Out of the lab and into the field: Validation of portable cell culture protocols

Abstract

Objective:

Field-based research on inflammation and health is typically limited to baseline measures of circulating cytokines or acute phase proteins, while lab-based studies can pursue a more dynamic approach with ex vivo cell culture methods. The laboratory infrastructure required for culturing leukocytes limits application in community-based settings, which in turn limits scientific understandings of how psychosocial, behavioral, and contextual factors influence the regulation of inflammation. We aim to address this gap by validating two “field-friendly” cell culture protocols, one using a small volume of venous whole blood, and another using finger stick capillary whole blood.

Methods:

We evaluated the performance of both protocols against a standard laboratory-based protocol, using matched venous and capillary blood samples collected from young adults (N=24). Samples were incubated with lipopolysaccharide (LPS) and hydrocortisone, and the production of pro-inflammatory cytokines IL-1β, IL-6 and TNFα was measured in response.

Results:

Comparisons indicate a high level of agreement in responses across the protocols and culture conditions. The overall correlation in results was 0.88 between the standard and small volume protocols, and 0.86 between the standard and capillary blood protocols. Repeatability for the small volume and capillary blood protocols was high, with mean coefficient of variation across five replicates of 6.2 and 5.4%, respectively.

Conclusions:

These results demonstrate the feasibility of culturing cells and quantifying the inflammatory response to challenge outside the lab, with a wide range of potential applications in biobehavioral research in community-based and remote field settings.

Introduction

Inflammation is critical for initiating life-saving immune responses to injury and infection. However, recent studies have emphasized the role of dysregulated, non-resolving inflammation in the development of chronic degenerative diseases ranging from autoimmunity to cardiovascular disease (CVD) (1–5). Inflammation has also been shown to be influenced by a host of environmental factors, including psychosocial stressors, nutritional and microbial exposures, and socioeconomic status (6–8). As a result, there is tremendous interest in inflammation as a potential mechanism linking contexts and experiences to health and disease over the life course.

The most common approach to investigating inflammation in this context is to measure circulating cytokines or acute-phase proteins. However, these tonic measures are limited in that they provide only a snapshot of complex processes that characterize the inflammatory response. For example, they cannot capture how leukocytes react to and recover from challenge, which may be important for disease risk. Furthermore, many inflammatory mediators are released from multiple tissues, so it is not clear to what extent they reflect immune functions per se (9–11). All of these features hamper understanding of the role inflammation plays in biobehavioral processes.

In pursuit of a more dynamic approach to immune activity, some researchers have applied ex vivo cell culture methods, which capture leukocyte responses to a microbial challenge. Here, leukocytes are incubated in whole blood with microbial stimuli to measure the strength of the inflammatory cytokine response, and in some cases the effectiveness of negative-feedback regulation. Lipopolysaccharide (LPS), or endotoxin, is a commonly used ligand: It is part of the outer membrane of gram-negative bacteria and simulates a bacterial challenge. In addition, glucocorticoids (GC) are critical down-regulators of inflammation in vivo, and can be incubated along with LPS to assess the sensitivity of leukocytes to negative feedback. Impaired sensitivity to this feedback reflects glucocorticoid resistance (GCR).

Cell culturing has provided important insights into the role that psychological and social factors play in influencing inflammatory responses. For example, multiple studies have found that leukocytes of chronically stressed individuals produce more pro-inflammatory cytokines after incubation with LPS and GC, suggesting that prolonged stress-related GC exposure leads to the development of GCR (12–16). However, these studies have been conducted exclusively in laboratory-based settings that contain biological safety cabinets for preventing contamination and large immobile incubators with CO₂ control for pH regulation. Replicating these conditions in community-based settings is difficult, if not impossible, which has prevented wider implementation of cell culture protocols by researchers working in diverse field settings, or trying to engage harder to reach populations. This methodological constraint, in turn, limits investigations into how psychosocial contexts and experiences may shape the regulation of inflammation.

We aim to address this gap by developing two “field-friendly” protocols for ex vivo cell culturing, one using a small volume of venous whole blood, and another using finger stick capillary whole blood. We validated these methods in reference to a standard laboratory-based protocol, and compared results across protocols using a matched set of samples collected from young adults. Leukocytes were incubated with LPS and GC, and we quantified the production of pro-inflammatory cytokines IL-1β, IL-6 and TNFα in response. We find a high level of agreement in results across the protocols, demonstrating the feasibility of culturing cells and quantifying the cytokine response to challenge in non-laboratory settings.

