Profiling migration of human monocytes in response to chemotactic and barotactic guidance cues

Summary

Monocytes are critical to innate immunity, participating in chemotaxis during tissue injury, infection, and inflammatory conditions. However, the migration dynamics of human monocytes under different guidance cues are not well characterized. Here, we developed a microfluidic device to profile the migration characteristics of human monocytes under chemotactic and barotactic guidance cues while also assessing the effects of age and cytokine stimulation. Human monocytes preferentially migrated toward the CCL2 gradient through confined microchannels, regardless of donor age and migration pathway. Stimulation with interferon (IFN)-γ, but not granulocyte-macrophage colony-stimulating factor (GM-CSF), disrupted monocyte navigation through complex paths and decreased monocyte CCL2 chemotaxis, velocity, and CCR2 expression. Additionally, monocytes exhibited a bias toward low-hydraulic-resistance pathways in asymmetric environments, which remained consistent across donor ages, cytokine stimulation, and chemoattractants. This microfluidic system provides insights into the unique migratory behaviors of human monocytes and is a valuable tool for studying peripheral immune cell migration in health and disease.

Graphical abstract

Highlights

• •The MAP chip profiles migration of human monocytes under various chemotactic and barotactic cues
• •Monocytes preferentially migrate toward CCL2 gradients, regardless of migration pathway and donor age
• •IFN-γ reduces human monocyte chemotaxis, velocity, and CCR2 expression
• •Human monocytes show biased migration toward low-hydraulic-resistance pathways

Motivation

Cell migration is fundamental to the biological processes that drive health and disease. While in vivo models provide invaluable insights into cell migration within complex biological environments, precise control over the microenvironment and single-cell tracking is essential to deepen our understanding of the fundamental characteristics of cell migration. We present a high-throughput microfluidic platform, termed the migration analysis of peripheral immune cells (MAP) chip, that features four distinct sets of microchannels designed to assess the effects of both chemotactic and barotactic stimuli on cell migration at a single-cell level. By profiling human monocyte migration using the MAP chip, we demonstrated the utility of this device in characterizing migration of human monocytes under diverse conditions.

Hall et al. introduce the MAP chip, a microfluidic platform for profiling human monocytes under chemotactic and barotactic guidance cues. It reveals biased migration toward low-hydraulic-resistance pathways, disrupted migration upon cytokine stimulation, and consistent chemotaxis and barotaxis across donor ages—enhancing our understanding of human monocyte migration characteristics.

Introduction

Monocytes play a critical role in innate immunity function, response, and resolution of many inflammatory disorders.^1,2 Upon tissue injury or inflammation, monocytes rapidly migrate to target sites, differentiate into macrophages, engulf cellular debris and pathogens, and release various cytokines that further drive the adaptive immune response. This intricate process is facilitated by chemotaxis—a process in which immune cells migrate along a chemotactic gradient released by tissue or other immune cells.^3,4 Animal and human studies have shown that while the recruitment and activation of monocytes can resolve acute threats, it can also aggravate existing pathologies, such as cancer,^2,5,6 neurodegenerative diseases,⁷ and autoimmune conditions.^8,9 Similarly, monocyte dysfunction has been implicated in aging. Monocytes from aged individuals show impaired proinflammatory cytokine responses¹⁰ and phagocytosis.¹¹ In neurologic disorders, immune cells infiltrate the brain through (compromised) brain barriers,^12,13,14 and monocyte-derived macrophages emerge as the dominant infiltrating immune cells, as observed in brain lesions of patients with multiple sclerosis (MS).^15,16 Among various chemokines, monocyte chemoattractant protein-1 (MCP-1/CCL2) is well known for guiding the chemotaxis of myeloid cells. CCL2 orchestrates the migration of monocytes not only during regular immune defense but also in a spectrum of pathological conditions, including MS and Alzheimer’s disease (AD).^17,18,19 The dual role of monocytes in the body underscores the importance of understanding and characterizing this critical innate immune cell. Additionally, gaining a better understanding of the impact of CCL2 on monocytes will play a pivotal role in enhancing current therapeutics. However, the principles that govern monocyte trafficking in various conditions remain relatively elusive.

Proinflammatory cytokines are key regulators of innate and adaptive immunity, driving myeloid cell activation and migration to inflammation sites.^20,21 Their role extends to various neuroinflammatory disorders. Among these cytokines, interferon (IFN)-γ, which belongs to the type II IFN protein family,²² is elevated in tissues, plasma, and cerebrospinal fluid (CSF) across neurologic conditions like AD, MS, Parkinson’s disease (PD), and amyotrophic lateral sclerosis (ALS).^{23,24,25,26,27,28,29,30,31,32,33} Yet, its precise biological functions and impact on monocyte migration are underexplored. Chronic inflammatory diseases are also influenced by the dysregulation of other proinflammatory cytokines, such as granulocyte-macrophage colony-stimulating factor (GM-CSF). GM-CSF operates in both innate and adaptive immunity, correlating with disease severity in neuroinflammatory and autoimmune disorders.^{34,35,36,37,38} With IFN-γ and GM-CSF impacting multiple diseases and biological processes, they are promising therapeutic targets. Nevertheless, the specific impact of IFN-γ and GM-CSF on monocyte migration signatures, whether systemically or locally within specific tissues, remains unclear.

Migration of immune cells is also influenced by various factors such as hydraulic resistance (barotaxis), matrix stiffness (durotaxis), and the geometric properties of the microenvironment (topotaxis).^4,39,40 In the presence of pathology, the microenvironment undergoes significant changes,^41,42 altering the chemical and physical cues that drive immune cell migration. Understanding how these alterations impact immune cell migration, particularly the innate immune cells, holds promise for devising therapeutic strategies to steer immune responses. Notably, it has been shown that neutrophils prioritize migration paths with low hydraulic resistance,⁴³ independent of chemotactic signals, providing crucial evidence of how competing microenvironmental cues can alter immune cell migration.⁴³ However, there are limited studies focusing on how a combination of these factors impacts human monocyte migration, partly due to the lack of a suitable model that enables studying the effects of multiple stimuli simultaneously. In vivo models offer invaluable insights into immune cell behavior within their native and intricate microenvironments. However, these models are limited in controlling specific aspects of the microenvironment and profiling immune cell migration at a single-cell level. In contrast, microfluidic systems provide powerful platforms for studying immune cell migration ex vivo under well-controlled experimental conditions,^{43,44,45,46,47,48,49} thereby facilitating the quantitative analysis of immune cell migration patterns.

Here, we developed a microfluidic system for migration analysis of peripheral immune cells (hereafter referred to as the MAP chip). Using the MAP chip, we explored migration characteristics of human monocytes in response to chemotactic (CCL2 and CCL5) and barotactic cues. We also investigated the impact of donor age and proinflammatory cytokine stimulation (IFN-γ and GM-CSF) on monocyte decision-making within the MAP chip. We discovered that human monocytes maintain their chemotactic ability and bias toward low-hydraulic-resistance paths regardless of donor age. However, stimulation with IFN-γ (but not GM-CSF) impairs the chemotaxis and speed of human monocytes in the MAP chip. Mechanistically, we found that IFN-γ stimulation significantly diminished monocyte CCR2 receptors, thereby hindering their CCL2 chemotaxis. Collectively, these data delineate distinct and/or biased migration patterns of human monocytes that are further modulated by chemical and physical cues. This exemplifies the utility of the MAP chip for studying migration patterns of human monocytes and other peripheral immune cells in response to various cues in health and disease.

Results

Development of the MAP chip

We engineered and validated a microfluidic system (the MAP chip) that enables quantitative analysis of peripheral immune cell migration behaviors ex vivo. The MAP chip contains four different compartments, including chemotaxis, chemotactic maze, dual-taxis, and distance dual-taxis microchannels (Figure 1A). The microchannels in each compartment contain a chemokine gradient between the CHEMOKINE chambers (higher chemokine concentration) and the CENTRAL immune cell loading chamber (lower chemokine concentration). The microchannels leading away from the CENTRAL chamber are 5.3 μm in height and 6 μm in width, creating a consistent, initial spatial environment for migrating monocytes. Then, in the dual-taxis and distance dual-taxis designs, the width of the microchannels varies (i.e., 3.8, 6.0, or 10 μm) to alter the hydraulic resistance. In the maze microchannels, the height and width of the microchannels remain constant at 5.3 × 6 μm. However, monocytes can take a short path (∼738 μm) or a long path (∼1,020 μm) to reach the CHEMOKINE chamber, with multiple dead-end paths on either route. Using the MAP chip, we tested the impact of donor age and proinflammatory cytokine stimulation on human monocyte migration characteristics. Specifically, CD14⁺ monocytes were isolated from human peripheral blood and loaded into the CENTRAL chamber of the MAP chip primed with 100 nM CCL2 or CCL5 chemokine. Monocyte directional migration was monitored using time-lapse confocal imaging for up to 18 h (Figure S1A). Next, the data were quantified using IMARIS software in an unbiased and semi-automatic manner (Figure 1B). We also modeled the CCL2 gradient using 10,000 molecular weight (MW) dextran fluorescein, which is a close match in size to the recombinant CCL2 used herein (∼10 kDa). This approach allowed us to directly image and quantify gradient formation and stability over 15 h in the MAP chips (Figures S1B and S1C; Videos S1, S2, S3, and S4). Gradients remained stable throughout the MAP chip over the time course, with changes in gradient over time corresponding directly to the number of microchannels. Based on the modeling concentration from fluorescence intensity, the gradient from the reservoir to exit is 25 nM per 100 μm at time 0 for all devices, which reduces to around 15 nM per 100 μm by 15 h for the chemotaxis design (11 microchannels per reservoir) but is maintained at ∼23.5 nM per 100 μm for the maze design (2 microchannels per reservoir) (Figure S1C).

