Deletion of GLUT1 and GLUT3 Reveals Multiple Roles for Glucose Metabolism in Platelet and Megakaryocyte Function

Deletion of GLUT1 and GLUT3 Reveals Multiple Roles for Glucose Metabolism in Platelet and Megakaryocyte Function

Article Deletion of GLUT1 and GLUT3 Reveals Multiple Roles for Glucose Metabolism in Platelet and Megakaryocyte Function Graphical Abstract Authors ...

4MB Sizes 1 Downloads 96 Views

Article

Deletion of GLUT1 and GLUT3 Reveals Multiple Roles for Glucose Metabolism in Platelet and Megakaryocyte Function Graphical Abstract

Authors Trevor P. Fidler, Robert A. Campbell, Trevor Funari, ..., Dipayan Chaudhuri, Andrew S. Weyrch, E. Dale Abel

Correspondence [email protected]

In Brief Fidler et al. show that glucose metabolism is essential for platelet production, activation, and clearance. Their findings reveal complementary roles for glycolysis versus mitochondrial metabolism in platelet viability. Blocking both metabolic pathways leads to complete clearance of platelets from the circulation, due to calcium overload and calpain activation.

Highlights d

Glucose metabolism is required for megakaryocytemediated platelet production

d

Platelet metabolism is essential for platelet survival

d

d

Energetically stressed platelets undergo Ca2+-calpainmediated necrosis and clearance Glucose metabolism modulates multiple nodes of platelet activation

Fidler et al., 2017, Cell Reports 20, 881–894 July 25, 2017 ª 2017 The Author(s). http://dx.doi.org/10.1016/j.celrep.2017.06.083

Cell Reports

Article Deletion of GLUT1 and GLUT3 Reveals Multiple Roles for Glucose Metabolism in Platelet and Megakaryocyte Function Trevor P. Fidler,1,2,3 Robert A. Campbell,2,4 Trevor Funari,3 Nicholas Dunne,3 Enrique Balderas Angeles,5 Elizabeth A. Middleton,2,4 Dipayan Chaudhuri,5 Andrew S. Weyrch,2,4 and E. Dale Abel2,3,6,* 1Department

of Pharmacology and Toxicology, University of Utah, Salt Lake City, UT 84112, USA in Molecular Medicine, University of Utah, Salt Lake City, UT 84112, USA 3Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Carver College of Medicine, University of Iowa, Iowa City, IA 52242, USA 4Department of Internal Medicine, University of Utah, Salt Lake City, UT 84112, USA 5Nora Eccles Harrison Cardiovascular Research and Training Institute, University of Utah, Salt Lake City, UT 84112, USA 6Lead Contact *Correspondence: [email protected] http://dx.doi.org/10.1016/j.celrep.2017.06.083 2Program

SUMMARY

Anucleate platelets circulate in the blood to facilitate thrombosis and diverse immune functions. Platelet activation leading to clot formation correlates with increased glycogenolysis, glucose uptake, glucose oxidation, and lactic acid production. Simultaneous deletion of glucose transporter (GLUT) 1 and GLUT3 (double knockout [DKO]) specifically in platelets completely abolished glucose uptake. In DKO platelets, mitochondrial oxidative metabolism of non-glycolytic substrates, such as glutamate, increased. Thrombosis and platelet activation were decreased through impairment at multiple activation nodes, including Ca2+ signaling, degranulation, and integrin activation. DKO mice developed thrombocytopenia, secondary to impaired pro-platelet formation from megakaryocytes, and increased platelet clearance resulting from cytosolic calcium overload and calpain activation. Systemic treatment with oligomycin, inhibiting mitochondrial metabolism, induced rapid clearance of platelets, with circulating counts dropping to zero in DKO mice, but not wild-type mice, demonstrating an essential role for energy metabolism in platelet viability. Thus, substrate metabolism is essential for platelet production, activation, and survival. INTRODUCTION Glucose enters cells via glucose transporters, of which platelets express the facilitative glucose transporter 1 (GLUT1) (Craik et al., 1995) and glucose transporter 3 (GLUT3) (Heijnen et al., 1997). Of GLUT3, 85% is located in a-granule membranes with 15% localized in the plasma membrane. Upon degranulation, GLUT3 translocates to the plasma membrane and is believed to account for the increased glucose uptake following

activation (Heijnen et al., 1997). Platelet activation also increases glycolysis and lactic acid production (Karpatkin, 1967), glycogenolysis (Scott, 1967), and glucose oxidation (Warshaw et al., 1966). In vitro inhibition of these processes blunts platelet activation (Akkerman and Holmsen, 1981; Yamagishi et al., 2001), while incubation under hyperglycemic conditions potentiates activation in response to ADP, arachidonic acid, thrombin, and collagen (Yamagishi et al., 2001). However, these associations have not been evaluated in vivo. Although the correlation between activation and increased metabolism is well established, the mechanisms underlying these relationships are unknown. Studies of permeabilized platelets, lacking cytosol, indicate that ATP is an essential cofactor for Ca2+-induced degranulation (Flaumenhaft et al., 1999). Administration of the competitive glucose uptake inhibitor 2-deoxyglucose (2-DOG) to platelets suggested that glucose uptake and metabolism play an important role in platelet aggregation and degranulation. Although these changes correlate with ATP content, a direct relationship between platelet glucose uptake and ATP content remains to be demonstrated (Akkerman et al., 1979). Conversely, platelets isolated from diabetic patients demonstrate dysfunctional Ca2+ signaling (Li et al., 2001), although it is unknown if platelet activation in this context arises from altered metabolism. Mitochondrial respiration has also been linked to platelet activation. Inhibitors of mitochondrial respiration blunt platelet activation (Garcia-Souza and Oliveira, 2014). Upon activation, a subpopulation of platelets demonstrate mitochondrial depolarization and generation of reactive oxygen species (ROS), which facilitates phosphatidylserine (PS) exposure to the outer leaflet of the plasma membrane that promotes platelet procoagulant activity (Garcia-Souza and Oliveira, 2014). In addition to thrombotic functions, platelet mitochondria occupy a central role in the intrinsic apoptosis pathway (Mason et al., 2007). The balance between BCL-XL activity and the formation of mitochondrial Bak/Bax pores regulates platelet apoptosis and platelet viability. However, the relationship between mitochondrial metabolism and apoptotic cell death pathways in platelets remains to be clarified. Moreover, mice infected with dengue demonstrate an additional mechanism of platelet clearance that correlated with

Cell Reports 20, 881–894, July 25, 2017 ª 2017 The Author(s). 881 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

B

C

Ponceau S

**

** l

Ia Ia IIa IIa KO -KO DKO l + I + I + + 13 o KO KO KO T r T t U U n D 31GL GL Co UT UT L GL G

Control

Control Control + IIa DKO DKO + IIa

#

ECAR mpH/min

e on

P

en

C

O

R

ot

om lig

FC

yc

sa Ba

DKO

in

l

12

***

#

****

****

# ****

C-Lactic Acid Production pg/minute/1x106 Platelets

E

Control

**

ro

nt

Co

D

***

*

DKO

F

Glycogen Content (pg)

GLUT3

13

GLUT1

C-Lactic Acid Production pg/minute/1x106 Platelets

# #

2-Deoxy-D-Glucose (pmole)/ 1x106 Platelets/30 min

G LU T G 1 -C LU on G T1- trol LU K O T 3 G L U -C T on D 3-K tro KO O l D Co KO nt ro l

A

** l

ro

nt

Co

O

DK

Figure 1. Glucose Metabolism Is Decreased in DKO Platelets (A) Representative western blot of protein lysates from GLUT1-KO, GLUT3-KO, DKO, and respective littermate control platelets. (B) [3H] 2-DOG glucose uptake in platelets. (C and D) 13C-Lactic acid production (C) and 12C-lactic acid (D) in the presence of 13C-Glucose exclusively in the extracellular media (n = 3). (E) Seahorse analysis of DKO platelet extracellular acidification rate (ECAR) under non-stimulated and thrombin (IIa)-stimulated conditions (n = 3). (F) Glycogen analysis of platelets normalized to cell number (n = 5). Error bars are SEM (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 with respect to control genotype; #p < 0.05 with respect to treatment, same genotype; two-way ANOVA followed by Bonferroni multiple comparison post hoc test, B and E; Student’s t test, C, D, and F).

