Metabolic acidosis is thought to exacerbate chronic kidney disease in part by stimulating the release of potentially injurious substances. To define the genes whose expression is affected by exposure to an acidic milieus, we examined the effect of exposure of MDCK cells to pH 7.4 and pH 7.0 for 24 h on gene expression using a canine derived microarray. Exposure to this pH stress for 24 h led to increased expression of 278 genes (2.2% of the transcriptome) by at least 2-fold and 60 of these (21%) were upregulated by >3-fold. On the other hand, 186 genes (1.5% of the transcriptome) were downregulated by at least 2-fold and 16 of these (9%) were downregulated by 3-fold or more. Ten percent of the genes upregulated by at least threefold encode proinflammatory cytokine proteins, including colony stimulating factor 2, chemokine ligand 7, chemokine ligand 20, chemokine ligand 8, and interleukin-1α. Two others encode metallopeptidases. The most highly upregulated gene encodes a protein, lubricin, shown to be important in preventing cartilage damage and in tissue injury or repair. Upregulation of four genes was confirmed by quantitative PCR. Housekeeping genes were not increased. To examine the effect of decreasing medium pH, we measured intracellular pH (pHi) using 2,7-bis (2-carboxyethyl)5-carboxyfluorescein. With extracellular pH (pHo) of 7.0, pHi fell and remained depressed. These findings suggest that a pH stress alone can increase renal expression of proinflammatory and other genes that contribute to renal injury.
- metabolic acidosis
- MDCK cells
- intracellular pH
- proinflammatory cytokines
metabolic acidosis causes renal injury and thereby contributes to the progression of chronic kidney disease (CKD) (16). Studies in animals and humans have suggested that one mechanism by which metabolic acidosis might contribute to progression of CKD is stimulation of endothelin production and release by renal tubular cells (25).
We postulated that acid exposure might stimulate renal tubular cells to produce other substances that might be injurious per se or reduce the production of substances that might protect again renal injury. To examine this possibility, we assessed changes in gene expression of Madin-Darby canine kidney (MDCK) cells, an immortalized cell line derived from the dog distal renal tubule exposed to an acidic pH for a period of 4 or 24 h. This model was chosen to examine the direct effects of a reduction in pH on gene expression by renal tubular cells in the absence of other exogenous factors. Furthermore, to determine if changes in gene expression were due to alterations in intracellular and/or extracellular pH, the intracellular pH of the MDCK cells was monitored for several hours while bathed in an acidic media.
The results of the study indicate that exposure of MDCK cells to decreased pH (from pH 7.4 to pH 7.0) for 24 h causes an increase in expression of genes encoding proteins with diverse functions. Prominent among them were genes encoding proteins involved in inflammatory responses capable of inducing cellular injury. Since the medium was free of exogenous injurious factors, the exposure of the cells to an acidic milieu alone was sufficient to activate endogenous factors capable of contributing to cellular dysfunction and injury. These results suggest an inflammatory mechanism by which metabolic acidosis could result in renal injury and thereby contribute to progression of CKD. Moreover, since the increased expression of cytokines was observed after only 24 h, this indicates that these mechanisms potentially contributing to renal injury with metabolic acidosis are operative very early in the course of this acid-base disorder.
MATERIALS AND METHODS
Effect of Acidic pH on Gene Expression
The MDCK parental cell line NBL-2 was obtained from the American Type Culture Collection. These cells have been shown to be of distal tubule cell origin (9). Cells were grown to confluence in 5 ml of DMEM with FBS (pH 7.3 to 7.4) for 3–5 days in an aerobic environment of 20% O2-5% CO2-75% N2 at 37°C. Twenty-four hours before the experiment, culture media were switched to serum-free DMEM/Ham's F-12 pH 7.4.
On the day of study, cells were divided and placed either in media with a pH of 7.4 or one with a pH of 7.0 produced by reducing media bicarbonate concentration. They were then returned to the incubator for periods of 24 h. The pH of the media while residing in the incubator was confirmed by measuring pH with an in situ pH electrode.