Materials and Methods

Participants and study design

An opportunistic sample of N=25 adults was recruited for the purpose of methods evaluation. One participant had significant clotting during capillary blood collection which precluded sample transfer and was therefore excluded from analysis. Graphical representations were used to compare results, as well as statistical analyses of correlation. With N=24 and alpha=0.05, statistical power to detect correlations of 0.70 or greater was 0.98. Following the administration of informed consent, participants completed a brief demographic and biomedical survey, and then had antecubital (venous) and finger (capillary) blood collected. All data were collected under conditions of written informed consent with institutional review board approval from Northwestern University.

Portable and laboratory-based culture systems

To assess the feasibility of field-based approaches, an established lab-based protocol (standard culture, SC) (16) was compared to two field-based variants. The mini culture (MC) protocol is a field-adapted version of the established lab protocol, using a smaller quantity of venous blood and scaled-down requirements for reagents and equipment. The capillary blood (CB) protocol extends these simplifying features by using an even smaller quantity of blood, collected from the tip of a participant’s finger.

All three protocols were designed to capture two features of the inflammatory process: how strongly leukocytes respond to a common bacterial stimulus, and how sensitive they are to inhibitory signals that normally suppress that response. Response to the bacterial component was assessed by culturing whole blood with LPS (Invivogen; San Diego, CA) and quantifying production of a panel of pro-inflammatory cytokines, which included IL-1β, IL-6, and TNFα. Sensitivity to inhibitory signals was assessed by co-incubating the whole blood with LPS and GC (in the form of hydrocortisone; Sigma Aldrich; St. Louis, MO), which provides systemic anti-inflammatory feedback.

Standard Culture (SC):

Six mL of antecubital blood was collected in sodium-heparin Vacutainers (Becton-Dickinson, Franklin Lakes, NJ) and then immediately diluted to 90% v/v with R10 (360 uL blood, 40 uL R10; RPMI 1640 media, fetal bovine serum, L-glutamine, penicillin, streptomycin, Corning; Corning, NY). Diluted blood was then added in 400 μL aliquots to 24-well culture plates (Falcon, Tissue Culture Treated, Polystyrene Microplates, Corning; Corning, NY) containing either (a) LPS (50 ng/ml) or (b) LPS (50 ng/ml) and hydrocortisone, diluted in R10 to concentrations of 10⁻⁵ (Cort-5) or 10⁻⁶ M (Cort-6) (all doses reflect final in-well concentrations). A negative control well, where cells were incubated with an equivalent volume of R10 media alone, was included to measure background cytokine release. The final in-well dilution of whole blood was 72% v/v. Cultures were incubated for 4 hours at 37°C in 5% carbon dioxide, and then centrifuged at 17,000 x g for five minutes. Supernatants were harvested and frozen at −80°C.

Mini Culture (MC):

For this simplified version of the established protocol, we collected six mL of antecubital blood in sodium-heparin Vacutainers (Becton-Dickinson, Franklin Lakes, NJ). 250 μL aliquots were promptly dispensed into 1.7 mL microfuge tubes (Corning Axygen MCT-175-C-S; Corning, NY) containing endotoxin-free saline (9 g/dL NaCl, Quality Biological, Gaithersburg, MD) and either (a) LPS (50 ng/ml) or (b) LPS (50 ng/ml) and hydrocortisone at doses of 10⁻⁵ or 10⁻⁶ M (all doses reflect final in-well concentrations). A negative control tube was also included, where cells were incubated with saline alone. The final in-well dilution of whole blood was 72% v/v. Cultures were incubated for 4 hours at 37°C in sealed microfuge tubes placed in one of two portable incubators chosen randomly (Minitube #19180, Verona, WI, or Thermo Scientific #88871001, Waltham, MA). After 4 hours, the vials were removed from the incubator, and centrifuged at 17,000 x g for five minutes. Supernatants were then harvested and frozen at −80°C. As an additional experiment, we considered whether a simpler harvest procedure would yield similar results. Instead of centrifuging following incubation, tube contents were gently mixed and 65 μL transferred to #903 Protein Saver Cards (Whatman #1053612, Cytiva, Marlborough, MA) (17). The cards were allowed to dry for approximately 18 hours at room temperature, then sealed in gas impermeable bags with desiccant, and stored at −20°C.

