Simple, specific, rapid, and pharmacopoeia-compliant qPCR approach for the detection of mycoplasma in biopharmaceuticals

Abstract

Cell substrate utilized in the production of biologics intended for human use need to be cleared from mycoplasma contaminant as described in pharmacopoeia. While the gold-standard method for mycoplasma detection involves lengthy microbiology and cell-based assays, nucleic acid technologies offer the possibility to accelerate testing. In this study, we develop a qPCR assay capable of specifically detecting 11 mycoplasma species relevant to pharmacopoeias with only two primers and two hydrolysis probes, which greatly facilitates operations. While the primers cross-react with bacterial species, the specificity conferred by the hydrolysis probes allows for confident interpretation of the qPCR result and unambiguous statement regarding the presence of mycoplasma in the test article. Amplicon sequencing can further confirm the identity of contaminants. This comprehensive assay can therefore be of great help to quality control laboratories embedded in biologics production sites, which need to provide results in a timely manner and contribute to root cause analysis in case of contamination.

Graphical abstract

Dos Santos and colleagues have developed a pharmacopoeia-compliant, mycoplasma-specific qPCR assay that uses two primers and two probes as a minimum requirement. The development involved selecting a target mycoplasma genomic sequence, designing primers for genomic DNA amplification, and designing hydrolysis probes specific to amplicons derived from the mycoplasma genome.

Introduction

Mammalian cell lines used for the manufacturing of therapeutic proteins and vaccines for human use must be thoroughly characterized and demonstrate no detectable microbial contamination to ensure patient safety (reviewed in¹). This biosafety concern is addressed in various guidance, guidelines, and recommendation documents from health agencies,^{2,3,4,5,6,7,8} which describe the nature of the biosafety testing panel to be applied on cell lines. Pharmaceutical industries are therefore expected to provide the results of such testing in investigational new drug (IND)/investigational medicinal product dossier (IMPD) and marketing applications, prior to engaging their product in clinical trials or going commercial, respectively.

Besides viral safety, which is described in specific documents^9,10,11 and clearly summarized by the “International Conference on Harmonization (ICH) of technical requirements for registration of pharmaceuticals for human use” in a single document,¹² cell substrates are also under scrutiny with respect to contamination with organisms of the bacterial class of Mollicutes, often referred to as Mycoplasma. Such microbes are widespread in species of the Animalia and the Plantae kingdoms and may enter the production workflow from contaminated raw material. Indeed, because mycoplasmas are very small (0.1–0.3 μm in dimeter vs. 1–2 μm in length for bacteria), 0.1–0.2 μm pore size sterilizing-grade filters used for cell culture media preparation may not completely prevent mycoplasma ingress into the manufacturing process.¹³ Contamination may also occur through the introduction of a contaminated cell substrate early in non-GMP cell line development laboratories or through operators since most mycoplasma species are commensals of humans.^14,15,16 Although not all mycoplasma species are pathogenic for humans,^17,18,19 guidelines require that cells used to manufacture biologics intended for human use be tested for the presence of mycoplasma.

In addition to patient safety concerns, mycoplasma ingress into the manufacturing process also represents a threat to the industry due to the impact on mammalian cell metabolism, production yields, and eventually product quality.^20,21,22,23 As documented in cases of viral contamination, a breach in the biosafety control strategy might lead to facility shut down and failure to match drug demand on the market (reviewed in²⁴). It is therefore in the interest of industries to set up strategies capable of promptly detecting contamination all along their manufacturing process, on top of the biosafety requirements for clinical trials and marketing applications. Unlike bacteria which growth outcompetes that of mammalian cells and makes contamination easily detectable, contamination with mycoplasma might not have such a noticeable effect and get detected too late. To be efficient, a biosafety control strategy should therefore be sensitive and use methods with short turnaround times to protect facilities from forward processing of contaminated material.

The reference method for mycoplasma testing is described in the Food and Drug Administration (FDA) Points to Consider,⁴ the United States, European, and Japanese Pharmacopoeias,^25,26,27 which include a list of organisms that the method should be able to detect with a detection limit of 10 CFU/mL. This method relies on mycoplasma growth on solid or liquid media (also known as “cultivable mycoplasma” method) and derives from the early work of Cross et al.²⁸ It also includes mycoplasma detection by staining techniques and microscopic observation upon cocultivation with reference cell lines (also known as “non cultivable mycoplasma” method). Although these culture-based approaches are considered gold standard, they are difficult to implement on manufacturing sites that, therefore, outsource the assay to companies specialized in biologics safety testing. In addition, their long turnaround time in the month range is not compatible with fast product release.

