TRPM2 ion channels steer neutrophils towards a source of hydrogen peroxide

Neutrophils must navigate accurately towards pathogens in order to destroy invaders and thus defend our bodies against infection. Here we show that hydrogen peroxide, a potent neutrophil chemoattractant, guides chemotaxis by activating calcium-permeable TRPM2 ion channels and generating an intracellular leading-edge calcium “pulse”. The thermal sensitivity of TRPM2 activation means that chemotaxis towards hydrogen peroxide is strongly promoted by small temperature elevations, suggesting that an important function of fever may be to enhance neutrophil chemotaxis by facilitating calcium influx through TRPM2. Chemotaxis towards conventional chemoattractants such as LPS, CXCL2 and C5a does not depend on TRPM2 but is driven in a similar way by leading-edge calcium pulses. Other proposed initiators of neutrophil movement, such as PI3K, Rac and lyn, influence chemotaxis by modulating the amplitude of calcium pulses. We propose that intracellular leading-edge calcium pulses are universal drivers of the motile machinery involved in neutrophil chemotaxis.

www.nature.com/scientificreports/ We next used the classical Boyden chamber method to measure neutrophil recruitment by H 2 O 2 in vitro. In this assay, a membrane with 3 µm pores, through which only neutrophils can penetrate, separates the neutrophil and chemoattractant compartments. H 2 O 2 down to 1 nM concentration caused significant neutrophil recruitment through the membrane, with maximal recruitment between 10 and 10 μM H 2 O 2 and a similar inhibition to that observed in vivo at 100 μM (Fig. 1B).
Similar results were obtained using a second in vitro method, the ibidi μ-slide, in which neutrophil motion is continuously tracked under a microscope (Fig. 1C,D and Supplementary Video 1). Figure 1C shows the speed of movement of neutrophils as a function of concentration of H 2 O 2 . Neutrophils undergo random movements in the absence of H 2 O 2 , so the average speed is not zero in the absence of H 2 O 2 even though there is no significant directed movement (see Fig. 1D). The average speed is increased in a gradient of 1 nM H 2 O 2 distributed over 1 mm, the distance between the compartments containing H 2 O 2 and DMEM. Average speed is maximal between 0.1 and 10 μM H 2 O 2 and is inhibited at 100 μM H 2 O 2 . Maximal speed in fMLP is not significantly different from that in H 2 O 2 .
An alternative method of quantifying neutrophil movement is to measure the forward migration index (FMI), a measure of directionality, that gives the ratio of linear distance travelled in the direction of the H 2 O 2 gradient to total distance travelled (Fig. 1D). A significant increase in FMI was observed at 10 nM H 2 O 2 , and FMI was maximal between 0.1 and 10 μM H 2 O 2 . As in the other assays, directed migration was inhibited at 100 μM H 2 O 2 .
All three independent measures agree in showing that H 2 O 2 is a highly potent neutrophil chemoattractant, from low nanomolar levels up to 10 μM, and that chemoattraction is inhibited at concentrations of H 2 O 2 above 10 μM. A similar concentration-dependence of directed migration in a gradient of H 2 O 2 was also observed with human blood neutrophils (Fig. 1E).

Sensitivity of neutrophil chemotaxis to single molecules of H 2 O 2 .
The evidence above shows that neutrophils exhibit a clear increase in both movement and direction-finding in concentration gradients as low as 10 nM H 2 O 2 , distributed over the 1 mm gap between the chemoattractant and DMEM compartments of the ibidi μ-slide. This gradient corresponds to a concentration difference between the leading and trailing edge of a 10 μm diameter neutrophil of 100 pM, and, assuming that H 2 O 2 equilibrates rapidly and completely between the extracellular and intracellular spaces, this is a difference on average of around five intracellular molecules of H 2 O 2 between the leading and trailing halves of the neutrophil (see calculation in legend to Supplementary  Materials Fig. 2). The most sensitive neutrophils are therefore able to detect and respond reliably to only a few molecules difference of intracellular H 2 O 2 between the leading and trailing halves of the cell. This extraordinary sensitivity to H 2 O 2 suggests that the detection of H 2 O 2 is a highly-evolved property, critical for survival.

