Nature | www.nature.com |
1
Article
Bacterial chemolithoautotrophy via
manganese oxidation
Hang Yu & Jared R. Leadbetter
✉
In the format provided by the
authors and unedited
Supplementary information
https://doi.org/10.1038/s41586-020-2468-5
Nature | www.nature.com/nature
1
Supplementary Information
Table of Contents
Supplementary Taxonomic Proposal
................................
................................
..........................
2
Supplementary Notes
................................
................................
................................
....................
5
Supplementary Note 1: Thermodynamic calculation of Mn(II) oxidation
................................
.......
5
Supplementary Note 2: Growth of Species B
................................
................................
.......................
5
Supplementary Note 3: Evaluation of FISH oligonucleotide probes for Species A and Species B
.
5
Supplementary Note 4:
Evaluation of the ICP
-
MS method to measure oxidised and reduced
pools of Mn
................................
................................
................................
................................
..............
6
Supplementary Note 5: Evaluation of quantitative PCR p
robes for Species A, Species B and
Bacteria
................................
................................
................................
................................
....................
7
Supplementary Note 6: Ammonia as the anabolic N source for the co
-
culture
................................
8
Supplementary Note 7: Oxygen respiration and energy conserving respiratory complexes in
Species A
................................
................................
................................
................................
..................
8
Supplementary Note 8: Electron balance between catabolism and anabolism in Species A
.........
10
Supplementary Note 9: Stable isotope probing and nanoSIMS analysis
................................
........
11
Supplementary Note 10: Carbon assimilation pathways in Species A
................................
............
13
Supplementary Note 11:
Unrestricted manganese oxidation calculation
................................
........
15
Supplementary Information References
................................
................................
...................
16
2
Supplementary Taxonomic Proposal
Description of
“
Candidatus
Manganitrophus
”
, gen. nov.
Manganitrophus [man.gan.i.tro’phus]. N.L. n. manganum, manganese; N.L. Gr. n. trophos,
feeder; N.L. masc. n. Manganitrophus
, manganese feeder. Cells are pleomorphic, crescent shaped
rods and exhibit a capacity for aerobic, chemolithoautotrophic growth using manganese carbonate
as sole source of carbon and energy. Colonies have not yet been observed to form from single cells,
a
nd cells have not yet been grown in the absence of a co
-
enriched strain of a putative new
betaproteobacterial species belonging to the genus
Ramlibacter
. Liquid cultures grown in MnCO
3
defined minimal medium generate small manganese oxide nodules or concre
tions to which the
cells associated. Based on phylogenetic reconstructions using 16S rRNA gene and/or protein
sequences, the genus affiliates with
in
the proposed phylum
Nitrospirota
(
Nitrospirae
), is distinct
from
the genera comprising the classes
Nitrospi
r
i
a
and
Leptospirill
ia
,
and clusters with sequences
from not yet cultivated organisms typically recovered from subsurface karst environments, such
as that of “
Ca.
Troglogloea absoloni”. The G+C content of the only representative, type species is
56.4 mol%.
The type candidatus species of the candidatus genus is “
Ca.
Manganitrophus
noduliformans”.
Description of “
Candidatus
Manganitrophus noduliformans” sp. nov.
Candidatus
“Manganitrophus noduliformans” [nod.ul.i.for’mans
]. L. gen. dim. n. noduli,
of a little knot or nodule; L. part. adj. formans, forming; N.L. part. adj. noduliformans, nodule
-
forming. Cells exhibit aerobic, chemolithoautotrophic growth on manganese carbonate, utilizing
the Mn(II) as sole energy source whi
le fixing inorganic carbon via the reverse Tricarboxylic Acid
Cycle. Cells form pleomorphic, crescent shaped rods, ca. 1.07 μm by 0.4 μm in dimension. Liquid
cultures grown in manganese carbonate media generate manganese oxide nodules and concretions
with
which the cells associated. Colonies from single cells have not yet been observed to form in
or on agar, or to grow in the absence of a co
-
enriched strain of a putative new betaproteobacterial
species belonging to the genus
Ramlibacter.
By genomics, cells
are predicted to be capable of
motility by flagella or twitching, and of chemotaxis, but these have not yet been observed.
The
genome encodes LuxR and LuxI homologs often associated with acyl
-
homoserine lactone based
quorum sensing.
Growth by Mn(II) oxidat
ion has been observed to occur a pH ranging between
3
5.7 and 7.4. Optimal growth occurs between 34
°C and 40
°C, with no growth observed at or above
44
°C. Cultures are pasteurized by incubation at 50
°C for 12 hours. Cells grow at NaCl
concentrations equa
ting to salinities ranging from 2 ppt to 16 ppt and are predicted by genomics
to be able to make the compatible solute hydroxyectoine. Respiratory quinones are predicted by
genomics to be of the naphthoquinone type. By genomics, cells are predicted to be a
ble to
assimilate sulfate, but unable to grow lithotrophically via the oxidation of either H
2
,
ammonia
,
nitrite, or reduced sulfur substrates, or to grow anaerobically via denitrification or dissimilatory
nitrate reduction. Based on phylogenetic reconstruc
tions using 16S rRNA gene and/or protein
sequences,
the genus affiliates with
in
the proposed phylum
Nitrospirota
(
Nitrospirae
), is distinct
from
the genera comprising the classes
Nitrospir
i
a
and
Leptospirill
ia
,
and clusters with sequences
from not yet cult
ivated organisms typically recovered from subsurface karst environments, such
as that of “
Ca.
