This study used gene profiling and functional analyses to examine relationships between microglial activation states, myelin phagocytosis and ROS production, and the role of selected ion channels in these processes. Because the treatments and outcomes examined were complex, for clarity in comparing with previous studies, the salient findings will be discussed under four topics: (1) outcomes of single stimuli used to polarize microglia to different activation states, (2) effects of myelin phagocytosis on their activation state, (3) attempts to re-polarize microglial activation, and (4) expression and contributions of the ion channels in different activation states.
Effects of single stimuli
The first step was to validate the procedures used to stimulate primary rat microglia. In agreement with well-known markers [
17‐
20], we found that IFN-γ combined with TNF-α (I + T) induced a pro-inflammatory M1-like state, while IL-4 induced an anti-inflammatory (M2a) state. Thus, the cytokine concentrations we used were effective in changing the molecular inflammatory profile, as measured at 24 h.
It is expected that phagocytosis by activated microglia will depend on the target, whether it is opsonized, and which phagocytosis-related receptors are engaged. For instance, in the damaged CNS, if extravasation of complement or antibodies occurs, this can engage CR3 and Fcγ receptors, respectively. Furthermore, under in vitro conditions, if complement is present (i.e., if serum is not heat-inactivated), this can greatly promote myelin phagocytosis [
32]. We previously showed that unstimulated rat microglia can phagocytose polymer beads, yeast, and
Escherichia
coli bacteria [
11,
29,
65] and that
E. coli phagocytosis was robustly increased by LPS and IFN-γ, separately or in combination [
29]. Here, untreated rat microglia robustly phagocytosed myelin, and this was modestly increased by I + T (M1) and IL-10 (M2c), but not by IL-4 (M2a). Our results are entirely consistent with an earlier study of rat microglia, in which stimulation with mouse recombinant IFN-γ, TNF-α, or IL-10 (note species mismatch) increased myelin phagocytosis, but IL-4 did not [
22]. In contrast, IL-4 or IL-13 stimulation increased phagocytosis of myelin by human microglia [
23] and of apoptotic cells by rat microglia [
66].
We next asked whether there were changes in expression of specific phagocytosis-related receptors in different activation states.
(i) For M1 stimulation, previous reports are inconsistent and possibly species dependent. Some changes seen in rat microglia are expected to increase their phagocytic capacity. Our earlier study found that LPS increased FcγRIa and FcγRIIIa, while phagocytosis of
E. coli increased CR3 and SR-A [
29]. The stimulatory phagocytic receptor, FcγRIIIa, is often used as an M1 marker [
67,
68]. We found that I + T (M1) increased FcγRIIIa, as well as TIM-3, which promotes phagocytosis of apoptotic neurons [
51]. Other changes might dampen the phagocytic capacity, but again, there are possible species differences, e.g., LPS reduced FcγRIIIa in murine microglia [
21]. The TREM2 receptor aids in target internalization by microglia [
69,
70]. We found that I + T dramatically decreased TREM2 in rat microglia; however, LPS increased it in murine microglia [
71]. Fractalkine (CX
3CL1) is an important chemotactic signal released by apoptotic cells [
72] but in rat microglia, LPS decreased its receptor, CX
3CR1 [
73], as did I + T in the present study. Based on the changes we observed with M1 stimulation (and lack of changes in FcγRIIb, CD11b, P
2Y
6), we suggest that the most likely contributors to increased myelin phagocytosis were the reduced inhibitory SIRPα signal and a known increase in the ability of CR3 to bind targets under pro-inflammatory conditions [
74]. Despite the increase in FcγRIIIa, it is not likely involved. It binds to the Fc component of antibodies but antibody-mediated opsonization of the myelin debris should not occur because the culture medium contained heat-inactivated serum.
(ii) For M2 stimulation, published data on phagocytosis-related receptors are very limited and, again, there might be species differences. For rat microglia, we found that IL-4 treatment (M2a) increased expression of the inhibitory receptor, FcγRIIb, and substantially decreased CD11b, SR-A, CD68, and TREM2 (no changes in SIRPα, TIM-3). For murine microglia, IL-4 did not change FcγRIIb (CD32) expression [
21]. Surprisingly, CD68 expression decreased in both M1 and M2a microglia and after myelin phagocytosis by unstimulated cells. CD68 is commonly used to identify activated, phagocytic microglia [
7,
42‐
44]. Together, these results suggest that changes in receptor expression are not reliable predictors of the degree of myelin phagocytosis. Instead, protein levels and modulation might be more important. For instance, CR3 can potentiate or inhibit myelin phagocytosis depending on its conformation [
32]. Although the changes we observed suggest a less phagocytic phenotype in the M2a state, myelin phagocytosis was comparable to untreated microglia. Of course, effects on phagocytosis of other targets after CNS damage (e.g., apoptotic neurons, cell debris, infiltrating blood cells) might differ by involving different receptors.