PS exposure in vivo (Alonzo et al., 2012). Given the diverse mechanisms leading to platelet clearance, the integration of these mechanisms in the regulation of platelet viability and the contribution of platelet metabolism to these processes remain to be clarified. Given the dearth of information regarding the role of energy metabolism in regulating platelet biology in vivo, we generated mice lacking GLUT1 and GLUT3 specifically in platelets. Analyses of these mice revealed a dynamic interaction between glucose utilization and mitochondrial energy metabolism to maintain platelet energy requirements. Importantly, glucose metabolism is essential for pro-platelet formation from megakaryocytes. Glucose metabolism is essential for agonist-mediated Ca2+ signaling and downstream signaling events leading to platelet activation. We demonstrate a critical role for glycolytic and mitochondrial metabolism in regulating platelet clearance, establishing a paradigm whereby impaired platelet energy metabolism undermines calcium homeostasis, leading to increased cytoplasmic Ca2+ that activates a calpain-mediated cell death and clearance pathway.

882 Cell Reports 20, 881–894, July 25, 2017

RESULTS Glucose Metabolism Is Abolished in DKO Platelets To investigate the contribution of glucose metabolism to platelet function, GLUT1 and GLUT3 were individually or simultaneously deleted from platelets by crossing mice expressing a Pf4 promoter-driven Cre recombinase to mice harboring homozygous floxed GLUT1 and GLUT3 alleles individually or in combination. Immunoblot analysis of platelet protein lysates isolated from GLUT1 single-knockout (GLUT1-KO), GLUT3 single-knockout (GLUT3-KO), and GLUT1/GLUT3 double-knockout (DKO) mice and their respective littermate controls (Figure 1A) confirmed the absence of the respective proteins. Glucose uptake was measured in control quiescent platelets at 10 pmoles/1 3 106 platelets/30 min, which after thrombin stimulation increased 2-fold. Deletion of GLUT3 reduced basal glucose uptake by 22%, and it abolished thrombin-stimulated glucose uptake (Figure 1B), illustrating that GLUT3 mediates post-activation glucose uptake. Surprisingly, GLUT1-KO platelets revealed no changes in basal or thrombin-mediated glucose uptake relative

to controls. However, deletion of both GLUT1 and GLUT3 completely abolished glucose uptake under basal and thrombin-stimulated conditions, indicating that GLUT1 and GLUT3 are the biologically relevant glucose transporters in platelets. These data also demonstrate that GLUT1 and GLUT3 have unique but also overlapping roles in the regulation of platelet glucose uptake. DKO platelets do not undergo glycolysis. Quantitative analysis of 13C-1,6-glucose flux demonstrated that DKO platelets failed to convert exogenous glucose to lactic acid (Figure 1C), whereas controls generated 0.0085 pg lactic acid/minute/1 3 106 platelets. In addition, with 13C-glucose as the exclusive exogenous glucose source, 12C-lactic acid release, which represents the contribution of glycogen to glycolysis, was also abolished (Figure 1D). Qualitative analysis of glycolysis determined that the extracellular acidification rate (ECAR) was markedly repressed in DKO platelets (Figure 1E). Of note, ECAR was not zero in DKO platelets, potentially reflecting acidification of the media by acids other than lactate (TeSlaa and Teitell, 2014). Consistent with studies in human platelets (Karpatkin and Langer, 1968), thrombin increased glycolysis by 2-fold in control murine platelets. In contrast, DKO platelets did not increase glycolysis following thrombin stimulation or in the presence of inhibitors of mitochondrial metabolism (Figure 1E). Platelets contain large glycogen stores, which can be mobilized upon activation (Scott, 1967). Glycogen content in DKO platelets was reduced 7-fold to levels near the limit of detection (Figure 1F). Thus, glucose uptake and metabolism are largely abolished in DKO platelets. DKO Platelets Increase Utilization of Alternative Mitochondrial Substrates In the absence of glucose uptake, alternative metabolic pathways may sustain cellular ATP to maintain cellular viability. DKO platelets displayed a significantly increased ratio of (AMP + ADP)/(ATP) (Figure 2A), suggesting metabolic stress. Consistent with this, freshly isolated DKO platelets demonstrated increased phosphorylated AMP-activated protein kinase (AMPK) and acetyl-CoA carboxylase (ACC) (Figure 2B). Interestingly, mitochondrial membrane potential was significantly increased in freshly isolated DKO platelets independent of exogenous substrate present in the media (Figure 2C). Seahorse analysis of mitochondrial oxygen consumption rates (OCRs) in the presence of glucose, glutamate, and pyruvate indicated that, at baseline and under maximal respiration conditions, OCRs in DKO platelets were 3-fold higher than in controls (Figure 2D). Increased mitochondrial respiration in DKO reflects increased capacity to utilize mitochondrial substrates. OCRs were equivalent in control and DKO platelets with glucose as the sole metabolic substrate; however, following the addition of glutamate and pyruvate, mitochondrial OCR increased by 70% in DKO, but it was unchanged in controls (Figure 2E). DKO platelets exhibited decreased intracellular lactic acid and relatively increased pyruvate leading to a significantly reduced lactate/pyruvate ratio (Figure 2F), suggesting the induction of alternative pathways of pyruvate generation such as the transamination of alanine that requires mitochondrial a-ketoglutarate (Gray et al., 2014). Mitochondrial content and gross structure of DKO platelets were unchanged (Figures 3A and 3B), suggesting a qualitative rather

than a quantitative change in mitochondrial function. Thus, the absence of glucose metabolism in DKO platelets increased metabolism of alternative mitochondrial substrates. Platelet Activation Is Decreased in DKO Platelets DKO platelets exhibited blunted platelet activation. Non-stimulated DKO platelets displayed similar organelle distribution, with the exception of decreased a-granule number (Figures 3A and 3B), which may suggest a biogenesis defect. In addition, following stimulation with 250 mM PAR4 peptide, degranulation was impaired in DKO platelets, as evidenced by the retention of a-granules in platelets (Figures 3A and 3C). In vitro analysis of platelets incubated in 5 mM glucose for 1 hr revealed decreased activation of GPIIbIIIa, measured by JonA geo. Mean fluorescence intensity (MFI) (Figure 3D), and a-granule degranulation, marked by CD62p surface translocation (CD62p geo. MFI) (Figure 3E) in response to PAR4 peptide, the thromboxane receptor agonist U46619, ADP, and convulxin. However, this loss of activation was only partially rescued (ADP and U46619, JonA) when glutamate and pyruvate were added to the media (Figures 3D and 3E), suggesting a direct link between glucose metabolism and platelet activation that is not rescued in the presence of alternate mitochondrial substrates in DKO platelets. Agonist-Mediated Calcium Signaling Is Impaired in DKO Platelets In response to activating stimuli, platelets increase cytoplasmic Ca2+ that activates the Ca2+-sensitive scramblase TMEM16f to facilitate the translocation of PS to the outer leaflet of the plasma membrane, leading to a procoagulant response (Jobe et al., 2008; van Kruchten et al., 2013). Although PS is exposed, this procoagulant process is distinct from apoptosis-mediated PS exposure, and it is not thought to lead to an ‘‘eat me’’ signal that results in platelet clearance (van Kruchten et al., 2013). In response to thrombin plus convulxin, DKO platelets demonstrated impaired exposure of PS to the outer leaflet of the plasma membrane measured by annexin V binding (Figures 4A and S1A–S1F). Restoring Ca2+ flux by administration of the Ca2+ ionophore ionomycin completely restored annexin V binding (Figure 4A). In response to thrombin, cytoplasmic Ca2+ flux was blunted in DKO platelets (Figure 4B). Importantly, following the rise in agonist-mediated cytoplasmic Ca2+, control platelets restored cytoplasmic Ca2+ to near basal levels (Figure 4B), whereas levels remained increased in DKO platelets for >10 min following stimulation. This inability to export/sequester Ca2+ following stimulation suggests a defect in sarco/endoplasmic reticulum Ca2+-ATPase (SERCA)-mediated and/or plasma membrane Ca2+-ATPase (PMCA)-mediated Ca2+ transport. Using an independent methodology, we confirmed impairment of platelet Ca2+ mobilization following combined stimulation by thrombin and convulxin of DKO platelets (Figures 4C and 4D). Following thapsigargin treatment, which induces store-operated Ca2+ entry, DKO platelets failed to restore Ca2+ concentrations to control levels (Figures 4E and 4F), revealing a defect in store-operated Ca2+ entry. DKO platelets exhibited impaired GPIIbIIIa activation and CD62p surface translocation following treatment with thrombin plus convulxin, the Ca2+ ionophore ionomycin, or the SERCA

