Stability of aerobic granular sludge for simultaneous nitrogen and Pb(II) removal from inorganic wastewater

ABSTRACT In this paper, we proposed a strategy for the establishment of an aerobic granular sludge (AGS) system for simultaneous nitrogen and Pb(II) removal from inorganic wastewater. AGS was stored in lead nitrate solution to select functional bacteria resistant to lead poison, and then an AGS system for ammonia nitrogen (180-270 mg/L) and Pb(II) (15-30 mg/L) removal was established based on carbon dosing and a two-stage oxic/anoxic operational mode. After storage for 40 days, the stability of AGS decreased because specific oxygen uptake rate, nitrification rate and abundance of Nitrosomonas decreased to different degrees compared with those before storage. During the first 70 days of the recovery process, AGS in R1 (the blank reactor) and R2 (the control reactor) both experienced a first breakage and then regranulation process. The main properties of AGS in reactors R1 and R2 tended to be stable after days 106 and 117, respectively, but the structure of steady-state AGS in R2 was more compact. The total inorganic nitrogen (TIN) in effluent from R1 and R2 basically remained below 25 mg/L after days 98 and 90, respectively. The Pb(II) concentration in effluent from R2 was always below 0.3 mg/L. On day 140, the relative abundance of Nitrosomonas in R2 (6.17%) was significantly lower than that in R1 (12.15%), whereas the relative abundance of denitrifying bacteria was significantly higher than that in R1 (62.44% and 46.79%). The system removed 1 kg of influent TIN only consuming approximately 1.85 kg of carbon source, demonstrating clear advantages in energy savings. GRAPHICAL ABSTRACT


Introduction
Biological nitrogen removal (BNR) has been widely applied due to its low operating cost and simple maintenance [1].The activated sludge process, a typical BNR technology, has been successfully applied worldwide for more than a century [2].Although BNR usually has a lower operating cost than physical and chemical technologies, it is generally considered only to be suitable for the treatment of nontoxic or lowtoxic organic wastewater.However, developing industries release many toxic and harmful pollutants into biochemical treatment systems along with sewage, which usually has an adverse impact on the stability of the BNR system.Among them, heavy metals, which can be concentrated in the food chain, have attracted increasing concern [3].The wastewater discharged by mining, electroplating, battery, paint and chemical industries often contains Pb(II), and most of the residue Pb(II) is concentrated in activated sludge after being treated by municipal wastewater systems [4].The toxic effect of Pb(II) on BNR systems will likely gradually emerge with the accumulation of Pb(II) [5].Unfortunately, methods for the selection of functional bacteria resistant to lead poison are lacking.
Aerobic granular sludge (AGS) is a granular aggregate formed by microorganisms under high selection pressure [6].Compared with activated sludge, AGS has a compact structure, high toxicity resistance and fast settling velocity (30-90 m/h) [7], making it a promising technology in wastewater treatment.The unique stratified structure of AGS allows different functional bacteria to coexist in one granule [8], making single-stage denitrification possible.Wang et al. [9] found that AGS could remove 92% of the total nitrogen (TN) in organic wastewater with a ratio of carbon to nitrogen (C/N) of 25.Wei et al. [10] reported that AGS had a TN removal efficiency of 89.8% in organic wastewater (C/N of 9).Cydzik-kwiatkowska et al. [11] enhanced the nitrogen removal performance of AGS through intermittent acetate feeding and intermittent aeration strategies, and the TN removal efficiency of organic wastewater (C/N of 2.2) reached 77%.Zhang et al. [12] found that autotrophic nitrifying granular sludge (ANGS) could remove 60% of the total inorganic nitrogen (TIN) from inorganic wastewater, but the increase in influent ammonia nitrogen led to a significant decrease in TIN removal efficiency.Therefore, the AGS system can achieve high TN removal efficiency in the treatment of nitrogen-containing wastewater when there are enough carbon sources.However, its nitrogen removal ability deteriorates rapidly with a decreasing influent C/N ratio, and it is still a challenge for the AGS system to achieve efficient nitrogen removal with a low C/N ratio or inorganic wastewater.Some reports on AGS for Pb(II) adsorption have been reported, which indicate that AGS has a certain Pb(II) adsorption capacity.Wang et al. [13] found that the Pb (II) adsorption capacity of AGS could reach 101.97 mg/ gSS.Yao et al. [14] found that the Pb(II) adsorption capacity of AGS was in the range of 20-100 mg/gSS at different pH values and temperatures.Liu et al. [15] extracted extracellular polysaccharide substances (EPSs) from AGS and found that its maximum Pb(II) adsorption capacity was up to 1587.3 mg/gSS, and HOOC-and HO-groups were the main adsorption sites.Huang et al. [16] prepared a new type of aminofunctionalized magnetic AGS-biochar; its maximum Pb (II) adsorption capacity was 127 mg/gSS, and the Pb(II) removal efficiency was still as high as 88% after five adsorption/desorption cycles.
The adsorption of heavy metals by AGS can be completed through one or multiple mechanisms.Liu et al. [16] found that the EPS of AGS provided a large number of HO-and COO-groups to combine with Pb(II) in wastewater to form stable complexes.Xu et al. [17] found that AGS adsorbed Ni(II) through its exchange with K(I), Mg(II) and Ca(II).Sun et al. [18] found that AGS modified by grafting polyethylenimine changed the valence state of Cr(VI) through redox to reduce the toxicity.
As mentioned above, AGS has the potential for simultaneous nitrogen removal and heavy metal adsorption.However, autotrophic nitrifying bacteria and heterotrophic denitrifying bacteria reside in different regions of AGS [19] (the former usually grow on the outer layer, and the latter are often detected in the internal layer), and their tolerance to heavy metals is different.Juliastuti et al. [20] found that nitrifying bacteria were sensitive to environmental change and had poor resistance to Zn(II) and Cu(II).Tan et al. [21] found that Pb(II) at a concentration of 6 mg/L affected the nitrifying performance, whereas the removal of chemical oxygen demand (COD) was improved.Other similar reports also showed that the activity of nitrifying bacteria is more easily inhibited than that of other heterotrophic bacteria [22][23][24].However, few reports concerning nitrogen and heavy metal removal of AGS from inorganic wastewater have been reported.
The AGS system also has an overhaul or idle period, and the change in operating conditions undoubtedly has a negative impact on the stability of AGS [25].To make full use of AGS resources, the storage and recovery of AGS have attracted some scholars' attention.The storage of AGS can be divided into two categories, namely wet storage and dry storage.Most scholars stored AGS in a wet nontoxic environment.Duarte et al. [26] found that AGS stored at room temperature and wet conditions for 40 days had high damage resistance, while AGS stored after 180 days had an irregular shape and fluffy structure.Adav et al. [25] found that wet storage of AGS after 60 days would lose its structural stability and biological activity, mainly due to cell and protein hydrolysis.Zhang et al. [27] used an agarembedding method for dry storage of AGS for 30 days and achieved granular structure recovery within 11 days.He et al. [28] stored AGS in wet conditions for 58 days, and the removal efficiencies of ammonia nitrogen and TIN were both more than 95% after 18 days of recovery.To date, the storage of AGS under Pb(II) stress has rarely been reported, and there are few studies on the recovery of AGS for simultaneous nitrogen and Pb(II) removal from inorganic wastewater.
There are many ionic rare earth minerals in South Jiangxi, China, and a large amount of tail water containing ammonia nitrogen and even Pb(II) is produced in situ leaching [29].Wastewater from ionic rare earth mines is an acidic inorganic wastewater with high ammonia nitrogen, which can surpass 1000 mg/L (Table.S1 and S2 in Supplimentary material).Traditional biochemical technology requires a continuous dosage of an external carbon source for the efficient treatment of ionic rare earth mine wastewater, which results in high operating costs.To solve this problem, we established an AGS coupling denitrification system based on endogenous/ exogenous nitrification and denitrification, which could significantly reduce the aeration consumption, external carbon source dosage and alkalinity consumption in the treatment of simulated ionic rare earth mine wastewater [30].Therefore, AGS was stored in Pb(II) solution to select functional bacteria resistant to lead poison during the overhaul of the reactor.The AGS was then recovered to construct a two-stage oxic/anoxic AGS system for simultaneous removal of Pb(II) and nitrogen from inorganic wastewater to facilitate efficient treatment of inorganic wastewater.

