Articleshttps://doi.org/10.1038/s41589-018-0132-2

Enzyme promiscuity drives branched-chain fatty acid synthesis in adipose tissuesMartina Wallace1, Courtney R. Green   1, Lindsay S. Roberts2, Yujung Michelle Lee3,4, Justin L. McCarville3, Joan Sanchez-Gurmaches5,6, Noah Meurs1, Jivani M. Gengatharan1, Justin D. Hover1, Susan A. Phillips7, Theodore P. Ciaraldi8,9, David A. Guertin10, Pedro Cabrales1, Janelle S. Ayres3, Daniel K. Nomura2, Rohit Loomba11 and Christian M. Metallo1,12,13*

1Department of Bioengineering, University of California, San Diego, La Jolla, CA, USA. 2Departments of Chemistry, Molecular and Cell Biology, and Nutritional Science and Toxicology, University of California, Berkeley, Berkeley, CA, USA. 3Nomis Center for Immunobiology and Microbial Pathogenesis, The Salk Institute for Biological Studies, La Jolla, CA, USA. 4Division of Biological Sciences, University of California at San Diego, La Jolla, CA, USA. 5Division of Endocrinology, Division of Developmental Biology, Cincinnati Children’s Hospital Research Foundation, Cincinnati, OH, USA. 6Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH, USA. 7Division of Pediatric Endocrinology, Department of Pediatrics, University of California at San Diego, La Jolla, CA, USA. 8Virginia San Diego Healthcare System, San Diego, CA, USA. 9Division of Endocrinology & Metabolism, Department of Medicine, University of California at San Diego, La Jolla, CA, USA. 10Program in Molecular Medicine, University of Massachusetts Medical School, Worcester, MA, USA. 11NAFLD Research Center, Division of Gastroenterology, Department of Medicine, University of California at San Diego, La Jolla, CA, USA. 12Moores Cancer Center, University of California, San Diego, La Jolla, CA, USA. 13Diabetes Research Center, University of California, San Diego, La Jolla, CA, USA. *e-mail: cmetallo@ucsd.edu

SUPPLEMENTARY INFORMATION

In the format provided by the authors and unedited.

NATuRE CHEMiCAL BioLoGY | www.nature.com/naturechemicalbiology

Supplementary Table 1. GC/MS fatty acid retention times. Metabolite Retention

time(mins) m/z

C15:0 16.85 256 Iso-C16:0 17.78 270 C16:0 18.73 270 Iso-C17:0 19.89 284 Anteiso-C17:0 20.35 284 C17:0 21.02 284

Supplementary Table 2. Baseline demographics and clinical characteristics of NAFLD patient cohort. Mean(range). NASH(n=8) NAFL(n=8) Gender 3 males, 5 females 5 males, 3 females Age 47 (35 – 70) 55 (31 – 69) BMI 34 (31 – 40) 31 (27 – 37) AST 48 (26 – 102) 30 (22 – 46) ALT 62(40 – 122) 45 (22 – 96) NAS (0 – 8) 5 (3-6) 3 (2- 4) Steatosis (0-3) 2(1-3) 2 (1-3) Fibrosis score (0-4) 2 (1-3) 0 (0-1) Lobular inflammation (0-3) 2 (1-2) 1 (1-2) Portal inflammation (0-2) 1 (1-2) 1 (0-1)

Supplementary Table 3. shRNA sequences.

Gene Name shRNA sequence Acyl-CoA Dehydrogenase 8 (Acad8) CCGGGCCTGGATGATTGATAGCTTTCTCGAG

AAAGCTATCAATCATCCAGGCTTTTTG

Acyl-CoA Dehydrogenase medium-chain (Acadm)

CCGGGCAAACTGCTATTGAAGCAAACTCGAGTTTGCTTCAATAGCAGTTTGCTTTTTG

Acyl-CoA Dehydrogenase short-branched-chain (Acadsb)

CCGGGCTTCTTAGTAGACCGAGATACTCGAGTATCTCGGTCTACTAAGAAGCTTTTTG

Isovaleryl-CoA Dehydrogenase (Ivd) CCGGGCTGATATCCTAGTCGTGTATCTCGAG

ATACACGACTAGGATATCAGCTTTTTG

Acyl-CoA Dehydrogenase short-chain (Acads)

CCGGCGTCAACAATTCTCTCTACTTCTCGAGAAGTAGAGAGAATTGTTGACGTTTTTG

Supplementary Table 4. CRISPR target sequences. Guide Name Forward Reverse NTC0095 CACCGGCCGTGTTGCTGGATACGCC

AAACGGCGTATCCAGCAACACGGCC

Fasn 1 CACCGTGTCTCCGAAAAGAGCCGGG

AAACCCCGGCTCTTTTCGGAGACAC

Fasn 2 CACCGTTGGTGGAGCCAATTAACAG

AAACCTGTTAATTGGCTCCACCAAC

Crat 1 CACCGTCCACAAGTGCAACTATGGG

AAACCCCATAGTTGCACTTGTGGAC

Crat 2 CACCGACCACCCACGCATATAACCG

AAACCGGTTATATGCGTGGGTGGTC

Crat 3 CACCGTTTGCTGCCAAACTCATCGA

AAACTCGATGAGTTTGGCAGCAAAC

Supplementary Table 5. Primer sequences used in gene expression analysis.