Cell Reports 20, 881–894, July 25, 2017 883

Figure 2. DKO Platelets Increase Mitochondrial Metabolism in response to Metabolic Stress (A) Estimation of AMP + ADP to ATP ratio in whole platelet lysates (n = 8). (B) Western blot analysis of protein lysates from freshly isolated platelets (n = 3). (C) Mitochondrial membrane potential of platelets incubated in the indicated media for 1 hr (n = 3). (D) Seahorse analysis of platelet O2 consumption in 25 mM glucose + 1 mM glutamate + 1 mM pyruvate media ± thrombin (IIa) (n = 3). (E) Seahorse analysis of DKO and control platelets under basal conditions in the presence of 25 mM glucose ± 1 mM glutamate and 1 mM pyruvate (n = 3). (F) Ratio of intracellular lactic acid to pyruvic acid (n = 6). Error bars are SEM (*p < 0.05, **p < 0.01, and ****p < 0.0001 with respect to control genotype; #p < 0.05 with respect to treatment, same genotype; one-way ANOVA followed by Tukey’s multiple comparison post hoc test, C and E; two-way ANOVA followed by Bonferroni multiple comparison post hoc test, D; Student’s t test, A and F).

inhibitor thapsigargin (Figures 4G and 4H). These data indicate that glucose metabolism regulates PS exposure to the outer leaflet of the plasma membrane via its effect on calcium mobilization; but, even when calcium flux is rescued, glucose metabolism mediates additional steps required for integrin activation and a-granule degranulation. Platelet Glucose Metabolism Is Essential for In Vivo Thrombosis The contribution of platelet glucose metabolism to thrombosis and hemostasis in vivo is unknown. In a tail-bleeding assay, DKO mice exhibited significantly longer time to bleeding cessation, with many mice failing to stop prior to assay completion (Figure 4I). Although spontaneous bleeding was not observed,

884 Cell Reports 20, 881–894, July 25, 2017

hematocrit was modestly yet significantly reduced in untreated DKO mice (Table S1). Arterial thrombosis was also evaluated using 7.5% ferric chloride. DKO mice had significantly increased time to arterial occlusion relative to littermate controls, with many failing to occlude after 20 min of observation (Figure 4J). In a collagen/epinephrine-induced pulmonary embolism model, which is dependent on in vivo platelet aggregation, DKO mice demonstrated increased survival (Figure 4K). Thus, platelet glucose metabolism is essential for in vivo thrombosis. DKO Mice Are Thrombocytopenic Platelet counts were reduced in DKO mice (Figure 5A; Table S1); however, platelet counts were unchanged in mice with GLUT1or GLUT3-deficient platelets, respectively. To determine if

B

A

Basal Organelle/Platelet

DKO

Basal

Control

**

ria S DT hond ti oc M

ule nule ra δ -G

ran

α -G

PAR4 Organelle/Platelet

PAR4

C

****

s

ria

le

nu

ra

G

α-

C Control + Glucose C Control + Glucose + Glutamate + Pyruvate D DKO + Glucose D DKO + Glucose + Glutamate + Pyruvate

ito

M

nu

G

δ-

nu

om

s so

ra

s le

es

s

le

nd

o ch

Ly

ra

G

ed

α-

t la

u

m

Si

D

S OC

E

**** ****

in

**** ****

**** ****

0µ M 25

L 20

0n

g/ m

PA R

vu lx on C

U M 2µ 3.

10

µM

46

AD

61 9

P

in e el Ba s

4

**** ****

PA R

25 0µ

L m 20 0n

g/

M

vu on C

U M 2µ 3.

4

n lx i

61 9 46

AD 10 µM

Ba s

el in

e

P

****

**** *

**** ****

**** ****

# #

CD62p Geo. MFI

JonA Geo. MFI

#

Figure 3. Impaired Activation of DKO Platelets (A) Transmission electron microscopy of washed platelets in 5 mM glucose media, treated in the presence or absence of 250 mM Par4 (scale bar, 2 mm). (B–E) Quantification of non-stimulated (B, n = 4) and Par4 peptide-stimulated (C, n = 4) electron micrographs of platelets. Washed platelets pre-incubated for 30 min in the presence of 5 mM glucose ± 1 mM glutamate and 1 mM pyruvate were stimulated with the indicated agonist and analyzed for GPIIbIIIa activation (D, JonA geo. MFI) and CD62p surface translocation (E, CD62p Geo. MFI) (n = 5). Data are mean ± SEM (*p < 0.05, **p < 0.01, and ****p < 0.0001 with respect to control genotype; #p < 0.05 with respect to treatment, same genotype; two-way ANOVA followed by Bonferroni multiple comparison post hoc test, B–E).

thrombocytopenia resulted from decreased platelet production from megakaryocytes, we administered antibodies to GPIba that depletes platelets within 18 hr (Figure 5B). In control mice, platelet counts completely recovered after 96 hr, whereas DKO mice required 168 hr for platelet recovery, suggesting that mega-

karyocyte-mediated platelet biogenesis may be impaired under conditions associated with increased platelet consumption. GLUT1 and GLUT3 single-knockout mice exhibited no changes in platelet recovery (Figures S2A and S2B). Cross-sectional analysis of megakaryocytes in femurs and spleens of mice under

Cell Reports 20, 881–894, July 25, 2017 885

A

B

D

G

J

C

E

F

H

I

K

(legend on next page)

