Metabolic signatures of amyotrophic lateral sclerosis ...

Metabolic signatures of amyotrophic lateral sclerosis reveal insights into disease pathogenesis James C. Dodgea,1, Christopher M. Treleavena, Jonathan A. Fidlera, Thomas J. Tamsetta, Channa Baoa, Michelle Searlesa, Tatyana V. Taksira, Kuma Misraa, Richard L. Sidmanb,1, Seng H. Chenga, and Lamya S. Shihabuddina a

Genzyme, a Sanofi Company, Framingham, MA 01701-9322; and bBeth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA 02215

Contributed by Richard L. Sidman, May 11, 2013 (sent for review March 6, 2013)

Metabolic dysfunction is an important modulator of disease course in amyotrophic lateral sclerosis (ALS). We report here that a familial mouse model (transgenic mice over-expressing the G93A mutation of the Cu/Zn superoxide dismutase 1 gene) of ALS enters a progressive state of acidosis that is associated with several metabolic (hormonal) alternations that favor lipolysis. Extensive investigation of the major determinants of H+ concentration (i.e., the strong ion difference and the strong ion gap) suggests that acidosis is also due in part to the presence of an unknown anion. Consistent with a compensatory response to avert pathological acidosis, ALS mice harbor increased accumulation of glycogen in CNS and visceral tissues. The altered glycogen is associated with fluctuations in lysosomal and neutral α-glucosidase activities. Disease-related changes in glycogen, glucose, and α-glucosidase activity are also found in spinal cord tissue samples of autopsied patients with ALS. Collectively, these data provide insights into the pathogenesis of ALS as well as potential targets for drug development.

|

carbonic anhydrase electrolyte motor neuron disease

| glucagon | glycoproteins |

A

myotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disorder characterized by the selective loss of motor neurons in the central nervous system (CNS). Approximately 10% of all familial cases of ALS are due to mutations in the gene encoding cytosolic copper–zinc superoxide dismutase (SOD1) (1). Similar to patients with ALS, transgenic mice that express the mutant human SOD1 protein (e.g., SOD1G93A mice) display progressive motor neuron degeneration, muscular atrophy, and a shortened lifespan (2). Although the exact mechanisms and pathological processes responsible for the initiation and progression of motor neuron degeneration remain largely unknown (3), it is becoming increasingly clear that the disorder’s basis is multimodal (4). Acidosis is a feature of several neuropathological conditions including ischemia, epilepsy, Parkinson disease, and multiple sclerosis (5–7). In ALS, an acidic environment has the potential to catalyze the onset and/or progression of several disease features including diminished glutamate reuptake (8), mitochondrial vacuolization (9), glial cell activation (10), endoplasmic reticulum stress (11), and impaired oxidative phosphorylation/ATP synthesis (12). Acidosis also activates acid-sensing ion channels (ASICs), which, when excessively stimulated, promote intracellular Ca2+ influx, neuroinflammation, axonal degeneration, demyelination, and neuronal death—features of several mouse disease models (6, 7), including ALS. Interestingly, gene expression profiling studies showed significant up-regulation (2- to 10-fold) of ASIC2 (also known as amiloride-sensitive cation channel neuronal 1; ACCN1) and ASIC3 (ACCN3) in laser-captured motor neurons from patients with ALS harboring mutations in SOD1 (13), suggesting that acidosis occurs in patients with ALS. In agreement, significant upregulation of ASIC2 expression was recently noted in the spinal cords of patients with sporadic ALS and SOD1G93A mice (14). Metabolic acid-base disorders are classically identified by measuring the concentrations of plasma or serum electrolytes that affect H+ concentration. A strong ion difference (SID), the summed ion concentrations of all strong base minus all strong acid, is a major determinant of H+ concentration. In healthy mammals, SID has a positive value of about 40 mEq/L (15). The apparent 10812–10817 | PNAS | June 25, 2013 | vol. 110 | no. 26

SID (SIDapp), which does not take into account the role of weak acids in determining SID, is calculated as [Na+] + [K+] + [Mg2+] + [Ca2+] − [Cl−] − [lactate]. A more complex Eq. (1,000 × 2.46 × 10−11 × pCO2/(10−pH) + [albumin] × (0.12 × pH − 0.631) + [phosphate] × (0.309 × pH − 0.469), the effective SID (SIDeff), incorporates the contribution of weak acids derived from carbon dioxide (CO2), proteins (albumin), and phosphate. The difference between SIDapp and SIDeff is called the strong ion gap (SIG), and represents the contribution of unmeasured (i.e., unidentified) ions to the SID. SIG values are reportedly a strong predictor of mortality in individuals exposed to pathological acidosis (16), with the unidentified ions perhaps passing from liver into blood (17). It is not known if disease-related changes in SIDapp, SIDeff, or SIG manifest during development of ALS. Pathological acidosis can be averted through compensatory mechanisms such as (i) lowering acid production (e.g., downregulating glycogenolysis and glycolysis to inhibit lactate synthesis from glycogen), (ii) promoting acid elimination through increased respiration and renal excretion, and (iii) resynthesis of acids back into substrate stores (18). Interestingly, gene expression studies on laser-captured motor neurons from patients with ALS showed reduced mRNA levels of α-glucosidase, an enzyme that degrades glycogen to glucose (13). Similar studies on lasercaptured astrocytes from SOD1G93A mice also demonstrated more than twofold increases in glycogen synthase mRNA (19). Therefore, we have now investigated whether SOD1G93A mice undergo metabolic changes favoring development of acidosis and whether they display disease-related pH changes in the CNS. Further, we examined whether the disease course affects the concentration of ions responsible for the SIDapp, SIDeff, and SIG, and whether SOD1G93A mice display compensatory increased glycogen or modifications in α-glucosidase activity to avert acidosis. We also examined human ALS autopsied cervical spinal cord for changes in glycogen content and α-glucosidase activity to determine whether the mouse data are relevant for humans. Results Mice Display Progressive CNS Acidosis as ALS Advances. Before de-

termining whether pH changes occur as a function of disease progression in ALS mice, we first measured pH in male wild-type hybrid (B6SJL) mice (n = 12). Within the CNS, significant (P < 0.001) regional variations in pH were observed with the spinal cord being the most acidic (Fig. 1A). To determine whether there were disease-related changes in CNS pH, measurements were made when the mice (n = 11 per disease phase) were at symptom onset (SYMP), end stage (ES), and moribund (MB). No significant differences in pH were observed between WT and SYMP SOD1G93A mice (Fig. 1 B–F). However, at the ES stage, significantly lower pH values (P < 0.0001) were observed in the right

Author contributions: J.C.D. designed research; J.C.D., C.M.T., J.A.F., T.J.T., C.B., M.S., T.V.T., and K.M. performed research; J.C.D., R.L.S., S.H.C., and L.S.S. analyzed data; and J.C.D., R.L.S., S.H.C., and L.S.S. wrote the paper. The authors declare no conflict of interest. 1

To whom correspondence may be addressed. E-mail: [email protected] or [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1308421110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1308421110

d

6.5

* 6.5

6.0 5.5

C

6.0

Blood CSF

CTX

BS

WT

SC

CTX

D

7.0

SYMP

ES

BS 7.5

*

7.0

pH

pH

6.8 6.6 6.4

MB

*

6.5

6.2 6.0

6.0

WT

SYMP

ES

MB

SC

E

WT

F

SYMP

ES

MB

Blood 8.0

6.4

7.5

6.0

*

5.8

*

pH

pH

6.2

*

7.0

*

6.5

5.6

6.0

WT

SYMP

ES

MB

WT

SYMP

ES

MB

Fig. 1. pH varies significantly as a function of sample matrix (blood, CSF, and CNS parenchyma) and ALS disease course. (A) pH measurements in the blood, CSF, motor cortex (CTX), brainstem (BS) (motor trigeminal nucleus), and lumbar spinal cord (SC) of WT mice. Letters not connected by the same letter differ significantly from each other (P < 0.0001). pH measurements in ALS mice as a function of disease state vs. WT littermate controls in (B) CSF, (C) cortex, (D) brainstem, (E) spinal cord, and (F) blood. Significantly different from WT (*P < 0.0001). Error bars represent the ±SEM.