Troglogloea absoloni”.
The type (meta)genome gene sequence for the candidatus species is:
DDBJ/ENA/GenBank VTOW00000000
TS
(DOE
-
JGI
Integrated Microbial Genome
ID
2784132095
TS
); the 16S rRNA gene sequence is GenBank MN381734
TS
. These were obtained
from a two species co
-
culture along with a putative new species representing the
betaproteobacterial genus
Ramlibacter
(Genome sequence: DDBJ/ENA/GenBank
VTOX00000000 and
DOE
-
JGI
Integrated Microbial Genome
ID
2778260901
; Genbank 16S
rRNA MN381735). The culture was refined from a mixed enrichment culture initiated with
manganese carbonate and ca. 250 ml of unsterilized Pasadena (California
, USA
) municipal
drinking water distributed from a tap located on the second floor of the North Mudd Laboratories
at the California Institute of Technology (34°08'12.1"N 118°07'38.3"W). The (meta)genome of
the type (resolved from the
Ramlibacter
species) is 5.1
7 Mbp in size, and has a G+C content of
56.4 mol%.
[
Classis ‘Ca.
Trogloglia’ is represented by “
Ca.
Troglogloea absoloni”;
while ‘
Ca.
Manganitrophaceae’, fam. nov., and ‘
Ca.
Manganitrophales’, ord. nov.
are
typified by
“
Ca.
Manganitrophus”
and the type (
meta)genome sequence
DDBJ/ENA/GenBank
VTOW00000000
TS
].
4
Description of
Ramlibacter lithotrophicus,
sp. nov.
Ramlibacter lithotrophicus
(li.tho.tro'phi.cus).
Gr. n. lithos stone; N.L. masc. adj. trophicus
(from Gr. adj. trophikos
), feeding; N.L. masc. adj. lithotrophicus, a capacity for lithotrophic
metabolism
. Cells exhibit aerobic heterotrophic growth on succinate or tryptone; anaerobic
heterotrophic growth with nitrate as electron acceptor; and aerobic chemolithotrophic growth
on
H
2
.
Cells
are
pleomorphic
, forming
short
rods
, ca.
1.22 μm by 0.56
μm
in dimension
, with single,
central phase bright inclusions at low cell densities; and fabric
-
like networks of long filaments of
varying width at high cell densities. Cells exhibit tw
itching surface motility.
Colonies
are small
and featureless on
succinate agar
defined
media
or
reddish brown on tryptone
agar defined media
incubated aerobically, and become leathery and adherent upon aging; colonies are red when grown
anaerobically with
nitrate.
By genomics, cells are predicted to be capable of twitching
and flagellar
motility and to encode
putative hydrogenases; putative
sox
and
dsr
MKJOP gene clusters for the
lithotrophic oxidation of reduced sulfur compounds; genes for denitrification;
and genes for a
putative Calvin Cycle. Anaerobic metabolism of aromatic compounds by the
tungsten
-
dependent
benzoyl
-
CoA reductase
pathway is predicted by genomics, as is acyl
-
homoserine lactone based
quorum sensing via LuxR and LuxI homologs.
Growth has be
en observed to occur a pH ranging
between 5.7 and 7.4
, and at temperatures ranging from 20
°C
to
40°C
.
Based on phylogenetic
reconstructions using 16S rRNA gene and/or protein sequences,
the
species
affiliates with
in
the
betaproteobacterial genus
Ramlibac
ter
, but is distinct in both gene content and in pairwise
similarities from other
Ramlibacter
species.
The type
strain for the species is strain RBP
-
1
T
; the
genome sequence for the
type
species
is: DDBJ/ENA/GenBank
VTOX00000000
TS
and at
DOE
-
JGI
Integrated Microbial Genome
ID
2778260901
TS
; the 16S rRNA gene sequence is GenBank
MN381735
TS
. The
strain was isolated
on succinate minimal agar media
from a two species
culture (the other organism being “
Ca.
Manganitrophus noduliformans
”)
refined from a m
ixed culture enriched using
manganese
carbonate and ca. 250 ml of unsterilized Pasadena (California
, USA
) municipal drinking water
that
had been
distributed from a tap located on the second floor of the North Mudd Laboratories at the
California Institute o
f Technology (34°08'12.1"N 118°07'38.3"W). The genome of the type is
5.
26
Mbp in size, and has a G+C content of
69.03
mol%.
