Impact of different trace elements on metabolic routes during heterotrophic growth of C. ljungdahlii investigated through online measurement of the carbon dioxide transfer rate

Synthesis gas fermentation using acetogenic clostridia is a rapidly increasing research area. It offers the possibility to produce platform chemicals from sustainable C1 carbon sources. The Wood‐Ljungdahl pathway (WLP), which allows acetogens to grow autotrophically, is also active during heterotrophic growth. It acts as an electron sink and allows for the utilization of a wide variety of soluble substrates and increases ATP yields during heterotrophic growth. While glycolysis leads to CO2 evolution, WLP activity results in CO2 fixation. Thus, a reduction of net CO2 emissions during growth with sugars is an indicator of WLP activity. To study the effect of trace elements and ventilation rates on the interaction between glycolysis and the WLP, the model acetogen Clostridium ljungdahlii was cultivated in YTF medium, a complex medium generally employed for heterotrophic growth, with fructose as growth substrate. The recently reported anaRAMOS device was used for online measurement of metabolic activity, in form of CO2 evolution. The addition of multiple trace elements (iron, cobalt, manganese, zinc, nickel, copper, selenium, and tungsten) was tested, to study the interaction between glycolysis and the Wood ljungdahl pathway. While the addition of iron(II) increased growth rates and ethanol production, added nickel(II) increased WLP activity and acetate formation, reducing net CO2 production by 28%. Also, higher CO2 availability through reduced volumetric gas flow resulted in 25% reduction of CO2 evolution. These online metabolic data demonstrate that the anaRAMOS is a valuable tool in the investigation of metabolic responses i.e. to determine nutrient requirements that results in reduced CO2 production. Thereby the media composition can be optimized depending on the specific goal.