Capillary Blood Culture (CB):

Finger sticks were performed with Microtainer contact-activated lancets (Becton-Dickinson #366594, Franklin Lakes, NJ), and 30 μL of capillary whole blood was collected in non-heparinized micropipettes (PTS Diagnostics, Whitestown, IN). To avoid clotting, capillary blood was then immediately transferred into a 1.5 mL sterile cryogenic vial (Nalgene #5000–1012, Waltham, MA) containing endotoxin-free saline (9 g/dL NaCl, Quality Biological, Gaithersburg, MD) and either (a) LPS (29.8 ng/ml) and sodium heparin (21.4 USP/mL, endotoxin-free, Sigma-Aldrich, St. Louis, MO) or (b) LPS (29.8 ng/ml), sodium heparin (21.4 USP/mL), and hydrocortisone at doses of 5.96 × 10⁻⁶ or 5.96 × 10⁻⁷ M (all doses reflect final in-well concentrations). A negative control well was also included, where cells were incubated with sodium heparin (21.4 USP/mL) in saline alone. The final in-well dilution of whole blood was 233% v/v. Cultures were incubated for 4 hours at 37°C in sealed cryogenic vials placed in the incubators described above. After 4 hours, the vials were removed, and the tube contents gently mixed prior to transferring 65 μL to 903 Protein Saver Cards (Whatman #1053612, Cytiva, Marlborough, MA). Cards were allowed to dry for approximately 18 hours at room temperature, then sealed in gas impermeable bags with desiccant, and stored at −20°C.

As is evident, the simplified MC and CB protocols differed in several respects from the established SC protocol. First, in the simplified protocols whole blood was diluted in saline rather than R10 media, because saline is easier to access and store than R10 media, which requires refrigeration. Second, the SC protocol was performed in a level-2 biosafety cabinet, whereas the simplified culture protocols were conducted on the open benchtop, which more closely resembles the non-sterile conditions in field settings. Third, in the simplified culture systems, cells were cultured in small microfuge tubes, rather than twenty four-well plates, because these tubes were compatible with the portable incubators. These incubators are readily applied in field settings because unlike a standard laboratory device, they weigh just a few pounds, and can be powered via rechargeable batteries, or through an adapter that plugs into an automobile cigarette lighter. Obviously, portable incubators lack a CO₂ supply, so they do not maintain pH as consistently as a conventional incubator in a laboratory setting.

Finally, the endpoint dilution of whole blood in the CB protocol differed from the other protocols. With limited finger stick blood volume, the need to prevent the specimens from clotting, and the need to transfer a sufficient volume to filter paper for subsequent cytokine analyses, the CB protocol applied a higher dilution of whole blood than the other protocols. However, it is important to note that the ligand concentrations across all three culture protocols were identical when normalized to the volume of whole blood used in each protocol. Although these variations are important to highlight, we emphasize that they introduce a conservative bias into the comparisons we present. In other words, the protocol variations should, if anything, result in more discrepant rather than similar findings.

Repeatability of responses

To evaluate the repeatability of the field-adapted protocols, we conducted a parallel set of experiments. From three separate individuals, we collected larger volumes of antecubital blood, and divided each into five separate aliquots. The CB and MC culture protocols were then performed five independent times with these aliquots. We considered the LPS and LPS + Cort-6 condition, and omitted the LPS + Cort-5 condition in order to conserve resources.

Reagent stability

Field-based investigators face logistical challenges associated with accessing, collecting, and processing biological materials. They may need to prepare reagents before entering the field, and once there, might lack access to refrigeration. Thus, we conducted another set of experiments to determine the viability of reagents that had been prepared in advance and stored for varying lengths of time. At the outset of the project, we prepared a large set of culture reagents and stored them at −20°C. At multiple points over the next eight weeks, we transferred the reagents to room temperature conditions, protected from light. This procedure yielded reagent pools that had been removed from freezer storage 8, 6, 4, 3, 2, and 1 weeks before MC and CB culture protocols were performed. The baseline condition used reagents that had been removed from freezer storage the day the cultures were performed. The MC and CB culture protocols were then performed on the same day, with venous blood from three individuals (the large volume of blood precluded the use of capillary blood in the CB culture tubes). We considered the LPS and LPS + Cort-6 condition, and omitted the LPS + Cort-5 condition in order to conserve resources.

