The morphology and metabolic potential of the Chloroflexi in full-scale activated sludge wastewater treatment plants

ABSTRACT

Filamentous bacteria belonging to the phylum Chloroflexi have received considerable attention in wastewater treatment systems for their suggested role in the operational problem of impaired sludge settleability known as bulking. Their consistently high abundance in full-scale systems, even in the absence of bulking, indicates that they make a substantial contribution to the nutrient transformations during wastewater treatment. In this study, extensive 16S rRNA amplicon surveys of Danish wastewater treatment plants (WWTPs) with nutrient removal were screened to identify numerically important Chloroflexi genera. Fluorescence in situ hybridization probes were designed for their in situ characterization. All abundant Chloroflexi phylotypes were putatively identified as facultative anaerobic chemoorganotrophs involved in sugar fermentation. They were all filamentous but differed in their morphology and spatial arrangement. ‘Candidatus Villigracilis’ was predominantly located within the activated sludge flocs, where they possibly have structural importance, and their abundance was relatively stable. Conversely, the abundance of ‘Candidatus Amarolinea’ was highly dynamic, relative to other genera, sometimes reaching abundances in excess of 30% of the biovolume, suggesting their likely role in bulking episodes. This study gives an important insight into the role of Chloroflexi in WWTPs, thus contributing to the broader goal of understanding the ecology of these biotechnologically important systems.

activated sludge, filamentous bulking, Chloroflexi, ‘Candidatus Amarolinea’, FISH, microautoradiography

INTRODUCTION

Members of the phylum Chloroflexi constitute a substantial proportion of the activated sludge community in full-scale activated sludge wastewater treatment plants (WWTPs), where they reportedly constitute up to 30% of the biovolume and often make up the majority of filamentous bacteria present (Beer et al. 2006; Morgan-Sagastume, Nielsen and Nielsen 2008; Mielczarek et al. 2012). Filamentous bacteria are generally believed to have structural importance for activated sludge flocs with good settling properties. However, overgrowth of certain filamentous species is associated with open and diffuse flocs as well as interfloc bridging, leading to a sometimes severe operational problem known as bulking (Wanner, Kragelund and Nielsen 2010). Some Chloroflexi species have also been proposed to be involved in the stabilization of problematic foams at WWTPs (Kragelund et al. 2011) and membrane fouling in membrane bioreactors (MBRs) (Ziegler et al. 2016). In addition, the high abundance of the Chloroflexi in wastewater treatment systems, often in the absence of bulking problems, indicates that they make a proportional contribution to the observed nutrient transformations of these systems. As such, the study of this phylum has implications for plant operation and our general understanding of the ecology of wastewater treatment.

Relatively little is known about the ecology of the Chloroflexi in nutrient removal WWTPs. In situ studies reveal a preference for the uptake of sugars, and a high level of surface associated hydrolytic enzymes indicates their involvement in the breakdown of complex organics (Kragelund et al. 2007, 2011; Xia, Kong and Nielsen 2007). Due to poor phylogenetic annotation of the routinely applied taxonomic databases, studies of activated sludge community dynamics often consider members of the Chloroflexi phylum as a whole—an approach that ignores the likely phenotypic diversity among its members (McIlroy et al. 2015). An understanding of the ecology of the phylum and how it relates to system function requires the identification and in situ characterization of the abundant genus-level phylotypes present.

Historically, the identification of bulking filaments has relied on classification keys based on morphological characteristics, with most morphotypes identified by an Eikelboom number (Eikelboom 2000; Jenkins, Richard and Daigger 2004). More recent phylogenetic identification of these filaments has relied on 16S rRNA gene-based clone library analyses coupled with fluorescent in situ hybridization (FISH) of plants with filamentous bulking. Most of the antecedent morphotypes are possessed by members of the Chloroflexi and include: 1851 (Beer et al. 2002); 0092 (Speirs et al. 2009); 0803 (Kragelund et al. 2011; Speirs, Tucci and Seviour 2015); 0914 (Speirs et al. 2011); 0041/0675 (Speirs et al. 2017) and several others (Kragelund et al. 2009)—noting that organisms possessing the same filamentous morphotype are often unrelated (Seviour et al. 1997; Speirs, Tucci and Seviour 2015), and closely related organisms can also possess different morphotypes (Speirs et al. 2017) (see Table S1, Supporting Information, for a summary of the phylogeny of different morphotypes). FISH probes available for known phylotypes reportedly cover many of the Chloroflexi present, but there is still a substantial portion of the phylum without genus-level probes (up to 90% by FISH in some plants) (Kragelund et al. 2011).

The recent extensive Microbial Database for Activated Sludge (MiDAS) 16S rRNA gene amplicon sequencing-based survey, covering >50 Danish full-scale WWTPs over a 10-year period, has given a comprehensive overview of the abundant core members of the full-scale activated sludge treatment plants (McIlroy et al. 2015). Importantly, the MiDAS taxonomy endeavours to provide genus-level classification of all abundant groups in wastewater treatment systems based on 16S rRNA gene sequence similarity cut-off recommendations for taxonomic ranks (Yarza et al. 2014). As such, we are now able to systematically characterise the numerically important Chloroflexi genus-level phylotypes that would otherwise be aggregated into higher taxonomic ranks such as family or order. The in situ physiology of some of these groups, such as ‘Candidatus Promineofilum’ (the B45 group) (McIlroy et al. 2016) and P2CN44 (Kragelund et al. 2011), has been determined, but several abundant phylotypes are novel and known only by their 16S rRNA gene sequence (McIlroy et al. 2015).