| INTRODUCTION
The climate goals defined by the European commission demand a drastic reduction of greenhouse gas emission with two targets set for strategies of "carbon capture and storage" and "carbon capture and utilization" are under investigation. One promising technology for future applications is gas fermentation. Gaseous carbon and energy sources from industrial exhaust gas or gasified organic materials like municipal waste or lignocellulosic biomass can be used as a feedstock to produce platform chemicals and fuels. [3][4][5][6] These gases have in common that they are composed of H 2 , CO 2 , and CO in differing ratios and are called synthesis gas or syngas. As a group generally referred to as acetogens, several bacterial species can use these gases as a substrate for growth and production of a wide array of industrially relevant chemicals, especially organic acids and alcohols. [7][8][9][10][11] Acetogens employ a linear carbon fixation pathway called the Wood-Ljungdahl pathway (WLP), yielding acetyl-CoA as an energy rich intermediate 12,13 (shown in Figure 1). The ability to capture carbon from industrial or agricultural byproducts turns these microorganisms into promising biocatalysts, to reduce our global carbon footprint through sustainable production of commodity chemicals.
Even though acetogentic bacteria are generally investigated concerning their ability to grow autotrophically and turn syngas into commodity chemicals, carbohydrates as a carbon source provides an easily accessible substrate that is neither toxic nor limited by gas-water transfer rates. While consumption of C6 sugars such as glucose or fructose leads to the emission of CO 2 in various anaerobically grown organisms, acetogens can convert fructose stoichiometrically into three molecules of acetate without net production of CO 2 by using the same metabolic route as used for autotrophic growth. 14 This is achieved by employing the WLP to reduce the CO 2 evolved during glycolysis with the electrons derived from the sugar. The formation of three moles of acetate per mole of fructose was reported early in the research of acetogens, when the second isolated acetogen, Clostridium thermoaceticum, was described in 1942. 15,16 Employing the WLP as a metabolic module to reestablish redox homeostasis during growth on fructose allows acetogens to conserve more energy than their competitors in the same environment, providing them with an evolutionary edge. 17,18 Furthermore, the WLP has been shown to allow acetogens to access growth substrates like lactate 19 or ethanol, [20][21][22] which are products of many anaerobic fermentations. Both compounds are relatively common in anaerobic environments and inaccessible as carbon and sources for many other organisms. It has been demonstrated for the model acetogen C. ljungdahlii, that the genes of the WLP pathway are constitutively transcribed. When grown on CO/CO 2 + H 2 , where the WLP acts as the key metabolic route, transcription of the WLP genes is only increased 1.31-2.06 fold compared to growth with fructose, where the WLP provides a supporting role for the hexose metabolism 23,24 (Figure 1).
During glycolysis, fructose is converted into two coenzymeAbound acetyl-groups, two molecules of CO 2, and eight electrons.
Electrons are carried by four redox equivalents, namely two NADH and two reduced ferredoxin. Both CO 2 molecules can then dependent on the activity of the WLP, be reduced with electrons from the redox equivalents and incorporated into one additional acetyl-CoA via the WLP. 14 This remarkable feature allows for reestablishing the redox homeostasis and conserves additional energy, if acetate is produced from the acetyl-CoA. Thus, acetogens are able to conserve more energy during growth on fructose, compared to organisms without the WLP, which produce, for example, ethanol + CO 2 or lactate from sugars. 17 If fructose is the only available electron donor, redox homeostasis allows for the formation of three molecules of acetate from fructose. However, at low CO 2 availability due to excessive stripping into the gas phase or in the presence of additional gaseous electron donors like H 2 or CO, a shift from CO 2 reduction toward reduction of acetate to ethanol allows the cells to conserve more energy per carbon. This in turn leads to net CO 2 emission instead of strict homoacetogenesis. During autotrophic growth with H 2 as electron source, acetogens employ a bifurcating hydrogenase to transfer the electrons from H 2 to NAD(P)H and Fd. 14,25 In theory, during growth with fructose, this enzyme might act in reverse and lead to the production of traces of H 2 . Moorella thermoacetica for example, is known to produce small traces of H 2 during growth with glucose (2 mmol H 2 per 100 mmol glucose) under CO 2 limitation. 26 This, however, might be neglected for the purpose of this study, since this amount of H 2 evolution only accounts for a small fraction of the electrons transferred during the glycolysis. Since eight electrons per hexose are transferred to electron carriers during glycolysis, 100 mM glucose provide 800 mM of electrons. Since two electrons are needed for the production of a H 2 molecule, the electrons derived from 100 mM of glucose would in theory allow for the production of 400 mM of hydrogen. Thus, the observed 2 mmol of H 2 produced from 100 mmol of glucose by M. thermoacetica only constitute 0.5% of the available electrons in the cited study. Hence, the interaction and activity of these pathways, glycolysis, WLP and reduction of acids to alcohols, determine carbon conversion efficiency and product ratios.
