RNA biosensor and screening method thereof, ectoine-producing bacterial strain and screening method thereof

Abstract

This invention relates to RNA biosensors and their screening methods, as well as ectoin-producing strains and their screening methods. The RNA biosensor is used to screen ectoin-producing strains, and the base sequence of the RNA sensor includes the sequence shown in SEQ ID NO.1. The aforementioned RNA biosensor exhibits high sensitivity and specificity; the fluorescence intensity of fluorescent detection cells constructed using it increases linearly with ectoin concentration without significant background interference. Combined with microfluidic technology, it can be used for high-throughput screening of ectoin-producing strains. This application innovatively combines RNA biosensors and microfluidic screening technology, and with the aid of AI prediction, successfully develops a highly efficient method for screening high-yielding naturally induced ectoin-producing mutant strains. The RNA biosensor generated through AI prediction can rapidly and with high throughput sort high-yielding strains produced by natural mutagenesis (non-transgenic).

Description

Technical Field

[0001] This invention relates to the field of biotechnology, and in particular to RNA biosensors and their screening methods, as well as ectoin-producing strains and their screening methods. Background Technology

[0002] Ectoine, also known as tetrahydropyrimidine, is a polar, water-soluble, small-molecule cyclic amino acid derivative. It is widely found in halophilic and halophilic bacteria. As an important compatible solute, it helps organisms resist various extreme environmental stresses such as high osmotic pressure, high temperature, high salinity, drought, and ultraviolet radiation, playing a protective and stabilizing role for proteins, nucleic acids, biomembranes, and the entire cell. Due to these excellent properties, ectoine is widely used in the cosmetics industry for moisturizing, anti-wrinkle, and repair products, providing comprehensive care for the skin. In the pharmaceutical field, it can also serve as a biological stabilizer and drug carrier to improve drug stability and bioavailability. Furthermore, ectoine plays an important role in the food and enzyme industries, showing broad application prospects.

[0003] However, while synthetic biology techniques for genetically modifying strains have yielded some progress in the development of high-yield ectoine strains, they often face numerous challenges. Genetic modification is complex and costly, requiring significant time and effort for optimization and validation. More importantly, genetically modified strains may be limited in certain applications, especially in a market environment with increasing demand for natural, non-GMO products. Therefore, developing non-GMO ectoine products is of great significance.

[0004] The production of ectoine from non-GMO sources mainly relies on high-yielding strains generated through natural mutagenesis (non-GMO). Currently, there is a lack of high-throughput screening methods for high-yielding ectoine strains. Traditional screening methods depend on modifying the strains before fermentation, followed by detection of ectoine content in the fermentation broth using liquid chromatography. However, this method involves complex sample processing, long processing time (approximately 10 minutes per sample), high detection costs, and difficulty in achieving 100,000 samples per test. 4 Rapid detection of the above samples. Furthermore, the probability of obtaining high-yielding ectoine-producing strains through natural mutagenesis is extremely low, typically requiring screening of 10... 6 Only by using different mutant strains can high-yield strains be obtained, which further highlights the urgent need to develop high-throughput screening methods. Summary of the Invention

[0005] Based on this, this application provides an RNA biosensor and its screening method, as well as an ectoin-producing bacterial strain and its screening method. Fluorescent detection cells constructed using this RNA biosensor exhibit a linear increase in fluorescence intensity with ectoin concentration and no significant background interference. Combined with microfluidic technology, it can be used for high-throughput screening of ectoin-producing bacterial strains.

[0006] An RNA biosensor for screening ectoin-producing strains, the base sequence of the RNA sensor comprising the sequence shown in SEQ ID NO.1.

[0007] The aforementioned RNA biosensor exhibits high sensitivity and specificity. The fluorescence intensity of the fluorescent detection cells constructed using it increases linearly with ectoin concentration without significant background interference. Combined with microfluidic technology, it can be used for high-throughput screening of ectoin-producing strains.

[0008] Furthermore, this application provides a method for predicting the activation intensity of RNA biosensors and its application in high-throughput virtual screening. The method introduces a sequence-structure dual-modal fusion mechanism based on deep learning to achieve accurate modeling of the activation response capability between single-stranded RNA and target small molecules, thereby assisting in the design and screening of novel RNA biosensors.

[0009] The data-driven modeling approach adopted in this invention combines a pre-trained RNA language model with the SMILES representation learning method to construct a cross-modal prediction framework. This framework can quickly predict the fluorescence activation intensity or functional response intensity of any RNA sequence in the presence of specific small molecules without the need for experimental screening. Furthermore, it obtains the optimal candidate sequence through sorting and screening, significantly improving the efficiency of RNA biosensor development.

[0010] A method for predicting the intensity of activation response between single-stranded RNA and small molecule compounds, comprising the following steps:

[0011] Obtain a dataset containing RNA sequences, SMILES structural formulas of small molecule compounds, and corresponding Kd dissociation constant labels, where Kd is the affinity between RNA molecules and small molecules measured in an experimental environment;

[0012] The RNA sequence is embedded and encoded using a pre-trained RNA language model, and semantic vector features representing its primary sequence and potential conformation-sensitive regions are extracted to obtain the RNA embedding vector.

[0013] The SMILES expression of the small molecule compound is embedded and encoded using a pre-trained SMILES molecular structure model to obtain a molecular vector representation containing its functional group features and molecular structure topology information, thus obtaining a small molecule embedding vector.

[0014] A neural network prediction model is constructed by inputting the RNA embedding vector and the small molecule embedding vector into a fusion network, and performing Kd regression prediction through a fully connected network layer.

[0015] The neural network prediction model was trained and validated using leave-one-out cross-validation to obtain a prediction model with generalization ability.

[0016] Based on the trained model, multiple different RNA sequences and the SMILES structure of the small molecule compound to be tested are input, Kd prediction values ​​are output in batches, and the prediction results are sorted and filtered to select the 100 candidate RNA sequences with the highest prediction values.

[0017] The 100 candidate RNA sequences were experimentally screened to obtain the RNA sequence with the highest fluorescence value, which was then used as a biosensor for detecting ectoin.

[0018] The advantage of the aforementioned prediction method lies in its ability to encode the interaction mechanism between RNA sequences and target small molecules into digital feature vectors, and then use deep learning techniques to perform nonlinear mapping modeling, thereby achieving efficient in vitro prediction and candidate screening. Compared to traditional in vivo or in vitro experimental screening procedures, this method features wet experiment independence, high-throughput prediction, and low time cost, making it suitable for various novel RNA biosensor design scenarios, especially for small molecule drug detection and metabolite sensing.

