Compound‐specific sulfur isotope analysis of cysteine and methionine via preparatory liquid chromatography and elemental analyzer isotope‐ratio mass spectrometry

This article has been accepted for publication and undergone full peer review but has not been

through the copyediting, typesetting, pagination and proofreading process

which may lead to

differences between this version and the Version of Record. Please cite this article as doi:

10.1002/rcm.9007

Phillips Alexandra (Orcid ID: 0000

0001

5959

5238)

Compound

specific sulfur isotope analysis of

cysteine and methionine

via

preparatory liquid chromatography and elemental analyzer isotope

ratio mass

spectrometry

Alexandra A Phillips

, Fenfang Wu, Alex L Sessions

California Institute of Technology, Department of Geological and Planetary Sciences,

Pasadena, CA, USA

*Corresponding author: alex.phillips@caltech.edu

ABSTRACT

Rationale:

Compound

specific isotope analysis (CSIA) of organic sulfur molecules

has previo

usly been hindered by challenging preparatory chemistry and analytical

requirements for large sample sizes. The natural

abundance sulfur isotopic

compositions of the sulfur

containing amino acids, cysteine and methionine, have

therefore not yet been invest

igated despite potential utility in biomedicine, ecology,

oceanography, biogeochemistry, and other fields.

Methods:

Cysteine and methionine were subjected to hot acid hydrolysis followed

by quantitative oxidation in performic acid to yield cysteic acid a

nd methionine

sulfone. These stable, oxidized products were then separated by reverse

phase high

performance liquid chromatography (HPLC) and verified via offline liquid

chromatography mass spectrometry (LC

MS). The sulfur isotope ratios (δ

S values)

of purified analytes were then measured via combustion elemental analyzer coupled

isotope ratio mass spectromet

(EA

IRMS). The EA was equipped with a

temperature

ramped chromatographic column and programmable He carrier flow

rates.

Results:

column focusing of SO

in the EA

IRMS system, combined with

reduced He carrier flow during elution, greatly improved sensitivity allowing precise

(0.1

0.3

‰

1s.d.)

S measurements of 1 to 10 μg sulfur. We validated that our

method for purification o

f cysteine and methionine was negligibly fractionating using

amino acid and protein standards. Proof

concept measurements of fish muscle

tissue and bacteria demonstrated differences up to 4

‰

between the

S values of

cysteine and methionine that can be

connected to biosynthetic pathways.

Conclusions:

We have developed a sensitive, precise method for measuring the

natural

abundance sulfur isotopic compositions of cysteine and methionine isolated

from biological samples. This capability opens up diverse a

pplications of sulfur

isotopes in amino acids and proteins, from use as a tracer in organisms and the

environment to fundamental aspects of metabolism and biosynthesis.

Keywords: Amino Acids, Cysteine, Methionine, Sulfur Isotopes, EA/IRMS, CSIA

This article is protected by copyright. All rights re

served.

1 |

INTRODUCTION

The sulfur isotopic compositions of amino acids (AAs) are virtually unexplored but

may hold significant utility across diverse scientific

disciplines. In biomedicine, pilot

studies have suggested th

cysteine and methionine δ

values could

indicat

disease progression

sulfur metabolism is dysregulated at the onset of liver

cancer

. In archeology, bulk protein δ

measurements

from mummy hair

and

mammalian collagen

been used to reconstruct

ancestral

migration and

reliance on fish p

rotein, i

ndic

ating this as a promising direction for targeted paleodiet

reconstruction

Mass

balance isotopic models in plants suggest

that

differences

related to metabolism

could exist between cysteine and methionine

values,

which in turn could infor

m agricultural sectors on

the

efficiency of sulfur uptake in

soils

. Cysteine and methionine also have potential in biogeochemical studies to

record redox conditions, for example direct incorporation of

depleted sulfide in

anoxic sediments has been

demonstrated in deep

reaching mangrove roots

Measuring the compound

specific S isotope ratios of cysteine and methionine offer

more powerful insights than would bulk protein analyses, disentangling the effects of

metabolism versus environmental change.

ere, we present the first method for

natural

abundance sulfur isotope characterization of these amino acids

, with

successful measurements of 1

μg

sulfur

(~4

40 μg analyte).