Figure 1.

A MAP chip for quantitative analysis of immune cell migration characteristics at a single-cell level

(A) The MAP chip contains four different sets of microchannels: chemotaxis, chemotactic maze, dual taxis, and distance dual taxis. Monocytes are loaded into the CENTRAL chamber of a primed MAP chip and can migrate through the microchannels into the CHEMOKINE chambers.

(B) An experimental pipeline illustrating how human monocytes from different donor groups were isolated, stimulated if applicable, and loaded into the MAP chip primed with either CCL2 or CCL5 chemokine. Cell motility was then tracked at a single-cell resolution using time-lapse imaging. The data were quantified to detail monocyte migration characteristics. Additionally, flow cytometry was performed on the monocytes to quantify specific receptor expression levels.

Video S1. Concentration gradients in the chemotaxis microchannels, related to Figure 1

Dextran fluorescein (500 mg/mL) was loaded into the MAP chip, and the concentration gradient was imaged over 15 h. Scale bar, 20 μm.

Video S2. Concentration gradients in the dual-taxis microchannels, related to Figure 1

Dextran fluorescein (500 mg/mL) was loaded into the MAP chip, and the concentration gradient was imaged over 15 h. Scale bar, 20 μm.

Video S3. Concentration gradients in the distance dual-taxis microchannels, related to Figure 1

Dextran fluorescein (500 mg/mL) was loaded into the MAP chip, and the concentration gradient was imaged over 15 h. Scale bar, 20 μm.

Video S4. Concentration gradient in the maze microchannels, related to Figure 1

Dextran fluorescein (500 mg/mL) was loaded into the MAP chip, and the concentration gradient was imaged over 15 h. Scale bar, 20 μm.

Human monocytes maintain CCL2 chemotaxis regardless of donor age

To assess human monocyte chemotaxis, a recombinant human CCL2 protein (100 nM) was loaded into the MAP chip to generate a chemotactic gradient (Figure 2A). Monocytes were isolated from healthy donors, loaded into the primed MAP chip, and imaged using time-lapse confocal microscopy for up to 18 h (Figure 2B) to probe their migration behavior in response to the CCL2 chemokine (Figure 2C; Video S5) or a media control of Iscove’s modified Dulbecco’s medium (IMDM) supplemented with 20% v/v fetal bovine serum (FBS) (Figure 2D). To quantify the migration of human monocytes in the chemotaxis microchannels, the number of monocytes that migrated into the CHEMOKINE chamber was normalized to the total number of cells loaded into the CENTRAL chamber (see the STAR Methods). As expected, human monocytes showed significant CCL2 chemotaxis (p < 0.0001, n = 10 donors) compared to IMDM +20% FBS (Figure 2E). Monocyte velocity in the chemotaxis microchannels averaged 6.36 ± 0.71 μm/min (n = 10 donors; Figure 2F). Additionally, we confirmed that a CCL2 chemokine gradient, as opposed to the presence of evenly distributed CCL2, was necessary to induce significant monocyte chemotaxis in these microchannels (p < 0.0001; Figure 2G). This indicates that the CCL2 chemokine gradient in the MAP chip effectively induces directional migration of human monocytes.

Figure 2.

IFN-γ stimulation, not aging, significantly diminishes monocyte CCL2 chemotaxis

(A) Schematic showing the chemotaxis microchannel design, which establishes a chemotactic gradient along a 400 μm channel for directional migration of monocytes.

(B) Representative time-lapse confocal microscopy imaging of monocytes (cell membrane, green; nuclei, blue) in the chemotaxis microchannels primed with 100 nM CCL2 chemokine. Scale bar, 10 μm.

(C and D) Representative imaging with tracks showing monocyte migration in the chemotaxis microchannels in the presence of CCL2 chemokine (C) or control of media (D). Scale bar, 100 μm.

(E and F) Quantification of human monocyte migration (E) and migration velocity (F) in response to 100 nM CCL2 chemokine (n = 10 deidentified healthy donors).

(G) The number of migrating monocytes in the presence of CCL2 gradient or evenly distributed chemokine (n = 6 independent MAP chips).

(H and I) Quantification of human monocyte migration (H) and migration velocity (I) from 19- to 27-year-old (n = 4–5) and 50- to 60-year-old (n = 5-6) healthy donors.

(J) Quantification of human monocyte migration and migration velocity upon stimulation with GM-CSF or IFN-γ (n = 8–16 donors for control and n = 6–8 donors for stimulated monocytes).

p values are from Mann-Whitney test (E and H), Welch’s t test (G and I), one-way ANOVA (J, left), and Brown-Forsythe and Welch ANOVA tests (J, right). Boxplots show the median and the range between the 25th and 75th percentiles. The whiskers stretch from the minimum and maximum values.

Video S5. Directional monocyte migration in the chemotaxis microchannels, related to Figure 2

Monocytes were stained (blue, nucleus; green, cell membrane) and loaded into the MAP chip. Representative time-lapse imaging of monocytes navigating the chemotaxis microchannels. Migration is shown over approximately 45 min. Scale bar, 30 μm.

We then asked whether donor age impacts monocyte chemotaxis in response to the CCL2 chemokine. To this end, monocytes from 19- to 27- and 50- to 60-year-old donors were isolated from human peripheral blood mononuclear cells (PBMCs) and tested for CCL2 chemotaxis using the MAP chip. Intriguingly, we discovered that donor age had no significant impact on monocyte chemotaxis in response to 100 nM CCL2 chemokine in the MAP chip (n = 5 donors for 19- to 27-year-olds and n = 6 donors for 50- to 60-year-olds; Figure 2H). We observed that the 19- to 27-year-old and 50- to 60-year-old donors had average normalized monocyte migrations of 0.00046 ± 0.00028 and 0.00067 ± 0.00021, respectively. In the 50- to 60-year-old donor group, we observed a lower number of migrated monocytes from females compared to males (not significant, data not shown), a trend that was not detected in the 19- to 27-year-old donor group. Additionally, we observed no significant difference in monocyte migration velocity toward 100 nM CCL2 chemokine between the two age groups (Figure 2I). Monocytes isolated from 19- to 27-year-olds (n = 4 donors) and 50- to 60-year-olds (n = 5 donors) migrated toward the CCL2 chemokine at average velocities of 6.1 ± 0.68 and 6.3 ± 0.53 μm/min, respectively (Figure 2I). Collectively, CCL2 induced a robust directional migration of human monocytes in the MAP chip regardless of donor age.

IFN-γ (but not GM-CSF) stimulation impairs CCL2 and CCL5 chemotaxis of human monocytes

Next, we tested whether stimulation with proinflammatory cytokines, including GM-CSF and IFN-γ, impacts monocyte chemotaxis toward 100 nM CCL2 or CCL5 chemokine in the MAP chip. Monocytes from healthy donors were added to the MAP chip, where they were stimulated with 10 ng/mL GM-CSF or 50 ng/mL IFN-γ according to previously published reports^50,51,52 and profiled for chemotaxis. While stimulation with GM-CSF slightly decreased the directional migration of monocytes compared to unstimulated monocytes, we observed no statistical difference between the two groups in response to the CCL2 chemokine (n = 8 donors; Figure 2J). Similarly, GM-CSF stimulation did not significantly impact human monocyte chemotaxis toward the CCL5 chemokine compared to the unstimulated cells (n = 8 donors; Figure S2A). Notably, monocytes stimulated with IFN-γ showed 3.2- (p = 0.0155, n = 8 donors) and 2-fold (not significant, n = 8 donors) decreases compared to unstimulated monocytes in CCL2 and CCL5 chemotaxis, respectively (Figures 2J and S2A).

Moreover, stimulation with IFN-γ, but not GM-CSF, significantly decreased monocyte migration velocity (p < 0.0001) in response to 100 nM CCL2 chemokine compared to unstimulated monocytes (n = 6 donors; Figure 2J). Unstimulated monocytes migrated at an average velocity of 6.9 ± 0.50 μm/min, while GM-CSF- and IFN-γ-stimulated monocytes migrated at average velocities of 5.4 ± 0.42 and 3.6 ± 0.19 μm/min, respectively. GM-CSF-stimulated monocytes (7.2 ± 0.52 μm/min average velocity) migrated significantly slower compared to the unstimulated control (9.2 ± 0.59 μm/min average velocity) in response to 100 nM CCL5 chemokine (p = 0.0292, n = 6 donors; Figure S2B). IFN-γ-stimulated monocytes (5.3 ± 0.33 μm/min average velocity) were also significantly slower compared to the unstimulated control (p = 0.0001, n = 6 donors) and GM-CSF (p = 0.0402, n = 6 donors) in the presence of 100 nM CCL5 (Figure S2B). The distinctive impacts of IFN-γ stimulation on human monocyte chemotaxis and velocity in the MAP chip highlight the platform’s potential for profiling peripheral immune cell migration behavior before and after exposure to proinflammatory cytokines, as well as in the presence of various chemoattractants.

