The positive effects of arbuscular mycorrhizal fungi inoculation and/or additional aeration on the purification efficiency of combined heavy metals in vertical flow constructed wetlands

Inoculation with arbuscular mycorrhizal fungi (AMF) and additional aeration (AA), as two approaches to improve the functioning of treatment wetlands, can further promote the capacity of wetlands to purify pollutants. The extent to which, and mechanisms by which, AMF and AA purify wetlands polluted by combined heavy metals (HMs) are not well understood. In this study, the effects and mechanisms of AMF and/or AA on combined HMs removal in vertical flow constructed wetlands (VFCWs) with the Phragmites australis (reeds) were investigated at different HMs concentrations. The results showed that (1) AA improved the AMF colonization in VFCWs and AMF accumulated the combined HMs in their structures; (2) AMF inoculation and/or AA significantly promoted the reeds growth and antioxidant enzymes activities, thereby alleviating oxidative stress; (3) AMF inoculation and AA significantly enhanced the removal rates of Pb, Zn, Cu, and Cd under the stress of high combined HMs concentrations comparing to the control check (CK) treatment (autoclaved AMF inoculation and no aeration), which increased by 22.72%, 30.31%, 12.64%, and 50.22%, respectively; (4) AMF inoculation and/or AA significantly promoted the combined HMs accumulation in plant roots and substrates and altered the distribution of HMs at the subcellular level. We therefore conclude that AMF inoculation and/or AA in VFCWs improves the purification of combined HM-polluted water, and the VFCWs-reeds-AMF/AA associations exhibit great potential for application in remediation of combined HM-polluted wastewater.


Introduction
Nowadays, heavy metal (HMs) pollution mainly caused by human activities such as mining and highways, due to its high toxicity, non-biodegradability, and persistence nature, has posed an increasingly serious threat to human health and the stability of aquatic ecosystems even at trace levels (Jia et al. 2020;Yin et al. 2019;Yu et al. 2020). Compared with single HM contamination, combined HM contamination in wastewater exhibits higher toxicity and may lead to more serious environmental challenges (Shi et al. 2020;Wang et al. 2018). Therefore, there is an urgent need to develop effective techniques to relieve combined HM negative influences. In general, the remediation of combined HMs by conventional technologies not only needs high energy consumption, but also generates a large amount of secondary wastes (Jacob et al. 2018). Conversely, engineered systems mimicking natural systems, such as constructed wetlands (CWs), rely mainly on wetland plants, microbes, and substrates for wastewater purification, providing a highefficiency, economical, and eco-friendly mean for treating combined HM wastewater (Ayaz et al. 2020;Papaevangelou et al. 2017;Zhang et al. 2020). What's more, the wetland plants play important roles in the process of wastewater purification utilizing CWs. The roots of wetland plants can absorb HMs from wastewater and transfer them upward to Responsible Editor: Alexandros Stefanakis the aboveground parts, providing the necessary conditions for rhizosphere microbial activities to stabilize pollutants (Hu et al. 2021;Kochi et al. 2020). However, the resistance of wetland plants to HMs toxicity is limited. Xin et al. (2020) reported that the high concentration of Cd 2+ inhibited photosynthesis by disturbing the biosynthesis of chlorophyll precursors in the leaves, thereby leading to a decline in biomass. Therefore, improving the resistance of wetland plants to HMs stress can help to promote the removal efficiency of wetlands.
Arbuscular mycorrhizal fungi (AMF), a key group of soil microorganisms, can be symbiotic with more than 80% vascular plants in terrestrial (Trappe 1987). The significant roles of AMF in enhancing plants nutrition uptake, increasing plants resistance, and stimulating rhizospheric microbial activity under biotic and abiotic stresses, have been thoroughly reported (Camenzind et al. 2016;Huang et al. 2018;Riaz et al. 2021). Over the last two decades, it has been gradually reported that AMF could form mycorrhiza with wetland plants, which made it possible for them to perform their functions in aquatic habitats (Hu et al. 2020b;Xu et al. 2016). However, unlike the terrestrial environment, wetlands had a low AMF colonization due to inadequate dissolved oxygen content (Calheiros et al. 2019;Xu et al. 2016). With the in-depth research on mycorrhizal fungi by numerous researchers, inoculation with AMF, as an effective strategy of bioaugmentation, could become a promising tool to improve the functions of treatment wetlands (Tondera et al. 2021). Surprisingly, a number of studies have proved that plants assisted by AMF have better performance than non-inoculated plants in CWs, for example in the purification of HMs wastewater (Hu et al. 2020c;Li et al. 2016). However, the role of AMF on the treatment of combined HM pollution in aquatic environment remained poorly understood. At the same time, aeration is usually used to increase the amount of dissolved oxygen in water bodies, which is also well known as a common physical method to improve the ability of wetlands to remove pollutants . Xin et al. (2019) reported that pre-aeration of the rhizosphere promoted the uptake and bioaccumulation of Cu and Cd in seedlings. Nevertheless, there is no information so far about the extent to which AMF or combined with aeration plays an ecological role in removing combined HMs in CWs and whether the combination improves the treatment efficiency. As a result, the applications of AMF inoculation and additional aeration (AA) in the treatment of combined HMs in CWs require further study.
In this study, the effects and mechanisms of AMF and/or AA on combined HMs removal in vertical flow constructed wetlands (VFCWs) were investigated at different HM concentrations. Phragmites australis (reeds), an emergent plant, is commonly used for phytoremediation of wastewater in CWs (Rezania et al. 2019). The main aims of this study were to (1) assess the effects of AMF inoculation and/or AA on the growth and physiological indexes of Phragmites australis (reeds) under the stress of combined HMs; (2) evaluate the removal efficiencies and distribution of combined HMs in CWs with AMF inoculation and/or AA; (3) explore the role of AMF inoculation and/or AA on the removal of combined HMs in CWs. The results of this study can provide an understanding of the combination with AMF inoculation and/or AA in CWs for combined HM removal.

