X-ROS Signaling: Rapid Mechano-Chemo Transduction in Heart

See allHide authors and affiliations

Science  09 Sep 2011:
Vol. 333, Issue 6048, pp. 1440-1445
DOI: 10.1126/science.1202768


We report that in heart cells, physiologic stretch rapidly activates reduced-form nicotinamide adenine dinucleotide phosphate (NADPH) oxidase 2 (NOX2) to produce reactive oxygen species (ROS) in a process dependent on microtubules (X-ROS signaling). ROS production occurs in the sarcolemmal and t-tubule membranes where NOX2 is located and sensitizes nearby ryanodine receptors (RyRs) in the sarcoplasmic reticulum (SR). This triggers a burst of Ca2+ sparks, the elementary Ca2+ release events in heart. Although this stretch-dependent “tuning” of RyRs increases Ca2+ signaling sensitivity in healthy cardiomyocytes, in disease it enables Ca2+ sparks to trigger arrhythmogenic Ca2+ waves. In the mouse model of Duchenne muscular dystrophy, hyperactive X-ROS signaling contributes to cardiomyopathy through aberrant Ca2+ release from the SR. X-ROS signaling thus provides a mechanistic explanation for the mechanotransduction of Ca2+ release in the heart and offers fresh therapeutic possibilities.

In the heart, mechanical stretch during diastolic filling activates mechanotransduction signaling pathways that have broad implications for cardiac health and disease (1, 2). A small diastolic stretch (8%) (3) of a ventricular myocyte causes a burst of Ca2+ sparks (4), the elementary events corresponding to the release of free Ca2+ from intracellular stores (Fig. 1, A to C) (5, 6). The mechanism by which this occurs has remained elusive. To address this question, we created tools to improve investigation of single-cell function (Fig. 1A, fig. S2, and movies S1 and S2) (4, 7, 8). We firmly attached single myocytes to stiff glass micro-rods with a biological adhesive (MyoTak) (8). This allows precise control of cell length and the measurement of isometric force, thus permitting us to examine the details of stretch-dependent signal transduction. Such work reveals that stretch-activated Ca2+ sparks are triggered by a mechano-chemo signaling pathway that regulates local production of reactive oxygen species (ROS) in heart cells.

Fig. 1

Stretch-activated Ca2+ release in healthy and dystrophic cardiac myocytes. (A) Single rat ventricular myocyte attached to stiff glass micro-rods coated with MyoTak (8). (B) Fluorescent surface plot of stretch-activated Ca2+ sparks in myocyte subject to 8% axial stretch. (C) (Top) Average positional output from piezo-electric length controller upon 8% diastolic stretch, which imposes a sarcomere length change of 1.84 ± 0.02 to 1.99 ± 0.03 μm. (Bottom) Ca2+ spark histogram (500ms bins) before (black), during (red), and after (blue) stretch (n = 52 rat cells). (D) Frequency of Ca2+ waves before and during stretch, under control conditions (1.8 mM [Ca2+]o, n = 52 rat cells) and under conditions of Ca2+ overload (5 mM [Ca2+]o, n = 10 rat cells). (E) Fluorescent surface plot demonstrating stretch-induced Ca2+ wave in dystrophic (mdx) myocyte. (F) Frequency of Ca2+ waves in mdx myocytes (n = 38 cells) compared with WT myocytes (n = 25 cells) before and during stretch. *P < 0.05 compared with rest value. #P < 0.05 compared with control conditions value (D) or WT value (F); paired t test.

Stretching of myocytes initiated a burst of Ca2+ sparks that was rapid in onset (milliseconds) and large in magnitude (a nearly twofold increase) (Fig. 1C). Although diastolic stretch in normal myocytes produces the spark burst, under pathological conditions the identical stretch generated transient increases in the intracellular concentration of free Ca2+ ([Ca2+]i) that propagate as waves through the cell (Ca2+ waves). Such conditions include Ca2+ overload of the sarcoplasmic reticulum (SR) (Fig. 1D) and pharmacological sensitization of ryanodine receptors (RyR2s) (fig. S3). Stretch-activated sparks are not influenced by stretch-activated channels, Ca2+ influx, Na+ influx, nitric oxide signaling (4), nor acute stretch-induced increases in SR [Ca2+] (9). To explore how stretch activates Ca2+ signals under normal and pathological conditions, we examined cardiomyocytes from the stretch-sensitive muscular disorder, Duchenne muscular dystrophy (DMD). The frequency of Ca2+ waves produced by cellular stretch was increased in cardiomyocytes from the mouse model of DMD, the mdx mouse, compared with wild-type (WT) cells (Fig. 1, E and F). Recent reports that ROS-dependent nitrosylation (10) and oxidation (11, 12) of RyR2s contribute to aberrant Ca2+ release from the SR and arrhythmia in mdx myocytes encouraged us to examine ROS signaling.