As predicted by S. Razin in 1994,²⁹ nucleic acid technologies (NAT)-based methods have been developed as an alternative solution to the cell-culture- and dye-based mycoplasma detection assays. Many PCR methods have been designed, capable of specifically detecting mycoplasma of clinical interest such as Mycoplasma pneumonia^30,31,32,33 and Mycoplasma genitalium^34,35,36 or species of concern for cattle and poultry such as Mycoplasma gallisepticum and Mycoplasma synoviae.^36,37,38,39 Methods with a wider detection spectrum have also been made available as commercial kits that guarantee detection of mycoplasma contamination with specificity and limit of detection (LOD) matching guidelines’ expectations.⁴⁰ However, these kits are expensive and most of the time use SYBR Green for amplicon detection, which is a concern in the sense that non-specific DNA amplification might be detected, lead to false-positive signal, and trigger long investigation campaigns. Many articles in the literature also describe PCR,^{41,42,43,44,45,46,47,48} real-time qPCR,^49,50 reverse transcription PCR,⁵¹ LAMPCR,⁵² and hybrid methods⁵³ with broad range mycoplasma detection, which can be applied by any laboratory for in-process monitoring. Digital PCR has also been described, although with limited detection spectrum.^54,55,56 These methods are cheaper than commercial kits but require supply chain efforts since reagents should be ordered separately, with PCR-based methods sometimes relying on the use of 10 sets of primers or more to ensure coverage and specificity.

In this work, we describe the design of a real-time qPCR method that relies on the use of two primers, together with two hydrolysis probes as the unique specific reagents capable of detecting the mycoplasma species listed in relevant pharmacopoeias, with demonstrated absence of detection on bacterial germs, human, and Chinese hamster ovary (CHO) cell substrates.

Results

Identification of genome regions to be targeted by PCR

PCR primers and probes were designed from the reference sequences of germs listed in Table S1. Cricetulus griseus and Homo sapiens genomes were also included in this study to counter select sequences with potential cross-reactivity with test articles (CHO or human cell lines) or potential contamination by operators during handling. In order to identify the best genomic region to be targeted by PCR and ensure specific mycoplasma detection with minimal signal from other organisms, 16S and 23S ribosomal RNA (rRNA) gene sequences were compared. Sequences from 23S rRNA genes are too unique and did not allow the identification of all mycoplasma species of interest with a limited number of primers and probes (not shown). Sequences from the 16S rRNA genes, however, which are known to be species specific,^57,58 display some specificity across species but retain enough homology within the mycoplasma genera. It was therefore decided to focus on the 16S rRNA gene sequences to identify regions common to all mycoplasma species that a primers/probe set should recognize and significantly different from all other species. A first pass analysis aimed at identifying ∼100 bp stretches with 80% homology across species where primers and hydrolysis probes could be designed for mycoplasma detection. As shown in Figure 1A, three stretches corresponding to nucleotides ∼500–600, ∼750–850, and ∼1050–1150 in the 16S rRNA gene of the Mycoplasma hyorhinis strain BTS7 reference sequence were identified, which met the criteria mentioned above. Sequences were analyzed within these three regions to select primers and probes suitable for all mycoplasma targets detection, with Tm of 60°C and 70°C, respectively, and with no primer dimers and hairpin structures formation at 45°C. Primers and probes were also checked against sequences from other species listed in Table S1 to minimize cross-reactions. While specific primers/probes could not be designed within the first two regions, the third region gave satisfactory results after its range was extended to nucleotides 899–1075 of the 16S rRNA gene sequence (nucleotides 94660–94836 of the M. hyorhinis BTS7 genome) to accommodate all three reagents. As shown in Figure 1B (lower panel), eukaryotic and prokaryotic sequence mismatches within the green and blue regions allow for the design of primers with limited cross-reactivity with CHO and human sequences. Sequence homology is high, however, within these regions across mycoplasma and other prokaryotic species (Figure 1B, middle panels), suggesting that primers designed within these regions might also amplify non-mycoplasma prokaryotic DNA. In order to ensure PCR specificity, the hydrolysis probe was designed within the red region where there is sufficient homology among mycoplasma species (Figure 1B, upper panel) yet significant differences with other prokaryotic organisms (Figure 1B, middle panel).