Total RNA was isolated from the MDCK cells held at pH 7.4 or 7.0 for 24 h using TRIzol reagent (Invitrogen, Carlsbad, CA) combined with RNeasy columns (Qiagen, Valencia, CA) and the Qiagen eukaryotic RNA isolation kit. The cells were collected from the experimental flask and lysed in 1 ml of TRIzol reagent. To the extract the RNA from the lysed cells, 100 μl of chloroform were added per 500 ul of cell lysate. The cell extraction mixtures were repeatedly inverted in a microcentrifuge tube for 15 s, after which they were centrifuged at 17,900 g at 4°C for 10 min. The clear supernatant was mixed with 250 μl of ethanol and transferred to the RNAeasy Spin column, and RNA purification was performed following the manufacturer's instructions. The RNA concentration was quantified by absorbance at 260 nm, and the quality was evaluated by capillary electrophoresis using a model 2100 Bioanalyzer with a RNA 6000 Nano Assay kit (Agilent Technologies).
Microarray analysis was performed using the Canine V2Gene Expression MicroArray from Agilent (Santa Clara, CA). This chip contains 45,220 probes each 60 oligonucleotides in length representing 12,494 genes.
Fluorescent cRNA labeling.
All total RNA samples were serially diluted to a concentration of 1,000 ng in 5.3 ul for use with the Quick Amp Labeling kit. In preparation for the microarray, the Agilent One-Color Spike-In kit was used as positive controls. As per the manufacturer's protocol, known synthetic RNAs were added to the sample RNA isolated from cells exposed to pH 7.4 or 7.0 and amplified and labeled using the Quick Amp Labeling kit protocol (Agilent).
Following purification of the labeled cRNA, the sample was hybridized as recommended in the Agilent One-color quick amp protocol by using the Qiagen RNeasy spin columns.
To prepare the cRNA for hybridization, the Agilent fragmentation mix was used according to the Quick Amp Protocol. After a 30-min incubation period at 60° C, 55 ul of 2× GE Hybridization Buffer HI-RPM were added to each reaction tube. Samples were then vortexed and centrifuged for 1 min at 17,900 g to collect the sample at the bottom of the tube. Samples were then loaded onto the microarray slides and incubated in a hybridization chamber for 17 h at 65 ° C. The preparation was then washed with Agilent's Gene Expression Wash Buffer 1 and 2 (with Triton X) according to the protocol and scanned using the DNA Microarray Scanner 11.0.2.
Data were analyzed using Agilent's Feature Extraction Software as well as GeneSpring GX software. Data were filtered by expression (20–100%). All data with at least a twofold change compared with control were subjected to a t-test. Only those genes whose expression was altered by at least twofold with a P value <0.05 were considered to represent significantly regulated genes.
Gene annotations were obtained for the Agilent Canine Gene Expression Microarray from Agilent, NCBI: http//www.ncbi.nlm.nih.gov/gene, and Gene Ontology (GO) (http://www.geneontology.org/index.shtml). To identify enriched biological themes and further examine the putative functions of the genes whose expression were altered with acid exposure, we used the gene functional classification tool from the Database for Annotation, Visualization and Integrated Discovery (DAVID) website (http://david.abcc.ncifcrf.gov/home.jsp) (11,12). To identify significantly overrepresented GO terms in different categories, a modified Fisher's exact test was performed using the DAVID GO functional classification browser tool. P values <.05 were considered statistically significant.
Microarray Data Repository
The microarray data from this study are available at NCBI Gene Expression Omnibus Data Repository (GEO) under accession no. GSE36848.
Primers based on the sequence of 100- to 300-bp regions of selected upregulated genes seen in the microarray and several housekeeping genes were designed for real-time PCR analysis. Primer design was aided by the Primer 3 software available at website: http://frodo.wi.mit.edu/. Standard PCR was performed with the primers to ensure the production of only one product. The template for the real-time PCR reaction was the cDNA derived from the RNA isolated from cells exposed to pH 7.4 or 7.0 medium. A standard curve was produced using dilutions of genomic DNA as templates. A reaction mixture containing the master mixture with CYBR green label (Qiagen) and the primers was added to a 96-well plate, along with 1 μl of cDNA or genomic DNA template, for a final reaction volume of 50 μl per well. The negative control contained the reaction mixture but no DNA. Samples were run in a real-time PCR thermocycler (Applied BioSystems StepOnePlus) for 40 cycles (30 s at 95°C, 30 s at 57–60°C, and 40 s at 72°C). Data were collected during the extension step and expressed as arbitrary fluorescence units per cycle. A melting curve was run at the end to ensure that there was only one peak and only one product.
Measurement of intracellular pH.
