Structure and predictive functional profiling of microbial communities in two biotrickling filters treated with continuous/discontinuous waste gases

Volatile organic compounds (VOCs) are generated from many industrial activities such as petroleum refining, coating and furniture manufacturing (Lee et al. 2009; Cheng et al. 2016). The release of VOCs into the atmosphere poses a severe threat to human health and welfare (Gallastegui et al. 2011). Furthermore, many kinds of VOCs are primarily responsible for the formation of photochemical smog (Guo 2012; Wang 2015). Large quantities of waste gases containing VOCs are emitted each year in China, leading to increasingly severe air pollution; in response to this, Chinese authorities are reinforcing the management of pollution control (Tong et al. 2013; Zhu et al. 2016). Benzene and toluene, together with ethyl benzene and xylene constitute BTEX, account for approximately 59% (w/w) of gasoline pollutants (Rahul et al. 2013). BTEX ranked 78 of 275 substances that were identified as showing the most significant potential threat to human health based on the Comprehensive Environmental Response, Compensation and Liability Act (CERCLA 2007). Toluene, ethyl benzene and o-xylene are highly toxic and mutagenic, and many health problems, including kidney failure, heart attack, and liver disease, are likely to be caused by chronic exposure to these substances (Wilbur et al. 2008; Singh et al. 2010; Rahul et al. 2013).

Biofiltration is generally considered a cost-saving and environmentally friendly approach for the abatement of air pollution (Leson and Winer 1991; Nikiema et al. 2007). Many efforts have been made to optimize the configuration of bioreactors to enhance their overall performance (Irvine and Moe 2001; Kim et al. 2005). Effects of various environmental factors such as EBRT, pH, temperature, moisture content, and pressure drop, as well as the microbial communities inside, have been extensively studied (Baquerizo et al. 2009; Cabrol et al. 2012; Pérez et al. 2016; Copelli et al. 2017; Ordaz et al. 2018). These studies were conducted under steady conditions. However, many actual emission sources generate waste gases in a discontinuous mode, and the performance of bioreactors under this unsteady condition has been rarely reported (Lee et al. 2009; Lebrero et al. 2013; San-Valero et al. 2013; Alinejad et al. 2017). According to Qi and Moe (2006), a biofilter operated under discontinuous mode required more time to adapt to the pollutants compared with one that was continuously fed with pollutants. There are still large uncertainties concerning the impact of discontinuous mode on the removal performance, function and composition of microbial communities as well as biofilm formation, especially under discontinuous conditions from the beginning of operation. Studies related to these issues are very scarce in the literature.

The abatement performance of biofiltration reactors relies heavily on the biofilm formed on the surface of packing materials (Acuña et al. 2002). Biofilm formation allows a microbial lifestyle that is entirely different from the planktonic state (Flemming and Wingender 2010). The formation of mature biofilm on various surfaces is a major consideration when running biological reactors in discontinuous mode since some of the attached cells can detach from the surface during the idle phase and return to the planktonic form. In this situation, biofilm formation is extremely difficult. Microorganisms differ greatly in their ability to form biofilm or to degrade pollutants (Lee et al. 2009; Alinejad et al. 2017). Some types of microorganisms may play a leading role in biofilm formation, whereas others possess strong degrading capabilities but have a weak aptitude for forming biofilm. Therefore, the understanding of the specific functions of different microorganisms will provide us with insight into the biofiltration process of VOC-containing waste gases during discontinuous operation mode.

In this study, we used two biotrickling filters fed with a mixture of five aromatic compounds in continuous and discontinuous modes. The removal efficiencies of each bioreactor were compared for five aromatic compounds. High-throughput sequencing of 16S rRNA genes was employed to further understand the effect of the operation mode on the composition and ecological function of the microbial communities. The structure of microbial communities was compared before and after the switch of operation mode, and important species that contributed to the differences between continuous and discontinuous conditions were identified using the random forest classifier (Breiman 2001). The functions of the microbial communities were also predicted using the phylogenetic investigation of communities by the reconstruction of unobserved states (PICRUSt) (Langille et al. 2013). We aimed to identify the specific microorganisms capable of adapting adverse condition, and then reveal the mechanisms of biofilm formation under stress conditions.

