Cytokine Diversity in Human Peripheral Blood Eosinophils: Profound Variability of Interleukin-16

Abstract

Eosinophilic leukocytes develop in the bone marrow and migrate from peripheral blood to tissues where they maintain homeostasis and promote dysfunction via release of pre-formed immunomodulatory mediators. Here we explore human eosinophil heterogeneity with a specific focus on naturally-occurring variations in cytokine content. We found that human eosinophil-associated cytokines varied on a continuum from minimally (CV ≤ 50%) to moderately variable (50% < CV ≤ 90%). Within the moderately variable group, we detected immunoreactive IL-27 (953 ± 504 pg/mg lysate), a mediator not previously associated with human eosinophils. However, our major finding was the distinct and profound variability of eosinophil-associated IL-16 (CV = 103%). Interestingly, eosinophil IL-16 content correlated directly with body mass index (R² = 0.60, ***p < 0.0001) in one donor subset. We found no direct correlation between eosinophil IL-16 content and donor age, sex, total leukocytes, lymphocytes or eosinophils (cells / μL), nor was there any relationship between IL-16 content and the characterized −295T/C IL-16 promoter polymorphism. Likewise, although eosinophil IL-1β, IL-1α, and IL-6 levels correlated with one another, there was no direct association between any of these cytokines and eosinophil IL-16 content. Finally, a moderate increase in total dietary fat resulted in a 2.7-fold reduction in eosinophil IL-16 content among C57BL/6-IL5tg mice. Overall, these results suggest that relationships between energy metabolism, eosinophils and IL-16 content are not direct or straightforward. Nonetheless, given our current understanding of the connections between asthma and obesity, these findings suggest important eosinophil-focused directions for further exploration.

Introduction

Eosinophils have been historically perceived as end-stage, cytotoxic cells with a limited range of effector responses. Numerous observations have profoundly altered these views and have provided a spotlight on eosinophils as complex cells with critical immunomodulatory functions [1, 2]. While eosinophils are activated in and contribute to the pathophysiology of a wide spectrum of allergic and non-allergic disorders [3 - 7], evolution tells us that the ability to induce pathology cannot be the “raison d’être” for any existing cell lineage [1]. Recent studies that explore the role of eosinophils at homeostasis in the gastrointestinal tract [8, 9], that identify distinct eosinophil-mediated antimicrobial activities [10, 11] and that reveal specific eosinophil subtypes [12, 13] suggest that there are new and profound complexities remaining to be unraveled.

Among the most prominent features of eosinophils are their distinct cytoplasmic granules, which contain unique cationic proteins [14] and immunomodulatory mediators [15 – 17]. Moqbel, Lacy and colleagues [18, 19] were among the first to characterize these mediators in eosinophils, which include cytokines, chemokines, lipids, growth factors and their receptors [20]. Physiologic roles for these mediators and their importance in eosinophil function was highlighted by Lee and colleagues [21] in their manuscript entitled “The LIAR hypothesis,” in which they proposed that eosinophils were recruited to tissues to promote local immunity, remodeling and repair. This hypothesis has been strongly supported by recent studies on eosinophil function in health and disease in vivo; for example, studies featured in reviews on eosinophil function in helminth infection and innate immunity [22 – 24].

As such, a better understanding of eosinophil mediators and the mechanisms that promote their synthesis and release is currently a topic of significant interest [25 – 27]. Toward this end, Melo and colleagues [28] have characterized the tubulovesicular networks that serve as conduits for mediator secretion, and Lacy and colleagues [29, 30] have examined signaling factors that promote degranulation. The potential for distinct cytokine profiles, i.e., differences among eosinophils from human subjects, was first addressed by Spencer and colleagues [31], who focused on seven individual cytokines identified in eosinophil lysates from eighteen non-allergic and allergic donors. Among their findings, the Th1 cytokine IFNγ was identified as a prominent component of human eosinophils, although no significant differences were reported in comparisons between allergic vs. non-allergic donors.

In this manuscript, we build on these findings using dual-antibody proteome profiling technology. With this method we examined differential expression of 36 cytokines in lysates prepared from peripheral blood eosinophils purified from whole blood from normal donors. As a component of this study, we identified naturally-occurring variations in cytokine content, including the unique pattern displayed by the dual-function cytokine, interleukin-16 (IL-16) and its unanticipated correlation with donor body mass index (BMI) in one donor subset.

Materials and Methods

Blood donors.

Single samples of approximately thirty (30) mL EDTA-anti-coagulated whole blood were obtained from de-identified adult normal donors from the National Institutes of Health Clinical Center (Protocol Identifier ) and from the Laboratory of Allergic Diseases normal volunteer protocol (#09-I-0049). All donors agreed to donation via standard informed consent protocols. Eligibility for donation via the Clinical Center protocol includes age greater than or equal to 18 years, weight greater than 110 pounds, no known history of heart, lung, kidney disease or bleeding disorders, and no current pregnancy; samples used in our study were also negative for cytomegalovirus. Eligibility for donation via the LAD protocol includes age between 18 – 65, no known history of chronic virus infection, anemia, bleeding disorder or use of therapies that would have an impact on hematopoiesis, and no current pregnancy.

Mice.

