Oral NaHCO₃ Activates a Splenic Anti-Inflammatory Pathway: Evidence That Cholinergic Signals Are Transmitted via Mesothelial Cells

Chronic inflammation has been implicated in both acute and chronic kidney injury (1). The Chronic Renal Insufficiency Cohort study found that elevated inflammatory markers fibrinogen and TNF-α were associated with rapid loss of kidney function in patients with chronic kidney disease (CKD) (2). Furthermore, treatment with TNF-α antagonists has been associated with an attenuation in renal functional decline in CKD patients (3). Activation of the innate cholinergic anti-inflammatory pathway via stimulation of the vagal nerve, which suppresses proinflammatory cytokines and promotes anti-inflammatory macrophage cell polarization via activation of α-7-containing nicotinic receptors on splenic macrophages (4), has also been reported to ameliorate acute kidney injury (5).

Evidence from a number of small clinical trials as well as experimental models indicates that supplementation with oral sodium bicarbonate (NaHCO₃) may slow the decline in kidney function in CKD patients (6), yet the physiological mechanisms mediating this beneficial effect remain unclear. As inflammation has been associated with CKD progression, we speculated that NaHCO₃ may act to protect the kidneys by reducing inflammation. Therefore, we tested the hypothesis that “Oral NaHCO₃ intake promotes M2 macrophage polarization by activating splenic anti-inflammatory pathways.”

In the current study, we used flow cytometry as well as mRNA markers in isolated splenic macrophages to determine whether oral NaHCO₃ intake promotes M2 macrophage polarization in the kidney and spleen in both hypertensive Dahl salt-sensitive (SS) rats, in which significant inflammation is known to be present (7), as well as normotensive Sprague Dawley rats, in which baseline renal inflammation has been reported to be low. We also investigated the effect of acute oral NaHCO₃ loading on inflammatory cell profiles in the blood of healthy human subjects. Furthermore, as we found that gentle manipulation to visualize the spleen at midline during surgical laparotomy (sham splenectomy) was sufficient to abolish the anti-inflammatory response to oral NaHCO₃, we investigated the pathways through which signaling of NaHCO₃ intake may be transmitted to the splenic parenchyma.

Studies used 8–12-wk-old male Dahl SS or Sprague Dawley rats (Charles River Laboratories, Wilmington, MA). Rats were maintained ad libitum on water and a pellet diet containing low 0.4% NaCl (AIN76A; Dyets, Bethlehem, PA; low salt [LS] 0.4% NaCl). Rats were age matched for all protocols. All studies were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All of the protocols were approved in advance by the institutional animal care committee at Augusta University.

Rats were anesthetized with isoflurane (2–5%), and a midline incision was performed. Using a stereoscope, the vagal nerves were visualized immediately below the diaphragm and transected. Any nervous tissue around the esophagus was also cleared by dissection. When visualizing the esophagus, care was taken to limit any horizontal movement of the stomach and to avoid movement of the spleen. After wound closure, animals were allowed to recover for 2 wk before tissue was harvested under isoflurane anesthesia. Bloating of the stomach was used to confirm sub diaphragmatic transection of the vagal nerves at the time of sacrifice.

Dahl SS rats were anesthetized with isoflurane (2–5%), and a midline incision was performed. The spleen was located and gently moved toward the incision site by hand or with cotton tip applicators. The poles of the spleen were visualized and the spleen returned to its original position. Following the surgical procedure, rats were allowed to recover for 7 d before entering the high-salt (HS) protocol described above.

Esomeprazole (20 mg/kg; Sigma, St Louis, MO) was dissolved in DMSO before being diluted in saline and given as a final volume of 0.2 ml i.p. once daily. Methyllycaconitine (MLA: 5 mg/kg; Sigma) was dissolved in saline and given i.p. (0.2 ml twice a day). Control rats received DMSO and saline as vehicle (0.2 ml i.p. twice a day). All rats in drug treatment protocols received either 0.1 M NaHCO₃ or NaCl in drinking water and were treated daily for 3 d prior to tissue harvest.

With the exception of one group of rats (n = 5 + 5), in all HS-fed Dahl SS rats, blood pressure telemetry devices were implanted prior to beginning the study. All protocols were initiated at least 7 d after surgery to allow time for rats to recover, and no significant differences in flow cytometry results were observed between rats that had telemeters implanted compared with rats without prior surgery. No surgical procedures were performed on LS-fed Dahl SS rats or Sprague Dawley rats prior to tissue harvest. On the day of tissue harvest, rats were anesthetized with isoflurane (2–5%), and a midline incision was performed. The abdominal aorta was catheterized for the collection of ∼2 ml of arterial blood in a heparinized tube, and the left kidney was excised and immediately processed for flow cytometry. Ab markers used for flow cytometry as well as dilutions and references are given in Table I.

To identify and evaluate immune cells in renal and splenic tissues, we employed a flow cytometry-based assay. Briefly, kidneys were harvested and placed in RPMI 1640 (Thermo Fisher Laboratories) + 10% FBS (Atlanta Biologicals, Lawrenceville, GA), minced, and single-cell suspensions were achieved using a 100 μM cell strainer (BD Biosciences, San Diego, CA) followed by centrifugation (1400 rpm, 5 min) and lysis of erythrocytes by incubation with ACK Lysing Buffer (3 min at room temp; Quality Biological, Gathersburg, MD). Spleens were harvested and placed in RPMI 1640 + 10% FBS. Spleens were then injected with 1 ml of 100 collagen digestion units/ml Collagenase IV (Sigma) solution in three regions, then placed in 1 ml of 400 collagen digestion units/ml Collagenase IV, incubated at 37°C for 30 min, and minced. Single-cell suspensions were then achieved as described for the kidney, and all cells were washed twice (PBS).