Gene Name Forward sequence Reverse Sequence Pparγ (Pparg2) TTCGCTGATGCACTGCCTAT

ACAGACTCGGCACTCAATGG

Adiponectin (Adipoq) GACACCAAAAGGGCTCAGGA

GCCCTTCAGCTCCTGTCATT

Leptin (Lep) GGGAGGAAAATGTGCTGGAGAC

AAGCCCAGGAATGAAGTCCAA

Acyl-CoA dehydrogenase 8 (Acad8)

TGGCGGAGTGGGATCAGAA

CCACATCTGTTCGCACATAGAC

Acyl-CoA dehydrogenase short-branched-chain (Acadsb)

CCCAACCTGCTTGTCTCCTTG

ATCCCTGGATCACCGATTTCT

Acyl-CoA dehydrogenase short-chain (Acads)

GACTGGCGACGGTTACACA

GGCAAAGTCACGGCATGTC

Isovaleryl-CoA dehydrogenase (Ivd)

GGACGGCGAGTTTCCAGTT

CCACAGGCAATATCGAGTGGG

Branched chain amino acid transaminase 2 (Bcat2)

CAGCCACACTAGGACAGGTCT CAGCCTTGTTATTCCACTCCAC

Branched chain keto acid dehydrogenase (Bckdh) Carnitine acetyltransferase (Crat)

CTCCTGTTGGGACGATCTGG GCTGCCAGAACCGTGGTAAA

CATTGGGCTGGATGAACTCAA CCTTGAGGTAATAGTCCAGGGA

18S rRNA AGTCCCTGCCCTTTGTACACA

CGATCCGAGGGCCTCACTA

Supplementary Table 6. GC-MS metabolite fragment ions used for analysis. Metabolite Carbons Formula m/z Acetate 1-2 C4H9O2Si 117 Isovalerate 1-5 C7H15O2Si 159 Leucine 2,3,4,5,6 C13H32NOSi2 274 Isoleucine 2,3,4,5,6 C13H32NOSi2 274 Valine 1,2,3,4,5 C13H30NO2Si2 288 Citrate 1,2,3,4,5,6 C20H39O6Si3 459 Iso-C16:0 1-16 C17H34O2 270 Iso-C17:0 1-17 C18H36O2 284 Anteiso-C17:0 1-17 C18H36O2 284 Iso-C18:0 1-18 C19H38O2 298 C15:0 1-15 C16H32O2 256 C16:0 1-16 C17H34O2 270 C17:0 1-17 C18H36O2 284 C18:0 1-18 C19H38O2 298 C16:1n7 1-16 C16H34O2 268 C18:1n9 1-18 C19H38O2 296

Supplementary Table 7. Simplified network for Isotopomer Spectral Analysis (ISA) used to determine contribution of oxPPP derived NADPH to palmitate synthesis. Contribution to Palmitate synthesis (%) Description NADPH.l (a) à NADPH (a) (NADPH containing 2H tracer label) NADPH.d (a) à NADPH (a) (unlabeled NADPH) 14*NADPH (a) à Palm.s (abcdefghijklmn) de novo lipogenesis (g(t) value) Palm.s à Palm Newly synthesized palmitate Palm.d à Palm Pre-existing (unlabeled) palmitate 0*Palm.s + 0*Palm.d à Palm.m Mixing of pools for measurement

Supplementary Table 8. Simplified network for Isotopomer Spectral Analysis (ISA) used to determine contribution of oxPPP derived NADPH to synthesis of branched chain fatty acids and odd chain fatty acids. Contribution to fatty acid synthesis (%) Description NADPH.l (a) à NADPH (a) (NADPH containing 2H tracer label) NADPH.d (a) à NADPH (a) (unlabeled NADPH) 12*NADPH (a) à FA.s (abcdefghijkl) de novo lipogenesis (g(t) value) FA.s à FA Newly synthesized FA FA.d à FA Pre-existing (unlabeled) fatty acid 0*FA.s + 0*FA.d à FA.m Mixing of pools for measurement

Supplementary Figure 1.

a b

e

OH

O

OH

O

OH

O

iso-C16:0

iso-C17:0

anteiso-C17:0

c

f

M0 M1 M2 M3 M4 M50

20

40

60

80

Labe

ling

from

[3-2 H

]glu

cose

(%)

C16:0 iso-C16:0

M0 M2 M4 M6 M8M10M12 M14M16

0

5

10

15

20

25

30

35

C16

:0 %

enr

ichm

ent

from

[U13

C6]

Glu

cose

M0 M2 M4 M6 M8M10M12 M14M16

0

5

10

1520

40

60

80

iso-

C17

:0 %

enr

ichm

ent

from

[U13

C6]

Glu

cose

M0 M2 M4 M6 M8M10M12 M14M16

02468

10

40

50

60

ante

iso-

C17

:0 %

enr

ichm

ent

from

[U13

C6]

Glu

cose

C15:0

C16:0

C17:0

iso-C

16:0

iso-C

17:0

antei

so-C

17:0

0

20

40

60

80

100

% n

ewly

syn

thes

ized

in

in 3

T3L1

ove

r 72

hour

s (+

B12

)

d

Supplementary Figure 1. mmBCFAs are de novo synthesized in 3T3L1 cell culture. a. Structures of mmBCFAs identified in 3T3L1 adipocytes. b-d. Isotopologue distribution from [U-13C6]glucose in palmitate, iso-C17:0, and anteiso-C17:0, respectively in 3T3L1 cells. e. The % of each fatty acid newly synthesized, determined using isotopomer spectral analysis (ISA) of [3-2H]glucose traced 3T3L1 adipocytes for 72 hours. f. Iso-C16:0 and palmitate labeling from [3-2H]glucose in 3T3L1 adipocytes cultured for 72 hours. All data are representative from three cell replicates and each experiment has been repeated 3 independent times with the same results obtained. Data are presented as means ±�SEM with dot plots overlaid with the exception of e, where 95% confidence intervals from ISA model are shown.