lateral ventricle, brainstem, and lumbar spinal cord, as well as in the blood (Fig. 1 B–E). No significant differences in pH were observed between WT and ES SOD1G93A within the motor cortex. MB SOD1G93A mice displayed even lower pH values (P < 0.01) in the right lateral ventricle, brainstem, and blood, compared with ES SOD1G93A mice (Fig. 1 B–F). Measurements in lumbar spinal cord were comparable in ES and MB SOD1G93A mice (Fig. 1E). Our data on increasing acidosis within the lumbar spinal cord agree with recent findings in C57BL/6 SOD1G93A mice by an alternative pH measuring method (14). Thus, as the mouse ALS disease advanced, the CNS developed progressively greater acidosis. Impact of Disease Course in ALS Mice on Major Determinants of H+ Concentration. Blood samples were collected from male WT and

ALS (at SYMP and ES) mice at 10:00 AM and 6:00 PM (n = 6 per group per time point) and their serum chemistries (concentration of Na+, K+, Mg2+, Ca2+, Cl−, lactate, CO2, albumin, and PO4−3 ions) were analyzed (Table S1). Blood from ES mice (i.e., mice that displayed significant declines in CNS pH) had significant reductions in K+, Mg2+, and Ca2+ cations at 10:00 AM (P < 0.01) compared with WT mice. No significant changes in cation levels were observed in ALS mice at 6:00 PM except for Ca2+, which was significantly lower (P < 0.01) in both SYMP and ES mice. The concentrations of Cl– and lactate anions were also impacted during the course of the disease, with significant reductions (P < 0.01) in both Cl– and lactate levels. Reductions in Cl− were limited to SYMP mice at 6:00 PM, whereas reductions in lactate were observed in both SYMP and ES mice at 10:00 AM. Blood levels of CO2, PO4−3 and albumin (ions used to calculate the SIDeff) in ALS mice at 10:00 AM were not affected during the disease course. However, significant reductions in albumin Dodge et al.

Carbonic Anhydrase Inhibition Accelerates Mouse ALS. Carbonic anhydrase catalyzes the conversion of CO2 and H2O to HCO3− and its inhibition leads to elevated CO2 levels (acidosis). Cats and rabbits given acetazolamide (a carbonic anhydrase inhibitor) show significant reductions in pH within the brain (20, 21). Therefore, we examined the impact of acetezolamide on ALS mice. ALS mice (n = 10) administered acetazolamide (beginning on day 60) displayed a more rapid decline in hindlimb grip strength (P < 0.01), an earlier onset of paralysis (114 vs. 120 d; P < 0.01), and a trend toward accelerated death (122 vs. 134.5 d) vs. control siblings treated with vehicle (Fig. 2 B–D). Grip strength (Fig. 2B) and lifespan were unaffected in WT mice treated with acetazolamide. These findings indicate that promotion of acidosis in ALS mice accelerates disease progression. Furthermore, prescribing carbonic anhydrase inhibitors to treat hypertension,

A

60

WT SYMP ES

50 40

*

30 20

*

10

B 80

Grip strength SOD1 SOD1 ATZ

60 40

WT WT ATZ

**

20

0

0

-10

SIDapp

C

SIDeff

SIG

SOD1 SOD1 ATZ

80 60 40 20 0

80 100 120 Age (Days)

D

Onset of Paralysis 100

60

Survival 100

SOD1 SOD1 ATZ

80 60 40 20 0

0

90

100 120 Age (Days)

140

0 100

110

130 150 Age (Days)

170

Fig. 2. Impact of ALS on the major determinants of [H+] and the rate of disease progression in ALS mice following administration of acetazolamide (pH lowering agent). (A) Calculated apparent strong ion difference (SIDapp), effective strong ion difference (SIDeff), and strong ion gap (SIG) in ALS mice as function of disease state vs. WT littermate controls. (B) Hindlimb grip strength, (C) median end stage (partial paralysis in one limb) age, and (D) median survival age following daily treatment with acetazolamide (ATZ, ∼30 mg/kg). Significantly different from WT (*P < 0.01). Error bars represent ±SEM.

PNAS | June 25, 2013 | vol. 110 | no. 26 | 10813

NEUROSCIENCE

*

7.0

Force(g)

pH

b c

Percent survival

b

7.0

(P < 0.01) were observed in SYMP and ES mice at 6:00 PM. Alterations in serum ion levels are limited (in some cases) to specific circadian time points in ALS mice for reasons that are not well understood, perhaps attributable to various factors including potential disease-related differences in metabolic activity at the sampled time points (e.g., 10:00 AM = postprandial per rest phase; 6:00 PM = preprandial per onset of activity phase). Ion concentrations were then used to calculate SIDapp, SIDeff, and SIG in serum by the equations given in the Introduction. In the first reported measurements in mice, we found that WT mice display a SIDapp value of 43.02 ± 1.73, and similar values were found in SYMP (44.26 ± 1.65) and ES (41.60 ± 1.70) ALS mice (Fig. 2A), consistent with previous measurements in normal mammals (13). Interestingly, calculations of SIDeff, which takes into account the additional contribution of weak acids, showed significant (P < 0.01) differences between WT and ES mice (46.46 ± 1.85 vs. 30.03 ± 1.87; Fig. 2A). The SIG was then calculated by subtracting SIDeff from SIDapp. A SIG value greater than zero indicates presence of unexplained anions, whereas a SIG of less than zero suggests presence of unexplained cations. We found that ES mice displayed a positive SIG value compared with their WT counterparts (11.57 ± 0.22 vs. −3.44 ± 0.16; Fig. 2A). Hence, careful examination of pH and SIG may represent a unique approach to track disease progression in patients with ALS.

mEq/L

a

7.5

CSF 7.5

Percent of mice without partial paralysis in one limb

B 8.0

pH

A

glaucoma, or migraine headaches in patients with ALS may be contraindicated. ALS Mice Display Aberrant Glycogen Increases in CNS and Viscera.

Pathological acidosis is typically averted by decreasing acid synthesis and/or enhancing acid elimination (18). Therefore, we determined whether ALS mice harbored changes in glucose and glycogen levels interpretable as responses counteracting pathological acidosis. Glucose levels in the CNS of WT mice (n = 6) showed significant regional variation (P < 0.0001), with levels in cerebral cortex (CTX) > midbrain (MB) > cerebellum (CB) = brainstem (BS) > lumbar spinal cord (SC) (Fig. 3A). Similar regional variations in glucose levels were observed in the CNS of ALS mice (n = 6 per disease phase) (Fig. 3A). In contrast, glycogen levels were comparable throughout the CNS of WT and ALS mice (Fig. 3A). In cerebral cortex, although the glucose levels in ALS mice were lower (P < 0.05), the glycogen levels in WT and ALS mice were similar (Fig. 3A), as they were in midbrain and cerebellum. However, in brainstem and lumbar spinal cord, ALS mice displayed significantly higher levels of glycogen compared with WT (P < 0.0001) (Fig. 3A). Notably, the timing of abnormal glycogen accumulation in the CNS of ALS mice was coincident with significant declines in pH (shown above). Indeed, linear regression analysis showed that disease-related changes in pH and glycogen were significantly (P < 0.004) correlated (r2 = 0.7714, Fig. S1). As increased excretion of acids from the CNS into the periphery might occur (protecting the CNS from aberrant carbohydrate metabolism), we also measured glucose and glycogen levels in liver, muscle, and kidney of ALS mice at two circadian times (10:00 AM and 6:00 PM; n = 6 per disease phase per time point). In liver and muscle, circadian variations in glucose and glycogen were observed, with carbohydrate levels higher at 10:00 AM than at 6:00 PM (Fig. 4 A and B). Glucose levels in liver were comparable between ALS and WT mice at all sampled

