Search tips
Search criteria 


Logo of aacPermissionsJournals.ASM.orgJournalAAC ArticleJournal InfoAuthorsReviewers
Antimicrob Agents Chemother. 2005 January; 49(1): 388–397.
PMCID: PMC538851

Hemolysis of Erythrocytes by Granulysin-Derived Peptides but Not by Granulysin


Granulysin, a 9-kDa protein localized in human cytolytic T lymphoctyes and natural killer cell granules, is cytolytic against tumors and microbes but not against red blood cells. Synthetic peptides corresponding to the central region of granulysin recapitulate the lytic activity of the intact molecule, and some peptides cause hemolysis of red blood cells. Peptides in which cysteine residues were replaced by serine maintain their activity against microbes but lose activity against human cells, suggesting their potential as antibiotics. Studies were undertaken to determine the mechanism of resistance of red blood cells to granulysin and sensitivity to a subset of granulysin-derived peptides. Granulysin lyses immature reticulocytes, which have mitochondria, but not red blood cells. Granulysin lyses U937 cells but not U937 cells lacking mitochondrial DNA and a functional respiratory chain (U937ρ° cells), further demonstrating the requirement of intact mitochondria for granulysin-mediated death. Peptide G8, which corresponds to helix 2/loop 2/helix 3, lyses red blood cells, while peptide G9, which is identical except that the cysteine residues were replaced by serine, does not lyse red blood cells. Granulysin peptide-induced hemolysis is markedly inhibited by an anion transporter inhibitor and by Na+, K+, and Ca2+ channel blockers but not by Na+/K+ pump, cotransport, or Cl channel blockers. Although recombinant granulysin and G9 peptide do not induce hemolysis, they both competitively inhibit G8-induced hemolysis. The finding that some derivatives of granulysin are hemolytic may have important implications for the design of granulysin-based antimicrobial therapeutics.

Granulysin, a lytic molecule expressed by human cytolytic T lymphocytes and natural killer cells, is colocalized in granules with perforin and granzymes and is active against tumor cells and a variety of microbes, including Mycobacterium tuberculosis (12, 18, 34, 37). Recombinant 9-kDa granulysin disrupts artificial liposomes and cell membranes, damages mitochondria, and activates caspase-9 to induce apoptosis in nucleated cells but does not lyse red blood cells (20).

The 9-kDa granulysin contains two disulfide bonds, and reduction of recombinant granulysin enhances its lytic activity against tumor targets but does not affect its activity against bacteria (39). Peptides corresponding to the central region of granulysin lyse bacteria, human cells, and synthetic liposomes, while peptides corresponding to the amino or carboxyl regions are not lytic (39). Most peptides corresponding to either helix 2 or helix 3 lyse bacteria, while lysis of human cells and liposomes is dependent on the helix 3 sequence. Peptides in which positively charged arginine residues were replaced by neutral glutamine exhibit reduced lysis of all three targets (39). Granulysin and its constituent peptides bind to the cell surface based on charge and cause an increase in the intracellular Ca2+ concentration and a subsequent decrease in the intracellular K+ concentration (20, 31). Both calcium and potassium channel blockers inhibit granulysin/peptides-induced apoptosis in tumor cells (31).

Studies were undertaken to determine the mechanism of resistance of red blood cells to granulysin along with sensitivity to a subset of granulysin derived peptides. We report here that some granulysin-derived peptides, but not granulysin, lyse red blood cells. Target sensitivity to granulysin depends on the presence of mitochondria. In contrast, some peptides lyse red blood cells even in the absence of mitochondria. The roles of Ca2+, K+, Na+, and Cl in granulysin peptide-mediated hemolysis are characterized. These findings suggest possible toxicities of synthetic peptides considered as potential therapeutics.



Reagents used include the calcium channel blockers econazole and nickel sulfate (Sigma, St. Louis, Mo.); the potassium channel blockers tetraethylammonium chloride (TEA) (Calbiochem, La Jolla, Calif.), apamin (Sigma), and BaCl2 (Sigma); the sodium channel blockers QX222 (2-[(2,6-dimethylphenyl)amino]-N,N,N-trimethyl-2-oxoethaniminium chloride; Tocris, Ellisville, Mo.) and amiloride hydrochloride [3,5-diamino-N-(aminoiminomethyl)-6-chloropyrazinecarboxamide hydrochloride; Calbiochem]; the Cl channel blockers 9-anthracenecarboxylic acid (9-AC; Tocris) and NPPB [5-nitro-2-(3-phenylpropylamino)benzoic acid; Tocris]; the anion transport inhibitor 4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid (DIDS; Sigma), the [Na+/K+] pump inhibitor ouabain (Sigma); the [Na+/K+/2Cl] cotransport inhibitor bumetanide (Sigma); and the [K+/Cl] cotransport inhibitor R(+)-[2-n-butyl-6,7-dichloro-2-cyclophenyl-2,3-dihydro-1-oxo-1H-inden-5-il,oxy] acetic acid (DIAO). Dyes used for measuring intracellular Ca2+, Na+, Cl, and K+ concentrations were fluo-3/AM, sodium green, MQAE [N-(ethoxycarbonylmethyl)-6-methoxyquinolinium bromide], and K+-binding benzofuran isophtalate (PBFI) (all from Molecular Probes), respectively. Reticulocytes were stained with brilliant Cresyl Blue (Merck, Darmstadt, Germany).


The human monoblastic cell line (U937) was obtained from the American Type Culture Collection (Manassas, Va.).

Granulysin and granulysin-derived peptides.

The 9-kDa recombinant granulysin was expressed and purified as previously described (18). Granulysin peptides were synthesized as previously described (39) (Table (Table1).1). Peptides G8 and G9 include residues 23 to 51 (helix 2/loop 2/helix 3) of granulysin and are identical, except that cysteine was replaced by serine at positions 34 and 45 in G9 (39).