Due to the increased industrial value and the broader field of applications of alcohols compared to organic acids, research generally focuses on increasing yield and selectivity for alcohol production. [27][28][29][30][31][32][33] Various methods to genetically increase the spectrum of available products and to improve specificity for the targeted product have been reported. [34][35][36] Another route for higher selectivity and improved titers of targeted products is the optimization of fermentation conditions, especially media composition. Variation of the gas composition and gas availability has also been reported to influence the product ratio of acetate to ethanol. For example, an increased CO availability resulted in higher alcohol and decreased acid production. 14,25,26 An essential issue in alcohol production is the availability of necessary cofactors for enzymes involved in the required pathways. Since many enzymes in acetogens depend on metallic cofactors 37 (Figure 1), optimization of trace element compositions is crucial for optimal enzyme activity in syngas fermentation. The metal involved in the most significant number of metabolic conversions in acetogens is iron(II). 38 It is required in the electron carrier ferredoxin, 39,40 in many enzymes catalyzing reactions of the WLP 37 and alcohol dehydrogenases and aldehyde oxidoreductases 13,41 (Figure 1). Additionally, it was found to be required in concentrations an order of magnitude higher than almost all other trace elements in an optimized trace element solution for alcohol production in C. carboxidivorans. 42 But also other metals like nickel, selenium, tungsten, zinc, and cobalt are essential, 31,42,43 due to their involvement in several reactions.
In the first step of the WLP and the reduction of necessary electron carriers, iron(II) and nickel(II) are essential. The genome of the model acetogen C. ljungdahlii encodes four [FeFe] hydrogenases and one [NiFe] hydrogenase. These hydrogenases are either involved in reducing electron carriers like NAD(P)H, and ferredoxin (containing FeS-cluster) with electrons derived from H 2 or are part of the enzyme complex catalyzing the initial step of the WLP, reduction of CO 2 to formate. 14 Both metals are also essential for several further reactions F I G U R E 1 Carbon and electron flux during growth on fructose of C. ljungdahlii with online CTR measurement using the anaRAMOS. CO 2 evolved during glycolysis can be fixed in the Wood-Ljungdahl pathway or it can be exhausted into the gas phase. Exhausted carbon dioxide is detected by the anaRAMOS (1). For detailed explanation please see main text. Ack: acetate kinase, ACS/CODH: bifunctional acetyl-CoA synthase/CO dehydrogenase, AdhE: bifunctional aldehydeÀ/alcohol dehydrogenase, AOR: aldehyde-ferredoxin oxidoreductase, PFOR: pyruvateferredoxin oxidoreductase, pta: phosphotransacetylase. If the electron carriers NADH or ferredoxin are involved in reactions, only the reduced form of the electron carrier is shown for easier readability of the figure. For the same reason, free CoA, THF and ADP + Pi were omitted.
[H] is used as a placeholder for electron carriers (proton + electron), if the electron carrier involved in the reaction is variable. Possible electron transfer between molecular hydrogen, NADH and ferredoxin via bifurcating hydrogenase and rnf complex are not shown. Products are not stoichiometric, as dependent on multiple parameters. Source: Figure modified from Köpke et al. 13 in the WLP, like the final fusion of both reduced C1 intermediates to the coenzymeA-bound acetyl-group by the CO-dehydrogenase/acetyl-CoA-synthase (CODH/ACS) 13 (Figure 1). In this final reaction of the WLP, a cobalamin corrinoid iron(II)-sulfur protein is essential for transferring the methyl group from THF to form acetyl-CoA. 37,44 From here on, two distinct routes are possible. Either acetyl-CoA is directly reduced to ethanol via acetaldehyde by an iron(II)dependent bifunctional alcohol-/aldehydedehydrogenase (adhE), with electrons derived from two NADH, or acetate is produced first by phosphotransacetylase and acetate kinase, leading to the formation of one ATP (Figure 1). Acetate can then be further reduced to acetaldehyde via a tungsten-dependent aldehyde-ferredoxin-oxidoreductase (aor) with electrons derived from ferredoxin. Finally, ethanol is produced in a second step from acetaldehyde via adhE. 41  To study the impact of trace metals on metabolic activity, the anaerobic Respiration Activity Monitoring System (anaRAMOS) device was used. The anaRAMOS offers the possibility of real-time online monitoring of the carbon dioxide transfer rate (CTR) in up to eight parallel shake flasks. 49,50 The measured CTR is a direct quantitative indication of metabolic activity and can be used to calculate the total amount of produced net CO 2 (CT). As already discussed, in hexose metabolism, acetogens first produce two mol CO 2 per sugar, which can subsequently be reduced to acetyl-CoA in the WLP with electron carriers reduced during glycolysis. Any relative reduction of the observed CO 2 production by the anaRAMOS is an indicator of the WLP activity. This is the case, because in all experiments fructose was fully consumed. Hence, the carbon converted into biomass, acetate and ethanol was the same in all experiments.
The anaRAMOS has previously been demonstrated to be a helpful tool to optimize growth conditions. 49 Through this method an iron(II)-deficiency in YTF medium for C. ljungdahlii was detected. A model was introduced, which allows determination of the necessary amount of iron(II) for unlimited growth on a given amount of fructose. 49 In the present study, the addition of metal salts from PETC 1754 medium (iron, cobalt, manganese, zinc, nickel, copper, selenium, and tungsten) to YTF medium was analyzed to demonstrate the influence of different trace elements on the cultivation of C. ljungdahlii in complex medium. For this goal the anaRAMOS was deployed to monitor changes in metabolic activity through online measurement of CO 2 evolution. These results highlight the benefit of the anaRAMOS in understanding microbial metabolism. Additionally, due to semi-continuous ventilation, the anaRAMOS is closer to industrial fermentation conditions, compared to non-ventilated, but commonly used serum bottles. A special focus is set on the interaction of glycolysis and WLP. This is done by evaluating the CO 2 production, dependent on the experimental setting. Single trace element solutions were prepared to contain per liter 0.2 g respectively. The TE solution and single trace elements were sterile filtered.
Before each experiment, the main stock solution, fructose stock solution, and cysteine stock solution were mixed. Trace element solution or single trace elements were added according to the experiment's aim (details are given for each experiment). The pH was set to a value of 6 with HCl, and the final volume was adjusted using sterile deionized water. The completed medium was transferred into serum bottles or shake flasks under sterile conditions. Serum bottles were flushed for 20 min before inoculation using 90% ultra-high purity (99.999%) nitrogen (Praxair, Germany) and 10% carbon dioxide Cryo stocks of cells were stored at À80 C after the addition of 10% anoxic DMSO to an actively growing culture in YTF medium at OD 600 = 0.8.