time points; however, glycogen levels were four- to fivefold higher in ALS mice at 10:00 AM (P < 0.001) (Fig. 4A). Interestingly, increased glycogen was observed in SYMP mice, which was before detection of disease-related changes in pH. In muscle, significant reductions in glucose and concomitant elevations in glycogen were also observed in SYMP and ES mice. At 6:00 PM, glucose levels were significantly lower (P < 0.05) and glycogen levels higher (P < 0.05) in both SYMP and ES mice than in their WT counterparts (Fig. 4B). In kidney, significantly (P < 0.01) higher levels of glucose and significantly (P < 0.01) lower levels of glycogen were detected in ALS mice, perhaps reflecting that more glucose and lactate were excreted into urine with less being stored as glycogen (Fig. 4C). Collectively, our findings showed that changes in glucose and glycogen levels occurred in peripheral tissues before any overt signs of disease. Patients with ALS Harbor Increased Glycogen in the Spinal Cord. To determine whether carbohydrate levels are also modified in humans with ALS, we examined cervical spinal tissue samples collected at autopsy from male patients with ALS (n = 6) and aged matched male controls (n = 6). For patient donor information please see Supporting Information (Table S2). Glucose levels were significantly (P < 0.01) elevated in tissue homogenates of gray matter (GM) from ALS vs. control individuals (Fig. 3B), although levels in ventral white matter (VWM) homogenates were comparable in ALS and control samples. Similar to what was observed in SOD1G93A mice, glycogen levels were significantly elevated in both GM (P < 0.01) and VWM (P < 0.001) ALS vs. control tissue homogenates (Fig. 3B). To determine which cell types harbored increased glycogen, periodic acid-Schiff (PAS) staining on paraffinembedded cervical spinal cord sections showed increased glycogen in residual GM neurons and glia (Fig. 3C) and in VWM glia (Fig. 3D). Consistent with our biochemical findings, PAS staining appeared to be both stronger and more widespread in ALS vs. control

Fig. 3. Tissue levels of glucose and glycogen are affected by ALS disease course in mice and humans. (A) Glucose and glycogen levels as a function of CNS region (CTX, cerebral cortex; MB, midbrain; CB, cerebellum; BS, brainstem; and SC, lumbar spinal cord) in ALS mice as function of disease (SYMP, symptomatic stage; and ES, end stage vs. WT littermate controls. Significantly different from WT (*P < 0.05, **P < 0.0001). Bars with different letter are significantly different from each other (P < 0.0001). (B) Glucose and glycogen levels in human ALS and control (CTRL) cervical spinal cord tissue homogenates (GM, gray matter; VWM, ventral white matter). Significantly different from CTRL (*P < 0.001). Error bars represent ±SEM. (C and D) Paraffin-embedded, PAS-stained cervical spinal sections of human ALS and CTRL donors (C, GM; D, VWM; arrowheads point to glycogen storage; all images are at 20× cells except for panels at 60× in the Lower Right corner in C and D. (Scale bars, 50 μm.)

10814 | www.pnas.org/cgi/doi/10.1073/pnas.1308421110

Dodge et al.

500 250

0

0 10 am

6 pm

Kidney 800

**

200

600

150

** **

400 200

100 50

0

0 10 am

Glycogen (ug/g tissue)

Glucose (ug/g tissue)

C

6 pm

100

WT GLC SYMP GLC ES GLC WT GLY SYMP GLY ES GLY

6 pm

Fig. 4. Glucose (GLC) and glycogen (GLY) levels in peripheral tissues are affected by ALS disease course. GLC and GLY levels measured in (A) liver, (B) muscle, and (C) kidney of ALS mice as a function of disease phase vs. WT littermate controls at two circadian times (SYMP, symptomatic; ES, end stage). Significantly different from WT (*P < 0.05, **P < 0.01, ***P < 0.001). Error bars represent ±SEM.

tissue (Fig. 3 C and D). Collectively, these findings suggest that glycogen accumulation in spinal cord is a common feature of ALS. ALS Mice Show Changes in α-Glucosidase Activity. Reduced α-glucosidase mRNA levels have been reported in laser-captured motor neurons from spinal cords of patients with ALS (13). Deficiency in lysosomal acid α-glucosidase activity leads to Pompe disease, a “neuromuscular” lysosomal storage disease characterized by aberrant glycogen accumulation in both viscera and CNS (22, 23). Neutral (endoplasmic reticulum, ER) α-glucosidase trims glucose residues from N-glycans of glycoproteins, which is part of an elaborate and important quality-control system that promotes the proper folding and oligomerization of newly formed proteins and their transport from the ER to their final destination (24). We examined acidic and neutral α-glucosidase enzyme activities in spinal cord, muscle, liver, and serum as a function of disease progression in ALS (n = 5 per disease phase) vs. WT (n = 5) mice. All samples were collected at 10:00 AM. No significant changes in enzyme activity were observed between WT and ALS mice in serum, but in spinal cord, a significant (P < 0.01) reduction in acid α-glucosidase activity was found in SYMP and ES vs. WT mice (Fig. 5A). In contrast, neutral enzyme activity was significantly (P < 0.001) elevated more than twofold in ES (i.e., in mice, at the onset of paralysis) vs. WT counterparts (Fig. 5A). In muscle and liver, both acidic and neutral α-glucosidase activities were significantly (P < 0.001) elevated two- to eightfold (Fig. 5 B and C) in ES mice vs. WT counterparts. Interestingly, enzyme activity levels varied across different tissues, and were significantly greater in liver (P < 0.001) and muscle (P < 0.01) than in spinal cord (Fig. 5D). Neutral α-Glucosidase Activity Is Elevated in Patients with ALS. Our

findings of altered α-glucosidase enzyme activity in SOD1 mice prompted us to examine acidic and neutral enzyme activities in GM and VWM tissue homogenates from male patients with ALS (n = 6) vs. age-matched male controls (n = 6). In contrast to our findings in ALS mice, we observed a trend toward an elevation in acidic α-glucosidase activity (Fig. 5 E and F), suggesting that a reduction in acidic α-glucosidase activity may be limited to the SOD1G93A model. Alternatively, disease-related changes in enzyme activity might be species specific. However, consistent with our observations in SOD1G93A mice, a significant elevation in neutral α-glucosidase activity was noted in human ALS cervical spinal cord GM samples (Fig. 5E), and not in the VWM (Fig. 5F). Importantly, linear regression analysis showed that glucose levels in spinal cord GM correlated significantly (P < 0.001) with neutral (r2 = 0.6477) and acidic (r2 = 0.6147) glucosidase enzyme G93A

Dodge et al.

Discussion We show here that the SOD1G93A mouse model of ALS enters a state of progressive acidosis associated with manifold metabolic alterations present also in autopsied human ALS spinal cord samples. Several lines of evidence indicate that acidosis is not merely a consequence of disease, but is a significant modulator of disease progression. For example, in a controlled clinical trial, the intake of branched chain amino acids (exogenous acids) led to accelerated loss of pulmonary function in patients with ALS (26). Also, a faster rate of decline in muscle strength was observed in patients with ALS treated with Topiramate (Apotex Corp.), a known inhibitor of carbonic anhydrase (a crucial enzyme that buffers pH and helps maintain acid-base balance) (27). Consistent with these observations, we showed that motor functional decline and onset of paralysis were accelerated in ALS mice upon administration of acetazolamide, another carbonic anhydrase inhibitor. The carbonic acid (H2CO3)–bicarbonate (HCO3−) system (CO2 + H20↔H2CO3↔H+ + HCO3−) is central to maintaining acid-base balance through its ability to provide rapid adjustments to changing levels of volatile [partial carbon dioxide tension (pCO2)] and nonvolatile acids (hydrochloric, sulfuric, lactic, etc.).

