Appendices - 1-s2.0-S0016703720305901-mmc1.pdf

Appendix A: Calibration of alanine standards for

Site

Specific Isotope Ratio (

SSIR

)

Measurements

The molecular

average δ

C values of pure alanine standards were measured on a Thermo Fisher

Scientific Flash Elemental Analyzer (EA) coupled to a

Delta

V isotope ratio mass spectrometer

(IRMS) at Caltech. Alanine standards are described above (See Materials: Derivatization). A lab

acetanilide standard served as check on accuracy of δ

C measurements. The

values and

associated uncertainties for

the alanine standards are

19.4 ± 0.1

‰,

20.0 ± 0.2

‰, and

32.9 ±

0.2

‰ for Alfa Aesar, VWR, and Strecker alanine respectively

(Eiler et al., 2017)

(Table

);

acetanilide was measured to have a

C value of

27.6 ± 0.1

‰ in good agree

ment with its prior

measured value of

27.7 ± 1.7‰.

The Alfa Aesar and Strecker alanine standards were also analyzed at GSFC following protocols

from

(Elsila et al.,

2012)

using coupled GC

combustion

IRMS (GC

IRMS), which enables

isotopic analysis of individual amino acids in mixtures such as those from the Murchison

extracts. After accounting for dilution effects from the derivative methyl and isopropyl groups

(See Data

Processing and

(Elsila et al., 2012)

for details on dilution effects), the standards’

values were

19.4 ± 0.2

‰ and

33.3 ± 0.1

‰ for Alfa Aesar and Strecker

respectively, which is

within two standard errors of those measured at Caltech (Table

We also measured the δ

C values of C

1 in all 3 alanine standards via ninhydrin

decarboxylation, following methods from

(Van Slyke et al., 1941)

and

(Abelson and Hoering,

1961)

. Resulting

VPDB

values for C

1 were

28.5 ± 0.1

‰,

29.5 ± 0.3

‰,

43.5 ± 0.1

‰, for

the Alfa Aesar, VWR, and Strecker standards, respectively

(Eiler et al., 2017)

. Combining these

ata with the molecular

average

C values from above allowed us to calculate the average

of their combined C

2 and C

3 sites (See

Section 2.3:

Data Processing for calculations and

Figure

in main text for alanine with labelled carbon sites) as

14.

0.6

‰,

27.6

0.3

‰,

and

15.6 ± 0.3

‰. At the time of this publication,

we have no independent evidence regarding

the individual isotopic compositions of the C

2 and C

3 sites in these standards; however, NMR

studies of site

specific carbon isotope r

atios of amino acids (R. Robins pers. com.) indicate that

all common terrestrial forms of these amino acids, including standards purchased from Sigma

Aldrich (BioUltra, >99% Purity, Lot# BCBM6312V), have

VPDB

fractionations between C

and C

3 in each

molecule that are 10

‰ or less, which is in the upper range of differences

between methyl and adjacent cites for other small organics

(Gilbert et al., 2011)

The differences

we observe in the Murchi

son sample relative to the Alfa Aesar standard C

2 and C

3 are on the

order of 170

‰, and the error of the C

3 calculation (10

‰) is within error of the 10

‰

difference found between C

2 and C

3 in other alanine samples. Consequently, the potential

‰ di

fference is negligible in our study, and for this study we assume our standards have C

and C

3 sites that are identical in

C. Future measurements of one or more of the standards

used in this study could be used to refine the data presented here in ord

er to account for the likely

small differences between C

2 and C

3 in our alanine standards, but we think it implausible that

our conclusions could be influenced by the small isotopic differences between these sites likely

present in our terrestrial standa

rds.

Site

specific

C values for the Methods Development samples measured in December

and March are within error of one another (Table

). We interpret the differences in site

specific

isotope ratios between methods development and analytical samples as being due to terrestrial

contamination (though it is also possible that they partially reflect differences in isotopic

composition between

the alanine native to these two Murchison samples or fractionations arising

from chemical reactions of sample alanine during storage). Regardless, we base our discussion of

the Murchison sample only on the analytical sample. We present the data for the me

thods

development sample only in order to document the development of the methods used in this

study.

