Engineered bacterial therapeutics from synthetic biology to clinical translation: A multi-database scientometric analysis, 2000–2026 Authors: Hanyu Zhou, Jiaqi He, Jiabin Chen, Shuyu Ge, Ping Yu, Weibo Zhang Published: 2026 ABSTRACT ABSTRACTEngineered bacterial therapeutics integrate programmable strain engineering with biomedical functions such as sensing, delivery, immune modulation, microbiome intervention, and disease management. This study conducted a multi-database bibliometric analysis to map the knowledge structure, collaboration patterns, and translational direction of this field. Bibliographic records were retrieved from Web of Science Core Collection, Scopus, and PubMed for publications from 1 January 2000 to 1 June 2026. After document-type filtering, time restriction, deduplication, language screening, and manual eligibility assessment, 1,730 English-language articles and reviews were included. CiteSpace 6.4.R1, VOSviewer 1.6.20, and the bibliometrix R package were used to analyze publication trends, collaboration networks, journal co-citation, dual-map citation trajectories, citation bursts, keyword co-occurrence, and thematic evolution. Publication output increased steadily, with faster growth after 2020; 2025 was the highest-output complete year, while 2026 represented a partial retrieval year. China and the United States were the leading contributors, although institutional collaboration remained regionally clustered. Co-citation, citation-burst, and keyword analyses showed a shift from bacterial chassis construction and circuit design toward engineered probiotics, gut microbiota-related therapy, cancer immunotherapy, drug delivery, live biotherapeutic products, and clinical translation. Persistent translational priorities include controllability, safety, manufacturing consistency, and regulatory alignment. Introduction Engineered bacterial therapeutics are no longer only a synthetic biology concept. Gene editing, synthetic circuits, and chassis engineering have made it possible to build bacteria that sense disease-associated signals, release therapeutic payloads, or alter local host–microbe interactions. These systems are now being studied in cancer therapy, gut microbiome intervention, infection control, and bacteria-derived delivery platforms.1,2 The literature has also become difficult to synthesize as a single field. One branch focuses on engineered probiotics and live biotherapeutic products.3,4 Other work concerns material-assisted bacterial systems and bacteria-based cancer immunotherapy.5,6 A broader delivery-oriented line treats engineered bacteria as drug-delivery vehicles.7 These directions overlap technically, but they do not always share the same disease settings, safety questions, or translational endpoints. Separate reviews have described several of these areas, but a systematic mapping of the field across research fronts, collaboration networks, and the balance between technical refinement and clinical translation is still lacking. This distinction matters for medical translation. Living bacterial systems are not conventional delivery vehicles. They may colonize, persist, mutate, activate immunity, interact with resident microbes, or be shed into the environment. Their development therefore raises questions about genetic stability, biocontainment, manufacturing consistency, dose definition, and regulatory classification. Recent work on live biotherapeutic delivery and macroencapsulation shows that formulation behavior and in vivo exposure need to be considered together.8,9 Genetic containment studies further indicate that safety control must be designed into the organism rather than added only at the end of development.10 For this reason, the progress of engineered bacterial therapeutics cannot be judged only by the number of new circuits, strains, or proof-of-concept designs. It is also necessary to ask where the literature has accumulated, which countries and institutions shape the field, which journals and disciplines provide its knowledge base, and whether recent work is moving mainly toward technical refinement or clinical testing. We therefore performed a multi-database scientometric analysis using records from Web of Science Core Collection, Scopus, and PubMed for the period 2000 to 2026. After deduplication, language screening, and manual eligibility assessment, 1,730 English-language articles and reviews were included. Publication trends, citation structure, collaboration networks, journal co-citation, dual-map citation trajectories, reference bursts, keyword co-occurrence, thematic evolution, and author-level productivity were analyzed to characterize the development and translational direction of the field. Materials and methods Data sources and search strategy Bibliographic records were retrieved from Web of Science Core Collection (WoSCC), Scopus, and PubMed. The search was completed on June 1, 2026, and covered publications from January 1, 2000 to June 1, 2026. The search fields were adapted to each database, including Topic Search in WoSCC, TITLE-ABS-KEY in Scopus, and Title/Abstract in PubMed. The following search terms were used: (“engineered bacter” OR “genetically engineered bacter” OR “genetically modified bacter” OR “programmable bacter” OR “synthetic bacter” OR “engineered probiotic” OR “synthetic probiotic” OR “programmable probiotic” OR “engineered microbe” OR “programmable microbe” OR “bacterial therapeutic” OR “live biotherapeutic product”) Only English-language articles and reviews were included. Study selection The screening process was conducted as follows. A total of 12,348 records were initially retrieved, including 4,155 from WoSCC, 4,787 from Scopus, and 3,406 from PubMed. We excluded meeting abstracts, proceedings papers, letters, editorial materials, corrections, and retracted publications. Early access records were excluded only when they lacked complete bibliographic metadata or duplicated subsequently indexed records. After document-type and metadata screening, 3,769 records remained in WoSCC, 4,162 in Scopus, and 3,295 in PubMed. After applying the time restriction, 3,559 records remained in WoSCC, 3,917 in Scopus, and 3,014 in PubMed, see Figure 1. Figure 1.Flow diagram of record identification, screening, and inclusion.The flowchart details the process of identifying research through Web of Science, Scopus and PubMed. The database search includes terms like ‘engineered bacter*’, ‘genetically engineered bacter*’, ‘synthetic bacter*’. Initial search results are WOS: 4155, Scopus: 4787, PubMed: 3406. The screening process excludes meeting abstracts, proceeding papers, early access, letters and retracted publications, retaining only articles and reviews. After this, WOS: 3769, Scopus: 4162, PubMed: 3295. The year range filter is 2000.01.01 to 2026.06.01, resulting in WOS: 3559, Scopus: 3917, PubMed: 3014. Duplicate literature is removed and language is filtered to English, leaving 3992 articles. Finally, artificial screening for biomedical and clinical applications and engineering bacteria results in 1730 