5
Supplementary Notes
Supplementary Note 1:
Thermodynamic calculation of Mn(II) oxidation
The Gibbs free energy change of
Mn(II) oxidation at pH 7 (ΔG°’) of the following reaction
is determined to be
-
68 kJ/mol Mn(II) using the standard Gibbs free energy of formation
105
:
Mn(II
) + ½O
2
+ H
2
O → MnO
2 (s)
+ 2H
+
Values used are: G°
f
of Mn(II) =
-
228 kJ/mol, G°
f
of
훿
-
MnO
2
=
-
453.1 kJ/mol, G°
f
of O
2
(g)
= 0 kJ/mol, G°
f
of H
+
= 0 kJ/mol, G°
f
of H
2
O (l) =
-
237.18 kJ/mol. The reaction is under standard
condition except [H
+
] = 10
-
7
M. Not
e that the composition of birnessite (
훿
-
MnO
2
) is not well
defined and could range from MnO
1.74
to MnO
1.99
106
. The Gibbs free energy change of Mn(
II)
oxidation could be
-
80 kJ/mol Mn(II), if G°
f
of β
-
MnO
2
(pyrolusite, the most defined and stable
form) =
-
465.1 kJ/mol were used.
The Nernst equation was used to calculate the reduction potential at pH 7 (E°’) of
MnO
2
/
Mn(II), based on
Δ
G°’ of
훿
-
MnO
2
fr
om above and E°’ of O
2
/
H
2
O as
+
8
18
m
V
43
, to be
+
46
6
m
V.
Supplementary Note 2:
Growth of Species B
While Species B could be isolated from the Mn
(II)
-
oxidising consortia heterotrophically,
its growth varies on differ
ent carbon sources tested. On agar plates, Species B grows modestly
well with succinate and tryptone, poorly with acetate, and weakly with yeast extract. In liquid
cultures, Species B grows well with tryptone with doubling time of 2.89 ± 0.03 hr (n=3,
Exte
nded
Data Fig. 6c
), but poorly with acetate (
n=2,
Extended Data Fig. 6d
), succinate and yeast extract.
Species B also grows in liquid cultures lithoautotrophically with hydrogen as the energy source
without any organic carbon addition in the medium.
Supp
lementary Note 3:
Evaluation of FISH oligonucleotide probes for Species A and
Species B
Three oligonucleotide probes, two newly designed (NLT499
,
targeting
most members of
the
phylum
Nitrospirae
(
Nitrospirota
) to the exclusion of
Nitrospiria
, and BET
8
67
,
targeting
Betaproteobacteria
) and one previously designed (BET359 targeting
Betaproteobacteria
), was
evaluated using the Clone
-
FISH method
67
. No significant fluoresce
nce intensity decrease was
6
found for probes NLT499, BET359 and BET867 up to 45%, 35% and 35% formamide
concentrations in the hybridization buffer, respectively (
Extended Data Fig. 10a
-
d
). No cross
-
reaction was found when the two Betaproteobacterial probes
were applied to 16S rRNA gene clone
of Species A (
Extended Data Fig. 10a
,b
), and vice versa when NLT499 probe was applied to 16S
rRNA gene clone of Species B (
Extended Data Fig. 10
c,d
).
Supplementary Note 4:
Evaluation of the ICP
-
MS method to
measure oxidised and reduced
pools of Mn
The ICP
-
MS method used in this study measured the Mn concentration in the “acid
-
soluble
fraction” and the “acid
-
insoluble fraction” similar to a previous study
6
. Mn(II) compounds,
including MnCl
2
, MnSO
4
, MnCO
3
, should be soluble or become soluble in acid (0.5 M HCl). On
the other hand, Mn(IV) oxides s
hould not be acid soluble but solubilized upon reaction with
hydroxylamine. Accordingly, almost all Mn content for the Mn(II) compounds were found in the
“acid
-
soluble fraction”, whereas almost all the Mn content for various preparations of Mn(IV)
oxides w
ere found in the “acid
-
insoluble fraction” (
Extended Data Fig. 10e
).
Mn(III), which could be the predominant Mn species in the environment
107,108
, could not
be clearly distingui
shed in our method of analysis. Commercial Mn(III)
-
containing compounds
including Mn
3
O
4
and Mn
2
O
3
showed partitioning in both the “acid
-
soluble fraction” and the “acid
-
insoluble fraction” (
Extended Data Fig. 10e
). Mn
2
O
3
, with two Mn(III) per compound, partitioned
almost equally in the two fractions. Mn
3
O
4
, with one Mn(II) and two Mn(III) per compound,
partitioned about 20:80 in the acid
-
soluble:acid
-
insoluble fractions. It seems based on these two
test chemicals, Mn(III) partitioning into the two Mn fractions could not be easily predicted based
on the average or individual Mn oxidation states. The stru
cture in which hosts Mn(III) likely also
contribute to the partitioning as well. In our experiments, we would not able to distinguish Mn(III)
pools and that of Mn(II) and Mn(IV), if Mn(III) were abundant in our cultures. At the end of our
kinetic experimen
ts after which the cultures have reached stationary phase with no further cell
growth, part of the total Mn was still measured in the reduced pool (
Fig. 2; Extended Data Fig. 2
).