Figure 1.

Target identification within the 16S rRNA gene for mycoplasma-specific qPCR design

(A) Sequences from the mycoplasma species listed in Table S1 were aligned and similarities calculated. When several copies of the 16S rRNA gene were present, all copies were taken into account in the alignment. The genomic regions boxed in red (plain) represents ∼80 bp stretches with >80% global sequence homology over a 20-nucleotide sliding window. The dotted red box indicates the region used to design primers and probes. (B) Nucleotides 899–1075 of the 16S rRNA gene of M. hyorhinis BTS7 reference sequence alignment with the corresponding eukaryotic and prokaryotic sequences. Sequence alignment for all mycoplasma species relevant to the European Pharmacopoeia (upper panel), for mycoplasma and bacteria species (middle panel) and for prokaryotic and eukaryotic species (lower panel). Regions where forward primers, reverse primers, and probes sequences were designed are boxed in green, blue, and red, respectively. In all three panels, M. hyorhinis is used as a reference for mismatch identification. Mollicutes: M hy, Mycoplasma hyorhinis; A l, Acholeplasma laidlawii; M a, Mycoplasma arginini; M b, Mycoplasma buccale; M f, Mycoplasma fermentans; M g, Mycoplasma gallisepticum; M ho, Mycoplasma hominis; M hy, Mycoplasma hyorhinis; M o, Mycoplasma orale; M p, Mycoplasma pneumoniae; M sa, Mycoplasma salivarium; M sy, Mycoplasma synoviae; S c, Spiroplasma citri. Bacteria: E c, Escherichia coli; M l, Micrococcus luteus; C p, Clostridium perfringens; C s, Clostridium sporogenes; S b, Streptococcus bovis; S p, Streptococcus pneumoniae; L a, Lactobacillus acidophilus; L p, Lactobacillus plantarum; S a, Staphylococcus aureus; S s, Staphylococcus saprophyticus; B c, Bacillus cereus; B s, Bacillus subtilis. Others: H s, Homo sapiens; C g, Cricetulus griseus.

Forward and reverse primer design

Several primers were designed within the regions identified in Figure 1, and their fit for qPCR was determined. Regarding the forward primers, no sequence strictly identical across species could be identified, which imposed the design of primers capable of accommodating mismatches with target DNA. As shown in Figure 2A, the region to be targeted has variants in positions 933 (C/T), 940 (G/T), 946 (G/T), and 948 (T/C). Primers were thus generated with the most frequent nucleotide combinations at these four positions to select a sequence that could fit all targets. A similar approach led to the identification of reverse primers with sequences and characteristics presented in Figure 2B. In silico screening was then used to exclude sequences prone to primer dimer formation and with temperature-resistant hairpin structures (Tm > 50°C) and led to the selection of a limited number of primers for subsequent steps of the method setup. Tm were calculated assuming no mismatch with targets. It should be noted however that each mismatch lowers the Tm by ∼5°C according to the formula published in the Agilent QuickChange mutagenesis kit. For example, as shown in Figure 2A, the FwMyco30 primer that belongs to the CTTT group would have a Tm of 64.7°C with DNA from organism #13 (no sequence mismatch) but a Tm of ∼55°C with DNA from organisms # 17 and 9 (CGGT group, two mismatches) and a Tm of ∼50°C with DNA from organisms # 8 and 10–12 (TGTC and TGGT groups, three mismatches).

Figure 2.

Primer design

The forward (A) and reverse (B) primers were designed in the 16S rRNA genes sequences corresponding to nucleotides ∼910–960 and ∼1040–1080 of the M. Hyorhinis sequence taken as a reference. The nucleotide sequence in this region is indicated with letter sizes corresponding to the frequency of each nucleotide within the target mycoplasma sequences. The primer type column indicates the number of mycoplasma species which 16S rRNA gene sequence matches with the corresponding primer. Primer sequences are oriented 5′ to 3’. Nucleotides highlighted in yellow indicate sequence diversity within the selected region. Nt, primer length; GC%, percentage of G and C in the primer sequence; Dimer, propensity of the primer to form dimers; Tm, primer melting temperature; TmHP, hairpin melting temperature. GC%, Tm and TmHP were determined from the IDT website with qPCR parameters selected on batch analysis. Tm were calculated assuming no mismatch with target DNA. Parameters highlighted in gray led to primer counterselection; primers in bold were selected for the next steps.