To test whether cytokine stimulation may promote monocyte differentiation into macrophages in the MAP chip and alter their migration characteristics, we evaluated the impact of GM-CSF or macrophage CSF (M-CSF) on human monocyte differentiation. GM-CSF and M-CSF are well-known drivers of monocyte differentiation into proinflammatory (M1) and anti-inflammatory (M2) macrophages.⁵³ Monocytes from healthy donors were stimulated with 50 ng/mL M-CSF for 6 days or 10 ng/mL GM-CSF for 15 h (i.e., the time course of stimulation within the MAP chip). We then performed flow cytometry to determine the percentage of M1 and M2 macrophages after stimulation (Figures S2C–S2K). As expected, we found significantly more CD86⁺ cells (M1 macrophages) compared to unstimulated monocytes (p = 0.0131, n = 3 donors; Figures S2D and S2E) after 6 days of M-CSF stimulation. Similarly, the percentage of CD14⁺CD86⁻ monocytes was significantly decreased compared to unstimulated monocytes after M-CSF stimulation (p = 0.034, n = 3 donors; Figures S2D and S2E). We also noticed a slight increase in M2 macrophages (CD163⁺ cells) upon M-CSF stimulation for 6 days—31% of M-CSF-stimulated cells and 4% of unstimulated cells, on average, were CD163⁺ (not significant, n = 3 donors; Figures S2F and S2G). In contrast, when human monocytes were stimulated with GM-CSF for 15 h, an average of 7% of cells were CD86⁺ and an average of 0.5% were CD163⁺, neither of which was significantly different compared to the unstimulated control (n = 3 donors; Figures S2H–S2K). We also observed no significant change to the percentage of CD14⁺CD86⁻ monocytes compared to unstimulated monocytes after 15 h of GM-CSF stimulation (n = 3 donors; Figures S2H–S2K). While IFN-γ stimulation may promote monocyte differentiation, it usually requires longer than the 15 h of stimulation we used in the MAP chip and is often used to prime monocytes for differentiation with other cytokines.^51,54 It is unlikely that 15 h of IFN-γ stimulation alone led to adequate monocyte differentiation. Thus, cytokine stimulation does not likely modulate monocyte migration in the MAP chip via differentiation into macrophages.

Biased migration of monocytes toward low hydraulic resistance regardless of donor age

We next used the dual-taxis microchannels in the MAP chip to investigate whether barotactic changes influence monocyte migration behaviors. Each dual-taxis microchannel contains a straight channel connecting the CENTRAL chamber to a bifurcation where the hydraulic resistance in the right path increased by 1×, 6×, 14×, or 51× compared to the left path (Figures 3A–3C). The hydraulic resistance of each microchannel was determined according to the equation outlined in the STAR Methods, following the approach used in previous studies.^43,45,46

Figure 3.

Monocytes migrate through the path of lower hydraulic resistance regardless of donor age or proinflammatory cytokine stimulation

(A) The dual-taxis microchannels contain a chemotactic gradient and varying hydraulic resistances. At the bifurcation of each microchannel, the hydraulic resistance is increased by 6×, 14×, or 51× in the right path compared to the left path.

(B) Schematic showing a magnified view of the bifurcation within the 6× hydraulic resistance design.

(C) Representative time-lapse confocal microscopy imaging of monocytes (cell membrane, green; nuclei, blue) in the dual-taxis microchannels primed with 100 nM CCL2 chemokine. Scale bar, 10 μm.

(D) The number of migrating monocytes in the presence of CCL2 gradient or evenly distributed chemokine (n = 6 independent MAP chips).

(E–I) The migration ratio of monocytes in response to 100 nM CCL2 chemokine and varying hydraulic resistances (n = 17–18 healthy unstimulated, n = 3–5 19- to 27-year-old, n = 6 50- to 60-year-old, n = 7 GM-CSF-stimulated donors, and n = 7–8 IFN-γ-stimulated donors).

(J–L) The migration ratio of monocytes in response to 100 nM CCL5 chemokine and varying hydraulic resistances (n = 6–8 healthy unstimulated, n = 7 GM-CSF-stimulated, and n = 7–8 IFN-γ-stimulated donors).

p values are from Welch’s t test (D) and one-sample t and Wilcoxon tests (E–L). Boxplots show the median and the range between the 25th and 75th percentiles. The whiskers stretch from the minimum and maximum values.

The hydraulic resistance was modulated by altering the channel width at the bifurcation. We first confirmed that a CCL2 chemokine gradient, as opposed to the presence of evenly distributed CCL2, significantly induced monocyte migration in the dual-taxis microchannels (p = 0.0003; Figure 3D). Then, to quantify the migration bias of human monocytes, the migration ratio was calculated by dividing the number of monocytes that opted for the higher-hydraulic-resistance path by the number of monocytes that migrated through the lower-hydraulic-resistance path. Monocytes were loaded into MAP chips primed with 100 nM CCL2 chemokine, and cell migration was tracked through dual-taxis microchannels for up to 18 h using time-lapse confocal imaging. In the presence of CCL2, monocytes (n = 17–18 donors) showed a robust bias by 1.5-, 3-, and 5.9-fold toward the lower-hydraulic-resistance path compared to the 6×, 14×, and 51× higher-hydraulic-resistance paths (p < 0.0001), respectively (Figure 3E).

Similarly, we found that monocytes from 19- to 27- and 50- to 60-year-olds displayed an increased bias toward the lower-hydraulic-resistance path compared to the higher-hydraulic-resistance paths. Notably, monocytes from 19- to 27-year-olds (n = 3–5 donors) exhibited a greater propensity to migrate through the channel with lower hydraulic resistance, particularly when the alternative path presented with increased hydraulic resistance by 14× (8-fold, p < 0.0001) or 51× (7.5-fold, p = 0.0011; Figure 3F). Monocytes from 50- to 60-year-olds (n = 6 donors) exhibited a significantly higher migration rate through the channel with lower hydraulic resistance when the hydraulic resistance in the alternative path increased by 14× (2.5-fold, p = 0.0157) or 51× (5.7-fold, p < 0.0001; Figure 3G). Overall, across deidentified healthy donors (n = 17–18 donors) and different age groups (n = 9–11 donors), we found that human monocytes consistently exhibited a substantial bias toward the lower-hydraulic-resistance path as the hydraulic resistance in the alternative path increased. Interestingly, this bias remained consistent regardless of whether human monocytes were isolated from fresh whole blood or cryopreserved PBMCs.

GM-CSF- and IFN-γ-stimulated monocytes maintain migration bias toward low-hydraulic-resistance paths

Next, we examined whether GM-CSF or IFN-γ stimulation impacts the migration bias of human monocytes in dual-taxis microchannels. We found that GM-CSF- or IFN-γ-stimulated monocytes from deidentified healthy donors exhibited an increased bias toward the lower-hydraulic-resistance path compared to the higher-hydraulic-resistance path, regardless of the chemoattractant. In the presence of 100 nM CCL2 chemokine, GM-CSF-stimulated monocytes (n = 7 donors) showed a significant preference for the path with lower hydraulic resistance when the alternative path presented with increased hydraulic resistance by 14× (2-fold, p = 0.0312) or 51× (3-fold, p = 0.0312; Figure 3H). Similarly, monocytes stimulated with IFN-γ (n = 7–8 donors) chose the path of lower hydraulic resistance significantly more when the alternative path presented with greater hydraulic resistance by 6× (1.4-fold, p = 0.024), 14× (2.6-fold, p < 0.0001), or 51× (5.3-fold, p < 0.0001; Figure 3I).

GM-CSF- and IFN-γ-stimulated monocytes also maintained a bias toward the lower-hydraulic-resistance path when the dual-taxis microchannels were primed with 100 nM CCL5 chemokine. Notably, unstimulated monocytes (n = 6–8 donors) showed a substantial decrease in their migration toward channels with 6× (p = 0.0073), 14× (p = 0.0016), and 51× (p = 0.0002) greater hydraulic resistance compared to the 1× hydraulic resistance channel by 1.8-, 2.7-, and 5.2-fold, respectively (Figure 3J). GM-CSF-stimulated monocytes (n = 7 donors) chose the path of lower hydraulic resistance significantly more when the alternative path had a 6× (p = 0.0042) or 51× (p = 0.0006) greater hydraulic resistance (Figure 3K). While preference for the 1× hydraulic resistance path was not significant compared to the 14× migration path, GM-CSF-stimulated monocytes still showed an increasing preference for the lower-hydraulic-resistance path as the hydraulic resistance in the alternative path increased. In the case of IFN-γ-stimulated monocytes (n = 7–8 donors), we observed a significant bias for the lower-hydraulic-resistance path when the alternative path exhibited a greater hydraulic resistance by 14× (2.2-fold, p = 0.0078) or 51× (4.4-fold, p = 0.0078; Figure 3L).

Overall, human monocytes consistently displayed a robust bias toward migration paths with lower hydraulic resistance, regardless of donor age, GM-CSF or IFN-γ stimulation, cell freshness, or exposure to chemical cues such as CCL2 and CCL5 chemokine. This consistency suggests that human monocyte sensitivity to barotactic stimuli is an inherent and conserved behavior.