System construction
The experiment was carried out on an open-air balcony with a transparent shelter at Wuhan University of Technology in Wuhan, China (30ding of r > 2019 < /Year > < R. Twelve parallel miniature vertical water columns (VFCWs) with a height of 65 cm and a diameter of 20 cm were made of plexiglass (Fig. S1A). The gravel layers (with a diameter of 2-3 cm), zeolite (diameter of 1-2 cm) and quartz sand (diameter of 1-2 mm) were sequentially packed into the VFCWs, with heights of 15, 30, and 15 cm, respectively. In addition, the required aeration in VFCWs was achieved by programmable timing aeration discs.

AMF inoculum, wetland plants, and synthetic wastewater
The AMF Funneliformis mosseae (BGC XJ01A) were purchased from the Institute of Plant Nutrition and Resource, Beijing Academy of Agriculture and Forestry Sciences, China. The F. mosseae inoculum (ca. 70 spores/g), consisting of roots, spores, and hyphae, was obtained by trap culture with Zea mays L. The F. mosseae inoculum was obtained according to the method described in Xu et al. (2021).
A common wetland plant species, Phragmites australis (reeds), were chosen as the host plant in this study. The seeds, purchased from Fuliya Seed Industry Co., Ltd, China, were sterilized in 70% ethanol solution for 1 min, then buried in a tray filled with sterilized fine sand, and kept moist with sterilized water every day. Two weeks later, reed seedlings were transplanted into a sterile nursery cup with a diameter of 9 cm and a depth of 13 cm. The growth substrate was composed of river sand, vermiculite, and humus (5:1:1 in mass). Half of the seedlings were inoculated with F. mosseae inoculum, and the other half were inoculated with autoclaved F. mosseae inoculum (Hu et al. 2020c). Tap water was added to keep moist before the seedlings were transplanted to the VFCWs. Three seedlings were transplanted into each VFCW system on October 26, 2019, and acclimated to the VFCWs under flooded conditions (2-cm depth of water above the surface of quartz sand) for 10 days.
The influents were synthetically prepared using Pb(NO 3 ) 2 , Zn(NO 3 ) 2 , Cu(NO 3 ) 2 , Cd(NO 3 ) 2 , and ultra-pure water in this study. Additionally, each VFCW was irrigated with 50 ml modified Hoagland solution every week, in which KH 2 PO 4 concentration was adjusted to 0.01 mM in order to prevent precipitation of Pb (Islam et al. 2008). Considering the limited tolerance of reeds to combined HMs, 2, 5, 2.5, and 0.5 mg/L of Pb, Zn, Cu, and Cd were artificially prepared as the low concentrations of combined HMs, and 20, 50, 25, and 2.5 mg/L were set as the high concentrations of combined HMs, according to Cheng et al. (2002) and Fritioff and Greger (2006) with minor modifications.
Once the plants were transplanted into the VFCWs on October 26, 2019, aeration was carried out in the wetland for 4 h (1 L/min). Tap water was added to a height of 2 cm above the quartz sand layer, and the outlet valve was kept closed for the next 10 days. Then, the VFCWs began to be operated without combined HMs loading. Tap water was fed in at 10:00 am, and discharged at 10:00 am in the next morning. The VFCWs were cyclically operated for 2 weeks to promote the coating of microorganisms on the substrate or plant roots surface, with a hydraulic retention time (HRT) of 24 h. During the operation of the VFCWs with combined HM loading, the synthetic combined HM-polluted water was fed in and discharged with an HRT of 24 h. The influent of the next cycle was the effluent of the previous cycle, and the ultra-pure water was replenished to the original height (62 cm in height) (Fig. S1B). The purification experiment of simulated combined HM-polluted water was operated for 15 cycles, e.g., 30 days. All water samples were collected and stored at 4 °C.