A

B

Spinal Cord 400 300 200 150 125 100 75 50 25 0

*

*

*

pH 4.3

C 200

*

100 50

E

pH 7.0

Human Spinal Cord - GM 400 300

CTRL ALS

*

200 100 0

pH 4.3

pH 7.0

* *

400 200

D

WT SYMP * ES

pH 4.3

600

0

150

0

WT SYMP ES

800

pH 7.0

Liver 250

Muscle 1000

WT SYMP ES

pH 4.3

pH 7.0

Relative Enzyme Activity 500 400

spinal cord muscle b liver

b

300

c

200 100 0

F

a

a

pH 4.3

a pH 7.0

Human Spinal Cord - VWM 300

CTRL ALS

200 100 0

pH 4.3

pH 7.0

Fig. 5. α-Glucosidase activity is affected by ALS disease course in mice and patients. Lysosomal (pH 4.3) and neutral (pH 7.0) α-glucosidase activities were measured in ALS mice as a function of disease phase vs. WT littermate controls (SYMP, symptomatic; ES, end stage) in (A) spinal cord, (B) muscle, and (C) liver. Significantly different from WT (*P < 0.001). (D) Relative enzyme activity across different tissues (in WT mice) graphed relative to spinal cord levels. Bars with different letters are significantly different from each other (P < 0.01). (E and F) Lysosomal (acidic) and neutral α-glucosidase activities in human ALS and control (CTRL) cervical spinal cord tissue homogenates (GM, gray matter; VWM, ventral white matter). Significantly different from CTRL (*P < 0.01). Error bars represent ±SEM.

PNAS | June 25, 2013 | vol. 110 | no. 26 | 10815

NEUROSCIENCE

10 am

**

alpha glucosidase activity % of WT

0

1000 750

200

alpha glucosidase activity % of Spinal Cord

0

1250

**

300

alpha glucosidase activity % of CTRL

20000

5000

*

400

activity. In VWM, a significant (P < 0.001) correlation was also observed between glycogen levels and neutral (r2 = 0.4142) and acidic (r2 = 0.8172) glucosidase activities (Fig. S2). Recently developed human cell culture models of ALS (25) could be used to further elucidate the relationship between glucose, glycogen, and glucosidase enzyme activity within specific cell types in gray and white matter.


10000

1500


40000

Muscle

alpha glucosidase activity % of CTRL

15000

B500

Glycogen(µg/g tissue)

60000

Glucose (µg/g tissue)

Liver

******

20000

Glycogen(µg/g tissue)

Glucose (µg/g tissue)

A

The relationship between volatile (i.e., respiratory) and nonvolatile (i.e., metabolic) acids is best understood through an extension of the Henderson–Hasselbalch equation (pH = pK × log [HCO3−/ (0.03 × pCO2)]) (28). Either an increase in pCO2 or a drop in [HCO3−] will trigger a decrease in pH. Hyperventilation is a compensatory response initiated by nonvolatile acid accumulation, because it decreases pCO2, which in turn promotes the dissociation of H2CO3 into CO2 and H20. As H2CO3 dissociates (and CO2 is expelled by the lungs) H+ removal (alkalization) is promoted as H+ combines with HCO3−. Given these processes, it is not surprising that an increase in breathing frequency, tidal volume, and CO2 production was found to occur in SOD1G93A mice several weeks before death (29). Sustained hyperventilation leading to CO2 depletion within the CNS, however, will initiate vasoconstriction, reduced blood flow, ischemia, and edema (30)—all occurring in the spinal cord of SOD1G93A mice before motor neuron degeneration (31, 32). Pathological acidosis in ALS could be attributed to a number of events arising as the disease progresses. These include metabolic abnormalities, increased muscle catabolism, impaired respiratory function (although the latter does not occur until a few days before death in SOD1G93A mice (29), whereas we observed acidosis ∼20 d before death) and the accumulation of an as yet undetermined anion. Indeed, we showed that SOD1G93A mice present several changes indicative of altered lipid metabolism (e.g., reduced circulating triglyceride levels, reduced fat mass, increased circulating levels of the lipolytic hormone glucagon, and decreased levels of the antilipolytic hormone insulin), which, when prolonged, can trigger acidosis (Fig. S3). Interestingly, the changes in glucagon and insulin preceded reductions in fat mass and the development of pathological acidosis. We also recently reported that SOD1G93A mice displayed pathological elevations in the lipolytic hormone, corticosterone, which significantly correlated with accelerated disease progression (33). The reason why increases in lipid metabolism manifest during ALS remains unclear. During ALS progression the CNS may become more dependent on alternative fuels (lipids, ketones, amino acids) to generate energy when glucose flux through the glycolytic pathway is diminished (34). Our finding of an increased SIG also suggests that an unknown anion may be responsible, in part, for the development of acidosis in ALS mice. We showed that at the ES stage, the SIG value approaches 12 mEq/L, which is particularly interesting given that SIG values greater than 10 are reportedly the strongest predictor of mortality in individuals exposed to pathological acidosis (16). A variety of compensatory mechanisms that either lower acid production, promote acid elimination, or both may help to avert pathological acidosis (18). Glucose utilization (35) and monocarboxylate transporter 1 (MCT1)-mediated lactate uptake (36) within the CNS are inhibited by acidosis. Indeed, glucose uptake and/or levels within the CNS are significantly diminished in patients with ALS (37), SOD1G93A mice (32, 38), and in WT mice exposed to cerebrospinal fluid (CSF) from patients with ALS (39). Moreover, lactate levels are reduced in spinal cords of SOD1G93A mice (19), as is the expression of MCT1 (36). Our results extend these findings. Concomitant with the reductions in pH, significant accumulation of glycogen was observed in both the CNS and visceral organs of SOD1G93A mice. Within the ALS mouse CNS, reductions in pH were highly correlated (r2 = 0.7714) with glycogen accumulation. Importantly, spinal cord glycogen accumulation also appears to be a prominent feature of human ALS. Specifically, we found elevated glycogen in gray and ventral white matter samples from autopsied patients with ALS, confirmed by histological analyses showing elevated glycogen in both neurons and glia. The increased glycogen in the CNS of SOD1G93A mice and humans with ALS is consistent with spinal cord gene expression profiling studies from ALS mice and humans that showed elevated glycogen synthase mRNA levels and altered α-glucosidase mRNA (13, 19). Indeed, we show here that lysosomal (glycogen degrading) α-glucosidase activity is significantly reduced within the lumbar 10816 | www.pnas.org/cgi/doi/10.1073/pnas.1308421110

spinal cord of SOD1G93A mice in advance of the onset of hindlimb paralysis. However, lower acidic α-glucosidase activity was not observed in the spinal cord samples from patients with ALS, so this feature may be limited to mutant SOD1-related ALS. Nevertheless, therapeutic possibilities are suggested, given that rapamycin, a drug recently reported to lower glycogen levels in α-glucosidase knockout mice (40) (a model of Pompe disease) was also shown to accelerate disease in SOD1G93A mice (41). Accelerated disease progression in ALS mice following treatment that reduces glycogen levels is consistent with the data indicating that glycogen increase slows the disease course. We also measured the activity of neutral α-glucosidase, an ER resident enzyme that plays a critical role in the processing of N-glycans attached to glycoproteins to ensure proper folding. Removal of glucose residues by neutral α-glucosidase generates a monoglycosylated epitope that is recognized by lectin chaperones (i.e., calnexin and calreticulin) to maintain the protein within the ER until proper folding is achieved. Removal of the last glucose residue leads to the misfolded protein being transported to the proteasome for degradation (24). Enzyme activity was elevated more than twofold in the spinal cords of ES mice, as well as being significantly elevated in spinal cords of patients with ALS. We speculate that the continuous trimming of glucose residues from misfolded proteins by neutral α-glucosidase (to promote protein refolding) may also contribute to aberrant glucose homeostasis (and potentially to glucose neurotoxicity) (42) within the spinal cord. Indeed, our measured significant glucose accumulation within spinal cords of patients with ALS significantly correlated (r2 = 0.6477) with α-glucosidase enzyme activity. Changes in glycogen content were also observed in peripheral tissues of ALS mice, perhaps due in part to glucose and/or lactate export from the CNS, serving to protect the CNS from pathological acidosis and neurotoxicity. We were surprised to find elevated acidic (lysosomal) α-glucosidase activity in peripheral tissues and speculate that lysosomal α-glucosidase activity may increase in peripheral tissues when glycogen levels in the CNS become pathogenic, at which point glycogen must be metabolized to glucose and excreted. In agreement with this hypothesis, we observed that SOD1G93A mice had significantly increased glucose levels in kidneys (despite normal serum levels) and had 2.2–8.6fold increases in α-glucosidase activities in liver and muscle of ES vs. WT mice. We noted also that basal neutral α-glucosidase activity in WT mice is 1.7- and 4-fold greater in muscle and liver, respectively, vs. spinal cord. Both the basal- and disease-induced differences in neutral α-glucosidase activity within muscle and liver may explain in part why misfolded SOD1 protein was detected at significantly higher levels in CNS than in peripheral tissues in mouse ALS (43, 44). Our point of emphasis is that this mouse ALS model displays significant glycogen increases in both CNS and peripheral tissues. Developing drugs to alleviate acidosis and slow progression of ALS will be challenging. Although we also found significant upregulation of ASIC mRNA in the spinal cords of SOD1G93A ALS mice, treatment with amiloride, an ASIC blocker that can lessen disease severity in rodent models of multiple sclerosis, Parkinson disease, and ischemia (diseases also featuring acidosis) (5–7), failed to slow disease progression in the ALS mice model (Fig. S4). This may reflect the fact that several neuropathological features catalyzed by acidosis also occur independently of ASIC activation. Given our current findings, it is also no surprise that assisted ventilation (e.g., with bilevel positive airway pressure, BiPAP) has been linked to extended survival rates in patients with ALS (45, 46). Clearly, assisted ventilation would be most helpful during periods of acidosis. In summary, our data showed that SOD1G93A mice and humans with ALS share several metabolic abnormalities indicative of increased lipolysis and pathological acidosis. The disease also affected serum concentrations of ions used to calculate the SIDapp, SIDeff, and SIG, strongly suggesting that development of pathological acidosis was likely due in part to the presence of an unidentified anion. Furthermore, SOD1G93A mice display metabolic changes indicative of a compensatory response that may help Dodge et al.