Appendix B: Blanks

Multiple procedural blanks were carried through the workup and analyzed alongside the Methods

Development and Analytical samples. Al

l blanks start at their listed step (

e.g.

extraction,

transfer, derivatization; see Table S

) and follow all subsequent steps through derivatization as

outlined in Figure S

. As an example, blanks designed to test the extraction of amino acids had

water a

dded to an empty ampoule after which all subsequent extraction, transfer, and

derivatization steps were followed. Thus, all blanks should only contain derivatizing reagents,

the products of their reactions with one another, and hexane if sample processing

produced no

contamination. Procedural Blanks are summarized in Table S

and consisted of the following:

(1a and 1b) blanks that tested reagents used in the derivatization of alanine (our ultimate

analytical target), (2) a blank that starts with water leach

ing at GSFC and continues through

chemical derivatization at Caltech, (3) a blank that starts with the water:methanol transfer of the

meteorite extract into a GC vial at Caltech, and (4) a blank that starts with analyte derivatization

at Caltech (See Figur

e S

). Procedural Blanks 1a and 1b occurred prior to the day of meteorite

extract derivatization while Procedural Blanks 2

4 occurred on the same day as the

corresponding meteorite extract derivatization. Additional solvent blanks (injections of hexane

int

o the Orbitrap) and instrument blanks (temperature ramps with no injection) were run prior to

each meteorite analysis to test the instrument background.

Each procedural blank was analyzed in Direct Injection mode on the Orbitrap, and signals were

integr

ated between 6.5 and 8.5 minutes after injection for

C and

C counts from

m/z

140.032

and 141.035 fragment peaks (for conversion from signal intensity to counts see

(Eiler et al.,

2017)

. Alanine elutes at ~7.5 minutes and is

typically transferred into the reservoir from

approximately 7

–

8 minutes retention time, so counting the background over 2 minutes

overestimates possible contamination. As with the sample data (see Site

Specific Isotope

Analysis and Data Processing) data us

ed to calculate

R was culled only to include scans that

contained both the monoisotopic and singly

substituted fragment and was computed using a

counts

weighted average of all

R values in the blank. Reported sums of

C and

C counts

(Dataset S1)

use all scans including those which have only the monoisotopic or the singly

substituted fragment without the other in order demonstrate the maximum possible error in our

measurements. When compared to samples measured with the Reservoir Elution mode,

the

overestimation is even greater because in Reservoir Elution mode measurements are broadened

over many tens of minutes, giving them a lower signal

noise ratio (which is inversely

proportional to counts reported). The procedural blank for analytical M

urchison that had the

highest amount of contamination in all metrics was Procedural Blank 2 (Table S

), which started

with the meteorite extraction at GSFC. However, compared to the 15 pmol/μL alanine in the

analytical sample, Procedural Blank 2 contained

0.15 pmol/μL and could account for only 1.9

of the integrated

C counts, 0.7% of the integrated

C counts, and 0.3

% of the integrated

signal intensity relative to the directly injected Murchison sample.

The 140.032 and 141.035

m/z

fragments are th

e most abundant ones in the mass spectrum of alanine. Maximum abundances of

m/z

140.032 and 141.035 ions in blanks were low (see Dataset S1) and did not appear during the

7.41

7.73 window during which alanine elutes, so these background signals likely eith

represent other compounds derived from column bleed, reagents, etc., and/or part of the

instrument background. For chromatograms and spectra of blanks and Murchison, see Figure S

Solvent blanks and instrument blanks were run prior to meteorite sample analyses and also

processed for integrated

C and

C counts from 6.5 to 8.5 minutes elution time (Table S

These measurements find background

C and

C counts arising from the in

jector, column,

transfer lines, etc. to typically account for less than 0.5% of the measured

C and

C counts in

Murchison samples and a <0.05

‰ change in

R values. Of the fragments used to calculate the

site

specific isotope ratios of alanine, the hig

hest background signals were observed for the

m/z

184.021 fragment. In this case, the background counts account for approximately 0.5

% of the

measured signal but change the

R value by only ~0.03

‰, which is well within the ~10

‰

standard error of the me

asurements at the 184.021 fragment. The low procedural blanks and

instrument background demonstrate that our

R values reflect alanine from the meteorite rather

than background or contamination.