[0019] The virtual screening system built based on the above prediction model, combined with a random RNA sequence generator and a predictive sorting module, can rapidly screen a given small molecule target from millions of RNA sequences. Combined with the subsequently constructed fluorescent detection cells or synthetic biological circuits, it can realize the design and verification of a full-process RNA biosensor.

[0020] A method for screening an RNA biosensor, characterized in that the RNA biosensor is used to screen ectoin-producing bacterial strains, and the screening method includes the following steps:

[0021] The RiboBindNet model was used to perform high-throughput screening of saturated RNA mutant libraries, and the top N candidate RNA sequences were selected by sorting them from high to low according to their prediction scores. The RiboBindNet model is an artificial intelligence model system used to predict the binding ability between single-stranded RNA and ectoine in the saturated RNA mutant library.

[0022] Construct corresponding fluorescent detection cells from the first N candidate RNA sequences;

[0023] The fluorescent detection cells corresponding to each candidate RNA sequence were cultured in a culture medium containing ectoine at different concentrations. The changes in fluorescence intensity of the fluorescent detection cells at different ectoine concentrations were analyzed by flow cytometry, and Kd prediction was performed. The RNA biosensor was obtained based on the prediction results.

[0024] In some embodiments, the RiboBindNet model includes: an RNA sequence encoding module, a small molecule structure encoding module, a neural network module, and a Kd prediction module;

[0025] The RNA sequence encoding module is constructed based on a pre-trained RNA language model. It uses an improved Transformer architecture to perform context learning on the input single-stranded RNA base sequence and combines optional RNA secondary structure prediction information to extract potential conformation-sensitive regions, fold-driven domains and binding site information, and outputs a set of vector representations characterizing the semantics of RNA structure.

[0026] The small molecule structure encoding module is based on the idea of ​​molecular language modeling. It uses the SMILES Transformer model to embed and encode the SMILES string of the target small molecule, fully learns its atomic arrangement order, functional group distribution and spatial topological features, and obtains a continuous molecular representation vector.

[0027] The neural network module is used to perform feature concatenation, weighted fusion, or cross-attention matching between RNA embedding vectors and small molecule embedding vectors.

[0028] The Kd prediction module uses a fully connected regression network structure to output the Kd prediction value of the corresponding RNA-small molecule combination, so as to reflect the change in fluorescence expression intensity of the RNA sequence in the presence of the target small molecule in the cellular environment.

[0029] In some embodiments, the RiboBindNet model includes:

[0030] The RNA sequence encoding module, combined with a pre-trained RNA language model, extracts features from the RNA sequence and converts the sequence information into a numerical vector.

[0031] The small molecule structure encoding module is used to extract features of the chemical structure of ectoin based on SMILES characterization, and to convert small molecule compound information into numerical vectors.

[0032] A neural network module is used to learn the characteristics of the single-stranded RNA and ectoin to simulate the activation behavior that occurs during the actual binding process;

[0033] The Kd prediction module is used to predict the continuous variable Kd for RNA and small molecules in cell experiments.

[0034] In some embodiments, the step of using the RiboBindNet model to perform high-throughput screening of saturated RNA mutant libraries and selecting the top N candidate RNA sequences by ranking them from high to low based on their prediction scores includes:

[0035] Obtain a dataset containing RNA sequences, SMILES structural formulas of small molecule compounds, and corresponding Kd tags, where Kd represents the affinity between the RNA molecule and the small molecule measured in the experimental environment.

[0036] Based on the RNA sequence encoding module, the RNA sequence is embedded and encoded using a pre-trained RNA language model, and semantic vector features representing its primary sequence and potential conformation-sensitive regions are extracted to obtain the RNA embedding vector.

[0037] Based on the small molecule structure encoding module, the SMILES expression of the small molecule compound is embedded and encoded using a pre-trained SMILES molecular structure model to obtain a molecular vector representation containing its functional group features and molecular structure topology information, thus obtaining a small molecule embedding vector.

[0038] A neural network prediction model is constructed by inputting the RNA embedding vector and the small molecule embedding vector into a fusion network, and performing Kd regression prediction through a fully connected network layer.

[0039] The neural network prediction model was trained and validated using leave-one-out cross-validation to obtain a prediction model with generalization ability.

[0040] Based on the trained model, multiple different RNA sequences and the SMILES structure of the small molecule compound to be tested are input, Kd prediction values ​​are output in batches, and the prediction results are sorted and filtered to obtain the top N candidate RNA sequences.

[0041] In some embodiments, the RiboBindNet model performs as follows on the test set: the Pearson correlation between the predicted and actual values ​​is 0.53;

[0042] And / or, the base sequence of the RNA sensor includes the sequence shown in SEQ ID NO.1.

[0043] A recombinant vector carrying the base sequence of the RNA biosensor described above;

[0044] Alternatively, the recombinant vector carries the base sequence of the RNA biosensor obtained by the screening method described above.

[0045] A fluorescent detection cell, wherein the fluorescent detection cell carries the recombinant vector described above.

[0046] A method for screening ectoin-producing strains, comprising the following steps:

[0047] Fluorescent detection cells were mixed with the strain to be screened and added to a microfluidic chip for high-throughput screening. The mixture of cells with high fluorescence intensity was screened out, and then the cells were separated to obtain the ectoin-producing strain.

[0048] The fluorescent detection cells carry the base sequence of the RNA biosensor described above, or the base sequence of the RNA biosensor obtained by the screening method described above.

[0049] In some embodiments, prior to the step of mixing the fluorescent detection cells with the strain to be screened, a step of constructing the fluorescent detection cells is included:

[0050] The base sequence and ribozyme sequence of the RNA biosensor are linked together and then inserted into the fluorescent protein gene to obtain the fusion sequence;

[0051] The fusion sequence was digested with enzymes, ligated with ligase, and inserted into an empty vector to obtain a recombinant vector;

[0052] The recombinant vector was transformed into host bacteria to obtain the fluorescent detection cells.