Quantification of cytokine response

The samples were assayed for IL-1β, IL-6 and TNF-α in duplicate, using a multiplex fluorescence immunoassay (Luminex Performance Human XL Cytokine Discovery Panel; Bio-techne) on a Luminex MAGPIX instrument (Austin, TX). Supernatant samples were subjected to a 1/20 dilution with Assay Diluent provided by the manufacturer. Samples stored as DBS on Protein Saver Cards were analyzed using an updated version of a previously validated DBS protocol for Luminex (18, 19). For each DBS sample, a single 5 mm punch was removed and eluted in 120 μL Assay Diluent (~1/22 dilution) for a period of approximately 16 hours at 4°C. On the day of the multiplex assay, 50 uL of eluate was transferred in duplicate to the assay plate. For duplicate pairs, the average intra-assay coefficients of variation were 1.59%, 1.61%, and 1.81% for IL-1β, IL-6, and TNFα, respectively. The corresponding inter-assay coefficients of variation were 4.03%, 2.87% and 2.67%. Assay sensitivity for IL-1β, IL-6, and TNFα are reported by the manufacturer as 2.33, 6.75, and 8.55 pg/mL, respectively. However, the data reduction software (xPONENT 4.2 for MAGPIX, Luminex) will extrapolate values below the lowest calibrator, except for fluorescence readings approaching background. For these determinations we assigned a value at the midpoint between zero and the lowest returned value for each cytokine. All stimulated samples produced cytokine concentrations that were well above the lowest calibrator.

Collectively, these experiments yielded an enormous amount of data, which raises concerns about false discovery. To alleviate these concerns, and streamline the analyses, we made an a priori decision to aggregate cytokine concentrations. Therefore, within each culture protocol, values of IL-1β, IL-6, and TNFα were transformed into z-scores (mean = zero, SD =1), and then summed into a composite variable. The composites displayed high levels of internal consistency (Cronbach’s alpha = 0.71, 0.77, and 0.81 for SC, MC, and CB cultures, respectively), reflecting strong inter-correlation among the cytokine responses. Results for individual cytokines are reported as supplementary material.

Analysis of culture performance

Agreement in results across culture protocols was evaluated graphically with scatterplots and Bland-Altman plots of agreement (20), as well as well as least squares and and Passing Bablok regression (21, 22). Multi-level regression models were used to calculate correlations that accounted for the clustering of observations within individuals. Passing Bablok regression models are non-parametric and robust to outliers, and are commonly used for methods comparison studies. Bland-Altman plots consider the difference in results as a function of the mean result for matched samples across protocols, and we inspected the 95% limits of agreement across the measurement range for evidence of bias or inconsistent variability.

The repeatability of results for each culture protocol was operationalized as intra-culture variation, or imprecision, and calculated as the percent coefficient of variation (CV; SD/mean x 100) for five replicates of the LPS and LPS + 10⁻⁶ hydrocortisone conditions for the MC and CB protocols.

Reagent stability was analyzed by plotting cytokine response as a function of storage time and visually inspecting for decay in cytokine response. Least squares regression analyses were used to test for statistically significant trends. Storage time in weeks was modeled as a continuous variable, as well as a categorical variable to consider non-linear trends, with the “Day 0” baseline condition as the reference.

With the exception of initial descriptive results (Figure 1, Table S1), all analyses are based on cytokine response variables that account for background cytokine production. Specifically, for each culture protocol and condition, the level of cytokine production in the unstimulated condition was subtracted from the cytokine concentration in the experimental conditions. All statistical analyses were conducted with Stata/SE for Windows, version 15.1 (StataCorp, College Station, TX).

Figure 1.

Box plot of cytokine response across culture protocols and ligands. Median (25^th percentile, 75^th percentile) standardized composite cytokine value is represented by each box.

LPS: LPS only condition

LPS CRT-5: LPS with hydrocortisone 10⁻⁵

LPS CRP-6: LPS with hydrocortisone 10⁻⁶

UNSTIM: unstimulated control condition

Results

The protocols were evaluated with venous and finger stick blood samples collected from 24 participants (mean age = 29.2 years, range 19 to 50; 29% male). Figure 1 presents the pattern of cytokine response across culture protocols and ligands, using the aggregate cytokine composite. Disaggregated results for IL-1β, IL-6, and TNF-α are presented in Table 1. Responses within each culture protocol were as expected, with substantial cytokine production in the LPS-treated conditions, and negligible production in the unstimulated condition. Furthermore, cytokine responses were substantially attenuated in the presence of glucocorticoids, with greater attenuation in the higher dose condition (hydrocortisone 10⁻⁵).