Seed sludge
Dark brown AGS (termed S 0 ) was collected from a sequencing batch reactor (SBR, termed R 0 ) in the laboratory [30] (inner diameter 29.2 cm, height 200 cm, effective volume 120.5 L).Its properties were as follows: SV 30 /SV 5 of 0.98, SVI of 35 mL/g, average particle size of 1.72 mm, granulation rate of 97.3%, MLVSS/MLSS of 0.79, EPS content of 25.59 mg/gSS and PN/PS of 0.74.The cyclic time was 6 h, including: filling (5 min), aerobic reaction (145 min), anoxic reaction (150 min), aerobic reaction (50 min), settling (5 min) and drainage (5 min).The media was inorganic wastewater with high ammonia nitrogen.The influent ammonia nitrogen was 120 mg/L, and sodium acetate solution (the corresponding COD was 250 mg/L) was added at the beginning of the anoxic reaction to promote denitrification.The removal efficiencies of ammonia nitrogen and TIN were more than 90% and 80%, respectively.Nitrifying bacteria in the AGS were mainly Nitrosomonas (relative abundance of 14.6%), and denitrifying bacteria were mainly Thauera (relative abundance of 31.2%).

Storage of AGS
AGS was stored in lead nitrate solution at room temperature in four measuring cylinders with a volume of 500 mL (Fig. S1 in Supplementary material) in winter (the temperature was between 9 and 26 °C, and average temperature was 16 °C) to select functional bacteria resistant to lead poison.The operating procedure was as follows: one litre of sludge-water mixture was collected from R 0 and washed three times with deionized water; then, the supernatant was removed, and Pb(II) solution prepared by Pb(NO 3 ) 2 was added to fix the volume to 500 mL.The MLSS was approximately 17 g/ L. Four parallel samples were prepared (named as S 1 , S 2 , S 3 and S 4 ), and the Pb(II) concentrations were controlled at 0, 5, 10 and 20 mg/L, respectively.The samples were placed in a dark place to avoid algae growth, and the top of the cylinders was sealed by food wrap to prevent dust from falling into the liquid.After 40 days of storage, the water level decreased by less than 0.5 cm, the supernatant was collected for water quality analysis, and the granules were used for recovery.