Interleukin-5 transgenic mice on the C57BL/6 and BALB/c backgrounds (B6-IL5tg NJ.1638 [32]) and BALB/c-IL5tg [33] mice, respectively) were maintained on-site (14BS vivarium, NIAID/NIH). Mice were maintained on normal house diets (LabDiet, Advanced Protocol 5V0T, 12.5% kcal from fat) or a diet with moderately increased fat (LabDiet, Select Mouse 5B0G, 22% kcal from fat), the latter for 5 – 6 weeks prior to evaluation. The National Institute of Allergy and Infectious Diseases Division of Intramural Research Animal Care and Use Committee, as part of the National Institutes of Health Intramural Research Program, approved all the experimental procedures as per protocols LAD7 and LAD 8E.

Human eosinophil isolation.

Eosinophils were isolated from whole blood using the Miltenyi Human Eosinophil isolation kit (#130–104-466) according to manufacturer’s instructions as described [34]. Briefly, particle-bound antibodies included in the MACS Cell Isolation Cocktail are added to whole blood which was then subjected to a magnetic field using the MACSxpress Separator (#130–098-308). Peripheral blood erythrocytes and leukocytes other than eosinophils were bound to antibodies and were held within the magnetic field while unbound eosinophils were carried through in the mobile buffer phase; residual erythrocytes were lysed with cold endotoxin-free water. Isolated eosinophils (typically ~ 2.0 – 8.0 × 10⁶ from ~30 mL blood, at > 98% purity and viability) were washed and resuspended at 10⁷ cells/mL phosphate buffered saline with 0.1% bovine serum albumin. Cytospins were prepared and stained with modified Giemsa (Diff-Quik; ThermoScientific) and a minimum of 200 cells were scored per preparation. Photo-microscopy was performed with a DMI4000 light microscope (Leica Microsystems, Wetzlar, Germany) equipped with a Retiga 2000R camera and analyzed using QCapture software (both from QImaging, Surrey, BC, Canada).

Preparation of lysates from human eosinophils:

Isolated eosinophils were collected by centrifugation, washed, and resuspended in lysis buffer (20 mM tris pH 8.0, 139 mM NaCL, 0.5 μM EDTA, 1% Igepal and 10% glycerol with 1 tablet of protease inhibitor mix (Roche #04693159001) per 10 mL lysis buffer) at 10⁷ cells / mL. After 30 min incubation at 2 – 8°C, insoluble material was removed by centrifugation. Supernatants were stored in aliquots at −70°C until use. Protein concentration in supernatant was determined by MicroBCA assay (Thermo Scientific Cat. #23235) typically at 600 – 1000 μg/mL.

Cytokine profiling and analysis.

The cytokine contents in the lysates from individual human donors were evaluated using the Proteome Profiler Array Panel A (ARY005B, R&D Systems) which permits simultaneous detection of 36 distinct human cytokines. Briefly, biotinylated detection antibody cocktail was added to eosinophil lysates prepared as described above which were then used to probe membranes embedded with anti-human cytokine capture antibodies. Detection is carried out with Streptavidin-conjugated to Infra-red Dye 800CW (Li-Cor, Cat #926–32230) and signals were detected using a LiCor Odyssey CLx and Image Studio Software version 5.2 (LiCor). As information from cytokine profiling is generated as mean pixel density, one donor was chosen to serve as a standard which enabled us to normalize data from multiple trials.

Characterization of a standard donor lysate and normalization of profiling data.

The cytokine profiler generates raw data in the form of mean pixel density (mpd). As such, we used a standard donor lysate to normalize results which enabled us to compare findings from multiple trials. We demonstrated that consistent findings (mpd detected per cytokine) were obtained in a single trial in which profiler membranes were probed with 100 μg of eosinophil lysates prepared from three independent blood samples from the standard donor [Fig. 1A]. We also established a critical lysate concentration vs. response profile [Fig. 1B]. Eosinophil lysates from the standard donor were included in all experimental trials and were used to normalize the mean pixel density data as shown in the text. An example of the raw data obtained from the profiler for the cytokine interleukin-13 (IL-13) is shown in Fig. 1C. In order to compare data, the technical replicates from two independent trials are normalized as follows; for example from trial #1, technical replicates as shown, (d1a + d1b) / (s1a + s1b) = (1423 + 1338) / (1289 + 1191) = 1.11, and from trial #2, (d8a + d8b) / (s2a + s2b) = (780 + 740) / (703 + 740) = 0.98, and continuing, generating the data shown in Fig. 1D, in units of normalized pixel density (npd).

Fig. 1. Validation of cytokine profiling and normalization to standard donor lysate.

A. Reproducibility. Lysates were prepared from eosinophils isolated from from the whole blood drawn on three separate occasions from the standard donor; 100 μg from each lysate were evaluated by cytokine profiling in a single trial, and results (mean pixel density, mpd) for a selection of seven (7) cytokines from the full set are shown here. B. Dose-response. 250 μg, 100 μg and 75 μg samples from standard donor eosinophil lysate were evaluated in a single trial; results in mpd are as shown. C. and D. Sample data and normalization. The variability determinations presented in Tables 2 and 3 and in Figure 2 were generated from two experimental trials, each performed with 7 unique donor lysates (100 μg each) a single standard sample replicated in each trial. As mpds differ between trials (shown in C. for the cytokine, IL-13), the technical replicates from each donor sample were averaged and normalized to the standard donor in each trial as follows: Donor #1 (d1) is normalized via the calculation from technical replicates (error bars) (d1a + d1b) / (std a + std b) = (1423 + 1338) / (1289 + 1191) = 1.11 normalized pixel density (npd), with a and b representing technical replicates. The same calculations are performed for each donor in trial #1 and trial #2 (shown in D.). Once the data are normalized (npd), the range, mean, standard deviation, and coefficient of variation (CV, %) are calculated. For IL-13 as shown, the CV (%) is 26.8%, which places it within the minimum variability group (Table 2).