Table 1.

Mean IL-1β, IL6, and TNFα concentrations (pg/mL) for each ligand, across cell culture protocols.

Standard culture	IL1-β		IL6		TNFα
	mean	sd	mean	sd	mean	sd
LPS	6211.7	2444.2	94046.3	19352.7	43703.5	9562.9
LPS CRT-5	1539.4	518.0	34464.8	6468.8	16089.4	4790.0
LPS CRT-6	2320.1	727.7	49521.5	10646.7	22786.1	5867.9
UNSTIM	30.6	65.9	277.2	272.7	549.5	421.1
Mini culture	IL1-β		IL6		TNFα
	mean	sd	mean	sd	mean	sd
LPS	12877.1	4168.4	86885.4	21436.0	41441.1	9922.3
LPS CRT-5	7383.5	3010.9	38497.9	10156.0	21735.8	5302.1
LPS CRT-6	8438.6	3831.0	54167.1	12820.4	28089.5	6220.9
UNSTIM	143.4	163.1	1323.0	596.3	2185.0	802.8
Capillary blood	IL1-β		IL6		TNFα
	mean	sd	mean	sd	mean	sd
LPS	6952.6	2280.2	35580.7	9896.0	19392.9	5996.2
LPS CRT-5	2815.4	789.1	14382.3	4729.8	8121.8	2771.4
LPS CRT-6	3669.5	1131.1	19606.8	6785.1	11136.3	3497.3
UNSTIM	769.8	651.5	869.6	554.4	1342.2	795.5

LPS: LPS only condition

LPS CRT-5: LPS with hydrocortisone 10-5

LPS CRT-6: LPS with hydrocortisone 10-6

UNSTIM: unstimulated control condition

Subsequent analyses used results with background levels of cytokine production—as measured in the unstimulated control wells—subtracted out for each condition. Cytokine responses were quite similar across culture protocols. Aggregating across conditions, the correlation between SC and MC results was 0.88 (p<0.001, n=70). The corresponding correlation between SC and CB results was 0.86 (p<0.001, n=70). Figure 2 depicts the concordance graphically in scatterplots. When the cytokines were considered independently, the correlations across culture protocols were similarly strong (Table S1).

Figure 2.

Scatterplot and Passing Bablok regression line (95% confidence interval) of the association between standard culture and mini culture results (top), and between standard culture and capillary blood culture results (bottom), using the standardized composite cytokine response variable.

As Figure 2 (top) illustrates, the pattern of association between MC and SC results was linear across the range of values. However, cytokine production in the MC protocol trended lower than SC results at the high end of the measurement range, as indicated by the Bland-Altman plot of agreement (Figure S1) and the Passing Bablok regression equation, which had a positive intercept and slope less than one.

Similarly, Figure 2 (bottom) indicates that the association between CB and SC results was linear across the measurement range, with no systematic shifts in relative responses across the protocols. The interval between the upper and lower 95% limits of agreement, as indicated in the Bland-Altman plot (Figure S2) was 1.65. The interval for the mini culture protocol was 1.55, indicating a slightly higher level of variation for the capillary blood protocol in comparison with the standard culture protocol.

In some field settings, access to laboratory equipment needed for typical culture harvesting techniques may be limited. We therefore evaluated a simplified cell culture processing protocol in which we transferred the post-incubation contents of culture tubes—prior to centrifugation—directly to filter paper for storage as DBS. As shown in Figure 3, DBS and supernatant yielded very similar estimates of aggregate cytokine production, with a correlation of 0.91 (p<0.001, n=70). Correlations were also very high for individual cytokines (Table S2). The pattern of association was linear across the range of measurement, with no evidence for bias or shifts in values in the Passing Bablok regression equation or the Bland-Altman agreement plot (Figure S3).

Figure 3.

Scatterplot and Passing Bablok regression line (95% confidence interval) of the correspondence between the standardized composite cytokine response for mini culture samples processed as supernatant and dried blood spots (DBS).