Phagocytosis is associated with elevated NOX-mediated ROS production to help kill engulfed pathogens [
4,
54]. It was previously reported that untreated rat microglia robustly express the NOX2 isoform, with much lower NOX1 and NOX4 levels, and undetectable NOX3 [
53]. Our results confirm this pattern and extend it to the M1 and M2a states.
(i) Increased ROS production is a hallmark of M1 activation (reviewed in [
20]), and we found that it was increased in I + T-treated microglia. NOX2 was likely responsible because expression of both NOX2 and its regulatory subunit, Ncf1, were increased to very high levels, while NOX1 and NOX4 remained at very low levels.
(ii) IL-4 slightly increased ROS production while IL-10 had no effect, and this is consistent with our recent report [
34]. IL-4 decreased expression of NOX2 and did not change Ncf1; however, both remained at moderate levels and could account for the ROS production.
Effects of myelin phagocytosis
There is some evidence that myelin phagocytosis can affect the M1 activation state. Effects are potentially time dependent, and negative self-regulation might protect the cells from “overeating” during extended exposures to targets [
2]. When murine microglia were stimulated with IFN-γ or LPS, a short exposure to myelin (≤6 h) exacerbated the pro-inflammatory response [
10], while longer exposures (16–24 h) dampened this response [
10,
75]. Using the short exposure time (6 h), which was sufficient for optimal myelin uptake (see the “
Methods” section); we found little effect on the molecular profile of unstimulated rat microglia. In contrast, myelin increased expression of pro-inflammatory cytokines in M1 (I + T-treated) rat microglia. This is consistent with the previous short exposure study [
10]. In IL-4-treated cells, myelin reduced several M2a-associated molecules. These results suggest that myelin can skew activated rat microglia toward a pro-inflammatory state.
Because levels of phagocytosis-related receptors were not well predicted by the microglial activation state (above); it was important to ask whether they were altered by exposure to myelin debris. In unstimulated microglia, myelin slightly decreased SIRPα and considerably increased CX
3CR1 and P
2Y
6, which are all expected to promote phagocytosis, especially of apoptotic cells.
(i) In I + T-treated (M1) cells, myelin did not alter expression of receptors known to be involved in myelin phagocytosis (CD11b, SR-A, TREM2, CX
3CR1, SIRPα). Instead, it increased P
2Y
6, FcγRIIIa, C1r, and TIM-3 and slightly decreased FcγRIIb, changes that could promote phagocytosis of other targets.
(ii) In IL-4-treated (M2a) cells, myelin greatly increased CX
3CR1 and slightly increased P
2Y
6. CD68 was slightly decreased, and CD11b or SR-A were unchanged. Interestingly, CX
3CR1 [
76] and P
2Y
6 [
7] promote microglial migration toward damaged cells, and we previously found that M2a-activated rat microglia migrate better [
25]. Thus, exposure to myelin debris might further potentiate the migratory capacity of M2a-activated microglia.
Myelin phagocytosis increases ROS production by unstimulated microglia [
10,
22]. We confirmed this and showed that the myelin-evoked ROS production required NOX activity under all activation conditions tested (I + T, IL-4, IL-10). Myelin did not affect expression of NOX enzymes, but it increased expression of the proton channel, Hv1, which could contribute to the increased ROS production seen in unstimulated and M1-activated cells. Moreover, myelin binding to Mac1 (CD11b) can activate NOX2 and promote ROS generation [
77]. Interestingly, inhibiting NOX activity reduced myelin phagocytosis under all activation conditions tested. While we do not know the mechanism, an earlier study of rat macrophages suggested that NOX-mediated ROS production promotes signaling mechanisms involved in myelin phagocytosis [
78].
Sequential cytokine stimulation
After acute CNS injury, the inflammatory milieu changes over time [
79,
80]. However, whether microglial activation states are functionally plastic is poorly understood. It is important to determine if, once polarized, they can respond to new signals. Based on the few studies that have addressed time-dependent changes in the overall inflammatory state in vivo, the outcome might depend on the type of injury. In the cuprizone-induced de-myelination model, an overall M1 state gave way to an M2a phenotype at the time of re-myelination [
68]. In the first week after intracerebral hemorrhage, we observed concurrent elevation of pro- and anti-inflammatory mediators [
81,
82]. However, after cerebral ischemia or traumatic brain injury, murine microglia exhibited an early M2 state, followed by M1 [
67,
83]. An in vitro study of rat microglia found that adding IL-4 (M2a) before LPS (M1) decreased expression of the M1-associated molecules, COX-2, iNOS, and TNF-α compared with LPS alone [
84]. Similarly, in mixed rat glial cell cultures, simultaneous addition of LPS and IL-4 (or IL-10; M2c) reduced IL-6, TNF-α, and NO production, compared with LPS alone [
85]. Both studies assessed pro-inflammatory mediators only. For murine microglia, when LPS was followed by IL-4, NOS2 and COX-2 expression decreased, while CD206 (MRC1) and Arginase 1 (M2a markers) increased compared with LPS alone [
21,
86]. The present study greatly extends these previous reports.