Sample handling and measurement
Plant shoots and roots were harvested separately at the end of the experiment (60 days after transplanting), then washed with deionized water and dried in an oven. The root length, shoot height, and fresh weight of the roots and shoots were measured directly with the tape measure and electronic balance. After being rinsed with sterile water, the roots of different reeds from the same wetland column were preserved in formalin-acetic acid-alcohol solution and then cut into 1-cm fragments. A total of 30 root fragments of 1 cm/sample were observed through the steps of transparence -cleaning-softening-cleaning-acidification -trypan blue staining-decolorization-preparation-observation (Phillips and Hayman 1970). Mycorrhizal frequency (F%), mycorrhizal colonization intensity (M%), and arbuscular mycorrhizal abundance (A%) in roots were calculated by the method of Trouvelot et al. (1986). The superoxide dismutase (SOD) and peroxidase (POD) activities were determined using the nitro blue tetrazolium (NBT) method (Giannopolitis and Ries 1977) and the method by Chance and Maehly (1955), respectively. The reactive oxygen species (ROS) content and catalase (CAT) activity were determined by spectrophotometry, the root activity determined by the triphenyl tetrazole chloride (TTC) method, and the malondialdehyde (MDA) content was determined by the thiobarbituric acid colorimetric method (Gao 2006). Besides, chlorophyll content was determined by portable chlorophyll meter (SPAD-502 Plus, KONICA MINOLTA, Japan). Dry weight of the shoots and roots was measured using an analytical balance after drying at 80 ℃ to constant weight.
The bioconcentration factor (BCF) and translocation factor (TF) were calculated using the following equation to evaluate the effect of AMF on the Pb, Zn, Cu, and Cd translocationabsorption capacity in wetland plants (A et al. 2017).

Detection of Pb, Zn, Cu, and Cd
The substrates were digested with the mixture of concentrated HF-HNO 3 -HClO 4 mixed acid in a 5:2:1 volume ratio (Gao and Chen 2012). The harvested reeds were rinsed repeatedly with deionized water, and then exchanged with 20 mmol/L Na 2 -EDTA for 20 min to remove the HM ions adsorbed on the tissue surface, and finally digested on the electric heating plate with concentrated HNO 3 -HClO 4 mixed acid by volume 4:1 (Favas 2019

Statistical analysis
A three-way analysis of variance (ANOVA) was carried out to test the effects of main factors (inoculation of F. mosseae, aeration, and combined HMs) and their interactions on the growth, antioxidant enzyme activities, MDA, and ROS of reeds. The Ducan's testing was used to compare treatment differences with P < 0.05 set as a significant difference. All statistical analyses were performed using the IBM SPSS statistics version 26.0 software package for windows (IBM Corp., Armonk, NY, USA). The "Pheatmap" and "Factoextra" packages were used by R Software (version 4.1.0) to visualize the experimental data among the plant growth and physiological indexes in plant roots in all treatments. Origin 2021 (OriginLab, USA) was used to plot the figures.
When the experiment ended at the 60th day, the F% values of the corresponding treatment increased by 40.00%, 25.14%, 20.00%, 18.76%, 13.40%, and 7.07%, respectively, indicating that AMF could continue to infect the root system even under flooded conditions. In particular, the F% values of the "IA + FM" treatment at three concentrations (0, low, and high) were significantly higher than those of the "NA + FM" treatment (P < 0.05), which increased by 10.58%, 17.64%, and 20.00%, respectively. Meanwhile, the intensity of mycorrhizal colonization (M%) and arbuscular abundance (A%) also showed a similar trend with the F%. These results suggested that AA had a positive effect on improving AMF colonization in VFCWs, especially under the stress of high combined HMs concentrations, which was consistent with the results of Xu et al. (2021). Lower oxygen content in wetland habitats may be one of the most important factors leading to AMF not to colonize roots at a high level like in the terrestrial ecosystems Wang et al. 2015). Similarly, Hu et al. (2020b) reported that AMF colonization under fluctuating water depth was higher than high water regimes, which was ascribed to the increasing oxygen content of the rhizosphere by constantly changing water regime. In addition to increasing dissolved oxygen in wetlands, the addition of aeration also promoted the growth of wetland plants, thus providing more surface areas for microorganisms growth, and also altered the diversity of microbial communities distributed in the rhizosphere . Ferreira et al. (2021) and Viollet et al. (2017) revealed that the diverse microbial communities (e.g., plant growth-promoting rhizobacteria, nitrogen-fixing bacteria) had a positive role in promoting AMF colonization by helping AMF obtain more nutrients and facilitating hyphal growth to produce spores, and we reasoned that it also happened in aquatic environment. As the concentration of combined HMs increased, the F%, M%, and A% of the two inoculation treatments all showed a decreasing trend. A similar study by Hu et al. (2021) showed that the F% were 49.70% at 0 mg/kg Cr concentration, but decreased to 40.90% and 23.90% at 5 and 25 mg/kg Cr concentration, respectively. This may be attributed to the negative influence of HMs on the germination and growth of hyphal (Göhre and Paszkowski 2006). What's more, HMs had a great negative impact on the establishment of symbiotic relationship between AMF and plant roots, thereby decreasing root infectivity and colonization (Wu et al. 2019). Therefore, the AA played a promising role in promoting AMF colonization in wetland plant roots.

