**2. Results**

Strain 172205Δ*enc* was obtained by the enterocin biosynthetic gene cluster deletion (Figure 1a) and confirmed by PCR (polymerase chain reaction) amplification (Supplementary Figure S1 and Table S1). The HPLC profile of crude extract in D.O. medium (Figure 1b) showed that enterocin biosynthesis were totally blocked in mutant strain 172205Δ*enc*, including its intermediate metabolite-cinnamic acid.

**Figure 1.** (**a**) The organization of enterocin biosynthetic gene cluster before and after deletion. Amplification fragments by verifying primers were labeled by fronts in red; (**b**) HPLC detection of crude extracts of strain 172205 wild type and Δ*enc* in D.O. medium.

Almost 80 grams crude extract was obtained from extraction of 60 L fermentation broth of mutant 172205Δ*enc* and subjected to column chromatography and semi-preparative HPLC purification to afford compounds **1**–**5.** The chemical structures were showed in Figure 2.

**Figure 2.** Chemical structures of compound **1**–**5**.

15*R*-17,18-dehydroxantholipin (**1**) was obtained as a dark red powder with the molecular formula of C27H16ClNO9 (HRESIMS-high resolution electrospray ionisation mass spectrometry m/z 534.0587, calcd 534.0586 for [M + H]+), implying 20 degrees of unsaturation. Detailed comparison of 1D NMR (Table 1) between compound **1** and the reported xantholipin [10,11] showed that **1** had the same NMR data with xantholipin, except for the two sp2 quaternary carbons C-17 (δ<sup>C</sup> 136.0) and C-18 (δ<sup>C</sup> 141.5), implying a double bond. The mass spectrum suggested the loss of a H2O group to form a double bond between C-17 and C-18. The correlations in HMBC (heteronuclear multiple bond coherence) from H-16a to C-13, C-15, C-17 and C-18 also located the placement at C-17 and C-18. Thus, the planar structure of **1** was determined. The key HMBC and COSY (homonuclear correlation spectroscopy) correlations are shown in Figure 3. Furthermore, the quantum chemical ECD calculation method was also used to determine the absolute configuration. The calculated ECD spectrum of **1** was compared with the experimental one, which revealed an excellent agreement between them (Figure 4). Therefore, the absolute configuration of **1** was assigned to 15*R*. Thus, the structure of **1** was determined.


**Table 1.** 1H and 13C NMR data for Compound **1**–**2**.

\*, \*\* Assignments are made in comparison with literature data for similar reported compounds. **1** (800 and 200 MHz, DMSO-*d6*, δ in ppm); **2** (500 and 125 MHz, CD3OD-*d4*, δ in ppm).

(3*E*,5*E*,7*E*)-3-methyldeca-3,5,7-triene-2,9-dione (**2**) was obtained as a yellow powder. The molecular formula of **2** was determined as C11H14O2 (HRESIMS *m*/*z* 179.1064, calcd 179.1067 for [M + H]+), indicating 5 degrees of unsaturation. The 1D and HSQC (heteronuclear single quantum coherence) NMR data (Table 1) of **2** revealed the presence of three aliphatic methyls, two carbonyls (ketone) (δ<sup>C</sup> 202.0 and 201.3), five olefinic methines (C-4 to C-8) and six olefinic carbons. The 1H-1H COSY correlations (Figure 3) from H-4 to H-8 revealed the conjugated system indicated by bold lines in Figure 3. HMBC correlations from H-5 to C-3 and C-7, H-6 to C-4 and C-7, H-7 to C-5 and C-6 and H-8

to C-6 and C-7 suggested that **2** contained three conjugated double bonds. HMBC correlations from H-11 to C-2 and C-3, and H-4 to C-2 and C-11 located one methyl (H-11) at C-3. HMBC correlations from H-1 to C-2 and C-4, and H-11 to C-2 and C-3 supported one methyl (C-1) located at C-2 and the connection between C-2 and C-3. Moreover, HMBC correlation from H-10 to C-8 and C-9, H-8 to C-9, and H-7 to C-9 assigned another methyl ketone location at C-8. *J*H-5/H-6 = 14.6 Hz and *J*H-7/H-8 = 15.6 Hz revealed two (*E*)-alkene between C-5-C-8. The ROESY (rotating frame overhauser effect spectroscopy) correlation of H-4 and H-1 supported the (*E*)-alkene between C-3 and C-4. Thus, the (3*E*,5*E*,7*E*)-triene was identified and chemical structure of **2** was established.

