systemic inflammation and the breakdown of intestinal homeostasis are key events in chronic spinal cord injury patients.
Int. J. Mol. Sci.
22
:
744
.
27.
Liu
J.
,
H.
An
,
D.
Jiang
,
W.
Huang
,
H.
Zou
,
C.
Meng
,
H.
Li
.
2004
.
Study of bacterial translocation from gut after paraplegia caused by spinal cord injury in rats.
Spine
29
:
164
169
.
28.
Raz
R.
,
R.
Colodner
,
C. M.
Kunin
.
2005
.
Who are you–Staphylococcus saprophyticus?
Clin. Infect. Dis.
40
:
896
898
.
29.
Sadikot
R. T.
,
T. S.
Blackwell
,
J. W.
Christman
,
A. S.
Prince
.
2005
.
Pathogen-host interactions in Pseudomonas aeruginosa pneumonia.
Am. J. Respir. Crit. Care Med.
171
:
1209
1223
.
30.
Dietert
K.
,
B.
Gutbier
,
S. M.
Wienhold
,
K.
Reppe
,
X.
Jiang
,
L.
Yao
,
C.
Chaput
,
J.
Naujoks
,
M.
Brack
,
A.
Kupke
, et al
2017
.
Spectrum of pathogen- and model-specific histopathologies in mouse models of acute pneumonia.
PLoS One
12
:
e0188251
.
31.
Andonegui
G.
,
K.
Goring
,
D.
Liu
,
D. M.
McCafferty
,
B. W.
Winston
.
2009
.
Characterization of S. pneumoniae pneumonia-induced multiple organ dysfunction syndrome: an experimental mouse model of Gram-positive sepsis.
Shock
31
:
423
428
.
32.
Hoover
J. L.
,
T. F.
Lewandowski
,
C. L.
Mininger
,
C. M.
Singley
,
S.
Sucoloski
,
S.
Rittenhouse
.
2017
.
A robust pneumonia model in immunocompetent rodents to evaluate antibacterial efficacy against S. pneumoniae, H. influenzae, K. pneumoniae, P. aeruginosa or A. baumannii.
J. Vis. Exp.
2017
:
55068
.
33.
Prass
K.
,
J. S.
Braun
,
U.
Dirnagl
,
C.
Meisel
,
A.
Meisel
.
2006
.
Stroke propagates bacterial aspiration to pneumonia in a model of cerebral ischemia.
Stroke
37
:
2607
2612
.
34.
Quinton
L. J.
,
J. P.
Mizgerd
.
2011
.
NF-κB and STAT3 signaling hubs for lung innate immunity.
Cell Tissue Res.
343
:
153
165
.
35.
Quinton
L. J.
,
A. J.
Walkey
,
J. P.
Mizgerd
.
2018
.
Integrative physiology of pneumonia.
Physiol. Rev.
98
:
1417
1464
.
36.
Lucin
K. M.
,
V. M.
Sanders
,
T. B.
Jones
,
W. B.
Malarkey
,
P. G.
Popovich
.
2007
.
Impaired antibody synthesis after spinal cord injury is level dependent and is due to sympathetic nervous system dysregulation.
Exp. Neurol.
207
:
75
84
.
37.
Lucin
K. M.
,
V. M.
Sanders
,
P. G.
Popovich
.
2009
.
Stress hormones collaborate to induce lymphocyte apoptosis after high level spinal cord injury.
J. Neurochem.
110
:
1409
1421
.
38.
Yipp
B. G.
,
J. H.
Kim
,
R.
Lima
,
L. D.
Zbytnuik
,
B.
Petri
,
N.
Swanlund
,
M.
Ho
,
V. G.
Szeto
,
T.
Tak
,
L.
Koenderman
, et al
2017
.
The lung is a host defense niche for immediate neutrophil-mediated vascular protection.
Sci. Immunol.
2
:
eaam8929
.
39.
Terson de Paleville
D. G.
,
W. B.
McKay
,
R. J.
Folz
,
A. V.
Ovechkin
.