| Cultivation conditions
Two sequential pre-cultures were grown in serum bottles at 37 C and a shaking frequency of 220 min À1 at a shaking diameter of d 0 = 3 mm. The first pre-culture with a total filling volume of 10 ml

| Offline analysis
For offline analysis, shake flasks were removed from the anaRAMOS after the cultivation was terminated. The optical density (OD) was measured, after adequate dilution (OD 0.1-0.3), at a wavelength of 600 nm using a Genesys 20 Photospectrometer (ThermoScientific, USA). Subsequently, the OD was used to calculate the cell dry weight.
For this, the OD to cell dry weight (CDW) correlation was determined as shown in Figure S2. Equation (1) was used for calculation. For CDW measurement a pre-dried and pre-weight filter paper was used.
Ten-milliliter samples were transferred onto the filter paper while vacuum was applied. Filter papers were dried for 48 h and the weight was taken thereafter.   Figure S4. The second CTR peak (during phase II) of the cultivation with TE reached a maximum CTR of 3.2 mmol L À1 h À1 , while in standard YTF medium, the maximum CTR was only 2.1 mmol L À1 h À1 . In comparison, the cultivation without TE displayed a plateau after the second CTR peak for the remainder of phase II. In the cultivation with added TE, the CTR decreased sharply after the CTR peak at 19 h and reached 0 mmol L À1 h À1 after 23 h of cell growth. The progression of the CTR over time for the cultivation with TE did not adhere to the typical CTR curve shape expected for unlimited growth. 53 Reasons for this were investigated in the subsequent experiments reported in this manuscript.
To calculate the total CO 2 production throughout the cultivation, the CTR curve can be integrated to obtain the CT value ( Figure 2b).

| Influence of iron(II)
The first trace metal investigated was iron(II). Some of the data was previously shown in Reference [54].  (Figure 4b). The increase of the metabolic activity with increased iron(II) concentrations led to higher CO 2 production and higher CO 2 concentrations in the gas phase. In turn, this resulted in a net carbon loss in the cultivation since more CO 2 was removed via active ventilation during CTR measurement ( Figure 1).
Besides increasing CO 2 production, iron(II) addition led to a slight shift in the product spectrum. With an increasing concentration of iron(II), titers of the more reduced product ethanol increased at the expense of acetate formation (Figure 4c). This result was expected in light of the increased CO 2 production observed during cultivation.
Since CO 2 was not reduced in the WLP but released into the gas phase in the presence of increased iron(II) concentrations, acetate had to be reduced to ethanol to recycle electron carriers reduced during glycolysis. The shift toward ethanol in the presence of higher iron(II) concentrations occurs most probably since C. ljungdahlii converts acetate into ethanol using iron(II)-dependent alcohol dehydrogenase ( Figure 1). 13 Since the authors are not aware of a study reporting H 2 evolution from sugars by C. ljungdahlii, and the H 2 evolution from sugars reported for other acetogens is negligible as discussed before, 26 potential H 2 production is not considered in this study. In total, more biomass was produced with increased iron(II) availability ( Figure 4c). This finding correlates well with the non-inhibited exponential CTR increase (Figure 4a). These results also correspond well to the findings of our previous study, where iron(II) was reported to display a positive influence on the alcohol production in C. carboxidivorans. 42 Carbon balances were calculated with CT data and measured product titers ( Figure S6). With increasing iron(II) concentrations, total carbon in CO 2 and products slightly increased. Carbon balances were calculated for all experiments and revealed slightly higher amounts of carbon in the products than supplemented via fructose in the presence of added iron(II) (Figures S6-S8). This hints at an improved utilization of complex substrates like yeast extract in the presence of added iron(II).

| Influence of cobalt(II)
The second trace element investigated in detail was cobalt(II).