Cluster analysis.

Cytokine content data normalized as above were subjected to hierarchical cluster analysis (Partek Genomics Suite, v 7.18.0130, Partek Inc. St. Louis, MO). The clustering was performed using Euclidean dissimilarity with average linkage and gap statistics using five bootstrap iterations to produce the dendrograms and heat map. Missing metadata were imputed for two subjects.

Isolation of DNA and amplification of IL-16 promoter sequence for identification of −295T/C promoter polymorphism.

Donor DNA was isolated from frozen lysates of purified eosinophils using MinElute PCR purification Kit (Qiagen, #28004) or from frozen eosinophil cell pellets after treating with DNA lysis buffer (50mM Tris-Cl pH 8.0, 20mM NaCl, 1mM EDTA, 0.2% SDS with 200 μg/mL proteinase K.) A fragment of the promoter of the human IL-16 gene (bp 3280 to 3667, Genbank accession no. AY497901 http://www.ncbi.nlm.nih.gov/genbank)) was PCR-amplified with forward primer 5’-ATT GAC AAG CAT TTT CCT GAG T-3’ and reverse primer 5’-CGA GAC ACG CTT TGA TTG G-3’. The resulting 388 bp amplicon was sequenced (Genewiz, South Plainfield, NJ) to identify the nucleotides at the −295 site in the promoter region as described [35].

Isolation of mouse eosinophils and preparation of lysates.

Eosinophils were isolated from single cell suspensions prepared from lungs of C57BL/6-IL5tg and BALB/c-IL5tg mice; mice were perfused, lung tissue was excised, and single cell suspensions were prepared for isolation of eosinophils using the FACS protocol previously described [36]. Lysates were prepared from purified eosinophils (CD45⁺CD11c^-MHCII^-SiglecF⁺Gr1^-/lo) using the method described above for human cells at 10⁷ cells / mL lysate and used to determine immunoreactive IL-16 content by ELISA as described below; total protein concentration was determined by MicroBCA assay (Thermo Scientific Cat. #23235).

ELISAs.

Absolute levels of IL-16, IL-27, IL-1α, IL-1β and IL-6 in lysates prepared from human eosinophils and IL-16 in mouse eosinophils was determined using human DuoSet ELISAs from R&D Systems, including DY316 for IL-16; 15.6–1000 pg/mL range, DY2526 for IL-27; 156 – 10,000 pg/mL range; DY206 for IL-6, 9.38 – 600 pg/mL range; DY200 for IL-1α, 7.8 – 500 pg/mL range, and DY201 for IL-1β, 3.91 – 250 pg/mL range and mouse DuoSet ELISA DY1727 for IL-16; 46.9–3000 pg/mL range) respectively.

Statistics.

Statistical analyses were performed as indicated in text using algorithms in Graphpad PRISM 7 and 8.

Results

Normal donors.

Eosinophils were isolated from EDTA-anticoagulated whole blood from normal adult volunteer donors. The eligibility and exclusion criteria for donation are described in the Methods. As shown in Table 1, the full donor group including metadata (n = 37) included 22 males and 15 females between the ages of 22 – 75 and 20 – 66 years of age, respectively. The mean body mass indices (BMIs) were determined as 28.5 ± 5.4 kg / m² and 27.9 ± 5.9 kg / m² for male and female donors, respectively. These average values are within the range defined as overweight (25 < BMI ≤ 29.9) by the U. S. Centers for Disease Control, which are findings consistent with the current prevalence of overweight body habitus as well as obesity (BMI ≥ 30) in the United States [37]. Hematologic parameters for all donors were within or near our laboratory normal range. Of note, two donors had peripheral eosinophil counts that were slightly higher than the laboratory normal range, although both values were <500 eosinophils/μL, and thus not within the definition of peripheral eosinophilia. The donor selected to be the source of eosinophil lysate standard as described in the Methods and Fig. 1 was Female, Caucasian, age 60, BMI 19.2, hematocrit 42%, total leukocytes at 6.8 × 10³ /μL and 4.8% eosinophils.

Table 1: Normal donors.

Eosinophil lysates were prepared from thirty-seven (37) normal blood donors¹ (24 Caucasian, 6 African-American, 3 Hispanic, 1 Asian and 3 Mixed Race or Unrecorded). Also included are the NIH Clinical Center Laboratory normal range for each parameter. There were no statistical differences noted in any comparison of male vs. female donors, data shown as ± standard deviation. ²Two donors with peripheral eosinophil counts above normal range.