Peptide sequences used in this studya

Measurement of granulysin and granulysin peptide-induced hemolysis.

Peripheral blood from healthy donors was obtained with informed consent and collected in tubes containing heparin. Red blood cells were separated by centrifugation at 1,000 × g for 5 min and washed three times with RPMI 1640 without phenol red. The washed red blood cells were suspended in RPMI 1640 or HEPES buffer (5 mM HEPES, 150 mM NaCl [pH 7.4]). Red blood cells (2 × 106/ml in RPMI 1640) were treated with granulysin or granulysin peptides at the indicated concentrations at 37°C for 4 h. In some cases, granulysin was boiled for 10 min before the assay (boiled granulysin). Hemolysis was determined by measuring the release of lactate dehydrogenase (LDH) by using a kit (Roche).

Treatment of peptides with DTT.

Peptides G8 and G9 were incubated with 1 mM dithiothreitol (DTT) at room temperature for 10 min prior to addition to red blood cells. Because DTT affects the LDH assay, hemolysis was determined by measuring the supernatant optical density at 540 nm (OD540). DTT itself did not affect the OD540.

Generation of rabbit reticulocytes and hemolysis assay.

Japanese White rabbits, weighing 1.9 to 2.0 kg, were made anemic by subcutaneous injection of a 2.5% phenylhydrazine-HCI (Sigma) solution as follows: 1.0 ml on day 1, 0.8 ml on day 2, 0.6 ml on day 3, 0.8 ml on day 4, and 1 ml on day 5 (19). Control rabbits were injected with the same volume of sterile distilled water. On day 6 or 7, 5 ml of blood was collected from the ear into heparinized tubes. Cells were stained with Brilliant Cresyl Blue (36) to enumerate reticulocytes and red blood cells. Red blood cells and reticulocytes were isolated by centrifugation at 1,000 × g for 5 min and washed three times with RPMI 1640 without phenol red. Rabbit red blood cells and reticulocytes (2 × 106/ml in RPMI 1640) were treated with granulysin or granulysin peptides at 37°C for 4 h, and LDH release was measured.

Generation of ρ° cells.

U937 cells were cultured in RPMI 1640 supplemented with 50 ng of ethidium bromide per ml, 10% fetal bovine serum, 2 mM l-glutamine, 200 μg of penicillin per ml, 100 μg of streptomycin sulfate per ml, 100 μg of pyruvate per ml, and 50 μg of uridine per ml (29). After 3 months, these cells, designated U937ρ°, were maintained in medium without ethidium bromide. Exposure of U937 cells to antimycin A (50 to 100 μM) induced cell death, while the viability of U937ρ° cells was unaffected by antimycin A, demonstrating that the mitochondrial respiratory activity of U937ρ° cells was significantly reduced.

AnnexinV/PI staining.

U937 or U937ρ° cells (106/ml) were treated with granulysin or granulysin peptides for 3 h and washed twice with phosphate-buffered saline (PBS). Cell pellets were suspended in 150 μl of binding buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl2 [pH 7.4]) and stained with 5 μl of annexin V-fluorescein isothiocyanate (FITC) in the dark at room temperature for 15 min. Binding buffer (250 μl) and 5 μl of propidium iodide (250 μg/ml) were added prior to fluorescence-activated cell sorter analysis (BD Biosciences, San Jose, Calif.).

Ion channel blockers or transport inhibitors.

Red blood cells suspended in RPMI 1640 were preincubated with calcium channel blockers (econazole or nickel sulfate), potassium channel blockers, (TEA, BaCl2, or apamin), sodium channel blockers (QX222 or amiloride hydrochloride), anion transport inhibitor (DIDS), [Na+/K+] pump inhibitor (ouabain), [Na+/K+/2Cl] cotransport inhibitor (bumetanide), [K+/Cl] cotransport inhibitor (DIOA), or Cl channel blockers (9-AC or NPPB) for 30 min at 37°C. G8 or G9 (5 μM) was then added, and the incubation was continued for another 4 h at 37°C. The 50% inhibitory dose (ID50) was calculated as the concentration of inhibitor that reduced hemolysis by 50%.

Intracellular Na+, Ca2+, and Cl concentrations.

Intracellular Na+ and Ca2+ concentrations were measured using sodium green (27) or Fluo-3 AM (27, 35), respectively, as described previously. Briefly, red blood cells (2 × 106/ml in RPMI 1640) were loaded at 2 μM dye at room temperature for 60 min. The cells were then washed twice with RPMI 1640; G8 or G9 (5 μM) was added, and the cells were analyzed by flow cytometry using a Becton Dickinson FACScan (excitation at 488 nm and emission detected at 530 nm). Data were analyzed using CellQuest software. The intracellular Cl concentration was measured using MQAE (2). Red blood cells (2 × 106/ml in RPMI 1640) were loaded at 5 mM at 37°C for 2 h and washed three times with RPMI 1640. After addition of G8 or G9 (5 μM), fluorescence was monitored using a Becton Dickinson FACSDesk (excitation at 350 nm and emission detected at 450 nm). Data were analyzed using FlowJo software.

K+ efflux.

