Two sets of ten pots were installed. One set was treated with sulfamethoxazole solution (75 mg/L) and the other with water. Within each set, three pots were assigned to each vegetable (spinach as Spinacia oleracea, cilantro as Coriandrum sativum, and basil as Ocimum basilicum), and one pot was designated for soil as control. All pots (22 cm in diameter, 21 cm in height) were filled with a layer of 18 cm organic soil as the substrate for plant growth. SMX solution (75 mg/L) was prepared using 98% SMX from Shanghai Aladdin Biochemical Technology Co., Ltd. The experiment was conducted under a temperature range of 27°C–33°C. Two sets of pots were treated with respective solutions (avg. 60mL/day) to keep them moist in an environment with a relative humidity of 57%.

The spinach seeds were directly dispersed into the pots. But basil and cilantro seeds had to break dormancy for all plants to sprout at a relatively similar time. Each cilantro seed was a pod of two or three individual seeds. The pods were broken gently using the handle of a metal spoon and soaked in 38°C water for 24 h. The basil seeds were placed on a sheet of moist paper towels for 24 h. After the pre-germination process, 12 seeds were dispersed evenly in each pot, different by plant type. The seeds were covered with an additional 3 mm layer of soil. Shortened names are assigned to each of the eight experimental groups (Table 1).

After 19 days, four plants of medium height in each group were selected to track height and leaf growth. Heights were measured from the surface of the soil to the highest point of the plant. An original soil sample was collected. Three weeks after sowing, soil samples were collected from each group by either gathering soil of the plant rhizosphere or if no plant was present they were directly taken from the pot. All samples were kept at −80°C before being processed and analyzed at Wekemo Tech Co., Ltd. Shenzhen, China.

DNA was extracted from the soil samples using the CTAB method. First, the soil samples were transferred into 2.0-mL Eppendorf tubes containing 1,000 µL of CTAB extraction buffer. The tubes were then bathed at 65°C and gently mixed until adequate homogenization. After centrifugation, the supernatants were mixed with an equal volume of phenol chloroform isoamyl alcohol (25:24:1) which were centrifuged at 12,000 rpm for 10 min. Chloroform isoamyl alcohol (24:1) was added to supernatants, followed by mixing and 10 min centrifugation at 12,000 rpm. Next, the supernatants were transferred into 1.5-mL tubes with isopropyl alcohol, mixed by inversion, and incubated at −20°C. After the precipitation and solution separated, the tubes were centrifuged at 12,000 rpm for 10 min and the precipitations were washed with 1 mL 75% ethanol twice. After drying the precipitation, ddH₂O and 1ul of RNase were added. The mixtures were incubated at 37°C for 15 min.

Vegetable growth under SMX stress was accessed through two parameters visualized in Figure 1: stem height (graph a-c) and leaf length (graph d-f). At the start of the measuring period, all plants treated with SMX solution had similar lengths compared to those under normal conditions. A linear regression analysis showed that SMX solution improved basil growth on average by 0.05 cm/day and its final medium height (10.5 cm) surpassed the height under control conditions (9.3 cm) by 12.9%. However, SMX only changed the growth of cilantro and spinach by about 0.01 cm/day. A further significance analysis corresponded to the linear regression, revealing that 75 mg/L SMX solution did not pose significant impacts on basil, cilantro, and spinach growth (p-value >0.05 for all three plants).

Compared to the effect on stem growth, SMX had a clearer impact on leaf growth in all vegetables. Graph d-f in Figure 1 shows the leaf lengths at the start, middle, and end of the measuring period. Opposite to SMX’s facilitation in basil stem development, antibiotics led to smaller basil leaves throughout the measuring period. Its final medium length (3.92 cm) under SMX treatment was 13.8% shorter than the average leaf length (4.55 cm) of the control group. SMX also adversely affected the development of cilantro leaves with a final inhibition rate of 5.9%. Spinach leaf lengths were similar between the antibiotic group and the control group at the start and middle of the growth period. However, by the end of the period, Spinach leaves in the antibiotics group (12.13 cm) significantly outperformed the control group (9.05 cm) by 34.0% (p-value = 0.042 < 0.05).

Stem and leaf growth are indicators to assess the phytotoxicity of environmental pollutants (Mei et al., 2021). The data showed that SMX insignificantly promoted the basil stem growth. This result corresponded to Zhang et al.‘s study on pakchoi (B. chinensis) where SMX improved pakchoi height (Zhang et al., 2017). Interestingly, our findings showed the opposite of the general understanding that SMX adversely affects plant growth (Cheong et al., 2020). For instance, high SMX concentration in the soil could lead to a 44.7% decrease in tomato plant height and a 38.2% decrease in cucumber plant height. Similarly, Lv et al. found that exposure to 100 mg/L SMX solution resulted in a 28.13% decrease in strawberry (Fragaria ananassa) stem weight (Lv et al., 2021). This dichotomous effect of SMX between our results and the general knowledge could be explained by the plant’s hormetic response to antibiotics. Researchers found that while high concentrations of antibiotics tend to inhibit plant growth, low concentrations actually facilitate stem development (Migliore et al., 2000). This phenomenon was supported by recent studies. The low dosages of SMX led to longer stems in both pakchoi and strawberry, while higher concentrations reduced the stem length (Zhang et al., 2017; Lv et al., 2021). Therefore, it is possible that under SMX stress basil exhibited an insignificant height increase, while cilantro and spinach were not affected at all.