Gender	# Donors1	Age (years)	Age range (years)	BMI (kg / m2)	BMI range (kg / m2)
Male	22	43 ± 16	22 – 75	28.5 ± 5.4	18.9 – 41.3
Female	15	38 ± 16	20 – 66	27.9 ± 5.9	19.2 – 38.3
Hematology		Mean ± SD		Donor Range	Normal Lab Range
Hemoglobin (g/dL)		13.7 ± 1.1		11.4 – 16.3	11.2 – 15.7
Hematocrit (%)		39.9 ± 3.0		33.1 – 44.3	34.1 – 44.9
White Blood Cells (K/μL)		5.5 ± 1.1		3.1 – 7.50	3.98 – 10.04
Neutrophils (K/μL)		3.39 ± 0.84		1.52 – 4.72	1.56 – 6.13
Neutrophils (%)		61 ± 6.8		37.3 – 73.8	34 – 71.1
2Eosinophils (K/μL)		0.17 ± 0.10		0.05 – 0.45	0.04 – 0.36
Eosinophils (%)		3.2 ± 1.8		1.0 – 9.4	0.7 – 5.8
Basophils (K/μL)		0.05 ± 0.03		0.01 – 0.12	0.01 – 0.08
Basophils (%)		0.94 ± 0.56		0.20 – 3.1	0.1 - 1.2
Monocytes (K/μL)		0.34 ± 0.09		0.18 – 0.55	0.24 – 0.86
Monocytes (%)		6.2 ± 1.4		3.7 – 9.1	4.7 – 12.5
Lymphocytes (K/μL)		1.55 ± 0.42		0.83 - 2.49	1.18 – 3.74
Lymphocytes (%)		28.5 - 6.7		16.8 – 51.5	19.3 – 15.7
Platelets (K / μL)		239 ± 57		120 – 374	173 – 369

Relative expression of eosinophil cytokines.

Lysates were generated from eosinophils isolated from an initial fourteen normal donors and relative expression of 36 cytokines was determined by profiling in two separate trials, each including the standard lysate; the method for evaluating profiling data with this standard and presentation of data in units of normalized pixel density (npd) was described in the Methods and Legend to Fig. 1. These findings were used to determine range (minimum and maximum) and mean ± standard deviation for all cytokines evaluated. All thirty-six (36) cytokines and soluble mediators included on the profiler were detected in the donor eosinophil lysates, including IL-27, which, to the best of our knowledge, has not been identified previously as a component of human eosinophilic leukocytes. We confirmed expression of immunoreactive IL-27 in the standard and ten additional donor lysates at 953 ± 504 pg/mg lysate, with a coefficient of variation (CV) = 53%. Numerous and complex immunomodulatory activities have been described for IL-27, primarily those directed at regulating the activities of T lymphocyte subsets [38, 39].

As shown in Table 2, twenty-two (22) of the 36 mediators detected by profiling display minimal variability, defined here as those with a calculated coefficient of variation (CV) less than or equal 50%. In addition to IL-13 (see Fig. 1D), this group also includes IL-5, CCL2, CCL1, IL-4, TNFα and IL-17.

Table 2: Cytokines detected that display minimal variability.

Twenty-two (22) of the 36 cytokines detected in eosinophil lysates by proteome profiling display minimal variability (n = 14 donors). Profiling included the standard donor lysate described in Fig. 1 which was included as a normalization control; see Fig. 2 for an example of how this analysis was performed. Minimal variability is defined as coefficient of variation (CV) ≤ 50%; npd, normalized pixel density.

Cytokine	CV (%)	Mean ± SD (npd)	Range (npd)
MIF	19.4	1.11 ± 0.22	0.80 – 1.52
IL-5	21.5	1.23 ± 0.25	0.63 – 1.58
IL-13	26.8	0.94 ± 0.25	0.40 – 1.28
PAI-1	29.8	0.83 ± 0.25	0.36 – 1.16
G-CSF	30	0.86 ± 0.25	0.37 – 1.28
IL-17E	30.1	0.80 ± 0.23	0.31 – 1.14
CCL2	30.8	0.86 ± 0.24	0.40 – 1.26
CCL3 / CCL4	30.8	0.87 ± 0.27	0.30 – 1.24
CXCL1	31.7	0.98 ± 0.31	0.41 – 1.49
CCL1	32.5	0.91 ± 0.29	0.38 – 1.30
IL-21	33.9	0.90 ± 0.30	0.34 – 1.28
CXCL10	37.5	1.12 ± 0.42	0.25 – 1.82
CXCL12	38.3	1.00 ± 0.37	0.41 – 1.64
IL-32A	39	0.98 ± 0.38	0.31 – 1.39
CCL5	39.4	1.26 ± 0.50	0.48 – 2.16
GM-CSF	39.5	1.19 ± 0.47	0.51 – 2.23
ICAM-1	40.6	1.34 ± 0.54	0.49 – 2.51
TNFα	41.9	1.17 ± 0.49	0.42 – 2.25
IL-17	43.1	1.57 ± 0.68	0.59 – 2.90
C5 / C5a	44.8	1.04 ± 0.46	0.20 – 1.69
IL-4	48.5	1.54 ± 0.75	0.44 – 2.82
CD40L	48.7	0.99 ± 0.48	0.24 – 1.67

The thirteen (13) cytokines detected in eosinophil lysates at levels that vary more moderately (CV > 50% but < 90%) are listed in Table 3. Among these are IFNγ, IL-8, IL-10, and near the higher end of this group, CXCL11 and IL-1Ra, the latter two with CVs at 74% and 84%, respectively.

Table 3: Cytokines detected that display moderate to profound variability.

Thirteen (13) of the 36 cytokines detected in eosinophil lysates that display moderate variability (n = 14 donors). Profiling included standard donor lysate (see Fig. 1) as a normalization control as described in the Legend to Table 2. Moderate variability is defined as coefficient of variation (CV) > 50% and ≤ 100%; see Fig. 2 for an example of how this analysis was performed; npd, normalized pixel density. One cytokine (IL-16) of the 36 detected displays profound variability, with a CV > 100%.