K+ efflux was measured by atomic absorption spectrometry (model AA-6800 instrument; Shimadzu, Kyoto, Japan) as described previously (6, 14, 16). In brief, red blood cells were suspended at 1% in 5 mM HEPES-0.15 M NaCl buffer without K+. They were then treated with various concentrations of G8 or G9 for 0 to 10 min at room temperature. At the conclusion of the incubation, the cells were centrifuged immediately and the supernatant was assayed for K+. Hemolysis was determined by measuring the OD540. Control cells (no peptide) were lysed to determine the total K+ and hemoglobin levels; G8-induced K+ efflux was calculated as % K+ efflux = [(K+ efflux of G8-treated sample −K+ efflux of control sample)/total K+ after cell lysis] × 100. To assess the effect of DIDS on G8-induced K+ release, red blood cells were suspended at 1% in 5 mM HEPES-0.15 M NaCl buffer without K+ and preincubated with 25 μM DIDS for 30 min at 37°C. They were then treated with 20 μM G8 for 0 to 10 min at room temperature. At the conclusion of the incubation, the cells were centrifuged immediately and the K+ concentration in supernatant was assayed by atomic absorption spectrometry.

Osmotic protection.

Red blood cells (2 × 106/ml) were suspended in 0.135 M NaCl-5 mM HEPES-NaOH (pH 7.4) supplemented with 30 mM d-mannitol, sucrose, raffinose, or polyethylene glycol (molecular weight 600, 1,000 or 2,000). G8 was added, and hemolysis was determined after incubation for 4 h at 37°C. Molecular diameters were from Katsu et al. (21).

Size of red blood cells.

The size of the red blood cells after treatment with G8 or G9 was determined by flow cytometry. Forward scatter, a measurement of red blood cell size, was calculated using FlowJo software.

Binding of G8 and granulysin to red blood cells.

Binding of G8 and granulysin to red blood cells was analyzed by confocal microscopy, as previously described (20). Briefly, red blood cells were placed on an inverted coverslip chamber (pretreated with l-polylysine), incubated with G8 or granulysin for 2 to 4 h, washed with PBS, and fixed in 4% paraformaldehyde in PBS for 1 h at room temperature. After being washed, the cells were incubated for 1 h in permeabilize/block solution (5% goat serum, 5% human AB serum, 0.1% NP-40 solution, 0.01% saponin, and 1% milk in PBS) at room temperature, washed three times with PBS, and incubated with rabbit anti-human granulysin polyclonal antibody followed by FITC-anti-rabbit immunoglobulin G. Finally, the chamber was washed with 0.1% saponin-5% milk in PBS, mounted onto a glass slide using a drop of Mowiol (Calbiochem, Madrid, Spain), and stored at 4°C in the dark until observation. Preparations were observed in a Zeiss 310 confocal microscope and analyzed using LSM 3.95 software.


Granulysin exhibits enhanced hemolytic activity after being boiled.

We previously reported that denatured and/or reduced recombinant granulysin exhibits enhanced lytic activity against Jurkat, a human T-cell tumor line, and liposomes (39). We asked here whether denatured granulysin also exhibits enhanced hemolytic activity. As shown in Fig. Fig.1A,1A, boiled granulysin is more potent than refolded granulysin at lysing red blood cells, consistent with its activity against Jurkat cells (39). These findings indicate that the hemolytic activity of recombinant granulysin does not depend on its three-dimensional structure and that short linear regions of the granulysin sequence might mediate hemolysis. Therefore, we assessed the hemolytic activity of synthetic peptides corresponding to linear sequences of granulysin (39).

FIG. 1.
Granulysin and a subset of granulysin-derived peptides induce hemolysis. Human red blood cells were treated with granulysin (GRA) or boiled granulysin (GRA-B) (A), granulysin-derived peptides (B), or DTT-treated G8 or G9 (C) at 37°C for 4 h, and ...

A subset of granulysin peptides causes hemolysis of red blood cells.

Peptides G8 and G20, corresponding to helix 2/loop 2/helix 3 and loop 2/helix 3/helix 4, respectively, exhibit strong lysis of red blood cells (Fig. (Fig.1B).1B). Peptide G14 (loop 2/helix 3) showed intermediate hemolysis. Peptide G15, which is identical to G14 but contains two R-to-Q substitutions, exhibited similar lysis of red blood cells to that of G14, suggesting that arginine residues in loop 2 do not contribute to hemolytic activity. Peptide G13 (helix 2/loop 2) caused a low level of hemolysis. Peptides G1 (helix 1), G10 (helix 2), G11 (helix 3), and G21 (helix 4/loop 3/helix 5) did not lyse red blood cells. Peptide G9, identical to G8 but with two C-to-S substitutions, did not lyse red blood cells, indicating a critical role for cysteine residues in the activity of peptide G8. Thus, lysis of red blood cells is confined to peptides derived from the central region of granulysin and is similar to their ability to induce the apoptosis of nucleated cells such as Jurkat (39).

Previously we reported that reduced granulysin peptides lost their ability to lyse the human T-cell tumor line Jurkat (39). Therefore, we asked here if the reduction of peptides also affects their hemolytic activity. As shown in Fig. Fig.1C,1C, DTT-treated G8 lost most of its hemolytic activity while DTT treatment of G9 had no effect, indicating that intra- or intermolecular disulfide bonds are important for G8-mediated hemolysis.

Granulysin, but not peptide G8, requires intact mitochondria to induce target cell lysis.

We hypothesized that red blood cell resistance to granulysin-mediated cell death may be due to the absence of mitochondria. We previously showed that mitochondria were important targets of granulysin in tumor cell lysis (20). Reticulocytes, the precursors of mature erythrocytes, do contain mitochondria, allowing us to compare their sensitivity with that of red blood cells to granulysin-mediated lysis. Reticulocytes and erythroblasts were isolated from rabbits recovering from phenylhydrazine-HCl-induced anemia. Cresyl Blue staining indicates that nearly all of the red blood cells from these animals were nucleated erythroblasts or reticulocytes (Fig. (Fig.2B),2B), whereas only about 1 to 2% of the red blood cells from the control rabbits (Fig. (Fig.2A)2A) were reticulocytes. Reticulocytes and erythroblasts, but not red blood cells, were lysed by granulysin (Fig. (Fig.2C).2C). Although peptide G8 lysed both cell types, reticulocytes and erythroblasts were more sensitive than red blood cells (Fig. (Fig.2D),2D), suggesting that mitochondria contribute to target sensitivity. To confirm this, U937 cells were incubated with ethidium bromide to generate U937ρ° cells, deficient in mitochondrial DNA and respiration (29). U937 cells undergo apoptosis when cultured with granulysin, but U937ρ° cells do not. Interestingly, U937ρ° cells were sensitive to peptide G8, although the extent of apoptosis was less than in U937 cells (Fig. (Fig.2E2E).