[0053] In some embodiments, at least one of the following features is also included:

[0054] Feature 1: After the step of transforming the recombinant vector into the host bacteria, the following steps are also included: culturing the transformed recombinant bacteria in a culture medium containing a concentration gradient of ectoine, and analyzing the change in fluorescence intensity of the recombinant bacteria at different ectoine concentrations using flow cytometry; if the change in fluorescence intensity of the recombinant bacteria corresponds to the concentration of ectoine, then the recombinant bacteria are the fluorescent detection cells.

[0055] Feature 2: The empty vector is the pCDFduet plasmid, and the host bacterium is Escherichia coli;

[0056] Feature 3: The strain to be screened is a mutated halophilic bacterium;

[0057] Feature 4: The fluorescent protein gene includes the GFP gene, and the base sequence of the RNA biosensor and the ribozyme sequence are linked and inserted into the 5'UTR region of the GFP gene.

[0058] An ectoin-producing strain was obtained by screening using the screening method described above. Attached Figure Description

[0059] Figure 1 This is a schematic diagram illustrating the construction of a plasmid for fluorescent detection of cells in the example.

[0060] Figure 2 This is a flowchart of the microfluidic platform operation in Example 2;

[0061] Figure 3 The graph shows the yield of shake-flask fermentation of Ectoin mutant bacteria in Example 3. Detailed Implementation

[0062] The present application will be further described in detail below with reference to the embodiments and examples. It should be understood that these embodiments and examples are for illustrative purposes only and are not intended to limit the scope of the present application. The purpose of providing these embodiments and examples is to enable a more thorough and comprehensive understanding of the disclosure of the present application. It should also be understood that the present application can be implemented in many different forms and is not limited to the embodiments and examples described herein. Those skilled in the art can make various modifications or alterations without departing from the spirit of the present application, and the equivalent forms obtained also fall within the protection scope of the present application. Furthermore, numerous specific details are set forth in the following description to provide a more complete understanding of the present application. It should be understood that the present application can be implemented without one or more of these details.

[0063] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this application and in its specification is for descriptive purposes only and is not intended to be limiting of the application.

[0064] the term

[0065] Unless otherwise stated or in case of conflict, the terms or phrases used in this application shall have the following meanings:

[0066] The terms "and / or," "or / and," and "and / or" as used in this application encompass any one of two or more of the related listed items, as well as any and all combinations of the related listed items. These arbitrary and all combinations include any two related listed items, any more related listed items, or a combination of all related listed items. It should be noted that when at least three items are connected using at least two conjunctions selected from "and / or," "or / and," and "and / or," it should be understood that in this application, the technical solution undoubtedly includes solutions connected by "logical AND," and also undoubtedly includes solutions connected by "logical OR." For example, "A and / or B" includes three parallel solutions: A, B, and A+B. For example, the technical solution of "A, and / or, B, and / or, C, and / or, D" includes any one of A, B, C, and D (that is, a technical solution that is connected by "logical OR"), as well as any and all combinations of A, B, C, and D, that is, combinations of any two or three of A, B, C, and D, and also combinations of all four of A, B, C, and D (that is, a technical solution that is connected by "logical AND").

[0067] In this application, terms such as "preferred," "better," "more suitable," and "ideal" are merely used to describe implementation methods or embodiments that achieve better results, and should be understood not to limit the scope of protection of this application.

[0068] In this application, terms such as "further," "even further," and "particularly" are used to describe purposes and indicate differences in content, but should not be construed as limiting the scope of protection of this application.

[0069] In this application, "optionally," "optionally," and "optional" mean that something is optional, that is, it means that it is selected from either "with" or "without." If there are multiple "optional" entries in a technical solution, unless otherwise specified, and there are no contradictions or mutual constraints, each "optional" entry shall be independent.

[0070] In this application, the terms "first aspect," "second aspect," "third aspect," "fourth aspect," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or quantity, nor should they be construed as implicitly indicating the importance or quantity of the indicated technical features. Moreover, "first," "second," "third," "fourth," etc., serve only as a non-exhaustive enumeration and should be understood not to constitute a closed limitation on quantity.

[0071] In this application, the technical features described in an open-ended manner include both closed technical solutions consisting of the listed features and open technical solutions that include the listed features.

[0072] In this application, % (w / w) and wt% both represent weight percentage, % (v / v) refers to volume percentage, and % (w / v) refers to mass-volume percentage.

[0073] All references to this application are incorporated herein by reference as if each document were individually incorporated herein by reference. Unless they conflict with the purpose and / or technical solution of this application, all cited references are incorporated herein by reference in their entirety and for all purposes. When references are cited in this application, the definitions of relevant technical features, terms, nouns, phrases, etc., are also incorporated herein by reference. Examples and preferred embodiments of the cited technical features may also be incorporated herein by reference, but only to the extent that they enable the implementation of this application. It should be understood that when the cited content conflicts with the description in this application, this application shall prevail or modifications shall be made adaptably to the description in this application.

[0074] Currently, there is a lack of high-throughput screening methods for high-yield ectoine strains. Traditional screening methods rely on modifying the strains before fermentation, followed by liquid chromatography to detect the ectoine content in the fermentation broth. However, this method involves complex sample processing, long processing time (approximately 10 minutes per sample), high detection costs, and difficulty in achieving 100,000 samples per test. 4 Rapid detection of the above samples. Furthermore, the probability of obtaining high-yielding ectoine-producing strains through natural mutagenesis is extremely low, typically requiring screening of 10... 6 Only by using different mutant strains can high-yield strains be obtained, which further highlights the urgent need to develop rapid, high-throughput, and low-cost detection and screening methods.

[0075] To overcome the above problems, a first aspect of the present application provides an RNA biosensor for screening ectoin-producing strains, wherein the base sequence of the RNA sensor includes the sequence shown in SEQ ID NO.1.

[0076] Specifically, the sequence shown in SEQ ID NO.1:

[0077] GGGAAGGAUCCGUUGGCUUCGGUGAAACGAUCUUACACUUGAGGC UAGAAAUACUCCCGAAAGGCUUGAGCCGUAGGCCUU.

[0078] RNA biosensors are sensors that utilize the properties and functions of RNA molecules to detect specific target substances within organisms or the environment. RNA molecules can specifically bind to target substances such as certain small molecules, ions, and proteins. By designing or screening RNA sequences that can specifically recognize target substances, when the target substance is present, the RNA sequence binds to it, triggering a series of structural or signal changes. These changes can further affect related signal transduction elements, such as fluorophores and enzymes, thereby generating detectable signals, such as fluorescence or chemiluminescence signals. This detection principle based on RNA-specific recognition provides a new approach for high-throughput screening.