Hydraulic resistance, not spatial confinement, defines the human monocyte migration path

Next, we investigated whether the robust biased migration of human monocytes toward the path with lower hydraulic resistance is due to their sensitivity to hydraulic resistance or a response to a confined space at the decision-making point. For this purpose, we created the distance dual-taxis microchannels in the MAP chip. In each microchannel, a straight channel leads from the CENTRAL chamber to a bifurcation where the dimensions of the left and right path after the bifurcation are the same. The dimensions of the left and right paths remain the same for another 30, 60, 120, or 180 μm before the right path narrows, increasing the hydraulic resistance compared to the left path. The hydraulic resistance of each path was calculated across the entire channel, from the bifurcation to the edge of the CHEMOKINE chamber. Compared to the left path, the right path had a greater relative hydraulic resistance by 9.2×, 7.6×, 5.3×, or 3.5× (Figures 4A and 4B). Migration of monocytes in the distance dual-taxis microchannels was monitored for up to 18 h using confocal time-lapse imaging (Figure 4C; Video S6). We again showed that a CCL2 chemokine concentration gradient induced significantly more monocyte migration than an evenly distributed CCL2 chemokine concentration (p < 0.0001; Figure 4D). Next, we tested the directional migration of human monocytes from deidentified healthy donors (n = 17–18 donors) or 19- to 27- (n = 4–5 donors) or 50- to 60-year-olds (n = 6 donors) in the distance dual-taxis microchannels in the presence of 100 nM CCL2. Despite equivalent spatial confinement in the left and right paths at the bifurcation, monocytes from deidentified donors exhibited a dramatic bias toward the path with lower hydraulic resistance when the alternative path had 9.2× (p < 0.0001), 7.6× (p < 0.0001), 5.3× (p = 0.0016), and 3.5× (p = 0.0237) greater hydraulic resistance (Figure 4E). This indicates that human monocytes are sensitive to hydraulic resistance and not the confinement of the migration path. Monocytes from 19- to 27-year-olds migrated through the lower-hydraulic-resistance path significantly more when the alternative path had 9.2× (p = 0.0056, Figure 4F) greater hydraulic resistance. Monocytes from 50- to 60-year-olds migrated significantly more through the lower-hydraulic-resistance path in the presence of 9.2× (p = 0.0312), 7.6× (p = 0.0312), and 5.3× (p = 0.0312) greater hydraulic resistance paths (Figure 4G). Interestingly, both age groups showed an increasing preference for the lower-hydraulic-resistance path as the hydraulic resistance in the alternative paths increased (Figures 4F and 4G). These data further reinforce that monocyte migration behavior is altered by hydraulic resistance, not the spatial confinement at the decision-making points.

Figure 4.

Spatial confinement does not dictate the migration of human monocytes toward lower hydraulic resistance paths

(A) The distance dual-taxis microchannels contain equal dimensions at the bifurcation and varying distances to a change in width and, consequentially, hydraulic pressure. The left channel has lower hydraulic resistance, and the right path has a 9.2×, 7.6×, 5.3×, or 3.5× greater hydraulic resistances.

(B) Schematic showing a monocyte experiencing equal spatial confinement at the bifurcation but different hydraulic resistances.

(C) Representative time-lapse confocal microscopy imaging of monocytes (cell membrane, green; nuclei, blue) in the distance dual-taxis microchannels primed with 100 nM CCL2 chemokine. Scale bar, 10 μm.

(D) The number of migrating monocytes in the presence of CCL2 gradient or evenly distributed chemokine (n = 6 independent MAP chips).

(E–G) The migration ratio of monocytes in response to CCL2 and varying hydraulic resistances (n = 17–18 healthy unstimulated, n = 4–5 19- to 27-year-old, and n = 6 50- to 60-year-old donors).

(H and I) Quantification of the migration velocity of monocytes from 19- to 27-year-old (n = 5) and 50- to 60-year-old donors (n = 5) in response to 100 nM CCL2 chemokine.

(J and K) The migration ratio of stimulated monocytes in response to CCL2 and varying hydraulic resistances (n = 8 GM-CSF-stimulated and n = 7–8 IFN-γ-stimulated donors).

(L–N) The migration ratio of stimulated monocytes in response to CCL5 and varying hydraulic resistances (n = 8 healthy unstimulated, n = 7–8 GM-CSF-stimulated, and n = 8 IFN-γ-stimulated donors).

(O and P) Quantification of monocyte migration velocity toward (O) 100 nM CCL2 or (P) CCL5 chemokine when unstimulated (n = 16 donors for CCL2 and n = 6 donors for CCL5), GM-CSF stimulated (n = 6 donors for CCL2 and CCL5), or IFN-γ stimulated (n = 4 donors for CCL2 and n = 6 for CCL5).

p values are from Mann-Whitney test (D and I), one sample t and Wilcoxon tests (E–G and J–N), Welch’s t test (H), and two-way ANOVA (O and P). Boxplots show the median and the range between the 25th and 75th percentiles. The whiskers stretch from the minimum and maximum values.

Video S6. Directional monocyte migration in the distance dual-taxis microchannels, related to Figure 4

Monocytes were stained (blue, nucleus; green, cell membrane) and loaded into the MAP chip. Representative time-lapse imaging of monocytes navigating the distance dual taxis microchannels. Migration is shown over approximately 2.4 h. Scale bar, 30 μm.

We also found that monocyte migration velocity, regardless of donor age, was dependent on the hydraulic resistance of the migration path. Monocytes from donors aged 19–27 years (n = 5 donors) migrated significantly faster through the higher-hydraulic-resistance paths at an average velocity of 7.4 ± 0.23 μm/min compared to 5.6 ± 0.33 μm/min in the lower-hydraulic-resistance paths (p = 0.0027; Figure 4H). Similarly, monocytes from donors aged 50–60 years (n = 5 donors) migrated significantly faster through the higher-hydraulic-resistance paths at an average of 7.4 ± 0.51 μm/min compared to 5.0 ± 0.17 μm/min in the lower-hydraulic-resistance paths (p = 0.0079; Figure 4I).

We next tested whether GM-CSF or IFN-γ stimulation alters the migration bias of human monocytes in the distance dual-taxis microchannels. In the presence of CCL2, GM-CSF-stimulated monocytes (n = 8 donors) showed significant bias toward the path of lower hydraulic resistance only when the alternative path had 9.2× greater hydraulic resistance (p = 0.0204; Figure 4J). However, similar to our previous observations, GM-CSF-stimulated monocytes showed an increasing trend toward lower-hydraulic-resistance paths as the hydraulic resistance in the alternative path increased. Additionally, monocytes stimulated with IFN-γ (n = 7–8 donors) showed a significant preference for the path with lower hydraulic resistance when the alternative path presented with increased hydraulic resistance by 9.2× (p = 0.0312), 7.6× (p = 0.0156), or 5.3× (p = 0.0156; Figure 4K). This demonstrates that human monocytes maintain their bias toward lower-hydraulic-resistance pathways in the presence of CCL2, regardless of stimulation.

Next, we assessed whether CCL5 chemokine impacts migration characteristics of monocytes in distance dual-taxis microchannels. Unstimulated monocytes (n = 8 donors) significantly sought out the path of lower hydraulic resistance in the presence of CCL5 chemokine only when the alternative path had 9.2× greater hydraulic resistance (p = 0.003; Figure 4L). We observed similar trends in stimulated monocytes in which GM-CSF- (n = 7–8 donors) and IFN-γ-stimulated (n = 8 donors) monocytes chose the path of lower hydraulic resistance significantly more only when the alternative path presented with 9.2× increased hydraulic resistance (p = 0.0183 and p = 0.0234, respectively; Figures 4M and 4N). While monocytes showed a trend of migrating toward lower-hydraulic-resistance paths in the presence of CCL5, it was not as robust as CCL2 chemokine. This is likely due to the comparatively low number of monocytes that express the chemokine receptor CCR5 (Figure S3D), as well as the narrow range of relative hydraulic resistances we tested in the distance dual-taxis microchannels.

We next set out to determine whether GM-CSF and IFN-γ stimulations affect monocyte migration velocity in paths with different hydraulic resistances in the presence of 100 nM CCL2. Unstimulated monocytes (n = 16 donors) migrated significantly faster through the higher-hydraulic-resistance path at an average of 7.0 ± 0.55 μm/min (p = 0.0023) compared to 4.8 ± 0.35 μm/min in the lower-hydraulic-resistance path (Figure 4O). Treating human monocytes (n = 6 donors) with GM-CSF did not significantly alter monocyte migration velocity in the low (4.8 ± 0.24 μm/min) or high (6.7 ± 0.44 μm/min) hydraulic resistance channels compared to the unstimulated control (Figure 4O). Like the unstimulated condition, GM-CSF-stimulated monocytes showed increased migration velocity in the higher-hydraulic-resistance path compared to the lower-hydraulic-resistance path (not significant; Figure 4O). Similarly, IFN-γ-stimulated monocytes (n = 4 donors) migrated through the lower-hydraulic-resistance channel at an average velocity of 3.3 ± 0.37 μm/min and the higher-hydraulic-resistance channel at 4.3 ± 0.26 μm/min (Figure 4O). IFN-γ-stimulated monocytes exhibited significantly slower migration velocity in the higher-hydraulic-resistance path compared to unstimulated monocytes (p = 0.0363; Figure 4O). Overall, GM-CSF and IFN-γ abrogated the greater migration velocity of monocytes in higher-hydraulic-resistance paths versus lower-hydraulic-resistance paths. However, migration velocity in higher-hydraulic-resistance microchannels was still increased compared to lower-hydraulic-resistance microchannels after stimulation. Additionally, IFN-γ treatment led to noticeably decreased migration velocity regardless of hydraulic resistance.