The aim of this study was to determine the in situ physiology of selected abundant novel Chloroflexi-activated sludge phylotypes. The extensive MiDAS survey of full-scale Danish nutrient removal activated sludge plants was used to identify the most abundant genus-level taxa belonging to the phylum Chloroflexi. FISH probes were designed for these phylotypes and applied in combination with microautoradiography (MAR) and histochemical staining for their in situ characterization.

METHODS

Biomass sampling and fixation for FISH

Biomass samples from the aerobic stage of selected full-scale activated sludge WWTPs with nutrient removal were fixed with 4% paraformaldehyde (PFA) for 3 h at 4°C. After fixation, samples were washed three times in sterile-filtered tap water, re-suspended in 50% ethanol in 1 × PBS solution [v/v], and stored at −20°C. For basic operational information for WWTPs sampled in this study, see Mielczarek et al. (2013).

Phylogenetic analysis and FISH probe design

Phylogenetic analysis and FISH probe design were performed with the ARB software package (Ludwig et al. 2004) with the MiDAS database (Release 2.1), which is a version of the SILVA database (Release 123 NR99) (Quast et al. 2013) curated for activated sludge and anaerobic digester sequences (McIlroy et al. 2015, 2017b). Target sequences selected for probe design were all those classified to the target genus in the MiDAS database (Release 2.1). The binding efficiencies of probes to their target and single-base-mismatched-non-target-sequences were assessed in silico with the mathFISH software (Yilmaz, Parnerkar and Noguera 2011). The Ribosomal Database Project (RDP) PROBE MATCH function (Cole et al. 2014) was used to identify non-target sequences with indels (McIlroy et al. 2011). Probe validation and optimization were based on generated formamide dissociation curves (Daims, Stoecker and Wagner 2005), where average relative fluorescent intensities of at least 50 cells calculated with ImageJ software (National Institutes of Health, Maryland, USA) were measured for varied hybridization buffer formamide concentration in increments of 5% (v/v) over a range of 0%–65% (v/v) (data not shown). In the absence of pure cultures being available for any of the phylotypes, each probe was optimised with activated sludge samples for which amplicon sequencing indicated high abundance of their respective target group (see Table 1). Unlabelled competitor probes were designed for all single base mismatched sequences to minimise the chance of mis-hybridization (Manz et al. 1992). Details for probes designed in this study have been deposited in the ProbeBase database (Greuter et al. 2016).

Table 1.

Probes designed and optimised in this study.

Probe	E. coli pos.	Target group	Coverage*	Non-target hits	Sequence (5’-3’)	[FA]%**
CFX64	64–99	‘Ca. Amarolinea spp.’	1/2	0	TCT ACC TAA GCA GAC CGT TC	30
CFX64_H1#	41–63	Helper for CFX64	N/A	N/A	AAC TTG CAT GTG TTA AGC ACG CC	N/A
CFX64_H2#	100–116	Helper for CFX64	N/A	N/A	TCA CCC GTG CGC CAC TG	N/A
CFX763A	763–783	‘Ca. Villigracilis spp.’ sub-group A	68/260	1	GTT TAC TAC CCT AGC TTT CGC	45
CFX763A_C1	763–783	Competitor for CFX763A	N/A	N/A	GTT CAC TAC CCT AGC TTT CGC	N/A
CFX763A_C2	763–783	Competitor for CFX763A	N/A	N/A	GTT TAC TCC CCT AGC TTT CGC	N/A
CFX763A_C3Δ	763–783	Competitor for CFX763A	N/A	N/A	GTT TGC TAC CCT AGC TTT CGC	N/A
CFX763A_C4Ψ	763–783	Competitor for CFX763A	N/A	N/A	GTT TAC TAC CCT AGC TGT CGC	N/A
CFX763AB_H1A	784–808	Helper for CFX763A, CFX763B	N/A	N/A	TAG GAT TAC CGG GGT CTC TAA TCC C	N/A
CFX763AB_H1B	784–808	Helper for CFX763A, CFX763B	N/A	N/A	TAG GAT TAC CSG GGG TCT CTA ATC CC	N/A
CFX763B	763–783	‘Ca. Villigracilis spp.’ sub-group B	91/260	0	GTT TAC TAC CCT AGC TGT CGC	45
CFX763B_C1§	763–783	Competitor for CFX763B	N/A	N/A	GTT TAC TAC CCT AGC TTT CGC	N/A
CFX763mix	763–783	‘Ca. Villigracilis spp.’	159/260	1	GTT TAC TAC CCT AGC TKT CGC	45
CFX449	449–491	‘Ca. Sarcinithrix spp.’	12/14	1	GGG ATA CCG TCC TTG TCT CT	50
CFX449_C1	449–491	Competitor for CFX449	N/A	N/A	GGG GTA CCG TCC TTG TCT CT	N/A
CFX449_H1	492–516	Helper for CFX449	N/A	N/A	ACG TAG TTA GCC GAG ACT TAT TCC T	N/A
CFX449_H2	422–448	Helper for CFX449	N/A	N/A	TCT CCC AGA AAA GRR GTT TAC GAC CCG	N/A
CFX1151	1151–1170	‘Ca. Sarcinithrix spp.’	12/14	1	TTG ACT CCG GCA GTC CCA CT	50
CFX1151_C1	1151–1170	Competitor for CFX1151	N/A	N/A	TTG ACA CCG GCA GTC CCA CT	N/A
CFX1151_H1	1171–1189	Helper for CFX1151	N/A	N/A	ATC CCC ACC TTC CTC CGG T	N/A
CFX1151_H2A	1102–1127	Helper for CFX1151	N/A	N/A	TAA CTA GTA GBG AGG GTT GCG CTC GT	N/A
CFX1151_H2B	1102–1127	Helper for CFX1151	N/A	N/A	TAA CTA GTA GCA GGG GTT GCG CTC GT	N/A
CFX67⌘	67–89	‘Ca. Sarcinithrix spp.’	6/14	0	TTC CGA AGA TYA GGT TCG	35
CFX67 comp⌘	67–89	Competitor for CFX67	N/A	N/A	TTC CGA AGA TCG GGT TCG	N/A
CFX67-H46⌘	46–66	Helper for CFX67	N/A	N/A	TTC GAC TTG CAT GTG TTA RGC	N/A
CFX67-H95⌘	95–112	Helper for CFX67	N/A	N/A	CCG TRC GCC ACT AAC CYT	N/A