Equipment
The recovery and long-term operation was carried out in two identical SBRs (named as R 1 and R 2 ).The effective volume of R 1 and R 2 is 27.7 L each (inner diameter of 14 mm, effective height of 180 mm), and the water exchange ratio is 60%.Granules in S 2 , S 3 and S 4 were inoculated into R 2 for recovery, and granules in S 1 were inoculated into R 1 for recovery.In addition, granules stored in situ in R 0 (the same storage conditions as S 1 ) for 40 days were inoculated into R 1 to make the sludge concentrations in R 1 and R 2 equal to each other.Aeration of R 1 and R 2 was provided by an air pump (ACO-012A).The aeration rate of each reactor was 42.5 L/min, and the liquid level rise was approximately 10 cm.The water temperature was maintained at 25-30 °C by using a heating rod (P-200W).The cyclic time was 6 h (4 cycles per day), including filling, reaction, settling and drainage (Table 1).Intermittent aeration was set in anoxic reaction to realize solid⍰ liquid mixing.The aeration rate was 21.3 L/min, and the liquid level rise was 3 cm.The intermittent aeration lasted for 10 min every 40 min from day 1-65 and 30 s every 15 min from day 66-140.The solids retention time of both R 1 and R 2 was maintained at approximately 50 days.

Media
The composition of the inorganic wastewater with high ammonia nitrogen is shown in Table S3 in Supplementary material.The main pollutants were ammonia nitrogen and Pb(II), while other elements such as calcium, magnesium and phosphorus, were also added.The alkalinity of the influent was 10-15 mmol/L (adjusted by sodium bicarbonate), the pH was approximately 8.5, and the concentrations of other pollutants are shown in Table 2. Sodium acetate solution was added to the anoxic reaction (1 g sodium acetate contributed 0.68 g COD) to enhance the denitrifying performance of AGS: sodium acetate solution with a volume of 250 mL was added at 150-155 min from day 1-40 (contributed 180 mg/L COD), and 420 mL was added at 150-155 min from day 41-65 (contributed 300 mg/L COD).
Particle size distribution was measured by wet sieving separation method, and the procedure was carried out as suggested by Zeng et al. [30].Granules with particle size larger than 0.3 mm were defined as AGS, and the granulation rate was the mass percentage of AGS relative to the entire sludge sample.It is assumed that there is a sieve that the mass proportions of filtered sludge and unfiltered sludge were both 50% (average particle size equals to the aperture of the sieve), which can be calculated from the mass distribution curve.
Granular morphology was qualitatively observed by a scanning electron microscope (SEM, FEI, MLA650F, USA), and the sampling procedure was carried out as suggested by Tay et al. [32].Heat extraction method was adopted to extract EPS from AGS, as suggested by Wang et al. [33].PS and PN contents were determined by phenol-sulphuric acid method [34] and modified Lowry method [35], respectively.EPS was the sum of PS and PN.Specific oxygen uptake rate (SOUR) produced by heterotrophic bacteria, ammonium oxidizers and nitrite oxidizers, namely, SOUR H , SOUR NH4 and SOUR NO2 , were determined by the methods developed by Ochoa et al. [36].SOUR N , namely the sum of (SOUR) NH4 and (SOUR) NO2 , was the overall activity of nitrifying bacteria.The Pb(II) content in the ash of AGS from R 2 was extracted by a digestion method as suggested by Pokhrel et al. [37]: the sludge was incinerated at 600°C for 2 h; the ash was dissolved in 0.5 mol/L HNO 3 and stirred for 12 h; the supernatant was collected for Pb(II) determination.The surface of AGS before and after adsorption was scanned by an Xray photoelectron spectrometer (Thermo Scientific K-Alpha+, ThermoFisher).

Microbial communities
Completely mixed AGS mixture (2 mL) was collected into a 2 mL centrifuge tube and centrifuged at 10000 rpm at room temperature for 3 min.The supernatant was discharged and replaced by distilled water, and AGS sample was obtained after the above procedure was repeated for three times.The deoxyribonucleic acid (DNA) was extracted by an E.Z.N.A.TM Soil DNA Kit (Omega, Bio-Tek, Norcross, GA, USA).Then, a Qubit 2.0 DNA detection kit was employed to exactly quantify the amount of DNA for polymerase chain reaction (PCR).The PCR primers were V3-V4 universe primers 341F (CCCTACAC-GACGCTCTTCCGATCTG (barcode) CCTACGGGNGGCWG-CAG) and 805R (GACTGGAGTTCCTTGGCACCCGAG AATTCCAGACTACHVGGGTATCTAATCC).The detailed first and second amplification process could be seen as described by Chen et al. [38].Finally, the extracted DNA was subjected to sequencing analysis by Miseq sequencing platform (Illumina, Inc., San Diego, CA, USA) of the V3_V4 region of 16S rDNA gene (Sangon Biotech Co., Ltd., Shanghai, China).

Determination of reaction rate
Batch experiments were carried out in conical flasks with effective volume of 300 mL.Sludge-water mixture with volume of 150 mL was collected from R 1 and R 2 , respectively.The sludge was settled and washed three times with deionized water.After removal of the supernatant, the washed sludge and nitrogen-containing solution were filled into the conical flasks and fixed to 300 mL.The reaction parameters were shown in Table 3.The alkalinity of the mixture was adjusted to 8 mmol/L by dosage