SMX impacted the leaf differently compared to its effect on plant height. SMX’s inhibition effect on basil and cilantro leaves could be due to the antibiotic’s higher concentration in leaves than stems. Although the sulfamethoxazole uptake mechanism is not fully understood, the majority of studies have pointed to passive diffusion as the main pathway (Franklin et al., 2016). Thus, the transpiration stream plays a crucial role in SMX’s translocation. Since stems are the main antibiotic channels rather than reservoirs, stems do not retain as many antibiotics as leaves (Liu et al., 2013). Several studies have shown that leaves accumulate more antibiotics than stems (Herklotz et al., 2010; Liu et al., 2013; Zhao et al., 2019). Therefore, basil and cilantro leaves could accumulate higher antibiotic concentrations than stems, which inhibits leaf development. On the other hand, spinach tends to accumulate fewer antibiotics in its tissues than other plants (Kodešová et al., 2019). Thus, SMX could exist in spinach leaves at a relatively lower concentration just enough to stimulate leaf growth.

The results showed that plants had different antibiotic tolerance. While SMX led to longer basil stems and larger spinach leaves, it suppressed cilantro’s growth. However, SMX’s impact on plant growth was overall insignificant, which suggested that vegetable stem and length lengths were not reliable indicators of SMX presence in the soil. Despite these insignificant impacts, SMX could still noticeably affect rhizobacteria diversity. Therefore, a further investigation into the microbial community under SMX stress is needed.

Raw sequence libraries were constructed based on the soil sample from eight pots. The number of raw reads ranged from 22884329 to 33657674 (avg. 26714901). Low-quality or short sequences were filtered out with the Trimmomatic algorithm, followed by the removal of host gene. As a result, 21521875 to 32085255 clean reads (avg. 25269515) were obtained for each group. A comparison with kraken2 and a self-built database (Wekemo Tech Co.) revealed the taxonomic distribution of each sample.

Alpha diversity indices, calculating each sample based on 5329 identified operational taxonomic units (OTUs), are shown in Table 2. Both the abundance-based coverage estimator (ace) and chao1 assessed the species richness. Compared to ace, chao1 gave more weight to the species of low abundance (Bo-Ra et al., 2017). Higher ace and chao1 values were observed for soil, basil, and cilantro treated with SMX. This result suggested that the presence of SMX in the sample increased the species richness except for spinach. Shannon and Simpson’s indices reflected bacterial diversity through richness and evenness. While Shannon emphasized more on bacterial species richness, Simpson focused on evenness (Suturina et al., 2022). As both Shannon and Simpson indices were generally higher for SMX groups than control, SMX in this experiment increased bacterial richness and evenness in all samples.

The microbial composition at phylum, order, family, and genus levels are presented in Figure 2. SMX changed microbial composition in pots with the vegetables. At the phylum level, Pseudomonadota was identified as the dominant phylum for all samples, followed by Actinomycetota. Pseudomonadota, better known as proteobacteria, is the most prevalent bacterial phylum in the environment (Leão et al., 2023; Selmani et al., 2023). This phylum includes many rhizobacteria associated with plant growth promoting activities, including nitrogen fixation, phosphorus uptake, and antifungal effect (Ramakrishna et al., 2019; Gómez-Godínez et al., 2023). At order rank, Xanthomonadales, Burkholderiales, and Pseudomonadales are the top three most dominant orders all of which belong to the phylum Psedomonadota. SMX increased the richness of Xanthomonadales in BAA (16.4%) and CIA (16.8%), compared to Xanthomonadales proportion in BAT (9.2%) and CIT (10.3%). At the family level, Xanthomonadacae, under order Xanthomonadales, was the most dominant family. It existed in a larger abundance in BAA and CIA than BAT and CIT. Although SPA had a similar soil profile to other plants, SPT altered the microbial composition significantly. Xanthomonadacae took up 42.1% in SPT, compared to the average abundance of Xanthomonadacae in the rest of the samples (13.8%). Spinach also facilitated the proliferation of Rhizobiaceae (15.9%), a family under Hyphomicrobiales, which is a symbiotic nitrogen-fixing bacteria (Ormeño-Orrillo et al., 2015). Gómez et al. found that Rhizobium could colonize spinach roots and promote root, shoot, and leaf growth (Jimenez Gómez et al., 2016).