Appendix C: Potential additional constraints for alanine SS

IR measurement

We attempted to add a fourth constraint to our characterization of the carbon isotope structure of

alanine by measuring the

R of a fragment ion having a monoisotopic mass of 113.0208

). The straightforward fragment

suggested by this mass would be CF

CH(O)CH

using

2 from the parent alanine. However, our studies of labeled alanines suggest that this fragment

only receives sample carbon atoms from C

3 of the parent alanine along with two carbons from

the TFAA derivat

izing reagents and none from C

2 of the parent alanine. The stoichiometry of

this ion suggests it is a recombination product (

i.e.

, because direct fragmentation of the parent

molecule cannot create a single piece containing these sites). We infer C

3 of al

anine recombines

with COH and CF

from the TFAA derivatizing reagent either as a two

body reaction or as two

stepwise reactions. This complexity calls into question whether such a measurement could yield

a consistent constraint on the

R of C

3 because th

e yields of recombination reactions generally

depend on source pressure and other analytical variables (

i.e.

, we can imagine the same ion

might be produced through other pathways when analytical conditions are varied). In any case,

when we attempted to app

ly this method to the derivatized Murchison extract our peak captures

of alanine were contaminated by at least one subsequent peak of a different compound. We

recognize one such candidate contaminant peak also produces a 113.0208 Da fragment ion.

Thus, we

consider these measurements to have failed for reasons having to do with our

chromatographic separations and peak trapping. We report these results in the for

completeness,

but we do not use these data as constraints on the Murchison sample carbon isotope

structure.

Appendix

: Error Analysis

Errors for the Total Orbitrap and the Combined Orbitrap/GC

MS calculations were weighted

according to the proportion effect of their value on the final calculation and then added in

quadrature (Eqn.

A1a

A1c

Combined Orbitrap(140,184)/GC

IRMS Calculation Error

{(3 x

molec avg

)

+ (2 x

2+C

)

}

0.5

(

A1a

)

{(2 x

1+C

)

}

0.5

(

A1b

)

{(2 x

2+C

)

}

0.5

(

A1c

)

It is important to note that the resulting computed errors for the three alanine sites are highly

correlated with one another due to interdependencies among the functions that relate them to the

various measured ratios. In particular, the

C of C

2 and C

3 are associated with large errors,

yet their average is known to within 1.5

‰ (1SE). The primary control on the error is the

experimental uncertainty in the

average C

1 + C

, which is doubled in

computing the site

specific uncertainty

of the C

2 si

(See Eqn.

)

and then propagated into the

calculated

the

site

. If future studies improve in the precision of the results presented here, it will be

productive to focus on these dependencies; in particular, a highly precise molecular

averag

measurement that includes the derivative carbons

a high precision analysis of the

m/z

21,

and a high precision analysis of the

fragment

m/z

113.032 fragment with peak capturing that

excludes subsequent peaks. These improvements were not

possible during this study due to

limited sample sizes, but a more ambitious effort to extract and purify alanine from Murchison

might achieve errors on the order of ~1

‰ for all sites (see

(Neubauer et al.,

2018)

for an

example of high precision amino acid C isotope structures measured using our techniques).

Appendix

: Alternative Pathways for Alanine Synthesis

In addition to acetaldehyde and cyanide reacting via Strecker synthesis, the alanine carbon

sotope structure could be explained by the reductive amination of pyruvic acid

(Rustad, 2009;

Robins e

t al., 2015)

. In this case, the pyruvic acid would form from a ketene (ethenone) which

sources its alkyl group (C

2) from the same

deplete CH

pool and its CO (C

1) from the

same

enriched CO pool described in the main text (See Figure S

The ethenone would then

react with CN and water to form pyruvic acid that could react with NH

on later to form alanine.