*Coverage of groups as defined in MiDAS database (Release 2.1) (McIlroy et al. 2015). Values given as group hits / group totals; NA = Not applicable; **Recommended optimal formamide concentration for use in FISH hybridizations. #These helper probes are not required for optimal fluorescent signal of their respective probe. ΔTested against pure culture of Ornatilinea apprima possessing CFX763A_C3 binding site; no fluorescent signal observed when O. apprima was hybridised with probe CFX763A at the recommended 45% FA indicating that this competitor probe is not required. No other single mismatched sequences were validated and thus it is recommended to use all other listed competitor probes to minimise the chance of mis-hybridisation to non-target organisms. Mismatched bases are underlined in the competitor probe sequences. ΨDo not use this competitor if CFX763A is applied with CFX763B. §Do not use this competitor if CFX763B is applied with CFX763A. ⌘These probes are taken from Speirs et al. (2011). Fluorescently labelled probes are indicated in bold typeface.

FISH

FISH was performed as detailed by Daims et al. (2005) using the probes designed in this study as well as CFX197 and CFX223, targeting ‘Ca. Promineofilum’ (Speirs et al. 2009); CFX1223 (Björnsson et al. 2002), and GNSB941 (Gich, Garcia-Gil and Overmann 2001), applied as a mix to target the phylum Chloroflexi; EUB-338-I, EUB338-II and EUB338- III (Amann et al. 1990; Daims et al. 1999), applied as a mix (EUBmix) to cover all bacteria; NON-EUB as a negative control for hybridization (Wallner, Amann and Beisker 1993). The hybridization conditions applied for each probe are given in Table 1 or as recommended in their original publications. Quantitative FISH (qFISH) biovolume fractions of individual Chloroflexi genera were calculated as a percentage area of the total biovolume, hybridizing the EUBmix probes, that also hybridises with the specific probe. qFISH analyses were based on 25 fields of view taken at 630× magnification using the Daime image analysis software (Daims, Lücker and Wagner 2006). Microscopic analysis was performed with an Axioskop epifluorescence microscope (Carl Zeiss, Oberkochen, Germany), an LSM510 Meta laser scanning confocal microscope (Carl Zeiss), and a white light laser confocal microscope (Leica TCS SP8 X).

Morphological classification

Microscopic observations of wet mount preparations and Gram and Neisser staining were performed according to the methods of Eikelboom (2000). Morphotype classification was carried out conforming to the described classification keys, based on morphological features of the filaments: shape, length and diameter of filamentous bacteria, motility, presence of branching, attached growth of other bacteria to the filaments, septa between adjoining cells (visible or not visible), shape of cells, presence of a sheath and sulphur granules (Eikelboom 2000).

Histochemical staining

Following FISH, polyphosphate inclusions were stained with 28 μM 4′,6-diamidino-2-phenylindole (DAPI) for 1 h at 4°C in the dark. After staining at such relatively high DAPI concentrations, the polyphosphate granules appear bright yellow with fluorescence microscopy (Streichan, Golecki and Schon 1990). Polyhydroxyalkanoates (PHA) were stained with 1% [w/v] Nile Blue A for 10 mins at 55°C essentially as described previously (Ostle and Holt 1982). FISH images were acquired prior to staining and the same fields of view relocated.

Microsphere adhesion to cells (MAC)

MAC was performed on a sample of fresh sludge to identify the hydrophobicity of target cells, applying the method of Kragelund et al. (2005) using a sonicated solution of 0.2 µm FluoSpheres fluorescent sulphate-modified microspheres with excitation and emission properties (505/515 nm) (Life Technologies Corporation, Eugene OR, USA).