Change in sludge morphology
After 40 days of storage, the colour of the inner core of most granules changed from yellow to black (Fig. S2 in Supplementary material).According to our previous studies, the black core was related to the formation of sulphde [27].However, AGS still maintained a complete structure without obvious disintegration under the stress of different Pb(II) concentrations.SEM indicated that a large number of bacilli resided in AGS before and after storage.After 10 days of recovery, AGS with a black core disappeared, and its colour turned brownish yellow (Fig. S3 in Supplementary material).During this period, a large number of broken granules appeared in both R 1 and R 2 , and the phenomenon was more pronounced in R 2 .The results indicated that 40 days of storage significantly reduced the structural stability of the AGS, and Pb (II) stress further exacerbated the damage to the AGS.A significant reduction in fine granules was observed on day 30, and regranulation was completed on day 70 as the reactors were dominated by compact granules, which was consistent with the phenomenon observed by Zhang et al [39].Therefore, the stored AGS first experienced disintegration and then regranulation in R 1 and R 2 .
On day 140, the AGS in R 2 had a denser structure than that in R 1 .SEM indicated that a large number of bacilli and some cocci were inhibited inside the granules from R 1 and R 2 , while crystals appeared on the surface of AGS from R 2 , which was speculated to be chelates formed by Pb(II), polysaccharides, proteins and other macromolecules [21].1(c)), which was mainly ascribed to EPS served as a carbon source for microbes in a starving environment [40].With the increase in Pb(II) concentration, the EPS and PN/PS showed an overall increasing trend, indicating that the microbes secreted more PN in S 2 , S 3 and S 4 .It was found that a large number of HO-and NH 3 -groups in PN could complex with heavy metal ions [41].Therefore, microbes secreted more PN to reduce the toxic effect of Pb(II).
The SOUR of the stored AGS decreased significantly compared with that of the seed AGS (Figure 1(d)).The SOUR of the heterotrophic bacteria and autotrophic bacteria in the granules from S 1 all decreased significantly; the SOUR of the granules from S 2 , S 3 and S 4 also decreased, but it was mainly ascribed to the decrease in the SOUR AOB and SOUR NOB .With the increase in Pb (II) concentration, the SOUR AOB showed a decreasing trend (6.4-2.2 mgO 2 /gVSS•h), the SOUR NOB changed little (1.6-3.4 mgO 2 ) increased (3.9-14.1 mgO 2 /gVSS•h) in S 2 , S 3 and S 4 .The changes in the SOUR H , SOUR AOB and SOUR NOB resulted in an overall decreasing trend of the SOUR AOB /SOUR NOB and an overall increasing trend of the SOUR H /SOUR N during storage.In conclusion, Pb (II) stress had a significant inhibitory effect on AOB but promoted the activity of aerobic heterotrophic bacteria.

Settling performance
The SVI of R 1 fluctuated on days 1-50 (40.7-61.9mL/g, Figure 2(a)), showed an increasing trend on days 51-75 (47.7-68.2mL/g), and stabilized at approximately 68 mL/g after day 76.The SVI of R 2 decreased rapidly from days 1-4 (70.25-46.0mL/g), showed an overall increasing trend on days 5-117 (46.17-65.97mL/g), and stabilized at approximately 55 mL/g after day 117.On days 1-140, the SV 30 /SV 5 values of R 1 and R 2 were close to 1.The steady-state SVI of R 2 was smaller than that of R 1 .It is speculated that Pb(II) acts as the adsorption bridge between negatively charged cells, making AGS in R 2 have a denser structure.

MLSS and MLVSS/MLSS
The MLSSs of R 1 and R 2 both experienced two increasing processes in the first 91 days (Figure 2(b)), and tended to be stable from day 92.The sudden decrease in MLSS on days 40-45 was because of the artificial discharge of sludge to maintain a stable food-to-microorganism ratio (F/M).The MLVSS/MLSS of both R 1 and R 2 decreased first and then tended to be stable: the former stabilized at approximately 0.88 after day 32, and the latter stabilized at approximately 0.85 after day 89.

EPS and PN/PS
The EPSs of R 1 and R 2 both decreased first, then increased and eventually tended to be stable (Figure 2 (c)).The EPS of R 1 stabilized at approximately 40 mg/ gSS after day 97, and that of R 2 basically remained between 30 and 40 mg/gSS after day 69.The PN/PS of R 1 and R 2 both fluctuated in the first 114 days and basically stabilized at 1.8-2.0 and 1.3-1.6 after day 115.The steady-state PN/PS of R 2 was much lower than that of R 1 , presumably because the addition of Pb(II) inhibited the secretion of PN.The granulation rate of R 1 remained above 90% during the operation, and that of R 2 remained above 90% after day 10 (Figure 2(d)).The average particle size of R 1 first increased, then decreased, and eventually stabilized between 1.0 and 1.2 mm after day 106.The average particle size of R 2 first decreased, then increased, and eventually stabilized between 0.97 and 1.08 after day 67.
In R 1 , the proportion of granules less than 1.0 mm changed little and was usually less than 10% (Figure 2 (e)).Different trends of the proportion of granules with diameters larger than 1.0 mm were observed: the most significant change took place in the proportion of granules with particle sizes larger than 2.4 mm.The proportion of granules with particle sizes of 1.4-2 mm was relatively stable, which replaced the granules greater than 2.4 mm to become the largest proportion of granules.In R 2 , the proportion of granules with particle sizes less than 0.6 mm and larger than 2 mm changed little and was usually less than 10%.Different trends of the proportion of granules with diameters of 0.6-2 mm were observed, and the proportion of granules with particle sizes of 1.4-2 mm eventually became the largest.The proportion of granules larger than 2 mm in R 2 was significantly smaller than that in R 1 , indicating that Pb (II) had a significant negative effect on their stability.