[0079] The aforementioned RNA biosensor exhibits high sensitivity and specificity. The fluorescence intensity of the fluorescent detection cells constructed using it increases linearly with ectoin concentration without significant background interference. Combined with microfluidic technology, it can be used for high-throughput screening of ectoin-producing strains.

[0080] A second aspect of this application provides a method for predicting the activation intensity of RNA biosensors and its application in high-throughput virtual screening. The method introduces a sequence-structure dual-modal fusion mechanism based on deep learning to achieve accurate modeling of the activation response capability between single-stranded RNA and target small molecules, thereby assisting in the design and screening of novel RNA biosensors.

[0081] The data-driven modeling approach adopted in this invention combines a pre-trained RNA language model with the SMILES representation learning method to construct a cross-modal prediction framework. This framework can quickly predict the fluorescence activation intensity or functional response intensity of any RNA sequence in the presence of specific small molecules without the need for experimental screening. Furthermore, it obtains the optimal candidate sequence through sorting and screening, significantly improving the efficiency of RNA biosensor development.

[0082] Specifically, the second embodiment of this application provides a method for predicting the intensity of the activation response between single-stranded RNA and small molecule compounds, comprising the following steps:

[0083] Obtain a dataset containing RNA sequences, SMILES structural formulas of small molecule compounds, and corresponding Kd dissociation constant labels, where Kd is the affinity between RNA molecules and small molecules measured in an experimental environment;

[0084] The RNA sequence is embedded and encoded using a pre-trained RNA language model, and semantic vector features representing its primary sequence and potential conformation-sensitive regions are extracted to obtain the RNA embedding vector.

[0085] The SMILES expression of the small molecule compound is embedded and encoded using a pre-trained SMILES molecular structure model to obtain a molecular vector representation containing its functional group features and molecular structure topology information, thus obtaining a small molecule embedding vector.

[0086] A neural network prediction model is constructed by inputting the RNA embedding vector and the small molecule embedding vector into a fusion network, and performing Kd regression prediction through a fully connected network layer.

[0087] The neural network prediction model was trained and validated using leave-one-out cross-validation to obtain a prediction model with generalization ability.

[0088] Based on the trained model, multiple different RNA sequences and the SMILES structure of the small molecule compound to be tested are input, Kd prediction values ​​are output in batches, and the prediction results are sorted and filtered to select the 100 candidate RNA sequences with the highest prediction values.

[0089] The 100 candidate RNA sequences were experimentally screened to obtain the RNA sequence with the highest fluorescence value, which was then used as a biosensor for detecting ectoin.

[0090] The advantage of the aforementioned prediction method lies in its ability to encode the interaction mechanism between RNA sequences and target small molecules into digital feature vectors, and then use deep learning techniques to perform nonlinear mapping modeling, thereby achieving efficient in vitro prediction and candidate screening. Compared to traditional in vivo or in vitro experimental screening procedures, this method features wet experiment independence, high-throughput prediction, and low time cost, making it suitable for various novel RNA biosensor design scenarios, especially for small molecule drug detection and metabolite sensing.

[0091] The virtual screening system built based on the above prediction model, combined with a random RNA sequence generator and a predictive sorting module, can rapidly screen a given small molecule target from millions of RNA sequences. Combined with the subsequently constructed fluorescent detection cells or synthetic biological circuits, it can realize the design and verification of a full-process RNA biosensor.

[0092] The development of traditional biosensors typically relies on natural transcription factors (TFs) that specifically respond to corresponding compounds. Existing transcription factors are then engineered, for example, through saturation mutagenesis and flow cytometry sorting (FACS) to alter their ligand specificity, enabling them to respond to corresponding ligand compounds. However, this modification process is complex, requires extensive screening, has a long development cycle, and is costly, making it difficult to meet the needs of efficient screening.

[0093] To overcome the above problems, a third aspect of this application provides a method for screening RNA biosensors, wherein the RNA biosensors are used to screen ectoin-producing strains, and the screening method includes the following steps S110-S130:

[0094] S110. The RiboBindNet model is used to perform high-throughput screening of the saturated RNA mutant library, and the top N candidate RNA sequences are selected by sorting them from largest to smallest according to the prediction score. The RiboBindNet model is an artificial intelligence model used to predict the binding ability between single-stranded RNA and ectoine in the saturated RNA mutant library.

[0095] S120. Construct corresponding fluorescent detection cells for the first N candidate RNA sequences respectively;

[0096] S130. The fluorescent detection cells corresponding to each candidate RNA sequence are cultured in a culture medium containing ectoine at different concentrations, and the changes in fluorescence intensity of the fluorescent detection cells at different ectoine concentrations are analyzed by flow cytometry and Kd analysis is performed. The RNA biosensor is obtained based on the analysis results.

[0097] This application presents a groundbreaking method for predicting and screening RNA biosensors based on deep neural networks. By combining the programmability of RNA sequence structures with the chemical spatial expression characteristics of small molecules, a cross-modal artificial intelligence model system is constructed. This system can accurately predict the activation response intensity between any RNA sequence and small molecules within a virtual model. Compared to traditional methods that rely on in vivo library construction and wet experimental screening, this approach significantly lowers the screening threshold and improves the design efficiency and scalability of RNA biosensors.

[0098] In some embodiments, the RiboBindNet model includes: an RNA sequence encoding module, a small molecule structure encoding module, a neural network module, and a Kd prediction module.

[0099] The RNA sequence encoding module is constructed based on a pre-trained RNA language model (such as RNA-FM). It uses an improved Transformer architecture to perform context learning on the input single-stranded RNA base sequence and combines optional RNA secondary structure prediction information to extract potential conformation-sensitive regions, fold-driven domains and binding site information, and outputs a set of vector representations characterizing the semantics of RNA structure.

[0100] The small molecule structure encoding module is based on the idea of ​​molecular language modeling. It uses the SMILES Transformer model to embed and encode the SMILES string of the target small molecule, fully learning its atomic arrangement, functional group distribution, and spatial topological features to obtain a continuous molecular representation vector. Optionally, it can be replaced with ChemBERTa, MolBERT, or graph neural networks (such as GAT, MPNN) for graph structure modeling to enhance the structure generalization ability.