Finally, we evaluated the repeatability of cytokine production in the MC and CB protocols by conducting the experiments five times in three participants. We calculated the percent coefficient of variation (%CV; SD/mean x 100) for the composite cytokine variable for each condition and participant (Table S3). Repeatability was high for both protocols: For MC, %CV values ranged from 2.5 to 9.00, with a mean of 6.2 across participants and conditions. For the CB protocol, mean %CV was 5.4, with a range of 0.3 to 9.9. Figure S4 presents these results graphically, demonstrating relatively tight clusters of results around the mean for each participant.

The ability to prepare reagents in batches, and store them in ambient environmental conditions, may facilitate application in some field-based settings. We therefore examined how cytokine production varied when reagents had been stored at room temperature for periods of 8, 6, 4, 3, 2, and 1 weeks before the culture protocols were performed. For MC, there was no consistent pattern of change in results in either the LPS or LPS + Cort-6 conditions (Figure S5). Regression analyses confirmed that storage time (weeks) was not a significant predictor of composite cytokine response (LPS: B=−0.02, 95% CI=−0.12, 0.08, p=0.71); LPS + Cort-6: B=−0.00, 95% CI=−0.07, 0.07, p=0.99).

Longer storage times were associated with lower cytokine production in the CB protocol (Figure S6). This pattern was evident in both the LPS (B=−0.32, 95% CI=−0.39, −0.26, p<0.001) and LPS + Cort-6 conditions (B=−0.19, 95% CI=−0.24, −0.15, p<0.001). When storage time was modeled as a categorical variable with zero weeks as the reference, cytokine responses in both the LPS and LPS + Cort-6 conditions did not decay significantly in tubes stored up to 3 weeks (all p>0.05).

Discussion

Inflammation is a dynamic and responsive system that plays a central role in host defense against injury and infection, but it also has the potential to contribute to the pathophysiology of a wide range of chronic degenerative diseases (1, 3, 4). It is therefore important to understand how inflammation is regulated, yet epidemiological and other field-based studies typically rely on one-time, baseline measures of inflammatory biomarkers (23). Cell culture systems are routinely used in immunology as an ex vivo model for quantifying the inflammatory response to challenge, but standard protocols require large volumes of venous blood and access to laboratory infrastructure for sample processing. We aimed to bridge these gaps by developing a “field-friendly” approach to cell culture that can be applied across a wide range of biobehavioral and community-based research settings.

We developed two versions of the protocol: MC represents a scaled down version of the standard cell culture approach for use with venous blood, whereas the CB protocol requires only a few drops of blood from a simple finger stick. Both versions provided results that were comparable to the gold standard SC protocol, with large increases in cytokine production in the presence of LPS, negligible production in the absence of LPS, and attenuation of response to LPS in the presence of glucocorticoids. Furthermore, the pattern of association in results was linear across the measurement range, with high levels of agreement in relative responses across the protocols. The level of repeatability was comparably high for all protocols. In short, an investigator can expect to generate a similar pattern of results, and draw the same substantive conclusions, from the MC and CB protocols as the SC protocol.

In some settings, the laboratory infrastructure needed for harvesting cultures may be limited. We therefore evaluated a simplified approach to processing cultures, and determined that transferring the contents of culture tubes directly to filter paper for storage as DBS provided results that were virtually identical to results from cultures harvested as supernatants. DBS sampling is widely used in newborn screening, as well as field-based research in anthropology, demography, and epidemiology, because samples can be collected, transported, and stored at low cost and low biohazard risk (17, 24). These advantages may further increase the feasibility of applying cell culture protocols outside of the laboratory setting.

With that aim in mind we also considered the possibility of preparing MC and CB reagents in batches in advance of use. Longer storage times did not influence the pattern of MC results. However, there was clear evidence of reduced potency with longer storage times for the CB cultures. Further investigation is required to determine why the reagents were less stable in the CB condition. Nonetheless, the current results indicate that CB reagents can be safely prepared up to 3 weeks in advance of blood collection, and still yield findings comparable to same-day preparation.