Materials and Methods Transgenic male ALS mice that strongly express the mutant SOD1G93A transgene were divided into equivalent groups. Mutant SOD1G93A gene copy number and protein expression were confirmed by PCR and Western blot analysis, respectively. Animals were housed under light:dark (12:12 h) cycles and provided with food and water ad libitum. At 75 d of age, food pellets were place on the

1. Rosen DR (1993) Mutations in Cu/Zn superoxide dismutase gene are associated with familial amyotrophic lateral sclerosis. Nature 364(6435):362. 2. Gurney ME, et al. (1994) Motor neuron degeneration in mice that express a human Cu,Zn superoxide dismutase mutation. Science 264(5166):1772–1775. 3. Bruijn LI, Miller TM, Cleveland DW (2004) Unraveling the mechanisms involved in motor neuron degeneration in ALS. Annu Rev Neurosci 27:723–749. 4. Teng YD, et al. (2012) Multimodal actions of neural stem cells in a mouse model of ALS: A meta-analysis. Science Transl Med 4(165):165ra164. 5. Ames A, 3rd, Maynard KI, Kaplan S (1995) Protection against CNS ischemia by temporary interruption of function-related processes of neurons. J Cereb Blood Flow Metab 15(3): 433–439. 6. Arias RL, et al. (2008) Amiloride is neuroprotective in an MPTP model of Parkinson’s disease. Neurobiol Dis 31(3):334–341. 7. Friese MA, et al. (2007) Acid-sensing ion channel-1 contributes to axonal degeneration in autoimmune inflammation of the central nervous system. Nat Med 13(12):1483–1489. 8. Swanson RA, Farrell K, Simon RP (1995) Acidosis causes failure of astrocyte glutamate uptake during hypoxia. J Cereb Blood Flow Metab 15(3):417–424. 9. Khan NA, Guevara P, Sotelo J (1989) Influence of in vitro lactic acidosis on central nervous system neurons. Pathol Biol (Paris) 37(6):725–729. 10. Morihata H, Nakamura F, Tsutada T, Kuno M (2000) Potentiation of a voltage-gated proton current in acidosis-induced swelling of rat microglia. J Neurosci 20(19):7220–7227. 11. Aoyama K, et al. (2005) Acidosis causes endoplasmic reticulum stress and caspase12-mediated astrocyte death. J Cereb Blood Flow Metab 25(3):358–370. ska M, Nelson D, Dagani F, Deas J, Silver IA (1993) Relations between in12. Erecin tracellular ions and energy metabolism under acidotic conditions: A study with nigericin in synaptosomes, neurons, and C6 glioma cells. J Neurochem 61(4):1356–1368. 13. Kirby J, et al. (2011) Phosphatase and tensin homologue/protein kinase B pathway linked to motor neuron survival in human superoxide dismutase 1-related amyotrophic lateral sclerosis. Brain 134(Pt 2):506–517. 14. Behan AT, et al. (2013) Acidotoxicity and acid-sensing ion channels contribute to motoneuron degeneration. Cell Death Differ 20(4):589–598. 15. Rinaldi S, De Gaudio AR (2005) Strong ion difference and strong anion gap: The Stewart approach to acid base disturbances. Curr Anaesth Crit Care 16(6):395–402. 16. Kaplan LJ, Kellum JA (2004) Initial pH, base deficit, lactate, anion gap, strong ion difference, and strong ion gap predict outcome from major vascular injury. Crit Care Med 32(5):1120–1124. 17. Kaplan LJ, Kellum JA (2010) Fluids, pH, ions and electrolytes. Curr Opin Crit Care 16(4): 323–331. 18. Sestoft L, Bartels PD (1983) Biochemistry and differential diagnosis of metabolic acidoses. Clin Endocrinol Metab 12(2):287–302. 19. Ferraiuolo L, et al. (2011) Dysregulation of astrocyte-motoneuron cross-talk in mutant superoxide dismutase 1-related amyotrophic lateral sclerosis. Brain 134(Pt 9):2627–2641. 20. Bickler PE, Litt L, Banville DL, Severinghaus JW (1988) Effects of acetazolamide on cerebral acid-base balance. J Appl Physiol 65(1):422–427. 21. Heuser D, Astrup J, Lassen NA, Betz BE (1975) Brain carbonic acid acidosis after acetazolamide. Acta Physiol Scand 93(3):385–390. 22. Hirshhorn R, Reuser AJJ (2001) Glycogen storage disease type II: Acid alpha-glucosidase (acid maltase) deficiency. The Metabolic and Molecular Bases of Inherited Disease, eds Scriver CA, Beaudet AL, Valle D, Sly WS (McGraw–Hill, New York), 8th Ed, pp 3389–3420. 23. Sidman RL, et al. (2008) Temporal neuropathologic and behavioral phenotype of 6neo/6neo Pompe disease mice. J Neuropathol Exp Neurol 67(8):803–818.

Dodge et al.

cage floor to facilitate access to food by weakening mice. Mice were scored as symptomatic (SYMP), showing abnormal hindlimb splay, median age 82 d; end stage (ES), with onset of limb paralysis (typically hindlimb), median age 103 d; and moribund (MB), unable to right themselves within 30 s after being placed on their backs, median age 122 d. All procedures were performed using protocols approved by Genzyme’s Institutional Animal Care and Use Committees. Please see SI Materials and Methods for further details. ACKNOWLEDGMENTS. Human tissue was obtained from the National Institute of Child Health and Human Development Brain and Tissue Bank for Developmental Disorders at the University of Maryland, Baltimore, Contact HHSN275200900011C, Ref. N01-HD-9-0011.