ROS and reactive nitrogen species (RNS) react with cysteine residues on RyR2 to rapidly and reversibly modulate RyR2 [Ca2+]i sensitivity (10, 1315). Application of the antioxidant N-acetylcysteine (NAC) to healthy myocytes blocked the stretch-induced burst of Ca2+ sparks (Fig. 2A), suggesting that stretch-dependent oxidation of RyR2s (11, 14, 15) may underlie the increase in spark frequency. To assay whether such a mechanism could account for the rapid changes in Ca2+ spark rate observed (Fig. 1C), we superfused cells with 200 μM H202 and assayed spark activity (fig. S4A). Superfusion of H202 reversibly increased Ca2+ spark rate, with kinetics and magnitude similar to that observed with physiologic stretch (fig. S4, A to C), confirming that an oxidative mechanism can account for rapid and reversible regulation of RyR2 [Ca2+]i sensitivity during stretch.

Fig. 2

ROS generation and microtubule integrity underlie stretch-activated Ca2+ sparks. (A) Stretch-activated Ca2+ sparks in 14 rat cardiomyocytes in normal Tyrode’s (NT), after 5 min local superfusion with 10 mM NAC-NT, and upon washout in NT. (B) (Top) Ca2+ spark histogram in untreated control rat myocytes (n = 11 cells, filled boxes) and in myocytes treated with 3 μM DPI (n = 11 cells, hashed boxes) before (black), during (red), and after (blue) stretch. (Bottom) Quantitation of spark rate in these cells. (C) Average baseline-corrected (fig. S5) DCF fluorescence time course in 36 rat myocytes subjected to 8% stretch (black). The derivative (red trace) of a polynomial fit (green trace, r2 = 0.96) to the DCF data reveals the time course of ROS production. (D) Average DCF fluorescence time course in control (black, n = 19 cells), colchicine-treated (10 μM, blue, n = 9 cells), and DPI-treated (3 μM, red, n = 14 cells) rat myocytes. (E) Quantification of DCF slope before (black), during (red), and after (blue) stretch. Data normalized to pre-stretch slope. *P < 0.05 compared with rest value; paired t test.

Because NAC indiscriminately scavenges all ROS and reactive nitrogen species (RNS), we sought to identify the specific source of RyR2 oxidation. The reduced-form nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX) family of enzymes represents a major source of ROS in the cardiovascular system (16) and has a central role in the pathology of DMD (12, 17) and heart disease (18). Unlike other ROS sources (for example, xanthine oxidases, the mitochondrial electron transport chain, or uncoupled nitric oxide synthases), NOX generates superoxide (O2) in a highly regulated manner, which is ideal for a role in cell-signaling cascades (16). Furthermore, NOX-dependent ROS activates RyR1 in skeletal muscle in vitro (19). Therefore, we tested for a potential role of NOX by applying the inhibitor diphenyleneiodonium (DPI). DPI blocked the stretch-dependent burst of Ca2+ sparks (Fig. 2B), which is consistent with the hypothesis that physiologic stretch triggers rapid production of ROS by NOX.

To test this hypothesis, we measured ROS generation during stretch of ventricular myocytes with the fluorescent ROS sensor 2′,7′-dichlorofluorescein diacetate (DCF). The rate of change (slope) of the DCF fluorescence signal reports the rate of ROS generation within the cell (fig. S5). An 8% stretch produces an immediate increase in DCF slope. Upon return to the original cell length, the DCF slope rapidly returns to the prestretch level (Fig. 2C, black trace). The derivative (Fig. 2C, red trace) of a polynomial fit [Fig. 2C, green trace; correlation coefficient (r)2 = 0.96] to the DCF data reveals that ROS production is increased immediately upon stretch, gradually decreases during the prolonged stretch duration, and returns to its prestretch level upon relaxation (Fig. 2C). This correlates with the time course of stretch-dependent Ca2+ sparks depicted in Fig. 2B and fig. S5C. Treatment of cells with DPI fully blocked the stretch-induced increase in ROS production (Fig. 2, D and E), suggesting that NOX activation underlies stretch-dependent ROS production. However, DPI also inhibits mitochondrial ROS production (20). Mitochondrial oxidative metabolism results in constant ROS production, which is consistent with the mitochondrial-like pattern of DCF fluorescence in a resting cell (fig. S6). We therefore uncoupled mitochondrial metabolism and subsequent ROS production by applying the proton ionophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP). The strong mitochondrial fluorescence pattern with DCF was lost after CCCP application, but the stretch-dependent ROS burst and stretch-dependent Ca2+ sparks were maintained (fig. S6, C to E). Thus, mitochondria appear not to contribute to stretch-induced ROS production and resulting Ca2+ sparks.