Control0.16 mM 0.5 mM 1.5 mM 4.5 mM

***

**** *

+ isovalerate

C16:0

iso C

16:0

iso C

17:0

antei

so C

17:0

0

5

10

15

fatty

aci

d co

mpo

sitio

n re

lativ

e to

con

trol

*

***

******

Control0.16 mM 0.5 mM 1.5 mM 4.5 mM

+ 2-MB5

C16:0

iso C

16:0

iso C

17:0

antei

so C

17:0

0

1

2

3

4fa

tty a

cid

com

posi

tion

rela

tive

to c

ontro

l

Control0.16 mM 0.5 mM

1.5 mM 4.5 mM

**

*

*

******

*****

********

******

*****

+ propionate

***

iso C

16:0

iso C

17:0

antei

so C

17:0

C15:0

C16:0

C17:0

0

1

2

3

4

fatty

aci

d co

mpo

sitio

n re

lativ

e to

con

trol

FASN

GAPDH

NT G1 G2

iso-C

16:0

C16:0

0

10

20

30

40

% e

nric

hmen

t fro

m

[U-13

C6]

Glu

cose

(1-M

0)

p=0.014

p=0.028Control100 nM TVB3166

Fabp4

PPAR

Adiponectin

Leptin

Bcat2

Bckdha

0.0

0.5

1.0

1.5

rela

tive

mR

NA

expr

essi

on

sgNTsgFasn 1sgFasn 2

sgNT

sgFas

n 1

sgFas

n 20

10

20

30

C16

:0 %

enr

ichm

ent f

rom

[U

-13C

6] G

luco

se (1

-M0)

N.D

a b c

d e f

gp=0.00011

p=0.00027

sgNT

sgFas

n 1

sgFas

n 2

Pre-ad

ipocytes

0

1

2

3

iso-

C16

:0%

of t

otal

fatty

aci

ds

N.D N.D

Control+ isobutyrate

h

Supplementary Figure 2.

MWLd.

250 kD

Supplementary Figure 2. mmBCFAs are de novo synthesized via FASN. a. FASN levels in pooled CRISPR/Cas9 Fasn KO 3T3L1 adipocytes (see 12a for full blot). b. % enrichment of C16:0 after 48 hours in [U-13C6]glucose in pooled CRISPR/Cas9 Fasn KO 3T3L1 adipocytes. c. Relative gene expression of adipocyte differentiation markers in Fasn KO 3T3-L1 adipocytes. d-f. Relative abundances of mmBCFAs and C16:0 in differentiated 3T3L1s (day 8 post induction) following addition of d, isovalerate, e, 2-methyl-butyrate, or f, propionate, for 24 hours. Fatty acids calculated as % total fatty acids and normalized to control conditions. Two-tailed students t-test was performed on three cellular replicates for each comparison with no adjustment for multiple comparisons. g, Iso-C16:0 levels as % total fatty acid in pooled CRISPR/Cas9 Fasn KO 3T3L1 adipocytes following addition of isobutyrate for 24 hours. h, % enrichment of fatty acids in 3T3L1 adipocytes after 48 hours in [U-13C6]glucose +/- 100 nM TVB3166. Two-tailed students t-test was performed on three cellular replicates for each comparison with no adjustment for multiple comparisons. Two-tailed students t-test was performed on three cellular replicates for each comparison with no adjustment for multiple comparisons. All data are representative from three cell replicates and each experiment has been repeated 2 independent times with the exception of d-f which have been repeated 3 independent times. Data are presented as means ±�SEM with dot plots overlaid. *p<0.05, **p<0.01, ***p<0.001.

a b c

d e f

NT1NT2

Acad8-1

Acad8-2

Acadm-1

Acadm-2

Acads-1

Acads-2

Acadsb

-1

Acadsb

-2Ivd

-1Ivd

-20

2

4

6

8

iso-

C16

:0 a

s pe

rcen

tof

tota

l FA

(%)

NT

shIvd

-1

shIvd

-20.0

0.5

1.0

1.5

Rel

ativ

e Iv

d ge

ne e

xpre

ssio

n

NT

shAca

d8-1

shAca

d8-20.0

0.5

1.0

1.5

Rel

ativ

e A

cad8

ge

ne e

xpre

ssio

n

NT

shAca

dsb-1

shAca

dsb-2

0.0

0.5

1.0

1.5

Rel

ativ

e A

cads

b ge

ne e

xpre

ssio

n

Supplementary Figure 3.

0

1

2

3

4

iso-

C17

:0 a

s pe

rcen

tof

tota

l FA

(%)

NT1NT2

Acad8-1

Acad8-2

Acadm-1

Acadm-2

Acads-1

Acads-2

Acadsb

-1

Acadsb

-2Ivd

-1Ivd

-20.0

0.5

1.0

1.5

ante

iso-

C17

:0 a

s pe

rcen

tof

tota

l FA

(%)

NT1NT2

Acad8-1

Acad8-2

Acadm-1

Acadm-2

Acads-1

Acads-2

Acadsb

-1

Acadsb

-2Ivd

-1Ivd

-2

Supplementary Figure 3. Identification of ACADs that catabolize BCKDH products in 3T3L1 cells. a-c. mRNA levels of a, Acad8, b, Ivd, c, Acadsb relative to NT shRNA in 3T3L1 adipocytes. Representative data from three cell replicates. Data are presented as means ±�SEM with dot plots overlaid. d-f. Relative mmBCFA abundance in 3T3L1 adipocytes following knockdown of Acad enzymes. Fatty acids are calculated as % total fatty acids. Representative data from six cell replicates, data presented as box (25th to 75th percentile with median line) and whiskers (min to max values). 1 or 2 refers to different hairpins. Each experiment has been repeated 3 independent times.