At the genus level, the effect of antibiotics was registered across all plant types. Lysobacter under the family Xanthomonadacae was the dominant genus. Its abundance in basil, cilantro, and spinach with SMX (14.9%, 14.8%, 9.0% respectively) was consistently higher than in samples treated with water (7.2%, 8.8%, 7.7%). Lysobacter protects plants through the release of antimicrobial compounds and facilitates the degradation of macromolecules with extracellular enzymes (Gómez Expósito et al., 2015; Brescia et al., 2020). Its high concentration in sulfamethoxazole-contaminated soil could be explained by plants’ need for antibiotic degradation. Pseudomonas and Hydrogenophaga were the second and third most abundant genus. While Pseudomonas promotes plant growth and phytopathogen control, Hydrogenophaga modifies pollutants through enzymic activities and transforms them into less toxic substances (Sah et al., 2021; Majeed et al., 2022). SMX did not have noticeable impacts on the abundance of Pseudomonas and Hydrogenophaga. Three genera emerged from SPT, including Pseudoxanthomonas, Sphingomonas, and Neorhizobium. Peseudoxanthomonas and Neorhizobium facilitate plant growth through molecule degradation and nitrogen-fixing respectively. Sphingomonas increase plant’s stress tolerance and production of phytohormone (Sah et al., 2021; Lombardino et al., 2022; Marzba et al., 2022).

From the heat map in Figure 3, SPA had high abundances of Sphigobium, Hydrogenphaga, Acidvorax, Stutzerimonas, and Bordetella, which could induce spinach’s larger leaf length. On the other hand, SMX led to decreases in the abundance of Streptomyces and Sulfuricaulis in the basil and cilantro rhizosphere. Streptomyces aids the plant with nitrogen, phosphorus, and potassium uptake while serving as a biocontrol agent against plant pathogens (Amaresan et al., 2018). Sulfuricaulis plays the role of a sulfur-oxidizer. The decrease in Streptomyces and Sulfuricaulis under SMX stress might eventually lead to smaller basil and cilantro leaves.

Overall, SMX impacted the bacterial composition and increased the diversity in all samples of this experiment. Past studies have shown that antibiotics altered the bacteria diversity through the elimination of sensitive bacteria and the outgrowth of resistant bacteria (Cycoń et al., 2019). Corresponding to the study by Reichel et al., Pseudomonas were insignificantly impacted by SMX, possibly because of the prevalence of resistance genes in Pseudomonas (Reichel et al., 2014). SMX’s impact on bacterial composition could contribute to the changes in plant growth. For instance, a decrease in the abundance of Streptomyces and Sulfuricaulis could have led to smaller basil and cilantro leaves, while the increase in Sphigobium, Hydrogenphaga, Acidvorax, Stutzerimonas, Bordetella might facilitate spinach leaf growth. Although the analysis of microbial diversity demonstrated SMX’s impact on soil bacteria, an examination of ARGs was still needed to study the correlation between antibiotics and ARB.

The cleaned sequence library was compared with the protein database UniRef 90. ARGs in each sample were identified through the ARBD database. According to Figure 4, MacB was the most abundant gene in every rhizosphere sample (average 19.4%). This gene encodes for an ATP-binding cassette (ABC) transporter that exports antibiotics specifically macrolide through mechanotransmission (Greene et al., 2018). Among the top 10 genes, 7 genes encoded for antibiotics efflux pump (MacB, BcrA, MexF, CeoB, MexW, RosB, AcrB), and 3 genes encoded for antibiotics target alternation (VanRA, BacA, Pbp1a) (Charlebois et al., 2012; Stogios and Savchenko, 2020; Fauzia et al., 2023). The presence of SMX spiked the amount of Sul1 on average by 108% and increased the amount of Sul2 by 33.7% (p-value = 0.006). Both Sul1 and Sul2 protect the bacteria through antibiotic target alternation by encoding for dihydropteroate synthase (DHPS) with low affinity to sulfamethoxazole. This discovery supported the fact that antibiotics could establish selective pressure that led to the proliferation of specific ARGs (Skandalis et al., 2021).

KEGG orthology database identified host bacteria and annotated functional genes similar to known resistance genes (Martínez et al., 2015). Figure 5 reveals the concentration of functional genes among bacterial genera. Due to its greatest abundance, Lysobacter hosted most functional genes. Among the top 200 species, genus Pseudomonas held 13,778 reads on average, while Hydrogenophaga contained 84,529 reads. Pseudomonas’s abundance was only 3.6% higher than Hydrogenophaga’s, but Pseudomonas held 62.4% more functional genes. This finding revealed that Pseudomonas tend to host more functional genes than Hydrogenphaga. Correspondent to our discovery, Ashy et al. also concluded that Pseudomonas was a key host for viral genes and antibiotic resistance genes in the rhizosphere of moringa (Moringa oleifera) (Ashy et al., 2023). In the soil sample of spinach treated with water, Stenotrophomonas and Agrobacterium were not abundant genera, but hosted relatively large amounts of functional genes. The presence of those functional gene hosts, namely, Pseudomonas, Stenotrophomonas, and Agrobacterium, could facilitate the proliferation of ARGs. Figure 6 visualizes the interactions among bacterial genera. Shingomonas and Alfipia are two common bacteria that did not carry many functional genes but hindered other key hosts. Shingomonas competed with Pseudomonas, while Alfipia limited the growth of Stenotrophomonas and Agrobacterium. Therefore, it would be potentially beneficial to monitor and facilitate the growth of Shingomonas and Alfipia to outcompete functional gene hosts.