Primer assessment by dye-incorporation-based PCR product detection

Selected primers were tested in SYBR-Green-based qPCR with 20,000 copies of genomic DNA as described in the materials and methods section. As shown in Figure 3A, the use of FwMyco27, 30, 39, and 46 resulted in the highest Cq values across most mycoplasma species investigated. On the contrary, Cq values obtained with FwMyco11, 14, and 17 were below 20 in most instances. All Cq values obtained with FwMyco7 were below 20 except for Mycoplasma hominis. Quantitative PCR performed on DNA from other organisms amplified target DNA, although to a lower extend, with Cq in the 23–30 range (Figure 3A). This was expected given sequence homologies shown in Figure 1 and is not a concern since the final qPCR method is intended to use hydrolysis probes to specifically detect DNA amplified from mycoplasma. Besides SYBR Green incorporation readout, amplicons generated were also resolved by 15% PAGE to check the quality of the amplification and potential presence of primer dimers. As shown in Figure 3B, PCR performed with the FwMyco7/RevMyco1∗ primer set resulted in an expected profile with unique ∼130 bp bands of similar intensities across species and with no primer dimer visible at the bottom of the gel. On the contrary, PCR performed with the FwMyco30/RevMyco1∗ primer generated more primer dimers and species in the 200 bp range, on top of the expected amplicon of 130 pb. In addition, the amplification had various efficiencies across species, with specific bands barely detected for PCRs with Cq > 34 and clearly visible when Cq < 33 in most instances. Amplicons generated by all other forward primers together with RevMyco1∗ resulted in profiles close to that obtained with FwMyco30 (not shown), consistent with the qPCR results. The FwMyco7/RevMyco1∗ primer set that gave the best results was therefore chosen for further method development.

Figure 3.

Primer evaluation in SYBR-Green-based PCR

Reference DNA from mycoplasma and other organisms were used in qPCR reactions with the RevMyco1∗ primer and the indicated forward primers. (A) Cq values are reported and colored from efficient detection (green) to poor or no detection (red). ND, not detected. (B) Amplicons obtained from the qPCRs with the FwMyco7/RevMyco1∗ (upper panel) and FwMyco30/RevMyco1∗ (lower panel) primer sets were resolved by 15% PAGE and visualized under UV light. Molecular weight marker size (base pairs) is indicated on the left of each gel.

Hydrolysis probe design

Following the demonstration of efficient target DNA amplification with the selected primer set, the qPCR method was further improved to ensure detection restricted to mycoplasma genera. To this aim, mycoplasma-specific hydrolysis probes were designed within the relevant genomic region shown in Figure 1B. Given the nature of the genomic sequences, no conventional 23-nucleotide probes could be designed with a Tm of ∼70°C, suitable for all mycoplasma species detection. Probes were thus synthesized with modified chemistry such as locked nucleic acids (LNAs) or minor grove binding (MGB) moieties to tighten the probe/target interaction. Longer probes of up to 40 nucleotides were also designed but invariably encompassed non-homologous sequences in the 5′ region of the targeted mycoplasma DNA region (Figure 4). Finally, due to significant mismatches between the Acholeplasma laidlawii sequence and that of other mycoplasma species, a specific probe was also designed for this particular species. For long probes, the ZEN technology was used on top of the strong Iowa Black fluorescence quencher (IabFQ) to ensure efficient quenching of the fluorochrome. For short probes, two configurations were used, either LNA modification with IabFQ or MGB modification with the black hole quencher-1 (BHQ). In silico analysis led to the elimination of probes with high-temperature-resistant (Tm>∼45°C) hairpin structures and dimer formation. The remaining SMyco5 LNA, SMyco5bis LNA, SMyco5 MGB, SMyco5bis MGB, and SAcho1 LNA probes were synthesized and used with the unique FwMyco7/RevMyco1∗ primer set in qPCR performed on mycoplasma DNA reference standards.

Figure 4.

