Document [original]

Enzyme Mechanisms

Coupling CO2Reduction and Acetyl-CoA Formation: The Role of a

CO Capturing Tunnel in Enzymatic Catalysis

Jakob Ruickoldt, Jae-Hun Jeoung, Maik Alexander Rudolph, Frank Lennartz,

Julian Kreibich, Reinhard Schomäcker, and Holger Dobbek*

Abstract: The bifunctional CO-dehydrogenase/acetyl-

CoA synthase (CODH/ACS) complex couples the

reduction of CO2to the condensation of CO with a

methyl moiety and CoA to acetyl-CoA. Catalysis occurs

at two sites connected by a tunnel transporting the CO.

In this study, we investigated how the bifunctional

complex and its tunnel support catalysis using the

CODH/ACS from Carboxydothermus hydrogenofor-

mans as a model. Although CODH/ACS adapted to

form a stable bifunctional complex with a secluded

substrate tunnel, catalysis and CO transport is even

more efficient when two monofunctional enzymes are

coupled. Efficient CO channeling appears to be ensured

by hydrophobic binding sites for CO, which act in a

bucket-brigade fashion rather than as a simple tube.

Tunnel remodeling showed that opening the tunnel

increased activity but impaired directed transport of

CO. Constricting the tunnel impaired activity and CO

transport, suggesting that the tunnel evolved to seques-

ter CO rather than to maximize turnover.

Introduction

The reductive acetyl-CoA pathway allows the simultaneous

fixation of two molecules of CO2to acetate, generating ATP

in the process.[1,2] It is one of the most energy-efficient

biological carbon fixation pathways, making it a promising

candidate for the production of renewable fuels. The CO-

dehydrogenase/acetyl-CoA-synthase (CODH/ACS) com-

plex catalyzes the final step of this pathway: the formation

of acetyl-CoA from CO2, reducing equivalents, a methyl-

cation, and coenzyme A. Bacterial CODH/ACS complexes

consist of a CODH-homodimer carrying two ACS subunits

on each side.[3–6]

The synthesis of acetyl-CoA takes place in the enzyme

complex in two partial reactions (reviewed by Can et al.[7]).

First, CO2is reduced to CO at the C-cluster (a [NiFe4(OH)-

(μ3-S)4] cluster) of the CODH subunit. The generated CO

then travels through a 70 Å long tunnel to the A-cluster in

the ACS subunit. The A-Cluster is a Ni,Ni-[4Fe4S] cluster

which catalyzes the condensation of CO, a methyl-cation

(donated by the corrinoid-FeS-protein (CoFeSP) and

CoA.[4,8,9] A scheme of the proposed catalytic cycle at the A-

cluster can be found in Figure S1.

The ACS subunit can adopt at least two conformations

that affect the CO tunnel: open and closed. In the open

conformation the A-cluster is solvent-exposed and the

tunnel between the C-cluster and the A-cluster is closed. In

the closed conformation the A-cluster is buried and the

tunnel is continuous.[3,4] So far, it is not known what triggers

the switch between these conformations.

The presence of the tunnel in the CODH/ACS complex

was proposed even before structure information was

available.[10,11] This was based on the fact that, despite the

very high affinity of hemoglobin for CO, hemoglobin did

not strongly inhibit the channeling of CO into acetyl-CoA.

The presence of a hydrophobic tunnel was later confirmed

by the first crystal structures.[3,4,12]

The role of the tunnel in catalysis has been investigated

in several studies using the CODH/ACS complex from

Moorella thermoacetica.[13–15] Constrictions in the tunnels

connecting the two C-clusters in the CODH-homodimer to

each other and to solvent decreased CO oxidation activity

and acetyl-CoA synthesis from CO2.[14] However, constric-

tions in the tunnel connecting the C-cluster and the A-

cluster did not severely affect the CO oxidation activity but

almost eliminated the production of acetyl-CoA from

CO2.[13] In addition, the activity in the condensation reaction

was no longer inhibited by CO in the narrow-tunnel variants

but was significantly lower than the wild-type activity. This

led to the dual-pathway hypothesis,[14] which states that the

CO coming through the tunnel reacts in the correct step of

catalysis while the CO approaching the A-cluster from the

[*] Dr. J. Ruickoldt, Dr. J.-H. Jeoung, Dr. J. Kreibich,

Prof. Dr. H. Dobbek

Humboldt-Universität zu Berlin

Institut für Biologie

Unter den Linden 6, 10099 Berlin, Germany

E-mail: holger[email protected]

M. A. Rudolph, Prof. Dr. R. Schomäcker

Technische Universität Berlin

Institut für Chemie - Technische Chemie

Straße des 17. Juni 124, 10623 Berlin, Germany

Dr. F. Lennartz

Helmholtz-Zentrum Berlin

Macromolecular Crystallography

Albert-Einstein-Straße 15, 12489 Berlin, Germany

published by Wiley-VCH GmbH. This is an open access article under

the terms of the Creative Commons Attribution Non-Commercial

License, which permits use, distribution and reproduction in any

medium, provided the original work is properly cited and is not used

for commercial purposes.