We examined effects of sequential addition of M1- and M2-inducing cytokines on the inflammatory profile, expression of phagocytosis-related receptors and ROS-related molecules, and on myelin phagocytosis and consequent ROS production. We employed four sequential treatment paradigms that address the possibility that the cytokine profile changes after injury or disease. The most convincing re-polarization of the inflammatory state was between I + T and IL-4 treatments, applied in both sequences. (i) M1→M2a paradigm: In I + T-primed microglia, adding IL-4 dampened the pro-inflammatory profile (NOS2, TNF-α, IL-6, COX-2) and increased M2a markers (MRC1, c-myc, CCL22). This is entirely consistent with previous studies using LPS and IL-4 (cited above). The relationship of phagocytosis and ROS production to expression of receptors and enzymes and to phagocytosis- and ROS-related molecules was complicated and, sometimes, unexpected. Although myelin phagocytosis was increased, several phagocytosis-promoting receptors decreased (CD11b, FcγRIIIa, CD68, C1r, TIM-3), compared with I + T alone. Among the inhibitory receptors, SIRPα decreased and FcγRIIb was unchanged. Thus, the decrease in SIRPα might have promoted phagocytosis. Despite the lack of change in ROS production, ROS-related molecules decreased (NOX enzymes, Ncf1, Hv1), suggesting that the remaining levels were sufficient. (ii) M2a→M1: In IL-4-treated cells, adding I + T skewed them toward an M1 profile. There was increased expression of most pro-inflammatory molecules (NOS2, TNF-α, COX-2), decreases in some M2 markers (MRC1, CD163), and increased myelin phagocytosis and ROS production. The outcome might depend on the exact stimulus paradigm and target type. For instance, we observed some changes in receptor expression that are expected to promote phagocytosis: an increase in P2Y6 and decreases in the inhibitory receptors, FcγRIIb and SIRPα. Most phagocytosis-promoting receptors decreased, particularly CD11b, TREM2, CD68, TIM-3, and CX3CR1. (iii) M1→M2c: In I + T-primed cells, adding IL-10 did not resolve the pro-inflammatory state and, surprisingly, it increased iNOS/NOS2 expression. While expression of several phagocytosis-related receptors increased (CD11b, FcγRIIIa, TIM-3), SIRPα increased slightly, while phagocytosis and ROS production were unchanged. (iv) M2c→M1: In IL-10-treated cells, adding I + T increased myelin phagocytosis and ROS production (gene changes were not examined, as explained above).
Overall, our results show malleability in re-polarization of rat microglia between M1 and M2a states. Their qualitative ability to respond to a new incoming signal was preserved but their quantitative response was often reduced. The amount of re-polarization could well depend on the experimental paradigm (cytokine concentrations, time course). For instance, although we did not observe resolution of M1 activation by IL-10, it is possible that the time course of treatment or monitoring was not optimal. In the future, in vivo spatial and temporal changes in the cytokine environment will need to be examined in each damage/disease model to further examine the re-polarization capacity of microglia.