FIG. 2.
Requirement of mitochondria for granulysin-induced hemolysis. (A and B) Cresyl blue staining of blood cells isolated from a control rabbit (A) and a rabbit recovering from anemia (B) (magnification, ×372). (C and D) Rabbit red blood cells (RBC) ...

Effects of K+, Ca2+, Na+, and Cl channel blockers on G8-induced hemolysis.

We reported that an increase in intracellular Ca2+ levels and a decrease in intracellular K+ levels were critical for granulysin and its derivative peptides to induce apoptosis in Jurkat cells (20, 31). Thus, we explored the effect of ion channel blockers on G8-induced hemolysis. In red blood cells, the predominant intracellular cations are K+ and Na+ while the predominant intracellular anions are Cl and HCO3. Three K+ channel blockers (TEA, BaCl2, and apamin) inhibited G8 lysis of red blood cells (Fig. 3A to C). The Cl/HCO3 anion transport inhibitor, DIDS, blocked hemolysis (Fig. (Fig.3D),3D), but two Cl channel blockers (NPPB and 9-AC) had no effect (data not shown). Na+ channel blockers (amiloride and QX222) (Fig. 3E and F) and Ca2+ channel blockers (econazole and Ni2+) (Fig. 3G and H) showed modest inhibition of G8-induced hemolysis. No effect on G8-induced hemolysis was observed in cells treated with ouabain, a Na+/K+ pump inhibitor; bumetanide, a Na+/K+/2Cl cotransporter inhibitor; or DIOA, a K+/Cl cotranspot inhibitor (data not shown). Thus, K+ channels and Cl/HCO3 anion transport appear to be involved in G8-induced hemolysis.

FIG. 3.
K+, Na+, and Ca2+ channel blockers, as well as an anion transport, inhibitor inhibit G8-induced hemolysis. Red blood cells were pretreated for 30 min with the indicated compounds. G8 (•) or G9 (○) (5 μM) ...

G8 increases intracellular Na+, Ca2+, and Cl concentrations and induces a K+ efflux.

We next measured the levels of intracellular ions after treatment of red blood cells with G8. G8 increased intracellular Na+ (Fig. (Fig.4A),4A), Ca2+ (Fig. (Fig.4B)4B) and Cl (Fig. (Fig.5A)5A) concentrations but decreased the intracellular K+ concentration (Fig. 4C and D). Peptide G9 did not affect the levels of these intracellular ions. The K+ efflux precedes hemolysis (Fig. (Fig.4D).4D). We also found that the Cl/HCO3 anion transport inhibitor DIDS significantly inhibits G8-induced K+ efflux (Fig. (Fig.4E4E).

FIG. 4.
G8 causes an increase in intracellular Na+ and Ca2+ concentrations and a decrease in K+ concentration. Intracellular Na+ (A) and Ca2+ (B), K+ efflux (C and D), hemolysis (D), and K+ efflux in the ...
FIG. 5.
G8 increases the intracellular Cl concentration and red blood cell size. Intracellular Cl concentration (A) and red blood cell size (B) after treatment with G8 or G9 (5 μM) are shown.

Effect of colloid-osmotic protectants on G8-induced hemolysis.

Disruption of the integrity of the erythrocyte lipid bilayer by a membrane attack complex such as complement creates an osmotic gradient because small ions can traverse the damaged membrane but large cytoplasmic components such as hemoglobin cannot. The consequent inflow of water causes the cell to expand rapidly. If membrane damage is sufficient, the cell ruptures, releasing hemoglobin. This process is called colloid osmotic lysis. If this colloid-osmotic process is involved in G8-induced hemolysis, the hemolysis will be reduced when a solute of an appropriate size is added to the extracellular aqueous solution (21). This is based on the observation that, if the osmotic pressure of intracellular hemoglobin is balanced with that of the solute in the outer solution, hemolysis is not induced. Thus, the size of the membrane lesion can be determined by examining whether the substances added can protect the membrane from hemolysis. However, no protection from G8-induced hemolysis was observed in media supplemented with mannitol, sucrose, raffinose, or polyethylene glycol (molecular weight 600, 1,000, or 2,000) (not shown), indicating that the colloid-osmotic process is not involved in G8-induced hemolysis.

G8 causes red blood cell swelling.

Because G8 increases intracellular Na+, Ca2+ and Cl concentrations, we speculated that water would enter the red blood cells due to the increased osmotic gradient, resulting in red blood cell swelling. To investigate this, we measured the sizes of red blood cells after treatment with G8. We found that red blood cells treated with G8, but not with G9, were larger than control red blood cells (Fig. (Fig.5B).5B). Pretreatment of red blood cells with the K+ channel blocker TEA prevented G8-induced swelling of RBC, while red blood cells treated with TEA alone showed a slight decrease in size (data not shown).

G8 and granulysin bind to red blood cells.

Because G8 and granulysin affect membrane channels of red blood cells, we investigated whether G8 and granulysin bind to red blood cells. We used a polyclonal rabbit anti-granulysin antibody (18) to detect granulysin. This antiserum binds to immobilized granulysin and G8 but does not bind to G9. As shown in Fig. Fig.6,6, fluorescence was detected in red blood cells treated with G8 and granulysin followed by anti-granulysin antiserum.