Neutrophil chemotaxis towards H 2 O 2 depends on TRPM2.
The TRPM2 ion channel is expressed in neutrophils 12,13 and is potently activated by H 2 O 2 14,15 , suggesting that it may be a candidate for the sensor of H 2 O 2 . H 2 O 2 is thought to act indirectly to generate ADPR, which in turn directly activates TRPM2 16,17 . TRPM2 −/− mice are highly susceptible to infection 18 , suggesting that the absence of TRPM2 causes a critical immune system defect. We found that genetic deletion of TRPM2 strongly inhibited H 2 O 2 -dependent neutrophil migration in all three assays ( Fig. 1A-D), supporting the idea that TRPM2 is the major physiological sensor of H 2 O 2 in neutrophils.
Neutrophil chemotaxis towards ADPR depends on TRPM2. The intracellular C terminus of TRPM2 contains a nudix hydrolase homology domain (NUDT9-H), named after the mitochondrial ADP-ribose (ADPR) pyrophosphatase NUDT9, to which ADPR binds to activate the ion channel 15,[19][20][21][22][23][24][25] . We found that neutrophils responded vigorously to an extracellular gradient of ADPR (Fig. 1F). Migration up a gradient of ADPR was strongly inhibited by deletion of TRPM2 (Fig. 1F), showing that neutrophil migration up a gradient of ADPR also depends on TRPM2. These experiments suggest that externally applied ADPR may be able to cross the cell membrane and activate the intracellular TRPM2 binding site. The idea that a large and hydrophilic molecule such as ADPR may be able to cross the cell membrane may seem unlikely, but it is supported by the activation of a TRPM2-dependent calcium influx when ADPR is applied externally (see Fig. 3G below), and also by work in other labs 25,26 . It is possible that a specific carrier could transport ADPR across the cell membrane in order to activate the intracellular TRPM2 binding site.
Chemotaxis towards most conventional chemoattractants does not depend on TRPM2. The marked inhibition of neutrophil migration up gradients of H 2 O 2 or ADPR, caused by deletion of TRPM2, could be due to an effect of the deletion on migratory ability per se. This possibility is not, however, supported by the observation that TRPM2 deletion does not affect chemotaxis to the known chemoattractants lipopolysaccharide (LPS), C5a and CXCL2 ( Fig. 2A). Deletion of TRPM2, however, reduces but does not completely abolish chemotaxis towards the bacterial/mitochondrial peptide fMLP ( Fig. 2A), as also noted in all assays shown in Fig. 1A-D, suggesting that fMLP exerts its chemotactic effect in part by activating TRPM2.

Neutrophil chemotaxis towards H 2 O 2 or ADPR is inhibited by blocking TRPM2.
To test further the dependence of neutrophil migration on TRPM2, we examined the effect of pharmacological block of TRPM2 with the inhibitor ACA (N-(p-amylcinnamoyl)anthranilic acid 27 ). Block of TRPM2 with ACA caused a significant inhibition of chemotaxis in response to both H 2 O 2 and ADPR ( Fig. 2B and Supplementary Materials Fig. 1B), similar in magnitude to the effect of genetic deletion of TRPM2 shown in Fig. 1. Block of TRPM2 with ACA had no effect, however, on chemotaxis towards LPS, C5a and CXCL2, but caused a partial inhibition of migration towards fMLP (Fig. 2B). ACA has off-target actions 27  www.nature.com/scientificreports/ with genetic deletion of TRPM2 supports the hypothesis that the effects of both genetic deletion and pharmacological block of TRPM2 are due to inhibition of TRPM2 itself and not to an off-target effect on another protein. Figure 2C shows similar experiments on human blood neutrophils. All chemoattractants strongly increased FMI to a similar extent, as was also observed with mouse peritoneal neutrophils ( Fig. 2A,B). In Fig. 2D, block of TRPM2 with ACA in human blood neutrophils largely inhibited migration up a gradient of H 2 O 2 but had no effect on migration up a gradient of IL-8, the human homologue of CXCL2, results closely resembling those in mouse neutrophils (Fig. 2B).