articles.Flowchart of research identification through Web of Science, Scopus and PubMed. Flow diagram of record identification, screening, and inclusion. Records from the three databases were merged, deduplicated, and screened for language. A total of 3,992 unique English-language records were retained. Two researchers independently screened the titles and abstracts. Studies were included if they focused on the biomedical or clinical applications of engineered bacteria, engineered probiotics, programmable bacteria, or live biotherapeutic products. Disagreements were resolved through discussion with a third researcher. Inter-rater agreement for title/abstract screening was assessed on a random subsample of 400 deduplicated English-language records. The overall agreement was 94.75%, with a Cohen’s kappa of 0.837 (95% CI: 0.77–0.90), indicating strong agreement. All disagreements were resolved through discussion with a third researcher. Finally, 1,730 eligible records were included in the bibliometric analysis. Data processing and analysis Bibliographic information, including titles, authors, affiliations, abstracts, keywords, source journals, publication years, DOI, database source, and cited references where available, was extracted and standardized. Author names, institutional names, country names, and keywords were checked and harmonized before analysis. CiteSpace 6.4.R1, VOSviewer 1.6.20, and the bibliometrix R package were used for bibliometric analysis. CiteSpace was used for co-citation analysis, keyword co-occurrence analysis, clustering, and citation burst detection. The time span was set from 2000 to 2026, with one year per slice. Link strength was calculated using the cosine algorithm, and the top 50 items per slice were selected. The scale factor K was set to 5. Burst detection parameters were set as alpha1/alpha0 = 2.0 and gamma = 1.0. These parameter settings were selected to balance network resolution and interpretability. The Top N = 50 selection per time slice retains the most frequently occurring or cited items within each annual slice while avoiding over-dense networks. The scale factor K = 5 was used to keep the top-tier nodes legible in the generated visualizations. The burst detection parameters (alpha1/alpha0 = 2.0 and gamma = 1.0) were retained as default settings for identifying references or keywords with sustained increases in attention. VOSviewer was used to visualize co-authorship, keyword co-occurrence, and journal citation networks. The bibliometrix R package was used to calculate descriptive bibliometric indicators, including publication trends, citation indicators, and author-level impact metrics such as the h-index and g-index. Publication count was used to describe research output. Citation count was treated as a descriptive indicator of scholarly attention, not as a direct measure of methodological quality. Total link strength represented the cumulative strength of collaboration, citation, or co-occurrence links in network analyses. Citation bursts identified references or keywords that showed a rapid increase in attention during a specific period. In this study, interdisciplinarity was operationally assessed through journal co-citation clustering and dual-map overlay trajectories; the field was considered more interdisciplinary when citing and cited journals were distributed across multiple disciplinary clusters and connected by cross-cluster citation trajectories. Results Publication output and citation trends in engineered bacterial therapeutics Annual publication output on engineered bacterial therapeutics remained limited before 2014, increased gradually in the following years, and accelerated more clearly after 2020. A further rise was observed from 2024 to 2025, making 2025 the highest-output year among complete publication years. The records for 2026 were retrieved only up to June 1, 2026, and should therefore be interpreted as a partial-year estimate. Annual citation counts were also higher in the recent period than in the early period, but the right side of the citation curve should be read descriptively because citations accumulate over time and the 2026 citation window is incomplete (Figure 2). Figure 2.Annual trends in publications and citations.This dual-axis chart shows annual publications (purple bars) and citations (blue line) in engineered bacterial therapeutics from 2000 to 2026. Publications remained low from 2000 to 2010, ranging from 6 to 21 articles per year, with corresponding citations below 540. Growth accelerated after 2015, with publications rising from 43 to 80 by 2018 and citations increasing from 1,157 to 2,272. A sharp upsurge occurred from 2020 onward: publications climbed from 94 to a peak of 250 in 2025, while citations grew from 3,888 to 12,189 in 2025. In 2026, both metrics declined to 177 publications and 6,147 citations, reflecting the partial-year data cutoff (June 1, 2026). Overall, the field expanded nearly 23-fold in annual output and over 2,000-fold in citations between 2000 and the 2025 peak.A combined bar and line graph showing annual publications and citations by year.Annual publication and citation trends in engineered bacterial therapeutics from 2000 to 2026. Bars show the number of publications per year, and the blue line shows total citations received by publications from each year. Data for 2026 cover publications retrieved up to June 1, 2026. Annual trends in publications and citations. Supplementary Table S1 summarizes descriptive bibliometric indicators for the complete dataset. The corpus contains 1,730 documents from 543 sources, involving 8,073 authors and 97,633 cited references. The annual growth rate is 11.28%, international co-authored publications account for 20.64% of the dataset, and the average citation count per document is 36.67. These citation-based indicators are reported as measures of scholarly attention rather than as direct measures of research quality. Global research distribution and international collaboration At the country level, China and the United States formed the two largest nodes in the collaboration network, indicating their leading publication output in this field. Visible collaboration links connected these two countries with multiple partners across North America, Europe, and the Asia-Pacific region. European countries, including Germany, Italy, Spain, and the United Kingdom, appeared as a relatively compact regional group, while China and the United States were connected with several country groups rather than being confined to a single regional cluster, see Figure 3. Figure 3.Global distribution and collaboration network of publications in engineered bacterial therapeutics.A world map illustrates the global distribution and collaboration network of publications in engineered bacterial therapeutics. The map features nodes representing countries, with China and the United States as the largest nodes, indicating their leading publication output. Collaboration links connect these countries with partners across North America, Europe, the Middle East and the Asia-Pacific