Consequently, assuming a low

C ISM CN pool, this reaction network could explain our results.

Furthermore, as the reaction network (Fi

gure S

) still involves the addition of CN to an sp

hybridized carbon and the oxidation of a nitrile to a carboxyl group

(Rustad, 2009)

, the isotope

effect and thus predicted initial carbon va

lues should not greatly change between the scenarios

(excepting possible changes in isotope effect due to physiochemical conditions).

Unlike the Strecker model, the pyruvate model would not provide clear pathways to amines,

aldehydes, or monocarboxylic a

cids. Furthermore, measured values of keto acids are, as of yet,

unavailable such that we could not compare predictions of this model to our data. For this reason,

we chose to focus on the Strecker synthesis possibility. The agreement between our predictio

and measured values across a wide

range of compound classes supports the possibility that

Strecker synthesis of aldehydes and cyanohydrins produced alanine and other organic

compounds.

We also considered whether Murchison alanine could be the product o

f a reaction network in

which alanine carboxyl is derived from high

C HCN, through Strecker chemistry. This

hypothesis could be indirectly supported by the observation that

monocarboxylic acids

Murchison have high molecular average

C values

(Yuen et al., 1984)

. If these carboxylic

acids formed by hydrolysis of nitriles, then those nitriles presumably could

have

been high in

C. And if that

C enrichment were hosted by the

terminal CN group, we should expect co

existing HCN would be

C enriched.

We are not aware of measurements of

C of Murchison

nitriles (and their terminal CN groups are certainly not known

). But if their

terminal

CN groups

were

enriched enough to account for the 10’s of per mil enrichment of carboxylic acids, it would

imply a

C value for that group of +100

‰ or more.

This hypothesis is speculative but based

on sound chemical

principles

nd so worth

considering. Nevertheless, it is strongly contradicted

by data (both from previous studies and our study), so we think it must be rejected.

Most simply,

HCN from Murchison is relatively low in

(Pizzarello, 2014)

, and our measurement of

alanine carboxyl indicates it is consistent with derivation by Strecker reaction from that

measured HCN. We

conc

lude

the most parsimonious interpretation is that alanine in fact did

form from the HCN present in Murchison, and that this HCN was not derived from a

strongly

enriched pre

solar pool.

Finally, we consider the IOM as source of organics.

(Huang et al., 2007)

argue that

monocarboxylic acids and other small organics could be produced by the hydrothermal

processing of IOM. Observations that might be taken as evidence of this idea include: 1)

Correlations of the

C values of monocarboxylic acids with their carb

on numbers are similar to

those for moieties from the IOM; and 2) our measurements demonstrate that the IOM has an

isotopic composition similar to the

C pool that was the source of the C

1 and C

3 sites of

alanine, perhaps suggesting alanine is also form

ed by hydrolysis of IOM. This second

observation could be understood in the context of the model we present if the IOM and alanine’s

1 and C

3 sites both derive from a primordial low

C pool (

i.e.

, hydrocarbons and HCN). If,

instead, alanine was made fr

om hydrolysis of the IOM, it is not obvious how it would have

acquired such an extraordinarily high

C value in its C2 carbon site without evidence of

enrichment in the C1 and C3 sites. We are aware of no high

C chemical moieties of the IOM

that could

readily explain this finding, and so we believe this idea could not be developed to

provide a satisfactory explanation of this study’s results.

Nevertheless, future compound

and site

specific measures may be able to identify IOM

processing as a source o

f soluble organics in Murchison (and perhaps other carbonaceous

chondrites). The site

specific δ

C isotope ratio for compounds produced by IOM processing

should mirror those found in the IOM aliphatic side chains (which have compound specific

molecular av

erage δ

C values of 57.9

‰ to 0.4

‰). In contrast, the reaction network we propose

predicts that the terminal carboxyl (C

1) sites of the carboxylic acids will be highly

C enriched

compared to all other CH

sites.