Cytokine	CV (%)	Mean ± SD (npd)	Range (npd)
IFN-γ	50.8	0.85 ± 0.41	0.23 – 1.63
IL-6	53.5	1.97 ± 1.05	0.44 – 3.39
IL-18	54.9	1.37 ± 0.75	0.39 – 2.56
IL-8	55.8	1.64 ± 0.91	0.56 – 3.85
IL-2	56.1	1.53 ± 0.86	0.47 – 2.82
IL-10	60.3	1.23 ± 0.74	0.24 – 2.62
IL-12p70	62.1	1.70 ± 1.05	0.47 – 3.76
IL-1β	63.1	1.77 ± 1.12	0.40 – 3.79
IL-1α	65.7	1.31 ± 0.86	0.33 – 3.11
TREM-1	66.7	1.29 ± 0.86	0.26 – 3.28
IL-27	72.5	1.80 ± 1.30	0.30 – 4.11
CXCL11	73.6	1.55 ± 1.14	0.31 – 3.94
IL-1Ra	84.1	1.22 ± 1.03	0.33 – 3.10
Cytokine	CV (%)	Mean ± SD (npd)	Range (npd)
IL-16	103	6.08 ± 6.3	0.49 – 19.1

By contrast, interleukin-16 (IL-16) was detected in eosinophil lysates at concentrations that varied profoundly, over a range of 0.49 – 19.1 npd, at a mean ± standard deviation of 6.08 ± 6.3, with a CV = 103%

Eosinophil IL-16 content was examined by cluster analysis with each cytokine content (npd) displayed for each donor (n = 14; [Fig. 2]). This analysis highlights the remarkable variability of eosinophil IL-16; the content pattern is unique and showed no correlations to those of any other cytokine or cytokines. Principal component analysis (PCA) revealed no associations with donor age, gender or BMI group (data not shown). Equally intriguing, the cluster analysis suggested coordinate regulation of eosinophil-associated cytokines (see Discussion) and revealed direct correlations among several in the moderately variable group (see Table 3), including IL-1β, IL-1α, and IL-6 as discussed further below.

Fig. 2. Cytokine variability in eosinophil lysates: heat map.

Variability (npd) determined as indicated in the Legend in Fig. 1. via clustering analysis. Donor (1 – 14) characteristics listed in columns include: race (see Legend to Table 1), BMI class (n, normal; o, overweight; b, obese, by standard criteria; age group (y, young 18 – 35 yrs; m, middle aged 36 – 55 yrs; o, old ≥ 56 yrs) and sex (male, female). Asterisk (*) indicates donor group in which one can detect coordinate up-regulation of a cohort of moderately variable proinflammatory cytokines.

Absolute quantitation of eosinophil-derived IL-16.

To continue our exploration of eosinophil IL-16 content, ELISA analysis was performed on the lysates from the original 15 donors (14 donors and the standard) together with lysates prepared from eosinophils from an additional 22 donors (total n = 37; see Table 1). First, as shown in Fig. 3A, we demonstrated that the absolute concentration of IL-16 (pg/mg eosinophil lysate) correlated directly with the mean pixel density readings from lysates from the original donor set. These findings which provide direct confirmation and validation of the findings from the cytokine profiling methodology. Immunoreactive IL-16 (pg/mg eosinophil lysate) for the complete set of eosinophil lysates is shown in Fig. 3B. As anticipated, immunoreactive IL-16 content varies dramatically among donors, from a low of 274 pg/mg lysate to a high of 13,300 pg/mg lysate; the mean ± standard deviation calculates to 2683 ± 3026 pg/mg lysate, and a CV of 113%. The eosinophils shown Fig. 3C were isolated from the donor indicated at the arrow in Fig. 3B, one of the donors with a relatively high concentration of eosinophil-associated IL-16, at 4970 pg/mg lysate; as shown, these eosinophils are pure preparations with normal morphology.

Fig. 3. Cytokine variability in eosinophil lysates: interleukin-16.

A. Concentration of IL-16 determined by ELISA (pg/mg lysate) correlates with data from cytokine profiling (mean pixel density) R² = 0.80, ***p < 0.0001. B. Immunoreactive IL-16 (pg/mg lysate) evaluated standard donor (pink bar) and extended donor set (n = 38); CV= 113%. C. Modified Giemsa staining of eosinophils isolated from donor indicated at arrow in B., IL-16 = 4971 pg/mg lysate; original magnification, 40X.

Eosinophil IL-16 content correlates with body mass index (BMI) for a subset of donors.

As shown, this analysis permits us to divide the donors into three groups. The first group (group A) includes donors with BMI ≤ 22 (n = 8). Interestingly, all donors with this low BMI maintain low levels of eosinophil-derived IL-16 (< 2500 pg/mg lysate). The second, group B (n = 18) are donors with a higher BMI (BMI ≥ 23) but who also maintain IL-16 < 2500 pg/mg eosinophil lysate. The final group, group C (n = 9), is most interesting, as these donors also have higher BMIs, in the same general range as group B (BMI ≥ 23) but have markedly elevated eosinophil IL-16 content (IL-16 ≥ 2500 pg/mg lysate). As shown in Fig. 4A, there is a significant correlation between eosinophil IL-16 content and donor body mass index (BMI) for donors in combined groups A/C (dotted line; R² = 0.60, ***p < 0.0001); there is also a correlation between eosinophil IL-16 content and BMI for donors in group C evaluated alone (R² = 0.39, *p < 0.05, data not shown). By contrast, no correlation between eosinophil IL-16 content and BMI was observed when evaluating combined group A/B, or group B alone.