region. European countries like Germany, Denmark, Italy, Spain and the United Kingdom, Ireland and Finland form a compact regional group. The map includes a legend for article count, with circles indicating 600 and 200 articles and a cluster legend with five categories: 1 (green), 2 (blue), 3 (orange), 4 (purple) and 5 (red). Clusters are distributed globally, with visible connections between countries such as Canada, Mexico, Brazil, Chile, Russia, Turkey, Iran, India, Singapore, Australia and New Zealand. The map highlights the extensive international collaboration in this scientific field.A world map showing global collaboration in engineered bacterial therapeutics publications.Node size represents the number of publications from each country, and connecting lines show international collaborations. Different colors denote collaboration clusters identified by network analysis. Global distribution and collaboration network of publications in engineered bacterial therapeutics. Temporal evolution of national research output Cumulative national publication output revealed a marked concentration of research activity in two leading countries. The United States remained the dominant contributor for most of the observation period, maintaining the highest cumulative output curve. China showed a comparatively lower cumulative output before 2018, but its growth accelerated notably after 2020. By the retrieval endpoint in 2026, China’s cumulative output had approached that of the United States, see Figure 4. Figure 4.Temporal trends in publication output among leading countries in engineered bacterial therapeutics.A line graph titled, Country Production over Time. The horizontal axis label is, Year, with values from 2000 to 2023. The vertical axis label is, Articles, with labeled ticks at 0, 500, 1000 and 1500. A legend titled, Country, lists: China, France, Germany, India, Italy, Korea, Singapore, Spain, United Kingdom, USA. Multiple country lines rise over time. Two lines dominate: one increases steadily from near 0 around 2000 to above 1500 by 2023; another stays low until about 2020 then rises sharply to near the same level by 2023. Other country lines remain below roughly 300 articles by 2023, with gradual increases after about 2015 and more noticeable rises after about 2020. No individual point markers with numeric labels are shown, so coordinate pairs are not readable from the plotted lines.A line graph showing country production over time measured by articles.The curves show cumulative publication output from the major contributing countries between 2000 and 2026. The 2026 endpoint corresponds to the retrieval date of June 1, 2026. Temporal trends in publication output among leading countries in engineered bacterial therapeutics. Other productive countries exhibited more moderate growth trajectories. Germany, France, Italy, Spain, and the United Kingdom represented the main European contributors, whereas India, South Korea, and Singapore constituted important contributors from Asia outside China. Although these countries showed steady increases over time, their cumulative outputs remained substantially below those of China and the United States throughout the study period. Temporal productivity of leading authors Among the leading authors shown in the timeline, publication activity was unevenly distributed across time, with some contributors appearing repeatedly across multiple years. Danino T, for example, appeared repeatedly from approximately 2015 onward, with visible publication output and citation accumulation in several years. Several other author nodes became more prominent after 2020, consistent with the recent expansion of the field (Figure 5). Because author names were standardized mainly by surname and initials, these patterns should be interpreted as descriptive author-level trends rather than precise individual-level rankings. Figure 5.Temporal distribution of publication output among leading authors.Authors′ Production over Time. The horizontal axis label is, Year, with tick labels 2001, 2003, 2005, 2007, 2009, 2011, 2013, 2015, 2017, 2019, 2021, 2023, 2025. The vertical axis label is, Author, listing LINDSEY JS, LI Y, WANG Y, DANINO T, HOLTEN D, ZHANG Y, WANG Q, BOCIAN DF, CHANG MW, LIU Y. Each author has a horizontal timeline with circular markers placed at specific years. Bubble size represents Number of Articles (labeled 2.5, 5.0, 7.5), while color intensity represents Times Cited per Year (labeled 0, 20, 40, 60). The plotted timelines show repeated markers across multiple years for several authors, with denser sequences of markers appearing from about 2015 through 2025 for multiple author rows and multiple markers also appearing around 2009 through 2015 for some author rows.A bubble chart showing authors′ production over time with bubble size for number of articles and TC per year.The size of each bubble represents the number of publications in a given year. Color intensity represents Times Cited per Year for publications from a given year and should not be used as a year-normalized quality comparison. Horizontal lines show the active publication period of each author. Temporal distribution of publication output among leading authors. Institutional collaboration network Institutional collaboration was organized around several major clusters rather than a broadly even network. The most prominent cluster was formed by high-output Chinese institutions, led by the Chinese Academy of Sciences and the University of Chinese Academy of Sciences, with several universities connected within the same group. A second visible cluster included major United States institutions such as Massachusetts Institute of Technology and Harvard University. Smaller European and Asia-Pacific institutional groups were also present, but their links were less dense than those of the largest clusters. The visually prominent nodes, including the Chinese Academy of Sciences, University of Chinese Academy of Sciences, Massachusetts Institute of Technology, University of California Berkeley, and Shanghai University Tong University, suggested that collaboration was concentrated around a limited number of productive institutions. Peripheral institutions had fewer connections, indicating that the institutional network remained cluster-based rather than uniformly integrated (Figure 6). Figure 6.Institutional collaboration network in engineered bacterial therapeutics.The network visualization depicts institutional collaboration in engineered bacterial therapeutics. No axes or units apply. Node size represents collaboration volume, with larger nodes indicating more publications or partnerships. Links denote co-authorship or collaboration, and colors signify distinct clusters or communities. Prominent hubs include CHINESE ACAD SCI, TIANJIN UNIV, MIT, HARVARD UNIV, and WASHINGTON UNIV. The network is dominated by a large China-centered cluster (green) and multiple US-centered clusters (red and purple), with limited inter-cluster links. The purple cluster includes WASHINGTON UNIV, N CAROLINA STATE UNIV, and