FAM-labeled hydrolysis probe design

Probes were designed in the 16S rRNA gene sequences corresponding to nucleotides ∼970–1,040 of the M. Hyorhinis sequence taken as a reference. The nucleotide sequences in these regions are indicated with letter sizes corresponding to the frequency of each nucleotide within the target mycoplasma sequences. The probe names and sequences are indicated in column 1 and 2, respectively. Underlined bases in the nucleotide sequences were LNA modified. Probe sequences are oriented 5′-3’. Nucleotides highlighted in yellow indicate sequence diversity within the selected region. Nt, primer length; GC%; percentage of G and C in the primer sequence; Dimer, propensity of the primer to form dimers; Tm, primer melting temperature; TmHP, hairpin melting temperature. GC%, Tm1 and TmHP were determined from the IDT website with qPCR parameters selected on batch analysis. Tm2 was calculated with PrimerExpress 3.0 and the values retained only for MGB probes (italicized). Tm were calculated assuming no mismatch with target DNA. Parameters highlighted in gray led to probe counterselection; probes in bold were selected for the next steps.

Probes were tested in qPCR with 20,000 copies of genomic DNA. Regarding the SMyco5 probes, A. laidlawii set aside, all probes efficiently detected all mycoplasma species DNA, with Cq ranging from ∼20 to 30, except the SMyco5 MGB probes that failed to detect M. gallisepticum DNA (Figure 5, left panel dataset). No or limited background signal (Cq > 37) was observed with DNA samples from other species although DNA amplification occurs as demonstrated previously with eukaryotic and prokaryotic species (Figure 3). This confirms the specificity of the hydrolysis probes for mycoplasma DNA detection, with the SMyco5bis LNA probe being the most efficient.

Figure 5.

Characterization of LNA and MGB probes on standard DNA samples

Reference DNA from mycoplasma or other organisms were used in qPCRs with the RevMyco1∗and FwMyco7 primers and the indicated probes. PCRs were performed with probes used either alone (left panel) or in combination (right panel). Cq values are reported and colored from efficient detection (green) to poor or no detection (red). ND, not detected. Italicized, failed or non-specific detection.

As mentioned previously, the A. laidlawii 16S rRNA gene sequence is significantly different from that of other mycoplasma species, which explains why SMyco5 probes failed to detect material amplified with the FwMyco7/RevMyco1∗ primer set when A. laidlawii DNA reference standard was used as a matrix. Quantitative PCRs were thus performed with the SAcho1 LNA probe and resulted in the opposite pattern, with efficient detection of DNA amplified from A. laidlawii reference standard but not from other mycoplasma reference standards (Figure 5, right panel dataset). In order to assess whether all relevant mycoplasma species could be detected in a single qPCR reaction, experiments were repeated with both SMyco5bis LNA and SAcho1 LNA probes added to the same reaction mix. As shown in Figure 5 (right panel dataset), all relevant mycoplasma species were detected in these experimental settings. In addition, the Cq values obtained demonstrate that the method is as efficient whether probes are combined or used in two distinct reactions, with undetectable signal on DNA from other species. This confirms the method’s specificity for mycoplasma species. It is therefore possible to use a single primer set together with two probes in a single qPCR reaction to specifically detect all mycoplasma species listed in various pharmacopoeias with no background signal on DNA from other species.

Of note, there is a correlation between the Cq obtained by SYBR-Green-based and hydrolysis-probe-based qPCR, which suggests that probes introduce no bias in amplicon detection. The range of Cq obtained across mycoplasma species is therefore mostly due to the differential annealing of primers to the 16S rRNA genes of each mycoplasma species addressed in this study, or duplication of these genes, which is specific to each species.

Method validation

The results presented above are representative of one set of experiment performed several times; however, on top of specificity, the final settings with the best primers and probes combination went through full validation for linearity and precision (repeatability and intermediate precision) as described in the ICH-specific document.⁵⁹ In the course of this specific study, focus was put on M. hyorhinis as a representative for all mycoplasma species of interest due to sequence similarity and one species each of Acholeplasma and Spiroplasma genera (A. laidlawi and Spiroplasma citri). Reference DNA standards for all three organisms were used, together with internal positive control (IPC), discriminatory positive control (DPC), and negative controls. Briefly, linearity was investigated by performing qPCR in triplicate, with 10-fold serial dilution of target DNA from 100,000 to one copy, with 3–8 replicates per dilution. Repeatability and intermediate precision were addressed by performing the experiments on three different days by two operators. For the sake of conciseness and because it is out of scope of this paper focused on method development, the complete dataset is not presented in this work. The data were used, however, to narrow down the range of DNA concentration to be used for LOD and limit of quantification (LOQ) determination described below.