Microautoradiography

Biomass was sampled from the aerobic stage of full-scale activated sludge WWTPs in Aalborg West, Bjergmarken, Ringkøbing and Odense North-West, Denmark. These plants were confirmed to contain the target organisms, including OTUs sharing ≥99% 16S rRNA gene sequence similarity with the proposed type sequences for each proposed genus (Tables S2 and S3, Supporting Information). All plants are designed for N removal and enhanced biological phosphorus removal (EBPR) and have stable performance. For further details on the plants assessed in this study, see Mielczarek et al. (2013). Biomass samples were stored at 4°C and all incubations performed within 24 h from sampling. The MAR incubation protocol was based on the method of Nierychlo, Nielsen and Nielsen (2015). Activated sludge was aerated for 40 min at room temperature prior to MAR incubation to reduce the residual substrates, oxygen and NO_x present. Sludge was then diluted with filtered sludge water from the same plant to yield a biomass concentration of 1 mgSSmL⁻¹ and transferred to 11 mL serum bottles. Radiolabelled substrates were added to yield a total radioactivity of 10 µCi mg⁻¹ SS. The following substrates were used: [³H]acetate, [³H]galactose (Amersham Biosciences, UK), [³H]glucose, [³H]mannose, [¹⁴C]pyruvate (Perkin-Elmer, Waltham MA, USA), [³H]amino acid mix, [¹⁴C]butyric acid, [³H]fructose, [³H]glycerol, [³H]ethanol, [³H]lactate, [³H]NAG, [¹⁴C]propionate (American Radiolabeled Chemicals Inc., Saint Louis MO, USA). The corresponding cold substrate was added to yield a total concentration of 2 mM organic substrate. To achieve anaerobic conditions, prior to substrate addition, oxygen was removed by repeated evacuation of the headspace and subsequent injection of O₂-free N₂. Anaerobic incubations with selected substrates were supplemented with 0.5 mM nitrite or 2 mM nitrate to assess their use as electron acceptors. The supernatant concentrations were monitored using Quantofix Nitrate/Nitrite strips (Macherey-Nagel, Düren, Germany) and readjusted to their initial concentrations anaerobically to prevent exhaustion. Samples were incubated for 3 h at room temperature (approx. 21°C) on a rotary shaker at 250 rpm. Incubations with [¹⁴C]carbonate (American Radiolabeled Chemicals Inc., Saint Louis MO, USA) contained 20 µCi mg⁻¹ SS of radiolabelled substrate. 1 mM NH₄Cl was added to half of the incubations to investigate ammonia and nitrite (produced from the oxidation of added ammonia) oxidation activity. These samples were incubated aerobically (same conditions as above) for 5 h, as suggested by Daims et al. (2001). A pasteurised biomass (heated to 70°C for 10 min) incubation was prepared as a negative control to assess possible silver grain formation due to chemography. Incubations were terminated by the addition of cold PFA to a final concentration of 4% [w/v]. Samples were fixed for 3 h at 4°C and subsequently washed three times with sterile-filtered tap water. Aliquots of 30 μL of the biomass were gently homogenised between glass coverslips. Following FISH (see earlier), coverslips were coated with Ilford K5D emulsion (Polysciences, Inc., Warrington, PA, USA), exposed in the dark for periods of 10 days, and developed with Kodak D-19 developer.

RESULTS

Distribution of Chloroflexi in Danish full-scale WWTPs

Phylogenetic tree based on 16S rRNA gene sequences shows all the abundant Chloroflexi groups found in Danish WWTPs with nutrient removal and their phylogenetic relationship (Fig. 1). The Chloroflexi phylum is among the most abundant phyla in full-scale systems in Denmark, constituting on average 10.6% of the total reads across all plants (Fig. 2a). The Chloroflexi classes Anaerolineae, Caldilineae, Ardenticatenia and SJA-15 made up the majority of members of the phylum present (Fig. 2b). All the abundant genus-level phylotypes (Fig. 2c) are novel, having no available cultured representatives, and were initially given provisional alphanumeric names in the MiDAS database (McIlroy et al. 2015) (not shown here). Based on their characterization, as presented in this study and previous publications, we propose new, previously unpublished, candidate names (Murray and Stackebrandt 1995). These names will be incorporated into future versions of the MiDAS taxonomy (McIlroy et al. 2015) and are used throughout this report. These include: ‘Candidatus Sarcinithrix’ (Sar.ci'ni.thrix. L. fem. n. sarcina, a package, bundle; Gr. fem. n. thrix, hair; N.L. fem. n. Sarcinithrix a hair bundle; formerly Ca. Sarcinathrix (release 2.1), Table S2, Supporting Information), ‘Candidatus Villigracilis’ (Vil.li.gra'ci.lis. L. masc. n., villus, tuft of hair; L. adj. gracilis, slim, slender; N.L. fem. n. Villigracilis a slender tuft of hair; formerly MiDAS taxon SBR1029 (release 1.21) and Ca. Villogracilis (release 2.1), Table S3, Supporting Information), ‘Candidatus Defluviifilum’ (De.flu.vi.i.fi'lum. L. neut. n. defluvium, sewage; L. neut. n. filum, a thread; N.L. neut. n. Defluviifilum a thread from sewage; formerly MiDAS taxon P2CN44 (release 1.21), Table S4, Supporting Information) and ‘Candidatus Amarolinea’ (A.ma.ro.li'ne.a. Gr. fem. n. amara, conduit, channel, sewer; L. fem. n. linea, a thread, line; N.L. fem. n. Amarolinea a thread from a sewer (Andersen et al. 2018); formerly MiDAS taxon C10_SB1A (release 1.21) and Candidatus Amarilinum (release 2.1)). ‘Kouleothrix spp.’, possessing the 1851 bulking filament morphotype, was present in low abundance with a median and mean of 0.04% and 0.4%, respectively. ‘Ca. Defluviifilum’, ‘Ca. Promineofilum’, ‘Ca. Villigracilis’ and ‘Ca. Sarcinithrix’ represent the four most abundant genera by median read abundance (Fig. 2c), collectively constituting on average 6.2% of the total reads across all Danish plants assessed in this study. These phylotypes were relatively stably present across the different WWTPs (Fig. S1, Supporting Information) and therefore represent core members of the microbial community of these systems. As little is known regarding the physiology of the latter two genera, they were selected for a detailed characterization in this study. Relative to these phylotypes, the ‘Ca. Amarolinea’ showed a much more dynamic distribution and periodically reached abundances in excess of 30% of the amplicon reads, which would indicate a likely role in bulking episodes in Denmark. As such, this genus was also selected for characterization in this study.

Figure 1.