Appendix

Parent

Body Organic

Reaction Model

Constraints on the Isotope Effects Associated with Syntheses

To calculate the

C values of alanine precursors and organic synthesis products other than

alanine, isotope effects of different synthetic steps were collated from literature rev

iew and those

for Strecker synthesis were measured via experimental work conducted as part of this study.

Isotope effects for Strecker synthesis were further validated by comparison to literature values

for isotope effects from similar reaction mechanisms.

The reduction of aldehydes into imines via reductive amination has a maximum measured

isotope effect of 0.6

‰

(Billault et al., 2007)

, which is lower t

han our measurement errors so was

treated as a 0

‰ fractionation in the model. Studies for carbon isotope effects during the

oxidation of aldehydes have observed a range of effects from negligible (aldehyde to

thiohemiacetal conversion)

(Canellas and Cleland, 1991)

large deuterium isotope effects that

suggest possible concurrent carbon isotope effects

(Wiberg, 1954)

; although these have not been

measured. To consider both possibilities, we consider two endmember cases of 1) no isotope

effect and 2) a 30‰ normal kinetic isotope effect, similar to intrinsic KIE’s associated with other

carbon oxidation reactions

(Cleland

, 2005)

Mechanisms and associated isotope effects are

portrayed in Figure

in the main text. Differences in our solution between the 0

‰ carbonyl

oxidation KIE and the 30

‰ normal KIE case are depicted in Figure

in the main text.

Experimental work

was conducted to constrain the isotope effects in Strecker synthesized alanine

from ammonium chloride, acetaldehyde, sodium cyanide, and water at temperatures ranging

from 20°C to 25°C for the creation of the aminonitrile and 80°C to 120°C for its acid hyd

rolysis.

We measured the average isotopic composition of solid reagents and products via EA

IRMS, of

acetaldehyde via combustion over CuO into CO

which was measured on a dual

inlet IRMS, and

the site

specific isotopic composition of alanine produced by th

e synthesis was measured for

C of the C

2 + C

3 (140.032 fragment) on the Orbitrap as described above. Our measurements

indicated that the average

C of C

2 and C

3 of alanine produced by Strecker synthesis

(

30.6

0.9

‰) is approximately 12‰ deplete

d in

C relative to the reactant acetaldehyde (

19.1

‰) regardless of yield. Because C

3 does not participate in the Strecker reaction, we

assumed the difference in the average

C for C

2 and C

3 is due to a

‰ isotope effect on

which is consistent with other CN addition reactions

(Lynn and Yankwich, 1961)

. C

(found by a subtraction of C

2 and C

3 from the molecular average) exhibited a normal KIE that

had an ave

rage value of 22

‰ for alanine produced between a 10

% and 55

% yield

(

54.1

3.2

‰ relative to a starting CN

C of

31.8

0.2

‰). This KIE also agrees with

literature values for amide oxidation

(Robins et al., 2015)

Our reaction network model assumes a low yield of products and unlimited supply of reactants

relative to the products such that isotope effects would be appare

nt in products and but would not

significantly alter the

C of the reactants (and, consequently, other compounds produced from

them). The agreement between ou

r predicted isotope ratios and measurements in literature,

particularly for acetaldehyde and HCN

, is consistent with this assumption. However, below we

analyze the possibility that variations in certain factors would impact our results:

Temperature:

The isotope effects associated with reactions in our hypothesized reaction network

range up to 30

‰.

Given that the temperatures of aqueous alteration of the CM chondrites have

been demonstrated to have varied between 20 and 71 ̊C (293.15

–

344.15

(Guo and Eiler,

2007)

) through clumped isotope thermometry, and given that chemical isotope effects commonly

exhibit approximately linear variations in amplitude with 1/T

, we estimate that these model

estimates could have varied by several per mil. For moderate variations in reaction progress

(below), these should lead to variations of just a few per mil in predicted

C values of products.