SOUR
The SOUR of granules from R 1 and R 2 both increased rapidly in the first 8 days (Figure 2(f)).Then, the SOUR H of R 1 and R 2 first decreased, then increased and gradually tended to be stable.The SOUR H of R 1 remained at approximately 13 mg O 2 /gVSS•h after day 55, and that of R 2 was between 17 and 23 mg O 2 /gVSS•h after day 97.The SOUR AOB and SOUR NOB of R 1 and R 2 all changed little from day 8.The SOUR H /SOUR N and SOUR AOB /SOUR NOB of R 1 were stable after day 8, remained below 0.5 and stabilized between 0.9-1.4.The SOUR H /SOUR N of R 2 showed a slowly increasing trend from day 8 (0.1-0.7), while the SOUR AOB /SOUR NOB of R 2 showed a slowly decreasing trend (2-1).Compared with the SOUR N of R 1 , the SOUR N of R 2 was lower, indicating that Pb(II) inhibited the activity of nitrifying bacteria.After day 80, the SOUR H of R 2 was significantly higher than that of R 1 , indicating that Pb(II) promoted the activity of aerobic heterotrophic bacteria.

Nitrogen and Pb(II) removal
The ammonia nitrogen, nitrite nitrogen, nitrate nitrogen and TIN in the effluent from R 1 were not stable in the first 100 days (Figure 3(a)).Then, they changed little in the subsequent days (the ammonia nitrogen and TIN in the effluent were usually less than 15 and 20 mg/L).The ammonia nitrogen, nitrite nitrogen, nitrate nitrogen and TIN in the effluent from R 2 were not stable in the first 98 days.Then, they remained relatively constant in the subsequent days (the ammonia nitrogen and TIN in the effluent were usually less than 14 and 25 mg/L).High ammonia nitrogen removal efficiencies of R 1 and R 2 were maintained in most of the days (Figure 3(b)), and they were always greater than 90% starting on days 46 and 83, respectively.The TIN removal efficiencies of both R 1 and R 2 fluctuated first and then tended to be stable, basically remaining above 90% starting on days 99 and 91, respectively.The Pb(II) concentration in the effluent from R 2 was always below 0.3 mg/L (Figure 3 (c)), and the corresponding removal efficiency was usually more than 98%.

COD utilization
The COD in the effluent from R 1 fluctuated between 12 and 154 mg/L (Figure 3(d)) on days 1-46, and the corresponding COD utilization efficiency also fluctuated greatly (1%−93%).Then, the COD in the effluent basically stabilized below 60 mg/L, and the corresponding COD utilization efficiency remained above 80%.The effluent COD of R 2 had a similar trend to that of R 1 : it fluctuated greatly on days 1-46 and stabilized below 70 mg/L after day 46, and the corresponding COD utilization efficiency basically remained above 80%.

Pollutants degradation in typical cycles
The ammonia nitrogen decreased and then tended to be stable in typical cycles (Figure 4(a,b)).On day 50, nitrate nitrogen clearly accumulated in the aerobic phase of R 1 and R 2 , and it decreased in the anoxic phase; nitrite nitrogen clearly accumulated in the anoxic phase and then decreased in the aerobic phase.On day 140, nitrite nitrogen clearly accumulated in the first aerobic phase of R 1 and R 2 and then decreased to the minimum; no obvious accumulation of nitrate nitrogen was observed.The TIN in R 1 and R 2 showed a decreasing trend in typical cycles, but the decreasing trend was more clear on day 140.Compared with different typical cycles, both R 1 and R 2 maintained good nitrifying performance.However, higher effluent TIN caused by the accumulation of nitrate nitrogen was observed in R 1 and R 2 on day 50.On day 140, both R 1 and R 2 realized partial nitrification, and nitrite nitrogen became the main contributor to the effluent TIN.The ammonia nitrogen and TIN in the effluent of R 2 were lower than those of R 1 , indicating that the functional bacteria related to nitrification and denitrification in AGS had gradually adapted to Pb(II) toxicity.The COD in the aerobic phase was usually less than 50 mg/L; the COD in the anoxic phase increased rapidly after carbon dosing but then decreased rapidly to approximately 50 mg/L.During the typical cycles, most Pb(II) was removed in the first 30 min, and then it tended to be stable thereafter (Figure 4(c,d)).
In the first 99 days, the denitrifying performance in the two reactors was not stable.On the one hand, the granular breakage after reaeration destroyed the microenvironment of the AGS, resulting in restricted growth of nitrifying bacteria and denitrifying bacteria.On the other hand, the aeration and carbon source dosage method also had a significant influence on the simultaneous nitrification and denitrification performance of the AGS, and two-stage oxic/anoxic combined with two doses of external carbon sources was more conducive to the removal of TIN.With the completion of regranulation, the nitrifying bacteria and denitrifying bacteria successfully colonized the granules again, so the denitrifying performance of the system gradually stabilized.According to the German ATV standard (ATV-DVWKA131E), the removal of 1 kg of nitrogen requires 5 kg COD.However, the two-stage granular oxic/anoxic process removed 1 kg of influent TIN only consuming approximately 1.85 kg of carbon source benefiting from endogenous nitrification and denitrification of AGS [30].Therefore, the two-stage granular oxic/anoxic process can not only efficiently remove nitrogen and lead but also has clear advantages in energy savings, which provides another option for ionic rare earth mine wastewater treatment.