Maximum-likelihood (PhyML) 16S rRNA gene phylogenetic tree of abundant activated sludge phylotypes (bold typeface) and isolated members of the phylum Chloroflexi. The alignment used for the tree applied a 20% conservational filter to remove hypervariable positions, giving 1108 aligned positions. Phylogenetic classification is indicated with black brackets and is based on the MiDAS database (Release 2.1). Coverage of probes relevant to the current study is shown with red-coloured brackets. Eikelboom morphotype numbers of previously described phylotypes are designated with blue-coloured brackets. Bootstrap values from 100 resamplings are indicated for branches with >50% (white dot), 50%–70% (gray) and >90% (black) support. Species of the phylum Cyanobacteria were used as the outgroup. The scale bar represents substitutions per nucleotide base.

Figure 2.

Distribution of Chloroflexi in 25 full-scale activated sludge WWTPs sampled 2–4 times per year from 2006 to 2015. (a), 10 most abundant phyla in Danish WWTPs. (b), 10 most abundant Chloroflexi classes in Danish WWTPs (c), 10 most abundant Chloroflexi genera in Danish WWTPs. X-axis shows the relative read abundance in percentage of total bacteria.

Figure 3.

Composite FISH micrographs of novel Chloroflexi genera in full-scale activated sludge. Specific probes (Cy3-label, red) target (a) ‘Ca. Amarolinea’, (b) ‘Ca. Villigracilis’ and (c) ‘Ca. Sarcinithrix’, and EUBmix probe (Cy5-label, blue) targets most bacteria. Activated sludge was sampled from (a) Bjergmarken WWTP, (b) Odense North East WWTP and (c) Aalborg West WWTP. Target filaments appear magenta, while all other cells appear blue. Scale bars represent 20 μm.

Figure 4.

FISH and corresponding bright-field MAR micrographs showing sugar uptake by three abundant Chloroflexi phylotypes (‘Ca. Amarolinea’ hybridised probe CFX64; ‘Ca. Villigracilis’ sub-group A hybridised probe CFX763A; ‘Ca. Sarcinithrix’ hybridised probe CFX1151). Activated sludge was sampled in Aalborg West or Odense North West WWTPs. Target cells in FISH micrograph overlays appear yellow: specific probe (red) + EUBmix (green); and non-target cells appear green (EUBmix only). Black silver granules indicate positive MAR signal. Scale bar represents 10 µm.

Phylogeny and FISH probe design

‘Candidatus Villigracilis’ are members of the Anaerolineaceae, which is currently the sole family of the class Anaerolineae in the MiDAS and SILVA taxonomies. These sequences fall within order SBR1031 in the Greengenes taxonomy (McDonald et al. 2012). The CFX763A and CFX763B probes were designed to cover separate sub-groups (A and B) of the ‘Ca. Villigracilis’ (Fig. S2, Supporting Information)—collectively covering >60% of the MiDAS database sequences classified to the genus. The target region is not covered by the V1–3 region amplicon sequences, although both probes match the full-length database sequences most closely related to the abundant OTU sequences (data not shown). When applied to full-scale activated sludge biomass, both probes exclusively hybridised thin filaments (0.3–0.4 µm wide and 15–50 µm long) that were often observed in bundles and almost entirely located within the flocs (Figure 3b). The CFX763AB_H1A and CFX763AB_H1B helper probes are recommended to give optimal fluorescence signal for both probes. Amplicon sequencing of the V1–3 region of the 16S rRNA cannot be used to confidently separate the A and B sub-groups, due to high sequence similarity of the sequence region, but qFISH indicates that the former is the more numerically important of the two.

The ‘Ca. Amarolinea’ genus falls within the novel MiDAS Chloroflexi class-level-group SJA-15, together with the also abundant genus ‘Ca. Sarcinithrix’ (Fig. 2). A probe to cover the entire ‘Ca. Amarolinea’ group was not identified, so the CFX64 (Table 1) was designed for the abundant amplicon OTUs (OTU_3 and OTU_4592, Fig. S3, Supporting Information) and the most closely related full-length database sequence (AF513086). These sequences share >97% similarity. As the probe covers these abundant OTUs, it should cover the majority of members of the genus in full-scale activated sludge in Denmark. When applied to activated sludge, the probe hybridised exclusively to filamentous bacteria (Figure 3a; see later for a detailed description of their morphology). A strong positive signal was obtained without addition of designed unlabelled helper probes CFX64_H1 and CFX64_H2 (Table 1), which did not noticeably improve fluorescent signal (data not shown). The ‘Ca. Amarolinea’ filaments constituted up to 30% of the community biovolume in some full-scale WWTPs in Denmark—confirming their high abundance with amplicon sequencing (see Table 2).

Table 2.

Abundance estimation: 16S rRNA amplicon sequencing and qFISH (percentage of total).

WWTP	Sample date	Abundance (%)
		Sequencing*	qFISH
‘Ca. Villigracilis’
Odense North East	September 2015	9.0	Sub-group A: 12 Sub-group B: 1
Aalborg East	August 2015	6.0	Sub-group A: 11 Sub-group B: > 1
Horsens	August 2006	6.0	Sub-group A: 11 Sub-group B: > 1
Fredericia	October 2015	3.3	Sub-group A: 6 Sub-group B: > 1
‘Ca. Amarolinea’
Aalborg West	May 2014	11.9	16
Bjergmarken	August 2013	33.7	30
Odense North West	August 2010	24.2	17
‘Ca. Sarcinithrix’
Viby	August 2008	2.3	2
Boeslum	August 2008	2.1	3
Bjermarken	February 2013	1.2	2
Ejby Mølle	May 2008	1.0	1
Aalborg West	February 2012	0.8	1

*Taken from the 16S rRNA gene amplicon sequencing MiDAS survey of Danish WWTPs (McIlroy et al. 2015).