In summary, the results in Figs Fig. 2). The reason for this diversity in response is unknown but could be related to the age or condition of different neutrophils in the pool. www.nature.com/scientificreports/ We tested whether the apparently lower maximal values of FMI in some experiments (for example with ADPR, Fig. 1F) was due to a real difference or simply reflected variability between different neutrophil samples. Neutrophil migration to H 2 O 2 depends strongly on temperature. TRPM2 is a member of the large TRP family of ion channels, and like several other members of this family, is strongly activated by increasing temperatures within the physiological range 15,[28][29][30][31] . Figure 3A shows that neutrophil migration towards very low concentrations of H 2 O 2 is potently enhanced by small increases of temperature and that the effect of temperature is abolished by deletion of TRPM2. The effect of temperature is particularly prominent in a gradient of 1 nM H 2 O 2 , when an elevation from 37 to 39 °C, corresponding to a mild fever, enhances the ability of neutrophils to migrate up a gradient of H 2 O 2 from an FMI of 0.052 ± 0.028 to 0.277 ± 0.059, a 5.3-fold increase. By contrast, the effect of temperature on migration towards the conventional chemoattractants CXCL2, C5a and LPS is much less marked (Fig. 3B-D). The strong temperature dependence of neutrophil migration towards H 2 O 2 provides further support for a critical role of TRPM2 in neutrophil migration in response to H 2 O 2 , and moreover suggests a reason why fever may be beneficial in fighting infection, because an elevation of temperature by only 2℃, corresponding to a mild fever, gives a five-fold increase in neutrophil migratory ability towards low levels of H 2 O 2 .
Temperature and chemoattractants elevate intracellular calcium by activating TRPM2. TRPM2 is permeable to calcium ions 14 , like other members of the TRP family 32,33 , suggesting that the striking dependence of neutrophil migration on temperature may be due to a calcium influx through activated TRPM2 ion channels. Figure 3E shows that elevations of temperature in the physiological range (from 37 to 41 °C) induce an intracellular calcium increase in neutrophils, and that low concentrations of H 2 O 2 (1 nM and 10 nM) enhance the amplitude of the calcium increase and lower its temperature threshold. Deletion of TRPM2 abolishes the thermallyinduced elevations of calcium (Fig. 3F). The strong temperature-dependence of calcium influx through TRPM2 supports the hypothesis that a calcium influx through TRPM2 underlies the striking temperature-sensitivity of neutrophil migration driven by H 2 O 2 (Fig. 3A). The results also support the idea that calcium influx through TRPM2 may be a driver of neutrophil motion, for which further evidence is presented below.
We next examined whether H 2 O 2 , ADPR and fMLP, whose effect on neutrophil migration depends on activation of TRPM2 (see above), also induce a calcium influx. Figure 3G and H show that in all cases these agonists cause an elevation of intracellular calcium, from the resting level of around 100 nM to 500 nM, and that the increase is dependent on TRPM2. There was a significant delay between application of H 2 O 2 or ADPR and activation of calcium influx, consistent with their known intracellular sites of action 16 . H 2 O 2 is able to cross the cell membrane through aquaporins 34 , and ADPR is also known to cross the neutrophil cell membrane, though the mechanisms are not clear 25,26 . There was a rapid drop in intracellular calcium when external calcium was removed (Fig. 3G, left-hand panel), showing that activation of TRPM2 causes an influx of calcium from the extracellular solution.