Determination of the limit of detection and quantification

While probes and primers were developed with purified genomic DNA used at known concentrations in qPCR (20,000 genome copies per reaction), guidelines ask for assays capable of detecting a defined number of colony-forming unit (CFU)/mL as part of method validation. Lyophilized mycoplasma samples of known CFU contents were therefore tested together with various amounts of qPCR DNA standards to bridge CFU to genome copies and determine the LOD and LOQ of the method, expressed in CFU/mL.

The first set of samples consisted in serial dilutions of reference DNA standards in which qPCR Cq results were used to calculate the parameters of a calibration curve (Figure 6A). The LOQ of the method expressed in genome copies per reaction (LOQ_CN/reaction) was determined from the lowest dilution with quantifiable signal for all three replicates and standard deviation <5%. As shown in Figure 6A (upper panel), the LOQ was 100 mycoplasma genome copies per reaction for all species except M. synoviae (1,000 mycoplasma genome copies per reaction), which is similar to the values observed during the validation study (not shown). Of note, all samples were positive in the IPC channel at the expected Cq value (not shown). The LOD was also determined from the lowest dilution with detectable but not necessarily quantifiable signal for all three replicates and was one 10-fold dilution below the LOQ.

Figure 6.

Calibration and method LOD/LOQ determination

(A) Specific amounts of DNA standards from the indicated organisms were used in triplicate qPCRs with the RevMyco1∗ and FwMyco7 primers and the Smyco5bis LNA and Sacho1 LNA probes. Top, Cq values; bottom, parameters of the calibration curve. Cq values highlighted in red and green were retained for LOQ/LOD determination, respectively, according to criteria described in the text. ND, no detectable amplification; NA, not applicable (an amplification signal is detected but no Cq value is returned by the software). (B) Specific amounts of mycoplasma 10CFU standards from the indicated organisms were used in hexaplicate qPCRs with the RevMyco1∗ and FwMyco7 primers and the Smyco5bis LNA and Sacho1 LNA probes. Mean Cq values are shown (SD were <5% for all nine samples, not shown). The genome copy equivalent and LOQ were determined as described in the materials and methods section.

The method was also applied to colony reference standards, with known CFU contents used as a second set of samples analyzed together with the first set of samples. The qPCR Cq values (Cq_{10CFU standard}) obtained on samples corresponding to 0.5 CFU were then used to calculate the LOQ expressed in CFU/mL from the calibration curve, as described in the materials and methods section. As can be seen in Figure 6B, the method LOQ was <100, which corresponds to an LOD <10 CFU/mL and matches the expectations of pharmacopoeias.

Case studies

Our method was then used to characterize cell culture samples derived from a human U2OS cell line culture tested positive for mycoplasma by Eurofins with the real-time PCR MycoSEQ kit from Thermo Fisher Scientific (Sample 1) and from a CHO cell line culture with altered cell growth metrics and unusually turbid cell culture supernatant (Sample 2). Both samples were tested by qPCR with the MycoTOOL kit and the method described in this paper. While both methods returned positive results on sample 1, sample 2 was positive for mycoplasma contamination with the MycoTOOL kit only (Table 1). In order to better understand these results, amplicons generated by the methods described in this work were sequenced and revealed that sample 1 was contaminated with M. hyorhinis, which is consistent with the high probability of cell culture contamination with this organism.⁶⁰ Because sample 2 was positive with the MycoTOOL kit but not with the method described in this work, we suspected a contamination with microbes distinct from all mycoplasma species for which our method has demonstrated specificity. The analysis of the PCR product confirmed the presence of an amplicon of a ∼130 bp fragment that sequencing identified to DNA from Micrococcus luteus, a germ commonly found in the environment, on human skin, and that of other animals.⁶¹ Altogether, these results confirm that our method is specific on experimental samples and can be used confidently to not only determine the status of mycoplasma contamination but also identify the nature of the contamination by amplicon sequencing if an investigation is needed.

Table 1.

Case study on laboratory samples

Method	Assay	Sample 1, U2OS (H. sapiens origin)	Sample 2, CHO (C. griseus origin)
MycoSEQ kit (Eurofins)a	qPCR/SYBR Green	positive	not done
MycoTOOL kitb	qPCR/probe	positive	positive
	amplicon size	∼460 bp	∼460 bp
Method developed in this workc	qPCR/probe	positive	not detectable
	amplicon size	∼130 bpd	∼130 bpd
	amplicon sequence homology	M. hyorhinis	M. luteus

^aSample analysis with the MycoSEQ kit was performed at Eurofins.

^bSample analysis with the MycoTOOL kit was performed in the Sanofi lab according the instruction of the manufacturer.