Five aromatic compounds were used as target pollutants in this study. Our findings demonstrated that toluene and ethyl benzene could be easily removed in the continuously operated BTF1. However, the presence of xylene inhibited the overall removal performance of BTF1. The decrease in the inlet concentrations of xylene reduced this inhibitory effect. Interactions between the components of mixed gases have been reported in the literature, but some reports were conflicting. For instance, Strauss et al. (2004) and Kim et al. (2009) reported that the removal of toluene was inhibited by p-xylene, while the removal of p-xylene was enhanced by toluene. Another study showed that xylene isomers decreased the RE of toluene, but the effect of toluene on xylene was negligible (Jorio et al. 1998). The inhibitory effect between these aromatic compounds was explained by their similarities in the chemical structure, the metabolic pathways and the enzymes (Gallastegui et al. 2011). However, the interaction during the removal of similar compounds was also influenced by other factors such as the inlet loading rates, the type of microorganisms and the configuration of the bioreactor.

Many industrial waste gases are emitted discontinuously. The purification of discontinuously emitted waste gases by biofiltration still remains challenging. Our findings suggest that BTF2 cannot be acclimated and cannot form a robust biofilm successfully under discontinuous operation mode. Biomass accumulation inside BTF2 was approximately 47 times lower than that observed in the continuously operated BTF1. EPS, which is the key component of biofilm, was approximately one hundred times higher in BTF1 than that in BTF2, indicating that the biofilm formation in BTF2 was severely affected. Under discontinuous operation mode, microbial cells usually enter a so-called endogenous metabolic state during idle times due to the insufficiency of carbon sources. The growth rates of microorganisms were thus lowered (Gallastegui et al. 2011). Additionally, biofilm formation is a highly complex process involving several steps, such as reversible adhesion to solid surfaces, irreversible attachment and EPS production (Molin and Tolker-Nielsen 2003; Zoubos et al. 2012). Those microbial cells reversibly adhered to the surface of packing materials can easily detach from the surface when nutrients are limited. In this situation, it is difficult to increase the density of microbial cells to a threshold level at which quorum sensing signal molecules can be produced (Kim et al. 2014).

The switch in the operation mode from continuous to discontinuous did not decrease the removal performance of BTF1, indicating that once enough biomass had accumulated or a mature biofilm had formed, its robustness was highly enhanced. In a well-organized biofilm, microbial cells aggregate closely with each other and embed in a sticky matrix that is primarily composed of extracellular polymeric substances (EPS). Microorganisms inside the biofilm can coordinate their behaviors synchronously when environmental conditions change (Cheng et al. 2016). Furthermore, with the protective effects provided by EPS, the resistance of microbial cells to various adverse conditions can be greatly enhanced (Dang and Lovell 2000; Jung et al. 2013). However, the microorganisms in the bottom layer would seriously suffer from the starvation shock in some extent, and need time to readapt to the changed environmental condition, leading to relative lower biomass than in the upper layer.

The switch in the operation mode from discontinuous to continuous greatly enhanced the removal performance of BTF2 and promoted the accumulation of biomass as well as the production of EPS in the bottom layer, suggesting that continuous operation mode stimulated the growth of microorganisms and the formation of biofilm in BTF2. It can be inferred that BTF2 must have harbored some kinds of microorganisms that potentially had the capability to degrade aromatic compounds. These microorganisms could not proliferate under the discontinuous operation mode due to the limitation of carbon sources. When continuously fed VOCs were loaded into BTF2 from the bottom, the numbers of these microorganisms adhered on the packing materials of the bottom layer increased rapidly to the threshold at which a biofilm began to form. It is well known that microbes rely heavily on quorum sensing systems to form biofilms (Rohwerder et al. 2003). As mentioned above, although members of the genus Comamonas predominated in BTF2, their growth rate was too slow to reach a sufficient number of cells to form a biofilm.

The analysis of high-throughput sequencing data revealed that the two BTFs formed very differently structured microbial communities. The operation mode had strong effects not only on the filter performance of the bioreactors but also on the composition of the microbial communities. However, it is worth noting that BTF2 had a similar removal performance on day 88 compared to BTF1. This finding suggests that the structure of the microbial communities was heavily influenced by the operation mode, but their degradation functions were determined more by the composition of the waste gases.