FISH probes were already available to target the ‘Ca. Sarcinithrix’ (CFX67), which was shown to possess the Eikelboom 0914 morphotype in nutrient removal activated sludge systems in Australia (Speirs et al. 2011). Application of CFX67 probe did not give significant positive fluorescence to bacterial cells in Danish plants. The probe misses several full-length sequences classified to the genus and does not cover any of the abundant OTU sequences (>0.1% average read abundance in at least 1 plant) from the MiDAS full-scale survey. As such, two new probes were designed to give better coverage of the clade. The designed CFX449 and CFX1151 individually target 85% of the full-length database sequences (Figure 3c). As such, they can be applied with different fluorochromes to confirm specific coverage of the genus, or together as a mix to give higher signal (Fig. S4, Supporting Information). Helper probes were not required for either probe, but did give a more even signal over the filament, which was also reported by Speirs et al. (2011) for the CFX67 probe. The few filaments positive for the CFX67 probe in the Danish WWTPs assessed in this study also hybridised the CFX449 and CFX1151 probes (Fig. S4, Supporting Information). The FISH signal for the CFX449 and CFX1151 overlap in tested samples (Fig. S4, Supporting Information) and Quantitative FISH values with these probes were similar to amplicon-sequencing based estimates (Table 2), which suggests that the probes provide good specificity in targeting the ‘Ca. Sarcinithrix’.

All three phylotypes studied gave positive hybridization signal with the EUBmix probe set, which is commonly applied as a universal probe targeting bacteria (EUB338, EUB338-II and EUB338-III). This is of interest, given that many Chloroflexi reportedly lack the target site for the EUBmix probe set and fail to hybridise the probe in situ (Kragelund et al. 2007, 2011; Speirs et al. 2009). Most 16S rRNA gene sequences of the ‘Ca. Amarolinea’, ‘Ca. Sarcinithrix’ and ‘Kouleothrix’ contain the site for the EUB338 probe, the ‘Ca. Villigracilis’ sequences possess the EUB338-III site, and all groups have been shown to hybridise the probe in situ. Most members of the ‘Ca. Defluviifilum’ have one mismatch, though in silico analysis with the MathFISH software (Yilmaz, Parnerkar and Noguera 2011) predicts positive binding (with a calculated melting formamide point ([ΔFA]_m) of 50%), which is confirmed in situ for filaments hybridizing the T0803–0654 probe designed to target the group (Kragelund et al. 2011). Thus, of the abundant phylotypes, only the ‘Ca. Promineofilum’ genus is not covered by the EUBmix probe set (Speirs et al. 2009).

Morphological description and classification

The morphological properties of the CFX64 positive filaments, representing the ‘Ca. Amarolinea’ genus, were investigated in detail for association to a morphotype of the well-known antecedent classification systems (Eikelboom 1975). Though not clearly visible, the cells appeared to be rectangular with no visible septa, a trichrome thickness of 1.0–2.2 μm, and a length in the 20–140 μm range. They were non-motile, Gram stain negative, with no branching or attached growth. The whole filaments stained blue/violet with the Neisser stain, with no visible volutin granules. Excess polyphosphate stores were not observed with DAPI staining. Cells did not appear to contain excess stores of PHAs, with negative results with Nile blue A staining and only small positive granules observed with the Sudan black stain. From these observations, primarily based on the characteristic violet color of the cells after Neisser staining, it is suggested that the morphology of the filament is most consistent with the Eikelboom type 0092 morphotype (Eikelboom 2000). The ‘Ca. Promineofilum’ also reportedly has the 0092 morphotype (Speirs et al. 2009), but there was no observed overlap between the CFX64 and the CFX197 probes targeting the ‘Ca. Amarilimum’ and ‘Ca. Promineofilum’ genera, respectively (Fig. S5, Supporting Information). The ‘Ca. Promineofilum’ phylotype is also thinner in appearance, with a trichome diameter of approx. 0.8 μm (Speirs et al. 2009). Very few of the fluorescent sulphate modified microspheres attached to the CFX64 positive filaments (data not shown), indicating that they do not have a hydrophobic surface and are likely not involved in foam formation.

Morphological descriptions are already reported for members of the ‘Ca. Sarcinithrix’ (Speirs et al. 2011) and ‘Ca. Defluviifilum’ (Kragelund et al. 2011; Speirs et al. 2017). The surface hydrophobicity of ‘Ca. Sarcinithrix’ was assessed for the first time here, where it was determined to be hydrophilic and therefore unlikely to be involved in foam formation. Morphological classification of the ‘Ca. Villigracilis’ was not successful due to their location within the floc, making interpretation of staining analyses difficult.