[0101] The neural network module is used to perform feature concatenation, weighted fusion, or cross-attention matching between RNA embedding vectors and small molecule embedding vectors. This fusion mechanism supports learning cross-modal recognition features and improves the model's fitting accuracy to the true activation response.

[0102] The Kd prediction module employs a fully connected regression network structure, outputting the predicted Kd value for the corresponding RNA-small molecule combination, reflecting the change in fluorescence expression intensity of the RNA sequence in the presence of the target small molecule in the cellular environment. Optionally, the model can also be extended to a multi-task learning framework, simultaneously outputting the predicted Kd value and binding affinity parameters (such as Kd or EC). 50 This allows for more refined RNA sequence screening.

[0103] The model was trained and tested on publicly available RNA-small molecule activation datasets and simulated datasets. Leave-one-ligand-out cross-validation was used to evaluate the model's generalization ability. The model achieved a Pearson correlation of 0.53 in the Kd prediction task, indicating that it possesses high structure transfer ability and prediction accuracy.

[0104] In virtual screening applications, the model provided in this application can automatically predict 100,000 or more randomly generated RNA sequences in batches after inputting a given small molecule structure, output the corresponding Kd score and sort and screen them, and select high-response RNAs for constructing downstream biosensor modules, significantly shortening the experimental development cycle.

[0105] This application combines AI technology with RNA structure modeling to provide a design method for RNA biosensors that is universal, scalable, and has high throughput capabilities, applicable to multiple application scenarios such as environmental monitoring, drug screening, and anabolic metabolism control.

[0106] In some embodiments, the step of using the RiboBindNet model to perform high-throughput screening of saturated RNA mutant libraries and selecting the top N candidate RNA sequences by ranking them from high to low based on their prediction scores includes:

[0107] Obtain a dataset containing RNA sequences, SMILES structural formulas of small molecule compounds, and corresponding Kd tags, where Kd represents the affinity between the RNA molecule and the small molecule measured in the experimental environment.

[0108] Based on the RNA sequence encoding module, the RNA sequence is embedded and encoded using a pre-trained RNA language model, and semantic vector features representing its primary sequence and potential conformation-sensitive regions are extracted to obtain the RNA embedding vector.

[0109] Based on the small molecule structure encoding module, the SMILES expression of the small molecule compound is embedded and encoded using a pre-trained SMILES molecular structure model to obtain a molecular vector representation containing its functional group features and molecular structure topology information, thus obtaining a small molecule embedding vector.

[0110] A neural network prediction model is constructed by inputting the RNA embedding vector and the small molecule embedding vector into a fusion network, and performing Kd regression prediction through a fully connected network layer.

[0111] The neural network prediction model was trained and validated using leave-one-out cross-validation to obtain a prediction model with generalization ability.

[0112] Based on the trained model, multiple different RNA sequences and the SMILES structure of the small molecule compound to be tested are input, Kd prediction values ​​are output in batches, and the prediction results are sorted and filtered to obtain the top N candidate RNA sequences.

[0113] In one embodiment, N is 100 in the first N candidate RNA sequences. It should be noted that N is not limited to 100, and an appropriate number of candidate RNA sequences can be selected as needed.

[0114] In some embodiments, steps S121-123 of constructing the fluorescent detection cells in S120:

[0115] S121. The base sequence and ribozyme sequence of the candidate RNA sequence are linked together and then inserted into the fluorescent protein gene to obtain the fusion sequence;

[0116] S122. The fusion sequence is digested with enzymes, ligated with ligase, and inserted into an empty vector to obtain a recombinant vector;

[0117] S123. The recombinant vector is transformed into the host bacteria to obtain the fluorescent detection cells.

[0118] The fluorescent protein gene includes the GFP gene, and the candidate RNA sequence and the ribozyme sequence are ligated and inserted into the 5'UTR region of the GFP gene. It should be noted that the fluorescent protein gene is not limited to the GFP gene; it can also be a fluorescent protein gene of other colors, such as a red fluorescent protein gene or a blue fluorescent protein gene.

[0119] The empty vector is the pCDFduet plasmid, and the host bacterium is *Escherichia coli*. Specifically, the host bacterium is *Escherichia coli* MG1655. It should be noted that the empty vector and host bacterium are not limited to the examples mentioned above; suitable empty vectors and host bacteria can be selected as needed.

[0120] In some embodiments, the concentration gradient of ectoine in the culture medium containing the concentration gradient is 0.01 μM, 0.1 μM, 1.0 μM, 10 μM, 100 μM, 1 mM, and 10 mM.

[0121] In some embodiments, a Beckman CytoFLEXS flow cytometer was used to analyze and record the changes in GFP fluorescence intensity at different ectoine concentrations, and a curve showing the relationship between fluorescence intensity and ectoine concentration was plotted. Sensitivity, specificity, changes in fluorescence intensity with ectoine concentration, and the presence of background interference were scored. The single-stranded RNA sequence with the highest score was selected as the desired RNA biosensor. The scoring method was as follows: if the fluorescence intensity and ectoine concentration showed a good linear relationship over a wide range, a linear regression analysis could be used to obtain a linear equation describing the relationship between fluorescence intensity and ectoine concentration, in the form I = mC + b, where I represents fluorescence intensity, C represents ectoine concentration, m represents the slope, and b represents the intercept. The slope m represents the rate of change of fluorescence intensity with increasing ectoine concentration, and the intercept b represents the background fluorescence intensity when the ectoine concentration is 0. The linear regression equation for the single-stranded RNA sequence with the highest score in this application is I = 73C + 6, with a correlation coefficient R² = 0.92, a linear range of 10 μM to 1 mM, and a dynamic range of 100-fold. These values ​​indicate that this RNA biosensor exhibits a very good linear relationship between fluorescence intensity and ectoine concentration within the concentration range of 10 μM to 1 mM, and also possesses high sensitivity and a wide dynamic range.

[0122] Specifically, a sequence with optimal sensitivity and specificity is selected as the desired RNA biosensor. This RNA biosensor corresponds to a sequence in which the fluorescence intensity of the fluorescence detection cells increases linearly with ectoin concentration and there is no significant background interference. In a specific example, the base sequence of the RNA sensor includes the sequence shown in SEQ ID NO.1.

[0123] A fourth aspect of this application provides a recombinant vector carrying the base sequence of the RNA biosensor described above.

[0124] Alternatively, the recombinant vector carries the base sequence of the RNA biosensor obtained by the screening method described above.