Progress towards the compound

specific isotopic analysis of organic sulfur

com

pounds has historically been hindered by mass spectrometric limitations (Table

1). S

ulfur isotope measurements

typically

rel

ied

on analyte combustion to SO

, a

highly

polar

, toxic, corrosive, and hygroscopic

gas

, before

online measurement

via

sotope

atio

ass

pectrometr

(IRMS)

To compensate for a host of analytical

difficulties resulting from these properties of SO

, analyses require

relatively

large

sample sizes

ranging from

100

μg S

even when using

specialized elemental

analyzer (EA) with

online combustion that improved on traditional dual

inlet designs

. Moreover, because the EA does not inherently separate different analyte

compounds, offline preparative purification is needed prior to analysis.

The

combination of these two requirements

presented a substantial barrier to

measurements

of analytes

such as

amino acids that exist in the environment in low

concentrations

. An alternative strategy for sulfur isotope determination used

luorination

of analytes

sulfur hexafluoride (

), which

required large sample

sizes but improved

the

precision due to the favorable properties of SF

. When

measurements of this inert gas were combined with a microvolume and tenfold

increased signal amplification, detection limits were lowered to 0.6

3.2 μg S

However, the

preparation

of SF

requires specialized vacuum lines and dangerous

reactants

and has not yet been demonstrated for organic analytes

. M

ulti

collector

inductively coupled plasma mass spectrometry

(

ICPMS

)

has

also

recently

demonstrated r

emarkably low

sensitivity for measuring

the sulfur isotopes of sulfate

and sulfur

bearing

minerals

, but thus far requires conversion of analytes to

sulfate

Direct c

oupling

of gas chromatography (GC)

to MC

ICPMS

was first reported

in 2009

, and

has

enabled highly sensitive

compound

specific

measurements of

organic sulfur compounds

, including

volatile species from crude oils

and mature

sediments

, as well as marine dimethylsulfonopropionat

(DMSP).

Unfortunately

for our application,

GC separatio

n of

cysteine and methionine

is not a viable option

because existing derivatization strategies are

not reliably quantitative and may

fractionate sulfur isotopes.

This article is protected by copyright. All rights re

served.

Simultaneous with

ICPMS

development,

there

has been a

parallel

renaissance

IRMS techno

logy

leading to significantly

reduced sample sizes: online ‘purge and

trap’ configurations

have

measure

350 μg

sulfur

and dual

column GC systems

have reached 30

μg

sulfur

. Most

recently

the Thermo

Scientific Flash

Isolink equipped with a

temperature

ramped

chromatographic

column

measured

in bone collagen

samples containing just

μg

sulfur

. This system

, which we

improved upon in the current study,

provides sufficient sensitivity to make offline

preparative isolation

of the sulfur

AAs much less tedious.

Analyses of cysteine and methionine have also faced significant difficulties in their

chemical separation. Isolation methods have typically employed hot acid hydrolysis

to release amino acid residues from proteins

. However, this

approach led to partial

or complete oxidation of cysteine and methionine to cysteic acid and methionine

sulfone (Figure 1), even when the headspace was flushed with argon or nitrogen

gas

. To avoid such problems, amino acid residues were often

oxidized

reduced

21,2

, or

alkylated

. However, alkylation only effectively targets cysteine

and reduction only methionine (Table S1

, supporting information

). Recent studies

have thus converged on oxidation with performic acid (CH

) to

quantitatively yield

cysteic acid and methionine sulfone prior to LC

MS separation and

quantification

Here we employed

a modified version of this oxidation strategy

We validated the

method as non

fractionating using

commercial

standards of cystein

methionine,

and

bovine serum albumin

(

a well characterized

sulfur

rich protein

), and established

performance characteristics of the methodology

then

applied our novel

approach

biomass from

two ubiquitous microbes,

Escherichia coli

and

Pseudomonas fluorescens

, and

to muscle tissue from

two ecologically important fish

species,

Oncorhynchus nerka

(salmon)

and

Thunnus albacares

(tuna)

. These

analyses reveal

offsets

up to 4

‰

the

cysteine and methionine δ

values

that can

probably

e traced to

metabolism. We

expect that

this new method

ology

will

augment

the growing

stable isotope

toolkit, with applications in biomedicine,

ecology,

agriculture,

oceanography, biogeochemistry,

and

other diverse scientific fields.