Angewandte

Chemie

Research Article www.angewandte.org

How to cite: Angew. Chem. Int. Ed. 2024,63, e202405120

doi.org/10.1002/anie.202405120

solvent inhibits catalysis by binding non-productively in an

earlier catalytic step. In what ways catalysis benefits from

carrying out the coupled reaction within a stable bifunc-

tional protein complex harboring a long gas tunnel in

contrast to two monofunctional enzymes, is not known.

To clarify this, we took advantage of the unique proper-

ties of the ACS from C. hydrogenoformans to act as a single,

monofunctional enzyme or to be part of a stable, bifunc-

tional complex with CODH-III.[9] We inferred the phylogeny

of the CODHs and show how the structure has adapted as

complex-forming CODHs evolved. We then assessed

whether the stable CODH/ACS complex is catalytically

superior to the monofunctional enzymes, and finally ana-

lyzed the role of the tunnel in the bifunctional complex by

site-directed mutagenesis.

Results and Discussion

Phylogeny of the CODH Subunit

We used the genomic proximity of the genes encoding

CODH and ACS as a guide to discriminate between mono-

and bifunctional CODHs (Figure S2). This approach cor-

rectly identified the sequences of currently known bifunc-

tional CODH/ACS complexes, including the M. thermoace-

tica CODH, C. hydrogenoformans CODH-III, and C.

autoethanogenum CODH. Conversely, none of the known

monofunctional CODHs were predicted to be bifunctional

or encoded by a gene proximal to a gene encoding an ACS.

The inferred phylogeny indicates that the CODH/ACS

complexes of M. thermoacetica and C. hydrogenoformans

(bifunctional type I), on the one hand, and the complex of

C. autoethanogenum (bifunctional type II), on the other

hand, evolved independently (Figure S2). This is consistent

with the various structural differences between the two

complexes, which differ in their binding interfaces, structural

motifs to stabilize the complex, and the tunnels within

CODH and connecting the active sites of CODH and ACS.

Here, we focus on the structural evolution of the Moorella /

Carboxydothermus type I bifunctional CODHs. Whereas

monofunctional CODHs have gas tunnels that span the

entire dimer and connect the active site cluster to the

molecular surface, bifunctional type I CODHs have blocked

channels to the surface and CO2diffusion into the active site

requires structural dynamics that control substrate access to

the C-cluster.[16] We followed the evolution of the residues

lining the tunnel of monofunctional CODHs and found that

the smaller side chains near the tunnel exit found in

monofunctional CODHs, such as CODH-II and the CODH

from Rhodospirillum rubrum, are present in the last

common ancestor of monofunctional CODHs and bifunc-

tional type I CODHs (AncB; see Figure 1, Figure S3), but

that these residues were replaced by bulkier side chains,

predominantly phenylalanines in the transition to the last

common ancestor of bifunctional CODHs (AncC; Figure 1,

Figure S3). These bulky side chains block the tunnel exit,

closing the connection between the C-cluster and the

surface.

Ancestral sequence reconstruction points to a second

hotspot of changes in the transition from mono- to bifunc-

tional CODHs (AncB to AncC, Figure 1). In the structures

of the CODH/ACS complex from M. thermoacetica and C.

hydrogenoformans, the N- and C-terminal extensions of the

CODH subunit interact closely with the ACS subunit,

stabilizing the complex. These extensions began to evolve

during the transition from the common ancestor of mono-

functional CODHs (AncB) to the ancestors of bifunctional

CODHs (AncC), with the C-terminal helix extending before

the N-terminal loop extension (Figure 1). These structural

elements are absent from the common ancestor of mono-

functional and bifunctional CODHs (AncB), which is more

similar in length to extant monofunctional CODHs (Fig-

ure S4).

What is the Advantage of a Bifunctional Complex?

We analyzed the transport of CO during catalysis using a

competition experiment. Here, the synthesis of acetyl-CoA

from CO2was followed in the presence of hemoglobin,

which binds free CO with high affinity (Kd=84 nM, Fig-

ure S5). For the CODH/ACS complex, the carboxy-hemo-

globin complex (Hb-CO) formed only after the rate of

acetyl-CoA synthesis ceased due to the consumption of

methylcobinamide and CoA (Figure 2b). Thus, CO was

channeled into acetyl-CoA as long as all substrates were

present, as found before.[10,11] However, we wondered if a

tight bifunctional complex is essential for efficient channel-

ing of CO into acetyl-CoA. Therefore, we performed the

same experiment with the monofunctional CODH-II and

the isolated ACS subunit. Surprisingly, even with the

monofunctional enzymes, Hb-CO formed only after acetyl-

CoA synthesis had stopped (Figure 2d). Apparently, the A-

cluster has a higher affinity towards CO than hemoglobin.

Figure 1. Phylogeny of type-I bifunctional CODHs. A, B and C denote

three ancestral states. Their sequence lengths are shown in green,

black numbers near internal nodes are non-parametric bootstrap

scores, showing that the proposed clades are well supported. CODH-

IICh, CODH-IVCh and CODHRr are exemplary monofunctional CODHs,

CODH-IIICh and CODHMt bifunctional CODHs. Lower case Ch denotes

Carboxydothermus hydrogenoformans CODHs, Rr stands for Rhodospir-

illum rubrum and Mt for Moorella thermoacetica. Numbers within the

collapsed clades give the number of sequences in the clade. Change in

color from violet to orange denotes the hypothetical switch from

monofunctional to bifunctional CODHs with the character state

changes shown above the branches. For the complete phylogeny, see

Figure S2.