**Figure 3.** Key heteronuclear multiple bond coherence (HMBC), homonuclear correlation spectroscopy (COSY) and rotating frame overhauser effect spectroscopy (ROESY) correlations of compound **1**–**5**.

Qinlactone A (**3**) was obtained as a colorless oil. Its HRESIMS data indicated its molecular formula is C16H22O4 (*m*/*z* 279.1587, calcd 279.1591 for [M + H]+) with 6 degrees of unsaturation. The 1D (Table 2) and HSQC NMR data of **3** revealed the presence of five aliphatic methyl groups, one carbonyl (ketone, δ<sup>C</sup> 202.2) and one carbonyl (ester, δ<sup>C</sup> 183.3) and five olefinic methines (C-5 to C-9). The 1H-1H correlations of H-5/H-6/H-7 and H-8/H-9 confirmed the same three conjugated double bonds as **2**. The HMBC correlations (Figure 3) from H-12 to C-11 (δ<sup>C</sup> 202.2) and C-10, H-9 to C-11 and H-16 to C-8/C-9/C-10/C-11 confirmed the location of one methyl ketone and a methyl (C-16) at C-10. ROESY correlations of H-12/H-9, H-8/H-16 and H-9/H-7 confirmed all the (E)-alkenes from C-5 to C-10. Additionally, HMBC correlations from H-13 to C-1/C-2/C-3/C-14 and H-14 to C-1/C-2/C-3/C-13 suggested two aliphatic methyl groups located at quaternary carbon C-2, and indicated the connection from C1 to C3. HMBC correlations from H-15 to C-3/C-4 located the last methyl (δ<sup>C</sup> 22.1) at C-4. Two carbons connected to oxygen atoms (δ<sup>C</sup> 81.5 and 88.0) suggested the OH group at C-3 and ester oxygen connected with C-4. Thus, based on the unsaturation and ester group (δ<sup>C</sup> 183.8, C-1), a γ-lactone structure was revealed. The HMBC correlations from H-3 to C-1/C-2/C-4/C-5/C-13/C-14 also confirmed γ-lactone in **3**. Meanwhile, the HMBC correlations of H-5 to C-4 and H-6 to C-4 revealed the connection of lactone and conjugated olefin part. Thus, the planar structure of **3** was established. The relative configuration of **3** was established by ROESY experiment. The ROESY correlation of H-3 and H-14/H-5 suggested the relative configuration of *3R\**, *4S\** (Figure 3). The absolute configuration was confirmed by a good agreement between the calculated ECD spectrum of **3** and experimental one (Figure 4). Therefore, the absolute configuration **3** was assigned to *3R, 4S*. Compound **3** was named qinlactone A.


1H and 13C NMR data for Compound **3**–**5** (500 and 125 MHz, CD3OD-*d4*, δ in ppm).

**Table 2.**

Qinlactone B (**4**) was obtained as a colorless oil and assigned the same molecular formula as **3** by HRESIMS (*m*/*z* 279.1587, [M + H]+). The 1D and 2D NMR data (Table 2) of **4** corresponded closely to those of **3**, which suggested **4** had the same planar structure with **3** as epimer instead of enantiomer. The ROESY correlations of H-3 and H-15/H-13 suggested the relative configuration of *3R\**, *4R\** (Figure 3). The absolute configuration was determined as *3R*, *4R* by the similar Cotton effects between the calculated ECD spectrum and experimental one of **4** (Figure 4). Thus, the structure of **4** was determined (Figure 2), named qinlactone B.

**Figure 4.** Experimental and calculated electronic circular dichroism (ECD) spectra for compound **1**, **3** and **4**.