^cSample analysis with the method developed in this work was performed as described in the text.

^dAmplicon size distinct from that generated by the IPC and DPC, ∼160 and 180 bp, respectively.

Comparison with commercial detection platforms

In order to assess the competitiveness of our method, we conducted a benchmarking analysis with kits commercially available, which can be used together with the work of Lawrence et al. and more recently by Toten at al., to help end users for qPCR method selection.^40,62 Four major items were retained for this analysis, the method itself, its compliance to various pharmacopoeias, the presence of extra reagents intended to secure data interpretation, and the price for 100 reactions.

As can be seen in Table 2, most kits rely on hydrolysis-probe-based qPCR that guarantees two levels of specificity: target amplification and detection. Applied Biosystems has recently developed a hydrolysis-probe-based qPCR MycoSEQ Plus kit in addition to the existing MycoSEQ kit; however, this gain is undermined by the need for a specific fluorescent detector that is not present in most qPCR machines. Although digital PCR is not the focus of this work, it should be noted that our method is fully functional on such platform. The digital PCR QIAcuity Mycoplasma Kit from Qiagen has therefore been included in this benchmark analysis. However, it is not as comprehensive as our method, as it does not include a DPC. The availability of documentation provided with kits is another important factor to consider when selecting kits. In this respect, apart from Minerva Biolabs, Roche, and Qiagen, most kits offer limited transparency, making investigations difficult in the event of out-of-specification results.

Table 2.

qPCR-based kits benchmarking

Method origin		Sanofi In-house	Sigma LookOut	Minerva Biolabs Venor GeM qEP	Roche MycoTOOL	Applied Biosystems MycoSEQ	Applied Biosystems MycoSEQ plus	Qiagen QiAcuity mycoplasma
Technical characteristics	detection	qPCR hydrolysis probe	qPCR hydrolysis probe	qPCR hydrolysis probe	qPCR hydrolysis probe	qPCR SYBR Green	qPCR hydrolysis probe	digital PCR hydrolysis probe
	channels	FAM/HEX/Cy5 +/−ROXa	FAM/HEX	FAM/HEX	FAM/HEX	FAM	FAM/VIC/NED + ROX	FAM/HEX
	apparatus compatibility	all	all	all	touchdown PCRb	all	NED mandatory	all
	transparencyc	+++	+	+++	+++	+	+	+++
Regulatory	pharmacopoeia compliance	EuPh 2.6.7 USP<63> JP18	no	EuPh 2.6.7 JP18	EuPh 2.6.7 USP<63> JP18	EuPh 2.6.7	EuPh 2.6.7 USP<63> JP18	EuPh 2.6.7 USP<63> JP18
	LOD	<10CFU	NA	<10CFU	<10CFU	<10CFU	<10CFU	<10CFU
Controls	IPC	yes	yes	yes	yes	no	yes	yes
	DPC	yes	no	no	no	yesd	yes	no
	PCe	NA	yes	yes	yes	NA	NA	yes
	UNG	yes	yes	yes	yes	no	no	no
Competitiveness	pricef	100€	3,320€g	1,200€	2,900 €	3,900€	4,100€	2,100€
Conclusion	scoreh	9	4	6	5	2	6	5
	decisioni	implementation ongoing	–	backup	in use	–	–	–

^aRox is a fluorescent dye present in the qPCR MasterMix and used to normalize all other fluorescence signals. Although the use of ROX is possible with the Sanofi in-house method, it has not been described in this work.

^bThe device must be able to perform an annealing temperature gradient in the course of the PCR.

^cTransparency ranking was based on the nature of documentation provided with the kit from the most (+++) to the less (−) documented. The items taken into account for such ranking were: general technical documentation, validation report, list of mycoplasma, bacteria, and other organisms or reference DNA standards used to develop the method and scientific publications related to the method.

^dThe DPC is detected in the same channel as that used for the detection mycoplasma. Discrimination between sample contamination with mycoplasma or with the discriminatory control requires amplicon melting curve analysis.

^eThe PCs included in the kits generate signal in the mycoplasma-specific channel. Discrimination between sample contamination with mycoplasma or with the discriminatory control requires amplicon analysis by electrophoresis.

^fPrices are indicated for 100 reactions, as published on the manufacturers’ website (as of July 2025). Regarding the in-house method, the price has been calculated from each reagent purchased independently.

^gThis kit is sold at 830€ but contains reagent for 25 PCR reactions.