In the presence of CCL5 chemokine, GM-CSF and IFN-γ stimulations altered monocyte migration velocity through higher- and lower-hydraulic-resistance channels. Unstimulated monocytes (n = 6 donors) migrated at an average velocity of 8.8 ± 0.39 μm/min through the higher-hydraulic-resistance channels compared to 8.1 ± 0.56 μm/min in the lower-hydraulic-resistance paths (Figure 4P). GM-CSF-stimulated monocytes (n = 6 donors) migrated at average velocities of 7.3 ± 0.28 and 6.2 ± 0.30 μm/min through the higher- and lower-hydraulic-resistance channels, respectively (Figure 4P). In the lower-hydraulic-resistance microchannels, GM-CSF-stimulated monocytes migrated significantly slower compared to the unstimulated monocytes (p = 0.014; Figure 4P). Similarly, IFN-γ-stimulated monocytes (n = 6 donors) migrated at average velocities of 5.7 ± 0.33 and 4.3 ± 0.24 μm/min through the higher- and lower-hydraulic-resistance channels, respectively (Figure 4P). We also observed that IFN-γ stimulation robustly decreased the migration velocity of monocytes. Specifically, in the lower-hydraulic-resistance microchannels, IFN-γ-stimulated monocytes traveled 1.9- and 1.4-fold slower than unstimulated monocytes (p < 0.0001) and GM-CSF-stimulated monocytes (p = 0.0086), respectively (Figure 4P). In the higher-hydraulic-resistance microchannels, IFN-γ-stimulated monocytes traveled 1.5- and 1.3-fold slower than unstimulated monocytes (p < 0.0001) and GM-CSF-stimulated monocytes (p = 0.0447), respectively (Figure 4P).

Overall, these data demonstrated robust migration bias of human monocytes toward lower-hydraulic-resistance pathways, regardless of chemoattractants and stimulation with GM-CSF or IFN-γ. Lastly, while GM-CSF and IFN-γ stimulations do not alter the migration bias of human monocytes toward lower-hydraulic-resistance paths, IFN-γ treatment robustly diminished the migration velocity of human monocytes in both low- and high-hydraulic-resistance microchannels, regardless of the chemoattractant present.

Human monocytes navigate effectively through complex chemotactic paths

To further characterize migration behaviors of human monocytes in the MAP chip, we assessed their ability to navigate through complex pathways in the presence of CCL2 (Figures 5A and 5B; Video S7). We first showed that a CCL2 concentration gradient induced significantly more monocyte migration compared to evenly distributed CCL2 in the maze microchannels (p < 0.0001; Figure 5C). Using the chemotactic maze in the MAP chip, we measured four different metrics: (1) the number of cells that solved the chemotactic maze, (2) track duration—the time cells took to solve the chemotactic maze, (3) track length—the distance cells traveled within the maze to solve it, and (4) migration velocity. Interestingly, monocytes from donors aged 50- to 60-years-old exhibited similar behaviors compared to the 19- to 27-year-olds in the maze in the presence of 100 nM CCL2. We found no significant difference in monocyte migration between 19- to 27- (n = 5 donors) and 50- to 60-year-olds (n = 6 donors), indicating that monocytes from both age groups could equally navigate the chemotactic maze in the presence of CCL2 (Figure 5D).

Figure 5.

IFN-γ stimulation, not aging, hinders human monocyte chemotaxis through complex pathways

(A) Schematic showing the chemotactic maze with several decision-making forks.

(B) Representative time-lapse confocal microscopy imaging of monocytes (cell membrane, green; nuclei, blue) in the chemotactic maze primed with 100 nM CCL2 chemokine. Scale bar, 10 μm.

(C) The number of migrating monocytes in the presence of CCL2 gradient or evenly distributed chemokine (n = 6 independent MAP chips).

(D–G) Quantification of normalized monocyte migration (D), track duration (E), track length (F), and migration velocity (G) in response to 100 nM CCL2 chemokine for 19- to 27-year-old (n = 3–5) and 50- to 60-year-old (n = 6) donors.

(H–K) Quantification of normalized monocyte migration (H), track duration (I), track length (J), and migration velocity (K) in response to 100 nM CCL2 chemokine for unstimulated (n = 17–18 donors), GM-CSF-stimulated (n = 8 donors), and IFN-γ-stimulated (n = 8 donors) monocytes from deidentified donors.

(L–O) Quantification of normalized monocyte migration (L), track duration (M), track length (N), and migration velocity (O) in response to 100 nM CCL5 chemokine for unstimulated (n = 8 donors), GM-CSF-stimulated (n = 7–8 donors), and IFN-γ-stimulated (n = 8 donors) monocytes from deidentified donors.

p values are from Mann-Whitney test (C), Welch’s t test (D–G), Brown-Forsythe and Welch ANOVA tests (H and L), Kruskal-Wallis test (I, J, M, and N), and one-way ANOVA (K and O). Boxplots show the median and the range between the 25th and 75th percentiles. The whiskers stretch from the minimum and maximum values.

Video S7. Directional monocyte migration in the maze, related to Figure 5

Monocytes were stained (blue, nucleus; green, cell membrane) and loaded into the MAP chip. Representative time-lapse imaging of monocytes navigating the maze with tracks displayed. Migration is shown over approximately 2.3 h. Scale bar, 30 μm.

To better dissect the migration characteristics of human monocytes in the maze, we tested the navigation specifics of monocytes from each age group. Human monocytes from 19- to 27- (n = 3 donors) and 50- to 60-year-old (n = 6 donors) groups showed similar average track durations of 133.4 ± 7.3 and 136.7 ± 11.8 min, respectively (Figure 5E), suggesting that monocytes from both age groups were spending similar amounts of time in the maze. We also found that monocytes from both age groups traveled similar distances in the chemotactic maze, as measured by track length. Specifically, monocytes from 19- to 27-year-olds (n = 3 donors) and 50- to 60-year-olds (n = 6 donors) had average track lengths of 1,084 ± 108 and 941.7 ± 31 μm, respectively (Figure 5F). Additionally, we found no significant difference in the monocyte migration velocity between the two age groups (Figure 5G). Monocytes migrated at average velocities of 8.9 ± 0.94 and 7.4 ± 0.44 μm/min from donors aged 19–27 (n = 3 donors) and 50–60 years (n = 6 donors), respectively (Figure 5G). Overall, these data indicate that donor age does not significantly influence the number of migrating cells, the time required for monocytes to navigate through a complex chemotactic maze, the chosen migration path, or cell velocity.

IFN-γ stimulation impairs monocyte ability to navigate through complex chemotactic paths

Next, we explored whether GM-CSF and IFN-γ stimulations impact monocyte ability to navigate through the maze. We observed that a substantially lower number of IFN-γ-stimulated monocytes (n = 8 donors) solved the maze compared to unstimulated monocytes (n = 18 donors, p = 0.0003; Figure 5H) in the presence of CCL2. IFN-γ-stimulated monocytes (n = 8 donors) showed an average track duration of 241 ± 21 min, which is significantly longer than unstimulated monocytes (n = 17 deidentified donors, 147 ± 17 min, p = 0.0148) and GM-CSF-stimulated monocytes (n = 8 deidentified donors, 113 ± 10 min, p = 0.0013; Figure 5I). These data indicate that IFN-γ stimulation substantially impacts monocyte navigation time through a complex chemotactic pathway.

Next, we asked whether the prolonged track duration observed in IFN-γ-stimulated monocytes was due to their utilization of indirect paths to navigate through the maze or lower migration velocity to reach the CHEMOKINE chamber. We found no significant difference in the track length among all three groups (i.e., unstimulated, GM-CSF stimulated, and IFN-γ stimulated) of monocytes, indicating that cells predominantly chose similar routes to navigate through the chemotactic maze (Figure 5J). However, IFN-γ-stimulated monocytes (n = 8 donors) migrated at a significantly slower velocity compared to unstimulated (n = 17 donors, p = 0.005) and GM-CSF-stimulated monocytes (n = 8 donors, p = 0.0004; Figure 5K). Overall, GM-CSF stimulation did not impact the ability of human monocytes to navigate through the chemotactic maze in the presence of the CCL2 chemokine. However, IFN-γ stimulation led to a dramatic decrease in the speed of monocytes navigating through the complex maze, but it did not compromise their chemotactic accuracy.