Angewandte

Chemie

Research Article

To test this, we performed ITC experiments (Figure S6). In

the reduced state, the A-cluster has a Kdfor CO of about

750 nM, which is not enough to outcompete hemoglobin.

However, in the methylated state the Kddrops to 12 nM,

probably due to the immediate insertion of the CO into the

NiCH3bond.[17] This is about 8-times lower than the Kdof

hemoglobin determined here (Figure S5). Still, even such a

low affinity of the methylated A-cluster for CO is insuffi-

cient to explain the observed kinetics: due to the approx-

imately 14-fold excess of hemoglobin over ACS in the assays

(assuming a degree of maturation of 50% for ACS), Hb-CO

should still be able to compete with methylated ACS for CO

binding.

Three hypotheses may explain this observation: First,

CODH-II and ACS form a complex with a continuous gas

tunnel, second, the binding of CO to the A-cluster is much

faster than that to Hb, or third, there is a partition

equilibrium for CO being in water or in the hydrophobic

tunnel creating a pool of CO connected to the A-cluster but

inaccessible to hemoglobin. The first hypothesis is unlikely,

as size-exclusion co-injection experiments of ACS and

CODH-II showed neither a peak corresponding to a stable

complex nor a change of retention times or peak shapes

indicative of transient complex formation (Figure S8).[18]

Furthermore, CODH-II lacks the structural elements that

have evolved in the bifunctional CODHs to stabilize a

complex with ACS. While we cannot completely rule out

the formation of weak transient CODH-II/ACS complexes,

the majority of CODH-II molecules would be uncomplexed

and dominate CO generation. At this point, we cannot

distinguish between hypothesis 2 and 3 and will return to

this question in the section “CO transport in the variants”.

Remodeling the Intramolecular CO Tunnel

To understand why such a long and complex tunnel evolved

in CODH/ACS, we used site-directed mutagenesis to

investigate its role in catalysis. Two amino acid exchanges in

the ACS subunit (A225L and A268M) were designed to

restrict the tunnel between the C-cluster and the A-cluster,

two others (F231A in CODH-III and F515A in ACS) were

designed to open the tunnel system and finally, the

combination of a restricting and an opening double

exchange (A225L/F231A) was investigated. All variants

were structurally and kinetically characterized.

We previously reported the structures of the wild-type

complex and the F231A variant at resolutions of 2.04 Å

(WT) and 2.07 Å (F231A).[6] In this study, we resolved the

crystal structures of the remaining variants at resolutions of

2.60 Å (A225L), 2.62 Å (A268M), 2.22 Å (F515A), and

2.10 Å (A225L/F231A)(See Table S3 for statistics of the X-

ray diffraction data and model refinement).

The overall structures of the tunnel variants and wild-

type CODH/ACS are indistinguishable. The RMSD of Cα-

atoms positions between the wild-type and variant structures

were not greater than the estimated coordinate error

(Table S1). With the exception of F515A, the environment

at the mutation site was not perturbed (Figure S9). The

F515A exchange resulted in a slight rearrangement of a

loop, which in turn caused a movement of a neighboring

arginine (Arg250; Figure S9b). However, this arginine is on

the surface of the protein and its rearrangement does not

cause any further changes in the tunnel. Indirect effects on

the tunnel dynamics, such as changes of the conformational

equilibrium between the open and closed conformation of

ACS in solution, cannot be ruled out.

The tunnel system was analyzed computationally using

CaverAnalyst 2.0[19] with a probe radius of 0.9 Å (Figure 3).

For the wild-type complex, in addition to the tunnel

connecting the A-cluster and the C-cluster, exits were

detected at the CODH/ACS interface and within the

CODH dimer (Figure 3d). The tunnel with the largest

bottleneck radius (1.03 Å) was the tunnel connecting the A-

cluster A and the C-cluster. The smallest cross-section of

CO measures 1.79 Å,[20] showing that conformational fluctu-

ations are necessary for CO transport, and thus the

relevance of each tunnel cannot be readily determined from

a static crystal structure. However, the computed tunnel

connecting the A-cluster and the C-cluster almost perfectly

overlays with the Xenon binding sites found in the crystal

structures of the homologous M. thermoacetica complex

(Figure S10).[4,12] Furthermore, the Fo-Fc-map shows negative

density in the tunnel region (Figure S10), indicating that the

concentration of disordered water molecules in the tunnel is

lower than in the bulk solvent. Disordered water molecules

are generally assumed to surround protein crystal structures

and are therefore taken into account in the bulk-solvent

correction to calculate Fo-Fc-maps.[21] These features are in-

Figure 2. CO transport during catalysis for the CODH/ACS complex

variants and the isolated ACS subunit coupled with CODH-II. The

competition experiment was always carried out in 0.1 M Mops/NaOH

pH 7.2 supplemented with 0.3 mg/L carbonic anhydrase, 7 μM Hb,

200 μM CoA, 50 μM methylcobinamide, 10 mM Ti(III)-EDTA, and

1.2 mM CO2, if not otherwise stated. a) Structure of the CODH/ACS

complex (PDB 7ZKJ). ACS red, CODH-III blue. b) Competition experi-

ment traces for the CODH-III/ACS complex (0.22 μM). c) CODH-II

(light and dark blue, PDB 4UDX) and the isolated ACS subunit (red,

PDB 7ZKJ). d) Competition experiment traces for the coupled action of

ACS (1 μM) and CODH-II (150 nM).