24. Herscovics A (1999) Importance of glycosidases in mammalian glycoprotein biosynthesis. Biochim Biophys Acta 1473(1):96–107. 25. Haidet-Phillips AM, et al. (2011) Astrocytes from familial and sporadic ALS patients are toxic to motor neurons. Nat Biotechnol 29(9):824–828. 26. Tandan R, et al. (1996) A controlled trial of amino acid therapy in amyotrophic lateral sclerosis: I. Clinical, functional, and maximum isometric torque data. Neurology 47(5): 1220–1226. 27. Cudkowicz ME, et al.; Northeast ALS Consortium (2003) A randomized, placebo-controlled trial of topiramate in amyotrophic lateral sclerosis. Neurology 61(4):456–464. 28. Chesler M (1990) The regulation and modulation of pH in the nervous system. Prog Neurobiol 34(5):401–427. 29. Tankersley CG, Haenggeli C, Rothstein JD (2007) Respiratory impairment in a mouse model of amyotrophic lateral sclerosis. J Appl Physiol 102(3):926–932. 30. Ainslie PN, Duffin J (2009) Integration of cerebrovascular CO2 reactivity and chemoreflex control of breathing: Mechanisms of regulation, measurement, and interpretation. Am J Physiol Regul Integr Comp Physiol 296(5):R1473–R1495. 31. Zhong Z, et al. (2008) ALS-causing SOD1 mutants generate vascular changes prior to motor neuron degeneration. Nat Neurosci 11(4):420–422. 32. Miyazaki K, et al. (2012) Early and progressive impairment of spinal blood flow-glucose metabolism coupling in motor neuron degeneration of ALS model mice. J Cereb Blood Flow Metab 32(3):456–467. 33. Fidler JA, et al. (2011) Disease progression in a mouse model of amyotrophic lateral sclerosis: The influence of chronic stress and corticosterone. FASEB J 25(12):4369–4377. 34. Chang Y, et al. (2008) Messenger RNA oxidation occurs early in disease pathogenesis and promotes motor neuron degeneration in ALS. PLoS ONE 3(8):e2849. 35. Newman GC, Hospod FE, Schissel SL (1991) Ischemic brain slice glucose utilization: Effects of slice thickness, acidosis, and K+. J Cereb Blood Flow Metab 11(3):398–406. 36. Uhernik AL, Tucker C, Smith JP (2011) Control of MCT1 function in cerebrovascular endothelial cells by intracellular pH. Brain Res 1376:10–22. 37. Dalakas MC, Hatazawa J, Brooks RA, Di Chiro G (1987) Lowered cerebral glucose utilization in amyotrophic lateral sclerosis. Ann Neurol 22(5):580–586. 38. Browne SE, et al. (2006) Bioenergetic abnormalities in discrete cerebral motor pathways presage spinal cord pathology in the G93A SOD1 mouse model of ALS. Neurobiol Dis 22(3):599–610. 39. Godoy JM, Skacel M, Lima JM, Andrade CM (1990) Arqu Neuro-Psiqu 48(4):473–477. 40. Ashe KM, et al. (2010) Inhibition of glycogen biosynthesis via mTORC1 suppression as an adjunct therapy for Pompe disease. Mol Genet Metab 100(4):309–315. 41. Zhang X, et al. (2011) Rapamycin treatment augments motor neuron degeneration in SOD1(G93A) mouse model of amyotrophic lateral sclerosis. Autophagy 7(4):412–425. 42. Tomlinson DR, Gardiner NJ (2008) Glucose neurotoxicity. Nat Rev Neurosci 9(1): 36–45. 43. Rakhit R, et al. (2007) An immunological epitope selective for pathological monomermisfolded SOD1 in ALS. Nat Med 13(6):754–759. 44. Zetterström P, et al. (2007) Soluble misfolded subfractions of mutant superoxide dismutase-1s are enriched in spinal cords throughout life in murine ALS models. Proc Natl Acad Sci USA 104(35):14157–14162. 45. Bach JR (1995) Amyotrophic lateral sclerosis: Predictors for prolongation of life by noninvasive respiratory aids. Arch Phys Med Rehabil 76(9):828–832. 46. Sancho J, Servera E, Bañuls P, Marín J (2010) Prolonging survival in amyotrophic lateral sclerosis: Efficacy of noninvasive ventilation and uncuffed tracheostomy tubes. Am J Phys Med Rehabil 89(5):407–411.

PNAS | June 25, 2013 | vol. 110 | no. 26 | 10817

NEUROSCIENCE

avert pathological acidosis. Our results also demonstrated that glycogen increase and fluctuations in α-glucosidase activity in the spinal cord are robust features of human ALS. Collectively, our work provides fresh insight into the pathogenesis of ALS as well as potential biomarkers and targets for drug development.

Supporting Information Dodge et al. 10.1073/pnas.1308421110 SI Results Increasing evidence indicates that alterations in energy metabolism play an important role in the pathogenesis of amyotrophic lateral sclerosis (ALS). Metabolic abnormalities in patients with ALS include development of a hypermetabolic phenotype, a gradual loss of body fat that cannot be attributed to decreased energy intake, hyperlipidemia, and a number of hormonal changes (e.g., insulin insensitivity and elevated circulating glucocorticoid and glucagon levels) that affect lipid catabolism (1–6). Similarly, transgenic mice over-expressing the G86R mutation of the Cu/Zn superoxide dismutase 1 (SOD1) gene display lower insulin levels, reduced adipose tissue mass, and an increase in peripheral lipid clearance (7, 8). Consistent with these observations, we recently reported that accelerated disease progression in SOD1G93A mice is positively correlated with circulating levels of the lipolytic hormone, corticosterone (9). Sustained triglyceride catabolism to nonesterified fatty acids followed by β-oxidation to ketone bodies (i.e., acetoacetate, β-hydroxybutyrate, and acetone) has the potential to lead to development of pathological acidosis (10). Interestingly, elevated ketone levels have been reported in both human and mouse models of ALS. For example, in early symptomatic SOD1G86R mice and humans with ALS, serum β-hydroxybutyrate levels are significantly elevated (7, 11). In patients with ALS, levels of acetone in cerebrospinal fluid (CSF) is increased more than twofold (12). Although these metabolic alterations observed during the course of disease appear to favor the development of acidosis, it remains unclear whether this aspect of the disease is pathogenic. Therefore, we determined whether SOD1G93A mice display metabolic changes (i.e., alterations indicative of increased lipolysis) that would favor the development of acidosis. EchoMRI whole body imaging was used to assess lean and fat mass in male SOD1G93A mice (n = 12) following onset and subsequent progression of the disease. Lean and fat mass values in symptomatic (SYMP, abnormal hindlimb splay; median age 75 d) ALS mice were comparable to WT littermate controls (Fig. S3 A and B). However, end stage (ES, onset of partial paralysis in one limb; median age 103 d) and moribund (MB, unable to right themselves within 30 s after being placed on their back; median age 122 d) ALS mice showed significant (P < 0.0001) reductions in both lean and fat mass (Fig. S3 A and B). It should be noted that ES mice displayed normal food consumption (food was placed on the cage floor starting at symptom onset); therefore, similar to patients with ALS, reductions in mass cannot be attributed to reduced caloric intake. The breakdown of triglycerides into fatty acids is primarily regulated by lipolytic (e.g., glucagon and corticosterone) and antilipolytic hormones (e.g., insulin) (13). Recently, we reported that higher corticosterone levels in SOD1G93A mice were highly correlated with accelerated paralysis and earlier death (9). Importantly, disease-related changes in serum corticosterone levels were limited to specific circadian time (CT) points, suggesting that ALS mice endure episodes of transient metabolic stress. Here, we determined whether SOD1G93A mice display changes in glucagon and insulin that are consistent with increased lipid metabolism. Trunk blood was collected from male WT, SYMP, and ES mice at CT10 (10:00 AM) and CT18 (6:00 PM) (n = 8 per group per time point) for analysis. Measurements were made at CT10 because maximal elevations in corticosterone were noted at this time point in our previous study (9). Additional samples were collected at CT18 because we previously observed no difference in circulating corticosterone levels between WT and SOD1G93A mice at this time point. Similar to what had been Dodge et al. www.pnas.org/cgi/content/short/1308421110