There could be
three
explanations to this observation: 1)
there was still uno
xidised Mn(II) left at
the end of the experiment and thus measured in the acid soluble pool; 2)
Mn oxide biologically
produced react and partition differently than that of Mn(IV) oxides tested, and part of biologically
produced Mn oxide solubilized and was
measured in the reduced pool;
3
) there was Mn(III) in the
7
Mn oxides at the end of the experiment and thus measured in both the acid
-
soluble and acid
-
insoluble pools. These
three
possibilities needs to be addressed in future measurements targeting
Mn(III)
107,108
.
However, since Mn(II) in its various forms can be almost entirely measured in the acid
-
soluble fraction with little in the acid
-
insoluble fraction,
any increase in the ac
id
-
insoluble fraction
is an indication of oxidised Mn(II). Therefore, i
n this paper, we refer to the “acid
-
soluble fraction”
as
Mn(II), and
the
“acid
-
insoluble fraction”
as
Mn(II) oxidised representing Mn(III/IV).
Supplementary Note 5:
Evaluation of quant
itative PCR probes
for
Species A, Species B
and Bacteria
In quantitative PCR analyses, the range of bacterial quantification was 1
×
10
4
to 1
×
10
8
copies per reaction tube (
Extended Data Fig. 10g
), similar to the results reported previously on this
set
of primer
-
probe combination
74
. In our assays, the background signal was equivalent to
approximately 7.5
×
10
3
copies per reaction tube, higher than previously reported value of 2.5
×
10
3
copies per reaction tube
74
. High amplificatio
n efficiency (101.1% to 102.5%.) was found in the
bacterial quantification range. There was no difference when either Species A or Species B 16S
rRNA gene was used as the template for amplification (
Extended Data Fig. 10g
). For specific
quantification of S
pecies A and Species B, two probes were designed to use in conjunction with
the bacterial assay. The quantification ranges of these new probes were found to be 1
×
10
4
to 1
×
10
7
copies and 1
×
10
3
to 1
×
10
7
copies per reaction tube for Species A and Spec
ies B, respectively
(
Extended Data Fig. 10h
). The amplification efficiencies in these ranges were found to be 91.7%
for the Species A probe and 94.6% for the Species B probe. While 16S rRNA gene copies could
still be detected above or below the quantificat
ion ranges, the amplification efficiencies decreased
below 90% and thus not used in our analyses. When templates for Species A and Species B were
mixed in the same reaction tube, no interference in the quantification of either species was found
with the sp
ecific probes. The measured 16S rRNA gene copies of Species A and Species B
matched closely to that expected in the template mixtures (
Extended Data Fig. 10i,j
). Overall, our
analyses suggest that the specific probes developed in this study work in conjunc
tion with a
previously developed assay to quantify 16S rRNA gene copies of Species A, Species B and all
Bacteria simultaneously in a single quantitative PCR reaction.
8
Supplementary Note 6: Ammonia as the anabolic N source for the co
-
culture
The performanc
e of
cultures provided with nitrate as anabolic N source
(Main Text)
were
compared with those supplied with
ammonia
.
Although Species A should be able to transport and
assimilate nitrite, cyanate, urea,
and
ammonia
for anabolic N, genomics predicts that it
is incapable
of assimilating nitrate (
Supplementary Table 4
). This predicts
a dependency on
Species B
for N i
n
nitrate medi
a
, as the latter encodes multiple pathways for nitrate reduction (
Supplementary Table
5
).
Ammonia
might be expected to relieve this dependency, meanwhile serving as a superior
anabolic N source for both bacteria, with resultant improvements in growth and Mn(II) oxidation.
Curiously, for unexplained reasons this was not the case
. Mn(II) oxidation rate i
s comparable if
not faster with nitrate than with ammonia: the oxidation rate doubled initially every 6.3 day (s.d.
= 1.0, n=4) and slowed to every 10.9 day (s.d. = 0.3, n=4) with nitrate, and doubled initially every
6.0 day (s.d. = 1.2, n=4) and slowed to
every 14.5 day (s.d. = 4.0, n=5) with ammonia (
Fig. 2a;
Extended Data Fig. 2e
-
l
).
For one of the biological replicates with comparable Mn(II) oxidation rate (1 mM nitrate
replicate 1 shown in
Fig. 2a
, and 1 mM ammonia replicate 5 shown in
Extended Data F
ig. 2l
),
qPCR assay was used to measure the growth of Species A and Species B. Cell growth is slower
with ammonia, with
Species A
doubling every
9.7
days
and
Species
B
doubling every
11.1 days
(
Extended Data Fig. 2m
), compared to Species A doubling every 6
.2 days and Species B doubling
every 7.4 days (Main Text; Fig. 2b). This slower cell growth also resulted in lower cell yield of
roughly half of that of nitrate (Fig. 2c;
Extended Data Fig. 2n
). Additional experiments are needed
to confirm and better under
stand the underlying reasons for this influence of N source on the co
-
culture.