2 |

EXPERIMENTAL

2.1 |

Method Overview

amples were freeze dried then homogenized with mortar and pestle prior to acid

hydrolysis (Figure 2). An aliquot was taken for bulk

analysis via EA

IRMS.

Filtered, hydrolyzed AAs were then heated in performic acid, where cystein

e and

methionine were quantitatively oxidized to cysteic acid and methionine sulfone.

Reverse

phase preparatory HPLC

UV was used to separate and purify the two sulfur

AAs. Aliquots were assayed for purity via a separate LC

MS analysis. Further

aliquots of

the purified AA’s were analyzed

via

IRMS to measure

values.

2.2 |

Experimental Reagents

Standards of cysteine, methionine, cysteic acid, methionine sulfone, and bovine

serum albumin (BSA) were purchased from Sigma Aldrich (

St. Louis, MO, USA;

all

>99% purity). All solvents used were ACS reagent grade, with the exception of

ammonium hydroxide and ammonium acetate, which were HPLC grade. All water

used was ultrapure (>18.2 M

). All glassware was combusted at 460 ̊C for seven

This article is protected by copyright. All rights re

served.

hours to remove organic

carbon contamination. Vials and syringes were additionally

washed with solvent before use (methanol, dichloromethane).

2.3 |

Sample Preparation

Filets of wild

caught

nerka

(sockeye salmon)

and

albacares

(yellowfin tuna)

were purchased at a grocery s

tore in Pasadena, CA

, USA

Bacterial cultures (

E. coli,

P. fluorescens

) were grown in our laboratory (details below).

iomass from all four

was

rinsed

with

water

five times, then freeze dried with a VirTis lyophilizer (SP

Scientific

, Stone Ridge, NY, USA

)

for 1

–

3 days until dry

(Figure 2, Step 1)

. Samples

were transferred to a solvent

washed ceramic mortar and pestle and ground under

liquid N

until homogenized

(Figure 2, Step 2)

. Homogenized samples were then

transferred to glass jars and 3 x 1 mg aliquots were taken for bulk

analysis via

IRMS

(Figure 2, Step 3)

2.4 |

Acid Hydrolysis

30 mg each of AA standards, BSA protein, and microbial biomass, and 100 mg of

fish tiss

ue were weighed directly into 60

mL vials. 10 mL of water was added and

samples were sonicated for 15 min before

the

addition of 10 mL 12N HCl. Vials were

placed on a hot plate in the fume hood (100 ̊C, 24 hrs

;

Figure 2, Step 4

). Following

hydrolysis, samp

les were vacuum filtered through baked Whatman GF/F glass fiber

filters (0.7 μm pore size) and rinsed with water into new 60

mL vials. Filtered

samples were dried to completion under a stream of N

in an acid

grade fume hood.

2.5 |

Performic Acid Oxidati

Performic acid was prepared immediately prior to use by mixing hydrogen peroxide

and formic acid in a 9:1 (v:v) ratio and incubating (30 min, 23 ̊C). 5

10 mL of

performic acid was added to dried samples, which were placed on a hot plate (70 ̊C,

60 min)

in the fume hood, with occasional stirring throughout the reaction before

quenching on ice

(Figure 2, Step 5)

. Oxidized samples were dried under a stream of

. Samples were then resuspended via vortexing in 1.5 mL ultrapure water and

filtered through a 13

mm 0.22 μm PVDF (polyvinylidene fluoride) syringe filter

(Millex) into a 2

mL vial for HPLC separation.

2.6 |

HPLC

UV Separation

Methionine sulfone and cysteic acid were separated with an Agilent

(Santa Clara,

CA, USA)

1100 HPLC

UV system coupled to a Gilson

(Middleton, WI, USA)

FC203B

fraction collector adapted from a previously described method

(Figure 2, Step 6)

Briefly, samples (100 μL) were separated

on a PRPX100 strong anion exchange

column (Hamilton,

Reno, NV,

USA)

250 mm x 4.6 mm x 5 μm, 30 ̊C) with isocratic

50 mM ammonium acetate, buffered to pH 8 with 25% ammonia solution, at a flow

rate of 1.0 mL/min. Hydrolyzed samples produced a high and continuous

background UV absorption signal, obscuring

the

peaks for

cysteic acid and

methionine. Fraction collection of samples was therefore based solely on time

windows derived from separate analyses of methionine sulfone and cysteic acid

standards monitored at 254 nm.

