Pretreatment strategies for microbial valorization of bio‐oil fractions produced by fast pyrolysis of ash‐rich lignocellulosic biomass

This work evaluates a biorefinery approach for microbial valorization of bio‐oil fractions produced by fast pyrolysis of ash‐rich lignocellulosic biomass. Different methods are presented for the pretreatment of the low‐sugar complex bio‐oil consisting of organic condensate (OC) and aqueous condensate (AC) to overcome their strong inhibitory effects and unsuitability for common analytical methods. Growth of Pseudomonas putida KT2440, which was chosen as a reference system, on untreated bio‐oil fractions was only detectable using solid medium with OC as sole carbon source. Utilization of a pretreated OC which was filtered, autoclaved, neutralized and centrifuged enabled growth in liquid medium with significant remaining optical instability. By subjecting the pretreated fractions to solid phase extraction, more stable and less inhibitory bio‐oil fractions could be obtained enabling the appliance of common analytical methods. Furthermore, this pretreatment facilitated growth of the applied reference organism Pseudomonas putida KT2440. As there is currently no convincing strategy for reliable application of bio‐oil as a sole source of carbon in industrial biotechnology, the presented work depicts a first step toward establishing bio‐oil as a future sustainable feedstock for a bio‐based economy.

cellulose and hemicellulose. In the last decades, a number of conversion methods have been developed mainly focusing on the production of second-generation biofuels (Brethauer & Studer, 2015;Dahmen et al., 2016;Maurya, Singla, & Negi, 2015;Seidl & Goulart, 2016). One of such a conversion method is fast pyrolysis-a thermochemical method which converts lignocellulosic biomass in the absence of oxygen mainly into an energy-rich liquid referred to as biooil, also referred as to bio-oil (Bridgwater, Meier, & Radlein, 1999).
One example for a fast pyrolysis process with fractional condensation, which addresses pyrolysis of nonwoody biomass such as herbaceous, ash-rich biomass like wheat straw, was shown by Pfitzer et al. (2016) (bioliq ® process, Karlsruhe Institute of Technology (KIT), Germany). The applied two-stage condensation allows a controlled formation of two condensates: a viscous condensate rich in organic substances with <20 wt% water (OC) obtained at condensation temperatures between 60 and 90°C and an aqueous condensate (AC) consisting of up to 85 wt% water and water-soluble organic compounds formed at around 30°C ( Figure 1; Pfitzer et al., 2016). An excerpt of their composition is shown in Table 1. These condensates are mixed with pyrolysis char in different proportions forming energy-rich and free-flowing bioslurries, which are further used for gasification to generate synthetic fuels (Dahmen, Henrich, Dinjus, & Weirich, 2012). Depending on the applied biomass and the mixing ratio, unexploited side streams can accrue. As a future bioeconomy relies on the efficient and sustainable utilization of renewable resources, it is important that for all products and side streams, which accrue during an overall conversion process of lignocellulosic biomass an adequate valorization is found. Furthermore, it should be taken into account that the use of biomass as a feedstock for energy and fuel production is restricted by its relatively low volumetric energy content, seasonality, and discrete geographic availability (FitzPatrick, Champagne, Cunningham, & Whitney, 2010;Lipinsky, 1981).