The above experiments using calcium imaging are in agreement with patch-clamp recordings both from HEK cells transfected with TRPM2 and from isolated TRPM2-expressing membrane patches, which show that TRPM2 is directly activated by heat and by intracellular ADPR 13,15,16,30,35 . Recordings from membrane patches have shown, however, that H 2 O 2 does not directly activate TRPM2 16 . The action of H 2 O 2 on intact cells is therefore likely to be caused by generation of intracellular ADPR which then activates the TRPM2 ion channel (see diagram in Supplementary Materials Fig. 8).
Leading-edge calcium pulses in H 2 O 2 and ADPR are driven by TRPM2. The results above suggest that preferential activation of TRPM2 channels, by elevated levels of H 2 O 2 at the neutrophil leading edge, could cause a localised calcium influx that guides neutrophil chemotaxis by coupling to the motile machinery. We used calcium imaging of migrating neutrophils and we observed that an elevated intracellular calcium concentration at the neutrophil leading edge is indeed seen during movement of a neutrophil up a gradient of H 2 O 2 (Fig. 4A, left panel). Leading-edge intracellular calcium elevations were seen in all migrating neutrophils, and they had a pulsatile appearance for which the name "calcium pulse" seemed appropriate (see for example Supplementary Video 2). The calcium pulse is most prominent at the base of the pseudopodium, but in some images it can be seen also invading the advancing pseudopodium itself (Figs. 4B,C and 5A,B). Genetic deletion of TRPM2 abolishes both calcium pulses and cell migration (Fig. 4A, centre, and Fig. 4B-E). At high concentrations of H 2 O 2 , maximal activation of TRPM2 over all of the cell membrane means that a uniformly high level of calcium floods the cell and neutrophil migration is abolished (Fig. 4A, right).
Pharmacological block of TRPM2 with the inhibitor ACA 36 has a similar effect to genetic deletion of TRPM2 in inhibiting calcium pulses and neutrophil chemotaxis towards H 2 O 2 and ADPR ( Fig. 4B-E). Both calcium pulses and neutrophil movement are abolished by withdrawal of extracellular calcium (Fig. 4B,D), consistent with an influx of calcium from the external medium acting as a trigger for the generation of intracellular calcium pulses. Lowering the temperature to 33 °C, which strongly inhibits neutrophil motility (Fig. 3A), also inhibits the generation of calcium pulses (Fig. 4B,D and Supplementary Materials Fig. 4B). These parallel observations using a range of approaches support the idea that leading-edge calcium pulses in both H 2 O 2 and ADPR are generated by an influx of calcium through TRPM2, and that calcium pulses are coupled to neutrophil migration.
Leading-edge calcium pulses in classical chemoattractants are not driven by TRPM2. www.nature.com/scientificreports/ suggest that calcium pulses "steering" neutrophil chemotaxis, by determining the future direction of establishment of pseudopodia, may be a universal mechanism. Block of TRPM2 with ACA has no effect on calcium pulses in the classical chemoattractants CXCL2, LPS and C5a ( Together with the data in Fig. 2, showing the lack of effect of TRPM2 deletion or block on neutrophil migration to classical chemoattractants, these experiments show that neither the generation of leading-edge calcium pulses nor neutrophil migration in these chemoattractants is driven by a TRPM2-dependent mechanism. Signalling pathways determining neutrophil motility. There have been several proposals regarding the mechanisms that couple chemoattractants to the intracellular motile machinery of immune cells, including activation of PI3K to generate PIP3 at the neutrophil leading edge, leading to activation of Rac 1,37-40 ; or, in the case of chemoattraction by H 2 O 2 , oxidative modulation of cysteine residues present in the Src family tyrosine kinase lyn 9,11 . , showing that the effect of SFK inhibition is specific to the H 2 O 2 /TRPM2 mechanism. One possibility is that lyn may promote trafficking of TRPM2 to the surface membrane, and thus that inhibition of lyn would reduce the calcium influx triggered when TRPM2 is activated by H 2 O 2 by reducing surface membrane expression of TRPM2. In support of this proposal, an SFK member has been shown to play a similar role in promoting trafficking of TRPV1 to the neuronal cell surface membrane 41 . PI3K and Rac have also been suggested as initiators of neutrophil chemotaxis 1,37-40 . However, the PI3K inhibitor wortmannin and the Rac inhibitor EHT1864 only partially inhibited neutrophil chemotaxis in response to H 2 O 2 , CXCL2, LPS and C5a ( Fig. 5 and Supplementary Materials Fig. 5, 6). In the case of chemotaxis in a gradient of H 2 O 2 , the residual motility in the presence of the PI3K and Rac blockers was completely inhibited by the TRPM2 blocker ACA (Supplementary Materials Fig. 6A). The PI3K and Rac inhibitors also reduced the amplitudes of leading-edge calcium signals, suggesting that PI3K and Rac play a modulatory role in the pathways upstream of the generation of calcium pulses, but that the amplitude of calcium pulses is the primary event controlling motility in response to all chemoattractants.