b c

-2 0 2 4 60

2

4

6

8

log2(FC) of M+4 ion13C Valine traced cells versus no trace

-log1

0(p

valu

e) o

f M+4

ion

13C

Val

ine

vers

us n

o tr

ace

Neutral lipids SphingolipidsPhospholipids

a

d e

f

C16 gluco

syl c

eramide

C16:0/

18:1

alkyl

PI

C16:0/

C18:1

PC

C16:0

acyl

carn

itine

C16:0/

C18:1

PI

C16:0/

C16:0/

C16:0

TAG

C16:0/

C18:1/

C16:0

TAG

C16:0/

C18:1

PS

C16:0

Sphingomyelin

C16:0/

18:1

alkyl

PC

C16:0/

18:1

alkyl

PS 0

2

4

6

8

% o

f lip

id w

ith M

+4 e

nric

hmen

t fro

m U

13C

Val

ine

Control +10µM isoproterenol

iso-C

16:0

iso-C

17:0

antei

so-C

17:0

0

200

400

600

800

1000

BC

FA a

bund

ance

in m

edia

(nM

)

p=0.0007

p=0.0005

p=0.0006

O

O

HHO

O ON+

P-O

O

C16:0 LPC

O

O

HHO

O ON+

P-O

O

iso-C16:0 LPC

Same m/z, difficult to resolve with LCMS.

12C-Valine

NH2

CH3

CH3OH

O 3T3L1 cells

13C incorporation from valine creates a m/z mass increase of +4 allowing detection of iso-C16:0

O

O

HHO

O ON+

P-O

O

C16:0 LPC

O

O

HHO

O ON+

P-O

O

iso-C16:0 LPC

[U13C5]-Valine

NH2

CH3

CH3OH

O 3T3L1 cells

Supplementary Figure 4.

M0 M1 M2 M3 M4 M5 M6 M70.0

0.5

1.0

MID

of C

16:0

No traceU13C Valine

M0 M1 M2 M3 M4 M5 M6 M70.0

0.5

1.0

MID

of i

so C

16:0

No traceU13C Valine

Supplementary Figure 4. mmBCFAs are incorporated into distinct lipid species. a. Differentiation between lipids (e.g. lysophosphatidylcholine; LPC) containing palmitate or iso-C16:0 in the acyl chain using stable isotope tracing and LC-MS analysis. b-c. Isotopologue distribution of b, C16:0, and c, iso-C16:0 from 3T3L1 adipocytes cultured in 12C or [U-13C5]valine and used for lipidomic analysis. Data from two cell replicates and presented as means ±�SEM with dot plots overlaid. d. Volcano plot depicting fold change of enrichment and p-value for the M+4 ion in [U-13C5]valine traced cells versus 12C valine cells (no isotope tracer) for 72 hours. Data from 3-4 cell replicates, p-value determined from two-tailed students t-test. e. % enrichment of the most significantly enriched lipid species from [U-13C5]valine. Data from 3-4 cell replicates and presented as box and whiskers with error bars representing min to max values. f. BCFA abundance in media of 3T3L1 cells treated with isoproterenol. Data are presented as means ±�SEM with dot plots overlaid. Two-tailed students t-test was performed on three cellular replicates with no adjustment for multiple comparisons. Each experiment in this figure has been repeated 3 independent times.

ba c

0 10 200

2

4

6

8

hours postprandial

μM

in fe

mal

e C

57B

L6J

plas

ma

iso C16:0

iso C17:0

anteiso C17:0

iso C18:0

d

Supplementary Figure 5.

gWATiW

ATBAT

Liver

Muscle

Brain

0.00

0.05

0.10

0.15

nmol

es/m

g an

teis

o C

17:0

FemaleMale

p=0.011

p=0.005

gWATiW

ATBAT

Liver

Muscle

Brain

0.0

0.1

0.2

0.3

0.4

0.5

nmol

es/m

g is

o C

18:0

FemaleMale

p=0.032

iso C

16:0

iso C

17:0

antei

so C

17:0

iso C

18:0

0

2

4

6

8

10

μM in

pla

sma

Male Femalep=0.00004

Supplementary Figure 5. Levels of mmBCFAs in vivo. a-c. The abundance of a, anteiso-C17:0, and b, iso-C18:0, in various tissues (gWAT indicates gonadal WAT), and c, plasma mmBCFA levels from 16-week-old C57BL/6J male (n=6) and female mice (n=6). Two-tailed students t-test was performed for statistical comparison with no adjustment for multiple comparison. Data presented as box (25th to 75th percentile with median line) and whiskers (min to max values). d. Plasma concentration of total hydrolyzed mmBCFAs from C57BL/6J female mice (n=8) following removal of food at time 0 (7 am) and sampled at 2hrs, 6hrs, 12hrs and 24hrs. Data presented as means ±�SEM.