[0125] It should be noted that the detailed description of the RNA biosensor and its screening method is provided above and will not be repeated here.

[0126] In some embodiments, the recombinant vector also carries a ribozyme sequence and a fluorescent protein gene.

[0127] Fluorescent protein genes include the GFP gene. It should be noted that fluorescent protein genes are not limited to GFP; they can also be fluorescent protein genes of other colors, such as red fluorescent protein genes or blue fluorescent protein genes.

[0128] In some embodiments, the step of constructing a recombinant vector includes: ligating the base sequence and ribozyme sequence of the RNA biosensor and then inserting them into the fluorescent protein gene to obtain a fusion sequence; digesting the fusion sequence with enzymes and then ligating it with a ligase and inserting it into an empty vector to obtain a recombinant vector.

[0129] The fluorescent protein gene includes the GFP gene, and the base sequence of the RNA biosensor and the ribozyme sequence are linked and inserted into the 5'UTR region of the GFP gene.

[0130] The empty vector is the pCDFduet plasmid, and the host bacterium is *Escherichia coli*. Specifically, the host bacterium is *Escherichia coli* MG1655. It should be noted that the empty vector and host bacterium are not limited to the examples mentioned above; suitable empty vectors and host bacteria can be selected as needed.

[0131] A fifth aspect of this application provides a fluorescent detection cell carrying the recombinant vector described above.

[0132] By constructing this RNA biosensor into fluorescent detection cells, its fluorescence intensity increases linearly with ectoin concentration without significant background interference. Combined with microfluidic technology, it can be used for high-throughput screening of ectoin-producing strains.

[0133] A sixth aspect of this application provides a method for screening ectoin-producing bacterial strains, comprising the following steps:

[0134] Fluorescent detection cells are mixed with the bacterial strain to be screened and added to a microfluidic chip for high-throughput screening. Cell mixtures with high fluorescence intensity are screened out, and then separated to obtain the ectoin-producing bacterial strain. The fluorescent detection cells carry the base sequence of the RNA biosensor described above, or the base sequence of the RNA biosensor obtained by the screening method described above.

[0135] It should be noted that detailed descriptions of RNA biosensors and fluorescence detection cells can be found above, and will not be repeated here.

[0136] Microfluidic screening technology is a modern biotechnology that utilizes microfluidic chips and related technologies for high-throughput, high-precision screening. It combines an aqueous phase containing cells and analytical reagents with oil that isolates individual cell droplets, forming tiny droplets that act as independent microreactors. By precisely manipulating these droplets (such as merging, splitting, and sorting), high-throughput separation and cell recovery are achieved. The key advantage of this technology lies in its ability to detect intracellular and extracellular signals within the droplets, enabling high-throughput screening of cell secretions (such as metabolites and enzymes), providing strong technical support for screening high-yielding bacterial strains.

[0137] The aforementioned screening method combines predicted specific RNA aptamers with fluorescent proteins and cleverly incorporates microfluidic technology, successfully developing a low-cost, high-throughput strain screening method. This method not only significantly improves design efficiency and accuracy, and substantially reduces development costs and time, but also greatly enhances detection performance and adaptability. It can rapidly and efficiently screen high-yielding ectoine strains from a vast number of mutagenic strains, providing strong technical support for the efficient production of ectoine, and demonstrating significant innovation and practicality.

[0138] In some embodiments, prior to the step of mixing the fluorescent detection cells with the strain to be screened, a step of constructing the fluorescent detection cells is included. This step is the same as the steps S121-123 described above for constructing the fluorescent detection cells, as detailed above, and will not be repeated here.

[0139] The procedure, following the step of transforming the recombinant vector into the host bacteria, further includes the following steps: culturing the transformed recombinant bacteria in a culture medium containing a concentration gradient of ectoine, and analyzing the changes in fluorescence intensity of the recombinant bacteria at different ectoine concentrations using flow cytometry. If the change in fluorescence intensity of the recombinant bacteria corresponds to the concentration of ectoine, then the recombinant bacteria are the fluorescent detection cells. Specifically, the concentration gradient of ectoine in the culture medium is 0.01 μM, 0.1 μM, 1.0 μM, 10 μM, 100 μM, 1 mM, and 10 mM.

[0140] In some embodiments, the strain to be screened is a mutated halophilic bacterium. The mutation method is natural mutagenesis or other mutagenesis, such as atmospheric and room temperature plasma mutagenesis (ARTP). It should be noted that the mutation method is not limited to natural mutagenesis; other mutation methods, such as transgenic mutation, can also be used.

[0141] A seventh aspect of this application provides an ectoin-producing bacterial strain obtained by screening using the screening method described above.

[0142] Experimental verification showed that the highest ectoin yield of the ectoin-producing strains screened in this application was 6 g / L.

[0143] In summary, this application innovatively combines RNA biosensors and microfluidic screening technology, and leverages AI prediction to successfully develop a highly efficient method for screening high-yielding naturally induced ectoine mutant strains. The RNA biosensor generated by AI prediction enables rapid and high-throughput sorting of high-yielding strains produced by natural mutagenesis (non-GMO). This method not only avoids the complexity and limitations of traditional synthetic biology gene modification but also fully utilizes the advantages of naturally induced mutant strains to develop non-GMO ectoine products. Simultaneously, this method significantly improves screening efficiency and reduces production costs, providing a more competitive solution for the industrial production of ectoine and potentially driving its further development in various application fields.

[0144] The embodiments of this application will be described in detail below with reference to examples. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of this application. For experimental methods in the following embodiments where specific conditions are not specified, please refer to the guidelines given in this application, or follow experimental manuals or conventional conditions in the art, or follow the conditions recommended by the manufacturer, or refer to experimental methods known in the art.

[0145] In the specific embodiments described below, the measurement parameters involving raw material components may have slight deviations within the weighing accuracy range unless otherwise specified. Temperature and time parameters are subject to acceptable deviations due to instrument testing accuracy or operational precision.

[0146] In the following embodiments, unless otherwise specified, the content involved is as follows:

[0147] 1. Strains:

[0148] In the examples, Escherichia coli MG1655 and Halofilum ochraceum were both purchased from the China Microbial Culture Collection Center.

[0149] 2. Culture medium:

[0150] LB medium: peptone 10 g / L, yeast extract 5 g / L, NaCl 10 g / L.