886 Cell Reports 20, 881–894, July 25, 2017

basal conditions revealed no significant change in megakaryocyte number in DKO mice (Figures 5C and 5D). Megakaryocytes cultured from DKO bone marrow displayed significantly reduced GLUT1 and GLUT3 protein content (Figures S2C and S2D), and they demonstrated virtually no glucose uptake (Figure 5E). Importantly, the ability of bone marrow-derived megakaryocytes to produce platelets was significantly impaired in DKO cultures (Figures 5F and 5G), even in the presence of glutamate and pyruvate. These observations define an obligate role for glucose metabolism in platelet generation from megakaryocytes. Phosphatidylserine Exposure Is Increased in DKO Platelets in Response to Metabolic Stress Circulating platelet counts reflect the balance between platelet production and clearance. Therefore, we investigated circulating half-life of platelets by the administration of a Dylight 488-labeled anti-GP1bb antibody. GLUT1 and GLUT3 single-knockout mice demonstrated no reduction in circulating half-life (Figures S2E and S2F). However, DKO platelets demonstrated significantly reduced circulating half-life compared to controls (Figure 6A), suggesting that increased clearance also contributes to the thrombocytopenia observed. An important mechanism of platelet clearance occurs via protein desialylation, which can be monitored by Ricinus Communis Agglutinin I (RCA-1) binding (Li et al., 2015). However, DKO platelets demonstrated no change in RCA-1 binding (Figure S3A), rendering this mechanism unlikely. We also ruled out increased autophagy or oxidative stress as mechanisms for increased clearance (Figures S3B and S3C). Platelet clearance can also be potentiated by increasing extracellular exposure of PS on plasma membranes (Alonzo et al., 2012). Therefore, we monitored annexin V binding on platelets in diluted whole blood, 72 hr post-administration of Dylight 488-labeled anti-GP1bb antibody. Interestingly, GP1bb-positive platelets, which represent older platelets, displayed increased relative annexin V binding marked by the geo. MFI (Figure 6B), but no significant increase was observed in GP1bb-negative platelets or in the population as a whole, suggesting that, as DKO platelets age, they increase PS exposure to the outer leaflet of the plasma membrane. In vitro, control platelets incubated in DMEM with 5 mM glucose ± 2 mM glutamate and 1 mM pyruvate had minimal annexin V-positive platelets after 22 hr of incubation (Figure 6C). In contrast, after 22 hr, DKO platelets incubated in glucose alone exhibited 70% annexin V positivity (Figure 6C). Supplementation of the media with the mitochondrial substrates

glutamate and pyruvate significantly reduced but did not normalize annexin V binding in DKO platelets (Figure 6C). Due to optimal signal resolution, a 6-hr in vitro incubation time was chosen for subsequent signaling studies. After 6-hr incubation, DKO platelets incubated in glucose alone exhibited 25% annexin V positivity, while the addition of glutamate and pyruvate decreased annexin V binding by half to 12%. Control platelets, regardless of exogenous substrates, demonstrated <5% annexin V binding (Figure 6D). Following a 6-hr incubation, annexin V-positive platelets displayed mitochondrial depolarization (Figure 6E). Western blot analysis indicated that AMPK and ACC phosphorylation, following 6-hr incubation, were significantly increased in DKO platelets incubated in glucose only, and the addition of glutamate and pyruvate to the media attenuated these changes (Figure 6F). Thus, energy deficits as indicated by AMPK activation correlate with increased annexin V positivity. If energy metabolism regulates platelet annexin V binding and clearance, we hypothesized that decreasing platelet energy metabolism would further increase annexin V binding. Thus, platelets were treated in vitro with the ATP synthase inhibitor oligomycin. DKO platelets demonstrated increased sensitivity to oligomycin as marked by annexin V geo. MFI (Figure 6G). Furthermore, in vivo intraperitoneal administration of oligomycin at 1 mg/kg, a concentration that did not lead to clinical toxicity or weight change (data not shown), reduced circulating platelet counts by 20% in controls and by 97% in DKO mice (Figure 6H) after 6 hr. Together these data indicate that platelet metabolic function directly modulates circulating platelet halflife. Oligomycin treatment of controls suggests that glycolytic metabolism might be sufficient to maintain platelet vitality. However, in the absence of glucose transport, oligomycin blocks alternative substrate utilization, leading to catastrophic ATP loss, annexin V exposure, and nearly complete platelet clearance. Increased Platelet Clearance in DKO Mice Is Partially Mediated by Calpain Activation Platelet PS exposure is facilitated by two distinct pathways. The first pathway is agonist-mediated mitochondrial permeability transition pore (mPTP) opening and cytoplasmic Ca2+ activation of the scramblase TMEM16f. The second is through the intrinsic apoptosis pathway, which requires BCL-XL degradation, mitochondrial depolarization, cytochrome c release into the cytosol, and caspase-3 activation (Schoenwaelder et al., 2009). To determine if opening of the mPTP resulted in increased

Figure 4. DKO Platelets Demonstrate Decreased Thrombosis (A) Washed platelets in 5 mM glucose media treated with the indicated agonist for 15 min were monitored for annexin V positivity (n = 6). (B) Representative tracing of thrombin-mediated cytoplasmic Ca2+ monitored via fluo-4 MFI, using flow cytometry (n = 3). (C) Fura-2-loaded platelets stimulated with 1 U/mL thrombin + 160 ng/mL convulxin, SEM (dashed lines) (n = 4). (D) Change in Ca2+ following the administration of thrombin + convulxin (n = 4). (E) Fura-2-loaded platelets in 250 mM EGTA, treated with thapsigargin (TG) and then 1 mM Ca2+ SEM (dashed lines) (n = 4). (F) Change in Ca2+ following the administration of thapsigargin (n = 4). (G and H) Relative GPIIbIIIa activation (G, relative JonA Geo. MFI) and CD62p surface translocation (H, relative Geo. MFI) following stimulation with the indicated agonist (n R 3). (I) Tail bleeding was assessed by monitoring time to bleeding cessation (n = 7). (J) Time to occlusion was determined in a 7.5% Ferric chloride-induced arterial thrombosis model (n = 6). (K) Length of survival in a collagen/epinephrine-induced pulmonary embolism model (n = 12). Data are mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; two-way ANOVA followed by Bonferroni multiple comparison post hoc test, A, G, and H; Student’s t test, D, F, I, and J; log rank [Mantel-Cox] test, K).

Cell Reports 20, 881–894, July 25, 2017 887

A

B

E C

F

D G

Figure 5. Platelet Production by DKO Megakaryocytes Is Decreased (A) Platelet counts in whole blood (n = 7). (B) Platelets were depleted through the administration of anti-GP1ba (black arrow), and platelet counts were monitored (n = 6). (C) Representative image of immunohistochemistry for vWF in femurs and spleens from DKO and control mice counterstained with eosin. (D) Quantification of megakaryocyte density in femurs (n = 4) and spleens (n = 3). (E) Glucose uptake in cultured megakaryocytes. (F) Representative image of megakaryocytes derived from control and DKO bone marrow (scale bar, 60 mm). (G) Quantification of pro-platelet formation by bone marrow-derived megakaryocytes (n = 5). Data are mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 with respect to control genotype; one-way ANOVA followed by Tukey’s multiple comparison post hoc test, A; two-way ANOVA followed by Tukey’s multiple comparison post hoc test, B; Student’s t test, D, E, and G).

annexin V binding, we co-incubated DKO platelets with cyclosporine A, and we monitored annexin V binding following 6-hr incubation in vitro (Figure S4A). Cyclosporine A did not alter the time-dependent increase in annexin V positivity. We next consid-

888 Cell Reports 20, 881–894, July 25, 2017

ered the intrinsic pathway. Following 6-hr incubation in vitro, DKO platelets demonstrated no change in BCL-XL protein expression (Figure S4B). No change in sensitivity toward the BH3 mimetic Abt-737 was observed, as measured by annexin

Control DKO

Control C DKO D

**



p1

p1

G

G

F

E Control + Glucose Control + Glucose + Pyruvate + Glutamate DKO + Glucose DKO + Glucose + Pyruvate + Glutamate

#

****

1 Hour

AMPK

$

pACC

$

ACC

Annexin V Positive

H

C Control DKO D Counts (K/dL)

Annexin V Geo. MFI

pAMPK (T173) p

$

Annexin V Negative

Control DKO

C

#

6 Hour

G

on

**** ****

TMRM Geo. MFI

% Annexin V Positive Platelets

****

tro l+ G C lu + ont Py ro cos e ru l + va G te lu + cos D G e KO lu ta + m at G lu e co D se + KO Py + ru G va lu te co + se G lu ta m at e

D

Hours

bβ -F IT C

-F IT C

Hours

To ta l

****

Ne g

****

at ive

*

Annexin V Geo. MFI

*

C Control + Glucose Control + Glucose + Pyruvate + Glutamate C DKO + Glucose D DKO + Glucose + Pyruvate + Glutamate D