In all flow studies, cells were then incubated with Abs for surface markers for 20 min on ice in the dark (all Abs from Pharmingen-BD-Biosciences, San Jose, CA). Cells were then washed with PBS, fixed and permeabilized using FOXP3/Transcription Factor Fix/Perm Buffer (eBioscience, San Diego, CA) for 15 min in the dark on ice, and washed once again with PBS before incubation with Abs for intracellular staining of Foxp3, TNF-α, IL-10, and IL-17, on ice in the dark for 20 min (BD Biosciences). Cells were then washed and run through a four-color flow cytometer (FACS Calibur; Becton-Dickinson), and data were collected using CellQuest software.

In all flow studies, group identifiers were removed and flow cytometry analysis was performed by an investigator unaware of the hypothesis. Samples were double stained with control IgG and cell markers and were used to assess any spillover signal of fluorochromes; proper compensation was set to ensure the median fluorescence intensities of negative and positive cells were identical and were both gated populations. Gating was used to exclude dead cells and debris using forward and side scatterplots. In each analysis, 100,000 total events were collected. As a gating strategy, for each sample, isotype-matched controls were analyzed to set the appropriate gates. For each marker, samples were analyzed in duplicate measurements. To minimize false-positive events, the number of double-positive events detected with the isotype controls was subtracted from the number of double-positive cells stained with corresponding Abs (not isotype control), respectively. Cells expressing a specific marker were reported as a percentage of the number of gated events. A representative example of gating is shown in Fig. 1 and Supplemental Fig. 2.

Markers for M1 and M2 macrophages were selected based on previous studies in the rat (8, 9); however, to confirm the specificity of these markers, in a subset of animals, these markers were compared with alternative markers (Supplemental Fig. 2) and depletion of kidney cells identified as macrophages was confirmed using clodronate liposomes (Supplemental Fig. 3). In these studies, a single dose of clodronate liposomes or liposome controls (20 mg/kg; Encapsula Nanosciences, Nashville, TN) was injected into the tail vein of rats 24 and 72 h prior to harvesting the kidneys for flow cytometry. In addition, rat spleen macrophages were isolated from Dahl SS rats treated with either NaCl or NaHCO₃ for 2 wk and fed an HS (8%) diet. Splenic macrophages were isolated by centrifugation in a Percoll density gradient and real-time PCR was performed for rat iNOS, arginase, and GAPDH to determine relative M1/M2 macrophage polarization.

The MLR is a commonly used functional assay to determine the proliferative capacity of T lymphocytes in response to Ag presentation. This assay requires the use of responder cells (T cells from the spleen of untreated Dahl SS rats) and stimulating cells (splenocytes) from Sprague Dawley rats drinking either NaCl (0.1 M) or NaHCO₃ (0.1 M) for 3 d prior to tissue harvest (n = 3 animals each group). The responder cells and stimulator cells were set up in triplicate wells in a RPMI 1640 medium, which was supplemented with FBS, penicillin, streptomycin, l-glutamine, and 2-ME. Responder T cells were enriched using magnetic assorted cell sorting and used at the ratio of 1:5 (responders to stimulators). After 96 h of incubation at 37°C in a humidified 5% CO₂ environment, all cells were harvested into flow cytometry tubes. After one wash with PBS, all samples were then incubated with anti- CD71 Ab (a marker for activated and dividing T cells) for 20 min in the dark on ice. Samples were then washed with PBS, and T-cell proliferation was measured by using flow cytometry analysis for CD71 expression. The average of the triplicate samples was recorded for each rat, and the average of all animals in each group then calculated and compared by unpaired t test.

Real-time PCR was performed using rat iNOS, arginase, and GAPDH primers listed below, as previously reported (9): 1) rat iNOS forward 5′-TCT TGG TGA AAG CGG TGT T-3′; 2) rat iNOS reverse 5′-TGT TGC GTT GGA AGT GTA GC-3′; 3) arginase forward 5′-TAT CTG CCA AGG ACA TCG TG-3′; and 4) arginase reverse 5′-GTT CTG TTC GGT TTG CTG T-3′.

For urine collection, rats were placed in rat metabolic cages for 24 h (Nalgene, Rochester, NY). Urine was collected and weighed for volume determination. Up to 10 ml of urine was stored at −80°C for later analysis. All urinary data are presented as 24 h urinary excretion. Urinary Na, K, and Cl measurements were obtained using an electrolyte analyzer (Easylite; Medica, Bedford, MA). Samples were diluted 1:10, one part urine to nine parts Easylyte Urine Diluent as per the manufactures instructions. Urinary titratable acids were determined with titration of 5 ml of urine with NaOH or HCl to pH 7.4. Urinary NH₄⁺ concentration was determined using an ammonia ion selective electrode (Orion high performance ammonia ion selective electrode [Thermo Fisher Scientific]). Urine samples were diluted in Orion ionplus Solution Alkaline Reagent immediately before measurement as per the manufacturer’s instructions. Measurements were compared with those of a standard curve using serial dilutions of the Thermo Scientific Orion Application Solution 0.1 M NH₄⁺ Ammonium Standard. Final concentrations were calculated using a log curve (Graphpad Prism 6; Graphpad Software, La Jolla, CA).

Preparation and imaging of tissue by electron microscopy was performed by the Augusta University Histology Core facility. Tissue was fixed in 4% paraformaldehyde, 2% glutaraldehyde in 0.1 M sodium cacodylate (NaCac) buffer, pH 7.4, postfixed in 2% osmium tetroxide in NaCac, stained en bloc with 2% uranyl acetate, dehydrated with a graded ethanol series, and embedded in Epon-Araldite resin. Thin sections of 75 nm thickness were cut with a diamond knife on a Leica EM UC6 ultramicrotome (Leica Microsystems, Bannockburn, IL), collected on copper grids, and stained with uranyl acetate and lead citrate. Tissue was observed in a JEM 1230 transmission electron microscope (JEOL USA, Peabody, MA) at 110 kV and imaged with an UltraScan 4000 CCD camera and First Light Digital Camera Controller (Gatan, Pleasantonm, CA).