Growth indexes, chlorophyll content, and root vitality
Under the three concentrations of combined HMs, compared with the control check (CK) treatment ("NA + NM"), the shoot height, root length, aboveground fresh weight, underground fresh weight, aboveground dry weight, underground dry weight, and chlorophyll content of AMF inoculation and/or AA treatments increased by 9.40-56.17%, 0.14-76.47%, 12.22-169.32%, 2.07-126.62%, 11.84-58.33%, 28.62-238.77%, and 36.52-519.67%, respectively (Table S1). Among them, those growth indexes and chlorophyll content of the "NA + FM" and "IA + FM" treatments were significantly higher than those in the CK treatment (P < 0.05), and the growth promotion of the "IA + FM" treatment was the best. The root vitality of the AMF inoculation and/or AA treatments was significantly higher than the non-inoculated or non-aerated treatments (P < 0.05). Compared with the reeds grown under 0 mg/L combined HMs, the root vitality of the reeds grown at low concentrations increased slightly, while that dropped significantly at high concentrations (P < 0.05). Principal component analysis (PCA) results showed that these parameters were largely determined by Dim1 (67.90%), implying there was a strong correlation among them. Meanwhile, the results of scores of 12 treatments showed that AMF inoculation and AA had different effects on wetland plant physiological functions under the three concentrations of combined HMs (Fig. S2).
The results of hierarchical clustering analysis indicated that AMF inoculation and AA showed obvious synergistic effects on the growth characteristics of wetland plants (Fig. 1). For aquatic plants, HM stress could inhibit the root vitality, decrease the chlorophyll content and plant biomass by reducing the uptake of essential macro and micronutrients (Wu et al. 2019;Zhao et al. 2019). The plant biomass and chlorophyll content of wetland plant (Iris wilsonii) inoculated with AMF under the stress of Cr-contaminated water were higher than those of non-inoculated control, which was consistent with our results (Hu et al. 2020c). Schück and Greger (2020) conducted a 5-day HM exposure experiment on 34 types of wetland plants and measured 20 traits of each plant, and finally found that high biomass was one of the most important characteristics of the plant's ability to remove HMs. Therefore, wetland plants inoculated with AMF could strengthen their ability to remove HMs by increasing plant biomass. Meanwhile, AMF-inoculated plants can also alleviate the damage of combined HMs to chloroplast in leaves by restricting the transport to the aboveground parts (Janeeshma and Puthur 2020).
Similar with our result, Ban et al. (2021) revealed that the root vitality of reeds increased significantly under the low concentrations of CuO-NP-polluted water (50 mg/L), but decreased significantly at high concentrations (500 mg/L). A likely explanation was that plants themselves had the potential to improve their physiological metabolism level by increasing the root vitality to resist combined HM stress at low concentrations. The widespread underground network between the mycelium and plant roots was beneficial for facilitating nutrient flow between roots, and preventing combined HMs from entering root cells (Alam et al. 2019). However, excessive HMs could destroy the root cell structure, significantly reduce root vitality, and cause obvious physiological and metabolic disorders (Liu et al. 2014).
Meanwhile, the AA not only increased the dissolved oxygen in the rhizosphere microenvironment and improved the vent ability in reed tissues, but also promoted the absorption of nutrients by roots, thereby promoting the growth of the reeds. There was a very marked synergistic effect of AMF inoculation and AA on growth indexes, chlorophyll, and root vitality of wetland plants for treating combined HM-polluted water. Three-way ANOVA analysis also showed that AMF, aeration, combined HM pollution, and the interaction of AMF with aeration had significant effects on the growth indexes, chlorophyll content, and root vitality (P < 0.05) ( Table 2).