C

% Annexin V Positive Platelets

B

Po sit ive

% Gp1bβ-FITC Positive Platelets

A

*

#

#

*** Oligomycin (µM)

Vehicle

Oligomycin (1mg/kg)

Figure 6. DKO Platelets Demonstrate Increased Clearance (A) Platelets were labeled with anti-Gp1bb-FITC-labeled antibody, then monitored for percentage CD41-positive platelets (n = 6). (B) Whole blood was collected at 72 hr post-anti-Gp1bb-FITC injection, then monitored for annexin V Geo. MFI. (n = 6). (C) Washed platelets incubated in 5 mM glucose media ± 1 mM glutamate and 1 mM pyruvate at 37 C with 5% CO2 were monitored for annexin V exposure (n = 3). (D) Platelet annexin V binding following 6-hr incubation in specified media at 37 C with 5% CO2 (n = 20). (E) Mitochondrial potential of platelets stratified for annexin V positivity, following 6-hr incubation (n = 3). (F) Western blot analysis of platelet proteins following 6-hr incubation (n = 6). (G) Annexin V binding in platelets incubated for 1 hr in vitro with oligomycin. (H) Mice were injected with 1 mg/kg oligomycin or vehicle and monitored for platelet count following 6 hr (n = 5). Data are mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 with respect to control genotype; #p < 0.05 with respect to treatment, same genotype, $p < 0.05 with respect to annexin V-negative platelets of equivalent genotype and treatment; two-way ANOVA followed by Bonferroni multiple comparison post hoc test, A–E and H).

V binding or caspase-3/7 activity (Figures S4C and S4D). DKO platelets co-treated with the caspase-3 inhibitor Z-DEVD-FMK demonstrated no change in annexin V binding after 6-hr incuba-

tion (Figure S4E). Western blot analysis of lysates from DKO platelets incubated in media, containing glucose as the sole substrate, did not display caspase-3 activation measured by the

Cell Reports 20, 881–894, July 25, 2017 889

formation of a 15- and 17-kDa cleaved caspase-3 band (Figure 7A). Thus, mPTP opening or activation of the intrinsic apoptosis pathway does not appear to facilitate annexin V binding. Immunoblots revealed a band 3 kDa smaller than procaspase-3 (Figure 7A), which is calpain dependent (Wolf et al., 1999). Calpain-mediated platelet death has not previously been described; however, in other cell types, necrosis is facilitated through calpain activation. Because calpains are Ca2+dependent proteases, we investigated cytoplasmic Ca2+ concentrations. Following 1-hr incubation, cytoplasmic Ca2+ as well as stored Ca2+ concentrations were unchanged in control and DKO platelets (Figure 7B). However, following 6-hr incubation, DKO platelets incubated in vitro in glucose alone demonstrated increased basal cytoplasmic Ca2+ levels, despite decreased stored Ca2+ (Figure 7C), in annexin V-negative platelets. Supplementation of the media with glutamate and pyruvate significantly restored cytoplasmic as well as stored Ca2+ levels, indicating metabolic regulation of platelet Ca2+. After 6-hr incubation in vitro, calpain activity was significantly increased in DKO platelets (Figure 7D). Immunoblot analysis of filamin A, a known calpain substrate, revealed increased cleavage compared to controls (Figure 7E). Co-administration of calpeptin to DKO platelets decreased filamin A cleavage (Figure 7E). Interestingly, the calpain inhibitor calpeptin significantly decreased annexin V binding in a dose-dependent manner in DKO platelets when co-incubated in vitro for 6 hr (Figure 7F). In vivo treatment of DKO mice with calpeptin significantly prolonged platelet half-life, relative to vehicle-treated DKO mice (monitored 55 hr post-Dylight 488-labeled anti-GP1bb antibody pulse) (Figures 7G and S5). Consistent with previous reports (Bonomini et al., 2004; Wolf et al., 1999), calpeptin did not reduce agonistor ionophore-mediated annexin V exposure (Figure S6), indicating calpeptin acts via a mechanism that is independent of TMEM16f. Thus, DKO platelets undergo a necrosis-like death, which is Ca2+ and calpain dependent. DISCUSSION Several in vitro studies revealed a relationship between glucose metabolism and platelet function (Paterson et al., 1974; Tang et al., 2011; Yamagishi et al., 2001); however, the specific roles of glucose transport and utilization in platelet function in vitro and in vivo are incompletely understood. Our study demonstrates critical roles for glucose metabolism in platelet production, circulating half-life, activation, degranulation, and thrombus formation. Moreover, we demonstrate that reduced energy metabolism leads to increased annexin V exposure and platelet clearance through a calpain-dependent mechanism. RNA sequencing (RNA-seq) analyses of platelets from mice and humans indicate that GLUT1 (Slc2a1) and GLUT3 (Slc2a3) are the only class I glucose transporters expressed in platelets (Rowley et al., 2011; Shi et al., 2014). Our glucose uptake data indicate that GLUT1 and GLUT3 exhibit overlapping as well as unique functions. Surprisingly, deletion of GLUT1 alone did not alter glucose uptake, indicating that GLUT3 may compensate for GLUT1, possibly by increased plasma membrane GLUT3 localization or by increased GLUT3 activity. In addition, deletion

890 Cell Reports 20, 881–894, July 25, 2017

of GLUT3 alone only slightly decreased basal glucose uptake, but it completely abolished agonist-mediated glucose uptake. This supports previous studies indicating that GLUT3 is primarily located on a-granule membranes and translocates to the plasma membrane when platelets degranulate (Heijnen et al., 1997). The complete loss of glucose uptake in DKO platelets confirms that GLUT1 and GLUT3 represent the biologically relevant glucose transporters in platelets. In the absence of glucose availability, platelets increase mitochondrial respiration by utilizing alternative mitochondrial substrates. This metabolic plasticity appears to be due to qualitative adaptations in mitochondrial function, as evidenced by unchanged mitochondrial number in the face of increased mitochondrial membrane potential and respiration. It is possible that increased mitochondrial respiration and membrane potential increased ATP production. When glutamate and pyruvate were supplemented in the media, AMPK activation was reduced, although not completely normalized. Thus, alternative mitochondrial substrates in vivo, such as glutamate, lactate, or fatty acids, could support platelet metabolism in DKO mice. Even if true, freshly isolated DKO platelets displayed increased AMPK activation relative to controls. Furthermore, in vivo platelet counts were decreased and thrombus formation was impaired, indicating that mitochondrial metabolism was insufficient to maintain platelet energy requirements of DKO mice in vivo. GPIIbIIIa activation, degranulation, and PS exposure to the outer leaflet of the plasma membrane were decreased in DKO platelets. These three markers of platelet activation are facilitated by agonist-mediated activation of IP3 receptors, leading to storeoperated Ca2+ mobilization that increases cytoplasmic Ca2+. Although thrombin and convulxin activate two distinct PLC isoforms, dual-agonist stimulation failed to increase Ca2+ flux in DKO platelets. Furthermore, DKO platelets exhibited decreased store-operated Ca2+ entry, which also contributes to decreased agonist-mediated Ca2+ mobilization. We also demonstrated that treatment with a Ca2+ ionophore restored annexin V binding to control levels, indicating that the reduction in TMEM16f-mediated PS translocation in DKO platelets is secondary to reduced Ca2+ availability. However, in DKO platelets, Ca2+ ionophore treatment, which restored Ca2+ signaling, did not increase GPIIbIIIa activation or platelet degranulation. Agonist-mediated platelet activation was partially rescued with the addition of mitochondrial substrates, underscoring the energy dependence of these processes. The energetic mechanisms downstream of Ca2+ signaling that lead to GPIIbIIIa activation were not identified. One possibility for impaired integrin activation could be impaired actin remodeling, which serves as a scaffold for integrin activation. Reduced activation of proteins required for degranulation, such as the ATPase N-ethylmaleimide-sensitive factor, could represent an energy-dependent mechanism for platelet degranulation following Ca2+ restoration. SNAP-23 is a known regulator of degranulation and is a known calpain substrate (Rutledge and Whiteheart, 2002). It is possible that basal calpain activation results in SNAP-23 cleavage, impairing the ability of DKO platelets to degranulate. Although we do not fully understand the mechanism by which glucose regulates platelet activation at these multiple nodes in the activation process, our data unambiguously