Expression and contributions of ion channels
Ion channels regulate numerous processes in cells that are relevant to phagocytosis, including cell volume, Ca
2+ signaling, and cytoskeletal re-organization [
87]. Microglia express a surprisingly large array of ion channels, some of which are involved in proliferation, migration, and Ca
2+ signaling (reviewed in [
42,
88]). Very little is known about roles of channels in phagocytosis by microglia; particularly in different activation states. We previously found that Cl
− channels regulate phagocytosis of
E. coli by rat microglia [
65]. Phagocytosis is a Ca
2+-dependent process that involves SOCE [
89,
90]. The CRAC channel is apparently the major SOCE pathway in rat microglia [
33,
35,
91,
92]. Interestingly, CRAC can be activated by P
2Y
6 and other G protein-coupled metabotropic receptors, and we found that phagocytosis of myelin debris increased P
2Y
6 expression under all activation states examined. We then focused on CRAC and three K
+ channels (KCa3.1, Kir2.1, Kv1.3) that are thought or known to regulate Ca
2+ entry. In rat microglia, KCa3.1 and Kir2.1 regulate CRAC-mediated Ca
2+ influx [
24,
33,
35], and Kv1.3 is involved in ROS production [
14,
37]. CRAC channels are comprised of a pore-forming Orai1 subunit and a Ca
2+-sensing STIM1 subunit [
64]. While other subunits are considered less important, SOCE in murine microglia might also involve STIM2 [
90], and Orai3 activity can also be regulated by STIM proteins [
64]. Rodent microglia express mRNA for Orai1, Orai3, STIM1, and STIM2 [
89‐
91]. In murine microglia, Orai1, STIM1, and STIM2 contribute to SOCE and phagocytosis [
89,
90]. There are few reports regarding changes in Ca
2+ signaling and expression of relevant Ca
2+-signaling molecules in specific activation states; and the results are somewhat inconsistent. For murine microglia, LPS (M1) increased STIM1 without affecting Orai1 or Orai3 in one study [
89], but reduced STIM1 and Orai3 without affecting Orai1 or STIM2 in another [
90]. Orai1 expression was not changed in either study but SOCE was reduced. The Ca
2+ entry pathway was not determined. An earlier study of murine microglia reported that LPS rapidly elevated basal Ca
2+ and reduced the UTP-induced rise [
93]. For human microglia, M1 stimulation (GM-CSF + LPS + IFN-γ) did not affect the Ca
2+ response to ADP; whereas, M2 activation (M-CSF + IL-4 + IL-13) increased it and this was attributed to increased P
2Y
12 receptor expression [
94]. For rat microglia, we recently found that the CRAC-mediated Ca
2+ rise was ~50 % lower after IL-4 treatment (M2a) but not affected by IL-10 (M2c) [
35].
In unstimulated rat microglia, myelin affected expression of Kv1.3 only (increased) but channel expression was strongly affected by the microglial activation state.
(i) I + T-treated (M1) cells had increased expression of Kv1.3, KCa3.1, Kir2.1, STIM1, STIM2, and Orai3. Myelin phagocytosis increased Kv1.3, KCa3.1, Orai1, and Orai3. The high STIM expression and increase in Orai1, the pore-forming subunit of CRAC, should facilitate Ca
2+ signaling.
(ii) IL-4-treated (M2a) cells had increased Kv1.3 and KCa3.1 compared with control cells. Myelin had no further effects. No Orai or STIM molecules increased but STIM2 decreased slightly. Thus, the previously observed decrease in CRAC signaling [
35] is not readily explained by Orai or STIM expression.
(iii) Sequential cytokine addition had complex effects on channel expression. M1→M2a: Compared with I + T stimulation alone, subsequent IL-4 addition reduced KCa3.1, Kv1.3, Kir2.1, Orai3, STIM1, and STIM2. M2a→M1: Compared with IL-4 alone, subsequent I + T addition decreased expression of KCa3.1, Kv1.3, Orai3, STIM1, and STIM2, but not Kir2.1. M1→M2c: Compared with I + T alone, subsequent addition of IL-10 only changed KCa3.1, which was increased. Overall, with respect to ion channels, rat microglia showed considerable re-polarization between M1 and M2a states, while IL-10 was ineffective.
The use of selective channel blockers showed that CRAC was important for phagocytosis under all activation conditions examined, while myelin phagocytosis and ROS production by activated microglia were both dependent on Kir2.1 (but not Kv1.3 or KCa3.1). What could account for a lack of contribution of Kv1.3 and KCa3.1 when, in principle, all routes of K
+ flux can control the membrane potential of cells? The simplest possibility is that, despite substantial transcript expression, Kv1.3 and KCa3.1 were not active during myelin phagocytosis. We propose a scenario in which Kv1.3 and KCa3.1 are inhibited, while Kir2.1 and CRAC are facilitated. All three K
+ channels are post-translationally regulated by signaling molecules downstream of the phagocytosis receptors, CR3 and SR-A. CR3 signaling activates Src family tyrosine kinases, phosphoinositide-3-kinase (PI3K) and phospholipase C (PLC) [
74,
95]. Kv1.3 is strongly inhibited by activated Src in rat microglia [
36]. Lipid phosphatases localize to phagosome cups (see reviews [
55,
74]), and the lipid phosphatase, myotubularin-related protein 6 (MTMR6) regulates macropinocytosis [
96], which uses similar machinery to phagocytosis [
97]. KCa3.1 is strongly inhibited by MTMR6 [
98]. How might Kir2.1 and CRAC channel activity be promoted? Both CR3 and SR-A signaling involve PLC [
99] and PI3K [
100]. PI3K generates PIP2, which stabilizes the open configuration of Kir2.1 channels [
101]. PLC activity generates diacylglycerol (DAG) and inositol triphosphate (IP
3), which depletes ER calcium stores and activates CRAC channels [
55]. In addition, DAG activates protein kinase C, which can stimulate NOX enzymes and increase ROS production [
54].