Angewandte

Chemie

Research Article

line with a highly hydrophobic tunnel between the A-cluster

and the C-cluster.

Overall, the mutations altered the tunnel system as

intended. The A268M and the A225L exchanges resulted in

a narrowing of the tunnel, which in the case of A225L

blocked the path from the C-cluster to the A-cluster

(Figure 3a–c). The opening at the A-cluster due to the

F515A exchange resulted in a widening of the tunnel system,

whereas the F231A exchange opened a new tunnel branch

connecting the tunnel to the solvent, as indicated by the

presence of 2-methyl-2,4-pentanediol (MPD) molecules

derived from the mother liquor in the tunnel system. And

the A225L/F231A double mutant had the combined charac-

teristics of both exchanges (Figure 3g–h).

Catalytic Activity of the Tunnel Variants

The catalytic activities in both partial reactions and in the

coupled reaction (synthesis of acetyl-CoA from CO2) were

analyzed as a function of substrate concentration (Figure 4;

Figures S11–S15). All tunnel variants were active catalysts

for both CO2reduction and CO condensation. Furthermore,

all variants were able to synthesize acetyl-CoA with CO2as

the educt (Figure 5, Figure S16). Therefore, a continuous

tunnel without branches leading to the solvent (as in the

F231A exchange) is not essential for catalysis.

To compare the catalytic activities in the partial

reactions, we normalized the values according to the degree

of maturation of the C- and A-clusters, as determined by X-

ray crystallography (Table S2, see Materials and Methods

section for more information). In a previous study, we could

Figure 3. The tunnel system in the CODH/ACS complex variants. The structures are shown as surfaces colored light (ACS) and dark gray (CODH).

The tunnel system is shown as a tube, the metal cluster (labeled in panel d), the MPD molecules (orange) and the respective exchanged amino

acids (either bright green or light blue) are shown as spheres. Fe atoms are colored orange, S, yellow, Ni, green. Note that all structures possess a

two-fold rotational symmetry and the rotation axis is located perpendicular to the paper plane at the D-cluster. Tunnels were computed using

CAVER analyst 2.0[19] with a probe radius of 0.9 Å. a,b) Structure and tunnel system of the variants with an opened tunnel: F515A (a) and F231A

(b). c) Comparison of the radius of the tunnel connecting the A-cluster and the C-cluster for the wild type (gray) and for the variants with an

opened tunnel (same color as tunnel in the structures). The radius of CO is shown as black line. The position of important features and the

locations of the A- and the C-clusters are indicated by labels. d,e) Structure and tunnel system of the wild type (d) and the A225L/F231A variant

(e). f) Comparison of the radius of the tunnel connecting the A-cluster and the C-cluster for the wild type and for the A225L/F231A variant (labeling

and coloring as panel c). g,h) Structure and tunnel system of the variants with a constricted tunnel: A268M (g) and A225L (h). i) Comparison of

the radius of the tunnel connecting the A-cluster and the C-cluster for the wild type and for the variants with a constricted tunnel (labeling and

coloring as panel c).

Angewandte

Chemie

Research Article

show that the occupancy of the Ni atom of the C-cluster

almost matches the relative CO oxidation activity compared

to that of CODH-III/ACS directly isolated from C.

hydrogenoformans.[6,9] The average occupancy of the Ni

atom in the C-cluster was 36.5�9.3%. The average ratio of

the Ni anomalous density at the proximal to that at the

distal position is 0.38�0.09, which can be translated into the

percentage of Ni assuming 100% Ni occupancy in the distal

position and similar B-factors at both positions.

In CO2reduction, the variants with an additional tunnel

entrance due to the F231A exchange were more active than

the wild type. The A225L variant with a blocked tunnel had

about 50% of the wild-type activity. The A225L/F231A

variant also showed a decrease in CO2reduction activity of

about 50% compared to the F231A variant, demonstrating

that the tunnel blockage by L225 reproducibly decreases the

CO2reduction activity. However, the F515A and the

A268M variants had similar CO2reduction activities as the

wild-type, indicating that the tunnel opening at the A-cluster

and the slight constriction by M268 had little effect on CO2

reduction. The CO2reduction activity of the CODH/ACS

complex seems to be regulated by the bottlenecks of the

tunnel system, as it can be increased or decreased by

manipulating the tunnel.

In the CO condensation reaction, the A268M and F231A

variants had significantly higher activity than the wild type

(determined as Vlim at infinite CO concentration, see Supple-

ment for more information). The A225L/F231A variant had

significantly lower activity. Unfortunately, we cannot ration-

alize the different activities in CO condensation. However,

they may be due to altered dynamics of the ACS subunit,

which is essential for catalysis.[4,22] Nevertheless, all variants

of the complex (Supporting Information) and the isolated

ACS subunit (Figure 4f) showed the same response pattern

to CO concentration: The rate of the condensation reaction

decreased steadily with increasing CO concentration, reach-

ing Vlim at infinite CO concentration. This indicates partial

substrate inhibition of the enzyme by CO, where CO is

bound with high affinity in an active state and with low

affinity in a less active state, as has been previously

described for the M. thermoacetica CODH/ACS

complex.[23,24] These observations, together with the ITC

data on the binding of CO to ACS, suggest that the

methylated state is most likely the active, high-affinity state

and the reduced state is the less active, low-affinity state that

inhibits catalysis.