MASSACHUSETTS GEN HOSP, while the red cluster centers on MIT, HARVARD UNIV, and COLUMBIA UNIV. Smaller international clusters (blue and yellow) are also visible. The visualization highlights concentrated regional collaboration patterns with relatively sparse cross-cluster connections.Visualizing institutional collaboration in bacterial therapeutics, focusing on clustered hubs.Node size represents the relative contribution of each institution, and connecting lines show collaborative relationships between institutions. Different colors represent collaboration clusters identified by VOSviewer. Institutional collaboration network in engineered bacterial therapeutics. Journal co-citation network and disciplinary structure The journal co-citation network outlined the disciplinary structure underlying engineered bacterial therapeutics. General science and high-impact biomedical journals, including Nature, Science, Cell, Proceedings of the National Academy of Sciences of the United States of America, Nature Communications, and Nature Biotechnology, occupied central positions, indicating that the field draws on broadly influential biological and translational research. The remaining network was organized into several thematic clusters. One cluster reflected the microbiological foundation of the field, with representative journals such as Journal of Bacteriology, Molecular Microbiology, and Applied and Environmental Microbiology. A second cluster was associated with host-microbe interactions, immunology, and cancer biology, represented by Cell Host & Microbe, Gut Microbes, Gastroenterology, Frontiers in Immunology, and Cancer Research. A third cluster linked synthetic biology, biomaterials, and drug delivery, with journals such as ACS Synthetic Biology, Advanced Materials, Biomaterials, and Journal of Controlled Release, see Figure 7. Figure 7.Journal co-citation network in engineered bacterial therapeutics.A VOSviewer bibliometric visualization shows journals as labeled circles connected by many curved lines, indicating how often they are cited together. Larger circles represent more prominent journals, with Nature, Science, Cell, PLOS ONE, Nature Communications and PNAS near the center. Distinct color clusters suggest topical groupings: a red cluster around immunology, gut microbiome and cancer-related titles (e.g., Immunity, Gut, Cancer Research); a green/yellow cluster emphasizing microbiology and environmental/biotechnology journals (e.g., Journal of Bacteriology, Applied and Environmental Microbiology, Biotechnology and Biofuels); and a blue cluster featuring materials, methods and engineering outlets (e.g., Biomaterials, ACS Nano); and a light blue cluster focused on clinical infection and antimicrobial research (e.g., Clin Infect Dis, Antimicrob Agents Ch). Dense cross-links between clusters highlight strong interdisciplinary overlap.Color-coded map of journals linked by shared references, with Nature and PNAS as major hubs.Node size represents journal co-citation frequency. Links represent co-citation relationships, and colors show journal clusters identified by network analysis. Journal co-citation network in engineered bacterial therapeutics. Dual-map overlay of journal citation trajectories The dual-map overlay showed a citation structure centered on biomedical and molecular research. The strongest trajectories originated from journals in molecular biology, immunology, and clinical medicine, and mainly pointed to cited journals in molecular biology, genetics, chemistry, and materials science. This pattern indicates that engineered bacterial therapeutics are reported largely in biomedical and translational journals, while their knowledge base remains closely linked to genetic engineering, microbial biology, and enabling chemical or materials technologies. The most prominent paths emphasized biological mechanisms and programmable strain design as the core foundation of the field. Additional trajectories toward chemistry, materials science, and physics-related clusters reflected the role of biomaterials, nanodelivery, imaging, and externally controlled therapeutic systems. Thinner links to systems science, environmental toxicology, ecology, and broader health-science clusters were also visible, suggesting secondary but potentially growing connections as computational design, ecological safety assessment, and microbiome-based intervention develop further (Figure 8). Figure 8.Dual-map overlay of journal citation trajectories in engineered bacterial therapeutics.An infographic displays citation flow between journal subject clusters, arranged left to right. Left clusters include: 1. MATHEMATICS, SYSTEMS, MATHEMATICAL; 2. MEDICINE, MEDICAL, CLINICAL; 3. ECOLOGY, EARTH, MARINE; 4. MOLECULAR, BIOLOGY, IMMUNOLOGY; 5. PHYSICS, MATERIALS, CHEMISTRY; 6. PSYCHOLOGY, EDUCATION, HEALTH; 7. VETERINARY, ANIMAL, SCIENCE; 9. MOLECULAR, BIOLOGY, IMMUNOLOGY; 10. ECONOMICS, ECONOMIC, POLITICAL. Right clusters include: 8. MOLECULAR, BIOLOGY, GENETICS; 5. HEALTH, NURSING, MEDICINE; 7. PSYCHOLOGY, EDUCATION, SOCIAL; 12. ECONOMICS, ECONOMIC, POLITICAL; 13. REVISTA, PSICOLOGIA, SAUDE; 18. HISTORY, PHILOSOPHY, RECORDS; 10. PLANT, ECOLOGY, ZOOLOGY. Curved bands connect clusters, with prominent links from MATHEMATICS, SYSTEMS, MATHEMATICAL and MOLECULAR, BIOLOGY, IMMUNOLOGY to MOLECULAR, BIOLOGY, GENETICS and others to HEALTH, NURSING, MEDICINE and PLANT, ECOLOGY, ZOOLOGY.An infographic dual-map citation flow map for engineered bacterial therapeutics journals.The left side represents citing journals, and the right side represents cited journals. Citation curves show knowledge flows between disciplinary journal clusters. Thicker curves represent stronger citation relationships. Dual-map overlay of journal citation trajectories in engineered bacterial therapeutics. Citation bursts and emerging research fronts The strongest citation bursts appeared predominantly after 2013 and unfolded in successive waves. Early- to mid-period bursts centered on foundational work in programmable bacterial sensing, lysis-based delivery, gut colonization, live diagnostics, and engineered probiotics (e.g., Kotula et al., Danino et al., Din et al., Riglar et al., Hwang et al., Isabella et al., Mimee et al.). More recent bursts, extending toward the 2026 endpoint, shifted toward cancer-targeted bacterial therapy and programmable living therapeutic systems (e.g., Gurbatri et al., Vincent et al., Ma et al., Savage et al.). This temporal pattern traces a progression from early proof-of-concept bacterial engineering toward disease-oriented, translational research (Figure 9). Figure 9.Top 25 references with the strongest citation bursts in engineered bacterial therapeutics.An infographic titled ′Top 25 References with the Strongest Citation Bursts′ features a table listing references with columns for Year, Strength, Begin, End and a timeline from 2000 to 2026. Each row highlights the burst period between Begin and End years. Notable entries include: Bruckner C (2012, Strength 9.92, 2013-2017), Kotula JW (2014, Strength 15.64, 2015-2019), Danino