The impact of a reduction of media pH on the intracellular pH (pHi) of the MDCK cells was examined using the pH-sensitive fluorescent dye 2,7-bis (2-carboxyethyl)5-carboxyfluorescein acetomethoxy ester (BCECF-AM). Briefly, cells were grown on glass coverslips for 3 to 5 days till confluent. On the day of study, the coverslip containing the cells was resuspended in the DMEM media at pH 7.4 and loaded with the dye by incubation with 10 uM BCECF-AM in the incubator for 30 min. Once inside the cells, the nonfluorescent ester is hydrolyzed producing intracellular BCECF that responds to pHi. Unincorporated dye was removed by washing with medium, and the coverslip was inserted into a specially modified cuvette containing medium at pH 7.4. The CO2 of the environment of the cells was maintained by constant gassing of the cuvette with 5% CO2. After a 10- to 15-min period, the solution in the cuvette was changed to one with a pH of 7.0 by lowering the HCO3− concentration The fluorescence of the cells was monitored with excitation at 436 (isobestic point) and 490 nm and emission at 530 nm to allow assessment of changes in pHi. The fluorescent signal was calibrated using the K-nigericin method. Briefly, acidic and basic endpoints were determined by titration with HCl and NaOH, respectively, in the presence of the ionophore Nigericin and KCl. The fluorescent signal was calibrated using the equation:
where the pKa of BCECF is 6.98, R is the ratio F(λ1)/F(λ2) of flourescent intensities (F) measured at two wavelengths λ1 (490 nm) and λ2 (436) and the subscripts A and B represent the limiting values at the acidic and basic endpoints following titration, respectively (7).
Data from six control and six acid-exposed microarrays were taken from three separate experiments each performed on a different day. Two-hundred seventy-eight genes (2.2%) were upregulated by 2-fold or greater and 60 of these upregulated genes (21%) were upregulated by >3-fold. On the other hand, 186 genes (1.5%) were downregulated 2-fold and 16 of these genes (9%) were downregulated by 3-fold or more.
Genes that were upregulated by threefold or greater and their purported functions are shown in Table 1. This and Tables 2, 3, and 4 contain only the identified genes that have an annotated function for the encoded protein. Two cytokines, colony stimulating factor 2 (CSF2) and chemokine ligand 7 (CCL7), were upregulated more than sevenfold. Three other cytokines, chemokine ligand 2 (CCL2), chemokine ligand 8 (CCL8), and interleukin-1α (IL-1α), were upregulated 3- to 3.5-fold. Thus 10% of the highly upregulated genes were proinflammatory cytokines. Other genes potentially contributing to renal injury included two genes encoding metalloproteinases, matrix metallopeptidase 9 (MMP-9) and MMP-13 (upregulated by >4.5-fold). Additionally, two cytokine genes and two metalloproteinase genes were upregulated at least twofold.
Two genes were downregulated by at least fivefold. One DDIT3 or CCAAT/enhancer binding protein (C/EBP) homologous protein (CHOP) is a member of the endoplasmic reticulum (ER) stress-mediated apoptosis pathway (6). Another, myosin light chain kinase, plays a role in producing inflammatory changes in the kidney and other organs (33). It can act as a scaffolding protein that tethers kinase proteins and phosphatase to distinct cellular compartments, thereby regulating their activity. GAS2, a gene encoding the growth arrest specific 2 protein was downregulated by more than threefold. This growth arrest specific gene is involved in apoptosis (34). Moreover, it inhibits cell division; therefore, suppression of expression could modulate the renal injury response (34).
Genes that were differentially expressed with acidosis by more than twofold (upregulated and downregulated) were subject to functional clustering analysis according to their GO classification with emphasis on biological process genes and molecular function genes as shown in Tables 3 and 4. Genes involved in the immune response, inflammatory response, chemotaxis, cytokine and chemokine synthesis, or metallopeptidase activity were highly represented in the upregulated genes. A lesser number of biological processes or molecular functions were downregulated by acid exposure. Prominent among the categories represented was the ER overload response, response to stress, and ATP binding.
Real-Time PCR (Quantitative PCR)
Real-time PCR [quantitative (q)PCR] was performed with a portion of the same total RNA used in the microarray experiments. For the qPCR, primers based on the sequence of four genes that were highly upregulated on the microarray including lubricin, CCL7, CSF2, and MMP-13; one downregulated gene, DNA damage inducible transcript 3; and three housekeeping genes, ACTA1, B2M, and GADPDH, were utilized. Three separate real-time PCR experiments were performed for each of the genes and the results were averaged.