Supplementary Note 7:
Oxygen respiration and energy conserving respiratory complexes in
Species A
On average, the two Mn(II)
-
derived electrons are generally considered to be of
high
potential [Mn(II)/Mn(IV), E° ́=+
466
mV;
above]. However, the energetics of each of the two
sequential one
-
electron transfers can be impacted by inorganic and organic binding ligands
22,37
,
leading to a degree of uncertainty h
ere. While the
entry point
for each Mn(II) derived electron i
nto
electron transport chains
remains uncertain, neither electron likely has sufficient negative
reduction potential for the reduction of
membrane quinone
s
such as ubiquinone (E°’=+113 mV
43
)
.
9
Of the respiratory complexes, canonical respiratory
Complex I is unlikely to be employed for
energy conservation, leaving canonical or alternative Complex III, Complex IV, or cytochrome bd
oxidases as possible candidates for generating a proton motive force during Mn(II)
chemolithotrophy.
While a
single C
omplex IV
(
cbb3
-
type
heme
-
copper oxidase)
was identified
in Species A
genome,
it was
not well
-
expressed (24th percentile)
and therefore likely not the main oxygen
reductase under our experimental conditions
. Instead, four unconventional terminal oxidase
(T
O)
complexes, each having
cytochrome
bd
oxidase like (bd
-
like)
proteins, dominated expression (
Fig.
3b; Supplementary Table 4
). Compared to
canonical
bd oxidases, bd
-
like proteins have 9
-
15 (vs
8
-
9) predicted transmembrane helixes
, are
phylogenetically dis
tinct (
Extended Data Fig. 7
a), are
missing
conserved residues
including the known
quinone
-
binding
sites
38
(
Extended Data Fig. 8
)
.
Also, the
oxygen reduction activity of these bd
-
like oxidase remains to
be
biochemically
confirmed
.
The
bd
-
like proteins from the 4 terminal oxidases are not alike
,
and represent 4 distinct
evolutionary subclades (
Extended Data Fig. 7
b
).
Similar observations have been reported for
nitrite
-
oxidising
Nitrospira
, which
showed high expression of these bd
-
like oxidases in active
cultures
38,39
. In particular,
Nitrospira mosco
viensis
: the high expression of bd
-
like oxidases and
low expression of conventional bd oxidases suggest that the bd
-
like oxidases are likely used in
nitrite oxidation, and the conventional oxidases are used for other low potential electron donors
39
.
The
most highly expressed
bd
-
like
terminal oxidase complex in Species A was “TO_1”
(99
th
percentile
; Fig. 3b
)
,
which is also the highest expressed bd
-
like ox
idase in nitrite
-
oxidising
N. moscoviensis
39
. Like homologous complexes from
ammonia
-
and nitrite
-
oxidi
s
ing
Nitrospira
(
Extended Data Fig. 7b
)
,
TO_1 con
tains a predicted heme b membrane
domain
homolog
ous to
ethylbenzene dehydrogenase gamma subunit EbdC
109
,
nitrite oxidoreductase gamma subunit NxrC
,
and other
CISM
family
membrane attachment subunit
s
110,111
.
However, CISM subunits with
molybdate
-
binding motifs (e.g. N
xrAB) typically accompanying these subunits were not identified
in Species A.
This suggests that
TO_1
may
receive electrons from
alternative
periplasmic carriers
,
such as ferredoxin or cytochrome c (Fig.
3b), perhaps avoiding
energy loss associated
with el
ectron
transfer
through the
quinone
.
Energy maybe conserved in the same manner as conventional bd
oxidases, which have two proton transfer pathways to generate a proton motive force through
scalar reactions without proton pumping
102
.
The second highest expressed terminal oxidase (TO_2
;
83
rd
percentile
; Fig. 3b
)
also contains
a bd
-
like protei
n and
a membrane
-
bound
,
single heme b
10
EbdC/NxrC
homolog
, but is otherwise
highly unusual amongst cultivated organisms
[
currently
can only be found in an ammonia
-
oxidizing
Nitrospira sp.
RCA
112
and 3 other MAGs in clade IIIb
(
Extended Data Fig. 7b
)].
The c
-
terminus of the bd
-
like protein of this complex has two unu
sual
heme c binding sites, and the co
-
transcribed gene cluster encodes for a split version of
the
membrane cytochrome b
of a
conventional Complex III
(or
cytochrome bf complex
)
, multiple
cytochrome
s
c with 1
-
4 predicted heme c binding sites, and two H
+
/Na
+
antiporter MrpD
-
like
subunits (
Supplementary Table 4
). MrpD subunits are homologous to the proton pumping subunits
of
Complex I (NuoL/M/N) and various hydrogenases (e.g. CooM, HyfB/D/F, EchA)
113
. The
presence of motive
-
force generatin
g subunits in a terminal oxidase
is unprecedented
.
Perhaps
,
complex TO_2 mediates a process similar to that of electron bifurcation in anaerobic archaea and
bacteria
114
, e.g. perip
lasmic carriers of Mn(II) derived electrons (E°’=+4
66
mV) might couple the
thermodynamically favourable reduction of O
2
(E°’=+81
8
mV)
43
to unfavourable Q reduction
(E°’~+113 mV)
(Fig. 3b)
.