FIG. 6.
G8 and granulysin bind to red blood cells. Immunofluorescence of red blood cells treated with medium, granulysin, or G8 and anti-granulysin antiserum is shown.

Granulysin and G9 block G8-induced hemolysis of red blood cells.

Although neither G9 nor granulysin induce hemolysis, red blood cells pretreated with G9 or granulysin were rendered less susceptible to subsequent treatment with G8 (Fig. (Fig.7),7), suggesting that G9 and granulysin compete with G8 for binding to the membranes of red blood cells.

FIG. 7.
Granulysin and G9 protect against G8-induced hemolysis. Red blood cells were pretreated for 30 min with granulysin (GRA) or G9. G8 (5 μM) was then added, and hemolysis was measured after 4 h.


Granulysin is an effector molecule present in human cytolytic T lymphocytes and natural killer cells (18, 33, 34). Both granulysin and a subset of its derivative peptides kill bacteria (39), disrupt synthetic liposomes (20, 39), and induce apoptosis of mammalian nucleated cells (18). This work was undertaken for two reasons: to explore the mechanism of resistance of red blood cells to granulysin-mediated lysis and to determine whether granulysin-based peptides (39) exhibit hemolytic activity that could limit their utility as therapeutic agents. Currently we are generating a large panel of granulysin peptide derivatives and evaluating them as candidate antibiotics. Understanding the events involved in granulysin peptide-mediated hemolysis may help in this process. We previously showed that granulysin peptides induce cell death by apoptosis (20, 31, 39). However, there is no nucleus in red blood cells, and therefore the mechanism of G8-induced hemolysis is not apoptosis and is different from that of G8-induced cell death in nucleated cells.

In this report, we show that granulysin kills only target cells that contain intact mitochondria and that this is a major factor underlying the resistance of red blood cells to granulysin. Nevertheless, some constituent peptides are hemolytic. Peptides from the central region of granulysin, corresponding to either helix 2 or helix 3 and loop 2, induced hemolysis. Replacement of cysteine by serine (G8 versus G9) causes a loss of hemolytic activity, indicating that cysteine residues are important for the hemolytic activity, consistent with the lytic activity on nucleated tumor cells (39). Peptide G8 is more potent than granulysin, and boiled granulysin is more potent than recombinant granulysin, also consistent with the lytic activity on nucleated tumor cells (39). To explore the mechanism of G8-induced hemolysis, we investigated the effect of G8 on ion channels and transporters on red blood cell membranes (7) by using specific channel blockers and transport inhibitors.

Chlorodinitrobenzene (41), gramicindin (21, 32), and ferriprotoporphyrin (8) induce hemolysis following efflux of K+ from red blood cells. Similarly, G8 induces an immediate K+ efflux preceding hemolysis, suggesting that G8 activates/opens K+ channels. K+ channels transport K+ from inside to outside the cells, and more than 30 different types of K+ channels have been identified in mammalian cells. The following K+ channels have been found in red blood cells: (i) voltage-gated channels (Kv) (7), (ii) Ca2+-activated channels with intermediate and small conductance (IKCa and SKCa, respectively) (9, 23), (iii) inwardly rectifying channels (Kir) (23, 38), and (iv) ATP-sensitive Kir (KATP) (24). To examine the contribution of K+ channels, we investigated the effect of K+ channels blockers, TEA, BaCl2, and apamin, on G8-induced hemolysis. While none of these compounds affected baseline hemolysis, they all inhibited G8-induced hemolysis in a dose-dependent manner, with TEA and BaCl2 inhibiting hemolysis by 90 to 100%. Both Kir and KATP channels are inhibited by Ba2+, with ID50s of 0.002 and 0.1 mM, respectively (30), and BaCl2 inhibited crude-venom-induced hemolysis (28). Kv, KCa, and KATP are also extremely sensitive to externally applied TEA (30), and apamin selectively blocks SKCa (4). Taken together, the data suggest that a variety of K+ channels on erythrocyte membranes are involved in G8-induced hemolysis.

G8 increases the size of red blood cells, whereas TEA decreases their size and protects against G8-induced cell swelling. Because TEA blocks K+ channels and decreases K+ efflux, TEA-treated cells are resistant to hemolytic agents such as gramicidin (21, 32) and G8 (this study). Gramicidin, a linear pentadecapeptide antibiotic, opens K+ channels and induces K+ efflux from, and Na+ flux into, red blood cells. We also found that gramicidin at very low concentration induced the swelling of red blood cells and hemolysis and that TEA completely blocked gramicidin-induced hemolysis and swelling of human red blood cells (data not shown).

The anion transporter (band 3 protein) exchanges Cl with HCO3 and accounts for the high permeability of erythrocytes to Cl ions (7). It also helps define erythrocyte shape and membrane stability. DIDS, an anion transport inhibitor, protects against hemolysis induced by glycerol (42) and by mannan extracted from the Candida albicans cell wall (40). We observed that G8 induced an increase of the intracellular Cl concentration in red blood cells and that DIDS at 50 μM completely blocked G8-induced hemolysis. The ID50 was only 2.42 μM, suggesting that anion transport is pivotal in G8-induced hemolysis. On the other hand, Cl channel blockers, NPPB and 9-AC, did not inhibit G8-induced hemolysis, suggesting that these agents affect different Cl channels (11) or have different effects on Cl channels (22). Guizouarn et al. showed that activating anion transport not only induces an increase in the Cl influx but also induces Cl-independent Na+ and K+ permeability, which can be inhibited by several anion transport inhibitors including DIDS (15). These data partially explain why DIDS completely blocked G8-induced hemolysis. We speculate that G8-activated anion transport of red blood cells induced a Cl influx and an accompanying increase in intracellular Na+ and K+ efflux and that DIDS prevented these G8-induced ion fluxes by inhibiting anion transport. This also explains why the Cl channel blockers, 9-AC and NPPB, did not block G8-induced hemolysis. 9-AC and NPPB inhibit only Cl influx, while DIDS inhibits both Cl and Na+ influx and K+ efflux. Taken together, these data suggest that G8 opens/activates anion transport on red blood cell membranes or induces a DIDS-sensitive Cl permeability with subsequent increases in intracellular Cl and Na+ concentrations and efflux of K+, mediating G8-induced hemolysis.