Figure 1 shows the real-time PCR measurements for each of the upregulated genes studied, In general, the upregulation of each of the genes was similar to the values for microarray as shown in Table 1, although there were differences in the magnitude of these changes. Thus lubricin was upregulated ∼9.6-fold in the microarray and 14.1 with qPCR. Similarly, CCL7, CSF 2, and MMP-13 were upregulated 6.9 vs. 5.8, 8.6 vs. 7.3, and 4.6 vs. 10.9 with the microarray compared with qPCR. On the other hand DDIT3 was downregulated by approximately fivefold compared with sixfold with qPCR. The fold change in qPCR of the three housekeeping genes was 1.6 for ACTA1, 1.2 for B2M, and 1.3 for GADPH below the twofold change considered significant. Hence, qPCR validated the microarray observations.
Intracellular pH of MDCK Cells
A typical pH study is shown in Figure 2. The baseline intracellular pH of the MDCK cells when exposed to a pH of 7.4 was steady. Exposure of the cells to a pH of 7.0 resulted in a rapid fall in pHi. The pHi remained depressed as long as extracellular pH remained at 7.0. With return of medium pH to a pH of 7.4, however, pHi rose towards baseline. Thus continued exposure of cells to a lower HCO3− concentration results in a stable, albeit lower pHi.
The metabolic acidosis commonly accompanying CKD has been shown to produce renal injury and contributes to the progression of CKD (16). The mechanism(s) by which metabolic acidosis produces renal injury, however, remain unclear. Based on studies done in rats and humans, it has been proposed that the renal damage induced by metabolic acidosis is mediated, in part, by increased secretion of endothelin by renal tubular cells (32). Endothelin may enhance release of tissue inhibitors of metalloproteinases, inducing the release of cytokines that stimulate matrix accumulation (27). There is compelling data to support the role of endothelin in progression of CKD (16).
In vitro and in vivo studies have indicated that a reduction in pH, whether it be in the interstitial or intracellular compartment, activates various cellular pathways, which can cause cellular dysfunction and cellular damage (17). Therefore, we postulated that metabolic acidosis might alter the production of factors by renal tubular cells that could contribute to renal damage and further accelerate progression of renal disease.
To examine this possibility, we examined changes in gene expression of MDCK cells exposed to an acidic milieu produced by lowering extracellular HCO3− concentration for a period of 24 h. The number of genes altered when exposed to acidic medium of only 4 h was small, and therefore, only data for the 24-h exposure are reported. After 24 h, there was increased expression (>2-fold) of ∼2.2% of the genes present in the MDCK transcriptome, while 1.5% of the genes in the transcriptome were downregulated. Most importantly, 8 or 13% of the genes that were upregulated by more than threefold were those that might participate in the process leading to glomerular sclerosis and fibrosis and could contribute to renal injury.
Sixty genes were upregulated by 3-fold and 19 or 32% of these genes were upregulated by >4-fold. The gene with the greatest increase in expression (∼9-fold) was lubricin, also known as superficial zone protein or PRG4 (29). It is a mucinous glycoprotein, originally isolated from synovial fluid and thought to lubricate articular cartilage. However, it is also expressed in the kidney, lung, liver, heart, brain, muscle, testis, small intestine, bone, and cartilage (29). Additional biological functions that have been identified include prevention of cell adhesion and regulation of cell growth (26). The majority of experimental studies examining its biological role have been performed using synovial cells or chondrocytes, but its role in the kidney and other solid organs has not been ascertained. Nevertheless, lubricin is upregulated in models of CKD (8) and we speculate that it may play a cytoprotective role. Studies in cell injury models would shed more light on this conjecture.
Two other genes were upregulated almost to the same degree as lubricin. CSF2 was upregulated by almost ninefold, and CCL7 was upregulated by almost sevenfold. The protein encoded by the former gene is a cytokine that controls the production, differentiation, and function of granulocytes and macrophages (1). Several studies have demonstrated it plays an important role in producing renal injury after various insults. Thus ischemia-reperfusion increased expression of CSF2 by thick ascending limb epithelial cells both in vitro and in vivo (35) and increased expression levels of this protein facilitates the known inflammatory response (31). CCL7 gene encodes monocyte chemotactic protein 3, a secreted chemokine that attracts macrophages during inflammation and metastasis (15). The protein is also an in vivo substrate of MMP- 2, an enzyme that degrades components of the extracellular matrix (18). This protein has also been shown to play an important role in progression of squamous cell carcinoma by facilitating tissue invasion (14). Its increased production with metabolic acidosis could promote an inflammatory reaction.