ROS levels, lipid peroxidation, and antioxidant response
It can be seen from Table S2 that compared with the CK treatment ("NA + NM"), AMF inoculation and/or AA reduced the contents of MDA and ROS, and the "IA + FM" treatment showed the lowest values at different combined HM concentrations. Conversely, the SOD, CAT, and POD activities of AMF inoculation and/or AA treatments were significantly higher than those of the CK treatment at three concentrations (P < 0.05), with an increase of 7.32-176.78%, 3.70-74.88%, and 3.74-40.89%, respectively. Among them, the SOD, CAT, and POD activities of reeds in the "IA + FM" treatment were the highest. PCA results showed the plant growth indexes had a strong negative correlation with ROS and MDA, which was consistent with Hu et al. (2020a) (Fig. S2). Those results of hierarchical clustering analysis indicated that inoculation with AMF and/or AA had a Fig. 1 The cluster analysis among plant growth (shoot height, root length, aboveground fresh weight, underground fresh weight, aboveground dry weight, underground dry weight, chlorophyll, and root vitality) and physiological indexes (MDA, SOD, POD, CAT, and ROS) in plant roots and colors in the heatmap indicate the correlation between the different data sets. IA, intermittent aeration; NA:, no aeration; FM, inoculation with F. mosseae; NM, inoculation with autoclaved F. mosseae Table 2 Three-way analysis of variance of AMF inoculation, aeration, and combined HM stress on reeds * P < 0.05, **P < 0.01, *** P < 0.001. NS significant difference. AMF represents inoculation of F. mosseae, C represents combined HM stress  (Fig. 1). HM pollution was one of the abiotic factors leading to oxidative stress in plants, which could induce excessive ROS and lipid peroxidation, thus resulting in cellular damage and disturbing cellular ionic homeostasis (Dubey et al. 2018). SOD, CAT, and POD were the main antioxidant enzymes to remove ROS in plants, which could protect the structure and function of the cell membrane system and maintain the redox state of plant cells. Nafady and Elgharably (2018) found that AMF inoculation significantly reduced MDA content when investigating the effect of mycorrhizal symbiosis on Zea mays growth under combined HMs (Fe, Mn, Zn, Cu, Cd, and Pb) stress, which was consistent with our results. Similarly, whether in aquatic habitat or terrestrial habitat, mycorrhizal plants had better ROS scavenging ability and antioxidant ability than non-inoculated plants under HM stress (Hu et al. 2020c;Zhan et al. 2018). In addition, AA had the potential to alleviate oxidative damage by enhancing plant antioxidant capacity and leaf photosynthesis, and upregulate the expression of the genes maintaining cell redox balance (Xiaochuang et al. 2020). Therefore, AMF inoculation and AA synergistically improved the physiological indexes of reeds and increased antioxidant enzyme activities (SOD, CAT, and POD) under the stress of combined HMs. Three-way ANOVA analysis also showed that the interaction of AMF with aeration had significant effects on antioxidant enzyme contents (P < 0.05) ( Table 2).

Distribution of Pb, Zn, Cu, and Cd in wetland plants
As shown in Fig. 2, the concentrations of Pb, Zn, Cu, and Cd in reed roots were higher than those in the shoots at low and high concentrations, indicating that HMs were mainly accumulated in the roots of wetland plants. Meanwhile, the concentrations of Pb, Zn, Cu, and Cd in the reeds were all increased greatly with the increase of the concentration. Under the two concentrations of combined HMs, the concentrations of Pb, Zn, Cu, and Cd in the roots of AMF inoculation and/or AA treatments were higher than those of the CK treatment ("NA + NM"), while the "IA + FM" treatment had the highest HMs contents, which increased by 28.07-168.98% and 42.83-486.31%, respectively (P < 0.05). Conversely, the concentrations of Pb, Zn, Cu, and Cd in the shoots of the "IA + FM" treatment at low and high concentrations were significantly lower than those of the CK treatment, which decreased by 48.80-89.82% and 54.56-76.49%, respectively (P < 0.05). Interestingly, the "IA + NM" treatment showed the highest concentrations of HMs in the shoots under the stress of high combined HMs concentrations (P < 0.05), which may be related to the improvement of water and nutrients delivery to the shoots by AA (Rucińska-Sobkowiak 2016). Meanwhile, the ranking of HM concentrations in reeds was Cu > Pb > Zn > > Cd. Moreover, it could also be seen from the Table S3 that AMF inoculation and AA deeply increased the BCF and decreased the TF of the reeds. Similar results were obtained by Han et al. (2021b), as AMF played a positive role in promoting the accumulation of HMs in roots and inhibiting their transfer to the shoots. In addition, Huang et al. (2017) reported that AMF symbiosis with higher BCF and lower TF promoted the tolerance of P. australis to Cd stress compared to the non-inoculated groups, which was consistent with our results. Meanwhile, previous studies also proved that the longer roots of mycorrhizal plants had the potential to increase the accumulation of combined HMs in roots, and combined HMs were preferentially accumulated in mycelium of fungi rather than root cells (Wu et al. 2016). This might be the underlying mechanism that AMF plants were able to increase HMs influx in roots. The differences between the results of the present study and Fritioff and Greger (2006) in the ability to accumulate Pb, Zn, Cu, and Cd might be determined by the characteristics of plants and HMs. AA accelerated the absorption and transport of water and nutrients, improving the growth of plants, which was beneficial to enhance the resistance to combined HM stress and the transfer of HMs from roots to shoots. More importantly, the AA improved AMF colonization in plant roots, resulting in higher accumulation of combined HMs in roots and improving plant physiological status. In summary, AMF inoculation and AA were beneficial for wetland plants to resist combined HM stress.