The characterization of the partial reaction also allows

us to determine the rate-limiting step in the synthesis of

acetyl-CoA from CO2. The activity in the coupled reaction

correlated more with the activity in the reduction of CO2

(Figure 5) than with the activity in the CO condensation

reaction. Furthermore, during acetyl-CoA synthesis, only

low concentrations of the intermediate CO can build up

transiently, as no CO leakage from the tunnel is

observed.[10,11] The condensation reaction activity at low CO

concentrations is much higher than the maximum CO2

reduction rate for all variants. These two points support that

in our experimental setup, CO2reduction was the rate-

limiting step in the coupled reaction. We did not observe

activation of CO2reduction when coupled to acetyl-CoA

synthesis, as reported for the CODH/ACS complex of M.

thermoacetica.[23]

Besides the effects on the activities, the tunnel manipu-

lation also changed the Kmvalues for CO2and the electron

donor Ti(III)-EDTA in both CO2reduction and acetyl-CoA

Figure 4. Catalytic activities of the wild type CODH-III/ACS complex

and the isolated ACS subunit. All data were measured in 100 mM

Mops/NaOH pH 7.2 supplemented with 3 mg/L carbonic anhydrase at

50°C. For acetyl-CoA synthesis, the buffer was further supplemented

with 200 μM CoA and 50 μM methylcobinamide. For the condensation

reaction also 1 mM of Ti(III)-EDTA was present. The lines are fits with

the equation shown in the materials and methods section. Relevant

parameters and substrate concentrations are depicted in the plot.

[CO2]0denotes the intrinsic CO2concentration of the assay buffer. TE

stands for Ti(III)-EDTA. a,b) Dependence of the rate of acetyl-CoA

synthesis from CO2on either [CO2] (a) or [Ti(III)-EDTA] (b). c,d)

Dependence of the CO2reduction rate on either [CO2] (c) or [Ti(III)-

EDTA] (d). e,f) Dependence of the rate of the condensation of CO to

acetyl-CoA on [CO] for the CODH/ACS complex (e) or the isolated ACS

subunit (f).

Figure 5. Normalized CO2reduction activities and correlation for the

tunnel mutant. The activities were normalized on the wild type activity

and the degree of cluster maturation. The activity of the wild type

CODH-III/ACS with fully matured clusters equals 1. Error bars were

calculated assuming a 5% error for the estimated maturation degree

using the error of the fit for the kinetic parameters. a) CO2reduction

activity of the variants. b) Correlation plot of the Vmax in the acetyl-CoA

synthesis from CO2and the Vmax in CO2reduction. The shaded area in

panel a indicates the wild-type activity and in panels b the region of the

plot in which Vmax in the acetyl-CoA synthesis from CO2is smaller than

that of CO2reduction. The V-[S]-characteristics and the respective

results for the Kmand Kivalues for all variants can be found in the

Supporting Information (Figures S11–S15).

Angewandte

Chemie

Research Article

synthesis (Figure S17a). Of the two apparent Km,TiEDTA and

Km,CO2, only the latter is physiologically relevant. Km,CO2 was

similar in the coupled and in the partial reaction and

differed in the CO2reduction by almost an order of

magnitude between the variants (Figure S17b).

CO Transport in the Variants

Finally, we analyzed the gas transport for the tunnel variants

(Figure 6). All substitutions decreased the efficiency of CO

channeling into acetyl-CoA (Figure 6b). We quantified the

tightness of CO transport by the percentage of CO-free

hemoglobin in the steady-state phase of the reaction (Fig-

ure 6c).

While the F515A variant was as efficient as the wild-

type, the A225L and A268M variants were much less

efficient at channeling CO into acetyl-CoA and showed a

biphasic appearance of Hb-CO in the reaction course.

Variants carrying the F231A substitution could not prevent

the loss of CO, as Hb-CO was already formed after the start

of the reaction.

While it is readily understandable that an opening

caused by the F231A exchange reduces the gas-tightness of

the tunnel and thus impairs directed CO transport, the

effects of the A225L and A268M exchanges cannot be

explained by a simple gas-tight tube model. However, an

overlay of the exchanged positions with the structure of the

M. thermoacetica complex from xenon-pressurized

crystals[4,12] may explain the findings. The overlay indicates

that the A225L and A268M exchanges may block access to

isolated hydrophobic binding sites for Xe in the tunnel

(Figure S10, inset).

Thus, the tunnel might be seen as a bucket brigade of

CO binding sites rather than as a simple tube connecting the

C-cluster and the A-cluster. The A225L and A268M

exchanges, or solvent molecules entering the tunnel through

the F231A hole, might block access to some of the binding

sites of this bucket brigade and thus affect the overall

efficiency of transport. The biphasic appearance of Hb-CO

for the A225L and A268M variant might be due to slightly

lowered affinity of the hydrophobic binding sites in the

tunnel: At reaction start, CO preferably binds to hemoglo-

bin and upon partial saturation of hemoglobin preferably in

the tunnel.