| Influence of nickel(II) and iron(II)
The third trace element investigated in more detail was nickel(II).  reference cultivation, the CTR dropped to 0 mmol L À1 h À1 directly after the maximum was reached. CTR curves in the presence of nickel(II) sloped down more gently, after reaching their maximum values (Figure 6a).
Compared to the reference, CO 2 production (CT) decreased by 28% in cultures with added 0.4 mg L À1 of NiCl 2 Á6H 2 O (Figure 6b), indicating higher WLP activity if nickel(II) is available. Since the crucial enzyme in the WLP, the CODH/ACS responsible for the formation of acetyl-CoA, is strictly nickel(II) dependent. 37 It can be assumed, that an increased nickel(II) availability results in an increased CODH/ACS activity and thus higher CO 2 reassimilation. Higher nickel(II) availability also led to an increase in acetate production at the expense of ethanol production and correlated to a downward shift in final pH (Figure 6b,c). The higher carbon utilization efficiency is explained by increased WLP activity. Since more electrons are diverted toward CO 2 reduction to acetate, fewer electrons are available to reduce acetate to ethanol. 55 This causes a higher ratio of soluble products (acetate and ethanol) to CO 2 with higher nickel(II) availability (Figure 6d). Biomass yields also increased with added nickel(II) compared to the reference cultivation, but there was no correlation between higher nickel(II) concentrations and biomass yields. Total product carbon balances were calculated and showed a slight increase in carbon metabolized into products (CO 2 , acetate, and ethanol) in the presence of added nickel(II) ( Figure S8). In conclusion, NiCl 2 Á6H 2 O (and Fe(II)) addition to complex medium resulted in higher growth yield. It increased acetate formation at the expense of ethanol and CO 2 formation, indicating higher WLP activity, most probably due to alleviated limitation of CODH/ACS.
To gain further information on the influence of nickel(II) availability on the growth and production behavior of C. ljungdahlii, additional cultivations were performed, and offline sampled to generate data for OD and products. The results were correlated with the CTR curves recorded by the anaRAMOS (Figure 7). When only Fe(II) (8 mg L À1 of (NH 4 ) 2 Fe(SO 4 ) 2 Á6H 2 O) was added, cells displayed a diauxic growth behavior with higher initial exponential growth for the first 8-10 h, followed by a second, slower exponential growth rate (indicated by log OD in Figure 7b,e). The same applies for online measured CTR (indicated by log CTR in Figure 7b,e). This was accompanied by a first CTR peak at 6 h and a subsequent short decline in CTR before it rose again to a second, much higher peak after approximately 22 h and a sharp decline toward 0 net CO 2 evolution (Figure 7a). However, if The CT for the 10 ml cultures (39 mmol L À1 ) was 13 mmol L À1 higher than values for the 50 and 80 ml cultures (both 52 mmol L À1 ).
In total, this accounts for a reduction of carbon loss by 25% at larger   (Figure 8c), the pH value decreased with larger filling volumes (Figure 8b). The results in the experiment with increased filling volume was similar to the results for increasing nickel(II) concentrations. Both, a larger filling volume ( Figure 8b) and a higher nickel(II) concentration (with iron(II) present) (Figure 6b), allow for an enhanced CO 2 utilization via the WLP and shift the product spectrum toward acetate production. Interestingly, an increase in filling volume in standard YTF medium without the addition of iron(II) and nickel(II) displayed the same trend ( Figure S5).
Even though WLP activity was limited due to a shortage of trace metals, increased availability of dissolved CO 2 (higher filling volume) led to shorter cultivation times and reduced CO 2 release. However, this effect was less pronounced than in the presence of added iron(II) and nickel(II). For C. ljungdahlii cultivations in YTF medium, iron(II), cobalt(II), and nickel(II) were identified to influence the CO 2 production, the product spectrum, and biomass formation. The addition of iron(II) led to increased ethanol production. The addition of nickel(II) and iron(II) led to a reduction of CO 2 production by 28%. In comparison, an increase of the filling volume resulted in a reduction of CO 2 emission by 25%. In both cases, the product spectrum was shifted toward acetate, indicating more efficient carbon utilization during recycling of V L = 10-80 ml, absolute gas flow rate: 5 ml min À1 N 2 , medium = YTF medium, initial fructose concentration c t0 = 5 g L À1 , inoculation density: OD t0 = 0.1.
NADH and ferredoxin reduced during glycolysis. The reported results help to better understand the metabolic activity of C. ljungdahlii and the influence of trace elements on product and biomass formation. With this study as a proof of concept, it is now feasible to evaluate different media compositions, growth substrates and cultivation conditions for acetogenic bacteria, to characterize their behavior and improve strains and fermentation conditions. In the next steps, a more detailed analyses concerning the influence of additionally supplemented CO 2 on heterotrophic growth needs to be conducted. Future experiments could also investigate other carbon sources. The RAMOS device was recently adjusted to syngas as carbon source. 56

ACKNOWLEDGMENTS
We would like to thank BMBF for funding this project (03INT513BE) and the European Union for funding this project (Agreement No. 761042).

CONFLICT OF INTEREST
The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT
All data is available on request to the corrisponding author.