Distribution of Pb, Zn, Cu, and Cd in the substrates
As shown in Fig. 3A and B, AMF inoculation and/or AA significantly promoted the adsorption of Pb, Zn, Cu, and Cd in the substrates than those in the CK treatment ("NA + NM") at low and high concentrations (P < 0.05), and the "IA + FM" treatment showed the highest Pb, Zn, Cu, and Cd accumulation in the substrates. Compared with the CK treatment, the concentrations of Pb, Zn, Cu, and Cd in the substrates of the "IA + FM" treatment at low concentrations were 44.31, 32.34, 42.28, and 1.58 mg/kg, respectively, with an increase of 45.40%, 6.52%, 47.20%, and 17.10%, respectively, and the concentrations at high concentrations were 85.47, 105.97, 159.44, and 3.89 mg/kg, respectively, with an increase of 50.86%, 61.30%, 59.49%, and 48.80%, respectively. What's more, the ranking of Pb, Zn, Cu, and Cd accumulation contents in the substrates was Cu > Pb ≥ Zn > > Cd. In this study, it can be seen that inoculation with AMF enhanced the adsorption of combined HMs by the substrates. We speculated here that this was related to the fact that AMF could secrete some active matters (e.g., glomalin-related soil protein (GRSP)) to chelate the combined HMs in substrates, absorb the HMs ions via their hyphae and change the rhizosphere microbial community Wang and Feng 2021;Wang et al. 2020). GRSP released by AMF via mycelium and spore walls was water-insoluble glycoprotein, performing a remarkable role in chelating HMs (Wu et al. 2014). It was confirmed by Nafady and Elgharably (2018) that AMF inoculation presented a noteworthy decrease of Cd, Pb, Zn, and Cu concentrations due to the release of GRSP. Meanwhile, AA increased the oxygen content in the VFCWs, leading to the growth of microorganisms on the surface of the substrates and roots to form more biofilms, which can produce more extracellular polymeric substances (EPS) to remove combined HMs through chelation, surface precipitation, and ion exchange (Li and Yu 2014;More et al. 2014;Riaz et al. 2021). What's more, AA could also increase the frequencies of mycorrhizal symbiosis, thereby promoting the HM adsorption of the substrates through the potential mechanisms mentioned above. Under the two concentrations of combined HMs, the adsorption difference of Cu, Pb, Zn, and Cd might be attributed to the pH, redox potential, different chemical composition, and structure of the substrate material. For example, Hernández-Montoya et al. (2013) reported that zeolite showed a selectivity in the adsorption process of Pb, Zn and Cd, and the ranking was Pb > Zn > Cd.

Removal rates of Pb, Zn, Cu, and Cd under different treatments
It can be seen from Fig. 4A that the removal rates of Pb, Zn, Cu, and Cd of the four treatments all exceeded 80% at low concentrations, and those of the CK treatment ("NA + NM") were 95.17%, 80.36%, 91.66% and 80.68%, respectively. Among the four treatments, the removal rates of the "IA + FM" treatment were the highest, being 98.49%, 92.64%, 97.72%, and 92.06%, respectively, increased by 3.48%, 15.28%, 6.61%, and 14.11%, respectively, comparing with the CK treatment. At high concentrations, the removal rates of Pb, Zn, Cu, and Cd were lower than those at low concentrations. Comparing with the CK treatment, the removal rates of the AMF inoculation and/or AA treatments increased by 7.84-22.73%, 16.34-30.31%, 5.62-12.63%, 27.88-50.22%, respectively (Fig. 4B). At high concentrations, AMF inoculation and/ or AA could significantly improve the removal efficiencies of Pb, Zn, Cu, and Cd (P < 0.05). However, AMF inoculation and/or AA could significantly improve the At low concentrations, the removal rates of Zn and Cd in inoculated treatments ("IA + FM," "NA + FM") were significantly higher than those in the CK treatment (P < 0.05), but not for Pb and Cu. This may be due to the fact that the removal rates of Pb and Cu had been at a very high level (more than 90%), which made the role of AMF not obvious in promoting the performance. However, with the increase of combined HMs concentration, the adsorption efficiency of the substrates would be lower, and the toxicity of HMs to plants would be greater. Therefore, compared with the CK treatment, AMF had a more obvious effect on the removal of Pb, Zn, Cu, and Cd at high concentrations than that at low concentrations, with an increase of 11.34%, 20.03%, 5.62%, and 32.62%, respectively. Gunathilakae et al. (2018) found that AMF inoculation showed a more obvious effect on the removal of the high Cd level (50 ppm) than the low Cd levels (0, 5, 10, and 20 ppm) when evaluating the role of AMF on the phytoremediation potential of water hyacinth in Cd uptake, which was consistent with our result.