[0151] Marine agar medium: yeast extract 1.0g, peptone 5.0g, ferric citrate 0.1g, NaCl 89.45g, MgCl2 5.9g, KCl 0.55g, Na2SO4 3.24g, CaCl2 1.80g, Na2CO3 0.16g, KBr 0.08g, SrCl2 34.0mg, H3BO3 22.0mg, NaSiO3 4.0mg, NaF 2.4g, NH4NO3 1.6mg, Na2HPO4 8mg, distilled water 1.0L, pH 7.6.

[0152] 3. Cell construction and detection methods for fluorescence detection:

[0153] The ectoin RNA biosensor sequence and amplified ribozyme sequence were synthesized. PCR amplification was performed using a plasmid containing the GFP gene as a template to obtain the GFP gene. The RNA biosensor-ribozyme sequence was first ligated, and then ligated to the 5'UTR region of the GFP gene. Restriction enzymes EcoRI and XhoI were used for digestion, followed by ligation into the pCDFduet plasmid using T4 DNA ligase (plasmid map shown). Figure 1 As shown in the image, the successful construction of the plasmid was verified by Sanger sequencing. All successfully constructed plasmids were transformed into *E. coli* MG1655 cells. *E. coli* cells that had been correctly verified by PCR were picked from the plate and cultured overnight in LB broth. Then, OD200 was used to further analyze the plasmids. 600The cells were transferred to LB liquid medium at an initial concentration of 0.2 μM and cultured for 4 hours. Then, 0.01 μM, 0.1 μM, 1.0 μM, 10 μM, 100 μM, 1 mM, and 10 mM ectoine were added. An appropriate amount of cell culture medium was taken, diluted with sterile water, and analyzed using a Beckman CytoFLEXS flow cytometer.

[0154] 4. ARTP mutagenesis method

[0155] Take the halophilic bacteria to be mutated and incubate overnight in 5 ml of marine agar medium. The next day, wash twice with 10% glycerol and dilute to a final bacterial concentration of OD. 600 0.6-0.8, mutagenesis was performed using an ARTP mutagenesis breeding instrument with a mutagenesis time of 30s. The mutagenized cells were collected in liquid culture medium for resuscitation culture, and finally diluted and plated onto marine agar medium to obtain single colonies.

[0156] 5. Microfluidic screening methods:

[0157] Fluorescent cells and ARTP-mutated halophilic bacteria cells were diluted to a cell concentration of OD0. 600 Approximately 0.5 μL was added to the microfluidic chip. By adjusting the flow rate, surfactant concentration, and chip design, the optimal conditions were determined to be an oil phase flow rate of 1200 μL / h and an aqueous phase flow rate of 680 μL / h, which could generate droplets with a diameter of 95 μm at a generation rate of 452 droplets per second. The surfactant in the oil phase was Span 80 with a concentration adjusted to 1 wt%, and a standard chip channel height of 50 μm was used. These conditions ensured that the generated droplets were of uniform size, improving the accuracy and repeatability of screening. Finally, the high-throughput screening function of the microfluidic chip was used to screen out cell mixtures with high fluorescence intensity, and strains producing high levels of ectoin were obtained after separation.

[0158] 6. Shake-flask fermentation method:

[0159] The isolated halophilic mutant strain was inoculated into a test tube containing marine agar medium (with 7% NaCl added) and cultured overnight at 37°C to obtain seed culture. 50 mL of marine agar medium was added to a 500 mL shake flask and inoculated at a 10% inoculation rate. The fermentation temperature was 37°C, the shaker speed was 200 rpm, and the fermentation time was 72 hours.

[0160] 7. Detection method for ectoine:

[0161] An Agilent 1290 liquid chromatograph equipped with a photodiode array detector was used. The chromatographic column was a Kinetex F5 (2.6 μm 2.1 mm × 100 mm, Phenomenex). The mobile phase was pure water containing 0.1% (v / v) formic acid. The run time was 5.1 min, with a gradient flow rate of 0.1-0.3 mL / min from 0 to 5 min, and a flow rate of 0.1 mL / min at 5.1 min. The detection wavelength was 210 nm. The column temperature was 20 °C. The injection volume was 1 μL. Quantification was performed using ectoine standards as a standard curve based on peak area.

[0162] Example 1: Construction of Fluorescent Detection Cells

[0163] 1. Using the RiboBindNet artificial intelligence model, a high-throughput screening of saturated RNA mutant libraries was performed, and the top 100 were selected by sorting them from largest to smallest based on their predicted scores for experimental verification.

[0164] The RiboBindNet model is an artificial intelligence model system used to predict the binding affinity between single-stranded RNA (ssRNA) and small molecule compounds. RiboBindNet achieved an ROC-AUC of 0.81 on publicly available databases and experimental validation datasets. The RiboBindNet model includes an RNA sequence encoding module, a small molecule structure encoding module, a neural network module, and a Kd prediction module.

[0165] The RNA sequence encoding module is constructed based on a pre-trained RNA language model (such as RNA-FM). It uses an improved Transformer architecture to perform context learning on the input single-stranded RNA base sequence and combines optional RNA secondary structure prediction information to extract potential conformation-sensitive regions, fold-driven domains and binding site information, and outputs a set of vector representations characterizing the semantics of RNA structure.

[0166] The small molecule structure encoding module is based on the idea of ​​molecular language modeling. It uses the SMILES Transformer model to embed and encode the SMILES string of the target small molecule, fully learning its atomic arrangement, functional group distribution, and spatial topological features to obtain a continuous molecular representation vector. Optionally, it can be replaced with ChemBERTa, MolBERT, or graph neural networks (such as GAT, MPNN) for graph structure modeling to enhance the structure generalization ability.

[0167] The neural network module is used to perform feature concatenation, weighted fusion, or cross-attention matching between RNA embedding vectors and small molecule embedding vectors. This fusion mechanism supports learning cross-modal recognition features and improves the model's fitting accuracy to the true activation response.

[0168] The Kd prediction module employs a fully connected regression network structure, outputting the predicted Kd value for the corresponding RNA-small molecule combination, reflecting the change in fluorescence expression intensity of the RNA sequence in the presence of the target small molecule in the cellular environment. Optionally, the model can also be extended to a multi-task learning framework, simultaneously outputting the predicted Kd value and binding affinity parameters (such as Kd or EC). 50 This allows for more refined RNA sequence screening.