Furthermore, this experiment provides a clue to the

observed coupled kinetics of the monofunctional enzymes

where we were wondering, what caused the preferential

channeling of CO into acetyl-CoA: Fast binding of CO at

the A-cluster or pre-concentration and sequestration of CO

in the tunnel? If CO binding to the A-cluster would be

much faster than to hemoglobin, we wouldn’t expect the

A225L, A268M and F231A variants to perform worse than

the wild-type in CO-channeling. In these variants the CO

binding rate to the A-cluster is probably not affected by

these mutations far from the A-cluster. Thus, if these

mutations would just simply create cracks in the tube, CO

could be still channeled to the A-cluster. This is not the

case. Thus, the rate of binding to the A-cluster is not the

reason for the channeling of CO and it might be rather the

formation of a separate CO pool through hydrophobic

binding in the tunnel, which is sequestered from bulk water

and thus from hemoglobin.

Overall, as the A268M, A225L and the F231A exchanges

led to a decreased channeling efficiency, while splitting the

complex did not, we conclude that the blockade of hydro-

phobic binding pockets in the tunnel by bulky amino acids

(A268M, A225L) or by the intrusion of solvent molecules

(F231A) caused a lower channeling efficiency.

Conclusion

Both Moorella and Carboxydothermus CODH/ACS com-

plexes are stable protein assemblies containing a tight tunnel

for CO diffusion between the two active sites. It is likely

that both traits evolved during the transition from mono- to

bifunctional CODHs. These features seem to be irrelevant

for catalytic efficiency, since the coupled reaction between

monofunctional CODH and ACS works efficiently, and

manipulations of the tunnel by opening it can even increase

the overall turnover. So, why have these stable complexes

with narrow tunnels evolved?

Figure 6. CO transport during catalysis for the CODH/ACS complex

variants. The competition experiment was always carried out in 0.1 M

Mops/NaOH pH 7.2 supplemented with 0.3 mg/L carbonic anhydrase,

200 μM CoA, 50 μM methylcobinamide, 10 mM Ti(III)-EDTA, and

1.2 mM CO2, if not otherwise stated. a) Structure and tunnel system of

the CODH/ACS complex (PDB 7ZKJ). The shape of the complex is

shown in gray. The exchanged residues are shown as sticks and labeled

at one site of the two-fold symmetric complex. The label C and A

indicate the position of the C-cluster and the A-cluster, respectively. The

presentation is otherwise the same as in Figure 3. b) Competition

experiment traces for the tunnel variants. [Hb-CO] blue, [AcCoA] red,

sum black. Deviating substrate and enzyme concentrations were

0.22 μM (WT), 0.55 μM (F515A), 0.5 μM (A268M), 0.55 μM (A225L),

and 2.32 mM CO2, 5.1 mM Ti(III)-EDTA (F231A) variant (0.38 μM

enzyme) and 0.46 μM (A225L/F231A) and ca. 7 μM Hb for all variants.

c) Tightness of the tunnel system in the variants determined as the

percentage of free hemoglobin when the acetyl-CoA production reached

the steady state.

Angewandte

Chemie

Research Article

By manipulating the tunnel, we showed that the flux

through the tunnel controls the output of CO from the

CODH subunit. Furthermore, the coupled reaction is

probably rate-limited by the reduction of CO2and not by

the condensation of CO at the A-cluster, which could occur

at much higher rates in our experimental setup. All tunnel

variants, even those with a blocked tunnel, catalyze the

coupled reaction - the synthesis of acetyl-CoA from CO2.

However, the efficiency of channeling CO into acetyl-CoA

was impaired. While there is essentially no CO leakage from

the tunnel during catalysis with the wild-type CODH/ACS,

significant leakage was found for the variants with a

constricted tunnel (A268M, A225L) and for the variants

whose tunnel was opened by the F231A exchange at the

CODH/ACS-interface. The F515A exchange near the A-

cluster left the CO transport almost unaffected.

While an increased leakiness due to an opening readily

explains our observed kinetics, the effect of the constrictions

seems to go beyond a simple hindrance of CO diffusion

within the tunnel. Rather, the constrictions reduce CO

channeling efficiency by blocking the access to hydrophobic

CO binding sites in the tunnel. This mode of CO transport is

also supported by the surprising observation that CO is

efficiently channeled into acetyl-CoA using the isolated

ACS subunit and the monofunctional CODH-II. Appa-

rently, an intact continuous tunnel between the C-cluster

and the A-cluster is not needed to achieve efficient

channeling of CO into acetyl-CoA, but rather bucket-

brigade-like transport of CO between internal hydrophobic

binding sites inside the ACS enzyme (Figure 7). Thus, the

evolution of a stable complex between CODH and ACS

may have simply extended and stabilized this transport

system.