T (2015, Strength 17.13, 2016-2020), Riglar DT (2017, Strength 19.14, 2017-2022), Hwang IY (2017, Strength 18.74, 2017-2022), Din MO (2016, Strength 13.95, 2017-2021), Daeffler KNM (2017, Strength 11.92, 2017-2022), Duan FF (2015, Strength 10.15, 2017-2020), Mimee M (2015, Strength 9.02, 2017-2020), Isabella VM (2018, Strength 21.35, 2019-2023) and Savage TM (2023, Strength 9.02, 2024-2026).An infographic ranks top 25 references by citation burst strength, with timelines from 2000 to 2026.The red bars mark the period during which each reference shows a strong citation burst, and the strength value gives the intensity of the burst. References with red bars extending to 2026 correspond to the most recent ongoing burst periods within the dataset. Top 25 references with the strongest citation bursts in engineered bacterial therapeutics. Keyword co-occurrence and thematic evolution Keyword co-occurrence analysis revealed a network organized around a small set of high-frequency, highly connected terms. Escherichia coli had the highest occurrence and synthetic biology had the strongest total link strength; other central nodes, like expression, engineered bacteria, gut microbiota, probiotic, delivery, immunotherapy, and cancer, bridged the engineering backbone and the biomedical application clusters. This structure positions chassis design, functional expression, and delivery technologies as the main organizing axes of the literature. The temporal overlay captured a clear shift in research vocabulary. Earlier topics clustered around E. coli-based expression, protein production, metabolic engineering, and fermentation. More recent terms converged on gut microbiota, engineered probiotics, live biotherapeutic products, immunotherapy, the tumor microenvironment, and delivery modalities such as nanoparticles and ultrasound-responsive strategies. The trajectory traces a movement from production-centered engineering toward disease-focused translational research, while E. coli and synthetic biology remain the durable technical foundations (Figure 10). Figure 10.Temporal overlay visualization of keyword co-occurrence in engineered bacterial therapeutics.A VOSviewer-style network map displays research keywords as labeled nodes connected by many thin lines, with node size indicating prominence and color indicating recency on a 2018 to 2024 gradient (blue/green earlier to orange/red later). Large central nodes include Engineered Bacteria, Synthetic Biology, Escherichia Coli, Gene Expression and Gut Microbiota, linking multiple thematic clusters. On the right, a mostly blue-green cluster focuses on biosynthesis, fermentation, metabolic engineering, pathways and organisms such as Bacillus subtilis, yeast and Pseudomonas putida. The center emphasizes genetic tools and functions (CRISPR-Cas9, circuits, transcription, protein expression). On the left, warmer-colored clusters highlight biomedical applications immunotherapy, cancer therapy, tumor microenvironment, drug/oral delivery, probiotics, inflammatory bowel disease and specific strains like Salmonella typhimurium and E. coli Nissle 1917. A horizontal legend at bottom right shows the year scale from 2018 to 2024.Color-coded keyword map showing clusters around engineered bacteria, E. coli, gut microbiota and cancer.Node size represents keyword occurrence frequency, and connecting lines show co-occurrence relationships between keywords. The color gradient shows the average publication year of each keyword, with earlier topics shown in blue or purple and more recent topics shown in yellow, orange, or red. Temporal overlay visualization of keyword co-occurrence in engineered bacterial therapeutics. Timeline visualization of keyword clusters The keyword timeline revealed a field that has preserved a strong technical core while progressively developing application-specific clusters. Early high-frequency terms centered on Escherichia coli, expression, metabolic engineering, microbial production, and biosynthesis, mapping onto clusters such as synthetic biology, E. coli, and microbial production. Synthetic biology persisted as an organizing theme throughout the timeline, bridging early strain construction and expression systems with later therapeutic design. A clinically oriented shift became apparent from the mid-2010s onward, when drug delivery, engineered bacteria, gut microbiota, and resistance-related terms appeared alongside emerging therapeutic clusters. In the most recent segment, terms extending toward the 2026 endpoint included live biotherapeutic products, immunoengineering, inflammatory bowel disease, tumor microenvironment, cancer immunotherapy, phage therapy, and clinical trial vocabulary such as phase I, double-blind, and immune checkpoint inhibitors. These temporal patterns suggest that the field has not moved away from its engineering foundation; rather, it has expanded from chassis construction and microbial production into delivery, host–microbe intervention, and translational therapeutic evaluation (Figure 11). Figure 11.Timeline visualization of keyword clusters in engineered bacterial therapeutics.The timeline visualization shows keyword clusters in engineered bacterial therapeutics from 2000 to 2026. The x-axis represents years, while clusters are arranged vertically along the y-axis, with nodes spreading horizontally across the timeline. Nodes vary in size and are connected by curved lines, grouped by cluster with a color scale. Clusters are numbered 0 to 10: Synthetic Biology, Saccharomyces Cerevisiae, Drug Delivery, Escherichia Coli, Immunoengineering, Metabolic Engineering, Live Biotherapeutic Products, Microbial Production, Engineered Bacteria, Resistance and Phage Therapy. Early terms (2000-2005) include escherichia coli, cloning and gene expression. Mid timeline terms feature metabolic engineering and synthetic biology. Later terms (2015-2026) include gut microbiota, probiotic, drug delivery, cancer immunotherapy and clinical trials, with many links converging towards 2025-2026.A timeline network plot of keyword clusters in engineered bacterial therapeutics, shifting from 2000 to 2026.Each horizontal line represents a keyword cluster identified by CiteSpace, with cluster labels shown on the right. Nodes indicate keywords appearing in different years, and larger nodes represent higher frequency or greater network prominence. Links show co-occurrence relationships between keywords. The timeline displays the persistence, overlap, and temporal extension of major research themes from 2000 to 2026. Timeline visualization of keyword clusters in engineered bacterial therapeutics. Discussion Bibliometric analysis reveals a discipline evolving from its foundations in chassis construction, circuit design, and microbial production toward a translational focus on host-facing functions, disease-specific delivery, and clinical testing. The vocabulary of the field changed along the same timeline. Early work was dominated by terms such as gene expression, metabolic engineering, biosynthesis, chassis construction, and circuit design. More