Calcium pulses determine the direction of neutrophil chemotaxis. Increases in intracellular calcium in neutrophils and other immune cells have been suggested to be a consequence of cell movement, as might arise if a calcium increase was required to retract the trailing edge of the cell 42 , or because an interaction between cell and substrate triggers calcium entry caused by the movement itself 43,44 . In these scenarios, the calcium increase would be expected to follow the movement, while if the calcium pulse directs the movement it would be expected to precede it. For this reason we investigated whether a change in the cellular location of calcium pulses precedes a subsequent change in the direction of cell migration (calcium drives migration), or whether a change in the direction of migration precedes a change in location of calcium pulses (migration drives calcium). Neutrophils in the absence of a chemotactic gradient still exhibit calcium pulses, which change cellular location at intervals, and also exhibit random migration (See Supplementary Video 2), providing a useful way of distinguishing these two possibilities.   www.nature.com/scientificreports/ We quantified the angle between the geometric centre of the cell and the centre of the calcium pulse occurring at the same moment (calcium vector angle, Fig. 6A) and compared this angle with the angle between the geometric centre of a neutrophil at the moment of the calcium pulse and at different times as the cell advances (movement vector angle, Fig. 6A). In the case of the neutrophil shown in Fig. 6B and Supplementary Video 2, the calcium vector turns from an overall upward direction to an overall downward direction halfway through the recording (shown by a star in Fig. 6B,C). As is also visible in Supplementary Video 2, the change in cell direction follows the change in calcium signal direction with a significant delay.
The calcium vector connecting the cell geometric centre and the calcium pulse is measured at a given moment (blue trace in Fig. 6C), but the movement vector (orange trace) can refer to the cell position at a variable past or future time. Figure 6C shows that a lag of + 75 s between calcium vector and movement vector best describes the moment at which the movement vector begins to reverse in response to the abrupt change in the direction of the calcium vector. This positive delay is consistent with calcium driving the direction of movement, while a negative delay gives a poor fit (shown for a -50 s delay). These results are consistent with a change in calcium pulse location driving a change in movement, with a delay of around 75 s, and does not agree with the idea that movement might cause (i.e. anticipate) the calcium pulse. There is a further significant delay (half-time ~ 83 s) before the cell movement fully aligns with the new direction of the calcium vector (see middle panel in Fig. 6C), giving a total time lag between change of the direction of the calcium pulse and half-time of alignment of movement with the new calcium pulse location of 158 s. Two further neutrophils analysed in the same way gave similar delays, with total half-time for the alignment of cell movement to a change in calcium direction of 146 s and 164 s. The neutrophil movement therefore reacts to a change in the direction of a calcium pulse with a surprisingly long delay, with total half-time of around 160 s, or almost 3 min.