In situ substrate uptake

The results for substrate uptake by probe-defined ‘Ca. Villigracilis’ sub-groups A and B, ‘Ca. Amarolinea’, and ‘Ca. Sarcinithrix’ using MAR-FISH under various conditions are shown in Table S5 (Supporting Information), selected micrographs showing MAR signal are shown in Figure 4 and a summary of known in situ traits of abundant Chloroflexi is given in Table 3. ‘Ca. Villigracilis’ sub-group A, ‘Ca. Amarolinea’, and ‘Ca. Sarcinithrix’ only utilised sugars of the 13 substrates tested, consistent with other characterised Chloroflexi genera. The phylotypes differed in the types of sugars taken up, noting that variation was also observed within the ‘Ca. Amarolinea’ genus—with some filaments strongly positive and others clearly negative for fructose uptake. ‘Ca. Villigracilis’ sub-group B filaments were negative for all substrates and conditions. It may be that they have a relatively lower activity than the much more abundant sub-group A filaments that is below the detection of MAR. Further analyses are required to assess the reason for the observed lack of substrate uptake. All three genera were able to take up substrates under anoxic conditions, suggesting fermentative metabolisms. Anoxic uptake of sugars in presence of nitrate/nitrite was also observed, but their potential for denitrification is unclear, given that uptake was also observed without nitrate/nitrite addition. The same ambiguous results were obtained for the ‘Ca. Defluviifilum’ (Kragelund et al. 2011). A recently published genome representing the genus ‘Ca. Amarolinea’ indicates that they perform dissimilatory nitrate reduction to ammonium (DNRA) (Andersen et al. 2018), similar to ‘Ca. Promineofilum breve’ (McIlroy et al. 2016). The ability for nitrification was also assessed for these groups in this study, given that Nitrolancetus hollandicus, a nitrite oxidizing member of the class Thermomicrobia of the Chloroflexi, was isolated from activated sludge (Sorokin et al. 2012); albeit in a different class to the abundant members of the Chloroflexi in Danish full-scale systems. None of the genera appeared to be behaving as nitrifiers, with no observed uptake of labelled CO₂ in the presence of ammonia (Table 3 and Table S1, Supporting Information), while a positive MAR signal was noted for the well-known nitrifiers Nitrosomonas and Nitrospira, targeted by the probe Cluster6a_192 (Adamczyk et al. 2003) and Ntsp662 (Daims et al. 2001), respectively.

Table 3.

Summary of the known in situ physiology for Chloroflexi commonly found in activated sludge.

Property	‘Ca. Amarolinea’	‘Ca. Villigracilis’	‘Ca. Sarcinithrix’	‘Ca. Promineofilum’	‘Ca. Defluviifilum’	‘Kouleothrix’	Chloroflexi summary
Substrate uptake
Formate	ND	ND	ND	ND	−6	−2	−2,6
Acetate	−1	−1	−1	−9	−6,8	−2	+2/−2,6,8,9
Propionate	−1	−1	−1	−9	−6,8	−2	−2,6,8,9
Butyrate	−1	−1	−1	−9	−6	+2/−2	+2/−2,6,9
Lactate	−1	−1	−1	ND	ND	ND
Pyruvate	−1	−1	−1	−9	−6,8	+2/−2	+2/−2,6,8,9
Oleic acid	ND	ND	ND	−9	+8/−6,8	−2	+5/−2,6,8,9
Glucose	+1	−1	+1/−1	+9	+6,8	+2	+2,3,6,8,9
Mannose	−1	−1	−1	ND	+8/−6	+2/−2	+2,7/−2,6
Galactose	+1/−1	−1	+1/−1	ND	+8/−6	−2	+2,7/−2,6
Fructose	+1/−1	+1/−1	+1	ND	ND	ND	ND
NAG	−1	−1	−1	−9	+6/−6	+2	+2,3,6/−2,9
Glycine	ND	ND	ND	ND	−6,8	−2	−2,6,8
Leucine	ND	ND	ND	ND	−6,8	+2/−2	+2/−2,6,8
Amino acid mix	−1	−1	−1	−9	ND	ND	−9
Thymidine	ND	ND	ND	ND	−6	ND	+2/−6
Ethanol	−1	−1	−1	−9	−6,8	−2	−2,6,8,9
Glycerol	−1	−1	−1	−9	−8	ND	−8,9
Chemoautotrophy	−1*	−1*	−1*	ND	−6	−2**	−2,6
Electron acceptor conditions
Aerobic	+1	+1/−1	+1	+9	+6,8	+2	+2,6,8,9
Anoxic	+1	+1/−1	+1	+9	+6	−2	+6,9/−2,6
Anoxic + NO2−	+1	+1/−1	ND	ND	+6	−2	+6/−2,4,6
Anoxic + NO3−	+1	+1/−1	ND	ND	+6	−2	+6/−2,6
Storage polymers (staining)
PHA	−1	ND1	−7	−5	−6	ND	−5,6
Polyphosphate	−1	ND1	−7	−5	+6/−6	ND	+6/−5,6
Hydrophobic surface (MAC)	−1	ND1	−1	−9	+6	−2	+6/−2,9

References: 1. This study; 2. (Kragelund et al. 2007); 3. (Miura, Watanabe and Okabe 2007); 4. (Morgan-Sagastume, Nielsen and Nielsen 2008); 5. (Speirs et al. 2009); 6. (Kragelund et al. 2011); 7. (Speirs et al. 2011); 8. (Kindaichi et al. 2013); 9. (McIlroy et al. 2016). *carbonate with and without ammonia; **anaerobic with 2 mM thiosulfate present.

DISCUSSION

The results of this study, detailing the distribution and in situ morphology and physiology of individual abundant phylotypes of the phylum Chloroflexi, give valuable insight into their potential importance to the operation of WWTPs. The ‘Ca. Promineofilum’, ‘Ca. Defluviifilum’, ‘Ca. Villigracilis’ and ‘Ca. Sarcinithrix’ appear to be consistently abundant filamentous members of the full-scale WWTP community, where they are very likely to make an important contribution to the bulk nutrient transformations. Relative to other abundant Chloroflexi genera, the distribution of the ‘Ca. Amarolinea’ is much more dynamic, reaching levels of >30% of the biovolume in some plants (Fig. S6, Supporting Information). As such, it is more likely that the ‘Ca. Amarolinea’ are responsible for acute bulking episodes in Danish WWTPs, while other abundant phylotypes are core members of the community that are possibly important to floc structure and the breakdown of organics (see later). While the ‘Ca. Villigracilis’ are filamentous, they are almost exclusively found within the flocs and are unlikely to contribute to bulking episodes. This observation explains why they have never been detected before and highlights the value of large-scale full-scale surveys, such as MiDAS, for describing the abundant members of wastewater treatment systems and the importance of in situ analyses to evaluate the role of organisms in floc structure and settleability. In addition to a potential contribution to bulking, members of the ‘Ca. Promineofilum’ are also putatively involved in membrane fouling in MBR systems (Ziegler et al. 2016), and ‘Ca. Defluviifilum’ has a hydrophobic cell surface that has been implicated in the stabilization of problematic foams on the surface of reactor tanks (Kragelund et al. 2011).