Na+ channels transport Na+ from the extracellular fluid to the inside of the cell, and blocking the Na+ channels of red blood cells with amiloride attenuates oxidation-induced hemolysis (10). We found that the Na+ channels blockers amiloride and QX222 inhibited G8-induced hemolysis and that G8 induced an increase in intracellular Na+ levels. Thus, G8 activates Na+ channels and increases intracellular Na+ levels, causing passive entry of water into the cell and swelling of red blood cells, eventually leading to hemolysis.

Because mature red blood cells lack intracellular calcium stores, increases in intracellular calcium levels stem from calcium influx from extracellular calcium. When the intracellular free Ca2+ concentration increases in human erythrocytes, a large K+ loss is observed. This effect is mediated by a Ca2+-dependent K+ channel, first described by Gardos (13). We previously showed that G8 induced an increase in the intracellular Ca2+ concentration and that the Ca2+ channel blockers econazole and Ni2+ (17) inhibited G8-induced apoptosis in tumor Jurkat cells (31). Likewise, we show here that econazole and Ni2+ also inhibit G8-induced hemolysis. G8 induces an increase in the intracellular Ca2+ concentration in human red blood cells, suggesting that this effect occurs via Ca2+-activated K+ channels (13). On the other hand, Alvarez et al. (1) reported that the econazole, which interferes with a number of cytochrome P-450-dependent molecules, especially CYP51, affects receptor-operated Ca2+ channels in human red blood cells with an IC50 of 0.2 to 3 μM and also affects the calcium-dependent K+ channel with an IC50 of 1.8 μM. Aussel and Breittmeyer (5) found that in Jurkat cells, imidazole antimycotic inhibitors of P-450 such as econazole increase phosphatidylserine synthesis similarly to the effect of K+ channel blockers. Taken together, the modest inhibition of G8-induced hemolysis observed with econazole in the present study may be due to its effects on Ca2+ channels but also to its effects on calcium-dependent K+ channels.

The [K+/Cl] cotransport system utilizes the high K+ content of human erythrocytes to extrude K+ and Cl from the erythrocytes. One K+ ion and one Cl ion are transported from inside cells to the exterior fluid (26). This [K+/Cl] cotransport system plays an important role in the pathologic dehydration of sickle erythrocytes (7). DIOA is one of the most potent inhibitors of [K+/Cl] cotransport in human red blood cells (4). We found that DIOA did not inhibit G8-induced hemolysis even at very high concentrations, suggesting that [K+/Cl] cotransport is not involved in the G8-induced hemolysis.

[Na+/K+/2Cl] cotransport allows two Cl ions to be carried into cells along with one Na+ ion and one K+ ion, all moving in the same direction (25). In human erythrocytes, the [Na+/K+/2Cl] cotransport causes a small net extrusion of Na+, K+, and Cl under physiologic conditions and plays only a minor role in the regulation of human erythrocyte volume (7). In the present study, we observed a minimal (less than 10%) inhibitory effect on G8-induced hemolysis at 50 to 200 μM bumetanide, a [Na+/K+/2Cl] cotransport inhibitor, suggesting that this cotransport system does not play a role in G8-induced hemolysis.

The [Na+/K+] pump exchanges three Na+ ions outside the cell with two K+ ions going into the cell (7). In this study, ouabain, a Na+/K+ pump inhibitor, does not affect G8-induced hemolysis at 10 to 200 μM, indicating that this transport system is not involved in protecting erythrocytes from G8.

We also showed that colloid-osmotic protectants, such as d-mannitol, sucrose, raffinose, and polyethylene glycols with molecular weights of 600, 1,000 and 2,000 did not inhibit G8-induced hemolysis, indicating that this lysis is not colloid osmotic in nature. In contrast, these agents protect against gramicidin S, an amphipathic peptide previously shown to induce hemolysis (21).

Surprisingly, the hemolytic activity of G8 was significantly inhibited by pretreatment of red blood cells with either granulysin or peptide G9, even though neither of these agents is hemolytic. Because both G8 and granulysin can bind red blood cells, we speculate that G9 and granulysin compete with G8 for the same binding sites on red blood cells. Thus, binding is required but not sufficient to induce hemolysis.

In this paper, we show that granulysin-mediated cell death requires intact mitochondria in target cells but that some constituent granulysin peptides can directly induce hemolysis by activating ion channels. Recent crystal structure data indicate that granulysin binds to the cell membrane through a cluster of positive residues that appear to aggregate and then tunnel into the membrane through its hydrophobic core to cause “molecular electroporation” (3). Recombinant granulysin protects erythrocytes from damage, but constituent peptides can be hemolytic. Our data indicate that G8, a constituent peptide, interferes with the ability of the red blood cell membrane to maintain ion gradients by activating channels for K+, Na+, Cl, and Ca2+. This results in a Na+, Cl, and Ca2+ influx into and K+ efflux from red blood cells. The intracellular Na+, Ca2+, and Cl concentrations increase and the intracellular K+ concentration decreases. Subsequently, water enters the cell following the osmotic gradient, the cell swells, and hemolysis occurs.


This work was supported by NIH grant R01 AI43348 to A.M.K. A.M.K. is the Shelagh Galligan Professor of Pediatrics.