Acetat

e

Propionate

Butryate

Valerat

e

Isobutyr

ate 2MB

Isova

lerate

0

4

6

8

2

nmol

e/m

g w

et w

eigh

t c

ecal

con

tent

Germ freeSPF

SPF chowGerm free chow

C15:0

C17:0

iso-C

16:0

iso-C

17:0

antei

so-C

17:0

iso-C

18:0

0.0

0.2

0.4

0.6

% to

tal f

atty

aci

d in

food

iso-C

16:0

iso-C

17:0

antei

so-C

17

iso-C

18:0

C15:0

C16:0

C17:0

0

1

210

40

70

100

nmol

es/m

g liv

er Germ free

SPF

iso-C

16:0

iso-C

17:0

antei

so-C

17

iso-C

18:0

C15:0

C16:0

C17:0

0.0

0.2

0.4

0.65

2035506580

nmol

es n

ewly

sy

nthe

size

d/m

g liv

er Germ free

SPF

3 x 50mls D2O/day

2 x 50mls D2O/day

Day 0 4 7 14 21

Plasma

0 5 10 15 20 250

1

2

3

Plas

ma

D2O

enr

ichm

ent (

%)

Day

ba c d

fe g h

ji k l

iso C

16:0

iso C

17:0

antei

so C

17:0

0

5

10

15

20

25

μM in

hum

an p

lasm

a

m

Supplementary Figure 6.

iso C

16:0

iso C

17:0

antei

so C

17:0

iso C

18:0

C15:0

C16:0

C17:0

0.0

0.1

0.2

0.3

0.4

0.5

fract

ion

of fa

tty a

cids

ne

wly

syn

thes

ized

in e

WAT Germ free

SPF

p=0.0001

p=0.0045

p=0.048p=0.003

p=0.00005

p=0.005

iso C

16:0

iso C

17:0

iso C

18:0

C15:0

C16:0

C17:0

0.0

0.2

0.4

0.6

0.8

1.0

fract

ion

of fa

tty a

cids

ne

wly

syn

thes

ized

in li

ver

Germ freeSPF

p=0.013

iso-C

16:0

iso-C

17:0

antei

so-C

17

iso-C

18:0

C15:0

C16:0

C17:0

0.0

0.5

1.0

20406080

100

2

nmol

es n

ewly

sy

nthe

size

d/m

g eW

AT Germ free

SPF

p=0.031

p=0.016

p=0.0005

p=0.009

iso-C

16

iso-C

17

antei

so-C

17

iso-C

18:0

C15:0

C16:0

C17:0

012345

180

220

260

300

nmol

es/m

g iW

AT

Germ free

SPF

p=0.001

p=0.0000005

p=0.00005

p=0.000005

p=0.037

p=0.001

iso-C

16:0

iso-C

17:0

antei

so-C

17

iso-C

18:0

C15:0

C16:0

C17:0

01234

205230255280

180

nmol

es/m

g B

AT

Germ free

SPF

p=0.044

p=0.003 p=0.040

iso-C

16

iso-C

17

antei

so-C

17

iso-C

18:0

C15:0

C16:0

C17:0

02468

140180220260300340

nmol

es/m

g eW

AT Germ free

SPF

p=0.011 p=0.041

Supplementary Figure 6. De novo synthesis of mmBCFAs in germ free mice and humans. a. Short chain fatty acid (SCFA) levels in the cecum of germ free (n=3) and SPF (n=5) mice. Data presented as means ±�SD with dot plot overlaid. b. mmBCFA and OCFA composition of chow fed to germ free and control mice (n=3 food pellets). Data presented as means ±�SD with dot plot overlaid. c-d Fractional synthesis of fatty acids in c, eWAT, and d, liver, e-f. Amount of newly synthesized fatty acids present in e, eWAT, and f, liver, g-j, abundance of fatty acids in g, iWAT, h, BAT, i, liver, and j, eWAT of germ free (n=6) and specific pathogen free (SPF) mice (n=6). For c-j, two-tailed students t-test was performed for statistical comparison with no adjustment for multiple comparison. Data presented as box (25th to 75th percentile with median line) and whiskers (min to max values). k. Overview of the study design for 2H2O administration and plasma collection from NAFLD patients. l. Plasma 2H2O enrichment during the 21 days of 2H2O administration (n=16). m. Levels of total plasma fatty acids in NAFLD patients (n=16). Data presented as means ±�SD with dot plot overlaid.

a b

c d

Supplementary Figure 7.