Supporting Information

The authors have cited additional references within the

Supporting Information.[25–48]

Acknowledgements

We acknowledge access to beamlines of the BESSY II

storage ring (Berlin) through the Joint Berlin MX-Labora-

tory sponsored by Helmholtz Zentrum Berlin für Materi-

alien und Energie, Freie Universität Berlin, Humboldt-

Universität zu Berlin, Max-Delbrück-Centrum, and the

Leibniz-Institut für Molekulare Pharmakologie. This work

was funded by the Deutsche Forschungsgemeinschaft (DFG,

German Research Foundation) under Germany’s Excel-

lence Strategy—EXC 2008-390540038-UniSysCat. The au-

thors acknowledge the funding agency DFG for support of

the project through grant DO785/6-2. Open Access funding

enabled and organized by Projekt DEAL.

Conflict of Interest

The authors declare no conflict of interest.

Data Availability Statement

Coordinates and structure factor amplitudes are available in

the Protein Data Bank (https://www.ebi.ac.uk/pdbe/) under

accession numbers 8CJB, 8CMW, 8CJA and 8CJC.

Keywords: Carbon monoxide dehydrogenase ·Catalytic

coupling ·Nickel ·CO2·Substrate tunnel

[1] S. W. Ragsdale, Ann. N. Y. Acad. Sci. 2008,1125, 129–136.

[2] G. Fuchs, Annu. Rev. Microbiol. 2011,65, 631–658.

[3] T. I. Doukov, T. M. Iverson, J. Seravalli, S. W. Ragsdale, C. L.

Drennan, Science 2002,298, 567–572.

[4] C. Darnault, A. Volbeda, E. J. Kim, P. Legrand, X. Vernède,

P. A. Lindahl, J. C. Fontecilla-Camps, Nat. Struct. Mol. Biol.

2003,10, 271–279.

[5] O. N. Lemaire, T. Wagner, Biochim. Biophys. Acta Bioenerg.

2020,1862, 148330.

[6] J. Ruickoldt, Y. Basak, L. Domnik, J.-H. Jeoung, H. Dobbek,

ACS Catal. 2022,12, 13131–13142.

[7] M. Can, F. A. Armstrong, S. W. Ragsdale, 2014,114, 4149–

4174.

[8] S. Gencic, D. A. Grahame, J. Biol. Chem. 2003,278, 6101–

6110.

Figure 7. The role of the tunnel in catalysis. The tunnel is shown as a

stylized tube connecting the C- and A-cluster. The exchanged residues

are drawn schematically as white ellipsoids and the introduced residues

as yellow ellipsoids. CO and CO2are shown as spheres and the

hydrophobic binding sites as red circles. The introduction of bulky

residues might decrease the accessibility of the binding sites in the

tunnel, explaining why the constricting amino acid exchanges A225L

and A268M led to an increased leakiness of the tunnel. The exchanges

F231A and F515A open the tunnel. However, only the F231A exchange

led to a considerable leakage of CO, which could then reach the A-

cluster through the solvent. Furthermore, even with the isolated ACS

subunit, CO in solution is probably captured by the hydrophobic

binding sites, ensuring efficient channeling of CO into acetyl-CoA.

Angewandte

Chemie

Research Article

[9] V. Svetlitchnyi, H. Dobbek, W. Meyer-Klaucke, T. Meins, B.

Thiele, P. Römer, R. Huber, O. Meyer, Proc. Natl. Acad. Sci.

USA 2004,101, 446–451.

[10] E. L. Maynard, P. A. Lindahl, J. Am. Chem. Soc. 1999,121,

9221–9222.

[11] J. Seravalli, S. W. Ragsdale, Biochemistry 2000,39, 1274–1277.

[12] T. I. Doukov, L. C. Blasiak, J. Seravalli, S. W. Ragsdale, C. L.

Drennan, Biochemistry 2008,47, 3474–3483.

[13] X. Tan, H.-K. Loke, S. Fitch, P. A. Lindahl, J. Am. Chem. Soc.

2005,127, 5833–5839.

[14] X. Tan, A. Volbeda, J. C. Fontecilla-Camps, P. A. Lindahl, J.

Biol. Inorg. CheJ. Biol. Inorg. Chem.m.m. 2006,11, 371–378.

[15] X. Tan, P. A. Lindahl, J.Biol. Inorg. Chem. 2008,13, 771–778.

[16] P.-h. Wang, M. Bruschi, L. De Gioia, J. Blumberger, Journal of

the American Chemical Society 2013,135, 9493–9502.

[17] M. Can, M. J. Abernathy, S. Wiley, C. Griffith, C. D. James, J.

Xiong, Y. Guo, B. M. Hoffman, S. W. Ragsdale, R. Sarangi, J.

Am. Chem. Soc. 2023,145, 13696–13708.

[18] J. Bao, S. M. Krylova, L. T. Cherney, J. C. Y. LeBlanc, P.

Pribil, P. E. Johnson, D. J. Wilson, S. N. Krylov, Anal. Chem.

2014,86, 10016–10020.

[19] A. Jurcik, D. Bednar, J. Byska, S. M. Marques, K. Furmanova,

L. Daniel, P. Kokkonen, J. Brezovsky, O. Strnad, J. Stourac,

A. Pavelka, M. Manak, J. Damborsky, B. Kozlikova, Bio-

informatics 2018,34, 3586–3588.

[20] R. Banerjee, J. D. Lipscomb, Acc. Chem. Res. 2021,54, 2185–

2195.

[21] P. V. Afonine, R. W. Grosse-Kunstleve, P. D. Adams, A.

