In plant genetic transformation, one of the historically most useful – due to its versatility and simplicity – and widely employed reporter genes is the Escherichia coli beta (β)-glucuronidase gene (senso lato, GUS; EC 3.2.1.31) (Jefferson, 1987; Jefferson et al., 1987). To appreciate how widely GUS has been used as a reporter gene, a search at Google Scholar revealed that the Jefferson et al. (1987) and Jefferson (1987) papers had been cited 11,455 and 6169 times, respectively (July 6, 2021). As one random and fairly recent example, GUS was used to confirm the tissue-specific expression of a small subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase (RbcS) in transgenic sweet potato (Tanabe et al., 2015). 4-Methylumbelliferyl β-D-glucuronide (MUG) is a fluorescent substrate of the GUS fluorometric assay that is used to measure and quantify GUS activity, typically using the Jefferson (1987) or Jefferson et al. (1987) protocol. It is also common to observe the use of 5-bromo-4-chloro-3-indoxyl-β-D-glucuronic acid (X-gluc) in a histochemical assay to detect GUS activity in plant tissues (e.g., Dang et al., 2014). On occasion, X-gluc was abbreviated incorrectly, as 5-bromo-4-chloro-3-indolyl glucuronide (i.e., without beta) (e.g., Nyman and Wallin, 1992; Zeng et al., 2016), but this error was not quantified in the analyses conducted later on in this paper.
There are only two forms of glucuronidase, alpha (α-glucuronidase, or AGUS), and beta (BGUS). In humans and bacteria, BGUS is coded for by the BGUS (Oshima et al., 1987) and uidA (Martins et al., 1993) genes, respectively. BGUS is also used as a reporter gene in mammalian cells to explore the metabolism of drugs and other substances, such as the catalytic deconjugation of β-D-glucuronides, and serves as “a viable molecular target for several therapeutic applications” (Awolade et al., 2020). As demonstrated in AGUS and BGUS, Greek letters are used to differentiate the two forms.
Recently, it was discovered that the biomedical literature may be riddled with a unique typographic or linguistic error involving confusion between the Greek letter beta (β) and a German special character, Eszett (uppercase or capital ẞ; lowercase ß, or ß when written in italics), leading to the existence of an unknown number of undisclosed or unintentional errors (Teixeira da Silva, 2021a). As one example, a false positive of β-carotene was found, Eszett-carotene or ẞ/ß-carotene, which does not exist, but which may impact the claims made in food science and nutrition studies (Teixeira da Silva, 2021b). Similarly, in this paper, select cases of a false positive of beta (β)-glucuronidase, namely Eszett (ẞ/ß)-glucuronidase, are highlighted.
Incidence of Eszett (ẞ/ß)-glucuronidase in plant science using PubMed as the litmus test
Using a similar method as Teixeira da Silva (2021b), a search (July 6, 2021) was made on Google Scholar, PubMed, Elsevier’s sciencedirect.com and Springer Nature’s Springer Link for the false positive, ẞ/ß-glucuronidase, revealing 10,600, 114, 4083, and 1084 results, respectively (Table 1). Similarly, approx. 83,000, 21,681, 22,506 and 11,124 results for the corresponding control, β-glucuronidase, were found. These likely represent crude searches that themselves may include false positives (i.e., results not related to either β-glucuronidase or ẞ/ß-glucuronidase). Of note, whereas PubMed and sciencedirect.com did not differentiate the use of the alphabetized form of beta and the Greek letter itself (β), vastly different results were observed for Springer Link (6001 vs. 11,124, respectively) and Google Scholar (55,900 vs. 83,000, respectively). This is important because academics searching for literature on these two platforms/databases might find very different results, thereby potentially biasing their selection of literature to cite. Scientists are thus cautioned about their choice of search word, and it is suggested that they search using both the form that includes the Greek letter and also the alphabetized form, to ensure a maximum array of results. Curiously, none of the four platforms and/or databases that were searched recognized the term “Eszett”. Finally, all four platforms or databases recognized the uppercase and lowercase Eszett as equal terms, giving identical outputs when searching for ẞ/ß-glucuronidase, sometimes representing this as “ss” in the alphabetized form of the German Eszett.