2011
.
Respiratory motor control disrupted by spinal cord injury: mechanisms, evaluation, and restoration.
Transl. Stroke Res.
2
:
463
473
.
40.
Bleecker
E. R.
1986
.
Cholinergic and neurogenic mechanisms in obstructive airways disease.
Am. J. Med.
81
(
5A
):
93
102
.
41.
Pincus
A. B.
,
A. D.
Fryer
,
D. B.
Jacoby
.
2021
.
Mini review: neural mechanisms underlying airway hyperresponsiveness.
Neurosci. Lett.
751
:
135795
.
42.
Mazzone
S. B.
,
B. J.
Canning
.
2013
.
Autonomic neural control of the airways.
Handb. Clin. Neurol.
117
:
215
228
.
43.
Mazzone
S. B.
,
B. J.
Undem
.
2016
.
Vagal afferent innervation of the airways in health and disease.
Physiol. Rev.
96
:
975
1024
.
44.
Engel
O.
,
L.
Akyüz
,
A. C.
da Costa Goncalves
,
K.
Winek
,
C.
Dames
,
M.
Thielke
,
S.
Herold
,
C.
Böttcher
,
J.
Priller
,
H. D.
Volk
, et al
2015
.
Cholinergic pathway suppresses pulmonary innate immunity facilitating pneumonia after stroke.
Stroke
46
:
3232
3240
.
45.
Liu
T.
,
L.
Yang
,
X.
Han
,
X.
Ding
,
J.
Li
,
J.
Yang
.
2020
.
Local sympathetic innervations modulate the lung innate immune responses.
Sci. Adv.
6
:
eaay1497
.
46.
Tracey
K. J.
2002
.
The inflammatory reflex.
Nature
420
:
853
859
.
47.
Borovikova
L. V.
,
S.
Ivanova
,
M.
Zhang
,
H.
Yang
,
G. I.
Botchkina
,
L. R.
Watkins
,
H.
Wang
,
N.
Abumrad
,
J. W.
Eaton
,
K. J.
Tracey
.
2000
.
Vagus nerve stimulation attenuates the systemic inflammatory response to endotoxin.
Nature
405
:
458
462
.
48.
Rosas-Ballina
M.
,
K. J.
Tracey
.
2009
.
Cholinergic control of inflammation.
J. Intern. Med.
265
:
663
679
.
49.
Bernik
T. R.
,
S. G.
Friedman
,
M.
Ochani
,
R.
DiRaimo
,
L.
Ulloa
,
H.
Yang
,
S.
Sudan
,
C. J.
Czura
,
S. M.
Ivanova
,
K. J.
Tracey
.
2002
.
Pharmacological stimulation of the cholinergic antiinflammatory pathway.
J. Exp. Med.
195
:
781
788
.
50.
Borovikova
L. V.
,
S.
Ivanova
,
D.
Nardi
,
M.
Zhang
,
H.
Yang
,
M.
Ombrellino
,
K. J.
Tracey
.
2000
.
Role of vagus nerve signaling in CNI-1493-mediated suppression of acute inflammation.
Auton. Neurosci.
85
:
141
147
.
51.
Maouche
K.
,
M.
Polette
,
T.
Jolly
,
K.
Medjber
,
I.
Cloëz-Tayarani
,
J. P.
Changeux
,
H.
Burlet
,
C.
Terryn
,
C.
Coraux
,
J. M.
Zahm
, et al
2009
.
α7 Nicotinic acetylcholine receptor regulates airway epithelium differentiation by controlling basal cell proliferation.
Am. J. Pathol.
175
:
1868
1882
.
52.
Kawashima
K.
,
T.
Fujii
.
2000
.
Extraneuronal cholinergic system in lymphocytes.
Pharmacol. Ther.
86
:
29
48
.
53.
Chu
C.
,
C. N.
Parkhurst
,
W.
Zhang
,
L.
Zhou
,
H.
Yano
,
M.
Arifuzzaman
,
D.
Artis
.
2021
.