Priming the chemotactic maze with CCL5 revealed an additional mechanism through which IFN-γ stimulation alters how human monocytes navigate through complex pathways. In contrast to CCL2, we observed a slight decrease in the number of stimulated monocytes (GM-CSF or IFN-γ) that solved the chemotactic maze in the presence of CCL5, but it was not significant (n = 7–8 donors; Figure 5L). However, IFN-γ-stimulated monocytes (n = 8 donors) took significantly longer to navigate through the maze compared to unstimulated (n = 8 donors, p = 0.0062) and GM-CSF-stimulated (n = 7 donors, p = 0.0009) monocytes (Figure 5M). Next, we tested if the prolonged track duration observed in IFN-γ-stimulated monocytes was due to the lower migration velocity of monocytes, similar to CCL2, or their utilization of indirect routes to navigate through the chemotactic maze. Interestingly, we found that both the track length and migration velocity of monocytes were modulated by IFN-γ stimulation. Human monocytes stimulated with IFN-γ (n = 8 donors) migrated 1.3- and 1.4-fold further compared to unstimulated monocytes (n = 8 donors, p = 0.0266) and GM-CSF-stimulated (n = 7 donors, p = 0.0012) monocytes (Figure 5N). Additionally, IFN-γ-stimulated monocytes (n = 8 donors) traveled significantly slower at an average velocity of 6.3 ± 0.60 μm/min compared to unstimulated monocytes (n = 8 donors, 11.1 ± 0.64 μm/min, p = 0.0001) and GM-CSF-stimulated monocytes (n = 7 donors, 12.2 ± 1.22 μm/min, p = 0.0402; Figure 5O). Collectively, GM-CSF exhibited no impact on monocyte migration through the chemotactic maze primed with CCL5. However, IFN-γ stimulation resulted in a substantial increase in monocyte navigation time through the chemotactic maze, impaired ability to select direct chemotactic paths, and diminished migration velocity.

Chemokine receptor CCR2 is altered by donor age and proinflammatory cytokine stimulation

The MAP chip allowed us to profile migration characteristics of human monocytes under various conditions, such as different age groups and proinflammatory cytokine stimulation, as well as in response to chemical and physical stimuli (summarized in Figures 6A and 6B). We consistently found that donor age did not impact monocyte migration in the MAP chip, whereas IFN-γ stimulation resulted in unique migration characteristics. Flow cytometry was then used to quantify CCR2 and CCR5 expression levels on human monocytes (Figure S3A) from different age groups and stimulation conditions to better understand the observed migration profiles in the MAP chip. We found that an average of 91.8% ± 0.67% of monocytes from donors aged 19–27 (n = 4 donors) expressed CCR2. This percentage was significantly lower compared to the average expression of CCR2 in monocytes from donors aged 50–60 (n = 4 donors), which was 94.6% ± 0.92% (p = 0.0484; Figure S3B). CCR5 expression levels on monocytes from 50- to 60-year-olds (10.2% ± 1.8%) were also elevated compared to 19- to 27-year-olds (5% ± 1.6%, not significant; Figure S3B). Despite the significant increase in CCR2 receptor expression observed in the 50- to 60-year-old age group, this preliminary finding was incongruent with the absence of notable differences in the migration of monocytes between age groups in the MAP chip. However, when we examined the relationship between CCR2 receptor expression and CCL2 chemotaxis in a donor-dependent rather than age-dependent manner, we found a positive correlation between CCR2 expression and CCL2 chemotaxis (Figure S3C). Moreover, the functional migration differences observed in the MAP chip were supported by the levels of CCR2 expression on monocytes following stimulation with GM-CSF or IFN-γ. On average, 97.5% ± 0.73% of unstimulated monocytes (n = 4 donors), 95.4% ± 1.5% of GM-CSF-stimulated monocytes (n = 4 donors), and 91.8% ± 2.7% of IFN-γ-stimulated monocytes (n = 4 donors) expressed CCR2. IFN-γ-stimulated monocytes showed a significant decrease in CCR2 expression (p = 0.04) compared to the unstimulated control (Figure S3D). Stimulation with GM-CSF or IFN-γ slightly increased monocyte CCR5 expression but not significantly (Figure S3D). Specifically, an average of 16.6% ± 6.9% of unstimulated monocytes (n = 4 donors), 22.6% ± 9.0% of GM-CSF-stimulated monocytes (n = 4 deidentified donors), and 19.4% ± 5.3% of IFN-γ-stimulated monocytes (n = 4 donors) expressed the CCR5 receptor (Figure S3D). The comparable levels of CCR5 expression observed in both stimulated and unstimulated groups imply that the distinctive behavior of IFN-γ-stimulated monocytes in the CCL5-primed MAP chip is not solely due to the CCL5-CCR5 signaling axis.

Figure 6.

The MAP chip is a high-throughput platform for profiling monocyte migration in response to chemotactic and barotactic cues

(A) Schematic illustrating the various conditions used in the MAP chip to profile human monocyte migration characteristics.

(B) Heatmaps showing percentage of changes in human monocyte migration characteristics for older donors compared to younger donors or upon stimulation with GM-CSF or IFN-γ under various MAP chip conditions.

p values are from one-sample t and Wilcoxon tests.

Discussion

We developed and implemented a high-throughput microfluidic platform (MAP chip) containing chemotactic and barotactic cues to characterize monocyte migration behaviors. We used the MAP chip to evaluate how a variety of factors, like donor age and proinflammatory cytokine stimulation, alter monocyte migration in the presence of two chemoattractant proteins and different hydraulic resistances. Previous microfluidic devices have utilized a limited number of microchannel designs in a single device to profile the migration signatures of leukocytes.^43,49,55 These microfluidic devices provide invaluable insights, but their functionality is limited by their design. To overcome this, we designed the MAP chip to include four unique sets of microchannels. The dimensions of the MAP chip’s microchannels are consistent with those used in previous microfluidic devices evaluating leukocyte migration.^45,46,56 A 48-well plate of MAP chips can generate a total of 720 unique data points, providing both chemotactic and barotactic insights at a single-cell level. Using the MAP platform, we discovered that donor age had no significant effect on the recruitment of human monocytes in response to the CCL2 chemoattractant. The CCL2 chemokine is a key recruitment factor for human monocytes in a myriad of diseases, like breast cancer,^6,19 MS,¹⁷ AD,⁵⁷ osteoarthritis,⁵⁸ and atherosclerosis.⁵⁹ Additionally, we observed a positive correlation between donor-specific CCR2 expression and CCL2 chemotaxis, indicating that CCL2 chemotaxis, in part, is dependent on CCR2 receptor levels rather than donor age. This further implies that modulation of the CCL2-CCR2 signaling axis may help identify therapeutic targets that could regulate the infiltration of monocytes and steer their impact on pathogenesis.

Monocyte stimulation with IFN-γ (not GM-CSF) significantly decreased monocyte CCL2 chemotaxis, consistent with previous studies.^60,61,62 IFN-γ has been linked to decreased CCR2 expression in monocytes via CCR2 mRNA destabilization.⁶² This was supported by the significant decrease in CCR2 expression levels we observed on IFN-γ-stimulated monocytes. Similarly, IFN-γ stimulation slightly decreased CCL5 chemotaxis in human monocytes compared to unstimulated monocytes. IFN-γ-stimulated monocytes showed a dramatic reduction in the migration velocity, regardless of the chemoattractant, while retaining chemotactic accuracy in the presence of CCL2. A similar migration pattern was observed in monocytes treated with ROCK and nonmuscle myosin II, showing a ∼40% reduction in migration velocity toward N-formyl-methionyl-leucyl-phenylalanine without affecting chemotactic sensitivity.⁶³ One possible explanation for the IFN-γ-triggered decrease in monocyte velocity is that IFN-γ retains monocytes at the site of inflammation to maintain the immune response.⁶³ The exact mechanisms driving IFN-γ-mediated migration behavior remain unclear, but our data and previous reports highlight the crucial role of IFN-γ in peripheral immune cell migration.

GM-CSF stimulation also decreased monocyte migration velocity but had no other effects on chemotaxis in the presence of the CCL5 chemokine. Previous studies on the effects of GM-CSF on monocyte migration have focused on inducing CCL2-CCR2 chemotaxis^64,65 or differentiating monocytes into macrophages.⁶⁶ However, in our study, GM-CSF stimulation did not significantly lead to monocyte differentiation into macrophages in the MAP chip. Overall, the effects of GM-CSF on monocyte CCL5 chemotaxis and migration velocity warrant further investigation, given the critical role GM-CSF plays in a multitude of pathogenic processes.

We also found that regardless of the chemoattractant present, donor age, or proinflammatory cytokine stimulation, human monocytes robustly prefer migration paths with lower hydraulic resistance. Previous reports similarly found that HL-60 neutrophils preferred to migrate through paths of lower hydraulic resistance in microfluidic devices.^43,45 In the MAP chip, we investigated how hydraulic resistance affects human monocyte migration through the dual-taxis and distance dual-taxis microchannels. It is important to note that by increasing the length of the microchannels in the dual-taxis design, we increase hydraulic resistance and decrease the chemokine concentration experienced by the monocytes at bifurcations. To isolate the impact of hydraulic resistance on monocyte migration, we designed the distance dual-taxis microchannels. In this design, the microchannels have equal lengths and, therefore, similar chemokine concentrations after the bifurcation. Monocytes preferentially chose the paths with lower hydraulic resistance in both the dual-taxis and distance dual-taxis microchannels. This indicates that monocyte migration at the MAP chip bifurcation is influenced more by hydraulic resistance than chemokine concentration. We also demonstrated that hydraulic resistance, rather than spatial confinement at the decision-making points, defines the biased chemotaxis of human monocytes toward paths with lower hydraulic resistance. This indicates that monocyte sensitivity to barotactic stimuli is an inherent and conserved behavior, independent of donor age and GM-CSF or IFN-γ stimulation. However, the biological mechanisms driving this barotactic sensitivity in monocytes and across other peripheral immune cells remain unclear. A recent study demonstrated that activation of TRPM7 channels in breast cancer and fibrosarcoma cells led to migration toward lower-hydraulic-resistance paths.⁴⁶ Future studies are needed to test whether TRPM7 is also a key mechanosensor driving circulating monocytes toward lower-hydraulic-resistance paths.