Distribution of Pb, Zn, Cu, and Cd in subcellular components
As shown in Fig. S3, Pb, Zn, Cu, and Cd were mainly distributed in the cell wall, especially in the underground part of the reeds. The proportions of Pb, Zn, Cu, and Cd in the cell wall in the aboveground part of the reeds under the two concentrations were 30. 39-73.93%, 46.74-58.43%, 43.90-72.73%, and 35.23-44.71%, respectively, and those of the underground part of the reeds were 71. 85-92.44%, 70.27-88.56%, 72.42-90.11%, and 48.40-69.07%, respectively. As the concentration of combined HMs increased, there was a decline in the proportions of Pb, Zn, Cu, and Cd in the cell wall in the aboveground part of the reeds, while a contrasting tendency was found in the underground part, except for "IA + FM" treatment. For the aboveground and underground parts, the Pb, Zn, Cu, and Cd proportions in the reed cell wall of the "IA + FM" treatment were higher than those of the CK treatment ("NA + NM") at low concentrations. The cumulative concentration of single Pb, Zn, Cu, and Cd in the aboveground and underground parts of the reeds all was cell wall > organelle ≥ soluble component.
The different types and concentrations of HMs induced a distinct subcellular combined HMs distribution. The components of cell wall (e.g., polysaccharides and proteins) provided a large number of retention sites to bind HMs and restricted their transmembrane transport, leading to a lower toxicity than the free state of that in soluble component and organelle (Han et al. 2021a;Mwamba et al. 2016). Gao et al. (2021) revealed that AMF inoculation improved the ability of the plant cell wall in fixing Cd by increasing the cell wall components, resulting in more Cd fixed in the cell walls, which was consistent with our study. More importantly, the fixation of HMs by cell wall was considered as the first barrier to protect protoplasts from the toxic effects of HMs in plants, while the mycorrhizal plants could promote the deposition of HMs on the plant cell walls and reduce the distribution in organelles, thereby alleviating the damage of HMs to the host plant organelles (Huang et al. 2018). Additionally, Fritz (2007) reported that the HM ions affinity in pectin (a component of cell wall) showed a difference depending on the characteristics of them (Al 3+ > Pb 2+ > Cu 2+ > Co 2+ > Ni 2+ > Zn 2 + > Cd 2 + > Ca 2+ ), which might account fo r t he differe nce s i n the distribution of Pb, Zn, Cu , a nd Cd in subcellular comp one nts in our study. According to the result of Considine and Foyer (2021), oxygen may be as a signal for plant growth, which affecting the root and shoot growth and structure, resulting in different distributions of combined HMs in plant subcellular. This study provides sufficient evidence at a subcellular level to illustrate how AMF inoculation and AA benefit plant cells to resist combined HMs tolerance.

Localization of Pb, Zn, Cu, Cd in ultrastructure of mycorrhiza
The normal root cells of reeds were presented in Fig. 5A (e.g., i: nucleolus, h: nucleus, g: cell membrane) applying transmission electron microscopy (TEM). However, under the stress of combined HMs, the root cell morphology of reeds not inoculated with AMF changed greatly, such as deformation of cell wall (k) and helical curl of cell membrane (l) (Fig. 5B). In addition, there were many thickened cell walls (j) and compartmentalized structures about 2-3 μm in diameter in reeds inoculated with AMF under the stress of high combined HM concentrations ( Fig. 5C-F). Meanwhile, there were many black deposits (a, b, c, d, e, and f) in cell walls and cell interiors, which were suspected to the Pb, Zn, Cu, and Cd. Interestingly, the mycelium of AMF penetrating the cell wall (red circle) of reeds from one cell to another were observed (Fig. 5F), which showed the process of AMF infecting plant cells. Additionally, energy dispersive spectrometer (EDS) was conducted for those selected black deposits (a, b, c, d, e, and f), and the results were shown in (Fig. S4). The presence of Pb, Zn, Cu, and Cd peaks indicated that the added HMs ions entered reed cells (a, b, c, d) and AMF structures (e, f). Since the samples were organic matter and the sample holder was made of aluminum material, there was no doubt that the C, O, and Al elements were also identified.
TEM can be used to observe the injuries to different organs of plants in response to combined HMs stress. In this part, we can clearly observe how AMF-inoculated plants resist HM damage at the cellular level. Huang et al. (2018) found that the organelle integrity of plant cells became worse with the increase of Cd concentration applying TEM, while the root cell structure of plants inoculated with AMF were more complete. In addition, Abdel-Fattah et al. (2011) and Xu et al. (2018) found that increasing cell wall thickness and altering subcellular compartmentalization by AMF was a significant mechanism for plants detoxification to HMs, which was consistent with our observations. Meanwhile, a similar result reported by González-Guerrero et al. (2008) that Zn, Cu, and Cd also could be accumulated in the cell walls of AMF when these HMs were added to the extra-root mycelium.