At the end of the study, rats were anesthetized with isoflurane (2–5%), and tissues were excised and placed in 10% formalin solution (Sigma) for 48 h before being paraffin embedded, blocked, and processed (Augusta University Core facility). Kidneys were paraffin embedded in an automatic tissue processor, and 3-μm cut sections were mounted on siliconized and/or charged slides. The slides were deparaffinized and hydrated, and Ag retrieval was performed using IHC-Tek Epitope retrieval solution at steaming for 40 min (catalog no. IW-1100; IHCWorld). Tissue was blocked with 3% hydrogen peroxide in methanol for 10 min. The primary Abs used are listed in the Table I. The goat anti-rabbit IgG-HRP–conjugated secondary Ab (catalog no. sc-2004, 400 μg/ml, 1:400 at 1 μg/ml; Santa-cruz) was used for 30 min at room temperature (RT). The slides were stained with Betazoid DAB Chromogen Kit (catalog no. BDB2004H; Biocare Medical). Omitting the first Ab served as a negative control and resulted in no positive staining in tissue. All end-point analysis was blinded to the investigators. For histological scoring, all identifiers were removed, and slides were given a number before being scored by an investigator that was unaware of the hypothesis being tested. Data were then compiled by the primary investigator who had access to the numbering key.

Untreated rats were anesthetized with isoflurane (2–5%), and the spleen was excised. The spleen was placed in HBSS with 20 mM HEPES buffer pH 7.40 at RT with Fluo4. Imaging acquisition was conducted using the Andor Revolution system (Andor Technology Belfast, U.K.) (10). A Nikon microscope (Eclipse FN 1; Nikon, Tokyo, Japan) was connected to a laser confocal spinning unit (CSU-X1; Yokogawa, Tokyo, Japan) attached to a Sutter filter wheel and an ultrasensitive EMCCD camera (iXon^EM; Andor Technology, Belfast, U.K.). The microscope chamber was continuously perfused, at a rate of 2 ∼3 ml/min, with HBSS pH 7.4 using a peristaltic pump (Miniplus 3; Gilson, Middleton, WI). Chamber temperature was maintained at 36 ± 1°C using a single line solution heater (SH-28B; Warner Instruments, Hamden, CT) connected to a direct current power supply (1735A; BK Precision, Yorba Linda, CA). Calcium imaging experiments were monitored from thick spleen slices incubated at RT in HBSS containing 5 μM Fluo-4 AM and pluronic acid (2.5 μg/ml). Following an hour incubation period, slices were placed in RT HBSS until needed. Fluorescence images were obtained using a krypton/argon laser (488 nm excitation and >495 nm emission). Images were acquired at ∼0.7 frames/s. 40× or 20× Nikon dipping objectives were used to visualize the spleen surface during electrical stimulation.

All rats were maintained on LS chow with ad libitum water (tap water). Following 7 d recovery from surgery, on day 1 of the HS diet protocol, tap water was replaced with either 0.1 M NaHCO₃ (n = 11; Sigma) or equimolar NaCl (0.1 M) made fresh daily, also ad libitum (n = 10). NaHCO₃ or vehicle-treated water was then maintained for the remainder of the protocol. Following 4 d of LS feeding, rats were placed on an 8% HS diet (AIN76A 8%; Dyets) for 2 wk. Twenty-four-hour urine collections were obtained on day 3 of LS feeding and day 7 and day 14 of HS feeding, if applicable. Following 14 d of HS feeding, rats were anesthetized with isoflurane (2–5%), and tissue was harvested for analysis.

Dahl SS rats on the LS protocol were treated identically to the HS protocol above except following 4 d of LS, rats were anesthetized and tissue was harvested for flow cytometric analysis prior to beginning an HS diet. n = 5/5 for NaCl and NaHCO₃ treatment, respectively.

Subdiaphragmatic transection of the vagal nerves was as follows: 8-wk-old Sprague Dawley rats were either subject to laparotomy only, laparotomy + manipulation of the spleen to midline, or laparotomy + transection of the vagal nerves immediately below the diaphragm or laparotomy + transection of the vagal nerves immediately below the diaphragm + manipulation of the spleen to midline. Rats were allowed to recover for 14 d before being anesthetized with isoflurane (2–5%) and tissue harvested. Successful transection of the vagal nerves was confirmed by stomach bloating on sacrifice, and animals without confirmation were excluded from analysis.

Dahl SS rats in which the spleen was removed (n = 5/7 for vehicle and NaHCO₃ treatment, respectively) or moved to midline during surgery (sham splenectomy [n = 5/5 for vehicle and NaHCO₃ treatment, respectively]) were entered into the HS protocol described above prior to tissue sacrifice. An additional sham control (CON) group (laparotomy only) was performed to control for movement of the spleen (n = 4/5 for vehicle and NaHCO₃ treatment, respectively). Tissue collected from animals in this group was used to compare Ab sets for identification of macrophage polarization (Supplemental Fig. 2).

Rats were maintained on standard laboratory chow (Teklad) and placed on either 0.1 M NaHCO₃, 0.05 M NaCl/0.05 M NaHCO₃, 0.09 M NaCl/0.01 NaHCO₃, or 0.1 M NaCl for 4 d prior to tissue harvest (n = 3 animals were used at each dose).