Finally, Fig. 6D shows a comparison between the external H 2 O 2 concentration surrounding a cell migrating half-way up a 10 nM gradient over 1 mm, and the internal calcium concentration measured with fura-2. The calcium gradient is very much steeper than the H 2 O 2 gradient, showing that there must be a powerful non-linear amplification process that converts the gentle H 2 O 2 gradient into the strongly enhanced calcium signal seen at the cell leading edge.

Discussion
The work described here confirms previous studies showing that H 2 O 2 is a potent neutrophil chemoattractant both in vitro 2 and in vivo 3 . We demonstrate an essential role for the TRPM2 ion channel in directing neutrophil chemotaxis towards H 2 O 2 . We propose that preferential activation of TRPM2 by the higher levels of H 2 O 2 at the neutrophil leading edge causes a calcium influx, triggering amplified leading-edge calcium pulses that determine the direction of extension of pseudopodia and therefore of future neutrophil motion. There is a significant delay, of around 160 s, between a change of location of a calcium pulse and the half-time of response of the neutrophil direction, in striking contrast to the millisecond reaction time of many other calcium-driven cellular processes such as skeletal muscle contraction or synaptic vesicle exocytosis.
We find that neutrophils are guided by H 2 O 2 over four orders of magnitude of concentration, from 1 nM to 10 μM. Higher concentrations of H 2 O 2 (> 10 μM) inhibit chemotaxis and cause an elevated calcium concentration throughout the cell. In agreement, previous studies have also found inhibition of chemotaxis in neutrophils and lymphocytes at levels of H 2 O 2 above 10 μM 45,46 . A probable cause of the inhibition of chemotaxis at higher H 2 O 2 is loss of the differential intracellular calcium gradient necessary for cell guidance, as shown in Fig. 4A.
Neutrophil guidance towards most conventional chemoattractants, such as LPS, C5a or CXCL2, is independent of TRPM2, but similar leading-edge calcium pulses are still observed, suggesting that guidance by calcium pulses is a universal mechanism. Signalling from G-protein coupled receptors activated by conventional chemoattractants to intracellular calcium stores is a likely mechanism 47,48 (see diagram in Supplementary Materials  www.nature.com/scientificreports/ Fig. 8). An interesting exception to the TRPM2-independence of conventional chemoattractants is chemotaxis towards the chemoattractant peptide fMLP, which is partially inhibited either by either pharmacological block or genetic deletion of TRPM2. The rapid elevation of intracellular calcium caused by fMLP (Fig. 3G) suggests that activation of FPR1, the receptor for fMLP, may directly activate TRPM2, in addition to coupling to downstream G-protein signalling pathways. This proposal is supported by the recently-discovered physical interaction between TRPM2 and the FPR1 receptor 46 . Previous studies of the involvement of TRPM2 in neutrophil chemotaxis have come to contradictory conclusions. Yamamoto et al 49 found that deletion of TRPM2 decreased neutrophil migration in response to an inflammatory supernatant formed by activated neutrophils in vitro, and also decreased movement to areas of inflammation in vivo. In support, Hiroi et al. 50 found that neutrophil invasion following myocardial ischaemia was lowered in Trpm2 −/− mice. Wang et al. 46 , in contrast, found that deletion of TRPM2 promotes chemotaxis both in vitro and in vivo and they propose that TRPM2 generates a "stop" signal to halt neutrophil migration in the vicinity of pathogens. In the present study we show that neutrophil motility both in vitro and in vivo is potently promoted by low levels of H 2 O 2 , and that motility is inhibited, both in vivo and in vitro, by either pharmacological inhibition or genetic deletion of TRPM2. Thus, at low concentrations of H 2 O 2 , TRPM2 mediates a powerful chemoattractant mechanism. In addition, we find that high concentrations of H 2 O 2 (above 10 µM) inhibit neutrophil motility, which may be critical in providing a "stop" mechanism in the vicinity of pathogens. We show that this inhibition is due to a rise of the internal calcium level throughout the cell when TRPM2 channels are maximally activated by H 2 O 2 , thus flooding the cell with calcium and abolishing the internal calcium gradient necessary for neutrophil navigation. Thus the apparently contradictory conclusions of previous studies can be reconciled by the data obtained in the present work: H 2 O 2 , acting via TRPM2, is a potent chemoattractant at low levels of H 2 O 2 , but at higher levels of H 2 O 2 over-activation of TRPM2 floods the cell with calcium, abolishing the calcium gradient necessary tio drive chemotaxis and halting cell movement.