recent work more often involved engineered probiotics, gut microbiota intervention, cancer immunotherapy, drug delivery, tumor-targeted bacterial systems, programmable living medicines, and live biotherapeutic products. Strain engineering remains the technical base of the field, but the standard for a successful design has become more demanding: the organism is now expected to produce a measurable function in a host, not only under culture conditions. CRISPRi and transcriptional-control methods have added finer control over bacterial gene expression.11,12 CRISPR-associated transposase systems have improved genome engineering precision.13 Biological-containment designs have tightened safety controls, while resistance-gene targeting has moved CRISPR-based systems into antimicrobial applications.10,14 CRISPR-based biosensing further extends these tools into diagnostic settings.15 Disease-responsive probiotic circuits and nanobody-secreting commensals illustrate the move toward host-facing functions.16,17 Temperature-controlled expression systems set a higher bar, because inserting a genetic module is no longer sufficient unless timing, stringency, and anatomical location can also be controlled.18 The co-citation and dual-map overlays reinforce this picture. Engineered bacterial therapeutics now draws on several disciplines, not microbiology alone. Microbiology and molecular biology still underpin chassis construction; immunology, oncology, materials science, synthetic biology, and drug delivery now shape how the field handles localization, release, and exposure control. The remaining difficulty is no longer limited to engineering bacteria. The field still lacks shared frameworks for comparing in vivo exposure, persistence, payload output, host clearance, and ecological risk across different strains and delivery routes. A major part of this gap lies in delivery engineering. A genetic circuit can specify an intended function, but it does not determine where bacteria persist, when payloads are released, or how quickly the host clears the organism. Tumor-delivery studies using hyperbaric oxygen or hypoxia-activated bacterial systems demonstrate that local physiological conditions can guide bacterial activity.19,20 Several platform-level analyses have argued that engineered bacteria should be characterized by their exposure, localization, release, and clearance profiles, much like a drug delivery system.21 In solid tumors, controlling bacterial localization and intratumoral penetration is the primary challenge. At mucosal surfaces, transit, mucus, immune tone, and resident microbiota shape exposure.22,23 Oral delivery studies further emphasize that route, secretion strategy, and gastrointestinal survival belong to therapeutic design, not downstream formulation.24,25 Delivery constraints therefore need to be built into strain and formulation design from the start. CRISPR-based transcriptional control and metabolic engineering add another layer by tuning strain behavior and functional output.26,27 Work on commensal nanobody deposition and optimized bacterial nucleic-acid delivery similarly shows that therapeutic effect depends on local molecular release rather than bacterial presence alone.28,29 Non-oncological bacterial drug-delivery studies extend the same principle to skin, respiratory, gastrointestinal, and urogenital sites.30 The citation-burst data provide a timeline of this shift. Early burst references asked whether engineered bacteria could perform a defined function inside a living host. Kotula and colleagues showed that engineered E. coli could record environmental exposure in the mammalian gut; Din and colleagues used synchronized lysis to release payloads in vivo.31,32 Riglar and colleagues extended this approach to long-term live diagnostics of intestinal inflammation,33 and Hwang and colleagues demonstrated that engineered probiotic E. coli could suppress Pseudomonas aeruginosa gut infection in animal models.34 Subsequently, engineered E. coli Nissle was used to metabolize phenylalanine in the setting of phenylketonuria, while ingestible bacterial-electronic systems combined sensor bacteria with electronic readouts for gastrointestinal monitoring.35,36 Biosensing-oriented probiotic studies later extended the same approach to diagnostic platforms.37 In this part of the literature, in vivo function is no longer treated as a sufficient endpoint by itself; it also needs to be durable, quantifiable, and connected to a therapeutic or diagnostic readout. In the keyword timeline and citation-burst data, cancer therapy stood out as one major application-facing branch. Bacterial systems can exploit hypoxic, metabolically abnormal, and immune-suppressed niches within tumors. Cancer-focused reviews describe a transition from passive bacterial tropism toward actively engineered bacterial and nano-bacterial systems.38,39 Bacterial nanotechnology studies extend this direction toward targeted therapeutic delivery and immunotherapy.40 A design-oriented review of live bacterial cancer therapeutics treats strain selection, delivery route, payload control, and safety gating as components of a single development problem.41 Earlier oncology studies used bacteria to colonize tumors, release payloads, or remodel the tumor microenvironment.42,43 Short-lived and glucose-sensing bacterial systems further illustrate how bacterial behavior can be programmed for tumor control.44 Later immunotherapy work moved toward immune-circuit control and microenvironmental reprogramming, including IL-10R-based strategies45 and IFN-β-expressing VNP20009 systems.46,47 Ultrasound-controlled bacterial systems then demonstrated that pharmacological output could be triggered from outside the body. Abedi and colleagues built a temperature-actuated genetic state switch activated by focused ultrasound48; Chen and colleagues used focused ultrasound to induce IFN-γ expression from engineered bacteria in tumor models.49 These studies move the problem beyond tumor access. For cancer applications, a more demanding question is whether colonization can be separated from pharmacological output and controlled externally. Cancer immunotherapy is likewise moving toward defined immune targets. Redenti and colleagues engineered probiotic E. coli Nissle 1917 as a neoantigen delivery vector, optimizing the strain for neoepitope production, cytosolic delivery, and safety via susceptibility to blood clearance and phagocytosis.50 In tumor models, this strategy promoted dendritic-cell activation and neoantigen-specific T-cell responses. The work illustrates a shift toward antigen-specific immune programming, where the relevant output is dendritic-cell activation and T-cell specificity, not tumor colonization alone. The main challenge is to coordinate colonization, payload output, immune stimulation, and bacterial clearance at the right site and within a usable time window. A system that is potent but poorly controlled can increase inflammatory risk without producing reliable clinical benefit, especially in patients with compromised immunity or