Fig. 4. Eosinophil interleukin-16 content correlates with body mass indices (BMI) in a subset of normal donors.

A. IL-16 vs. BMI, Group A (pink symbols), BMI ≤ 22; Group B (black symbols), BMI ≥ 23, IL-16 < 2500 pg/mg lysate; Group C (blue symbols), BMI ≥ 23, IL-16 > 2500 pg/mg lysate. Correlation (dashed line), R² = 0.60, ***p < 0.0001. B. Distribution of normal donors as in A. (IL-16 vs. BMI), documenting male (filled symbol) vs. female (open symbol). No sex prevalence in either group (AC or AB) as per χ² test (total group 22 male, 15 female, see text); C. Eosinophil IL-16 content (pg/mg lysate) vs. donor age. Groups color-coded as indicated in A; ns, no significance.

Interestingly, donor sex does not factor in to these observations. As shown in Fig. 4B, there are 16 males and 10 females in group A/B (62% and 38%, respectively) and 9 males and 10 females in group A/C (47% and 53%, respectively); although there are fewer males in the latter group, in neither group do findings deviate significantly from the null hypothesis (p > 0.05, χ² test) which was calculated given expected values of 59% males (22) and 41% females (15). Likewise, there was no significant correlation between eosinophil IL-16 levels and donor age [Fig. 4C].

Eosinophil IL-16 content: two distinct normal distributions and no correlations with hematologic parameters.

As shown in Fig. 5A, group A and group B as defined above, which comprise all donors with IL-16 ≤ 2500 pg/mg lysate generate a normal distribution, with a mean and standard deviation of 1138 ± 656 pg/mg IL-16, and a CV of 57.6%. Likewise, group C (all donors with IL-16 > 2500 pg/mg lysate) also generates a distinct normal distribution, with a higher mean (6554 ± 3259 pg/mg lysate), and CV of 49.7% (Fig. 5B). If Group A is included with Group C (as in Fig. 4A), the normal distribution remains, and the CV increases to 85.6% (data not shown). Evaluated as two distinct distributions, IL-16 is among the high minimal to low moderately variable cytokines as described in Tables 2 and 3.

Fig. 5. Normal distributions and statistical evaluations.

A. Groups A/B (pink and black symbols) IL-16 < 2500; normal distribution with parameters as indicated, CV = 57.6%. B. Group C (blue symbols) IL-16 > 2500; normal distribution with parameters as indicated, CV = 49.7%.

However, even when divided into these two groups (IL-16 > 2500, group C and IL-16 < 2500, group A/B), no distinct correlations with any hematologic parameters were revealed. No correlations were observed between eosinophil IL-16 content and total leukocytes (cells / μL; [Fig 6A and Fig. 6B]). Likewise, despite characterization of IL-16 as a lymphocyte chemoattractant [40], no correlations were observed between eosinophil IL-16 content and total lymphocytes (cells / μL; [Fig. 6C and Fig. 6D]), nor was there any correlation between eosinophil number (cells / μL) and IL-16 content [Fig. 6E and 6F]. Donor sex makes does not skew or contribute to any of these results (see Suppl. Fig. 1)

Fig. 6. Eosinophil interleukin-16 content and hematologic parameters.

Groups A/B and Group C, color coded as in Figs. 4 and 5. No correlations determined for eosinophil lysate IL-16 (pg/mg lysate) vs. (A. and B.) total white blood cells (× 10³/μL); (C. and D.) total lymphocytes (× 10³ / μL); (E. and F.) total eosinophils (× 10³ / μL); ns, no significance.

Eosinophil IL-16 content does not correlate with IL-1β, IL-1α or IL-6.

Cytokines IL-1β, IL-1α and IL-6 play critical roles in modulating energy metabolism [41 – 45]. As suggested by the findings presented in the cluster analysis in Fig. 2, and shown graphically in Fig. 7A - C, eosinophil levels of IL-1β, IL-1α and IL-6 correlate directly with one another (R² = 0.67, 0.72 and 0.85, respectively); these correlations were confirmed on another set of samples by ELISA (Fig. 7D and 7E; pg/mg lysate, R² = 0.73 and 0.86, respectively). As suggested by the findings in Fig. 2, and confirmed here graphically [Fig. 7F] and on another set of samples by ELISA [Fig. 7G and 7H], eosinophil IL-16 content does not correlate directly with any of these mediators; a minor inverse correlation was detected in a comparison between IL-16 content and IL-1α (pg/mg lysate, R² = 0.29, *p < 0.05) although not when comparing normalized pixel density data (data not shown). As such, these findings reveal no clear relationships between IL-16, body weight, inflammation and metabolism.

Fig. 7. Eosinophil cytokine content: no correlations with IL-1α, IL-1β, or IL-6.

Correlations between A. IL-1α and IL-1β (npd; R² = 0.67, ***p = 0.0004); B. IL-1α vs. IL-6 (npd; R² = 0.72, ***p = 0.0001); C. IL-6 vs. IL-1β (npd; R² = 0.85, ***p < 0.0001); D. IL-1α vs. IL-1β (pg/mg lysate, R2 = 0.73, ***p < 0.005); E. IL-6 vs. IL-1β (pg/mg lysate R² = 0.86, ***p < 0.005); F. IL-16 vs. IL-6 (npd, ns); G. IL-16 vs. IL-1β (pg/mg lysate, ns); H. IL-16 vs. IL-1alpha (pg/mg lysate; R² = 0.29, *p < 0.05).

The −295T/C promoter polymorphism has no impact on eosinophil IL-16 content.