The ChAT-acetylcholine pathway promotes group 2 innate lymphoid cell responses and anti-helminth immunity.
Sci. Immunol.
6
:
eabe3218
.
54.
Baral
P.
,
B. D.
Umans
,
L.
Li
,
A.
Wallrapp
,
M.
Bist
,
T.
Kirschbaum
,
Y.
Wei
,
Y.
Zhou
,
V. K.
Kuchroo
,
P. R.
Burkett
, et al
2018
.
Nociceptor sensory neurons suppress neutrophil and γδ T cell responses in bacterial lung infections and lethal pneumonia. [Published erratum appears in 2018 Nat. Med. 24: 1625–1626.]
Nat. Med.
24
:
417
426
.
55.
Moriyama
S.
,
J. R.
Brestoff
,
A. L.
Flamar
,
J. B.
Moeller
,
C. S. N.
Klose
,
L. C.
Rankin
,
N. A.
Yudanin
,
L. A.
Monticelli
,
G. G.
Putzel
,
H. R.
Rodewald
,
D.
Artis
.
2018
.
β2-Adrenergic receptor-mediated negative regulation of group 2 innate lymphoid cell responses.
Science
359
:
1056
1061
.
56.
Walter
U.
,
S.
Kolbaske
,
R.
Patejdl
,
V.
Steinhagen
,
M.
Abu-Mugheisib
,
A.
Grossmann
,
C.
Zingler
,
R.
Benecke
.
2013
.
Insular stroke is associated with acute sympathetic hyperactivation and immunodepression.
Eur. J. Neurol.
20
:
153
159
.
57.
Noble
B. T.
,
F. H.
Brennan
,
P. G.
Popovich
.
2018
.
The spleen as a neuroimmune interface after spinal cord injury.
J. Neuroimmunol.
321
:
1
11
.
58.
Dames
C.
,
K.
Winek
,
Y.
Beckers
,
O.
Engel
,
A.
Meisel
,
C.
Meisel
.
2018
.
Immunomodulatory treatment with systemic GM-CSF augments pulmonary immune responses and improves neurological outcome after experimental stroke.
J. Neuroimmunol.
321
:
144
149
.
59.
Jagdmann
S.
,
D.
Berchtold
,
B.
Gutbier
,
M.
Witzenrath
,
A.
Meisel
,
C.
Meisel
,
C.
Dames
.
2021
.
Efficacy and safety of intratracheal IFN-γ treatment to reverse stroke-induced susceptibility to pulmonary bacterial infections.
J. Neuroimmunol.
355
:
577568
.
60.
Hawiger
J.
2001
.
Innate immunity and inflammation: a transcriptional paradigm.
Immunol. Res.
23
:
99
109
.
61.
Kolls
J. K.
2013
.
CD4+ T-cell subsets and host defense in the lung.
Immunol. Rev.
252
:
156
163
.
62.
Curtis
J. L.
2005
.
Cell-mediated adaptive immune defense of the lungs.
Proc. Am. Thorac. Soc.
2
:
412
416
.
63.
Galle-Treger
L.
,
Y.
Suzuki
,
N.
Patel
,
I.
Sankaranarayanan
,
J. L.
Aron
,
H.
Maazi
,
L.
Chen
,
O.
Akbari
.
2016
.
Nicotinic acetylcholine receptor agonist attenuates ILC2-dependent airway hyperreactivity.
Nat. Commun.
7
:
13202
.
64.
Wang
H.
,
M.
Yu
,
M.
Ochani
,
C. A.
Amella
,
M.
Tanovic
,
S.
Susarla
,
J. H.
Li
,
H.
Wang
,
H.
Yang
,
L.
Ulloa
, et al
2003
.
Nicotinic acetylcholine receptor α7 subunit is an essential regulator of inflammation.
Nature
421
:
384
388
.
65.
Valdez-Miramontes
C. E.
,
L. A.
Trejo Martínez
,
F.
Torres-Juárez
,
A.
Rodríguez Carlos
,
S. P.
Marin-Luévano
,
J. P.
de Haro-Acosta
,
J. A.
Enciso-Moreno
,
B.