heavy pretreatment. A different set of challenges appears in gut-directed engineered bacteria. These systems enter a microbial ecosystem rather than a localized tumor niche, and colonization, metabolic output, immune regulation, and pathogen suppression all affect therapeutic performance. Current gut-directed studies use engineered bacteria for at least three purposes: local immune regulation, metabolic output, and selective pathogen suppression. IFNL1-producing probiotics and orally delivered programmable probiotics represent inflammation-oriented strategies.51,52 Butyrate-producing and GLP-1-producing strains extend the platform toward metabolic and cardiovascular phenotypes.53,54 Microcin MccI47-producing live biotherapeutics represent a targeted antimicrobial route against carbapenem-resistant Klebsiella pneumoniae colonization.55 These examples argue against treating gut-directed engineered bacteria as a single category. Immune modulation, metabolic remodeling, and pathogen suppression require distinct endpoints and safety readouts. Gut-directed work also extends outside inflammatory bowel disease and infection. Studies on microbiome-preserving secretion systems and engineered Bacillus subtilis underscore the importance of ecological compatibility.56,57 Work on probiotic Escherichia coli Nissle 1917 and dual engineered bacterial systems makes the same issue more concrete in inflammatory disease models.58,59 Broader reviews of genetically engineered bacteria for inflammatory bowel disease also point to the need for indication-specific endpoints and safety assessment.60 Recent gut-brain axis studies have begun to connect engineered probiotics with distal host modulation. Wang and colleagues developed an alternating magnetic field-responsive engineered probiotic system for anxiety therapy through gut-brain axis modulation.61 This design combines engineered probiotics, external control, and distal host modulation in a single preclinical study. For neurological indications, however, behavioral specificity, long-term exposure, reversibility, and safety are harder to establish than local intestinal endpoints, so these results should still be interpreted with caution. Vesicle and biomaterial platforms offer a different strategy: they retain selected bacterial functions while avoiding sustained bacterial replication. Outer membrane vesicles can carry microbial antigens, immune-stimulatory signals, or therapeutic payloads without requiring long-term bacterial persistence.62 Extracellular vesicle studies similarly support bacterial vesicles as biologically active carriers rather than incidental bacterial by-products.63,64 Diagnostic studies also support bacterial extracellular vesicles as measurable biological carriers.65 Engineered vesicle systems have been explored for tumor vaccination, antibody delivery, doxorubicin and CD47-siRNA co-delivery, and postoperative immune checkpoint therapy.66,67 Compared with replicating live organisms, vesicle-based systems may be easier to define in terms of dose, composition, storage, and batch release. Their limitation is that they lack the self-renewing, environment-responsive behavior of live therapeutics. Material-assisted biohybrids are distinct from vesicle-based systems. Magnetotactic bacteria and gold nanoparticle-carrying bacteria aim to make bacterial activity more localized, measurable, or externally controllable.68,69 Hydrogel-based living systems extend this strategy to engineered living bacterial therapeutics.70 Gradient microbiome models and bacterial cellulose platforms further move this direction into microenvironmental modeling and biomaterial construction.71,72 Nanocomposite wound materials show how bacterial cellulose-based systems may enter translation as wound-repair materials.73,74 Cardiovascular tissue-engineering studies extend bacterial cellulose applications into another material-based route.75 Purified bacterial biomaterials mainly face questions of reproducibility, sterilization, degradation, and biocompatibility. Genetically modified live therapeutics must also address persistence, shedding, mutation, horizontal gene transfer, and ecological containment. Clinical-stage programs provide a practical reference point for interpreting these bibliometric patterns. Several bacterial or live microbial therapeutic platforms have reached clinical or translational testing, as summarized in Table 1. Their outcomes have been mixed. Oncology-oriented programs such as SYNB1891 and VNP20009 show that immune activation, local and systemic administration, and local bacterial activity are achievable, but dose control, inflammatory toxicity, and antitumor efficacy remain limiting.76,77 Intratumoral Clostridium novyi-NT testing further illustrates the balance between local bacterial activity and inflammatory toxicity.78 Other platforms, including CRS-207 and SYNB1020, demonstrate that biological plausibility and local activity do not automatically translate into durable clinical benefit.79,80 AG013 and earlier IL-10-secreting Lactococcus lactis studies similarly illustrate the gap between local feasibility and sustained clinical efficacy.81,82 These experiences reinforce a practical point: engineered bacteria need to be defined by their pharmacological characteristics – exposure, potency, clearance, genetic stability, manufacturability, and batch-release specifications – before clinical endpoints can be reliably tested, see Table 1.Table 1.Representative clinical-stage or translationally relevant bacterial and live microbial therapeutic programs.Program/platformDesign/modalityClinical settingHuman or translational signalRef.SYNB1891 (Escherichia coli Nissle 1917)Intratumoral EcN producing STING-activating cyclic dinucleotidesAdvanced solid tumors or lymphomaPhase I testing showed immune target engagement and manageable early dosing, while cytokine-related toxicity required dose optimization.76VNP20009 (Salmonella Typhimurium)Attenuated tumor-colonizing strainMetastatic melanoma and renal cell carcinomaAdministration and some tumor colonization were feasible, but objective tumor regression was not observed.77Clostridium novyi-NT sporesObligate anaerobic spores that germinate in hypoxic tumor regionsTreatment-refractory solid tumorsEarly intratumoral testing supported local bacterial activity, while inflammatory toxicity and delivery control remained limiting issues.78CRS-207 (Listeria monocytogenes)Live-attenuated vector expressing mesothelinPancreatic and other mesothelin-positive malignanciesAntigen-directed delivery was biologically plausible, but durable survival benefit was not consistently established.79SYNB1020 (Escherichia coli Nissle 1917)Ammonia-consuming engineered EcNHyperammonemiaEarly testing indicated tolerability and biological activity, but the metabolic output was insufficient for clinical development.80AG013 (Lactococcus lactis)Topical Lactococcus lactis secreting human trefoil factor 1Oral mucositis during cancer therapyLocal mucosal protein delivery was feasible; later efficacy assessment showed the difficulty of meeting