Variants at the −295 position in the leukocyte promoter of the human gene encoding IL-16 have been associated with varied responses to several human disorders including asthma, atopy and gastrointestinal diseases [35, 46, 47]. Burkhart and colleagues [35] identified the C codon at this position as promoting 6-fold more promoter activity when evaluated in human bronchial epithelial cells. To determine whether our findings relate in any way to donor genotype, we isolated DNA from a series of donors, and amplified a 388 bp promoter fragment for direct sequencing. As shown in Fig. 8A, the ratios we obtained, 0.53 TT: 0.44 CT: 0.03 CC were consistent with those reported in multiple studies for similar human donor populations [see data at https://www.snpedia.com/index.php/Rs4778889]. However, as shown in Fig. 8B, there are no statistically significant differences in eosinophil IL-16 content when comparing donors of the −295TT genotype to those of the −295CT/CC genotype.

Fig. 8. Genotyping and impact of −295T/C IL-16 promoter polymorphisms.

A. Fractions of the donor population with the given genotype and example of sequence shown. B. Eosinophil IL-16 content vs. genotype (−295T/T or −295C/T or C/C); ns, no significant difference.

IL-16 content of mouse eosinophils: strain and diet.

Given the findings above, it was not possible to predict how mice might respond to a diet that is higher in fat content. High fat diets are typically used to promote weight gain in susceptible mice, including those on the C57BL/6 background strain [48, 49]; on the other hand, human subjects are substantially more complex and do not necessarily gain weight in response to similar regimens [50]. Furthermore, our data suggest that eosinophil IL-16 content correlates with BMI, but only among one donor subset. There is no evidence to date suggesting a cause-and-effect relationship even in this donor subset, and, despite the growing mouse literature [51 – 54], there is at present no clear evidence regarding a role of human eosinophils in the immunomodulation of adipose tissue.

Nevertheless, as an initial consideration, hyper-eosinophilic mice from two distinct background strains (BALB/c and C57BL/6) were maintained for 5 – 6 weeks on enriched (22% fat) diets (see Methods). Eosinophils were isolated from these mice and IL-16 content was compared to those from matched mice maintained on baseline diets. As shown, this acute dietary change had no impact on detection of eosinophils in the lung in either of in these mouse strains [Fig. 9A]. However, among mice maintained on the normal diet, IL-16 content in eosinophils isolated from the C57BL/6-IL5tg strain was significantly higher (5.2-fold) than that measured in eosinophils from the BALB/c-IL5tg strain [Fig. 9B]. More interesting, the IL-16 content in eosinophils isolated from C57BL/6-IL5tg mice maintained on the 22% fat diet underwent a 2.7-fold reduction, from 2380 ± 650 pg/mg lysate to 870 ± 360 pg/mg lysate. These results are consistent with the findings shown in Fig. 7, as they suggest that, while eosinophil IL-16 content in human subjects may correlate with BMI in one donor subset, this finding may not be directly related to energy metabolism.

Fig. 9. Mouse eosinophil IL-16 content: strain and diet.

A. Eosiniophils in lungs of C57BL/6-IL-5tg (filled circles) and BALB/c-IL-5tg (open circles) mice maintained on normal diet and 5 – 6 weeks of a diet moderately increased in fat content (22%). B. IL-16 in lysates of eosinophils isolated from lungs of B6IL-5tg and BALB/c-IL5tg mice maintained on diets as above; pg/mg lysate; n = 5 – 10 per group, **p < 0.01 ANOVA.

Findings from these data suggest that a fully unbiased approach to the question of mice, diet and eosinophil cytokines may be required.

Discussion

In this work, we used a proteome profiling method to examine natural cytokine variability in eosinophils isolated from normal human subjects. As eosinophils play important roles in maintaining homeostasis and promoting tissue repair [1, 2], differences in cytokine content may have a direct impact on an individual’s capacity to defend against infection, respond to medication, and recover from disease. Among our prominent discoveries was the profound variability in IL-16 content, and the correlation between IL-16 content and donor BMI in a subset of phenotypically normal donors.

Interleukin-16 was first identified by Center and Cruikshank [55] as a T lymphocyte chemoattractant factor, the first of such molecules isolated from mitogen-stimulated human peripheral blood mononuclear cells. IL-16 is a tetrameric glycoprotein with high interspecies sequence homology that is expressed widely in both neuronal [56] and non-neuronal cells, the latter including CD8+ and CD4+ T cells, mast cells, eosinophils, fibroblasts and bronchial epithelial cells [57, 58]. Initially of interest as a biological means to suppress replication of human immunodeficiency virus [59], IL-16 has also been associated with the pathogenesis of allergic disorders [60] and promoting cell growth in a wide array of neoplastic diseases [61]. Biochemically, IL-16 has features in common with dual function cytokines IL-1α, IL-33 and HMGB1, including N-terminal DNA binding domains with nuclear localization signals separated by protease activation sites from regions capable of interacting with extracellular receptors [62]; as a group, these proteins, also known as alarmins, function to promote sterile inflammation in response to internal stress and cell necrosis.

Lim, Weller and colleagues [63] reported immunoreactive IL-16 protein in and release from the cytoplasmic granules of human eosinophils. Interestingly, IL-16 is also an eosinophil chemoattractant, eliciting responses via CD4 on the eosinophil cell membrane [64, 65]. Interestingly, IL-16 does not promote eosinophil degranulation or production of reactive oxidants. Given its role as a chemoattractant for both eosinophils and CD4+ T cells, IL-16 blockade in respiratory allergies and asthma has been explored [66 – 68].