To examine the effects of NaHCO₃ on acute changes in parasympathetic activity, 12 healthy participants (six men, six women, age 27 ± 2 y, body mass index [BMI] 25.3 ± 1.2 kg/m²) were provided 2 g of NaHCO₃ dissolved in 250 ml of bottled water (treatment [TXT] group). An additional six participants (four men, two women, age 25 ± 1, BMI 25.7 ± 2.1 kg/m²) were recruited as controls and were provided 1.39 g of NaCl (equivalent molar load to 2 g of NaHCO₃) dissolved in 250 ml of bottled water (CON group).

Blood samples were collected via an i.v. catheter (Nexiva; Becton Dickinson, Franklin Lakes, NJ) at baseline and at 60 min intervals posttreatment to examine changes in serum electrolyte balance (Na, K, and Cl⁻).

In the NaHCO₃ TXT group, 10 of 12 subjects had blood drawn at 3 h posttreatment. Blood was taken at all time points for all control subjects. No data were excluded from the analysis. Flow cytometric analysis of heparinized whole blood was performed as described previously (11–13). Briefly, cells were incubated with Abs for surface markers (15 min on ice in dark) before incubation with Abs against intracellular cytokines and factors (after permeabilization for 15 min using fix/Perm mixture; eBioscience), including CD11b, CD68, TNF-α (for M1 macrophages); CD11b, CD68, CD206 and IL-10 (for M2 macrophages) (purchased from BD BioSciences); and CD16 and TNF-α (for neutrophils; from eBioscience). Cells were then washed and run through a four-color flow cytometer (FACS Calibur; BD Biosciences), and data were collected using CellQuest software. Samples were double-stained with control IgG and cell markers to assess any spillover signal of fluorochromes. Proper compensation was set to ensure the median fluorescence intensities of negative and positive cells were identical and then was used to gate the population. Gating excluded dead cells and debris using forward and side scatter plots. To confirm the specificity of primary Ab binding and rule out nonspecific Fc receptor binding to cells or other cellular protein interactions, negative control experiments were conducted using isotype controls matched to each primary Ab’s host species, isotype, and conjugation format. The control Abs had no specificity for target cells within our studies yet retain all the nonspecific characteristics of the Abs used in the experiments.

Data were analyzed using Graphpad Prism (Graphpad) software. All data are expressed as mean ± SE. Multiple comparisons were analyzed using two-way ANOVA. All other parametric comparisons were analyzed via unpaired two-sided Student t test. Categorical data were analyzed by Fisher exact test. Significance was considered to be p < 0.05. All end point analysis was blinded to the investigators at the time of measurement and compiled using an identifier key once analysis was complete. No data were excluded following end point analysis (some animals were excluded prior to analysis that met exclusion criteria, i.e., failure to confirm vagal denervation). Flow cytometry gating and analysis was performed by an investigator who was unaware of the hypothesis being tested. Rats were received from a commercial source (two to three rats per cage) and were randomly allocated to either vehicle or treatment, taking care to allocate at least one animal per cage to each group.

The study followed the principles of the Declaration of Helsinki. All study protocols were approved by the Human Assurance Committee at Augusta University (no. 1029130), and written and verbal informed consent was received from all participants prior to inclusion in the study.

Urine electrolytes were measured in HS-fed (8% NaCl chow) Dahl SS rats. The addition of NaHCO₃ (0.1 M) to the drinking water of Dahl SS rats had no effect on urinary Na or K excretion when compared with vehicle (0.1 M NaCl), indicating similar Na and K intake across the course of the study. Urinary Cl excretion was significantly lower (p = 0.03) in rats drinking NaHCO₃ solution, reflecting that a large portion of Na intake in these animals was from NaHCO₃.

Immune cell profiles were identified using Ab markers listed in Table I. Flow cytometry data from blood, spleen, and whole kidney of rats treated with either vehicle or NaHCO₃ after 2 wk of HS are presented in Table II. NaHCO₃ treatment resulted in a significant (p < 0.01) decrease in TNF-α–expressing macrophages (M1-polarized macrophages) and an increase in IL-10–expressing macrophages (M2-polarized macrophages) in the kidney (Fig. 1, Table II; p < 0.04). A significant (p = 0.005) polarization from predominately M1 to M2 macrophage polarization was also observed in splenic tissue of HS-fed Dahl SS rats treated with NaHCO₃ (Fig. 1, Table II). The percentage of T cells identified as FOX3⁺CD4⁺ T cells in the spleen, kidney, and blood of NaHCO₃-treated rats was also significantly increased when compared with NaCl-treated rats (Table II).

As the Dahl SS rat model develops hypertension and renal injury when fed an HS diet (14), to determine whether NaHCO₃ promotes anti-inflammatory responses independent of these changes, we examined tissues from LS-fed Dahl SS rats and normotensive Sprague Dawley rats. Importantly, in LS-fed Dahl SS rats, a significant (p = 0.007) polarization from predominately M1 to M2 macrophage polarization was still observed in kidney tissue (Fig. 2). Furthermore, we found that the effect of NaHCO₃ intake to promote M2 polarization in the kidney was also observed in outbred Sprague Dawley rats (Fig. 2). The effect of NaHCO₃ was found to be dose dependent, with changes in macrophage polarization identified with as little as 0.01 M NaHCO₃ in the drinking water following only 4 d of NaHCO₃ drinking in rats eating LS laboratory chow (Fig. 2).