a clinical mucosal endpoint.81LL-Thy12/IL-10-secreting Lactococcus lactisBiologically contained strain expressing human interleukin-10Crohn diseasePhase I testing supported short-term feasibility, but durable anti-inflammatory benefit required stronger validation.82 Table 1.Representative clinical-stage or translationally relevant bacterial and live microbial therapeutic programs.Program/platformDesign/modalityClinical settingHuman or translational signalRef.SYNB1891 (Escherichia coli Nissle 1917)Intratumoral EcN producing STING-activating cyclic dinucleotidesAdvanced solid tumors or lymphomaPhase I testing showed immune target engagement and manageable early dosing, while cytokine-related toxicity required dose optimization.76VNP20009 (Salmonella Typhimurium)Attenuated tumor-colonizing strainMetastatic melanoma and renal cell carcinomaAdministration and some tumor colonization were feasible, but objective tumor regression was not observed.77Clostridium novyi-NT sporesObligate anaerobic spores that germinate in hypoxic tumor regionsTreatment-refractory solid tumorsEarly intratumoral testing supported local bacterial activity, while inflammatory toxicity and delivery control remained limiting issues.78CRS-207 (Listeria monocytogenes)Live-attenuated vector expressing mesothelinPancreatic and other mesothelin-positive malignanciesAntigen-directed delivery was biologically plausible, but durable survival benefit was not consistently established.79SYNB1020 (Escherichia coli Nissle 1917)Ammonia-consuming engineered EcNHyperammonemiaEarly testing indicated tolerability and biological activity, but the metabolic output was insufficient for clinical development.80AG013 (Lactococcus lactis)Topical Lactococcus lactis secreting human trefoil factor 1Oral mucositis during cancer therapyLocal mucosal protein delivery was feasible; later efficacy assessment showed the difficulty of meeting a clinical mucosal endpoint.81LL-Thy12/IL-10-secreting Lactococcus lactisBiologically contained strain expressing human interleukin-10Crohn diseasePhase I testing supported short-term feasibility, but durable anti-inflammatory benefit required stronger validation.82 Representative clinical-stage or translationally relevant bacterial and live microbial therapeutic programs. The problem extends beyond dose selection. Live bacterial products can replicate, disappear, mutate, shed, exchange genetic material, interact with resident microbes, or be cleared before their intended function is achieved. In oncology, uneven tumor perfusion and immune clearance can alter bacterial distribution. In the gut, diet, transit time, inflammation, antibiotics, and baseline microbiota can shift exposure from one patient to another. Manufacturing introduces further challenges: potency assays, genetic stability, viable count, release specifications, and batch-to-batch consistency must be defined in ways that regulators and clinicians can assess. Studies on bacterial modulation of host metabolism and Listeria monocytogenes as an immunotherapy vector both return to the same development requirement: clinical performance depends on measurable persistence, payload output, host clearance, and pharmacodynamic activity.83,84 CRISPR-Cas3-enhanced bacteriophage therapy adds a related example from antimicrobial development.85 Recent conceptual work on living bacterial therapeutics makes the same point from a product-design perspective.86 For the field to move forward, different strains, routes, and payloads will need common metrics before preclinical success can be compared across studies. The collaboration data reveal a structural dimension of these challenges. Publication output was led mainly by China and the United States, while institutional networks remained regional rather than globally integrated. Such an uneven structure could slow the development of shared standards for dose control, manufacturing consistency, potency testing, ecological risk assessment, and host-safety monitoring – methods that need to be reproducible across institutions and regulatory settings. Cross-regional consortia may help set those standards and enable multicenter validation and regulatory alignment. If so, future progress will rely less on new genetic modules and more on building reproducible pharmacology, controllable strain behavior, and clinically testable performance standards. Several limitations of bibliometric mapping should be kept in mind. This approach describes indexed publications and citation networks, but cannot capture unpublished failures, negative animal studies, manufacturing decisions, regulatory correspondence, or company strategy. Citation indicators favor older publications, while newer work may not yet reflect its longer-term influence. Coverage and cited-reference metadata also differ across Web of Science Core Collection, Scopus, and PubMed. The field lacks stable terminology; “engineered commensal,” “microbial therapeutic,” “microbiome therapeutic,” “living medicine,” and “live biotherapeutic product” are used interchangeably. This study concentrated on engineered bacteria as therapeutic or diagnostic platforms, including live bacterial products, bacterial delivery systems, and selected bacteria-derived therapeutic materials. Industrial microbial chassis that manufacture drug intermediates, ADC cytotoxic payloads, natural-product analogues, or other pharmaceutical raw materials were outside the scope, because they follow different search vocabularies, regulatory questions, and development metrics. Studies on manufacturing control, regulatory science, ecological risk assessment, and long-term biocontainment may therefore be underrepresented in this dataset. Genetic expiry-date circuits, for instance, program bacterial lifespan and post-function clearance.87,88 Programmable living materials and bacteria-based nanoplatforms also fall near this boundary, where engineered bacteria overlap with hybrid materials and precision delivery.89,90 Bacteria-based nanoprobes and precision strategies that modify tumor-associated bacteria similarly sit at the edge of the present search vocabulary.91,92 A later update would be stronger if it linked bibliometric trends with clinical trial results, regulatory documents, manufacturing standards, and long-term safety evaluation. Conclusion This multi-database scientometric analysis mapped 1,730 publications on engineered bacterial therapeutics from 2000 to 2026. Output rose sharply after 2020, led mainly by China and the United States. The knowledge base is distinctly interdisciplinary, drawing on synthetic biology, microbiology, immunology, oncology, materials science, drug delivery, and microbiome medicine. The bibliometric patterns trace a shift away from chassis construction and proof-of-concept circuits toward disease-oriented applications and a small set of clinical-stage programs. Progress from here will depend on reproducible colonization, controllable payload output, genetic stability, immune safety, manufacturing consistency, and regulatory clarity. Supplementary Material Supplemental Material