Regarding bio-oil fractions produced from wheat straw by fast pyrolysis in the bioliq ® plant at Karlsruhe Institute of Technology (KIT, Karlsruhe, Germany), two studies have already been reported investigating their potential as carbon source for microbial cultivation (Dörsam et al., 2016;Lange et al., 2017). Noticeable is that the sugar content of these fractions is very low in the OC fraction or even nonexistent in AC. The major component present in these fractions is acetic acid (Table 1), which can be a potential carbon source for some microorganisms (Berg, Tymoczko, & Stryer, 2002;Gerstmeir et al., 2003;Li et al., 2016). In 2016, Dörsam et al. determined the bio-oil tolerance limits of fungi, as well as investigated the reason for their low tolerance level by assessing 11 substances from bio-oil in different concentrations regarding the production of malic acid. The authors used the organic-rich bio-oil fraction of the so-called organic condensate (OC). For tolerance experiments, cultivations were performed on solid medium. Among the 22 screened fungi, 15 showed tolerance toward OC in minimal medium agar plates while growing on glucose. Cultivations with OC as sole carbon source suggest that only Aspergillus oryzae is able to grow on up to 1% of OC; however, quantitative data are not provided. Beside OC also an aqueous bio-oil phase is formed during the two-stage condensation in the KIT fast pyrolysis plant. In 2017, Lange et al. evaluated the suitability of Corynebacterium glutamicum to grow on this so-called aqueous condensate (AC). The authors reported that after supplementation of minimal medium with AC, turbidity analyses were no longer possible. Therefore, they established a protocol for pretreatment of AC consisting of four steps: pH adjustment, centrifugation, filtration, and heat treatment, aiming for a fermentable substrate by removing residual oils and solids as well as diminishing volatile substances such as methanol. However, also with the addition of pretreated AC to the medium growth was not detectable. Only supplementation of yeast extract to pretreated AC enabled growth of an engineered Corynebacterium glutamicum and the production of 1,2-propanediol.
Even though the bioconversion of bio-oil was investigated in the past, studies are mainly based on cultivations with sugar-rich, hydrolyzed, and detoxified bio-oil fractions added to complex media, which mostly include yeast extract and peptone (Bennett et al., 2009;Chan & Duff, 2010;Lian et al., 2010;Luque et al., 2014;Prosen et al., 1993;Sukhbaatar et al., 2014;Wang et al., 2012;Yang et al., 2011;Yu & Zhang, 2003). Furthermore, their focus generally lies on product formation and less on the biomass generation. As of today, the effect of reactivity of bio-oil during cultivation by quantitative means such as suitable investigation of supplemented culture medium over time as negative control has not been reported in detail.
In this article, the challenges of using sugar-poor bio-oil fractions as sole carbon sources for bacterial cultivation are described. Different pretreatment methods are presented to address problems such as reactive and inhibitory substances in bio-oil fractions produced by fast pyrolysis of ash-rich lignocellulosic biomass. Using the example of Pseudomonas putida KT2440, which is generally regarded as an organic solvent tolerant bacterial strain capable of degrading aromatic hydrocarbons and using a wide range of carbon sources (Cruden, Wolfram, Rogers, & Gibson, 1992;Inoue & Horikoshi, 1989;Ramos, Duque, Huertas, & Haïdour, 1995;Weber, Ooijkaas, Schemen, Hartmans, & Bont, 1993), a potential biorefinery route for microbial valorization of bio-oil is shown.

| Chemicals
All chemicals used in this study were either purchased from Carl Roth GmbH (Karlsruhe, Germany) or Sigma-Aldrich (Munich, Germany) if not stated otherwise.
T A B L E 1 Composition of the two bio-oil phases determined by GC-MS and typical ranges extracted from the literature (Islam et al., 2015) Compound (