To determine whether oral NaHCO₃ had a similar anti-inflammatory action in humans as we found in rats, we evaluated blood samples at baseline and 1, 2, and 3 h following ingestion of a single dose (2 g) of NaHCO₃ (n = 11) or equimolar NaCl (n = 6), each dissolved in 250 ml of bottled water. Pre- and posttreatment values of serum electrolytes are presented in Table III. There was a significant group by time interaction for changes in serum potassium (p = 0.029, η²_P = 0.279). Specifically, serum potassium decreased with NaHCO₃ treatment (p = 0.008), but there was no change with NaCl treatment (p = 0.381). BMI and C-reactive protein levels were not significantly different at baseline between either group, indicating a similar baseline inflammatory state (Table IV). No other significant differences were observed between TXT groups at baseline in any variables tested (Table IV).

Baseline flow cytometry values of all subjects, before ingesting NaHCO₃ or NaCl in solution, are presented in Table IV. Prior to any treatment, the percentages of blood leukocytes that were TNF-α⁺ neutrophils, M1 macrophages, or M2 macrophages were all significantly higher in the NaHCO₃ TXT group when compared with baseline values obtained in the NaCl TXT group (Table IV). There was a significant TREATMENT × TIME effect on both M1 macrophages (p = 0.0004) and TNF-α–positive neutrophils (p = 0.0146), with the levels of these inflammatory cells in the plasma being reduced to a significantly greater degree following ingestion of NaHCO₃ when compared with NaCl (Fig. 3). The greatest decreases in blood inflammatory cells were observed at 2 and 3 h following NaHCO₃ ingestion. Similar to our observations in rats, oral NaHCO₃ ingestion increased the percentage of blood leukocytes identified by flow cytometry as M2 macrophages (p = 0.00165) (Fig. 3). Decreases in inflammatory TNF-α⁺ neutrophils and M1 macrophages in the NaHCO₃ TXT group did not appear to be related to the differing baseline levels observed between TXT groups. When comparing individual responses between subjects of different groups, subjects with similar baseline levels of blood leukocytes responded differently if they received NaHCO₃ compared with NaCl (Supplemental Fig. 1).

In the current study, we found that, prior to beginning NaHCO₃ or vehicle treatment, either complete removal of the spleen or simple manipulation of the spleen to midline during sterile surgical laparotomy completely abolished the effect of NaHCO₃ to promote M1 to M2 polarization in the kidney of Dahl SS rats fed an HS diet for 2 wk (Fig. 4). Furthermore, both of these maneuvers resulted in a significant decrease in renal M2 macrophages when compared with sham laparotomy only (p = 0.02 and 0.0002, comparing laparotomy only to sham splenectomy and splenectomy for vehicle- and bicarbonate-treated groups, respectively; Fig. 4). We confirmed a functional anti-inflammatory response using the MLR. Proliferation of responder T cells isolated from the spleen of Dahl SS rats, in response to stimulation with splenocytes from NaHCO₃-treated rats, was significantly lower than that of control animals drinking NaCl (n = 3 rats each group, p = 0.00003; Fig. 5). In contrast to our findings in rats, in C57Blk6 mice, 3 d of oral NaHCO₃ (0.1 M in drinking water) increased splenic M1 macrophages and had no effect on M2 macrophages (data not shown). Further confirming the involvement of cholinergic signaling, MLA, a potent and specific inhibitor of the α7 nicotinic Ach receptor, prevented anti-inflammatory macrophage polarization to oral NaHCO₃ (Fig. 6). Similarly, the gastric proton pump inhibitor esomeprazole also inhibited the anti-inflammatory response to NaHCO₃ (Fig. 6).

Markers for M1 (Fig. 4; CD11b/80/TNF-α) and M2 (Fig. 4; CD11b/80/CD206/IL-10) macrophages were selected based on previous studies on the rat (8, 9); however, to confirm the specificity of these markers, these Abs were compared with alternative markers (CD68/CD163/CD206/TNF-α: for M1) and (CD68/CD163/CD206/IL-10: for M2). Results using both Ab markers sets were highly consistent and confirmed our observations of M1 to M2 polarization in the spleen and kidney of NaHCO₃-treated rats as well as the effect of splenectomy to abolish this response (Supplemental Fig. 2). In addition, we confirmed the identity of renal macrophages by specific depletion with clodronate liposomes (Supplemental Fig. 3).

As we found that light manipulation to visualize the spleen at midline during surgical laparotomy (sham splenectomy) was sufficient to abolish the anti-inflammatory response to oral NaHCO₃, we investigated how this maneuver could disrupt signaling of this response to the splenic parenchyma. Although, outside of the splenic hilum, only light connective tissue was observed to be attached to the spleen in rodents, we found that gentle splenic manipulation during surgery resulted in marked fibrosis of the splenic capsule and hypertrophy or hyperplasia of the capsular mesothelial cell layer following 2+ wk of recovery in both Sprague Dawley and Dahl SS rats (Fig. 7A). In histological sections, numerous fragile connections from the surrounding fascia were found to directly connect to the mesothelial layer along the inferior edge of the spleen (Fig. 6A). Prior to manipulation, both these connections as well as capsular mesothelial cells nearest to these junctions stained strongly positive for the pan-neuronal marker PGP9.5 as well as acetylcholine esterase, indicating cholinergic neural function (Fig. 8C, 8D, 8F, 8I). Although PGP9.5 staining within the capsular mesothelial cells was maintained following manipulation of the spleen, acetylcholine esterase staining within these now hypertrophied mesothelial cells was markedly reduced (Fig. 7D, 7E). Both acetylcholine esterase staining and tyrosine hydroxylase staining within the splenic parenchyma were unaffected by manipulation of the spleen (Supplemental Fig. 4). PGP9.5 and acetylcholine esterase staining was localized primarily to mesothelial cells proximal to the inferior edge of the spleen where these thin connections were present (Fig. 8G, 8I). Mesothelial cells along the superior edge of the spleen most often stained negative for these markers.