7

Cleaved Caspase 3

Ponceau S

Basal

1µM Ionomycin

Control

D

E

Hours Incubation 1

Calpain Activity Normalized to Protein

Calpeptin 10µM

Filamin A

**

Full Length Cleaved

Ponceau S Control

DKO

-

**** **** Basal

#

#

1µM Ionomycin

DKO

6

6

1 6

6

-

+

-

+

-

F

C Control DKO D % Annexin V Positive Platelets

Caspase 3 (Full Length)

%Fluo-4 Geo. MFI Normalized to Control

%Fluo-4 Geo. MFI Normalized to Control

AB

6 Hour ****

+ D KO

C

1 Hour

T73

G lu

G lu + tro l C on

B

co

co

A

se

se

Control + Glucose Control + Glucose + Pyruvate + Glutamate DKO + Glucose DKO + Glucose + Pyruvate + Glutamate

Calpeptin (µM) 0

**** #

**** #

1

10

0

1

10

G % Gp1bβ-FITC Positive Platelets

****

in pt

+ C

al

KO D

KO

+

D

al C l+

tro C

on

pe

in pe

pt

Ve h l+ tro on C

Ve h

*** ***

Figure 7. Calpain Activation Regulates DKO Platelet Annexin V Binding and Clearance In Vivo (A) Western blot analysis of caspase-3 (black arrow indicates calpain cleavage product). (B and C) Annexin V-negative platelets loaded with Fluo-4 in media with 1 mM EGTA were analyzed for basal cytoplasmic Ca2+ concentration and following stimulation with 1 mM ionomycin at 1 hr (B) and 6 hr post-isolation (C), marked by Fluo-4 Geo. MFI (n = 6). (D) Calpain activity of platelets incubated in glucose media for 6 hr. (E) Western blot analysis of Filamin A cleavage (n = 3). (F) Platelets were incubated in the presence of vehicle (Veh) or calpeptin for 6 hr, then analyzed for annexin V binding (n = 6). (G) Mice were injected with calpeptin (1 mg/kg) daily for 7 days prior to injection with anti-Gp1bb-FITC-labeled antibody, and calpeptin administration was continued daily until the cessation of half-life experiments. Blood was assayed at 55 hr post-injection (n = 6). Data are mean ± SEM (**p < 0.01, ***p < 0.001, and ****p < 0.0001 with respect to control genotype; #p < 0.05 with respect to treatment, same genotype; Student’s t test, D; one-way ANOVA followed by Tukey’s multiple comparison post hoc test, G; two-way ANOVA followed by Bonferroni multiple comparison post hoc test, B, C, and F).

Cell Reports 20, 881–894, July 25, 2017 891

demonstrate that glucose metabolism is essential for platelet activation and thrombus formation. DKO mice developed thrombocytopenia through decreased platelet production and increased platelet clearance. Although megakaryocyte density in the bone marrow was unchanged, the ability of bone marrow-derived megakaryocytes to take up glucose was almost completely abolished, and their ability to produce platelets was reduced 2-fold, demonstrating an important relationship between megakaryocyte glucose metabolism and pro-platelet formation. Mechanisms regulating platelet clearance are not well understood. Here we demonstrate that energy metabolism plays an essential role in platelet annexin V binding and circulating half-life. Unexpectedly, we found that lack of glucose metabolism in vitro led to a rapid translocation of PS to the outer leaflet of the plasma membrane. This PS exposure appears to be energy dependent because the addition of mitochondrial substrates to the media normalized AMPK activation and partially rescued this effect, while the addition of oligomycin increased annexin V binding. Indeed, in vivo we demonstrate that, as the glycolysis-deficient platelets age, they increase annexin V binding, suggesting that, although the inhibition of metabolism might not be initially detrimental to platelets, over time this loss of energy metabolism increases PS exposure to the outer leaflet of the plasma membrane that increases clearance. Together these data indicate that energetically stressed platelets expose PS to the outer leaflet of the plasma membrane. In vivo, in the absence of glycolysis, platelet counts were significantly reduced, and the inhibition of mitochondrial function further reduced platelet counts to zero. These findings extend our understanding of the regulation of platelet circulating halflife. Although we cannot rule out additional defects in DKO platelets acquired during biogenesis, we excluded perturbed desialylation and gross perturbations in organelle number and structure, with the exception of a-granules. Taken together, we have identified a critical role for platelet metabolism in the regulation of circulating platelet half-life. This metabolism-mediated clearance is partially dependent on increased cytoplasmic Ca2+ concentrations in parallel with depleted platelet Ca2+ stores, which activates calpain and increases annexin V exposure. In addition, following agonist-mediated Ca2+ flux, DKO platelets were unable to pump Ca2+ out of the cytosol. Therefore, we believe that cytoplasmic Ca2+ accumulates in DKO platelets, because of reduced ATP availability for SERCA-mediated pumping of Ca2+ into storage organelles and/or inhibition of the PMCA, which pumps Ca2+ into the extracellular space (Varga-Szabo et al., 2009). This energy-dependent regulation of cellular Ca2+ homeostasis mirrors similar mechanisms that are implicated in necrosis (Jackson and Schoenwaelder, 2010), but it is distinct from the procoagulant necrotic platelet response, because calpeptin, although inhibiting spontaneous annexin V exposure, did not inhibit agonist- or ionophore-mediated annexin V exposure. Importantly, calpeptin did not influence the circulating half-life of control platelets. Thus, energy-Ca2+-calpain-mediated clearance may only occur under specific circumstances, such as energetic stress. Therefore, we believe this energy-Ca2+-calpain-mediated

892 Cell Reports 20, 881–894, July 25, 2017

platelet clearance represents a mechanism by which platelets can undergo necrosis and be cleared from the circulation. In conclusion, these studies reveal an essential role for glucose metabolism in regulating multiple facets of platelet function, including platelet activation, thrombosis, platelet production, and clearance from the circulation. Elucidating the fundamental roles of glucose metabolism in platelets provides the conceptual framework to better understand how the extracellular milieu could potentially alter platelet function in metabolic disorders such as diabetes, where dysfunctional Ca2+ signaling is a hallmark of platelet dysfunction. EXPERIMENTAL PROCEDURES Animals All mice were on a C57BL/6 background and were housed under standard conditions of temperature and lighting. Pf4 Cre transgenic mice were obtained from the Jackson Laboratories. GLUT1-floxed mice were generated as previously described (Young et al., 2011). GLUT3-floxed mice were obtained from the trans-NIH Knockout Mouse Project repository, Slc2a3tm1c(KOMP)Mbp. LoxP sites were inserted flanking Slc2a3 exon 7, and primers CCAACTTAAACA CAATTGCCTGGTG and GGCTCACAATTACCCATAATGA were used for PCR identification of the GLUT3-floxed allele. GLUT3-KO, GLUT1-KO, and DKO mice were generated by crossing mice with respective homozygous floxed alleles to Pf4 Cre transgenic mice. Experiments were conducted on male mice between the ages of 8 and 14 weeks, unless otherwise noted. The investigators conformed to the Guide for the Care and Use of Laboratory Animals published by the NIH and animal studies were approved by the institutional animal care and use committees (IACUC) of the University of Iowa and the University of Utah. Platelet Isolations Whole blood was isolated from isoflurane-anesthetized mice through carotid artery cannulation into 1:20 acid-citrate-dextrose (ACD), as previously described (Rowley et al., 2011). When noted, platelets were incubated with Ter119- and CD45-labeled microbeads (Miltenyi Biotec) and negatively depleted of red blood cells and leukocytes. Following isolation, platelets were allowed to recover for 30 min prior to experimental manipulation. Platelet counts were determined by Cellometer Auto M10 (Nexcelom Bioscience). Immunoblots Proteins were isolated by methanol-chloroform precipitation. Proteins for GLUT1 and GLUT3 western blots were obtained from CD45 and Ter119 bead-depleted platelets by lysis using radioimmunoprecipitation assay (RIPA) buffer. Analysis was normalized to ponceau S staining unless otherwise noted. Protein concentrations were determined by bicinchoninic acid (BCA) analysis. Detailed antibody information can be found in the Supplemental Experimental Procedures. Glucose Metabolism Platelet and megakaryocyte glucose uptake analyses were conducted with 10 mM 3H-2-Deoxy-D-Glucose. 13C-glucose flux was determined through incubations of platelets with 25 mM 1,6-13C-glucose for up to 1 hr; 13C-lactic acid was quantified with gas chromatography-mass spectroscopy. Glycogen content was determined in 2 3 108 CD45 and Ter119 bead-depleted platelets using a Glycogen Assay Kit (Cayman Chemical). Detailed protocols can be found in the Supplemental Experimental Procedures. Mitochondrial Assays Platelet bioenergetics were evaluated by Seahorse XF24 Analyzer (Agilent Technologies) as previously described (Fink et al., 2012). Leukocytes and red blood cells were depleted using Terr119 and CD45 micro beads, and data were normalized to platelet counts. Mitochondrial potential was determined by flow cytometric analysis of tetramethylrhodamine methyl ester (TMRM)-stained platelets. AMP, ADP, and ATP estimation was determined