The bd
-
like proteins of t
he
third
(TO_3) and fourth (TO_4) highest
expressed terminal oxidase complex
es
(52
nd
and 35
th
percentile
, respectively
;
Fig. 3b
)
are
unusually truncated
and
occur in each complex as
pairs (
Extended Data Fig. 7b
). Both TO_3 and
TO_4
share some resemblance t
o canonical C
omplex III
and
Alternative Complex III
115,116
, and
therefore are
predicted
to have quino
ne
interactions
(Fig.
3b
)
.
Their expression
suggest
s
that
they
are not the main terminal oxidases
, rather may interact
with complexes that generate
reduced
quinones,
such as TO_2.
Supplementary Note 8:
Electron balance between catabolism and anabolism in
Species A
T
he biomass yield of Species A is estimated to be 100 mg dry biomass
×
(mol Mn(II)
oxidised)
-
1
, based on DNA yield of
3.1
×
10
6
ng
DNA
×
(mol Mn(II)
oxidis
ed)
-
1
(
Extended Data
Fig. 2d
), and 3.1% of dry weight of
Escherichia coli
is DNA
27
. Assuming [CH
2
O] as a simplified
composition estimate for biomass (30 g dry weight per mol), and given 4 mol electrons needed to
reduce
1
mol of CO
2
to
1
mol of [CH
2
O], then approximately 0.013 mol of electrons are in the
biomass per mol Mn(II) oxidised. With 2 electron
s from the Mn(II) oxidation to Mn(IV)O
2
,
approximately 0.7% of the electrons from Mn(II) oxidation could be recovered in the biomass.
11
Supplementary Note 9:
Stable isotope probing and nanoSIMS analysis
Mn(II)
-
dependent CO
2
fixation was examined directly in
a closed system, using stable
isotope probing with Mn
13
CO
3
as
the
sole source of energy and carbon, after oxidation and growth
by cultures that had been initiated with a 10% inoculum from an unlabelled culture. If cells in the
culture had engaged in
carbo
n
fixation during Mn(II) oxidation, the
13
C
-
dissolved inorganic carbon
released from the substrate would be the source of label incorporated into the biomass. Before
examining for such, the extent of Mn residuum remaining in the samples was examined by
nan
oSIMS. This was done in order to assess the effectiveness of the
Mn oxide nodule
dissolution
procedure on a fine scale, and to rule out any colocalization of Mn with
13
C that would indicate
that unreacted Mn
13
CO
3
remained in the analysed samples. If associ
ated with cells, unreacted
Mn
13
CO
3
would conflict with any proper interpretation on the analysis of the composition of the
biomass, but none was observed (
Extended Data Fig. 9v
).
Because of the high
13
C amount in our labelled sample, to better resolve the
15
N
12
C
secondary ion from that of the
14
N
13
C (both having mass 27):
13
C
12
C and
12
C
12
C secondary ions
were
measured
instead
13
C and
12
C ions for the carbon stable isotopes. Since there are two possible
combinations to obtain
13
C
12
C ion of mass 25,
25
R is twice of
13
R based on probability (P):
R
"#
=
C
$%
C
$"
C
$"
C
$"
=
P
(
C
$%
)
∙
P
(
C
$"
)
+
P
(
C
$"
)
∙
P
(
C
$%
)
P
(
C
$"
)
∙
P
(
C
$"
)
=
2
∙
P
(
C
$%
)
P
(
C
$"
)
=
2
∙
R
$%
Carbon isotope was calculated as follows:
13
R =
13
C/
12
C =
25
R/2
13
F =
13
C/(
12
C+
13
C) =
13
R/(1+
13
R)
Atom percent
13
C =
13
F
⋅
100
Secondary ions
14
N
12
C and
15
N
12
C were measured on the nanoSIMS for nitrogen stable
isotopes
14
N and
15
N respectively.
Nitrogen isotope was calculated as follows:
15
R =
15
N/
14
N =
15
N
12
C/
14
N
12
C
15
F =
15
N/(
14
N+
15
N) =
15
N
12
C/(
14
N
12
C+
15
N
12
C) =
15
R/(1+
15
R)
Atom percent
15
N =
15
F
⋅
100
12
The application of nanoSIMS
revealed that both species fixed Mn
13
CO
3
-
derived inorganic
carbon into cell material. In cell preparations
hybridized
with species specific fluorescent
oligonucleotides, the
13
C atom percent in Mn
13
CO
3
-
grown biomass was determined to be 35.57%
(s.d=9.57,
n=107) and 7.93% (s.d=5.62, n=27) for cells of Species A and Species B, respectively.
These values are strikingly higher than that of
13
C natural abundance. Cells grown under the same
conditions, but with MnCO
3
synthesized with unlabelled inorganic carbon,
were of
13
C natural
abundance as expected (
Extended Data Fig. 9a
). Moreover, these values almost certainly
underestimate of the full extent of incorporation of inorganic C by both organisms (see discussion
on reagents, below). Taking into account the high
er
13
C atom percent and the markedly higher
abundance of Species A cells in the co
-
culture (
Extended Data Fig. 3a
), the results indicate that
the bulk of the labelled chemosynthate
is associated with, and can be attributed to, the activities of
Species A.