As the cholinergic anti-inflammatory pathway can be activated by vagal stimulation, and transection of the vagal nerves abolishes this response, we investigated whether vagal signaling contributed to acetylcholine esterase staining in splenic capsular mesothelial cells. Although PGP9.5 staining in mesothelial cells did not appear to be altered with vagotomy, acetylcholine esterase staining within mesothelial cell bodies on the splenic capsule was abolished with prior vagotomy, with five of five surgical control animals being found positive for mesothelial acetylcholine esterase and four out of four animals in which complete subdiaphragmatic vagotomy was confirmed, being found negative for mesothelial acetylcholine esterase staining by an observer unaware of the origin of the sections (Fig. 7F–I; p = 0.008).

To further investigate this phenomenon, we examined these fragile mesothelial connections to the splenic mesothelium with electron microscopy. We found that the connections contain collagen fibers lined by mesothelium on both sides (identified by their numerous microvilli), but we found no evidence of any neural tissue. Importantly, capsular mesothelial cells primarily in and around the junction of these connections were identified to contain many ultrastructural elements within their cytoplasm, which visually resembled neuronal structures. The image in Fig. 9G contains more than 15 of these pairings within the cytoplasm of the mesothelial cells pictured. These included low-density cylindrical structures that often contained a single mitochondrion, similar to the structure observed in dendrites (Fig. 9A, 9B). These dendritic structures were often adjacent to higher-density cylindrical structures, separated by an electron-dense band, similar to the arrangement of axons and dendrites in neural tissue separated by a synaptic density (Fig. 9A, 9B). No vesicular structures were identified within these “axonal” organelles. Perhaps the most remarkable structures identified were dark bands within the axonal structures that resembled synaptic ribbons (Fig. 9C, 9E), as these structures are normally only observed in nerves with rapid firing rates such as the eye and vestibular organs. To attempt to confirm the identity of these structures, we performed immunohistochemical labeling of paraffin-embedded sections of the spleen with an Ab targeted against synaptic ribbons. Immunohistochemical labeling of paraffin-embedded sections with anti-Ribeye Ab revealed positive staining of the parenchyma of the spleen. Importantly, within the capsular region, this positive staining was limited to mesothelial cells within these connections and immediately adjacent to areas where these connections made contact with the capsular surface (Fig. 9H, 9I).

While investigating the anatomical relationships of mesothelial cells on the splenic surface, using the Ca²⁺-sensitive dye Fluo-4, we observed a dense plexus of nerves immediately below the collagen layer (Fig. 10G) of the splenic capsule. Activation of these nerves could be promoted by electrical field stimulation (Supplemental Video). Immunohistochemical analysis of transverse sections through the splenic capsule revealed that the majority of these nerves stained only lightly positive for PGP9.5 and acetylcholine esterase but were negative for tyrosine hydroxylase (Fig. 10A–F). These structures were confirmed as neurons by high-power electron microscopy imaging (Fig. 10H, 10I).

Two weeks following transection of the vagal nerves below the diaphragm in Sprague Dawley rats drinking NaHCO₃, splenic weight was reduced in vagotomized animals (0.92 ± 0.05 g) when compared with surgical control animals (1.29 ± 0.04 g; p < 0.0001), with spleens accounting for 0.28 ± 0.01% and 0.35 ± 0.01% (p < 0.01) of total body weight in vagal denervated and control animals, respectively. In rats in which these connections had been disrupted, vagotomy had no further effect to reduce spleen weight, indicating that both maneuvers likely cause regression of the splenic parenchyma through a common pathway (Fig. 11).

The first major finding of our study is that oral NaHCO₃ promotes a powerful anti-inflammatory response including M1 to M2 macrophage polarization and increased FOXP3⁺ CD4⁺ T regulatory cells (Tregs) within the spleen. We confirmed that these phenotypic changes were associated with functional anti-inflammatory effects using the MLR. As the Dahl SS rat model develops hypertension and renal injury when fed an HS diet, to determine whether NaHCO₃ promotes an anti-inflammatory phenotype independent of these changes, we examined tissues from LS-fed Dahl rats and normotensive Sprague Dawley rats. Our data indicate that the effect of NaHCO₃ on macrophage polarization was independent of salt diet or elevated blood pressure. In agreement with our findings in rats, oral NaHCO₃ promoted a robust anti-inflammatory response in the blood of subjects (Fig. 3), which was evident in as little as 1 h after ingestion. To our knowledge, these data are the first demonstration that orally ingested NaHCO₃ can promote a powerful anti-inflammatory response in both rats and humans.

We speculate that the anti-inflammatory effects of oral NaHCO₃ ingestion are mediated by activation of the cholinergic anti-inflammatory pathway. The cholinergic anti-inflammatory pathway has been reported to be the efferent arm of the anti-inflammatory reflex (15), which acts via vagal efferents to promote M2 macrophage polarization in the spleen and limit activation of the innate immune system, thereby preventing damage caused by excessive cytokine production (4, 16). Inflammatory macrophages and excessive TNF-α production have been implicated in the pathology of a broad range of disease states, including rheumatoid arthritis (17), cardiovascular disease (18), atherosclerosis (19, 20), irritable bowel disease (21), type 2 diabetes (22), and neurodegenerative diseases as well as others (23–26). Conversely, FOXP3⁺ Tregs have been shown to be beneficial in a wide range of pathologies. FOXP3⁺ Tregs act to suppress activation of the immune system and induce immune tolerance (27). Evidence suggests that expansion of Tregs may be beneficial in a wide variety of disease states that involve pathological activation of the immune system, including allergy (28), asthma (28), multiple sclerosis (29), graft versus host disease (30), diabetes (31), and hypertension (32, 33) as well as many others. Given its therapeutic potential against inflammatory disease, there is currently much interest in methods to activate the cholinergic anti-inflammatory pathway. Activation of vagal efferents by noninvasive ultrasound has been demonstrated to reduce kidney injury following ischemia reperfusion (34). Furthermore, in humans, efforts to stimulate the cholinergic anti-inflammatory pathway chronically by implanting stimulating electrodes on the vagal nerves have shown promise in patients with rheumatoid arthritis (17). Consistent with activation of the cholinergic anti-inflammatory pathway, in rats, removal of the spleen or treatment with the α7 nicotinic Ach receptor antagonist MLA abolished the anti-inflammatory effect of oral NaHCO₃ intake. Our data indicate that oral NaHCO₃ loading may provide a cheap, relatively safe, effective, and easily accessible and/or noninvasive method to activate cholinergic anti-inflammatory pathways in humans, which may be of benefit to patients suffering from a multitude of inflammatory disease states. As such, our findings could potentially have significant clinical application to the treatment of human disease. Future studies testing the efficacy of oral NaHCO₃ to limit injury in models of inflammatory disease will be required to determine the therapeutic potential of this stimuli.