using an AMP-Glo kit (Promega). Detailed protocols can be found in the Supplemental Experimental Procedures. Transmission Electron Microscopy Washed platelets were incubated in DMEM ± 250 mM Par4 peptide for 10 min at room temperature, then fixed with equal volumes 4% glutaraldehyde. Following 30-min glutaraldehyde incubation, platelets were centrifuged at 120 3 g for 10 min, resuspended in 4% glutaraldehyde, and processed as previously described (Schwertz et al., 2010). Organelle quantification was performed by an investigator who was blinded to genotype or treatment condition. Platelet Activation Freshly washed platelets were incubated in 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) Tyrode’s (HT) buffer with 5 mM glucose ± 1 mM pyruvate and 2 mM glutamate for 30 min. Platelets were then treated with agonists at the indicated concentrations in the presence of JonA-PE (phycoerythrin), CD62p-fluorescein isothiocyanate (FITC) (Emfret Analytics), and CD41-antigen-presenting cell (APC) (eBioscience) for 10 min at 37 C. Annexin V binding was determined in a similar manner, except staining was conducted for 15 min. Samples were analyzed using flow cytometry LSR II (Becton Dickinson) gating for CD41-positive events. Detailed protocols can be found in the Supplemental Experimental Procedures. Platelet Ca2+ Content Fluo-4 or Fura-2 was used to determine Ca2+ signaling. Detailed methods for Ca2+ assays can be found in the Supplemental Experimental Procedures. In Vivo Thrombosis Detailed methods for in vivo thrombosis assays can be found in the Supplemental Experimental Procedures. Platelet Regeneration and Half-Life Baseline platelets counts were determined by flow cytometry counting of diluted whole blood, gated for CD41-APC-positive events and normalized to flow-count fluorospheres (BD Biosciences). Mice were injected intravenously (i.v.) with 2 mg/g anti-GP1ba antibody for depletion or anti-GP1bb-FITC antibody (Emfret Analytics) for circulation half-life. Platelet counts were obtained daily. Megakaryocyte Studies Bone marrow was flushed, filtered, and cultured in DMEM with 5 mM glucose, glutamate, and recombinant thrombopoietin (TPO) for 5 days. Megakaryocytes were then cultured overnight on fibrinogen-coated chamber slides for experiments. In a blinded manner, megakaryocyte-producing platelets were quantified and normalized to total megakaryocytes. Megakaryocyte density in femurs and spleens was determined by immunohistochemistry for von Willebrand factor (vWF). Detailed protocols can be found in the Supplemental Experimental Procedures. Oligomycin Treatments Freshly isolated washed platelets were incubated in 5 mM glucose DMEM at 37 C for 1 hr in the presence of oligomycin at the concentrations specified. After 45 min, samples were stained with annexin V APC for an additional 15 min, diluted 1:20, and analyzed immediately using flow cytometry. In vivo oligomycin administration was performed by injecting 100 mL total volume into mice i.p. with 1 mg/kg oligomycin in 7% DMSO:saline. Platelet counts were determined 6 hr later. Animal weight was determined prior to injection and 24 hr postinjection. Calpain Activity and Inhibition Calpain activity was determined in platelet lysates incubated in 5 mM glucose DMEM for 6 hr. After 6 hr, platelets were centrifuged at 13,000 3 g for 5 min, then lysed in assay buffer and analyzed for calpain activity (ab65308, Abcam), which was normalized to protein content. In vitro, calpeptin was co-incubated with platelets incubated in 5 mM glucose DMEM for 6 hr and analyzed for annexin V binding. In vivo, calpeptin was administered i.p. at 7 mg/kg in 5%

DMSO:saline (100 mL final volume) for 7 days. Mice were then pulsed with anti-GP1bb-FITC antibody and calpeptin administration continued. Blood was analyzed daily. Statistical Analysis Statistical analyses were performed using GraphPad Prism 7 or Microsoft Excel 2011. Data are presented as mean ± SEM. The statistical significance threshold of p < 0.05 was utilized. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, six figures, and one table and can be found with this article online at http:// dx.doi.org/10.1016/j.celrep.2017.06.083. AUTHOR CONTRIBUTIONS Conceptualization, E.D.A., A.S.W., and T.P.F.; Methodology, E.D.A., A.S.W., T.P.F., and R.A.C.; Investigation, T.P.F., R.A.C., T.F., E.B.A., E.A.M., and N.D.; Writing – Original Draft, T.P.F.; Writing – Review & Editing, T.P.F., E.D.A., A.S.W., and R.A.C.; Funding Acquisition, E.D.A. and A.S.W.; Resources, E.D.A. and A.S.W.; Supervision, D.C., E.D.A., and A.S.W. ACKNOWLEDGMENTS We would like to thank Jamie Soto for Seahorse analyses performed at the University of Iowa metabolic phenotyping core and Katherine Walters for transmission electron microscopy analysis of platelets. The data presented herein were obtained at the Flow Cytometry Facility, which is a Carver College of Medicine core research facility at the University of Iowa. GLUT3-floxed mice were generated by the trans-NIH Knockout Mouse Project (KOMP) and obtained from the KOMP Repository (https://www.komp.org). Targeted embryonic stem cells (ESCs) were generated by NIH grants to Velocigene at Regeneron Inc. (U01HG004085) and the CSD Consortium (U01HG004080), which are archived and distributed by the KOMP Repository at UC Davis and CHORI (U42RR024244). R.A.C. was funded by NIH grant T32 DK 007115. D.C. was funded by NIH grant R00HL124070. A.S.W. was funded by NIH grant R01 HL 126547. This work was supported by NIH grant U54 HL112311 to A.S.W. and E.D.A., who are both established investigators of the American Heart Association. Received: February 8, 2017 Revised: May 8, 2017 Accepted: June 27, 2017 Published: July 25, 2017 REFERENCES Akkerman, J.W.N., and Holmsen, H. (1981). Interrelationships among platelet responses: studies on the burst in proton liberation, lactate production, and oxygen uptake during platelet aggregation and Ca2+ secretion. Blood 57, 956–966. Akkerman, J.W.N., Holmsen, H., and Driver, H.A. (1979). Platelet aggregation and Ca2+ secretion are independent of simultaneous ATP production. FEBS Lett. 100, 286–290. Alonzo, M.T., Lacuesta, T.L., Dimaano, E.M., Kurosu, T., Suarez, L.A., Mapua, C.A., Akeda, Y., Matias, R.R., Kuter, D.J., Nagata, S., et al. (2012). Platelet apoptosis and apoptotic platelet clearance by macrophages in secondary dengue virus infections. J. Infect. Dis. 205, 1321–1329. Bonomini, M., Dottori, S., Amoroso, L., Arduini, A., and Sirolli, V. (2004). Increased platelet phosphatidylserine exposure and caspase activation in chronic uremia. J. Thromb. Haemost. 2, 1275–1281. Craik, J.D., Stewart, M., and Cheeseman, C.I. (1995). GLUT-3 (brain-type) glucose transporter polypeptides in human blood platelets. Thromb. Res. 79, 461–469.