We thank Masakatsu Kato, Department of Physiology, Nippon Medical School, for his help in measurement of intracellular K+ concentrations.


1. Alvarez, J., M. Montero, and J. Garcia-Sancho. 1992. High affinity inhibition of Ca2+-dependent K+ channels by cytochrome P-450 inhibitors. J. Biol. Chem. 267:11789-11793. [PubMed]
2. Amorino, G. P., and M. H. Fox. 1996. Effects of hyperthermia on intracellular chloride. J. Membr. Biol. 152:217-222. [PubMed]
3. Anderson, D. H., M. R. Sawaya, D. Cascio, W. Ernst, R. Modlin, A. Krensky, and D. Eisenberg. 2003. Granulysin crystal structure and a structure-derived lytic mechanism. J. Mol. Biol. 325:355-365. [PubMed]
4. Anfinogenova, Y. J., X. Rodriguez, R. Grygorczyk, N. C. Adragna, P. K. Lauf, P. Hamet, and S. N. Orlov. 2001. Swelling-induced K+ fluxes in vascular smooth muscle cells are mediated by charybdotoxin-sensitive K+ channels. Cell Physiol. Biochem. 11:295-310. [PubMed]
5. Aussel, C., and J. P. Breittmayer. 1993. Imidazole antimycotics inhibitors of cytochrome P450 increase phosphatidylserine synthesis similarly to K+ channel blockers in Jurkat T cells. FEBS Lett 319:155-158. [PubMed]
6. Brain, M. C., J. M. Prevost, C. E. Pihl, and C. B. Brown. 2002. Glycophorin A-mediated haemolysis of normal human erythrocytes: evidence for antigen aggregation in the pathogenesis of immune haemolysis. Br. J. Haematol. 118:899-908. [PubMed]
7. Brugnara, C. 1997. Erythrocyte membrane transport physiology. Curr. Opin. Hematol. 4:122-127. [PubMed]
8. Chou, A. C., and C. D. Fitch. 1981. Mechanism of hemolysis induced by ferriprotoporphyrin IX. J. Clin. Investig. 68:672-677. [PMC free article] [PubMed]
9. Dunn, P. M. 1998. The action of blocking agents applied to the inner face of Ca2+-activated K+ channels from human erythrocytes. J. Membr. Biol. 165:133-143. [PubMed]
10. Duranton, C., S. M. Huber, and F. Lang. 2002. Oxidation induces a Cl-dependent cation conductance in human red blood cells. J. Physiol. 539:847-855. [PubMed]
11. Egee, S., O. Mignen, B. J. Harvey, and S. Thomas. 1998. Chloride and non-selective cation channels in unstimulated trout red blood cells. J. Physiol. 511:213-224. [PubMed]
12. Ernst, W. A., S. Thoma-Uszynski, R. Teitelbaum, C. Ko, D. A. Hanson, C. Clayberger, A. M. Krensky, M. Leippe, B. R. Bloom, T. Ganz, and R. L. Modlin. 2000. Granulysin, a T cell product, kills bacteria by altering membrane permeability. J. Immunol. 165:7102-7108. [PubMed]
13. Gardos, G. 1958. The function of calcium in the potassium permeability of human erythrocytes. Biochim. Biophys. Acta 30:653-654. [PubMed]
14. Gray, M., G. Szabo, A. S. Otero, L. Gray, and E. Hewlett. 1998. Distinct mechanisms for K+ efflux, intoxication, and hemolysis by Bordetella pertussis AC toxin. J. Biol. Chem. 273:18260-18267. [PubMed]
15. Guizouarn, H., N. Gabillat, R. Motais, and F. Borgese. 2001. Multiple transport functions of a red blood cell anion exchanger, tAE1: its role in cell volume regulation. J. Physiol. 535:497-506. [PubMed]
16. Halperin, J. A., C. Brugnara, A. S. Kopin, J. Ingwall, and D. C. Tosteson. 1987. Properties of the Na+-K+ pump in human red cells with increased number of pump sites. J. Clin. Investig. 80:128-137. [PMC free article] [PubMed]
17. Han, J. H., J. H. Lee, Y. H. Choi, J. H. Park, T. J. Choi, and I. S. Kong. 2002. Purification, characterization and molecular cloning of Vibrio fluvialis hemolysin. Biochim. Biophys. Acta 1599:106-114. [PubMed]
18. Hanson, D. A., A. A. Kaspar, F. R. Poulain, and A. M. Krensky. 1999. Biosynthesis of granulysin, a novel cytolytic molecule. Mol. Immunol. 36:413-422. [PubMed]
19. Housman, D., M. Jacobs-Lorena, U. L. Rajbhandary, and H. F. Lodish. 1970. Initiation of haemoglobin synthesis by methionyl-tRNA. Nature 227:913-918. [PubMed]
20. Kaspar, A. A., S. Okada, J. Kumar, F. R. Poulain, K. A. Drouvalakis, A. Kelekar, D. A. Hanson, R. M. Kluck, Y. Hitoshi, D. E. Johnson, C. J. Froelich, C. B. Thompson, D. D. Newmeyer, A. Anel, C. Clayberger, and A. M. Krensky. 2001. A distinct pathway of cell-mediated apoptosis initiated by granulysin. J. Immunol. 167:350-356. [PubMed]
21. Katsu, T., C. Ninomiya, M. Kuroko, H. Kobayashi, T. Hirota, and Y. Fujita. 1988. Action mechanism of amphipathic peptides gramicidin S and melittin on erythrocyte membrane. Biochim. Biophys. Acta 939:57-63. [PubMed]
22. Kocic, I., Y. Hirano, and M. Hiraoka. 2001. Ionic basis for membrane potential changes induced by hypoosmotic stress in guinea-pig ventricular myocytes. Cardiovasc. Res. 51:59-70. [PubMed]