| Bio-oil fractions and their pretreatments
The two bio-oil fractions organic condensate (OC) and aqueous condensate (AC) used in this study were prepared from wheat straw by fast pyrolysis in the bioliq ® plant at KIT in Karlsruhe, Germany (Pfitzer et al., 2016). The substances within the two condensates were analyzed by Thünen Institute Hamburg by GC-MS. As a result of unspecific pyrolytic reactions, OC and AC contain hundreds of different molecules, many of which have not been identified. This is most prominent in OC, where pyrolytic lignin and unidentified components constitute a major part of dry weight. An excerpt is shown in Table 1 comprising the most abundant substances in either OC or AC (above 0.1 wt%) and compared with typical ranges in the literature (Islam et al., 2015). OC is low in sugar (in sum 1.63 wt% and thereof 0.965 wt% levoglucosan), and AC even contains no monomeric sugars. Solids in OC are usually <10 wt% and the water content is adjusted between 12 and 18 wt% by varying the condensation temperature (Pfitzer et al., 2016). AC consists of up to 85 wt% water including water-soluble organic compounds and practically no solids. Its pH is~2.5. To maintain integrity to favor stability, OC samples were stored at 4°C and AC at −21°C until further use. To obtain stable and fermentable bio-oil fractions, various pretreatment methods were applied. One of the first purification steps of OC was an extraction. The aim was to extract water-soluble substances such as organic acids and sugars from bio-oil and get an aqueous extract rich in these substances. For this, OC was slowly added to ice-cooled water at the ratio of 1:6 while mixing with a stirrer. A filtration with a bottle top filter with PES membrane (pore size: 0.22 µm; membrane Ø: 75 mm; volume: 500 ml) obtained from Carl Roth GmbH (Karlsruhe, Germany) was performed for both OC extract and AC to remove pyrolytic lignin (water-insoluble fraction) and other solid substances. This modified approach is described by Scholze and Meier (2001), who used this procedure for the characterization of pyrolytic lignin. The resulting filtrates were autoclaved, neutralized, and centrifuged at 4,700 rpm for 15 min, whereby the middle phase represents the final aqueous fractions OC FANC , respectively, AC FANC (Figure 1), which were sterile-filtered at the end. Neutral and aromatic compounds within OC FANC and AC FANC were removed using solid phase extraction (SPE) cartridges (Strata-X 33 µm Polymeric Reversed Phase, 200 mg stationary phase, specific surface of 800 m 2 /g, Phenomenex Ltd., Aschaffenburg, Germany). Extraction of each sample was performed with a fresh column, by employing 20 ml of OC FANC and AC FANC, which were passed through the column in a single run according to the manufacturers' recommendations. The neutrals and aromatics bind at the column and the sterilefiltered flow-through was used as a more purified fraction called OC SPE , respectively, AC SPE (Figure 1). All pretreated bio-oil fractions were stored at 4°C.

| Microorganism, culture medium, and conditions
The used bacteria strain P. putida KT2440 (DSM 6125) was obtained from the DSMZ strain collection (Deutsche Sammlung von Mikroorganismen und Zellkulturen, Braunschweig, Germany). P. putida was grown at 30°C and 120 rpm. For the preculture LB medium (5 g/L yeast extract, 10 g/L tryptone, 10 g/L NaCl; pH 7.0) was used. After~18 hr of incubation, the cells were centrifuged (4,700 rpm, 4 min, 20°C) and the cell pellet was washed with 1:1 (v/v) 9 g/L NaCl solution, centrifuged again and resuspended in fresh 9 g/L NaCl solution. The main medium was inoculated with OD 600 0.

| Sampling, sample processing, and analytics
All photometric measurements were taken using a spectrophotometer (WPA CO8000 Cell density meter, Biochrom Ltd., Cambridge, UK) by measuring the optical density at λ = 600 nm (OD 600 ). The remaining sample was centrifuged (12,500×g, 15 min) to obtain cell-free supernatant for acetate detection. The concentration of acetate was determined from the supernatant samples using acetate assay kit (Enztech yellow line, R-Biopharm AG, Darmstadt, Germany) according to the manufacturers' instructions.

| Data analysis
All experimental data were obtained at least as duplicates from at least two individual biological experimental setups and shown as average values ± standard deviation. Data were fitted using a logistic equation with four parameters in a scientific data analysis and graphing software (Sigma Plot 13.0, Systat, San Jose, USA). Obtained data and the resulting fits were used to calculate the specific growth rate µ (hr). Substrate-to-biomass yields Y X/S (g/g) were calculated using the contained amount of acetate.