Along with NOX, microtubules represent a second component required for the stretch-dependent signaling pathway (4). We confirmed that disrupting microtubule integrity blocks the stretch-activated burst of Ca2+ sparks in both rat and mouse myocytes (fig. S7). If microtubules contribute to the activation of NOX, then disrupting microtubules should also block the stretch-dependent production of ROS. Treatment of cells with colchicine to depolymerize microtubules fully blocked the stretch-induced increase in DCF fluorescence, indicating that mechanotransduction through microtubules is required for stretch-dependent ROS production (Fig. 2, D and E). When stretch-dependent ROS production was inhibited, stretch often produced a slight decrease in the rate of ROS production, suggesting that in the absence of this pathway stretch may uncouple some baseline ROS production (Figs. 2D, 3B, and 4A).

Fig. 3

NOX2 in the t-system produces stretch-dependent ROS that sensitizes nearby RyR2s. (A) Quantification of DCF slope before (black), during (red), and after (blue) stretch in rat myocytes treated with 1 μM gp91ds-tat–scramble peptide (n = 6 cells), 1 μM gp91ds-tat (n = 10 cells), 3 μM gp91ds-tat (n = 9 cells), 20 μM rac1 inhibitor (n = 7 cells), or in the nonstretched region of cells treated with scramble peptide (n = 6 cells). Data was normalized to prestretch slope. (B) DCF fluorescence time course with stretch in WT and NOX2−/− myocytes (n = 6 cells). (C) Quantification of DCF slope before (black), during (red), and after (blue) stretch. (D) (Top) WT myocyte pre-incubated with 1 μM FAM-labeled gp91ds-tat shows sarcolemma, intercalated disc, and Z-line staining. (Bottom) High-magnification image of WT myocyte co-stained with Di-D (t-tubules) and FAM-labeled gp91ds-tat. Fluorescence plot-profile from region enclosed by dotted lines shows overlay of t-tubules and NOX2 inhibitory peptide. (E) (Top) NOX2−/− myocyte pre-incubated with 1 μM FAM-labeled gp91ds-tat and imaged with identical settings as (D) shows only diffuse staining. (Bottom) High-magnification image of NOX2−/− myocyte co-stained with Di-D and FAM-labeled gp91ds-tat shows loss of t-tubule staining pattern for gp91ds. (F and G) Line-scan image of Fluo-4 signal from electrically stimulated rat ventricular myocyte in the presence of 2.5 μM nifedipine at resting length (F) and upon 8% diastolic stretch (G). Red mark denotes electrical stimulation (fig. S9, protocol details). White bars mark active calcium release units (CRUs) rapidly triggered by depolarization. (H) Quantification of the percentage of active CRUs triggered immediately (within 10 ms) upon depolarization in cells held at resting length (black) or upon stretch (red) in control myocytes (n = 11 cells) or those treated with 1 μM gp91ds-tat (n = 6 cells). *P < 0.05 compared with rest values; paired t test.

Fig. 4

Stretch-induced ROS generation and aberrant Ca2+ release in dystrophic myocytes. (A) Average DCF fluorescence time course in WT myocytes treated with 1 μM gp91ds-tat–scramble (black, n = 10 cells) or 1 μM gp91ds-tat (red, n = 9 cells) and mdx myocytes treated with 1 μM gp91ds-tat–scramble (blue, n = 9 cells) or 1 μM gp91ds-tat (green, n = 9 cells). (B) Quantification of DCF slope before (black), during (red), and after (blue) stretch. Data normalized to prestretch slope. (C) Ca2+ wave frequency before (black), during (red), and after (blue) stretch. (D) Percentage of cells demonstrating Ca2+ oscillations during stretch protocols. (E) Fluorescence surface plot of mdx myocyte treated with 1 μM gp91ds-tat–scramble demonstrating stretch-induced Ca2+ oscillations. *P < 0.05 compared with rest value; paired t test. #P < 0.05 compared with WT value; paired t test.