Our initial interest in oral NaHCO₃ was in relation to its protective effect against kidney functional decline. By limiting the ability of the kidneys to excrete an acid load, loss of kidney parenchyma in CKD patients can promote metabolic acidosis (35). Oral NaHCO₃ was first given to CKD patients to prevent the deleterious effects of metabolic acidosis on nonrenal functions; however, evidence from a number of small clinical trials as well as experimental models now indicates that oral NaHCO₃ supplementation may also slow the decline in kidney function in CKD patients, even in the absence of metabolic acidosis (6). As chronic inflammation has been implicated in both acute and chronic kidney injury (1–3), we investigated the effect of oral NaHCO₃ on inflammatory cell profiles within the kidney. Our finding that oral NaHCO₃ promotes anti-inflammatory immune polarization is consistent with the hypothesis that oral NaHCO₃ may protect the kidney, at least in part, by activating a cholinergic anti-inflammatory pathway. In the current study we did not maintain animals on oral NaHCO₃ long enough to determine if there were differences in the rate of glomerular filtration rate (GFR) decline. This experiment is particularly difficult in the Dahl SS rat model as these rats develop malignant hypertension when fed salt in the drinking water, which causes them to die prematurely of stroke, often before major reductions in GFR are observed. As such, long-term studies in a normotensive model of CKD may be required to determine whether the anti-inflammatory effects of NaHCO₃ administration contribute to maintenance of GFR in the rat. Interestingly, we found that inhibition of gastric proton pumps prevented oral NaHCO₃ from activating an anti-inflammatory response, suggesting that gastric H⁺ secretion is required. This finding may be particularly relevant to CKD, as long-term use of proton pump inhibitors has been associated with increased risk of developing CKD (36).

There has been much debate in regards to the mechanisms through which efferent cholinergic signals from the vagal nerves are transmitted to the spleen (37–39). At the heart of this debate is the apparent lack of vagal innervation to the spleen. Although some studies have reported evidence of vagal innervation at the splenic poles in rodents (38), this remains controversial, and most current reports favor a mechanism in which efferent vagal signals are transmitted to the spleen via sympathetic nerves [this hypothesis, however, has also been refuted (40)], which enter along the vessels of the splenic plexus and/or via nonneuronal acetylcholine signaling initiated by circulating T cells that receive vagal input outside of the splenic parenchyma (41, 42). Both of these pathways are thought to result in activation of α-7–containing nicotinic receptors on splenic macrophages. Acetylcholine is synthesized in practically all living cells (43), and many components of acetyl-choline signaling have been found in mesothelial cells of the lung (44) and intestine (45). In addition to a barrier function, mesothelial cells have been reported to express TLR (46), act as APC (47), and are known to produce a number of immunoregulatory factors including COX2, iNOS, IL-1Β, TNF-α, and MCP-1 (45). Taken together, our data indicate a previously unrecognized function of the mesothelium to mediate splenic anti-inflammatory responses via cholinergic signaling to the splenic capsule. Such a mechanism may resolve much of the current controversy regarding how vagal stimulation alters splenic function, suggesting cholinergic signals may be transmitted to the splenic capsule not by vagal efferent nerves but by the mesothelium. Our findings are consistent with a recent report by Mihara et al. (45) who found that stimulation of the α-7 nicotinic acetylcholine receptor on rat intestinal mesothelial cells blunted the inflammatory response of these cells to LPS. Interestingly, these investigators also reported that enteric nerves (presumed to be vagal efferents) adhered to the mesothelial cells of the ileal serosa, leading the authors to speculate that the anti-inflammatory effects of the mesothelium in the ileum may be stimulated by autonomic nerves (45). To our knowledge, our data are the first evidence that mesothelial cells may have a “neuronal-like” function innervating the splenic capsule. Although our histological findings provide an explanation for our immunological data, our finding of neuronal-like function in mesothelial cells is unprecedented and raises a number of important questions. Although we are unable to address these questions in the current study, further studies are warranted, as similar innervation by mesothelial cells in other organs could represent a previously unrecognized pathway of biological communication.