p=0.0005

p=0.015

p=0.0005

p=0.004p=0.00008

p=0.0001

iso C

16:0

iso C

18:0

iso C

17:0

antei

so C

17:0

C15:0C17

:0 C16

:0C18

:0

C16:1n

7

C18:1n

90.00.51.0

10203060

80

100

nmol

es/m

g iW

AT

p=0.00009

p=0.029

LFDHFD

iso-C

16:0

iso-C

17:0

antei

so-C

17:0

iso-C

18:0

C15:0

C17:0

0.00.10.20.3

1.01.21.41.61.82.0

nmol

es/m

g fo

od

p= 0.006p= 0.003

p= 0.026p= 0.010

p= 0.001

p= 0.001LFDHFD

p=0.002

p=0.037

p=0.0002p=0.001

p=0.011

iso C

16:0

iso C

18:0

iso C

17:0

antei

so C

17:0C15

:0C17

:0C16

:0C18

:0

C16:1n

7

C18:1n

9

C18:2n

60123

1020

50100150200

nmol

es/m

g eW

AT

p=0.0001

p=0.017

p=0.011

LFDHFD p=0.047

p=0.0002

iso-C

16:0

iso-C

18:0

iso-C

17:0

antei

so-C

17:0

C15:0

C17:0

C16:0

C18:0

C16:1n

7

C18:1n

90.0

0.5

1.0

1.5

2.0

rela

tive

abun

danc

e iW

AT WTob/ob

p=0.013

p=0.036

p=0.010p=0.015

p=0.000002

Supplementary Figure 7. mmBCFA levels change with high fat diet. a. mmBCFA and OCFA abundance in diets used (n=3 food pellets). b. Total fatty acid abundance in iWAT following LFD (n=5) or HFD (n=4) feeding for 15 weeks. c. Total fatty acid abundance in eWAT following LFD (n=6) or HFD (n=5) feeding for 15 weeks. d. Relative abundances of fatty acids in inguinal WAT from C57BL/6J(n=4) or ob/ob (n=4) mice. For a-d, two-tailed students t-test was performed for statistical comparison with no adjustment for multiple comparison. Data presented as means ±�SEM with dot plot overlaid (a-c) data presented as box (25th to 75th percentile with median line) and whiskers (min to max values) (d).

25% 13C6 15N Leucine and 13C5 15N Valine HFD High fat diet

Weeks old 6 21

8 x C57BL/6J

12 weeks 3 weeks

25% 13C6 15N Leucine and 13C5 15N Valine LFD Low fat diet8 x C57BL/6J

Leucin

e

Valine

0

10

20

30

40

% e

nric

hmen

t in

plas

ma(

1-M

0)

LFDHFD

a

d

Valine

Leucin

e

Isoleu

cine

0.00

0.05

0.10

0.15

0.20

mM

in p

lasm

a

LFD HFD

b c

h i j

k l m

# of isotopesM0 M1 M2 M3 M4 M5 M6 M7

0

5

10

15

75

100

iso

C17

:0 %

enr

ichm

ent f

rom

[U

-13C

6] L

euci

ne in

BAT

# of isotopesM0 M1 M2 M3 M4 M5 M6 M7

0

5

10

15

75

100

iso

C18

:0 %

enr

ichm

ent f

rom

[U

-13C

5] V

alin

e in

BAT

M0 M1 M2 M3 M4 M5 M6 M70

1

2

50

75

100

C17

:0 M

ID fr

om

[U13

C] V

alin

e &

Leu

cine

in

Bra

in

e f g

Supplementary Figure 8.

LFD HFD 0

50

100

150

200

250

Fast

ing

Glu

cose

mg/

dl p=0.0013

iWAT

eWAT

BATLive

r

Muscle

Heart

Brain

0

2

4

6

C15

:0 M

PE fr

om

[U13

C] V

alin

e &

Leu

cine

LFD

HFD

# #

p=0.000008

p=0.0000001

p=0.00009

iWAT

eWAT

BATLive

r

Muscle

Heart

Brain

0

1

2

3

4

C17

:0 M

PE fr

om

[U13

C] V

alin

e &

Leu

cine

LFD

HFD

#

p=0.016

p=0.00001

p=0.00001p=0.000000

p=0.003

iWAT

eWAT

BATLive

r

Muscle

Heart

Brain

0.000

0.005

0.010

0.015

0.020

nmol

es/m

g is

o-C

17:0

13

C e

nric

hed

LFD

HFDp=0.00001

p=0.004p=0.000006

p=0.00005

p=0.002

iWAT

eWAT

BATLive

r

Muscle

Heart

Brain

0.000

0.005

0.010

0.015

0.020

0.025

nmol

es/m

g is

o-C

18:0

13

C e

nric

hed

LFD

HFD

# # #

p=0.000003

p=0.0006p=0.00007

iWAT

eWAT

BATLive

r

Muscle

Heart

Brain

0.0

0.5

1.0

1.5

2.0

Leuc

ine

rela

tive

abun

danc

e/m

g tis

sue

LFDHFD

p=0.028 p=0.0006

p=0.022

p=0.024

p=0.014

iWAT

eWAT

BATLive

r

Muscle

Heart

Brain

0.0

0.2

0.4

0.6

0.8

1.0

Valin

e re

lativ

e ab

unda

nce/

mg

tissu

e

LFDHFD

p=0.038

p=0.0002

p=0.008

LFDHFD

LFDHFD

LFDHFD

Supplementary Figure 8. De novo synthesis of mmBCFAs from BCAAs in vivo. a. Overview of study design for administration of isotope-enriched chow to C57BL/6J mice in the context of LFD or HFD. b. Fasting plasma glucose levels (n=6 HFD, n=6 LFD) and c, BCAA levels at termination of study (n=8 HFD, n=8 LFD). d. Leucine and valine plasma isotope enrichment after 3 weeks on indicated diet (n=8 HFD, n=8 LFD). e-m. Metabolite levels and enrichment from C57BL/6J mice fed a HFD or LFD with 25% of valine and leucine supplied as [U13C,15N] isotope. e. leucine and f, valine relative abundance in tissue. N=8 for all comparisons except LFD iWAT (n=7), LFD eWAT (n=7), LFD BAT (n=7), LFD muscle (n=7), LFD heart (n=5), LFD brain (n=7), HFD iWAT(n=5), HFD BAT (n=7), HFD muscle (n=7) and HFD heart (n=5). g-h. Isotopologue distribution of iso-C17:0 and iso-C18:0 in BAT (n=8 HFD, n=8 LFD). i. 13C molar enrichment of iso-C17:0. N=8 for all comparisons except LFD eWAT (n=7), LFD heart (n=5), LFD brain (n=7) and HFD iWAT (n=7). j. 13C molar enrichment of iso-C18:0. N=8 for all comparisons except LFD heart (n=6), LFD brain (n=4) and HFD brain (n=4). k, Isotopologue distribution of C17:0 in the brain (n=8 HFD, n=7 LFD). l. Mole percent enrichment of C15:0 across tissues. N=8 for all comparisons except LFD eWAT (n=7), LFD heart (n=6), LFD brain (n=7) and HFD iWAT (n=7). m. Mole percent enrichment of C17:0 across tissues. N=8 for all comparisons except LFD eWAT (n=7), LFD heart (n=6), LFD brain (n=7) and HFD iWAT (n=7). Data are presented as means ± SEM with dot plots overlaid (c-d, k) or box (25th to 75th percentile with median line) and whiskers (min to max values). Two-tailed students t-test was performed for statistical comparisons with no adjustment for multiple comparisons.