Table 1
Incidence1 of false positive entry (ẞ/ß-glucuronidase) and control entry (β-glucuronidase) in four search engines or indexes/databases
| PubMed | sciencedirect.com | Springer Link | Google Scholar |
beta glucuronidase | 21,681 | 22,506 | 6,001 | Approx. 55,900 |
β-glucuronidase | 21,681 | 22,506 | 11,124 | Approx. 83,000 |
Eszett glucuronidase | 0 | 0 | 0 | 0 |
ẞ-glucuronidase2 | 114 | 4,083 | 1,084 | Approx. 10,600 |
ß-glucuronidase3,4 | 114 | 4,083 | 1,084 | Approx. 10,600 |
1 These are raw searches which themselves may contain false positive search results, so the numbers only offer a crude appreciation of the volume of possible papers in each category. A search was conducted on June 17, 2021 (and verified for PubMed, which was analyzed in more detail, on July 6, 2021) |
2 Upper-case Eszett |
3 Lower-case Eszett |
4 The appearance changes, resembling beta (β), when written in italics: ß (in contrast, the visual appearance of the upper-case ẞ does not change in its italicized form) |
Given its manageable sample size (n = 114), and a recent focus on the accuracy and reliability (or lack thereof) of PubMed in academic publishing (Teixeira da Silva, 2021c), a small analysis was conducted using the PubMed data set. Using PubMed’s 114 apparent β-glucuronidase false positives (i.e., ẞ/ß-glucuronidase), studies directly related to plant science were searched. Studies using plant-based products, for example, in medical studies, or applied studied, such as in microbiology, were not considered. A total of 81 papers on plant science carrying this error, were found (Suppl. Table 1). These represented errors in the title, abstract and keywords, which would be detected during a search on PubMed. The original texts, where available, were also analyzed, as were the source webpages, i.e., of the journal/publisher, to assess whether the error was exclusive to PubMed, or in common with the original publication and/or publisher website. Where possible, an automatic search of β and ẞ/ß was conducted in PDF files, but for older files that could not read text automatically, a thorough search of the entire text was conducted manually.
Except for two papers (Chateau et al., 2000; Karimi et al., 2002), which had no errors in the original publication or journal website (i.e., PubMed-specific errors), the remaining 79 papers had errors in the abstract, text, keywords and/or abbreviations.
On PubMed, the error (false positive) was found in 100% of the 81 (and 79 in refined dataset) abstracts. In the original publications, including website, html and PDF versions of the papers, the false positive was found in 100% of the abstracts and in 49/79 (62.0%) of the texts (any location excluding the abstract, references, abbreviations, and key words). In the 79 papers examined, 38 (48.1%) also carried this error in the abbreviations. To give readers some notion of the range of errors that exist in this sample set, some examples include: one error (abstract) in Kong et al. (2020); three errors in the abstract and four in the text and abbreviations (Kuai and Morris, 1996); a mixture of correct β-glucuronidase and the false positive ẞ/ß-glucuronidase (Markulin et al., 2019), with 6/19 mentions of β or ẞ/ß being attributed to ẞ/ß); 21 errors anywhere in the text (Obertello et al., 2005).
Moreover, a number of papers had errors in MUG and X-gluc (5.1% (4/79) and 24.1% (19/79) errors, respectively). PubMed also displayed another two categories of deficiencies that were not existent in the original publications, namely the lack of italics (91.1% (72/79) errors), or abbreviated authors’ names (62.0% (49/79) errors).
In summary, these examples all employed the erroneous false positive Eszett (ẞ/ß)-glucuronidase to some extent (title, abstract, and/or main text), on PubMed and/or the publisher’s website.