| Pretreatment of bio-oil fractions
The major component of the two bio-oil fractions produced in the bioliq ® process is acetic acid (Table 1), which can be a potential carbon source for some microorganisms (Berg et al., 2002;Gerstmeir et al., 2003;Li et al., 2016;Lian et al., 2012). However, solids and pyrolytic lignin in OC, and residual oil in AC, as well as reactive and inhibitory substances make their application as microbial substrates very challenging. By the first pretreatment step, water-soluble substances such as organic acids and sugars within the viscous OC were extracted by cold-water extraction to be separated from residual bio-oil, pyrolytic lignin, and solids. The resulting fractions were sterile-filtered and referred to as OC FANC , respectively, AC FANC (Figure 1). The acetate concentration of OC FANC and AC FANC was about 6 g/L, respectively, 43 g/L. The relatively low acetate content in OC FANC originates from its cold-water extraction resulting in a dilution of water-soluble substances.
A more purified fraction was achieved by solid phase extraction. Here, neutral and aromatic compounds within OC FANC and AC FANC were reduced. These fractions are called OC SPE , respectively, AC SPE ( Figure 1) and exhibit acetate concentrations of about 5 g/L and 42 g/L. There is a small decrease in acetate content when treated with solid phase extraction.
To see the effect of pretreatments on the reactivity of bio-oil fractions, incubations of cell-free medium supplemented with bio-oil fractions were performed and the drift of optical density was measured during incubation and depicted in Figure 2. Measuring the OD 600 , no utilizable results could be obtained due to strong drift and variability between individual measurements as well as control measurements. Incubation of cell-free medium with OC FANC or AC FANC (Figure 2a

| Cultivations with OC or AC as sole carbon source
Experiments with untreated bio-oil fractions in culture suspensions severely hinder analytical procedure and the determination of the optical density of the culture. Especially, the high viscosity of OC and the formation of elastic-type particles interferes with the measurement of the OD 600 ( Figure 3) as well as the dry weight. Also, cell counting with microscopy is difficult, because it is not feasible to differentiate between small OC particles and P. putida cells (Figure 3). But the major problem is the reactivity of biooil fractions. During cultivation, the culture suspension containing bio-oil fractions is getting darker and particles precipitate. This reactivity leads to strong variations within an OD 600 measurement as well as dry weight determination making it impossible to verify biomass formation. Growth of P. putida on untreated OC was only detected using Wilm's KP i agar plates containing OC as carbon source (Figure 4a). The bacteria seem to form biofilm to overcome the inhibitors within OC making growth possible (Khiyami, Pometto, & Brown, 2005). Growth of P. putida is neither detectable on liquid nor on solid medium containing AC as sole carbon source (Figure 4b). This may be explained by the increased bioavailability of growth inhibitory water-soluble compounds in AC such as furans and methanol (Table 2).

| Cultivation with OC FANC or AC FANC as sole carbon source
By mixing OC with water and removing the pyrolytic lignin, first enables the detection of growth in liquid medium (Figure 4c). This could be explained by the fact that the mixing with water caused a dilution, which leads to a lower concentration of inhibitors within OC FANC , but also to a lower concentration of microbial accessible components such as acetate (only 5 g/L). This low concentration of acetate was adjusted to a final acetate concentration of 20 g/L acetate in OC FANC , which again resulted in a slight dilution of the inhibitors. The time-course of the OD 600 of cultivations with 1 g/L acetate content in OC FANC or AC FANC as sole carbon source is depicted in Figure 4c microorganisms subtracted by values obtained from medium without inoculation (Supporting information Figure S1). P. putida was able to grow on a final concentration of 1 g/ L acetate content in OC FANC up to a ΔOD 600 of 1 in 45 hr (Figure 4c). At the maximal ΔOD 600, acetate was completely depleted. For higher concentrations of about 5 g/L acetate in OC FANC growth was not detectable. OD 600 values of the medium with 5 g/L acetate content in OC FANC inoculated with microorganisms and medium without inoculation increased equally (Supporting information Figure S1) and the acetate concentration remained unchanged during cultivation (data not shown).
As AC was not mixed with water to obtain AC FANC , there is no dilution and the concentration of inhibitors within AC FANC is much higher compared to AC FANC . In addition, no adjustment to a higher acetate concentration has to be performed, because the acetate concentration in AC FANC is~42 g/L. Therefore, it is obvious that growth on AC FANC poses a significantly higher challenge than on OC FANC . Cultivations with 1 g/L acetate content in AC FANC as sole carbon source showed no increase of ΔOD 600 values and no decrease in acetate concentration over time (Figure 4d). Also, OD 600 values of medium containing 5 g/L acetate in AC FANC inoculated with microorganisms and medium without inoculation increased equally (Supporting information Figure S1), while the acetate concentration remained unchanged during cultivation (data not shown).  As shown in Figure 2a, there is still a strong reactivity of the pretreated bio-oil fractions OC FANC and AC FANC during incubations, which results in a significant drift in optical density and hinders common analytical methods.