Rivas-Santiago
.
2020
.
Nicotine modulates molecules of the innate immune response in epithelial cells and macrophages during infection with M. tuberculosis.
Clin. Exp. Immunol.
199
:
230
243
.
66.
de Jonge
W. J.
,
E. P.
van der Zanden
,
F. O.
The
,
M. F.
Bijlsma
,
D. J.
van Westerloo
,
R. J.
Bennink
,
H. R.
Berthoud
,
S.
Uematsu
,
S.
Akira
,
R. M.
van den Wijngaard
,
G. E.
Boeckxstaens
.
2005
.
Stimulation of the vagus nerve attenuates macrophage activation by activating the Jak2-STAT3 signaling pathway. [Published erratum appears in 2005 Nat. Immunol. 6: 954.]
Nat. Immunol.
6
:
844
851
.
67.
Sun
Y.
,
Q.
Li
,
H.
Gui
,
D. P.
Xu
,
Y. L.
Yang
,
D. F.
Su
,
X.
Liu
.
2013
.
MicroRNA-124 mediates the cholinergic anti-inflammatory action through inhibiting the production of pro-inflammatory cytokines.
Cell Res.
23
:
1270
1283
.
68.
Ağaç
D.
,
L. D.
Estrada
,
R.
Maples
,
L. V.
Hooper
,
J. D.
Farrar
.
2018
.
The β2-adrenergic receptor controls inflammation by driving rapid IL-10 secretion.
Brain Behav. Immun.
74
:
176
185
.
69.
Kenney
M. J.
,
C. K.
Ganta
.
2014
.
Autonomic nervous system and immune system interactions.
Compr. Physiol.
4
:
1177
1200
.
70.
Mohammadpour
H.
,
C. R.
MacDonald
,
G.
Qiao
,
M.
Chen
,
B.
Dong
,
B. L.
Hylander
,
P. L.
McCarthy
,
S. I.
Abrams
,
E. A.
Repasky
.
2019
.
β2 adrenergic receptor-mediated signaling regulates the immunosuppressive potential of myeloid-derived suppressor cells.
J. Clin. Invest.
129
:
5537
5552
.
71.
Travaglini
K. J.
,
A. N.
Nabhan
,
L.
Penland
,
R.
Sinha
,
A.
Gillich
,
R. V.
Sit
,
S.
Chang
,
S. D.
Conley
,
Y.
Mori
,
J.
Seita
, et al
2020
.
A molecular cell atlas of the human lung from single-cell RNA sequencing.
Nature
587
:
619
625
.
72.
Stanley
D.
,
L. J.
Mason
,
K. E.
Mackin
,
Y. N.
Srikhanta
,
D.
Lyras
,
M. D.
Prakash
,
K.
Nurgali
,
A.
Venegas
,
M. D.
Hill
,
R. J.
Moore
,
C. H. Y.
Wong
.
2016
.
Translocation and dissemination of commensal bacteria in post-stroke infection.
Nat. Med.
22
:
1277
1284
.
73.
Crapser
J.
,
R.
Ritzel
,
R.
Verma
,
V. R.
Venna
,
F.
Liu
,
A.
Chauhan
,
E.
Koellhoffer
,
A.
Patel
,
A.
Ricker
,
K.
Maas
, et al
2016
.
Ischemic stroke induces gut permeability and enhances bacterial translocation leading to sepsis in aged mice.
Aging (Albany NY)
8
:
1049
1060
.
74.
Wen
S. W.
,
R.
Shim
,
L.
Ho
,
B. J.
Wanrooy
,
Y. N.
Srikhanta
,
K. P.
Kumar
,
A. J.
Nicholls
,
S.
Shen
,
T.
Sepehrizadeh
,
M.
de Veer
, et al
2019
.
Advanced age promotes colonic dysfunction and gut-derived lung infection after stroke.
Aging Cell
18
:
e12980
.

The authors have no financial conflicts of interest.

Supplementary data