Reaction rate
Compared with the seed AGS from S 0 , the ammonia oxidizing rate, nitrite oxidizing rate and nitrate denitrifying rate of the granules from S 1 , S 2 , S 3 and S 4 decreased to different degrees, while the nitrite denitrifying rate of the granules from S 1 , S 2 , S 3 and S 4 changed little and seemed unaffected after 40 days of storage (Figure 5).After 130 days of recovery, the ammonia oxidizing rate, nitrite denitrifying rate and nitrate denitrifying rate of the granules from R 1 and R 2 all recovered to the prestorage levels, and some of them were much larger than those of the seed AGS from S 0 .The nitrite reduction rate was the highest among the 4 reaction rates, which was consistent with the accumulation of nitrite nitrogen in R 1 and R 2 after day 100.
It was observed that part of the ammonia oxidizing capacity of the AGS was lost during storage.Pb(II) stress at concentrations of 5-10 mg/L slowed the loss of ammonia oxidizing capacity, whereas 20 mg/L Pb(II) accelerated the loss of ammonia oxidizing capacity.Pb (II) stress had no obvious effect on nitrite oxidizing capacity during storage because the denitrification rates of AGS after storage were all above 80% of the mature AGS, which might be due to organics released by some dead cells providing substrates for the growth of the denitrifying bacteria.The nitrate reduction capacity of AGS in S 2 , S 3 and S 4 was better, indicating that Pb(II) could slow down the loss of denitrifying capacity.The ammonia-oxidizing rate of R 2 was lower than that of R 1 , and the reduction rates of nitrite and nitrate were both greater than those of R 1 , also confirming that Pb(II) inhibited nitrification but promoted denitrification of AGS.Our previous study showed that the free ammonia produced by high influent ammonia nitrogen effectively inhibited the activity of nitrite oxidizing bacteria [12].

XPS Analysis
XPS was used to analyse the changes in the surface properties of AGS before and after adsorption on day 140 (Figure 6 and the percentage of the -C-O-peak area increased significantly (28%−42%).This phenomenon could also be observed from the fine spectrum peak of O 1s (Figure 6 (c)), indicating that the adsorption is mainly related to ester and carboxyl groups.N peaks showed no obvious change (Figure 6(d)).In addition, K, Mg and Na on the surface of the granular sample from R 2 decreased (Table 4), indicating that the adsorption of Pb(II) involved ion exchange.The phosphorous content in R 2 clearly decreased, which suggested that a Pb 3 (PO 4 ) 2 precipitate was formed.Pb(II) was extracted from the granular sample by digestion, and 26% of lead was detected in the sludge ash on day 140.The Pb(II) content in the granular sample was 26 mgPb/gSS, which was higher than that determined by XPS (1.1 mgPb/gSS), further confirming that Pb was trapped inside the AGS.

Abundance and diversity of the bacterial community
Bacterial communities of the AGS from S 0 , S 1 , S 2 , S 3 , S 4 , R 1 and R 2 were analysed by high-throughput  5).The coverage of each sample was greater than 99.7%, indicating that the sequencing depth could fully reflect the bacterial community composition.The operational taxonomic units (OTUs) increased after storage and then decreased after 140 days of recovery.The Ace and Chao indices had similar trends to those of OTUs, and their values after recovery were significantly lower than those after storage, indicating that the bacterial abundance of AGS from R 1 and R 2 was lower.In addition, the ACE and Chao increased with increasing Pb(II) concentration, indicating that the addition of Pb(II) increased the bacterial abundance in storage.Considering diversity, the Shannon index increased significantly after storage, and the Simpson index decreased significantly after storage.Among them, S 1 decreased slightly, whereas S 2 , S 3 and S 4 decreased significantly, indicating that Pb(II) stress increased the bacterial diversity of the AGS.On day 140, the Shannon index of the AGS from R 1 was significantly higher than that of the AGS from R 2 , while the Simpson index was significantly lower than that of R 2 , indicating that the bacterial diversity of R 1 was higher.After storage, the uniformity of the microbial distribution increased (0.56-0.68), and it increased with increasing Pb(II) concentration, indicating that Pb(II) stress was conducive to the uniformity of the microbial distribution during storage.After 140 days of recovery, the Shannoneven index of R 1 was significantly higher than that of R 2 , indicating that Pb(II) reduced the uniformity of the bacterial community during recovery.