To better track the synthesis of new biomass, the incorporation of
15
N
-
nitrate and its
colocalization with
13
C was also examined (and contrasted with the co
-
localization of Mn and
13
C,
above) during stable isotope probing analyses. The
15
N atom percent in cells were determined to
be 57.59% (s.d=8.20, n=107) and 15.77% (s.d=13.30, n=27) for Species A and Species B,
respectively (
Extended Data Fig. 9a
). These results are also consist
ent with Species A having
generated the bulk of the new biomass in the co
-
culture, given its
higher
15
N atom percent
(
Extended Data Fig. 9a
)
and higher cell abundance (
Extended Data Fig. 3a
). Such results are not
inconsistent with Species A having reductiv
ely assimilated the nitrate into its cell material. Yet
genomes predict interspecies N crossfeeding from Species B to Species A when growing with
nitrate: Species A lacks a recogni
s
able pathway for assimilatory nitrate reduction (
Supplementary
Table 4
), wh
ereas Species B is predicted to be capable of both assimilating and dissimilating nitrate
(
Supplementary Table 5
). Inspection of the FISH coupled nanoSIMS
12
C
14
N ion image revealed
that while
Species B cells can have higher
14
N contents,
Species A cells ha
ve higher
15
N and a
greater ratio of
15
N/
14
N (
Extended Data Fig. 9b
-
u
;
Fig.
4
). This indicates that either Species B
reduces nitrate and shares much of the reduced nitrogen with Species A without incorporating to
the same degree itself, or that Species A e
ncodes an unrecogni
s
ed enzyme for assimilatory nitrate
reduction.
It is worthnoting that
Nitrospira inopinata
, a relative of Species A in the phylum
Nitrospirae
, does not assimilate either nitrate or nitrite
9
. In any event, the nanoSIMS
results
confirms that nitrate can serve as the N source for both species in the co
-
culture, but the
13
interspecies interaction and effect of ammonia (
Supplementary Note 6 above
) on the co
-
culture
remain mysterious.
Different preparation methods for samples
of cells prior to nanoSIMS analyses can often
dilute the stable isotope labels in the biomass, resulting in an underestimate of the actual amounts
of
13
C and
15
N that had been incorporated
117
. We
therefore also
performed nanoSIMS analyses on
cells without paraformaldehyde fixation and
FISH
. T
he resulting atom percent for
13
C and
15
N in
Species A cells increased to 56.56% (s.d=6.51, n=394) and 69.01% (s.d=6.34, n=394), respectively
(
Extended Data Fig. 9a
). There a
re, however, two main steps in nanoSIMS sample preparation that
remain that could dilute the stable isotope label: 1) the dissolution procedure employed to obtain
cells from the Mn oxide nodules contained unlabelled reagents (0.05 M sodium dithionite, 0.1
M
sodium citrate, 0.1 M sodium bicarbonate and 0.1 M EDTA), and 2) the DAPI
-
citifuor staining
and localization of cells prior to nanoSIMS analysis. Both procedures can dilute the
13
C isotopic
label in the biomass, thus even these
13
C and
15
N atom percent v
alues (
Extended Data Fig. 9a
)
likely remain as underestimates of the actual degree of authentic
label
incorporation that had
occurred.
It is not possible from these analyses to ascertain whether or not absolutely 100% of all
biomass carbon in the co
-
cultu
re is derived from CO
2,
and thus
chemosynthetic in nature. However,
such a degree of incorporation is not necessary to conform to one existing definition of autotrophy
8
,
i.e. that a majority of cell carbon is derived from CO
2
-
fixation. In such a contex
t and in
combination
with
other evidence
,
including the
composition of the
cultivation medium and
the
expressed CO
2
-
fixation pathway
genes
(
M
ain
T
ext and below), we propose that Species A is
operating as a chemolithoautotroph in the
manganese
-
oxidising
co
-
culture.
Supplementary Note 10:
Carbon assimilation pathways in Species A
All components in the reverse tricarboxylic acid (rTCA) cycle to produ
ce acetyl
-
CoA from
2 CO
2
were identified in the Species A genome (Fig.