2.7 |

MS Verification

MS analysis of all samp

les and standards was used to ensure that the collected

analytes were pure. Fractions collected from HPLC

UV separation were derivatized

with FDAA (

fluoro

dinitrophenyl

alanine amide

) and separated following a

This article is protected by copyright. All rights re

served.

previously published procedure

(Fi

gure 2, Step 7)

. Briefly, 100 μL of aqueous

sample was reacted with 10 μL of 6% triethylamine and 10 μL of 1% (w/v) FDAA in

acetone at 50 ̊C for 60 min then quenched with 10 μL of 5% acetic acid. Aliquots (20

μL) were introduced to an Agilent 1100 Series L

C/MSD with a Zobraz 300SB

column (Agilent, 2.1 mm x 150 mm x 5 μm), housed in the Caltech Proteomics

Laboratory, for a 45 min gradient between 5% acetic acid and acetonitrile at a flow

rate of 0.25 mL/min.

Mass spectra were

obtained in positive ion mode

, scanning

between

m/z

300

and

450.

The e

lectrospray voltage

was 4

350

°C. The

diode

array

detector

measured

the

UV absorption

340

2.8 |

IRMS Measurement

Fractions collected from the HPLC

UV separation were transferred to tin capsules

(OEA labs,

Exeter, UK;

9mm x 5mm, pressed, ultra

clean) and dried overnight in an

oven at 50 ̊C. Samples were analyzed with a

Thermo Scientific

(Bremen, Germany)

EA IsoLink™

ombustion elemental analyzer

system

coupled to a

Delta V Plus

Isotope Ratio Mass Spectrometer

(Thermo Scientific)

via a ConFlo IV Universal

interface

(

Thermo Scientific,

Figure 2, Step 8;

Figure 3)

The

EA utilized a single

reactor

configuration

with

user

packed columns comprising 3 cm of quartz wool, 14

cm wireform copper (5 mm size), 2 cm quartz wool, 6 cm granular tungsten (III)

oxide, 1 cm quartz wool, and 0.5 cm additional tungsten (III) oxide. Sample

combustion was accompanied by a pulse (4 s) of

and carried in a high (100

mL/min)

helium carrier gas flow rate

. SO

is trapped on the GC column at 50°C,

helping to sharpen the SO

peak while allowing CO

and N

to elute. The helium

carrier flow is then reduced to 15 mL/min to improve

the

split rati

os, and SO

eluted as a sharp peak (<30 s FWHM) by ramping the GC column temperature to

240°C at

100°C

/s. A typical IRMS measurement (24 min) brackets the sample SO

peak between four SO

reference gas peaks, with no magnet jump (Figure 4).

2.9 |

Data

Processing

The S contents (typically < 0.25 μg S) and

values (typically 1

10‰) of empty tin

capsules were measured

IRMS and used to correct all subsequent analyses

for the blank contribution

. Different batches of capsules varied in

their

S isotope

composition by up to ~5‰ and therefore the same batch was used for all samples

and standards within a day’s run. In the current study this blank adjustment was

minimal (< 5%) as sample peaks were sufficiently large (

30 Vs, 5 μg S)

;

however

for

smaller sample sizes the blank correction can become precision

limiting. A

previous report concluded that oxygen isotope correction, i.e. for

O, had a

negligible effect on

values and therefore performed no explicit

O correction

In our da

ta processing, any

O effects are corrected for during calibration with

external reference materials:

values

were

measured

relative to

a lab

reference gas

that

was itself

calibrated against IAEA reference materials

S1, S2, and

S3 using the same EA

IRMS system

. IAEA

and

IAEA

2 standards

were

also

analyzed in triplicate at the beginning, middle, and end of daily sequences

further

calibrate

sample

values, which were reported as permil (

‰

) variations relativ

to the

Vienna

Canyon Diablo Troilite

(VCDT) reference frame.

Sayle et al

observed large (0.6

‰

per V) size

related errors for aliquots of bone

collagen analyzed for

with the same model

IRMS system. In our tests

with SO

reference gas, perfo

rmed daily prior to analyses, linearity effects were

consistently low (<0.1

‰

per V). We observed no significant size

related effects for