Urzhumtsev, Acta Crystallogr. Sect. D 2013,69, 625–634.

[22] S. E. Cohen, E. J. Brignole, E. C. Wittenborn, M. Can, S.

Thompson, S. W. Ragsdale, C. L. Drennan, Structure 2020.

[23] E. L. Maynard, P. A. Lindahl, Biochemistry 2001,40, 13262–

13267.

[24] A. Biester, S. Dementin, C. L. Drennan, J. Inorg. Biochem.

2022,230, 111774.

[25] J.-H. Jeoung, H. Dobbek, Science 2007,318, 1461–1464.

[26] U. Mueller, R. Förster, M. Hellmig, F. U. Huschmann, A.

Kastner, P. Malecki, S. Pühringer, M. Röwer, K. Sparta, M.

Steffien, M. Ühlein, P. Wilk, M. S. Weiss, Eur. Phys. J. Plus

2015,130.

[27] K. M. Sparta, M. Krug, U. Heinemann, U. Mueller, M. S.

Weiss, J. Appl. Crystallogr. 2016,49, 1085–1092.

[28] W. Kabsch, Acta Crystallogr. Sect. D 2010,66, 125–132.

[29] P. Emsley, B. Lohkamp, W. G. Scott, K. Cowtan, Acta

Crystallogr. Sect. D 2010,66, 486–501.

[30] D. Liebschner, P. V. Afonine, M. L. Baker, G. Bunkóczi, V. B.

Chen, T. I. Croll, B. Hintze, L.-W. Hung, S. Jain, A. J. McCoy,

N. W. Moriarty, R. D. Oeffner, B. K. Poon, M. G. Prisant, R. J.

Read, J. S. Richardson, D. C. Richardson, M. D. Sammito,

O. V. Sobolev, D. H. Stockwell, T. C. Terwilliger, A. G.

Urzhumtsev, L. L. Videau, C. J. Williams, P. D. Adams, Acta

Crystallogr. Sect. D 2019,75, 861–877.

[31] S. Keller, C. Vargas, H. Zhao, G. Piszczek, C. A. Brautigam, P.

Schuck, Anal. Chem. 2012,84, 5066–5073.

[32] C. A. Brautigam, H. Zhao, C. Vargas, S. Keller, P. Schuck,

Nat. Protoc. 2016,11, 882–894.

[33] H. Zhao, G. Piszczek, P. Schuck, Methods 2015,76, 137–148.

[34] W. D. Robertson, A. M. Bovell, K. Warncke, JBIC J. Biol.

Inorg. Chem. 2013,18, 701–713.

[35] M. Yoshino, K. Murakami, Springerplus 2015,4.

[36] A. Mitchell, H.-Y. Chang, L. Daugherty, M. Fraser, S. Hunter,

R. Lopez, C. McAnulla, C. McMenamin, G. Nuka, S. Pesseat,

A. Sangrador-Vegas, M. Scheremetjew, C. Rato, S.-Y. Yong,

A. Bateman, M. Punta, T. K. Attwood, C. J. A. Sigrist, N.

Redaschi, C. Rivoire, I. Xenarios, D. Kahn, D. Guyot, P. Bork,

I. Letunic, J. Gough, M. Oates, D. Haft, H. Huang, D. A.

Natale, C. H. Wu, C. Orengo, I. Sillitoe, H. Mi, P. D. Thomas,

R. D. Finn, Nucleic Acids Res. 2015,43, D213–21.

[37] K. Katoh, D. M. Standley, Mol. Biol. Evol. 2013,30, 772–780.

[38] K. Katoh, Nucleic Acids Res. 2002,30, 3059–3066.

[39] B. Q. Minh, H. A. Schmidt, O. Chernomor, D. Schrempf,

M. D. Woodhams, A. von Haeseler, R. Lanfear, Mol. Biol.

Evol. 2020,37, 1530–1534.

[40] S. Kalyaanamoorthy, B. Q. Minh, T. K. F. Wong, A. von Haes-

eler, L. S. Jermiin, Nat. Methods 2017,14, 587–589.

[41] S. Q. Le, O. Gascuel, Mol. Biol. Evol. 2008,25, 1307–1320.

[42] A. M. Kozlov, D. Darriba, T. Flouri, B. Morel, A. Stamatakis,

Bioinformatics 2019,35, 4453–4455.

[43] F. D. K. Tria, G. Landan, T. Dagan, Nat. Ecol. Evol. 2017,1,

193.

[44] S. A. Ishikawa, A. Zhukova, W. Iwasaki, O. Gascuel, T. Pupko,

Mol. Biol. Evol. 2019,36, 2069–2085.

[45] K. N. Orlandi, S. R. Phillips, Z. R. Sailer, J. L. Harman, M. J.

Harms, Protein Sci. 2023,32, e4551.

[46] C. K. Saha, R. Sanches Pires, H. Brolin, M. Delannoy, G. C.

Atkinson, Bioinformatics 2021,37, 1312–1314.

[47] I. Letunic, P. Bork, Nucleic Acids Res. 2021,49, W293–W296.

[48] P. Kuzmic, Methods Enzymol. 2009,467, 247–280.

Manuscript received: March 14, 2024

Accepted manuscript online: May 14, 2024

Version of record online: June 25, 2024

Angewandte

Chemie

Research Article