Composition of the microbial community
The mature AGS from S 0 mainly consisted of 3 dominant phyla: Proteobacteria, Bacteroidetes and Acidobacteria (Figure 7(a)).They were also the dominant phyla in granules from S 1 , S 2 , S 3 , S 4 , R 1 and R 2 .Proteobacteria is the largest phylum of bacteria.Most Proteobacteria bacteria live in facultative or obligate anaerobic and heterotrophic life, and some are chemoautotrophic bacteria and photosynthetic bacteria.Bacteroidetes is the most common phylum of gram-negative bacteria in the gut.Acidobacteria are important soil microbes that are acidophilic, oligotrophic and difficult to cultivate.Three new phyla appeared in granules from S 1 , S 2 , S 3 and S 4 during storage (Firmicutes, Verrucomicrobia and Ignavibacteriae), but their relative abundance was less than 1.4%.Firmicutes contains a large group of bacteria that are chemotrophic.Many species are capable of producing spores, which can resist dehydration and extreme conditions.Verrucomicrobia is a phylum of gram-negative bacteria that are found in freshwater and marine environments, soil and human feces and play an important role in biogeochemical cycles globally.Ignavibacteriae is a newly identified phylum for which little information has been reported.Compared with the stored granules, one new phylum, Planctomycetes, appeared in granules from R 1 and R 2 .Planctomycetes is a small phylum of aquatic bacteria found in seawater, brackish water and fresh water, and some species are capable of anaerobic ammonium oxidation.The mature AGS from S 0 mainly included 8 classes (Figure 7(b)), which were also the same dominant classes in granules from S 1 , S 2 , S 3 , S 4 , R 1 and R 2 .In addition, 3 new classes appeared in granules from S 1 , S 2 , S 3 and S 4 , but their abundance was less than 1.7%; 5 new classes appeared in R 1 and R 2 , and Flavobacteriia and unclassified_Planctomycetes were only detected in R 1 and R 2 .
After 140 days of recovery, there were mainly 24 genera in both R 1 and R 2 .Compared with granules from S 1 , S 2 , S 3 and S 4 , 8 genera disappeared, and 8 new genera appeared.The 24 genera of each sample mainly included 3 functional groups: denitrifying bacteria (such as Thauera [8], unclassified_Burkholderiales [42], unclassified_Comamonadaceae [43], unclassified_Rhodospirillales [49] , unclassified_Xanthomonadaceae [49], Flavobacterium [54] , unclassified_Bacteria [64] and so on), nitrifying bacteria (such as Nitrosomonas [8] and Paracoccus [47] ) , and heterotrophic bacteria without denitrifying and phosphorous removal ability (such as unclassified_Bacteroidetes [59] and Acetoanaerobium [60] ) .The relative abundance of unclassified_Cytophagales changed little in S 0 , S 1 , S 2 , S 3 and S 4 , but it increased significantly in R 1 and R 2 (especially for R 1 ).It is speculated that this genus belongs to fermentative bacteria since some species in the genus are capable of digesting complex compounds [65].In addition, the average particle size of AGS from R 1 was overall larger than that from R 2 , which provides a larger anaerobic space for the growth of these anaerobic bacteria.A genus of unclassified_Planctomycetes was only detected in R 1 and R 2 , and many of its close genera belong to anaerobic ammonium oxidation bacteria [66].Therefore, the genus is speculated to have anaerobic ammonium oxidation ability since R 1 and R 2 can provide a potential environment for the growth of anaerobic ammonium oxidation bacteria.
The abundances of autotrophic nitrifying bacteria, denitrifying bacteria and heterotrophic bacteria without denitrifying and phosphorous removal abilities in AGS from S 0 were 14.64%, 62.72% and 2.86%, respectively.After 40 days of storage, the abundance of autotrophic nitrifying bacteria decreased significantly, but the abundance of denitrifying bacteria, phosphorous removal bacteria and other heterotrophic bacteria all increased, indicating that Pb(II) did not inhibit their growth.The decrease in the abundance of autotrophic nitrifying bacteria is not surprising since there is no aeration or external nutrients during storage.After 140 days of recovery, the relative abundance of autotrophic nitrifying bacteria in R 1 and denitrifying bacteria in R 2 basically recovered to the prestorage level, whereas the abundance of denitrifying bacteria in R 1 and autotrophic nitrifying bacteria in R 2 was significantly lower than the prestorage levels, indicating that Pb(II) had a significant inhibition effect on nitrifying bacteria but had a certain promotion effect on bacteria (especially for Thauera).Under stress, Thauera in R 2 secreted EPS to resist toxicity, so it is more competitive than other heterotrophic bacteria.The relative abundance of other heterotrophic bacteria in R 1 was higher than that in R 2 , indicating that Pb(II) also inhibited these heterotrophic bacteria.The stability of AGS decreased after 40 days of storage, which was determined by the release of pollutants to the liquor and the decrease in EPS, SOUR, reaction rate and relative abundance of the functional bacteria (especially for autotrophic nitrifying bacteria).This phenomenon is also commonly observed in similar studies [25][26][27][28].The reason was mainly ascribed to the evolution of the original microbial community caused by the difference in operating conditions between S 0 and S 1 -S 4 .Pb(II) had little effect on the shape and settling performance of AGS during storage, but it was beneficial to AGS to secrete more EPS (especially for PN) and maintain its SOUR (especially for SOUR H ). It can be observed from the morphology of the sludge that most of the granules have black cores after storage.This phenomenon indicated that facultative or anaerobic microbes overproliferated inside the granules, which was consistent with the observed increase in SOUR H and denitrifying bacterial (such as Thauera) abundance after storage (the relative abundance of heterotrophic bacteria was approximately 69.32%, 81.52%, 77.49%, 78.38% and 76.45% in S 0 , S 1 , S 2 , S 3 and S 4 ).In the first 30 days after reaeration, a large amount of granular breakage occurred in both R 1 and R 2 , and the phenomenon was more obvious in R 2 , indicating that Pb(II) had a more detrimental impact on the structural stability of the AGS during storage.
It is generally believed that a toxic environment is not conducive to the growth of functional bacteria [5].During storage, the relative abundance of autotrophic nitrifying bacteria in AGS decreased more obviously under Pb(II) stress, but the SOUR AOB of granules under Pb(II) stress was higher than that of granules without Pb(II) stress.Similarly, the relative abundance of denitrifying bacteria from S 2 , S 3 and S 4 was lower than that from S 1 (the relative abundance was approximately 74.21%, 68.89%, 69.06% and 68.52% in S 1 , S 2 , S 3 and S 4 ), but the SOUR H of granules from S 1 , S 2 , S 3 and S 4 increased with the Pb(II) concentration.In addition, the ammonia oxidizing rate, nitrite oxidizing rate and nitrate reduction rate under different Pb(II) stresses changed little after storage.The results showed that nitrifying bacteria and denitrifying bacteria under Pb(II) stress had stronger substrate degradation abilities.
Interestingly, AGS recovered in R 2 had a smaller particle size and lower SVI but a more compact structure compared with those in R 1 .The granular breakage was more obvious in R 2 , which provided more and smaller nuclei for the secondary granulation.In addition, Pb(II), like Ca(II) and Mg(II), also helps negatively charged cells bind together.Therefore, the inert core formed by Pb(II) results in a denser AGS structure in R 2 .Under Pb (II) stress, the relative abundance of nitrifying bacteria in R 2 was significantly lower than that in R 1 , but the relative abundance of denitrifying bacteria in R 2 was significantly higher than that in R 1 (the relative abundance of denitrifying bacteria was approximately 46.79% and 62.44% in R 1 and R 2 ).The ammonia oxidizing rate and nitrate reduction rate in the two reactors had a similar trend to the relative abundance of nitrifying bacteria and denitrifying bacteria.The results showed that Pb(II) inhibited the growth of AOB, which usually inhabited the surface of AGS [67], but promoted the proliferation of denitrifying bacteria inside the granules [19].As there was little difference in steady-state nitrogen removal efficiency between the two reactors, the AGS in R 2 maintained stability under Pb(II) stress.

