Phagocytosis as an effector mechanism of the innate immune response could be triggered by attachment of antigens to the surface of macrophages. The protein-based understanding of the signal processing pathways of innate immunity to microorganisms like Toll-like receptors (TLR), nucleotide-binding oligomerization domain (NOD) proteins, and myeloid differentiation primary-response protein 88 (MyD88) families for pathogen-associated molecular patterns (PAMs), has contributed to the development of therapeutics for human immune diseases 1 2 . However, it is still hard to explain the variability of responses caused by a lack of knowledge of the modulation mechanism of the immune response of single macrophages against multiple antigen stimulations. In other words, we still do not know whether signal processing can work simultaneously and independently against a plurality of antigen stimulations in different places on the surface of a single macrophage.

To understand the mechanism of complex signal processing that occurs in phagocytosis when there are multiple stimulations to macrophages, we need to give a series of fully controlled stimulations to an isolated single macrophage step-by-step under isolated circumstances. This is because with conventional group-based cultivation in a dish, stimulation of antigens to the target macrophage is usually done in an uncontrolled probabilistic way. Moreover, the physical contact with other macrophages might also influence the phagocytic response of macrophages.

In this paper, we report the time course of phagocytosis of an isolated single macrophage against a plurality of stimulations with antigens. In the experiment, to prevent the effects of unexpected factors, we used our on-chip single-cell cultivation system to give fully controlled stimulations to the isolated macrophage, and we then measured its response to those stimulations.

First, after we started cultivation on the chip, we simultaneously stimulated the isolated macrophage with two zymosan particles from opposite sides using two optical tweezers, as shown in Fig. 2. The two zymosan particles were attached to the macrophage within 6 s of each other (Fig. 2(d)) from the opposite direction. Phagocytosis started within 30 s on both sides, and both zymosan particles were phagocytosed simultaneously. The process of phagocytosis proceeded in the same manner and finished at almost the same time (343 s from the start). Six more experiments produced the same results: when the second stimulation occurred within 10 s of the first stimulation, simultaneous phagocytosis occurred.

Next, we stimulated the isolated macrophage with two zymosan particles with different timing (Fig. 3). Just as in the previous experiment, we first stimulated one side of the macrophage with a zymosan particle using optical tweezers (Fig. 3(a–c)). Just 117 s after confirming the start of phagocytosis in the first attachment (Fig. 3(d)), we attached the second zymosan particle to the surface of the macrophage (Fig. 3(e)). Then, as shown in Fig. 3(e–g), even though the second zymosan was attached to the surface of the macrophage, phagocytosis did not start until the first phagocytosis process was finished (Fig. 3(h)). It should be noted that the required time to start the second phagocytosis process was less than 10 s after completion of the first process. Moreover, the time to complete phagocytosis for the first stimulation was about 590 s, whereas it took 1140 s for the second stimulation – about twice as long. When the second stimulation occurred more than 90 s after the first stimulation, the same delayed response of the second phagocytosis was observed in all four subsequent experiments.

To confirm the magnitude of variability of phagocytosis, we also measured the process of phagocytosis of single cells in the case of a single stimulation of zymosan. Figure 4 shows one example of the phagocytic process. The averaged time (from six samples) for the start of phagocytosis after attachment was 97 s, and it was 748 s for the complete phagocytosis process.

The results indicate that the delay of the second stimulation can produce a different response in the second phagocytic process of the macrophage depending on the timing of the second stimulation. If the second antigen stimulation started within 10 s of the first stimulation, the response of the macrophage was simultaneous. In contrast, if the second stimulation was delayed more than 47 s after the first stimulation, the phagocytosis of the second stimulant did not start until after the first phagocytosis was finished. As the waiting time for the second phagocytosis (480 s in Fig. 3) was much longer than the variability of the starting time of phagocytosis (average 97 s, max. 155 s in Fig. 4), the delay in the process after the second stimulation was not due to the variability of phagocytosis, but was apparently due to neglect during the first process even though the cell had been stimulated by the second stimulant. The two different macrophage responses to two stimulations indicate that some mechanism exists to control the timing of phagocytosis in the event of multiple stimulations. This shows the potential for simultaneous phagocytosis from two zymosan particles in different areas on the macrophage, as shown in Fig. 2. It also indicates that the initial phagocytic process can prevent a subsequent phagocytic process from occurring during the first process.

One possible explanation is that there may be a gathering of receptors on the cell membrane to the first antigen, and this may cause a lack of ability to sense the second stimulation at the opposite side until those receptors are released from the first antigen. The same gathering phenomena of sensor proteins were reported in T-cell receptors 10 11 12 . If the movement of sensing proteins on the macrophage is the explanation for these differences in response, the sensor proteins should move faster than 1 μm/s (10 μm of movement for less than 10 s) to respond to the second antigen within 10 s after the first phagocytosis is finished (see Fig. 3). That is, sensor molecules should disperse from one side of the macrophage to the other (ca. 10 μm in diameter) within 10 s. This diffusion velocity is within the magnitude of free diffusion velocity of cell membrane proteins, 5–10 μm²/s. In contrast, recent studies found that diffusion rates of many transmembrane proteins in the cell membrane are much lower than those in artificial reconstituted membranes by a factor of as much as 10 to 100, because the transmembrane proteins are corralled, or they undergo hop diffusion 13 14 . From this viewpoint, the movement of the sensor proteins for phagocytosis appears to resemble free diffusion rather than anchored transmembrane proteins or hop diffusion transmembrane proteins.

In conclusion, we applied an on-chip single-cell cultivation system to measure plural stimulation of antigens on the surface of isolated macrophages and found that a delayed second stimulation might be neglected until the first phagocytosis was complete. This phenomenon indicates that the phagocytic system does not work independently of the condition of the other side of the cell.

KM, KO, and YW carried out the microchamber design, cell preparation, single cell observation, image analysis and also drafted the manuscript. Their contributions were equal. KY conceived of the study and participated in its design and coordination. All authors (KO was represented by KY) read and approved the final manuscript.

Ethical Permission No. 42 (to Yasuda Lab., April 1 2005, to March 31, 2006) was obtained from The Ethical Permission Organization of Animal Experiments in the Graduate School of Arts and Sciences, The University of Tokyo.