Our findings also suggested an important role of hydraulic resistance in regulating cell migration velocity. We showed that human monocytes migrate significantly faster through the higher rather than the lower-hydraulic-resistance paths. Notably, this phenomenon was independent of donor age and chemoattractants. Other studies have shown that increased confinement of the migration space and extracellular hydraulic resistance lead to greater migration speed of metastatic adenocarcinoma breast cancer cells.^46,67 However, the potential effect of hydraulic resistance on human monocyte migration velocity has not been demonstrated. These data highlight the critical impact of chemical and physical guidance cues, particularly hydraulic resistance, in modulating innate immune cell migration behavior, warranting further investigation.

The chemotactic maze allowed us to profile monocyte migration through complex pathways. Human monocytes, regardless of donor age, showed equal ability to navigate the chemotactic maze. However, IFN-γ (not GM-CSF) stimulation significantly impaired monocyte CCL2 chemotaxis and velocity, consistent with other studies.^60,61 This aligned with the migration patterns of IFN-γ-stimulated monocytes observed in the chemotaxis and distance dual-taxis microchannels. The substantial decrease of CCR2 receptors likely plays a primary role in diminished CCL2 chemotaxis and velocity upon IFN-γ stimulation.⁶² Additionally, IFN-γ may enhance monocyte adhesion to the device surface, reducing chemotaxis and velocity.^51,68,69,70 These findings highlight the key role of the CCL2-CCR2 signaling axis in modulating human monocyte migration and its potential as a therapeutic target. Interestingly, the CCR5 receptor remained unchanged by IFN-γ stimulation, suggesting that the robust migration signatures of IFN-γ-stimulated monocytes are likely due to mechanisms beyond the CCL5-CCR5 axis.

In summary, we have described the creation and validation of a MAP chip for investigating human monocyte migration characteristics under various chemical and physical guidance cues. Using the MAP chip, we demonstrated a biased migration of human monocytes toward lower-hydraulic-resistance paths, regardless of donor age and chemoattractant. We also showed that IFN-γ stimulation substantially reduces monocyte chemotaxis and their migration velocity. Lastly, we demonstrated that monocyte migration was impacted more by the chemical and physical environment than donor age. While we focused solely on human monocytes, the MAP chip can now be employed to explore the migration signatures of other peripheral immune cells under various environments. These studies can also guide potential interventions targeting specific migration signatures of immune cells.

Limitations of the study

The MAP chip is a powerful tool for studying peripheral immune cell migration, but our study has several limitations. For example, we did not account for how monocytes might affect each other’s migration within the microchannels. Additionally, when a monocyte migrates through a microchannel, it may increase hydraulic resistance and decrease the chemokine concentration sensed by subsequent monocytes. This phenomenon has been described in neutrophils⁴⁵ and could provide valuable insights into the monocyte migration patterns, or lack thereof, seen in the MAP chip.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Mehdi Jorfi (mjorfi@mgh.harvard.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The authors declare that all data supporting the findings of this study are available within the paper.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This work was supported in part by the Cure Alzheimer’s Fund. Illustrations were created with BioRender (https://www.biorender.com). We would like to thank Dr. Daniel Irimia for his feedback and insightful comments on our manuscript.

Author contributions

M.J. conceived, designed, and microfabricated the MAP chip. C.K.H., A.D., A.B., A.T., Y.T., F.E.E., and B.M.L. conducted the experiments and analyzed data. O.M.B. performed data analysis. All authors discussed and interpreted the results. C.K.H. and M.J. wrote the manuscript. M.J. supervised the study.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Human TruStain FcX (Fc Receptor Blocking Solution)	Biolegend	Cat. #422302; RRID: AB_2818986
Spark UV 387 Anti-Human CD45 Antibody	Biolegend	Cat. #304085; RRID: AB_2922537
Pacific Blue Anti-Human CD14 Antibody	Biolegend	Cat. #301828; RRID: AB_2275670
APC Anti-Human CD192 (CCR2) Antibody	Biolegend	Cat. #357207; RRID: AB_2562238
FITC Anti-Human CD195 (CCR5) Antibody	Biolegend	Cat. #359119; RRID: AB_2564070
APC Anti-Human CD14 Antibody	Biolegend	Cat. #367117; RRID: AB_2566791
PE Anti-Human CD163 Antibody	Biolegend	Cat. #333605; RRID: AB_1134005
FITC Anti-Human CD86 Antibody	Biolegend	Cat. #374203; RRID: AB_2721573
Biological samples
Human Whole Blood	Research Blood Components, LLC	Cat. #016
Human PBMCs	BioIVT, LLC	Cat. #HUMANHLPB-0002392; HUMANHLPB-0002393; HUMANHLPB-0002389; HUMANHLPB-0002562
Chemicals, peptides, and recombinant proteins
Recombinant Human CCL2/MCP-1 Protein	R&D Systems	Cat. # 279-MC-010
Recombinant Human CCL5/RANTES Protein	R&D Systems	Cat. #278-RN-010
Dextran, Fluorescein, 10,000 MW, Anionic	Thermo Fisher	Cat. #D1821
Recombinant Human GM-CSF Protein	R&D Systems	Cat. #215-GM-010
Recombinant Human IFN-gamma Protein	R&D Systems	Cat. #285-IF-100
Human M-CSF Recombinant Protein	Thermo Fisher	Cat. #PHC9501
Human GM-CSF Recombinant Protein	Thermo Fisher	Cat. #PHC2015
Negative Photoresists	Kayaku Advanced Materials	Cat. #SU-8
4-Inch Silicon Wafer	NOVA Electronic Materials	Cat. #7375
Polydimethylsiloxane (PDMS)- Sylgard 184 Silicon with Curing Agent	Ellsworth Adhesives	Cat. #4019862
24-Well Glass Bottom Plate	Cellvis	Cat. #P24-1.5H-N
Iscove’s Modified Dulbecco’s Medium (IMDM)	Gibco	Cat. #12440053
Fetal Bovine Serum (FBS)	Gibco	Cat. #10082147
Dulbecco’s Phosphate-Buffered Saline (DPBS)	Gibco	Cat. #14190144
K2EDTA Blood Collection Tube	BD Vacutainer	Cat. #3-366643
EasySep Direct Human Monocyte Isolation Kit	STEMCELL Technologies	Cat. #19669
CellTracker Green 5-Chloromethylfluorescein Diacetate (CMFDA) Dye	Invitrogen	Cat. #C2925
EasySep Buffer	STEMCELL Technologies	Cat. #20144
EasySep Human CD14 Positive Selection Kit II	STEMCELL Technologies	Cat. #17858
Cell Staining Buffer	Biolegend	Cat. #420201
Penicillin-Streptomycin-Glutamine (100X)	Gibco	Cat. #10378016
LIVE/DEAD Fixable Blue Dead Cell Stain Kit	Invitrogen	Cat. #L23105
4% Paraformaldehyde	Electron Microscopy Sciences	Cat. #50-259-99
Software and algorithms
Fiji (Version 2.15.1)	ImageJ	N/A
IMARIS (Version 9.8.2)	Oxford Instruments	N/A
FlowJo (Version 10.10.0)	BD Biosciences	N/A
Prism (Version 10.1.2)	Graphpad	N/A
R (Version 4.2.1)	R Core Team	N/A
AutoCAD (Version 2019)	Autodesk	N/A

Experimental model and study participant details

Design and fabrication of MAP chips

The MAP chip was designed using AutoCAD software. Using standard photolithography techniques, negative photoresists (Kayaku Advanced Materials) were patterned on a 4-inch silicon wafer. The first layer (∼5.3 μm), comprising the migration microchannels, and the second layer (∼70 μm) containing cell loading and chemokine compartments, were sequentially spin-coated and patterned using a UV exposure through a chrome photomask. The wafer was used as a master mold for the polydimethylsiloxane (Ellsworth Adhesives, PDMS, Sylgard 184)-based microfluidic devices. Silicon elastomer was mixed in a 10:1 ratio with the curing agent. The mixture was cured at 75°C overnight and then peeled off the master wafer. The central cell loading compartment and whole microfluidic chip were punched with 1.2 mm and 6 mm biopsy punches, respectively. The microfluidic devices were bonded irreversibly after plasma treatment to 24-well glass-bottom plates (Cellvis, P24-1.5H-N) and incubated on a hot plate at 80°C for 20 min. Each MAP chip contains four sets of microchannels extending from the CENTRAL chamber to the CHEMOKINE chambers. All microchannels are 5.3 μm in height with varying lengths and widths. For the dual taxis and distance dual taxis microchannels, the hydraulic resistance was calculated as, R_h = (12 μL)/[wh³ (1–0.63h/w)], where L = length, w = width, h = height, and μ = fluid viscosity.⁷¹ It is assumed that fluid viscosity is 1 mPa×s.

Study participant information

Whole blood from assumed healthy deidentified donors was purchased from Research Blood Components, LLC (Brighton, MA). Blood samples were collected in EDTA tubes with K2EDTA (Potassium Ethylene Diamine Tetra Acetic acid) as an anticoagulant to prevent the blood from clotting. For additional experiments, peripheral blood mononuclear cells (PBMCs) from healthy individuals, both male and female, (19–60 years of age) were obtained from BioIVT (Westbury, NY). The number of donors used for each experiment can be found in the corresponding Results section and Figure Legends.