This is comparable to full proc

edural analytical precision and less than otherwise unexplained

variability in the data, and so we consider it insignificant (in the context of the constraints and

goals of our model).

Reaction progress

: Our model presumes that essential reactants (water

, aldehydes, ammonia and

HCN) are more abundant than products that are created in our reaction network. If the

proportions of these compounds in the Murchison parent body initially resembled those in

comets (

e.g.

Biver

et al.

(

2019)

)

, this assumption would be well justified. However, if organic

synthesis reactions such as the Strecker chemistry locally went to near completion (consuming

most

of reactants), isotope effects associated with synthesis reactions would be mitigated, as

isotopic proportions in products would approach those of reactants. The largest kinetic isotope

effects associated with our reaction network model (30

‰) could be di

minished in this way

—

the extreme limit of quantitative yield, reduced to nothing.

The limits one should place on this argument are difficult to evaluate because all of the reactants

are more volatile than the products (

e.g.

, alanine is essentially

involatile whereas its proposed

substrates, acetaldehyde and HCN have boiling points of 20 and 26 ̊C, respectively). Thus, the

abundance ratios of aldehydes to amino acids in the Murchison meteorite are likely a poor guide

to their proportions early in the

history of the Murchison parent body. If the synthesis chemistry

had yields comparable to laboratory Strecker synthesis (10’s of %), then the effective KIE’s

would be approximately halved, or reduced by approximately 10

‰. That would be degrade the

level of agreement between our model prediction and the measured

C of some compounds in

our model (and improve the level of agreement for others), but by amounts that are a small

fraction of the isotopic variations (

i.e.

, si

specific and intermolecular differences) that motivate

our model. We therefore consider it implausible that this factor significantly impacts the overall

reasonableness of our model.

Alanine destruction

: Free and total alanine in Murchison are about o

third as abundant as in

the most alanine

rich CM chondrite (~0.20 and ~0.65

ppm, respectively), implying that it could

be residual to 10’s of % destruction. If this destruction was accompanied by a

C kinetic isotope

effect in the range typical of irre

versible organic reactions (~10

‰) and operated on one or

two atomic sites, then the residual alanine could have been enriched in

C by several per mil up

to perhaps 10

‰. The most likely mechanisms for alanine destruction (NH

replacement with

OH, or

decarboxylation) should either enrich the C

2 site or enrich both the C

1 and C

2 sites

equally in the residue. These effects are less th

n or just at the margin of the level of significance

addressed by our model and are a small fraction of the 150

‰ sit

specific effect our model was

tailored to describe. Moreover, the

C values of alanine from CM chondrites do not exhibit an

inverse concentration with their concentration in the samples, so there is no empirical evidence

to suggest such a fractionating

loss mechanism. We conclude loss of alanine through these side

reactions is unlikely to significantly impact our conclusions.

Calculation of reactant

C values

To estimate the site

specific

C values of reactants in our network model, we subtracted site

specific isotope effects constrained by our Strecker synthesis experiments from the measured

C values for alanine in the analytical Murchison sample. Based on these results, the reactant

is estimated to have a δ

VPDB

value of

‰ and the initial acetaldehyde is estimated to

have δ

VPDB

values

of 166 ± 10‰ and

‰ for the carbonyl (C

acetaldehyde

) and methyl

acetaldehyde

) carbons, respectively. Combining our results with th

e ISM chemical networks

described in

(Elsila et al., 2012)

and references therein, we predict that the carbonyl carbon in all

aldehyde functional groups are from the

enriched CO pool in the ISM and that all alkyl

carbons are from another,

depleted pool (that include C

compounds). Thus, in our model

assigned

C values of 166

‰ to all carbonyl carbons and

‰ to all alkyl

carbons. Equivalently, we calculated the molecular

average

C values of aliphatic aldehydes

with two or more carbons by calculating the carbon

weighted average values

of acetaldehyde

(64.6

1.5

‰) and additional aliphatic carbons (

‰) (Eqn.