NOX2 and NOX4 are the most abundantly expressed isoforms of NOX in adult cardiomyocytes (21). Membrane-bound NOX2 (also known as gp91phox) requires association with cytosolic regulatory subunits, including p47phox, to produce O2, whereas NOX4 does not (22). Therefore, to examine the role of NOX2, we used a specific peptide inhibitor (gp91ds-tat) that prevents the interaction of gp91phox with p47phox (23). Inhibition of NOX2 by gp91ds-tat blocked stretch-dependent ROS production in a dose-dependent manner, and a control, scrambled peptide had no effect (Fig. 3A). To further confirm NOX2 as a molecular source of stretch-induced ROS production, we measured DCF signals in myocytes from NOX2−/− mice. Myocytes lacking NOX2 expression did not show stretch-dependent production of ROS (Fig. 3, B and C). Recruitment of the small guanosine triphosphatases rac1 or rac2 to the NOX2 complex is also required for enzyme activation (16, 22). Specific inhibition of rac1, which binds microtubules (24), blocked stretch-induced ROS production (Fig. 3A).

For NOX2 to produce a ROS signal that rapidly influences Ca2+ release, we expected it to be located very close to the junctional SR (jSR) that contains the RyR2 clusters. To test this, we used the fluorescently labeled 6-carboxyfluorescein (FAM)–gp91ds-tat. FAM-gp91ds labeled the sarcolemma and intercalated disk structures at the periphery of myocytes and colocalized with t-tubule structures that penetrate the interior of the myocyte at the Z-line, which is in agreement with previous findings (Fig. 3D and fig. S8A) (25). In contrast, FAM-gp91ds showed no such Z-line localization in NOX2−/− myocytes, confirming the specificity of this signal (Fig. 3E and fig. S8B). Co-localization analysis indicated that approximately 70% of the FAM-gp91ds signal from the interior of myocytes colocalized with the t-tubule marker Di-D (fig. S8D). NOX2 regulatory subunits also must be readily available to promote rapid signaling, which is supported by the preferential expression of heteromultimeric subunit complexes in t-tubule fractions (fig. S1) (19, 24).

If the ROS are produced locally at t-tubule structures and act rapidly, the measured increase in ROS should only occur in the stretched region of the cell because ROS are short-lived species typically confined to action within their immediate vicinity. ROS were measured with confocal line scans through the interior of myocytes (8 to 12 μm from the sarcolemma), in both the stretched (Fig. 3D, yellow dotted line) and nonstretched (within 15 μm of intercalated disc structures) (Fig. 3D, blue dotted line) regions of cells. No increase in ROS production was detected in the nonstretched region of the cell (Fig. 3A). Consistently, Ca2+ spark rate is unchanged in the nonstretched portion of a stretched cardiomyocyte (4). These findings suggest that the ROS responsible for oxidation of RyR2 are produced locally and probably do not arise from sarcolemma or intercalated disc sources. Taken together, the above implicates NOX2 as a source of rapid stretch-dependent ROS production at the jSR of cardiomyocytes (X-ROS signaling).

To assay whether X-ROS signaling regulates intracellular Ca2+ release upon electrical stimulation, we modified a protocol used by Santana et al. (1996) to evaluate the activity of individual Ca2+ release units (CRUs) during depolarization of a myocyte (26). A CRU consists of a cluster of RyR2s in the jSR that produce a Ca2+ spark, which summate during electrical stimulation to produce the global increase in [Ca2+]i (5). A low concentration of nifedipine blocks a large portion of the Ca2+ influx that enters the cell through voltage-gated L-type Ca2+ channels during depolarization. This approach reduces the number of CRUs triggered by Ca2+-induced Ca2+ release, thus enabling the visualization and quantification of fluorescence signals arising from individual CRUs in electrically stimulated myocytes (fig. S9, protocol details) (26). Using this protocol, field stimulation of a myocyte held at resting length triggered an immediate (<10 ms) spike in fluorescence in approximately 50% of the CRUs in a given line scan (Fig. 3F). Sites that demonstrated rapidly triggered Ca2+ release were deemed active CRUs (Fig. 3, F and G, white bars). The remaining 50% of sites showed a delayed increase in fluorescence that probably arose from the diffusion of Ca2+ from CRUs outside of the confocal image plane; these were deemed inactive CRUs (fig. S9, D and E). After physiologic stretch, the percentage of active CRUs rapidly triggered by field stimulation was significantly increased (Fig. 3, G and H). This is consistent with a stretch-dependent increase in RyR2 [Ca2+]i sensitivity enabling more reliable triggering of CRUs during electrical stimulation. Inhibition of NOX2 by gp91ds blocked the stretch-dependent increase in CRU triggering (Fig. 3H).