Combined HM mass balance
As shown in Fig. 6, it can be seen that combined HMs were mainly stored in the substrates, accounting for more than 50%, which was consistent with the results of Ban et al. (2021). Under the two concentrations, the proportions of combined HMs in the plants of the AMF inoculation and/or AA treatments were higher than those of the CK treatment ("NA + NM"), with the following ranking order: "IA + FM " > "NA + FM" > "IA + NM" > "NA + NM." Besides, the proportions of combined HMs in the roots of "NA + FM" treatment (more than 85%) were higher than those of the CK treatment at the low and high concentrations. Conversely, the proportions of combined HMs in the polluted water of the AMF inoculation and/or AA treatments than those of the CK treatment under the two concentrations decreased by 34.11-64.34% and 15.44-30.63%, respectively, and the "IA + FM" treatment was the lowest among the four treatments.
The removal rates of combined HMs in VFCWs were closely related to the synergistic effects of the plants, microorganisms, and substrates. Under the stress of combined HMs, AMF decreased the combined HM concentration in water by improving the plants uptake and promoting that absorbed in substrates, which was consistent with Riaz et al. (2021). In this study, the combined HM contents of the inoculated treatments in substrates were lower than that in the CK treatment at low concentrations, which was not consistent with Hu et al. (2021). The reason might be that the wetland plants adsorbed more HMs due to the low concentrations, thus resulting in the reduction of combined HMs in the substrates. Besides, AMF accumulated HMs in their intra-radical and extraradical mycelium, thus resulting in lower HM contents in the water. Wu et al. (2016) showed that the extraradical mycelium of AMF could absorb and transport Cr to mycorrhizal roots from a distance, which could contribute to promoting the ability of the wetland system to purify HM-polluted water. In addition, the effect of AA on the mass balance of combined HMs in aquatic habits was mainly related to the enhanced microorganism activities and improved HM uptake of wetland plants. Therefore, we found that AMF inoculation and AA showed significant synergistic effect on the removal of combined HMs from VFCWs.

Conclusions
In this study, the effects and mechanisms of arbuscular mycorrhizal fungi and/or additional aeration on combined heavy metal removal in vertical flow constructed wetlands were investigated at different heavy metal concentrations. The utilization of additional aeration in vertical flow constructed wetlands provided better performance for arbuscular mycorrhizal fungi colonization of reed roots. Compared with the CK treatment (autoclaved arbuscular mycorrhizal fungi inoculation and no aeration), the frequency of mycorrhiza values was increased by 10.58%, 17.64%, and 20.00% due to the addition of aeration under the combined heavy metal stress of three concentrations (0, low, and high), respectively. Additionally, arbuscular mycorrhizal fungi inoculation and/or additional aeration significantly promoted the reeds growth, antioxidant enzymes activities (SOD, POD, and CAT), and heavy metal accumulation in plant roots and substrates, and the promotion effect of arbuscular mycorrhizal fungi inoculation was greater than additional aeration. What's more, arbuscular mycorrhizal fungi inoculation and/or additional aeration also increased the removal rates of Pb, Zn, Cu, and Cd under the stress of combined heavy metals. In particular, the combination of arbuscular mycorrhizal fungi inoculation and additional aeration had the best purification performance, especially at high concentrations, the removal rates of Pb, Zn, Cu, and Cd were increased by 22.72%, 30.31%, 12.64% and 50.22%, respectively, and no obvious pattern was found between arbuscular mycorrhizal fungi inoculation and additional aeration. Besides, inoculation with arbuscular mycorrhizal fungi and/or additional aeration changed the subcellular structures distribution of combined heavy metals in reeds, and that mainly distributed in the cell walls. Transmission electron microscopy (TEM) and energy dispersive spectrometer (EDS) showed that arbuscular mycorrhizal fungi could induce thicker cell walls of plant roots and allow more combined heavy metals to be deposited in plant cell walls and fungal structures. Overall, the reeds-arbuscular mycorrhizal fungi/additional aeration system exhibits great potential for application in remediation of combined heavy metal-polluted wastewater.