The ratio between calcium at the leading and trailing edge of an advancing neutrophil is much greater than the ratio between H 2 O 2 concentrations (Fig. 6D), suggesting that a process of non-linear amplification enhances the leading-edge calcium signal. One likely possibility for such a mechanism is the highly non-linear process of calcium-induced calcium release from subcellular stores 25,26,47,48,51 .
A scheme consistent with the work reported here is shown in Supplementary Materials Fig. 8. In a gradient of H 2 O 2 , preferential activation of TRPM2 ion channels at the leading edge causes an influx of calcium from the extracellular solution, triggering calcium-induced calcium release from intracellular stores. The Src family kinase lyn modulates this pathway at an early stage, perhaps by regulating trafficking of TRPM2 ion channels to the surface membrane. The intracellular signalling molecules PI3K and Rac, which have been proposed as drivers of chemotaxis, do not appear to play a direct role in chemotaxis, but instead fine-tune motility to all chemoattractants by regulating a step common to all, for example refilling of intracellular calcium stores, and thus modulate the amplitude of the leading-edge calcium pulses.
Some studies have found no evidence for calcium gradients in neutrophils during chemotaxis 44,52 , though others have obtained evidence that leading-edge calcium signals play a critical role in chemotaxis of many different cell types 53,54 . In the present work, we observed leading-edge calcium signals in all motile neutrophils in vitro. Interestingly, leading-edge calcium signals have also recently been detected in migrating neutrophils in vivo 55 . These authors find that neutrophils can be guided in vivo by expressing TRPV1 and then imposing a directional gradient of the TRPV1 agonist capsaicin 55,56 . This work supports the idea that a TRP channel alone can mediate neutrophil guidance, but it does not identify the responsible TRP isoform. Here we identify TRPM2 as the ion channel mediating physiological neutrophil guidance towards a source of H 2 O 2 . TRPM2 may be an interesting future drug target, for instance in controlling excess neutrophil invasion in conditions such as sepsis.
Neutrophils are able to navigate up a gradient of H 2 O 2 in which there is an average difference of only a few molecules of intracellular H 2 O 2 between the leading and trailing halves of the cell. This extraordinary sensitivity suggests that navigation towards H 2 O 2 is vital for mammalian survival and has been highly honed by evolution. The strong temperature dependence of TRPM2 activation 15,28-30 suggests a novel function for fever: by enhancing TRPM2 activation, fever potently enhances the sensitivity to a gradient of H 2 O 2 , thereby enhancing neutrophil guidance and promoting the detection and killing of invading pathogens. H 2 O 2 may therefore play a more critical role than conventional cytokines in the early phase of the response of the innate immune system to tissue damage or pathogen attack.