; See

Appendix C

aldehyde

= (

)

molec avg, acetaldehyde

+ (

)

2, acetaldehyde

(Eqn.

)

where

is the carbon chain length and C

aldehyde is a molecule with one aldehyde carbon and

methylene carbons. All such calculations are made using

C mole fraction (“fractional

abundance”) rather than

C values to avoid systematic errors arising from non

linearities of the

scale.

In our model, amines form from a reactant aldehyde’s reductive amination (Figure

, main text),

which is proposed to have an insignificant KIE, so we estimated that the

C value of the amine

molecule is equal to that of an al

dehyde molecule with the same carbon backbone (See

Dataset

). Monocarboxylic acids formed from the oxidation of aldehyde precursors were assigned to

have isotope effects that range from 0

‰ to

‰. In the first case, the product carboxylic acids

have

C values equal to their aldehyde precursors (See

Dataset S2

). For the alternate case of a

fully expressed

‰ KIE during oxidation of the aldehyde’s carbonyl site, the isotope effect is

assumed to only occur on the C

1 carbon, so the molecular

average

C for acetic acid was

calculated accounting for the isotope effect only occurring on this site (Eqn.

). Higher carbon

chain carboxylic acids (C

and above) were calculated as

the carbon

weighted average values of

acetaldehyde (64.6

1.5

‰) and additi

onal CH

groups (

‰) to decrease error (Eqn.

molec avg, acetic acid

= (1

0.050/2)

molec avg, acetaldehyde

(

)

carboxylic acid

= (

)

molec avg, acetic acid

+ (

)

2, acetaldehyde

(

)

All

amino acids (

i.e.

not only alanine) were assumed to undergo fractionation in Strecker

synthesis as described above. Because our analytical Murchison alanine measurements include

C for sites that have undergone the same fractionations associated with their synthesis

(

e.g.

the C

1 and C

2 carbons of all alpha amino acids formed by Strecker synthesis are predicted to

be fractionated in the same way we predict for our model of alanine formation), we used

alanine’s site

specific isotopic composition as our building block

s for other amino acids.

Glycine’s

C was predicted based on the 184.021

m/z

fragment measurement (corrected for

dilution with carbons from derivatizing agents) and alanine was assigned to have the

C value

directly measured in this study, 25.5

‰ (

e.g.

, it is not predicted but serves as the basis for

predicting other species, particularly acetaldehyde and HCN). All amino acids with longer alkyl

chains than alanine were assumed to have additional alkyl carbons (

i.e.:

with a

C equal to that

of C

3 in a

lanine) comprising the balance of the molecular carbon inventory (Eqn.

amino acid

= (

)

molec avg, alanine

+ (

)

C3, alanine

(

)

In addition to Strecker synthesis, we also considered the possibility that C

1 in amino

acids could

equilibrate with the dissolved inorganic carbon (DIC) pool on meteorites (

e.g.

, the carbonate

pool). The DIC pool is

3000

times more abundant than all amino acids on Murchison combined

(Sephton, 2002)

. Consequently, in the case of equilibration between the

two reservoirs, the

value of DIC would control that of C

1 in amino acids. We assumed a DIC reservoir with a

of 80

‰, equal to the highest measured literature value for CM chondrites

(Sephton, 2002)

(an

thus the maximum effect on amino acids with which it equilibrates). Using

values for CO

and CO

amino acid carboxyl group equilibration from

(Rustad et al., 2008)

and

(Rustad,

2009)

respectively, we predicted the

C of different amino acids on Murchison that had

equilibrated with its carbonate pool (

Dataset S2

). Of am

ino acids with molecular

average

values measured on Murchison, only glycine and alanine also have

values for CO

and amino

acid carboxyl group in

(Rustad, 2009)

. These values are 4.4

‰ a

nd 4.9

‰, so we adopted an

average value of 4.65

‰ for

CO2

amino acid C

1 site

in our calculations for all amino acids.