Burkholderia and Mycobacterium were the most predominant genera in BTF1 and BTF2, respectively. The random forest model showed that they were also the two most important genera contributing to the differentiation of microbial communities between the two BTFs. The genus Mycobacterium and other members, of which relative abundance increased after the switch, together comprised the core microbial community of BTF2. Members of the genus Mycobacterium can retain viability under starvation conditions by entering a nonreplicating resting state (Gengenbacher et al. 2010; Wu et al. 2016). The degradation of toluene, ethyl benzene, and o-xylene by the genus Mycobacterium is also reported in the literature (Zhang et al. 2013). Additionally, the genus Mycobacterium predominated in BTF1 with relative abundances of 2.56% (first stage) and 1.88% (second stage). However, its relative abundance was far lower than the relative abundance of the genus Burkholderia. It is most likely that the growth rate of the genus Mycobacterium was lower than that of the genus Burkholderia when VOCs were continuously fed. This could be further confirmed by the fact that the relative abundance of the genus Burkholderia increased up to 10 times in BTF2 after the switch, while an increase of only 45% was observed for the genus Mycobacterium. Members of the genus Burkholderia are widely reported as having high degradation capabilities for numerous kinds of pollutants (Wang et al. 2002; Johnsen et al. 2005; Lee et al. 2016). However, unlike the genus Mycobacterium, some members of the genus Burkholderia may not be tolerant of the stress of starvation.

Biofilm formation is precisely regulated by a complex network involving multiple bacterial mechanisms such as quorum sensing and the two-component and c-di-GMP system (Coggan and Wolfgang 2012; Rasamiravaka et al. 2015). The metagenomics inferred by the PICRUSt method provided some helpful information for understanding the effect of operation mode on the ecological function of the microbial communities and on biofilm formation. Membrane transporters such as transporter, ABC transporter and secretion proteins are responsible for the transportation of a variety of substances across the cell membrane. Gene family encoding transporters were more abundant in BTF1 than those in BTF2, indicating that microorganisms grown inside BTF1 had high nutrient uptake rates. Hence, cell numbers increased to the threshold that the quorum sensing system required. Membrane transporters have reportedly played important roles during the biofilm formation of Pseudomonas fluorescens (Hinsa et al. 2003). The accumulation of exopolysaccharide was decreased significantly in a transporter-deficient mutant of Rhizobium leguminosarum compared to that of its wild type, and the mutant was severely impaired in terms of biofilm formation (Vanderlinde et al. 2010). Two-component signal transduction systems enable microorganisms, such as Pseudomonas, Escherichia and Vibrio, to regulate extracellular polysaccharide production and biofilm formation (Martinez and Vadyvaloo 2014; Rasamiravaka et al. 2015; Teschler et al. 2015). A review of the literature demonstrated that motility is a physiological and behavioral trait that is usually linked to the response to environmental gradients. The chemotactic response that is driven by environmental sensing and cell motility is important for microorganisms to search for available nutrients (Dang and Lovell 2016). The decrease in cell motility in BTF2 after the switch in the operation mode indicated a transition of microbial lifestyle from free-living forms to a biofilm.

Although PICRUSt revealed the changes in those gene families involving biofilm formation and pollutant degradation, it remains largely unknown to what degree the operation mode affected the biofilm formation. The determination of biomass showed that biofilm formation was greatly increased in BTF2 after the switch of operation mode, despite the fact that the relative abundance of the majority of these gene families decreased. In this sense, the quantitative analysis of the functional genes involved in biofilm formation is necessary. However, comparing all inferred gene families demonstrated that many gene families involved in the degradation of aromatic compounds and terpene series as well as the metabolism of amino acids and other carbohydrates were obviously upregulated in BTF2 compared to those in BTF1 after the switch in the operation mode. Linking this change to the microbial community in BTF2 could indicate that the genus Mycobacterium and the unclassified OTU 8121 may be the two most important degraders of aromatic compounds since their relative abundance dramatically increased after the switch in the operation mode.