Materials and methods
Animals. Black C57BL/6 WT and TRPM2 −/− K/O mice (6-8 weeks old) were bred in house from matings of TRPM2 +/− heterozygous mice to ensure that the genetic and environmental backgrounds of both WT and TRPM2−/− mice were as far as possible identical. TRPM2 −/− K/O mice were a gift from the Mori laboratory (Kyoto, Japan) 49   In vitro Boyden chamber neutrophil migration assay. QCM Boyden chamber chemotaxis neutrophil migration assay kits were purchased from Merck Millipore (ECM505, Massachusetts, USA). Neutrophil chemotaxis cell migration assays were conducted according to manufacturer's instructions. In brief, neutrophils, isolated and prepared as above were seeded at 2 × 10 5 cells per insert and each insert placed in a chamber containing either DMEM medium alone, 0.001 µM, 0.01 µM, 0.1 µM, 1 µM, 10 µM, 100 µM H 2 O 2 or 1 µM fMLP. Inserts and chambers were incubated at 37 °C in 95% air/5% CO 2 for 4 h to allow cells to migrate through neutrophilspecific 3 µm pores. Non-migrated cells were removed from the inserts and the inserts placed in cell-detachment   Calcium imaging of neutrophils. Neutrophils isolated as above were re-suspended in DMEM + 10% FBS at a concentration of 5 × 10 5 per ml. For experiments in which intracellular calcium was to be recorded while extracellular solutions were changed, neutrophils were plated onto a collagen-coated 13 mm round glass coverslip and incubated at 37 °C in 95%air/5% CO2 for 1 h to allow neutrophils to adhere. Fura2-AM (5 uM in DMEM) was then added to the cells on the coverslip for 30 min at 37 °C in 95% air/5% CO 2 . Fluorescence was measured during alternating illumination at 340 nm and 380 nm (OptoScan; Cairn Research Inc, Kent, UK) every 2 s using a Nikon Eclipse Ti inverted microscope with a 40 × lens and iXon 897 EM-CCD camera controlled by WinFluor 3.2 software. F 340/380 ratios were obtained using FIJI (ImageJ) and converted to calcium concentrations using the equation given by Grynkiewicz et al. with values R max = 2.501, R min = 0.103, both determined experimentally, see Fig. 3G. Full details in Vilar et al. 31 .
For experiments when calcium signals during chemotaxis up a gradient of chemoattractant were to be recorded, 1 µl of Fura-2 AM solution (50 µg Fura-2 AM + 10 µl pluronic F-127 + 10 µl DMSO) was added to 500 µl of neutrophil suspension and incubated for 1 h at 37 °C in 95%air/5% CO 2 . Fura-2 loaded cells in suspension were seeded into ibidi chambers as previously and imaged in a Nikon Ti-E microscope with a 40 × phase contrast lens. Fast-moving neutrophils located in the middle of the central cell migration strip were selected, with typically only one cell imaged per field. Calcium ratio images were obtained with alternating 340 nm and 380 nm epi-illumination supplied by stable LED light sources (Fura-LED, Cairn Research), typically at 5 s intervals. In some experiments acquisition of each pair of epifluorescence images was followed by a transmitted white-light phase-contrast image. All images were filtered by a broad-band 510 nm filter and captured with a Photometrics Prime 95B sCMOS camera. Stage movement, focus and image acquisition were controlled by Nikon NIS Elements software. The ImageJ Fiji RatioPlus plug-in was used to generate F 340/380 ratio images and a rainbow look-up table (LUT) was applied to the ratio images to indicate the level of calcium. Values of the ratio F 340/380 were converted to intracellular free calcium concentration using the equation given by Grynkiewicz et al. 59 . To determine the values of fluorescence ratios at maximal (R max ) and zero (R min ) calcium levels, neutrophils were imaged in the presence of ionomycin (10 μM) either in DMEM or in 0Ca DMEM with added 2 mM EGTA, giving mean values R max = 3.132 and R min = 0.803 (note that the differences between these values and those above arises because chemotaxis experiments were performed on different apparatus). To indicate the location of the maximum calcium level on a phase-contrast image, a 3-colour LUT was applied to an F 340/380 ratio image and the red channel (highest F 340/380 ) was selected and superimposed on the phase-contrast image. The coordinates of the region of highest calcium within the cell were identified in F 340/380 images using the Fiji maximum intensity identifying tool. The coordinates for the geometric centre of the neutrophil were identified using the cell edge and centre localising tool. In some images (for example, Fig. 4A) mean values of F 340/380 in leading and trailing halves was calculated by identifying the cell outline using the cell edge localising tool and dividing the cell by hand into leading and trailing halves in the direction of the imposed chemotactic gradient.