In recent studies Speirs et al. (2015; 2017) described filamentous Chloroflexi phylotypes in Australian WWTPs possessing the bulking Eikelboom morphotypes 0803 and 0041/0675 that are commonly observed by light microscopy in Danish plants (Mielczarek et al. 2012). These phylotypes classified within the Anaerlolineae and Caldilinaea, respectively. However, none of these phylotypes were found to be abundant in the amplicon-based survey of Danish systems (data not shown). Their FISH probe defined 0675 phylotype falls within the MiDAS defined ‘Ca. Defluviifilum’ genus (v. 2.1), shown previously to possess the Eikelboom 0803 phylotype (Kragelund et al. 2011), which they suggest should be split into a separate genus based on the divergence of its 16S rRNA gene sequence and the different morphology (only ca. 90% sequence similarity). Further characterisation of members of the current ‘Ca. Defluviifilum’ clade, including obtaining representative genomes, will help to resolve the phylogeny of these organisms. Members of the ‘Kouleothrix’ genus, shown to possess the Eikelboom 1851 bulking morphotype (Beer et al. 2002; Kohno, Sei and Mori 2002), were also relatively low in abundance in Danish plants (Kragelund et al. 2011) but are common in other countries—such as Japan, where they have been suggested as a major contributor to filamentous bulking episodes (Nittami et al. 2017). Global surveys will help to establish how relevant the abundant Danish phylotypes characterised in this paper are worldwide.

The abundant phylotypes in Danish WWTPs appear to be organoheterotrophic, fermentative, facultative anaerobes. This is suggested from their exclusive utilization of sugars and ability for anaerobic carbon uptake. Fermentative pathways were also annotated in the sole genomes from these genera—‘Ca. Promineofilum breve’ (McIlroy et al. 2016) and ‘Ca. Amarolinea aalborgensis’ (Andersen et al. 2018). The physiology determined for the ‘Ca. Defluviifilum’ and ‘Ca. Villigracilis’ is consistent with other members of their respective classes (Caldilineae and Anaerolineae), which are mostly fermentative organoheterotrophic filaments growing on sugars and/or amino acids (Sekiguchi et al. 2003; Yamada et al. 2006, 2007; Grégoire et al. 2011; Nunoura et al. 2013; Podosokorskaya et al. 2013; Imachi et al. 2014; McIlroy et al. 2017a). ‘Ca. Villigracilis’ is the first reported facultative anaerobic genus of the Anaerolineae, with all isolates of other described genera being obligate anaerobes.

In addition to the likely fermentation of sugars by the abundant Chloroflexi, metabolic diversity within the less abundant members of the phylum is evident—with reported assimilation of short and long chain fatty acids, amino acids and glycerol (Table 3). Some members of the phylum also appear unable to take up substrates anaerobically and may scavenge sugars released from aerobic breakdown of complex organic matter. Uptake of N-aminoglucosamine (NAG), a component of peptidoglycan and lipopolysaccharides, for some Chloroflexi, suggests a specific role in the breakdown of cellular material (Kragelund et al. 2007, 2011; Miura, Watanabe and Okabe 2007). Of the known Chloroflexi genera reportedly abundant in activated sludge, only the ‘Kouleothrix’ genus is seemingly unable to take up substrates anaerobically in situ (Kragelund et al. 2007), although isolates of the genus are capable of anaerobic fermentative growth on sugars (Kohno, Sei and Mori 2002).

In this study, the most abundant members of the Chloroflexi in Danish nutrient removal WWTPs were identified and their ecophysiology described. These phylotypes appear to differ in their impact on plant operation—with suggested importance in sludge settleability, foaming and membrane fouling being associated with different groups. All abundant members of the phylum likely ferment sugars, and future research should aim to obtain representative genomes for each in order to carry out more detailed comparison of their metabolic activities. Such an approach will explain important questions regarding how these organisms coexist, and what conditions determine their relative abundances. The FISH probes designed in this study will allow more hypothesis-based in situ investigation of their physiologies, based on genomic evidence. The taxonomic annotation and design of FISH probes for the abundant Chloroflexi, in combination with the high throughput nature of 16S rRNA gene amplicon sequencing, will also allow their routine observation and study. Defining and naming these novel genus level taxa importantly provides the foundation, upon which information on their morphology, distribution and physiology can be gathered for an in-depth understanding of their ecology and how this might relate to operational parameters.

ACKNOWLEDGEMENTS

We thank K. Vilstrup for help with the latin names.

FUNDING

This work was supported by the Danish Council for Independent Research (grant no. 4093–00127A); Innovation Fund Denmark (EcoDesign); The Obel Family Foundation; Danish wastewater treatment plants in MiDAS; and Aalborg University.

Conflicts of interest. None declared.

Author notes

Present address: Australian Centre for Ecogenomics, University of Queensland, Australia