Method details

Priming of MAP chips

CC motif chemokine Ligand 2 (CCL2) (R&D Systems, 279-MC-010) or CCL5 (R&D Systems, 278-RN-010) was reconstituted in IMDM +20% FBS to a final concentration of 100 nM. The chemokine was introduced to the microfluidic devices by pipetting 50 μL into the central compartment. Alternatively, 50 μL of IMDM +20% FBS was pipetted into the device as a control. The devices were then placed in a desiccator for 20 min, followed by washing each well containing one microfluidic device with 1 mL Dulbecco’s phosphate-buffered saline (DPBS). Then, 100 μL of IMDM +20% FBS was pipetted into the central compartment of each microfluidic device, washing off the chemokine from the central cell loading compartment and creating a chemokine gradient within the microchannels. The wells were washed once more with DPBS, and then 1.5 mL of IMDM +20% FBS and 10 ng/mL GM-CSF or 50 ng/mL IFN-γ, if applicable, was added to each well, submerging each microfluidic device.

For the experiment requiring evenly distributed chemokine in the MAP chip, the chemokine microchannels were not washed after priming with CCL2 or CCL5 to prevent the creation of a chemokine concentration gradient. Instead, the devices were submerged in 800 μL of IMDM +20% FBS and 100 nM CCL2 or CCL5.

Modeling CCL2 concentration stability

CCL2 diffusion modeling was conducted with 10,000 MW dextran fluorescein (Thermo Fisher, D1821) at 500 μg/mL in IMDM +20% FBS, imaged every 10 min for 15 h. Fluorescence intensity in the chemokine chamber and at the migration channel entrance was compared using FIJI software to estimate the chemokine gradient. Local gradients in the microchannels were visualized using the Fire LookUp Table on the fluorescein signal. Time-lapse imaging was aligned with the Linear Stack Alignment with SIFT Registration Plugin in FIJI. The CCL2 gradient estimation (nM per 100 μm) assumed initial dextran fluorescence corresponded to 100 nM CCL2 and similar diffusion rates between CCL2 and 10 kD dextran.

Monocyte isolation and loading into MAP chips

Human monocytes were isolated from whole blood within 4 h of the blood collection using the EasySep Direct Human Monocyte Isolation Kit (STEMCELL Technologies, 19669). Isolated monocytes were stained with 0.5 μM CellTracker Green 5-chloromethylfluorescein diacetate (CMFDA) dye (Invitrogen, C2925), centrifuged at 300G for 5 min at 4°C and resuspended in IMDM +20% FBS (Gibco, 10082147) at a concentration of 200,000 cells per 5 μL. For additional experiments, IMDM +20% FBS was slowly added to thawed PBMCs, and PBMCs were centrifuged at 300G for 15 min at room temperature. The PBMCs were then resuspended in EasySep Buffer (STEMCELL Technologies, 20144), and CD14⁺ monocytes were isolated using EasySep Human CD14 Positive Selection Kit II (STEMCELL Technologies, 17858). Isolated CD14⁺ monocytes were stained with 0.5 μM CellTracker Green CMFDA dye. After washing the CD14⁺ monocytes with EasySep Buffer, the cells were resuspended at a concentration of 100,000 cells per 5 μL in IMDM +20% FBS.

Monocytes were loaded into the microfluidic devices by carefully pipetting 5 μL of cells into the central compartment of each device using a gel loading pipette tip. For conditions that required cytokine stimulation, monocytes were resuspended in IMDM +20% FBS with 10 ng/mL recombinant human GM-CSF (R&D Systems, 215-GM-010) or 50 ng/mL recombinant human IFN-γ (R&D Systems, 285-IF-100). Monocytes did not incubate with the stimulating cytokines before loading into the MAP chip. Upon loading of human monocytes, microchannels were imaged every 1.5–3.5 min to track monocyte migration for up to 17 h or 18 h for the MAP chips containing CCL2 with a stable chemokine concentration. Images were captured at 10× magnification with 1.5x or 1.39x optical zoom using a Nikon A1R HD25 confocal microscope. An Okolab Stage Top Chamber was used to keep the cells at 37°C with 5% CO₂ and 95% relative humidity during imaging.

Image processing

Data were collected from the time-lapse confocal imaging and then analyzed using IMARIS software (Oxford Instruments, Version 9.8.2). First, each frame was defined as 512 x 512 pixels (px) and analyzed with the same parameters throughout the analysis. Migration data were collected using Brownian motion tracking, XY cell diameter was defined as 12 μm, the maximum distance was defined as 35, and the maximum gap size was defined as 3. For the chemotaxis design, a Region of Interest (ROI) of 420 x 220 px was used to count the number of monocytes that migrated into the chemotactic chamber. To determine the average velocity, an ROI of 380 x 220 px was used to track cell velocity during migration through microchannels. For the maze design, the number of cells that entered the maze and the number that solved the maze were counted manually. Normalized monocyte migration was calculated for the maze and the chemotactic design by dividing the number of cells that solved the maze or made it to the chemotactic chamber by the number of cells loaded into the device. Track duration was calculated according to the time each cell took to complete the maze by manually tracking the cells. Migration data were collected manually for the dual and distance dual taxis designs. Then a migration ratio was calculated by dividing the number of cells that took the higher hydraulic resistance path by the number of cells that took the lower hydraulic resistance path. Eight separate channel masks were generated to determine velocity in the dual taxis and distance dual taxis designs, and each was assigned an intensity value to sort cells into eight different ROIs. Then a “spots analysis” was performed, and the intensity mask values of the entire region were exported and sorted into the respective ROIs, along with their associated velocities. This allowed for the analysis of the velocity of all eight ROIs simultaneously. The data were not collected if < 3 cells migrated for a sample across all designs within the same MAP chip. Similarly, velocity, track duration, and solving ratio were not calculated if only one cell migrated in a design.

Flow cytometry of CCR2 and CCR5 receptors

Monocytes were isolated from the whole blood of healthy deidentified donors or frozen PBMC samples as described above. After isolation, every 200,000 monocytes were resuspended in 100 μL of cold Cell Staining Buffer (Biolegend, 420201) with 5 μL of Human TruStain FcX (Biolegend, 422302, 1:200 dilution). After 10 min of incubation at room temperature, 2.5 μL (1:40 dilution) of Spark UV 387 anti-human CD45 antibody (Biolegend, 304085), Pacific Blue anti-human CD14 antibody (Biolegend, 301828), APC anti-human CD192 (CCR2) antibody (Biolegend, 357207), and FITC anti-human CD195 (CCR5) antibody (Biolegend, 359119) were added and incubated for 20 min on ice in the dark. The stained monocytes were then washed with cold Cell Staining Buffer, centrifuged at 300G for 5 min at 4°C, and resuspended in 1 mL of cold Cell Staining Buffer. Samples were kept on ice until use. Flow cytometry was conducted on a BD LSRFortessa X-20 flow cytometer, and the data were analyzed using FlowJo software (10.10.0).

Stimulation of monocytes with GM-CSF or M-CSF

Human monocytes were isolated from the whole blood of healthy deidentified donors as previously described and resuspended at a concentration of 1,000,000 cells/mL in 10 ng/mL GM-CSF (Gibco, PHC2015) in IMDM +1% penicillin/streptomycin (Gibco, 10378016) for 15 h or 50 ng/mL M-CSF (Gibco, PHC9501) in IMDM +1% penicillin/streptomycin for 6 days according to the manufacturer’s protocol and previously published works.^72,73 The monocytes were then cultured in a round bottom 96-well plate with 200,000 cells in each well. At the desired time points, the monocytes were collected and prepared for flow cytometry.

Every 1,000,000 cells were resuspended in 1 mL cold Cell Staining Buffer (Biolegend, 420201) with 5 μL of Human TruStain FcX (Biolegend, 422302, 1:200 dilution). After incubation for 5 min at room temperature, 5 μL of APC anti-human CD14 (Biolegend, 367117), PE anti-human CD163 (Biolegend, 333605), FITC anti-human CD86 (Biolegend, 374203), and 1 μL of Live/Dead Blue (Thermo Fisher, L23105) were added and incubated for 30 min on ice in the dark. The stained monocytes were then washed with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde (PFA) for 20 min on ice in the dark. Following this, the monocytes were washed again with PBS, and resuspended in 500 μL cold Cell Staining Buffer. Samples were kept on ice until use. Flow cytometry was conducted on a BD LSRFortessa X-20 flow cytometer, and the data were analyzed using FlowJo software (10.10.0).

Quantification and statistical analysis

Multiple donors were used across the experiments, with no single donor used more than once. Monocytes were isolated for each experiment and randomly allocated to microfluidic devices. Data analysis was performed using GraphPad Prism 10 and R version 4.2.1. The Shapiro-Wilk test assessed data normality. In instances where the data did not show a normal distribution, nonparametric tests were applied for statistical analysis. A significance threshold of p value <0.05 was considered significant and ^∗p value <0.0332, ^∗∗p value <0.0021, ^∗∗∗p value <0.0002, ^∗∗∗∗p value <0.0001. Specific tests and donor numbers are detailed in figure legends. Unless stated otherwise, all data represent at least three independent experiments, with circles in graphs representing donors. Investigators were blinded during data collection and analysis to eliminate bias.

Published: September 5, 2024