Qinlactone C (**5**) was obtained as a light-yellow oil with the molecular formula C16H24O6, determined by HRESIMS *m*/*z* 313.1650 (calcd 313.1646 for [M + H]+), implying 5 degrees of unsaturation. Compared with **3**, 1D NMR data (Table 2) of **5** revealed the absence of two olefinic methines, but the presence of two methines connected with oxygen atoms (δC/<sup>H</sup> 72.8/4.44 and 77.9/3.63). Based on the formula and unsaturation analysis, **5** contained a vicinal diol at C-5 and C-6, which was also confirmed by 1H-1H COSY correlations from H-5 to H-9 and HMBC correlations from H-5 to C-3 and C-4 (Figure 3). Therefore, the planar structure of **5** was identified. ROESY correlations of H-3 and H-13/H-5 suggested the relative configuration of *3R\**, *4R\** in the lactone ring (Figure 3). However, due to the existing vicinal diol structure, the relative and absolute configuration of **5** could not be determined based on the present data. Thus, compound **5** was named qinlactone C.

In the anti-microbial bioassay test, only compound **1** exhibited strong bioactivity against *Staphylococcus aureus* and *Candida albicans*, with MIC values of 0.78 μg/mL and 3.13 μg/mL, respectively. Meanwhile, in antiproliferative bio-test, **1** showed strong inhibitory effects on MCF-7 and HeLa cell lines with IC50 values of 5.78 μM and 6.25 μM, respectively (Table 3). Compound **2**–**4** showed weak activities against MCF-7 and HeLa cell lines with IC50 values ranging from 129 to 207 μM (Table 4).


**Table 3.** MIC (μg/mL) against pathogenic microbes of compounds **1**–**5**. *E. coli*: *Escherichia coli*; *S. aureus*: *Staphylococcus aureus*; *C. albicans*: *Candida albicans*.



#### **3. Discussion**

In this study, we identified five new compounds with bioactivities from mangrove *Streptomyces qinglanensis* 172205 with "genetic dereplication." Compound **1** showed strong anti-*Staphylococcus aureus* and anti-*Candida albicans* activities with MIC values of 0.78 μg/mL and 3.13 μg/mL, respectively, and exhibited strong cytotoxicities against MCF-7 and HeLa cell lines with IC50 values of 5.78 μM and 6.25 μM, respectively. However, compound **2**–**4** exhibited only weak antiproliferative activities with IC50 values ranging from 129 to 207 μM.

Attempting to activate gene clusters of possible unknown secondary metabolites in strain 172205, we deleted the whole biosynthetic gene cluster of the main product enterocin. Through HPLC detection, we found that this strategy in strain 172205 did not clearly activate any new metabolites in several media. However, we still focused on some low-yield products, which produced in the wild type strain as well (Figure 1). However, the mutant strain without the main product enterocin facilitated detection and isolation of the low-yield products, from which we finally identified five new compounds. Thus, "genetic dereplication" did help simplifying the process of isolation and mining the low-yield products. This strategy would be more effective for identifying multiple types of metabolites in one strain, if combined with other genome mining tools or methods.

15*R*-17,18-dehydroxantholipin is an analog of reported xantholipin [10], which exhibited similar strong antiproliferative and anti-microbial activities. In fact, we firstly identified a gene cluster by antiSMASH analysis in the genome of strain 172205 (Supplementary Figure S41 and Table S2), which had a high similarity with the reported xantholipin gene cluster [12]. Then, we tried at least 10 media to detect the similar UV absorption of xantholipin by HPLC and characterized mass for halogen compounds by HRESIMS, and finally detected analogs in products from D.O. medium, which is the same recipe with the reported medium to produce xantholipin. Compound with the targeted UV absorption was identified as 15*R*-17,18-dehydroxantholipin. Lacking the oxidoreductase gene *xanZ2* which was proposed for the double band reduction at C-17 and C-18 [12], resulted in the production of 15*R*-17,18-dehydroxantholipin in strain 172205. Hence, genome-guided compound discovery combined with OSMAC is an effective method for isolation and identification of some well-known and valuable compounds or potential new analogs. Moreover, multiple strategies of genome mining will be helpful to mine the potential of natural products in one strain.