reported in SOD1G86R mice (7), the levels of the antilipolytic hormone insulin were significantly (P < 0.05) diminished in SYMP and ES SOD1G93A mice vs. WT mice at CT10, but not at CT18 (Fig. S3D). Interestingly, as observed in patients with ALS (6), the levels of the lipolytic hormone glucagon were elevated in SYMP and ES mice compared with WT controls at CT10 (Fig. S3C). The increase in glucagon was significant (P < 0.02) in ES mice. At CT18, no difference in circulating glucagon levels was observed between ALS and WT mice. In agreement with these findings of disease-associated changes in hormonal levels that favor lipolysis, we also recorded a significant decline (P < 0.01) in serum triglyceride levels in ES mice at CT10, but not at CT18 (Fig. S3E). Similar reductions in serum triglyceride levels have been reported in SOD1G86R mice (8). Collectively, our findings indicate that SOD1G93A mice display several metabolic changes (reduced fat mass, increased corticosterone, elevated glucagon, lower insulin, and decreased triglycerides) that are consistent with the hypothesis that the animals were in a state of heightened lipolysis. These results also highlight the importance of making experimental measurements at various circadian time points to uncover potential underlying neuropathological features in the course of the ALS disease. SI Materials and Methods Body Composition and Hormone Measurements. Lean and fat mass in ALS mice (as a function of disease stage) and age-matched WT counterparts were measured with an EchoMRI-100 (EchoMRI) according to the manufacturer’s instructions. Measurements were carried out in living, nonsedated mice in less than 1 min. ELISA kits were used to measure serum glucagon (R&D Systems) and insulin (EMD Millipore) levels according to the manufacturer’s instructions. pH Measurements. In vivo pH measurements were made in anesthetized ALS (at different phases of disease) and WT mice (agematched to ES and MB ALS mice) using a pH Optica system (World Precision Instruments) according to the manufacturer’s instructions (14) Briefly, a needle type fiber optic microsensor (with a 140-μm tip size and 0.01-pH resolution) was stereotaxically implanted into various CNS regions [motor cortex, ventral thalamus, lateral ventricle (CSF), brainstem (motor trigeminal nucleus), and lumbar spinal cord] under isoflurane anesthesia. Final measurements were recorded after a 10-min equilibration period (after the sensor was inserted into the targeted region). After the measurements were made, animals were rapidly decapitated (while still anesthetized) to collect trunk blood for pH measurements. Pilot studies (in which pH was measured over a 30-min period) were carried out in WT mice to confirm that isoflurane (0.8%) exposure did not produce measurable changes in pH. Serum Chemistry Analysis. The VetACE clinical chemistry system (Alpha Wassermann) was used to measure the concentration of analytes (albumin, calcium, and phosphate) in the serum of ALS and WT mice. These spectrophotometric analyses were performed at 37 °C. Sodium, potassium, chloride, and CO2 levels in serum were determined with a Beckman CX5 clinical chemistry analyzer (Beckman Coulter), which uses indirect ion selective electrode (ISE) methodology to determine the concentration of electrolytes in biological fluids. The Beckman CX5 determines ion concentration by measuring electrolyte activity in solution. When the sample/buffer mixture contacts the electrode, ions 1 of 5

undergo an ion exchange in the hydrated outer layer of the glass electrode, and as the ion exchange takes place, a change in voltage (potential) is developed at the face of the electrode. The potential follows the Nernst equation and allows calculation of electrolyte concentration in a solution. Glucose and Glycogen Levels. Glucose and glycogen levels in CNS

tissue (cerebral cortex, midbrain, cerebellum, brainstem, and lumbar spinal cord) and peripheral tissues (liver, quadriceps muscle, and kidney) were measured as described (15). Briefly, lysates were first digested with α-amyloglucosidase (0.54 mg/mL; Sigma-Aldrich) in 25 mM potassium acetate buffer, pH 5.5. The released glucose in the test samples and the levels of endogenous glucose from undigested samples were quantitated with the Amplex red glucose/ glucose oxidase assay kit (Invitrogen). Glycogen levels were determined by subtracting the glucose levels in the undigested samples from those in the digested samples. Bovine liver glycogen (Sigma-Aldrich) was used as a reference standard. α-Glucosidase Activity Measurements. Tissue α-glucosidase activities were determined using 4-methylumbelliferone-α-D-glucopyranoside as the artificial substrate. Samples were diluted 1:10 in 100 mM sodium acetate containing 0.01% BSA at a pH of 4.3 or 7.0 and then pipetted (15 μL) in duplicate into a 96-well plate. A total of 50 μL of 4-MU-α-D-glucopyranoside substrate was then added to each well and allowed to incubate at 37 °C in the dark for 2 h. The reaction was stopped with glycine-NaOH, pH 12.5 (185 μL per well) and the fluorescence determined with a flourometer at an excitation wavelength of 360 nm and an emission wavelength of 450 nm with a cutoff at 435 nm. Periodic Acid-Schiff Staining. Tissue sections were deparaffinized, hydrated in water, and oxidized in 0.5% periodic acid solution for 5 min. Sections were then rinsed in distilled water, placed in Schiff reagent for 15 min (during which time the sections become light pink), and then washed in lukewarm tap water for 5 min (sections immediately turn dark pink). Sections were then counterstained with Mayer’s hematoxylin for 1 min, washed in tap water for 5 min, dehydrated, and coverslipped with a synthetic mounting medium (Acrytol; Leica Microsystems). Acetazolamide Administration. Constant release pellets (Innovative

Research of America) were implanted in the back of the neck of anesthetized female mice using a trochar (according to the manufacturer’s instructions). Acetazolamide (Taro Pharmaceuticals) 1. Desport JC, et al. (2001) Factors correlated with hypermetabolism in patients with amyotrophic lateral sclerosis. Am J Clin Nutr 74(3):328–334. 2. Dupuis L, et al. (2008) Dyslipidemia is a protective factor in amyotrophic lateral sclerosis. Neurology 70(13):1004–1009. 3. Kasarskis EJ, Berryman S, Vanderleest JG, Schneider AR, McClain CJ (1996) Nutritional status of patients with amyotrophic lateral sclerosis: Relation to the proximity of death. Am J Clin Nutr 63(1):130–137. 4. Reyes ET, Perurena OH, Festoff BW, Jorgensen R, Moore WV (1984) Insulin resistance in amyotrophic lateral sclerosis. J Neurol Sci 63(3):317–324. 5. Patacchioli FR, et al. (2003) Adrenal dysregulation in amyotrophic lateral sclerosis. J Endocrinol Invest 26(12):RC23–RC25. 6. Hubbard RW, et al. (1992) Elevated plasma glucagon in amyotrophic lateral sclerosis. Neurology 42(8):1532–1534. 7. Dupuis L, Oudart H, René F, Gonzalez de Aguilar JL, Loeffler JP (2004) Evidence for defective energy homeostasis in amyotrophic lateral sclerosis: Benefit of a high-energy diet in a transgenic mouse model. Proc Natl Acad Sci USA 101(30):11159–11164. 8. Fergani A, et al. (2007) Increased peripheral lipid clearance in an animal model of amyotrophic lateral sclerosis. J Lipid Res 48(7):1571–1580.

Dodge et al. www.pnas.org/cgi/content/short/1308421110

pellets were designed to deliver 30 mg/kg of drug (n = 10) or placebo (n = 10) per day over a 90-d period. Groups were sibling matched and mice were implanted with pellets at 60 d of age. RT-PCR. Cervical spinal cord tissues from B6SJL-Tg(SOD1-G93A)

mice were homogenized in Trysol reagent (Sigma-Aldrich) and total RNA was extracted with the miRNeasy Mini kit following the manufacturer’s protocol (Qiagen). Total RNA concentration was determined with a NanoDrop 1000 spectrophotometer (ThermoFisher Scientific) by absorbance at 260 nm. cDNA was synthesized with the High-Capacity cDNA Reverse Transcriptase kit (Applied Biosystems), 1.5 μg total RNA serving as template. Gene expression levels were measured by real-time RT-PCR. TaqMan Gene Expression assays (Applied Biosystems) containing specific Mus musculus (Mm) probes and primers [acid sensing ion channel 1 (ASIC1), Mm01305997; ASIC2, Mm00475691; ASIC3, Mm00805460; Mm00448968; Peptidylprolyl Isomerase A (PPIA), Mm02342430] and TaqMan Gene Expression Master mix (Applied Biosystems) were used with 50 ng of cDNA template to determine expression levels of ASIC1, ASIC2, ASIC3, and endogenous control gene Ppia. PCR reactions were performed on the Applied Biosystems 7500 Real-Time PCR system with the following thermal conditions: 50 °C for 2 min, 95 °C for 10 min, and 40 cycles of 95 °C for 15 s, and 60 °C for 1 min. Each reaction was performed in duplicate and in a 96-well format. PCR data were acquired from Sequence Detector Software (SDS version 2.0; Applied Biosystems) and quantified by a standard curve method. Human Tissue Samples. Human cervical spinal cord segments from nine patients who died of respiratory failure caused by ALS were used (Table S2). Control samples were obtained from seven agematched individuals without evidence of neurological disease. Human tissue was obtained from the National Institute of Child Health and Human Development Brain and Tissue Bank for Developmental Disorders at the University of Maryland, Baltimore, Contact HHSN275200900011C, Ref. N01-HD-9–0011. Statistics. Error bars represent ±SEM. Statistical tests were per-