We next measured stretch-dependent ROS production and [Ca2+]i regulation in the mdx mouse. Mdx myocytes demonstrated significantly increased ROS production with stretch when compared with WT controls (Fig. 4, A and B). This effect was blocked by the NOX2 peptide inhibitor. Consistent with the large response in mdx myocytes, X-ROS signaling components are up-regulated in the dystrophic heart. Specifically, NOX2 expression and activity are increased (11), and we observed increased expression of all tubulin subunits, with a concomitant increase in microtubule network density (fig. S10).

In mdx heart cells, the occurrence of Ca2+ waves was increased at rest and further increased after stretch (Fig. 4C). Acute inhibition of NOX2 with gp91ds-tat blocked the stretch-induced waves (Fig. 4C). NOX2 inhibition did not affect the increased Ca2+ wave activity at rest (Fig. 4C), which is consistent with other chronic pathways stably promoting wave generation in the basal state (10, 27). A more dramatic form of aberrant Ca2+ regulation was also observed in mdx myocytes, in which stretch triggered Ca2+ oscillations. This behavior ceased soon after the stretch was released, which is consistent with aberrant reversible signaling and not mechanical damage (Fig. 4E). Inhibition of NOX2 reduced this behavior (Fig. 4D). Very large mechanical stresses of mdx myocytes, whether applied as osmotic shock (17) or large strain stretch (20%) (7), can lead to massive Ca2+ influx through membrane micro-tears, triggering rapid cell death. We used physiologic 8% stretch (3) and never observed myocyte death. Stretch that is physiologic in WT myocytes thus appears to be a pathological trigger in mdx heart cells through NOX2 ROS production.

ROS can drive the downstream production of other reactive oxygen or nitrogen species that also contribute to cardiac Ca2+ signaling and pathology. In DMD, chronic RyR2 S-nitrosylation is linked to “leaky” RyR2s and Ca2+-dependent arrhythmia (10). O2 (the ROS produced by NOX) is an upstream prerequisite for the S-nitrosylation of RyR2. ROS also function as an upstream regulator of various transcriptional and signaling defects underlying DMD cardiomyopathy (11, 12, 17). Enhanced X-ROS signaling with each diastolic stretch in mdx myocytes (Fig. 4A) would promote a chronic oxidative environment that likely contributes to disease progression.

The mechanical distortion produced by physiologic stretch regulates NOX2-dependent ROS production through an intact microtubule network (X-ROS signaling). ROS produced by stretch is strategically localized to the jSR to permit rapid redox modification of RyR2 and regulation of cardiac Ca2+ signaling. That ROS can dynamically tune RyR2 sensitivity in a physiological context reveals a signaling pathway that when dysregulated contributes to the defective [Ca2+]i handling inherent to cardiomyopathy in DMD. Given the predominant role of oxidative stress (14, 18) and cytoskeletal dysfunction (28) in the progression of cardiomyopathy, X-ROS signaling may have important pathophysiological effects in the context of heart disease.

Supporting Online Material

Materials and Methods

Figs. S1 to S11

References (29–31)

Movies S1 and S2

References and Notes

  1. Materials and methods are available as supporting material on Science Online
  2. Acknowledgments: We thank S. Martin for antibodies to tubulin subunits and for discussion, G. Shi for technical assistance, and M. Williams for NOX2−/− animals. B.L.P is supported by an NIH training grant (T32 HL072751-07) to the Training Program in Cardiovascular Cell Biology. This work was supported by NIH grants R01 HL106059, P01 HL67849, R01 HL36974, RC2 NR011968, and S10 RR023028; Leducq North American-European Atrial Fibrillation Research Alliance; European Union Seventh Framework Program (FP7), Georg August University, “Identification and therapeutic targeting of common arrhythmia trigger mechanisms”; and the Maryland Stem Cell Research Fund. World Precision Instruments (WPI, Sarasota, FL, USA and SI-Heidelberg), Gabe Sinclair (Fourhourday, Baltimore, MD, USA), and Siskiyou (Grants Pass, OR, USA) supported instrumentation and equipment development.
View Abstract

Stay Connected to Science

Navigate This Article