[0169] In the experimental section, the model was trained and tested on both publicly available RNA-small molecule activation datasets and simulated datasets. Leave-one-ligand-out cross-validation was used to evaluate the model's generalization ability. The model achieved a Pearson correlation of 0.53 in the Kd prediction task, indicating that it possesses high structure transfer capability and prediction accuracy.

[0170] In virtual screening applications, the model provided in this application can automatically predict 100,000 or more randomly generated RNA sequences in batches after inputting a given small molecule structure, output the corresponding Kd score and sort and screen them, and select high-response RNAs for constructing downstream biosensor modules, significantly shortening the experimental development cycle.

[0171] 2. Synthesis of predicted sequences and construction and screening of cells for fluorescence detection:

[0172] Based on the RiboBindNet prediction results, a professional DNA synthesis company was commissioned to synthesize the first 100 sequences predicted by Kd. Following the aforementioned method for constructing fluorescent detection cells, these 100 sequences were fused with ribozymes and GFP sequences and ligated into pCDFduet plasmids. After all plasmids were correctly sequenced, they were transformed into *E. coli* MG1655. PCR verification yielded strains JZ001-JZ100. *E. coli* strains JZ001-JZ100 were inoculated into LB medium and cultured overnight, then... 600 Ectocin was initially transferred to LB liquid medium at a concentration of 0.2 μM and cultured for 4 h. Then, 0.01 μM, 0.1 μM, 1.0 μM, 10 μM, 100 μM, 1 mM, and 10 mM ectoin were added. The changes in GFP fluorescence intensity at different ectoin concentrations were recorded using a Beckman CytoFLEXS flow cytometer, and a curve showing the relationship between fluorescence intensity and ectoin concentration was plotted. Based on the scoring principles described above, strain JZ086 showed the best linear relationship between fluorescence intensity and increasing ectoin concentration, and exhibited high sensitivity. Therefore, cells from this strain were selected as the fluorescence detection cells for subsequent screening of mutagenic strains. Sequencing revealed the following RNA biosensor sequence in these cells (sequence shown in SEQ ID NO.1):

[0173] GGGAAGGAUCCGUUGGCUUCGGUGAAACGAUCUUACACUUGAGGC UAGAAAUACUCCCGAAAGGCUUGAGCCGUAGGCCUU.

[0174] The scoring method involves determining the linear relationship between fluorescence intensity and ectoine concentration over a wide range. Linear regression analysis yields a linear equation describing this relationship, in the form I = mC + b, where I represents fluorescence intensity, C represents ectoine concentration, m represents the slope, and b represents the intercept. The slope m represents the rate of change of fluorescence intensity with increasing ectoine concentration, and the intercept b represents the background fluorescence intensity at ectoine concentration of 0. The linear regression equation for the single-stranded RNA sequence with the highest score in this application is I = 73C + 6, with a correlation coefficient R² = 0.92, a linear range of 10 μM–1 mM, and a dynamic range of 100-fold. These values ​​indicate that this RNA biosensor (sequence shown in SEQ ID NO. 1) exhibits a very good linear relationship between fluorescence intensity and ectoine concentration within the concentration range of 10 μM–1 mM, and possesses high sensitivity and a wide dynamic range.

[0175] Example 2: Screening of induced variants using microfluidics

[0176] like Figure 2 ( Figure 2 As shown in the operating procedure of the microfluidic platform, the fluorescent detection cells constructed in Example 1 and the halophilic bacterial cells after ARTP mutagenesis were both diluted to OD. 600 Approximately 0.5 μL of the mixture was added to the microfluidic chip. By adjusting the flow rate, surfactant concentration, and chip design, the optimal conditions were determined to be an oil phase flow rate of 1200 μL / h and an aqueous phase flow rate of 680 μL / h, which could generate droplets with a diameter of 95 μm at a generation rate of 452 droplets per second. The surfactant in the oil phase was Span 80 with a concentration of 1 wt%, and the standard chip channel height was 50 μm. These conditions ensured that the generated droplets were of uniform size, improving the accuracy and repeatability of the screening. Finally, the high-throughput screening function of the microfluidic chip was used to screen out the 10 cell mixtures with the highest fluorescence intensity, and the high-yielding ectoine-producing halophilic strain was obtained after isolation.

[0177] Example 3: Shake-flask fermentation of mutagenic halophilic bacteria

[0178] The halophilic strain isolated in Example 2 was inoculated into test tubes containing marine agar medium and cultured overnight at 37°C to obtain seed culture. 50 mL of marine agar medium was added to a 500 mL shake flask, inoculated at a 10% inoculation rate, and fermented at 37°C and 220 rpm for 72 hours. The ectoin yield of the wild-type strain (WT) and each mutant strain (H1 to H10) was determined. The results are shown below. Figure 3 .

[0179] from Figure 3 It was found that, compared with the wild-type strain, the mutagenic strains screened using microfluidic technology had a 3-12 fold increase in yield, with the highest detected ectoine yield being 6 g / L.

[0180] In summary, this application provides a novel ectoin screening method. An ectoin RNA biosensor is obtained through AI prediction and screening, integrated into a ribozyme sequence, and ligated to the 5'UTR of GFP to construct fluorescent detection cells. Microfluidic technology is then used to screen for high-yielding ectoin strains. Unlike existing genetic engineering methods, the strains obtained in this application do not require complex gene editing processes. The resulting strains and their products possess natural properties, meeting market and consumer demand for natural products and effectively mitigating the market risks associated with genetically modified products. This provides strong support for the market promotion of ectoin products. Furthermore, traditional natural mutagenesis screening methods are inefficient and struggle to obtain ideal strains. This application effectively addresses the pain point of difficulty in obtaining ectoin-producing strains through natural mutagenesis screening. High-throughput screening using AI prediction and microfluidic technology significantly improves screening efficiency and success rate, resolving the safety issues and complex production processes of genetically engineered strains in existing technologies. This method is expected to promote the large-scale application of ectoin in cosmetics, pharmaceuticals, and other fields.

[0181] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0182] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. An RNA biosensor, characterized in that, The base sequence of the RNA sensor for screening an ectoine-producing strain consists of the sequence shown as SEQ ID NO.

1.

2. A recombinant vector, characterized in that, The recombinant vector carries the base sequence of the RNA biosensor according to claim 1.

3. A fluorescent detection cell, characterized by, The fluorescent detection cell carries the recombinant vector according to claim 2.

Images