23. Lang, P. A., S. Kaiser, S. Myssina, T. Wieder, F. Lang, and S. M. Huber. 2003. Role of Ca2+-activated K+ channels in human erythrocyte apoptosis. Am. J. Physiol. Ser. C 285:C1553-C1560. [PubMed]
24. Lapaix, F., S. Egee, L. Gibert, G. Decherf, and S. L. Thomas. 2002. ATP-sensitive K+ and Ca2+-activated K+ channels in lamprey (Petromyzon marinus) red blood cell membrane. Pflugers Arch. 445:152-160. [PubMed]
25. Lauf, P. K., and N. C. Adragna. 2000. K-Cl cotransport: properties and molecular mechanism. Cell Physiol. Biochem. 10:341-354. [PubMed]
26. Lauf, P. K., J. Bauer, N. C. Adragna, H. Fujise, A. M. Zade-Oppen, K. H. Ryu, and E. Delpire. 1992. Erythrocyte K-Cl cotransport properties and regulation. Am. J. Physiol. Sect. C 263:C917-C932. [PubMed]
27. Li, F., K. Sugishita, Z. Su, I. Ueda, and W. H. Barry. 2001. Activation of connexin-43 hemichannels can elevate [Ca2+]i and [Na+]i in rabbit ventricular myocytes during metabolic inhibition. J. Mol. Cell. Cardiol. 33:2145-2155. [PubMed]
28. Long-Rowe, K. O., and J. W. Burnett. 1994. Characteristics of hyaluronidase and hemolytic activity in fishing tentacle nematocyst venom of Chrysaora quinquecirrha. Toxicon 32:165-174. [PubMed]
29. Marchetti, P., S. A. Susin, D. Decaudin, S. Gamen, M. Castedo, T. Hirsch, N. Zamzami, J. Naval, A. Senik, and G. Kroemer. 1996. Apoptosis-associated derangement of mitochondrial function in cells lacking mitochondrial DNA. Cancer Res. 56:2033-2038. [PubMed]
30. Nelson, M. T., and J. M. Quayle. 1995. Physiological roles and properties of potassium channels in arterial smooth muscle. Am. J. Physiol. Sect. C 268:C799-C822. [PubMed]
31. Okada, S., Q. Li, J. C. Whitin, C. Clayberger, and A. M. Krensky. 2003. Intracellular mediators of granulysin-induced cell death. J. Immunol. 171:2556-2562. [PubMed]
32. Otten-Kuipers, M. A., T. L. Beumer, N. A. Kronenburg, B. Roelofsen, and J. A. Op den Kamp. 1996. Effects of gramicidin and tryptophan-N-formylated gramicidin on the sodium and potassium content of human erythrocytes. Mol. Membr. Biol. 13:225-232. [PubMed]
33. Pena, S. V., D. A. Hanson, B. A. Carr, T. J. Goralski, and A. M. Krensky. 1997. Processing, subcellular localization, and function of 519 (granulysin), a human late T cell activation molecule with homology to small, lytic, granule proteins. J. Immunol. 158:2680-2688. [PubMed]
34. Pena, S. V., and A. M. Krensky. 1997. Granulysin, a new human cytolytic granule-associated protein with possible involvement in cell-mediated cytotoxicity. Semin. Immunol. 9:117-125. [PubMed]
35. Princen, K., S. Hatse, K. Vermeire, G. J. Bridger, R. T. Skerlj, E. De Clercq, and D. Schols. 2003. The antiviral activity of the CXCR4 antagonist AMD3100 is independent of the cytokine-induced CXCR4/HIV coreceptor expression level. AIDS Res. Hum. Retroviruses 19:1135-1139. [PubMed]
36. Sabath, D. E., S. T. Cross, and L. Y. Mamiya. 2003. An improved method for detecting red cells with hemoglobin H inclusions that does not require glass capillary tubes. Clin. Lab. Haematol. 25:87-91. [PubMed]
37. Stenger, S., D. A. Hanson, R. Teitelbaum, P. Dewan, K. R. Niazi, C. J. Froelich, T. Ganz, S. Thoma-Uszynski, A. Melian, C. Bogdan, S. A. Porcelli, B. R. Bloom, A. M. Krensky, and R. L. Modlin. 1998. An antimicrobial activity of cytolytic T cells mediated by granulysin. Science 282:121-125. [PubMed]
38. Virkki, L. V., and M. Nikinmaa. 1998. Two distinct K+ channels in lamprey (Lampetra fluviatilis) erythrocyte membrane characterized by single channel patch clamp. J. Membr. Biol. 163:47-53. [PubMed]
39. Wang, Z., E. Choice, A. Kaspar, D. Hanson, S. Okada, S. C. Lyu, A. M. Krensky, and C. Clayberger. 2000. Bactericidal and tumoricidal activities of synthetic peptides derived from granulysin. J. Immunol. 165:1486-1490. [PubMed]
40. Watanabe, T., M. Takano, M. Murakami, H. Tanaka, A. Matsuhisa, N. Nakao, T. Mikami, M. Suzuki, and T. Matsumoto. 1999. Characterization of a haemolytic factor from Candida albicans. Microbiology 145:689-694. [PubMed]
41. Zou, C. G., N. S. Agar, and G. L. Jones. 2002. Chlorodinitrobenzene-mediated damage in the human erythrocyte membrane leads to haemolysis. Life Sci. 71:735-746. [PubMed]
42. Zou, C. G., N. S. Agar, and G. L. Jones. 2000. Haemolysis of human and sheep red blood cells in glycerol media: the effect of pH and the role of band 3. Comp. Biochem. Physiol. Sect. A 127:347-353. [PubMed]

Articles from Antimicrobial Agents and Chemotherapy are provided here courtesy of American Society for Microbiology (ASM)