| Cultivation with OC SPE or AC SPE as sole carbon source
SPE facilitates more stable bio-oil fractions (Figure 2b). Using bio-oil fractions after SPE as sole carbon source, P. putida was able to grow on higher concentrations of OC (up to 3 g/L acetate content in OC SPE ), as well as the first time on AC (1 g/L acetate content in AC SPE ) ( Figure 5). This is attributed to the employed SPE, which retains aromatic and hydrophobic compounds that have been reported to interfere with microbial growth Jarboe et al., 2011;Lian et al., 2010;Prosen et al., 1993).
Considering the different concentrations of OC SPE, it is obvious that the higher the concentration of OC SPE the longer the duration of the lag-phase is ( Figure 5). Cultivations with higher acetate concentrations (5 g/L) in OC SPE showed no increase in OD 600 and no decrease in acetate during cultivation after incubation time of up to 7 days (data not shown). Using AC SPE as sole carbon source, P. putida grew on 1 g/L acetate content in AC SPE up to a maximal OD 600 of 1.3 within 147 hr (Figure 5a). While a total of 1 g/L acetate in AC SPE cultivations was depleted completely, acetate concentrations of AC SPE cultivations with higher AC SPE concentrations remain constant during cultivation supposing that P. putida is not able to grow on higher AC SPE concentrations ( Figure 5b). Reference cultivations on different acetate concentrations without any addition of bio-oil fractions were performed for comparison ( Figure 5). Growth of P. putida was detected up to an acetate concentration of 5 g/L.

| Comparison of cultivation parameters
As a first assessment of potential suitability of pretreated bio-oil fractions as carbon sources for biotechnology, process parameters of different cultivations using untreated and pretreated bio-oil fractions are compared (Table 2). While maximum growth rates on pretreated bio-oil fractions are in a similar range (0.1-0.3 hr -1 ), in contrast, the maximal growth rate of cultivations on pure acetate is significantly higher (0.6-0.9 hr -1 ). This may mainly be attributed to the fact that many inhibitory substances (Table 1) are not or only partially removed by the applied pretreatment methods. Substrate-to-biomass conversion yields Y X|S are in a similar range for cultivation on pure acetate, OC FANC and OC SPE of~0.3 g/g. Cultivation on AC SPE, however, shows significantly higher yields of 0.46 g/g. This can mainly be attributed to the fact that AC SPE as the aqueous fraction of pyrolysis contains additional potential carbon sources such as propionate, formate, and acetol compared to OC SPE .
Using bio-oil as an alternative carbon source for microbial cultivation is very challenging due to its instability and highly complex composition along with strong adverse effects on microbial growth and existing analytical procedures. In this study, it was shown that SPE may be a suitable tool to obtain stable bio-oil fractions with less inhibitory substances which facilitates growth of P. putida KT2440 on low-sugar bio-oil fractions.
This provides a first step toward establishing bio-oil as a feedstock for microbial valorization. Future work should address direct bioconversion and investigate appropriate fractionating condensation designs and conditions.