Bibliography

Abelson P. H. and Hoering T. C. (1961) Carbon isotope fractionation in formation of amino

acids by photosynthetic organisms.

Proc. Natl. Acad. Sci. USA

, 623

–

632.

Billault I., Courant F., Pasquereau L., Derrien S., Robins R. J. and Naulet N. (2007) Corre

lation

between the synthetic origin of methamphetamine samples and their

N and

C stable

isotope ratios.

Anal. Chim. Acta

593

, 20

–

29.

Biver N., Bockelée

Morvan D., Hofstadter M., Lellouch E., Choukroun M., Gulkis S., Crovisier

J., Schloerb F. P., Rezac

L., von Allmen P., Lee S., Leyrat C., Ip W. H., Hartogh P.,

Encrenaz P., Beaudin G. and the MIRO team (2019) Long

term monitoring of the

outgassing and composition of comet 67P/Churyumov

Gerasimenko with the

Rosetta/MIRO instrument.

A&A

630

, A19.

Canellas

P. F. and Cleland W. W. (1991) Carbon

13 and deuterium isotope effects on the

reaction catalyzed by glyceraldehyde

phosphate dehydrogenase.

Biochemistry

8871

–

8876.

Cleland W. W. (2005) The use of isotope effects to determine enzyme mechanisms.

Arch.

Biochem. Biophys.

433

, 2

–

12.

Eiler J., Cesar J., Chimiak L., Dallas B., Grice K., Griep

Raming J., Juchelka D., Kitchen N.,

Lloyd M., Makarov A., Robins R. and Schwieters J. (2017) Analysis of molecular

isotopic structures at high precision and accuracy by

Orbitrap mass spectrometry.

Int. J.

Mass Spectrom.

422

, 126

–

142.

Elsila J. E., Charnley S. B., Burton A. S., Glavin D. P. and Dworkin J. P. (2012) Compound

specific carbon, nitrogen, and hydrogen isotopic ratios for amino acids in CM and CR

chondrites and

their use in evaluating potential formation pathways.

Meteorit

Planet

Sci

, 1517

–

1536.

Gilbert A., Silvestre V., Segebarth N., Tcherkez G., Guillou C., Robins R. J., Akoka S. and

Remaud G. S. (2011) The intramolecular

distribution in ethanol reveals the influence

of the CO₂

fixation pathway and environmental conditions on the site

speci

fic

variation in glucose.

Plant Cell Environ.

, 1104

–

1112.

Guo W. and Eiler J. M. (2007) Temperatures of aqueous alteration and evidence for methane

generation on the parent bodies of the CM chondrites.

Geochim. Cosmochim. Acta

5565

–

5575.

Huang Y

., Alexandre M. R. and Wang Y. (2007) Structure and isotopic ratios of aliphatic side

chains in the insoluble organic matter of the Murchison carbonaceous chondrite.

Earth

Planet. Sci. Lett.

259

, 517

–

525.

Lynn K. R. and Yankwich P. E. (1961) Cyanide Carbon

Isotope Fractionation in the Reaction of

Cyanide Ion and Methyl Iodide. Carbon Isotope Effect in the Hydrolysis of Methyl

Iodide.

J. Am. Chem. Soc.

, 53

–

57.

Neubauer C., Sweredoski M. J., Moradian A., Newman D. K., Robins R. J. and Eiler J. M.

(2018) Sc

anning the isotopic structure of molecules by tandem mass spectrometry.

Int. J.

Mass Spectrom.

434

, 276

–

286.

Pizzarello S. (2014) The Nitrogen Isotopic Composition of Meteoritic HCN.

Astrophys. J.

796

L25.

Robins L. I., Fogle E. J. and Marlier J. F. (2015

) Mechanistic investigations of the hydrolysis of

amides, oxoesters and thioesters via kinetic isotope effects and positional isotope

exchange.

Biochimica et Biophysica Acta (BBA)

Prot

Proteom

1854

, 1756

–

1767.