formed either by an analysis of variance (ANOVA) or a two-way ANOVA followed by a Bonferroni post hoc test to find mean differences between groups (GraphPad; Prism software). Survival analysis was conducted by the Kaplan–Meier method. A value of P < 0.05 was considered statistically significant. 9. Fidler JA, et al. (2011) Disease progression in a mouse model of amyotrophic lateral sclerosis: The influence of chronic stress and corticosterone. FASEB J 25(12): 4369–4377. 10. Sestoft L, Bartels PD (1983) Biochemistry and differential diagnosis of metabolic acidoses. Clin Endocrinol Metab 12(2):287–302. 11. Kumar A, et al. (2010) Metabolomic analysis of serum by (1) H NMR spectroscopy in amyotrophic lateral sclerosis. Clin Chim Acta 411(7-8):563–567. 12. Blasco H, et al. (2010) 1H-NMR-based metabolomic profiling of CSF in early amyotrophic lateral sclerosis. PLoS ONE 5(10):e13223. 13. Jaworski K, Sarkadi-Nagy E, Duncan RE, Ahmadian M, Sul HS (2007) Regulation of triglyceride metabolism. IV. Hormonal regulation of lipolysis in adipose tissue. Am J Physiol Gastrointest Liver Physiol 293(1):G1–G4. 14. Friese MA, et al. (2007) Acid-sensing ion channel-1 contributes to axonal degeneration in autoimmune inflammation of the central nervous system. Nat Med 13(12):1483–1489. 15. Sidman RL, et al. (2008) Temporal neuropathologic and behavioral phenotype of 6neo/6neo Pompe disease mice. J Neuropathol Exp Neurol 67(8):803–818.

2 of 5

Fig. S1. Glycogen levels correlate with pH in the CNS of ALS mice. The percent reduction in pH (from WT) significantly (P < 0.004) correlated (r2 = 0.7714) with the percent increase (from WT) in glycogen accumulation observed within the CNS (cortex, midbrain, brainstem, and lumbar spinal cord) of SYMP and ES ALS mice.

Fig. S2. Glucose and glycogen levels correlate with α-glucosidase enzyme activity in human spinal cord tissue samples. (A) Neutral and (B) acidic α-glucosidase enzyme activity correlated with glucose levels in gray matter (GM) and ventral white matter (VWM) tissue samples collected from patients with ALS and agematched controls. Significant (P < 0.001) correlations were observed in GM tissue samples for both neutral and acidic enzyme activities. (C) Neutral and (D) acidic α-glucosidase enzyme activity correlated with glycogen levels in human GM and VWM tissue samples. Significant (P < 0.001) correlations were observed in VWM tissue samples for both neutral and acidic enzyme activities.


3 of 5

Fig. S3. ALS mice display metabolic alternations (muscle catabolism and increased lipolysis) that favor development of pathological acidosis. (A) Lean and (B) fat mass composition in ALS mice as a function of disease state vs. WT littermate controls. Serum levels of (C) glucagon (lipolytic hormone), (D) insulin (antilipolytic hormone), and (E) triglyceride were measured in ALS mice as a function of disease state vs. WT littermate controls at different circadian time (CT) points (SYMP, symptomatic, abnormal hindlimb splay, median age 82 d; ES, end stage, onset of limb paralysis (typically hindlimb), median age 103 d; and MB, moribund mouse unable to right itself within 30 s after being place on its back, median age 122 d.) Significantly different from WT (*P < 0.05,**P < 0.02, ***P < 0.01, ****P < 0.0001). Error bars represent ±SEM.

Fig. S4. Acid-sensing ion channel (ASIC) expression in ALS mice and the impact of amiloride treatment on disease progression. (A) ASIC3 mRNA (normalized to PPIA) measured in the spinal cords of WT (n = 8), SYMP (n = 8), and ES (n = 8) mice. Bars with different letters are significantly different from each other (P < 0.01). Error bars represent ±SEM. (B) Survival rates in ALS mice treated with amiloride pellets (n = 11) vs. ALS mice treated with placebo/control pellets (n = 11). Constant release pellets (Innovative Research of America) were implanted in the back of the neck of anesthetized mice with a trochar (according to the manufacturer’s instructions). Amiloride pellets were designed to deliver 10 mg/kg of drug per day over a 90-d period. Mice were implanted with pellets at 60 d of age. No statistical difference in median survival was observed between mice implanted with placebo vs. amiloride pellets (125 d vs. 120 d).


4 of 5

Table S1. Serum concentration of electrolytes that affect H+ levels 10:00 AM Electrolyte +

Na K+ Ca2+ Mg2+ Cl− Lactate CO2 PO4−3 Albumin SIDapp SIDeff SIG

WT 140.8 4.00 4.72 2.09 108.0 6.95 29.83 2.17 28.2 43.02 46.46 −3.44

± ± ± ± ± ± ± ± ± ± ± ±

6:00 PM

SYMP 3.59 140.2 0.14 3.82 0.02 4.68 0.07 2.22 2.02 106.2 0.44 4.12 2.54 31.8 0.15 2.28 0.97 29.8 1.73 44.26 1.85 45.42 0.56 −1.16

± ± ± ± ± ± ± ± ± ± ± ±

2.10 0.13 0.07 0.17 0.80 0.24* 0.80 0.15 0.94 1.65 1.78 0.16

ES 138.8 3.06 4.35 1.67 105.8 4.33 32.14 1.94 26.60 41.60 30.03 11.57

± ± ± ± ± ± ± ± ± ± ± ±

WT 1.74 0.43* 0.04* 0.06* 0.80 0.42* 1.96 0.16 1.96 1.70 1.87* 1.22*

136.6 ± 3.70 ± 4.75 ± 2.09 ± 110.6 ± 5.89 ± 27.85 ± 1.94 ± 29.60 ± — — —

SYMP 1.73 134.80 ± 1.49 0.02 3.94 ± 0.13 0.07 4.42 ± 0.02 0.04 1.97 ± 0.07 1.40 103.8 ± 1.14* 0.40 4.90 ± 0.25* 2.40 27.91 ± 1.8 0.87 2.13 ± 0.07 2.40 26.4 ± 0.8* — — —

ES 140.6 4.10 4.20 1.92 107.4 5.80 32.62 2.05 26.00

± 1.5 ± 0.18 ± 0.03 ± 0.08 ± 1.16 ± 0.40 ± 2.26 ± 0.07 ± 0.36* — — —

Electrolyte concentrations (expressed as mEq/L) were measured in ALS mice as a function of disease state vs. WT littermate controls at different circadian times. ES, end stage; SIDapp, apparent strong ion difference; SIDeff, effective strong ion difference; SIG, strong ion gap; SYMP, symptomatic. Significantly different from WT (*P < 0.01). Error bars represent ±SEM.

Table S2. Human tissue donor information Case no. Sex Age at death, y/d Postmortem interval 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

M M M M M M M M M M M M M M M M M

54/40 41/335 59/236 61/49 71/17 77/23 66/245 45/354 88/147 45/364 56/164 59/53 57/305 64/54 69/296 59/283 48/242


15 21 6 8 19 3 13 19 24 17 22 44 16 13 23 18 21

Tissue type

Cause of death

Fixed Fixed Fixed/frozen Fixed/frozen Fixed Fixed/frozen Frozen Frozen Frozen Fixed Frozen Frozen Frozen Frozen Frozen Frozen Frozen

ALS ALS ALS ALS ALS ALS ALS ALS ALS Cardiovascular disease Cardiovascular disease Pending Cardiovascular disease Multiple injuries Pending Heroin and alcohol intoxication Cardiovascular disease

5 of 5