It is worth noting that even though our validation was limited to LPS, the MC and CB protocols can likely be adapted for other ligands, so responses to other pathogen-associated and danger-associated molecular patterns can be interrogated (e.g., viruses, cell death signals, metabolic byproducts). Furthermore, longer incubations can be added to assess patterns of response at different time points, and different outcomes can be assessed (e.g., additional cytokines, mRNA expression). Our validation has demonstrated the feasibility of miniaturized cell cultures with small volumes of blood, in the absence of CO₂ incubators and other laboratory infrastructure. As such, it can serve as a foundation for more widespread application beyond LPS and the activation of inflammation.

However, it is important to acknowledge that our protocol relies on fixed volumes of whole blood, rather than fixed numbers of cells. Therefore, this approach may not be appropriate for studies focused on localizing cytokine release to specific leukocyte subsets. While localization is not the goal for most studies, it is still important to recognize the heterogeneity of cellular composition in peripheral blood, and the possibility that differential responses to stimulation protocols could reflect sample-to-sample variation in leukocyte subset content and/or per-cell cytokine production. This situation is characteristic of all whole blood assays, and in some cases can be addressed by performing a complete blood cell count and differential—which can be implemented at the point-of-care (25)—with subsequent statistical adjustment for leukocyte populations. Another potential limitation is our validation of a short incubation period: For applications requiring longer incubations it may be important to control CO2 levels and ensure the sterility of the cultures, which may be more difficult to achieve outside of the lab.

We envision three scenarios where the MC and CB protocols have the potential to extend the reach of current laboratory-based approaches to investigating the regulation of inflammation.

Scenario one: Clinical setting.

Researchers working in a clinic can readily draw venous blood, but do not typically have access to the expertise and equipment of an immunology lab. In this scenario, implementing the MC protocol is straightforward: After routine venipuncture, small volumes of heparinized whole blood are transferred into MC tubes, which are incubated at 37°C in a compact incubator on the bench top. After four hours, the cultures are transferred to filter paper as DBS, or centrifuged for harvesting of supernatant. Samples are stored frozen prior to shipping or analysis.

Scenario two: Community setting.

Here, research takes place in a centralized location and venous blood collection may be possible. In this case the MC protocol can be implemented as in scenario 1. If venipuncture is not possible, then the CB protocol can be implemented with finger stick blood samples. Cultures are incubated on site and then transferred to filter paper as DBS for transport to the lab.

Scenario three: Remote setting.

In this situation, samples are collected in the home or nearby in the community. Access to electricity may be limited, and cultures may need to be transported prior to the end of incubation. Here, the CB protocol begins with the transfer of finger stick blood to culture tubes. Cultures are incubated in a portable unit powered by a rechargeable battery or automobile charger. After incubation, cultures are transferred to filter paper as DBS for storage until they can be transported to a freezer.

Developmental plasticity and ecological sensitivity are defining features of the human immune system (6, 26, 27). Therefore, it is important to study immune function—and inflammation—in the field, and not just in the lab. For example, biological anthropologists working in remote international settings have shown that microbial and nutritional environments in infancy shape inflammation in adulthood (28), and that elevated inflammation does not predict cardiovascular disease risk in the same way as in the US (29). Similarly, life course approaches in the US and Canada, drawing on community-based samples, document long term impacts of harsh family environments and socioeconomic adversity on the regulation of inflammation in adulthood (16, 30–32). With the field-friendly protocols developed here, these studies can be extended beyond “snapshots” of inflammatory biomarkers to consider how social and physical environments shape leukocyte responses to challenge. Because this study focused on methods development, we did not collect biobehavioral or psychosocial measures from participants. Follow up studies—building on a solid foundation of lab- and clinic-based research—are needed to investigate how cytokine responses covary with biobehavioral and psychosocial characteristics across a wider range of populations and research settings. Furthermore, these methods can help advance scientific understanding of the roles that genetic and epigenetic factors may play in these processes (33, 34).

Much remains to be learned about the determinants of inflammation and its links to disease. Dynamic approaches that investigate the inflammatory response to challenge have the potential to generate important insights into the regulation of inflammation and the factors that shape it over the life course. Our hope is that the implementation of cell culture protocols in diverse research settings in the community, around the world, will catalyze that effort.

Acronyms

CVD

cardiovascular disease

LPS

lipopolysaccharide

GC

glucocorticoids

GCR

glucocorticoid resistance

IL-1β

interleukin-1 beta

IL-6

interleukin-6

TNFα

tumor necrosis factor-alpha

SC

standard culture

MC

mini-culture

CB

capillary blood culture