3b
; Supplementary Table
4
), except for one
of the subunits of fumarate reductase. The fumarate reductase of Species A conforms to the Type
E quinone:fumarate oxidoreductases, which cont
ain a catalytic flavoprotein subunit A (SdhA) and
an iron
-
sulfur subunit B (SdhB), but here the canonical membrane subunits C and D (SdhC/D)
have been replaced by non
-
homologous subunits E and F (SdhE/F)
118
. SdhE with a duplicated
motif rich in cysteines CX31
-
35CCGX38
-
39CX2C, homologous to heterodisulfide reductase
14
subunit B, was found in Species A as
well as other species in the phylum including
Nitrospira
and
Leptospirillum
38,41,119
. However, the small me
mbrane anchor (SdhF)
that is thought to link the
protein complex to the cell membrane is missing in Species A and its relatives in the phylum, as
has been noted previously
38,41
. The genes in the rTCA cycle of Species A were found to be
expressed, including the oft cited key genes for ATP
-
citrate lyase, oxoglut
arate:ferredoxin
oxidoreductase (of the O
2
tolerant form with 5 subunits) and fumarate reductase
42
(
Supplementary
Table 4
). Carbon assimilation from pyruvate is thought to proceed via the Embden
-
Meyerhof
-
Parnas (EMP) pathway in
Nitrospira
and
Leptospirillum
38,41,119
. However, one of the key genes
for the EMP pathway that is found in the related genomes, Class I f
ructose
-
bisphosphate aldolase,
was not found in the Species A genome. This may or may not be due to any incompleteness in the
genome assembly. A homolog of Class II fructose
-
bisphosphate aldolase (IMG gene ID
2784403887) was found in Species A, but sequenc
e comparisons with characterized class II
fructose/tagatose aldolases reveal significant differences, thus whether its substrate is fructose or
not remains unclear. In principle, synthesis of hexose sugars from pyruvate can proceed through
the Entner
-
Doudo
roff (ED) pathway
;
however
,
Species A is missing a recogni
s
able key gene for
6
-
phosphogluconate dehydratase. A dihydroxy
-
acid dehydratase
101
homolog of 6
-
phosphogluconate dehydratase can be identified in the Species A genome (IMG gene ID
2784405073), but a phylogenetic analysis building on that of a previous study
101
suggest that it is
not a good candidate for encoding missing ED pathway e
nzyme (
Extended Data Fig. 5d
). Thus
from genomics, it is not entirely clear how the biosynthesis of hexoses from pyruvate proceeds in
Species A.
In addition to the rTCA cycle, functions of relevance to a partial oxidative TCA (oTCA)
cycle could also be ide
ntified in Species A, including citrate synthetase, pyruvate carboxylase, and
pyruvate dehydrogenase (Fig.
3b
;
Supplementary Table 4
). The only exception in the oTCA cycle
not found in Species A is the 2
-
oxoglutarate dehydrogenase (OGDH) complex, which is
belongs
to the protein family of 2
-
oxo acid dehydrogenase complexes
120
. All known E1 component of
OGDHs are homodimers, but all 5 homologs of the E1 component in the
Species A genome were
heter
o
tetrameric versions found in pyruvate dehydrogenase and branched
-
chain 2
-
oxo acid
dehydrogenase complexes but not OGDH complexes
120
.
No OG
DH with homodimeric versions of
the E1 component were identified in characterized relatives in the phylum
Nitrospirae
such as
Ca.
Nitrospira defluvii
38
, or the genomes of
Nitrospira inopinata
and
Nitros
pira moscoviensis
. The 2
-
15
oxoglutarate:ferredoxin oxidoreductase that typical of rTCA cycle has been hypothesized to
substitute for this enzymatic step in oTCA cycle with the production of low
-
potential ferredoxin
rather than NADH
38
, but this scenario would be energetically challenging, if possible at all, under
physiological conditions.
A
more
likely scenario is that the oTCA
cycles in Species A and
members of the phylum
Nitrospirae
are incomplete, as they are so in many autotrophs and non
-
respiratory, fermentative bacteria. In that case, a partial oTCA branch is probably only for
anabolism to generate key intermediate such as
2
-
oxoglutarate from pyruvate, rather than for
respiration and acetate mineralization. Correspondingly, characterized
bacteria in the phylum
Nitrospirae
48,121
either cannot grow on organic compounds other than formate as the electron donor,
or they are incomplete oxidisers
122
producing acetate rather than CO
2
.
Of genes of the EMP, rTCA, and oTCA pathwa
ys
that
were identified, all were expressed.
Of the carbon metabolic genes, the most highly expressed in Species A were aconitate hydratase,
ATP
-
citrate lyase and the genes for the three complexes catalyzing CO
2
fixation steps (2
-
oxoglutarate:ferredoxin ox
idoredu
c
tase, isocitrate dehydrogenase and pyruvate:ferredoxin
oxidoreductase; Fig.
3b
). Thus, gene expression under the Mn(II)
-
oxidation
-
dependent growth
conditions is consistent with CO
2
fixation via the rTCA cycle.
Supplementary Note 11:
Unrestricted
m
anganese
oxi
dation
c
alculation
The world manganese ore reserve is estimated to be 630 million metric tonnes
123
. Starting
with a single cell each of Species A and B, and u
sing the estimated doubling time of 6.1 days for
Species A (Fig. 2
b
) and Mn(II) oxidation rate of 1.188
⋅
10
-
11
g substrate
⋅
cell
-
1
⋅
day
-
1
(converted
from 9.0
⋅
10
-
15
mol substrate
⋅
cell
-
1
⋅
hr
-
1
,
Extended Data Fig. 3g
), the time (t) needed to generate
MnO
2
equall
ing global Mn reserves with unrestricted chemolithotrophic growth could be
estimated by the equation below, which would be within 2 years:
푀푛
(
퐼퐼
)
표푥푖푑푖푠푒푑
=
:
푁
&
∙
k
∙
2
'
(
)
푑휏
*
&
where,
N
0
= starting cell number
t = time needed
k = Mn(II)
oxidation rate
g = doubling time