Cell Reports 20, 881–894, July 25, 2017 893

Fink, B.D., Herlein, J.A., O’Malley, Y., and Sivitz, W.I. (2012). Endothelial cell and platelet bioenergetics: effect of glucose and nutrient composition. PLoS ONE 7, e39430. Flaumenhaft, R., Furie, B., and Furie, B.C. (1999). Alpha-granule secretion from alpha-toxin permeabilized, MgATP-exposed platelets is induced independently by H+ and Ca2+. J. Cell. Physiol. 179, 1–10. Garcia-Souza, L.F., and Oliveira, M.F. (2014). Mitochondria: biological roles in platelet physiology and pathology. Int. J. Biochem. Cell Biol. 50, 156–160. Gray, L.R., Tompkins, S.C., and Taylor, E.B. (2014). Regulation of pyruvate metabolism and human disease. Cell. Mol. Life Sci. 71, 2577–2604. Heijnen, H.F.G., Oorschot, V., Sixma, J.J., Slot, J.W., and James, D.E. (1997). Thrombin stimulates glucose transport in human platelets via the translocation of the glucose transporter GLUT-3 from alpha-granules to the cell surface. J. Cell Biol. 138, 323–330. Jackson, S.P., and Schoenwaelder, S.M. (2010). Procoagulant platelets: are they necrotic? Blood 116, 2011–2018. Jobe, S.M., Wilson, K.M., Leo, L., Raimondi, A., Molkentin, J.D., Lentz, S.R., and Di Paola, J. (2008). Critical role for the mitochondrial permeability transition pore and cyclophilin D in platelet activation and thrombosis. Blood 111, 1257–1265. Karpatkin, S. (1967). Studies on human platelet glycolysis. Effect of glucose, cyanide, insulin, citrate, and agglutination and contraction on platelet glycolysis. J. Clin. Invest. 46, 409–417. Karpatkin, S., and Langer, R.M. (1968). Biochemical energetics of simulated platelet plug formation. Effect of thrombin, adenosine diphosphate, and epinephrine on intra- and extracellular adenine nucleotide kinetics. J. Clin. Invest. 47, 2158–2168. Li, Y., Woo, V., and Bose, R. (2001). Platelet hyperactivity and abnormal Ca(2+) homeostasis in diabetes mellitus. Am. J. Physiol. Heart Circ. Physiol. 280, H1480–H1489. Li, J., van der Wal, D.E., Zhu, G., Xu, M., Yougbare, I., Ma, L., Vadasz, B., Carrim, N., Grozovsky, R., Ruan, M., et al. (2015). Desialylation is a mechanism of Fc-independent platelet clearance and a therapeutic target in immune thrombocytopenia. Nat. Commun. 6, 7737. Mason, K.D., Carpinelli, M.R., Fletcher, J.I., Collinge, J.E., Hilton, A.A., Ellis, S., Kelly, P.N., Ekert, P.G., Metcalf, D., Roberts, A.W., et al. (2007). Programmed anuclear cell death delimits platelet life span. Cell 128, 1173–1186. Paterson, R.A., Heath, H., and Cranfield, T. (1974). The effect of O-(beta-hydroxyethyl) rutoside on platelet intermediary metabolism in normal and streptozotocin diabetic rats. Biochem. Pharmacol. 23, 1591–1597. Rowley, J.W., Oler, A.J., Tolley, N.D., Hunter, B.N., Low, E.N., Nix, D.A., Yost, C.C., Zimmerman, G.A., and Weyrich, A.S. (2011). Genome-wide RNA-seq analysis of human and mouse platelet transcriptomes. Blood 118, e101–e111.

894 Cell Reports 20, 881–894, July 25, 2017

Rutledge, T.W., and Whiteheart, S.W. (2002). SNAP-23 is a target for calpain cleavage in activated platelets. J. Biol. Chem. 277, 37009–37015. Schoenwaelder, S.M., Yuan, Y., Josefsson, E.C., White, M.J., Yao, Y., Mason, K.D., O’Reilly, L.A., Henley, K.J., Ono, A., Hsiao, S., et al. (2009). Two distinct pathways regulate platelet phosphatidylserine exposure and procoagulant function. Blood 114, 663–666. Schwertz, H., Ko¨ster, S., Kahr, W.H., Michetti, N., Kraemer, B.F., Weitz, D.A., Blaylock, R.C., Kraiss, L.W., Greinacher, A., Zimmerman, G.A., and Weyrich, A.S. (2010). Anucleate platelets generate progeny. Blood 115, 3801–3809. Scott, R.B. (1967). Activation of glycogen phosphorylase in blood platelets. Blood 30, 321–330. Shi, D.S., Smith, M.C., Campbell, R.A., Zimmerman, P.W., Franks, Z.B., Kraemer, B.F., Machlus, K.R., Ling, J., Kamba, P., Schwertz, H., et al. (2014). Proteasome function is required for platelet production. J. Clin. Invest. 124, 3757–3766. Tang, W.H., Stitham, J., Gleim, S., Di Febbo, C., Porreca, E., Fava, C., Tacconelli, S., Capone, M., Evangelista, V., Levantesi, G., et al. (2011). Glucose and collagen regulate human platelet activity through aldose reductase induction of thromboxane. J. Clin. Invest. 121, 4462–4476. TeSlaa, T., and Teitell, M.A. (2014). Techniques to monitor glycolysis. Methods Enzymol. 542, 91–114. van Kruchten, R., Mattheij, N.J.A., Saunders, C., Feijge, M.A.H., Swieringa, F., Wolfs, J.L.N., Collins, P.W., Heemskerk, J.W.M., and Bevers, E.M. (2013). Both TMEM16F-dependent and TMEM16F-independent pathways contribute to phosphatidylserine exposure in platelet apoptosis and platelet activation. Blood 121, 1850–1857. Varga-Szabo, D., Braun, A., and Nieswandt, B. (2009). Calcium signaling in platelets. J. Thromb. Haemost. 7, 1057–1066. Warshaw, A.L., Laster, L., and Shulman, N.R. (1966). The stimulation by thrombin of glucose oxidation in human platelets. J. Clin. Invest. 45, 1923– 1934. Wolf, B.B., Goldstein, J.C., Stennicke, H.R., Beere, H., Amarante-Mendes, G.P., Salvesen, G.S., and Green, D.R. (1999). Calpain functions in a caspase-independent manner to promote apoptosis-like events during platelet activation. Blood 94, 1683–1692. Yamagishi, S.I., Edelstein, D., Du, X.L., and Brownlee, M. (2001). Hyperglycemia potentiates collagen-induced platelet activation through mitochondrial superoxide overproduction. Diabetes 50, 1491–1494. Young, C.D., Lewis, A.S., Rudolph, M.C., Ruehle, M.D., Jackman, M.R., Yun, U.J., Ilkun, O., Pereira, R., Abel, E.D., and Anderson, S.M. (2011). Modulation of glucose transporter 1 (GLUT1) expression levels alters mouse mammary tumor cell growth in vitro and in vivo. PLoS ONE 6, e23205.