Conclusions
AGS can be preserved in Pb(II) solution, and Pb(II) stress is beneficial to select adaptable functional bacteria but detrimental to the structural stability of AGS during storage.Adaptable genera such as Thauera and Nitrosomonas were selected under Pb(II) stress in storage, and they played an important role in the secondary granulation of AGS and the simultaneous removal of Pb(II) and TIN from inorganic wastewater, which provided a useful method for the establishment of simultaneous nitrification and denitrification AGS systems under Pb (II) stress.

Figure 1 .
Figure 1.Release of pollutants to the liquid and changes in properties of the granules during storage: (a) pollutant concentration in the supernatant from S 1 , S 2 , S 3 and S 4 ; (b) SVI and SV 30 /SV 5 of granules from S 1 , S 2 , S 3 and S 4 ; (c) EPS and PN/PS of granules from S 1 , S 2 , S 3 and S 4 ; (d) SOUR of granules from S 1 , S 2 , S 3 and S 4

Figure 2 .
Figure 2. Evolution of granular properties during recovery and long-term operation: (a) SVI and SV 30 /SV 5 ; (b) MLSS and MLVSS/MLSS; (c) EPS and PN/PS; (d) average particle size and granulation rate; (e) particle size distribution; (f) SOUR (a)).The characteristic peaks of Pb4f 5/2 (143.0 eV) and Pb4f 7/2 (138.4 eV) were detected in the granular sample from R 2 , which proved Pb(II) adsorption on AGS.The change in peak shape and peak area in the C spectrum confirmed the presence of Pb(II) chemisorption on the AGS surface.After peak separating, fitting and content calculation of the C 1s fine spectrum (Figure 6 (b)), there are three different forms of C. At 284.8, 286.5 and 288.0 eV, the binding energies represent -C-C-, -C-O-and O = C-, respectively.In R 2 , the percentage of the O = C-peak area was significantly lower (25%−21%),

Figure 3 .
Figure 3. Reactor performance during recovery and long-term operation: (a) TIN, ammonia nitrogen, nitrite nitrogen and nitrate nitrogen in effluent; (b) removal efficiencies of TIN and ammonia nitrogen; (c) Pb(II) in effluent and its removal efficiency; (d) COD utilization efficiency

Figure
Figure Trends of pollutant concentrations in typical cycles: (a) changes in TIN, NH 4 + -N, NO 2 --N and NO 3 --N on day 50; (b) changes in TIN, NH 4 + -N, NO 2 --N and NO 3 --N on day 140; (c) changes in COD on day 50; (d) changes in COD on day 140; (e) change in Pb(II) on day 50; (f) change in Pb(II) on day 140

Figure 5 .
Figure 5. Nitrification and denitrification rate of granules from S 0 , S 1 , S 2 , S 3 and S 4

Figure 6 .
Figure 6.XPS spectra of AGS from R 1 and R 2 on day 140: (a) All Peaks; (b) C Peak; (c) O Peak; (d) N Peak

Table 1 .
Composition of cyclic time.

Table 2 .
Changes in the influent quality during recovery.NaHCO 3 .Sodium acetate solution (the corresponding COD was 200 mg/L) was added at the beginning of the denitrifying reaction to provide carbon sources for denitrifying bacteria.The reaction lasted for 1 h.The water temperature was maintained at approximately 28 °C by water bath heating, and the aeration rate was 2.8 L/min.
Time (day) Concentration (mg/L) COD dosage (mg/L) Organic loading rate (kg/m 3 •d) Nitrogen loading rate (kg/m 3 •d) of remained below 2 mg/L.The TP in the supernatant of S 1 , S 2 and S 3 also changed little and was approximately 50 mg/L.However, the TP in the supernatant of S 4 decreased to 35.1 mg/L, presumably due to phosphate deposition with Pb(II).The COD in the supernatant of S 1 and S 2 was approximately 160 mg/L, while the COD of S 3 and S 4 increased to approximately 200 mg/L.It was speculated that the toxic effect of a high concentration of Pb(II) on cells was greater, leading to more obvious cytolysis.
might be caused by the adsorption of AGS or the formation of insoluble lead precipitation.In the meantime, nitrogen, phosphorous and organics were released from AGS into the supernatant.With the increase in Pb(II) concentration, the ammonia nitrogen in the supernatant increased(26.4-35.1 mg/L), while the nitrite nitrogen and nitrate nitrogen changed little and

Table 3 .
Parameters for determining reaction rate in batch experiments N are provided by NH 4 Cl, NaNO 2 and NaNO 3 .

Table 4 .
Elemental composition of the AGS determined by XPS.

Table 5 .
Characteristic indexes of the sequencing results.

Table 6 .
Changes in microbial community composition during the operation.