iso-C17:0

a b

c d

iso-C18:0e f

anteiso-C17:0g

N.D N.D N.D N.D

Supplementary Figure 9.

p=0.031

p=0.0004

p=0.002

iso C

16:0

iso C

17:0

antei

so C

17:0

iso C

18:0C15

:0

C17:0

C16:0

C18:0

C16:1n

7

C18:1n

90.000.010.02

0.10.30.5

103050

nmol

es n

ewly

syn

thes

ized

/m

g liv

er ti

ssue

LFDHFD

p=0.002

p=0.003

p=0.018

p=0.002

p=0.006p=0.002

p=0.005

iso C

16:0

iso C

17:0

antei

so C

17:0

iso C

18:0

C15:0

C17:0

C16:0

C18:0

C16:1n

7

C18:1n

90.00.20.40.6

51015

20304050

nmol

es n

ewly

syn

thes

ized

/m

g iW

AT ti

ssue

LFDHFD

p=0.003

p=0.011

p=0.001

p=0.016

p=0.0005

p=0.001

p=0.0004

p=0.017

iso C

16:0

iso C

17:0

antei

so C

17:0

iso C

18:0

C15:0

C17:0

C16:0

C18:0

C16:1n

7

C18:1n

90.00.20.40.6

246

204060

nmol

es n

ewly

syn

thes

ized

/m

g eW

AT ti

ssue

LFDHFD

p=0.001 p=0.002

p=0.0001

p=0.001

p=0.006p=0.002

p=0.00003

p=0.017p=0.008

iso C

16:0

iso C

17:0

antei

so C

17:0

iso C

18:0

C15:0

C17:0

C16:0

C18:0

C16:1n

7

C18:1n

90.00.10.20.3

50

100

150

nmol

es n

ewly

syn

thes

ized

/m

g B

AT ti

ssue

LFDHFD

p=0.009p=0.004

p=0.001

p=0.006

p=0.0005p=0.00006

p=0.002p=0.003

LFDHFD

0.000

0.005

0.010

0.015

0.020

nmol

e de

nov

o is

o-C

18:0

/m

g tis

sue/

day iWAT

eWAT

BATLiver

p=0.004

p=0.0004

LFDHFD

0.000

0.005

0.010

0.015

0.020

0.025

nmol

e de

nov

o is

o-C

17:0

/m

g tis

sue/

day iWAT

eWAT

BATLiver

p=0.002

p=0.0004

p=0.006

LFDHFD

0.0000

0.0005

0.0010

0.0015

0.0020

0.0025nm

ole

de n

ovo

ante

iso-

C17

:0/

mg

tissu

e/da

y iWATeWAT

BATLiver

p=0.045

Supplementary Figure 9. Tissue specific mmBCFA synthesis. a-d. Amount of fatty acids newly synthesized over either a 3-week period (b, iWAT, c, eWAT, d, BAT) or 7-day period (a, liver) in HFD and LFD C57BL/6J mice as determined via incorporation of deuterium into newly synthesized lipids. LFD (n=6) and HFD (n=4) for all comparisons except LFD BAT iso-C16:0, iso-C17:0, anteiso-C17:0, C15:0, C17:0, C16:1n7 which had n=5. Data are presented as means ± SEM with dot plots overlaid. e-g. De novo lipogenic turnover of, e, iso-C18:0, f, iso-C17:0 and g, anteiso-C17:0 across the primary lipogenic tissues. Data presented as box (25th to 75th percentile with median line) and whiskers (min to max values). N.D. = not detected. Two-tailed students t-test was performed for all statistical comparisons with no adjustment for multiple comparisons.

b

Brain co

rtex

Brain m

edulla

Cerebell

um

MidbrainLive

rHea

rt

Muscle

White fa

t

Brown fa

t0

5

10

15C

rAT

prot

ein

leve

lsra

tio H

/L n

orm

aliz

ed

a

Geiger et al.

CrAT

Actin

NT G1 G2 G3

sgNT

sgCrat 3

sgCrat 1

sgCrat 2

c

Fabp4

PPAR

Adiponectin

Leptin

0.0

0.5

1.0

1.5

2.0

rela

tive

mR

NA

expr

essi

on

ControlsgCrat 1

sgCrat 2sgCrat 3

C14:0

C16:0

C16:1n

7

C18:0

C18:1n

90.0

0.5

1.0

1.5

rela

tive

abun

danc

e **

ControlsgCrat 1

sgCrat 2sgCrat 3

d e f

Supplementary Figure 10.