N-myc downstream regulated gene 2 (NDRG2) was upregulated almost sixfold. This gene belongs to the NDRG family, which is comprised of four members, NDRG1–4, and is expressed in the tissues of the brain, heart, skeletal muscle, and kidney (3). NDRG2 is also involved in cell growth (30), differentiation (10), and neurodegeneration (22) and acts as a cell stress response molecule that is sensitive to oxygen deprivation (31). The expression of NDRG2 was significantly upregulated in tumor cell lines exposed to hypoxia and radiation (20), and it is also upregulated by aldosterone, a response that is observed within 15 min of exposure to the hormone (3). Although some have speculated that NDR2 mediates changes in sodium transport, it may also play a role in the renal injury occurring with aldosterone.
Two additional cytokines, CCL8 and IL-1α, were upregulated by 3.7- and 3.6-fold, respectively. CCL8 is a member of the monocyte chemoattractant family and has been implicated in graft vs. host disease and in allergic dermatitis (13). IL-1α is a proinflammatory cytokine that is increased in disorders such as rheumatoid arthritis. Single nucleotide polymorphisms for this gene have also been associated with increased risk for CKD (2).
Two metalloproteinase genes, MMP9 and MMP13, were upregulated by 4.8- and 4.60-fold, respectively. Metalloproteinase genes are upregulated in various experimental models of acute and CKD (18). By enhancing the degradation of the extracellular matrix, they may contribute to renal injury. In addition to the increased abundance of these proteins, their activity could be enhanced by acidosis, as a reduction in pH is an important activator. Hence, exposure of MDCK cells to a low medium pH increases expression of several proinflammatory genes, suggesting that inflammation is a response to acid stress by these cells and perhaps by other tubular cells of the kidney.
To further analyze the response of MDCK cells to an acid stress, all the genes whose expression was increased by twofold were subject to analysis using the GO classification system with particular emphasis on changes in genes involved in biological processes or molecular function. In the former category, genes involved in the immune response, inflammatory response, and chemotaxis were highly represented. In the latter category, genes involved in cytokine synthesis, chemokine synthesis, or metallopeptidase activity were highly represented. Taken as a whole, acid exposure elicited an increase in genes potentially involved in producing cellular damage to the kidney via inflammatory pathways.
The theme of genes downregulated by exposure to an acidic milieu is not as clear as that of the upregulated genes. Two genes were downregulated by at least fivefold. One DDIT3 or CHOP is a member of the ER stress-mediated apoptosis pathway (6). The gene encodes a transcription factor belonging to the C/EBP family (6). It is elevated upon DNA damage in cellular stress conditions. The DDIT3 protein has a central role in endoplasmic reticulum stress and DNA damage response by inducing cell cycle arrest and apoptosis (19, 21). DDIT3 has recently been implicated in the stress response leading to death of pancreatic insulin producing β-cells (19), and it may also be a part of cellular stress conditions causing neurodegenerative disorders (24). Thus downregulation of this protein would seem counterproductive, and so its role in the response to acidosis is unclear.
Another downregulated gene that encodes myosin light chain kinase plays a role in producing inflammatory changes in the kidney and other organs (33). Changes in its expression could modulate the inflammatory response to acidosis. Finally the gene encoding growth arrest specific protein 2 was downregulated by more than threefold. This protein suppresses cell division (34) and its downregulation could favor cell growth during tissue repair in response to injury caused by acidosis.
Functional analysis of the downregulated genes according to the GO classification revealed a lesser number of biological and molecular functional categories that were impacted. Of interest, genes involved in ATP binding, the ER stress response, and ER overload response were highly represented. The meaning of these changes is not obvious, but they could modulate the cellular damage arising from acid exposure.