In this study, we also examined the −295 T/C polymorphism in the leukocyte promoter of IL-16, a region that has been implicated in differential transcription in cell culture studies (C > T: [35]). The −295T/C polymorphism has been associated with several eosinophil-related disorders, notably, asthma, atopic dermatitis, and Crohn’s disease [35, 46, 47], although in most cases, no direct associations between polymorphisms and cytokine levels have been reported. The genotypes of the donors in our study were distributed in a manner consistent with others in our region and included a ratio of 0.53 TT: 0.44 TC: 0.03 CC (n = 34). Despite the afore-mentioned tissue culture findings, this specific promoter polymorphism had no direct association with the IL-16 content of the human eosinophil lysates. We cannot rule out the possibility that eosinophil IL-16 content is regulated by an unrelated transcriptional mechanism [69, 70] or by epigenetic / post-transcriptional means. Interestingly, we note that the human eosinophils evaluated by Lim and colleagues showed marked variation in levels of amplified transcript encoding this cytokine (see Fig. 4 of reference [63]).

As part of this study, we found a significant correlation between eosinophil IL-16 content and donor body mass index (BMI; Fig 3A) in a subset of normal donors. BMI, which is a ratio of weight to height in metric units, was a measure devised by the Belgian sociologist and statistician, Adolphe Quételet, as part of an effort to generate standard data on human populations [71]. While not a fully perfect means to evaluate weight or health [72], the BMI remains in widespread use. Likewise, while the roles of cytokines, inflammation and obesity is a field of significant interest (reviewed in [73 – 75]), the specific role of IL-16 and its correlation with weight in human subjects has been the subject of only limited evaluation. In one such study, Lichtnauer and colleagues [76] evaluated plasma cytokine levels among 79 adolescents and found elevated levels of IL-16 among those who were overweight (as determined by BMI) compared to those who were at normal weight. At the cellular level, Alomar and colleagues [77] identified IL-16 as synthesized by human pre-adipocytes although the proinflammatory mediator, IL-1β, had no impact on its release from these cells.

By contrast, there is a large and growing literature on the role of mouse eosinophils in promoting metabolic homeostasis and the immunomodulation of adipose tissue [51 – 54, 78, 79]. Among these studies, Wu and colleagues [52] characterized eosinophil migration to adipose tissue in wild-type mice, and found that eosinophil-deficient mice gain weight, develop impaired glucose tolerance, and insulin resistance when maintained on a high-fat diet. Rao and colleagues [53] and Qiu and colleagues [54] found that eosinophils are critical in promoting anti-inflammatory programs in adipose tissue, specifically, generating beige fat via release of IL-4.

There is considerable literature focused on the roles of the largely proinflammatory cytokines IL-1β, IL-1α, and IL-6 in immunoregulation of metabolism and control of adipose function [41 – 44]. Although there was no direct correlation (suggesting potential cross-regulation) between eosinophil-content of IL-16 and any one of these mediators, levels of IL-1α, IL-1β and IL-6 in eosinophils from normal donors all correlate directly with one another. Furthermore, levels of these eosinophil-derived cytokines are all relatively higher (together with IL-2, IL-12p70, IL-18, and IL-27, among others) in one specific cluster of primary donors (see asterisk in Fig. 2). Given our ongoing interests in natural variation in eosinophils, it will be intriguing to explore this observation further and to identify unique features among this potentially important donor cohort.

At this writing, there is only a limited information on obesity in human subjects and its impact on eosinophils, and likewise, the impact of eosinophils on human metabolic homeostasis and regulation of adipose tissue in disease states. Kim and colleagues [80] and Takeda and colleagues [81] reviewed the evidence linking obesity, asthma, eosinophilic inflammation, and leptin, the proinflammatory mediator synthesized by and released from adipose cells and detected in serum at levels corresponding with BMI in human subjects [82]. Human eosinophils express the receptor for leptin [83] and respond to exogenous leptin by undergoing chemotaxis, adhesion, release of cytokines, and prolonged viability [83, 84]. While it would be premature to speculate too far at this juncture, it is interesting to consider the possibility that chronic elevations in serum leptin may have an impact not only on acute responses, but also on eosinophil development, resulting in differences in cytokine content in a subset of otherwise normal donors. Of interest, Suzukawa and colleagues [85] found that leptin promoted cytokine synthesis in isolated human basophils.

In summary, we have performed proteome profiling on eosinophil lysates prepared from normal donors and characterized eosinophil-associated cytokines that varied on a continuum from minimally (CV ≤ 50%) to moderately (50% < CV < 90%) variable. Among the latter group, we identified IL-27, a member of the IL-12 family, as a new and previously unrecognized component of eosinophils from normal donors. By contrast, eosinophil IL-16 content varied profoundly and via a unique pattern with no correlations to levels of other eosinophil-derived cytokines. The dataset generated two distinct normal distributions, with one donor subset demonstrating a direct correlation between eosinophil IL-16 content and donor BMI. Given our current understanding of the links between eosinophils, asthma and obesity, an improved understanding of eosinophils and IL-16 content may provide important connections worthy of further exploration.

Key Points.

• Eosinophils are heterogeneous with natural variations in cytokine content.

• Eosinophil IL-16 content varies dramatically among normal donors (CV = 103%).

• Eosinophil IL-16 content may correlate directly with donor BMI (R² = 0.60).