Our finding that PGP9.5 staining within the capsular mesothelial cells was maintained, and that following transection of the vagal nerves, these cells did not hypertrophy as they did following movement of the spleen indicates that these cells maintained a “neural phenotype” in the absence of vagal input. These findings raise the possibility that these cells may receive signals from sources other than the vagal nerves. Evidence for vagal regulation of immunity derives from observations that vagal stimulation promotes an anti-inflammatory response that originates in the spleen (48). Given that there is no known vagal innervation to the spleen, selective vagal denervation of the spleen has been impossible. Consequently, studies aimed at demonstrating the role of the vagal nerves in the splenic anti-inflammatory response have either also stimulated or also inhibited vagal signaling to the gastrointestinal tract (15, 49–52). Our data indicating that oral ingestion of NaHCO₃ promotes an anti-inflammatory response, which is inhibited by an antagonist of the gastric proton pump, raises the possibility that the effect of vagal stimulation or denervation to promote or inhibit the anti-inflammatory response, respectively, is secondary to the common denominator between these stimuli: the stimulation of acid secretion in the stomach (53–55). This hypothesis is consistent with findings that Ghrelin, which also stimulates acid secretion, can activate the anti-inflammatory pathway (56). Our finding that mesothelial cells are required to mediate this anti-inflammatory response provides a potential sensory mechanism for this alternative hypothesis, whereby stomach acid secretion alters some factor within the peritoneal milieu, such as pH, that is sensed by the mesothelium that lines this compartment. Such a mechanism may be of physiological importance in deciphering whether Ags absorbed by the gut are inert (coming after a meal) or represent a potential infection of the peritoneum with ensuing acid production by invading bacteria and providing the appropriate response, either tolerance or inflammatory immune response, respectively. This alternative hypothesis challenges our current understanding of how vagal nerve stimulation promotes the cholinergic anti-inflammatory response in the spleen, suggesting for the first time that there may be no direct interface between the nervous and immune systems. In light of our data, further studies are warranted to determine whether promotion of an anti-inflammatory effect following stimulation of vagal nerves (classical activation of the cholinergic anti-inflammatory response) occurs independent of a requirement to stimulate stomach acid secretion.

A number of previous studies have investigated potential cholinergic innervation of the spleen and found no evidence of significant innervation (37). For example, Bellinger et al. (57) found little effect of vagal transection on acetylcholine esterase staining or choline acetyltransferase activity in the spleen of adult rats. How then can we reconcile our data with the results of these previous studies? One possibility that may explain our differing findings is that mesothelial cells represent a small fraction of total splenic tissue and thus would be unlikely to contribute significantly to total splenic choline acetyltransferase activity. Furthermore, as discussed below, the mesothelial cells themselves may not be directly responsible for signaling immune cells within the splenic tissue, lessening the need for significant parenchymal innervation (57). In difference to previous reports, our immunohistochemical studies were focused on the splenic mesothelium. As mesothelial cells reside on the splenic surface and Abs can often accumulate around the edge of the tissue, it is possible that previous investigators disregarded positive staining for acetylcholine esterase in these cells as nonspecific. Our observation that only mesothelial cells along the inferior axis of the splenic capsule stain positive for acetylcholine esterase, as well as evidence that this staining is lost following transection of the vagal nerves, however, strongly refute this possibility. Furthermore, these same cells stain positive for the pan-neuronal marker PGP9.5. It is also possible that mesothelial staining was simply missed by previous investigators. In this regard, unless specifically focusing on the mesothelial layer on the inferior axis of the splenic capsule, our own observations indicate there is little difference in acetylcholine esterase staining following vagal transection in other regions of the spleen.

One question raised by our data is, “How can mesothelial cells alter the immune cell profile within the spleen?” Perhaps mesothelial cells directly secrete acetyl-choline onto the splenic capsule, altering local splenic monocyte or T cell activation state. Alternatively, mesothelial cells could act via a more complex signaling process. In this regard, it is interesting that we observe a dense network of nerves directly below the splenic capsule on which the mesothelial cells reside (Fig. 10, Supplemental Video). Although it has been previously reported that the splenic capsule is innervated, this innervation was reported to consist primarily of tyrosine hydroxylase–positive sympathetic nerves that arise from blood vessels that supply the splenic parenchyma (58). In difference, we find that the majority of capsular nerves stain negative for tyrosine hydroxylase and instead stain lightly positive for acetyl-choline esterase and PGP9.5. Importantly, and consistent with previous reports, acetylcholinesterase staining was not altered in this region of the spleen following either manipulation of the spleen or subdiaphragmatic vagal denervation, suggesting these nerves are not of vagal origin (57). Given their close anatomical arrangement, we speculate that mesothelial cells may release paracrine factors, such as acetylcholine, that alter the signaling of these capsular ([perhaps sympathetic (57)]) nerves, which may then modulate the anti-inflammatory response within the splenic parenchyma. Such a signaling pathway could potentially explain why sympathetic denervation of the spleen results in loss of cholinergic anti-inflammatory responses, as sympathetic denervation has been demonstrated to abolish acetylcholinesterase staining in the region in which these underlying nerves are located (57). Importantly, we found no evidence that simple manipulation of the spleen altered sympathetic innervation of the spleen, as indicated by tyrosine hydroxylase staining (Supplemental Fig. 4).

In summary, we report that oral NaHCO₃ activates splenic anti-inflammatory pathways in both rats and humans. Our novel finding provides a potentially practical and/or cost-effective and relatively safe method to activate splenic anti-inflammatory pathways in humans and therefore may have significant therapeutic potential for inflammatory disease. We provide both functional (flow cytometry) and anatomical and histological evidence that the signals that mediate this response are transmitted to the spleen via a novel neuronal-like function of mesothelial cells. To our knowledge, this is the first evidence that mesothelial cells may have a role in transmitting cholinergic signals to distal sites and, combined with evidence that gastric acid secretion is required to promote an anti-inflammatory response to NaHCO₃, raises the possibility that there may be no direct interface between the nervous and immune systems. Future studies testing the efficacy of oral NaHCO₃ to limit injury in models of inflammatory disease will be required to determine the therapeutic potential of this stimuli.

We thank the Augusta University Histology Core for help with sample preparation and analysis.

The authors have no financial conflicts of interest.