MMA-SA Prop-CoA Ac-CoA Acetoacetate

Citrate

Oxaloacetate

Mitochondria

CoA, NAD+

HCO3-,NADH + H+

MMSADH

Valine Isoleucine Leucine

αKIV

IB-CoA

αKMV

2MB-CoA

αKIC

IV-CoA

BCAT*

BCKDH BCKDH BCKDH

BCAT* BCAT*αKG

glut

αKG

glut

αKG

glut

CO2 CO2 CO2

NAD+

NADH + H+

NAD+

NADH + H+

NAD+

NADH + H+

CoA CoA CoA

Citrate

Ac-CoA

OxaloacetateBC-CoA Prop-CoA

Acetyl-ACP

Propionyl-ACP

Branched acyl-ACP

Mal-CoA

Palmitate Heptanoate mmBCFA

x 7

x 7

x 6

Fatty acid synthase

Cytosol

2 NADPH2 NADP, CO2

SCFA

MC-CoA Tig-CoA 3MC-CoA

ACAD8 ACADSB IVDFAD

FADH2

FAD

FADH2

FAD

FADH2

Carnitine

CoACrAT

sBCFA-carnitine

sBCFA-carnitine

Carnitine

CoA

100 μm

100 μm

100 μm

100 μm

MW Ld.

75 kD

50 kD

37 kD

Supplementary Figure 10. CrAT drives mmBCFA synthesis. a. Relative levels of CrAT across multiple tissue types determined via quantitative proteomics. Data generated from Geiger et al.36 b-c, CrAT levels (see 12b for full blot) and Oil Red O images (Scale bar represents 100 µm), of pooled CRISPR/Cas9 Crat KO 3T3L1 adipocytes (7 days post induction of differentiation). Representative images and data from one independent experiment, three independent experiments have been carried out with the same result. d. Relative mRNA expression of adipocyte differentiation markers and relative fatty acid abundance in pooled CRISPR/Cas9 CrAT KO 3T3L1 adipocytes (7 days post induction of differentiation). Representative data from three cell replicates, three independent experiments have been carried out with the same result. f. Pathway map depicting synthesis of OCFAs and mmBCFAs by FASN. Data are presented as means ±�SEM with dot plots overlaid in d-e. Two-tailed students t-test was performed for all statistical comparisons with no adjustment for multiple comparisons. *p<0.05, **p<0.01, ***p<0.001

ge f

BCKDHA

Actin

Normox Hypox

b c da

Supplementary Figure 11

Bckdh

a

Bcat2

0.0

0.5

1.0

1.5

rela

tive

mR

NA

leve

ls p=0.007 p=0.003

NormoxiaHypoxia

murine

BAT

human

myo

tubes

human

WAT

0

5

10

15

20

25C

itrat

e M

PE fr

om

[U-13

C6]l

euci

ne

p=0.004

p=0.002

p=0.004

Adipose depots

Epidid

Retrop

er

Mesen Ing

Muscle Liv

er0

10

20

30

40

50

arte

rial b

lood

flow

, m

L.m

in-1

.100

g tis

sue

LFDHFD

p=0.0009p=0.002

p=0.009

p=0.023

Actin 50 kD37 kD

50 kD

37 kDBCKDHA

LFD HFD

eWATMWLd.

Normox

ia

Hypox

ia0

1

2

3

BCKD

HA

expr

essi

on

rela

tive

to a

ctin

p=0.034

HFDLFD

Adipose depots

Epidid

Retrop

er

Mesen Ing

Muscle Liv

er0

10

20

30

40

50

pO2,

mm

HG

p=0.002

p=0.006

p=0.022

Supplementary Figure 11. Hypoxia suppresses BCAA catabolism. a. Citrate mole percent enrichment from [U-13C6]leucine in primary brown adipocytes (BAT), primary human myotubes and white adipocytes (WAT) following 24 hours of tracing in hypoxia (1%) or normoxia. Representative data from 3 cell replicates, at least two independent experiments have been carried out with the same result. b. Relative mRNA levels of BCAA related genes in 3T3L1 cells exposed to hypoxia (1%) or normoxia for 48 hours. Three independent replicates. c-d. BCKDHA protein levels in 3T3L1 cells following exposure to hypoxia (1%) or normoxia for 48 hours (see 12c for full blot). e. Oxygen tension in various tissues in HFD and LFD C57BL/6J mice (n=4). f. Blood flow to various organs in LFD or HFD C57BL/6J mice (n=4). g. BCKDHA protein levels (see 12d for full blot) in epididymal white adipose tissue from LFD or HFD C57BL/6J mice (n=3 mice). Data are presented as means ±�SEM overlaid with dot plots or box (25th to 75th percentile with median line) and whiskers (min to max values). Two-tailed students t-test was performed for all statistical comparisons with no adjustment for multiple comparisons.

Supplementary Figure 12

MW Ladder

LFD HFD

50 kD

37 kD

Full blot corresponding to Supplementary Figure 11g.

BCKDH

HFDLFD

50 kD

37 kD

ActinMW Ladder

BCKDHA

Actin

N= Normox, H= Hypoxia

Full blot corresponding to Supplementary Figure 11c.

N H N H

N H N H

FASN

GAPDH

NT G1 G2

NT G1 G2

MW Ld.

250 kD

Full blot corresponding to Supplementary Figure 2a. Full blot corresponding to Supplementary Figure 10b.

MW Ld.

MW Ld.

50 kD

37 kDActin

75 kDCrAT

NT G1 G2 G3

NT G1 G2 G3

a b

c d

Supplementary Figure 12. Full blots of a, corresponding to supplementary figure 2a, b corresponding to supplementary figure 10b, c corresponding to supplementary figure 11c, d corresponding to supplementary figure 11g.

36. Geiger,T.etal.Initialquantitativeproteomicmapof28mousetissuesusingthe

SILACmouse.MolCellProteomics12,1709-22(2013).