Other investigators have examined changes in renal gene expression in animals with metabolic acidosis. Nowik et al. (23) examined gene expression using microarray analysis in whole kidneys and microdissected proximal tubule nephron segments taken from mice with metabolic acidosis produced by ammonium chloride feeding for 2 or 7 days. They found the most highly upregulated genes included solute carriers, acid transporters, and mitochondrial carriers. Proinflammatory cytokines and metalloproteinases were not reported as being highly upregulated. However, there were several differences in these studies compared with ours. Their studies were done in whole animals; therefore, any change in tubular gene expression produced by exposure to an acidic environment would be modified by changes in hormonal or other factors seen with metabolic acidosis. Also, either the whole kidney or proximal tubule segments were examined, whereas in our study we examined the MDCK cells that are derived from dog renal distal tubule. In addition, the duration of acid exposure in was substantially longer than used in the present study. Finally the severity of the acid exposure of tubular cells in our study is likely to be greater than in the study by Nowik et al. (23).
Similarly, Cheval et al. (5) examined changes in gene expression using Serial Analysis of Gene Expression (SAGE) analysis in outer medullary collecting ducts dissected from mice fed ammonium chloride for 3 days. They found increased expression of genes encoding proteins with diverse functions including solute and electrolyte transport but also oxidative stress, cell proliferation, and apoptosis. The increased expression of the latter group of genes could be consistent with a heightened renal stress response to the acid exposure. Again, as in the study by Nowik et al. (23), conditions were distinctly different from our study including use of whole animals, a longer duration of acid exposure, study of cells taken from different nephron segments, and less severe acid exposure. Given the variability in the gene changes reported in the different studies, further studies will be needed to determine what level of pH is necessary to stimulate proinflammatory genes and what factors might important in modulating the gene response of renal tubular cells to acid exposure.
Signal for Increased Gene Expression
A reduction in intracellular pH and/or extracellular pH could theoretically be the signal to increased expression of proinflammatory genes. Indeed, it has been shown that an increase in interstitial acidity activates receptors on collecting duct cells such as the proton activated G protein-coupled receptor GPR4 (28, 32). In the present study, the pHi fell and remained depressed as long as extracellular pH was low. Similar findings were demonstrated with neurons exposed to an acidic HCO3−/CO2 solution (4). Therefore, the MDCK cells in the present study were likely exposed to both a low extracellular and intracellular pH throughout the study, either one of which could be a signal for changes in expression of various genes.
The demonstration that an acidic milieu can contribute to renal injury by inducing the production of proteins involved in inducing an inflammatory response or accelerating the destruction of the renal matrix is of great importance. Metabolic acidosis is common in many clinical situations including acute kidney injury, CKD, and shock states (17). Recent studies in both animals and humans have confirmed that even mild metabolic acidosis can contribute to progression of CKD (16), although its role in renal damage with acute kidney injury or shock states has not been explored.
The results of the present study that demonstrate upregulation of genes encoding several inflammatory cytokines and metalloproteinases could suggest treatment with anti-inflammatory agents might abrogate the negative effects of these gene products providing ancillary therapy of patients with various kidney diseases.
In summary, exposure of renal tubular cells to an acidic milieu alone for a period of 24 h was sufficient to strikingly increase the expression of several proinflammatory cytokines and metalloproteinases as determined by microarray analysis. Furthermore, many of the genes upregulated were those affecting cytokine or chemokine production and the inflammatory response or immune response. These findings support a potential role of an inflammatory response to acid exposure in contributing to cellular damage to the kidney.
The signal for this effect could be a reduction in intracellular and/ or extracellular pH. The increase in cytokines is likely to recruit inflammatory cells to the kidney and contribute to the production of renal damage eventually resulting in glomerular sclerosis and renal fibrosis. Moreover, the data give further support for the potential benefits of treating patients with base geared to neutralize interstitial or intracellular acidity to slow the progression of CKD. They also provide a further rationale for administration of base in patients with acute metabolic acidosis and even anti-inflammatory drugs.
These studies were supported by funds from the Veterans Administration and a grant from the University of California, Los Angeles, Academic Senate. Brad Sherman is a Bioinformatics Analyst of the Laboratory of Immunopathogenesis and Bioinformatics at the National Institute of Allergy and Infectious Diseases and assisted in performance of the GO analysis.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: S.R., D.R.S., T.N., G.S., and J.A.K. conception and design of research; S.R., D.R.S., and T.N. performed experiments; S.R., D.R.S., T.N., G.S., and J.A.K. analyzed data; S.R., T.N., G.S., and J.A.K. edited and revised manuscript; D.R.S., T.N., G.S., and J.A.K. interpreted results of experiments; D.R.S. and J.A.K. drafted manuscript; T.N. and J.A.K. prepared figures; T.N., G.S., and J.A.K. approved final version of manuscript.