^hThe scoring was calculated by attributing one point to each item as follows: hydrolysis-probe-based qPCR; method compatibility with most qPCR devices; thorough documentation; method compliance to at least EuPh and LOD<10; presence of an IPC; presence of UNG. Three points were attributed to methods with a DPC detected in a fluorescence channel distinct from that used to detect mycoplasma contamination.

ⁱPractice for non-GMP samples at Sanofi R&D, Vitry-sur-Seine, France.

Besides technical specificities, compliance to regulatory guidelines is a key criterion when kits are intended to be used in a GMP environment. While not all kits claim compliance to the three most relevant pharmacopoeias for the drug market, most report an LOD below 10 CFU/mL required by all three pharmacopoeias.^25,26,27

The use of additional reagents, although not directly contributing to mycoplasma detection, is also crucial for demonstrating method specificity and handling accuracy. While most commercial kits come with an IPC and a PC, the value-added of our method lies on the presence of a DPC in which signal is detected on a separate channel from that specific for the detection of mycoplasma contamination. This helps prevent false-positive results at the level of the qPCR readout in case of handling errors and accidental contamination of samples with the DPC, while other methods require additional analysis (amplicon melting curve analysis or resolution by electrophoresis).

Finally, while kits come with all reagents, they are up to 40 times more expensive than our method. In addition, they can be discontinued, which is a major concern in GMP settings (this is the case for the PromoKine Kit I/RT VariantC, which has not been included in Table 2). Overall, this led our organization to analyze in-process control samples with the in-house method.

Discussion

Biosafety testing methods as described in pharmacopoeias are continuously challenged by technical evolution; however, given the time it takes from technological breakthrough and accumulation of knowledge to acceptance from health authorities and implementation in guidelines, archaic methods might persist several years before replacement. As such, animal testing has been widely used until the so-called 3R strategy (Replace, Reduce, Refine) encouraged the use of alternative method.^63,64 In line with this initiative, the European Directorate for the Quality of Medicines & HealthCare (EDQM) and European Medicines Agency (EMA) documents dedicated to viral safety included mass parallel sequencing (MPS) as an alternative to in vivo viral testing in 2018 and 2024, respectively,^12,65 up to 16 years after the first MPS methods were described for genome sequencing.^66,67,68 Regarding mycoplasma biosafety testing, although the possibility to use NAT-based methods was granted earlier in guidelines, “historical” cell-based methods are still considered gold standard and widely used in clinical trial and marketing application dossier. Nevertheless, PCR-based method designed for mycoplasma detection by 16S rRNA gene amplification is a valuable alternative to the lengthy, laborious microbiology and cell culture imaging techniques. Not only is this method advantageous because of its competitive turnaround time (1 day, including DNA extraction, instead of >28 days for microbiology method as described in guidelines) but it also avoids the use of experimental systems favorable for mycoplasma/bacteria growth. It is therefore the best solution for method internalization at production sites.

In the course of developing the method described in this paper, we opted for qPCR previously referred to as “TaqMan PCR,” in which specificity is conferred by primer selection and hydrolysis probe design, with a highly specific readout solely based on fluorescence intensity measurements. This is in sharp contrast with classical endpoint PCR that requires amplicon resolution by gel electrophoresis and staining with intercalating agents or SYBR-Green-based qPCR. Such methods can guarantee specificity toward mycoplasma only if several sets of specific primers are used^45,69,70 or if the readout includes amplicon sequencing, in cases where a “fit-for-all” limited primer set is used, with lesser specificity, leading to amplification of DNA from species other than mycoplasma.^46,71,72 Of note, Wong-Lee et al. have developed a PCR assay that only requires two primers for the detection of mycoplasma species relevant to various pharmacopoeias.⁷³ Much work has been based on these “universal” primers to develop and in some instances validate simple mycoplasma detection PCR or qPCR-based assays^{41,42,43,44,47,48}; however, we and others have reported false-positive detection with these reagents,^41,42,44 which advocates for the use of methods that ensure two levels of specificity such as hydrolysis-probe-based PCR. Our method was thus developed to bring specificity at the 16S rRNA DNA target amplification step and through the selection of probes specific for mycoplasma 16S DNA amplicons. A first level of specificity was reached with primer design and selection as demonstrated by amplification from mycoplasma DNA (Cq = 17.79 ± 2.54) and limited amplification from Escherichia coli, H. Sapiens, and C. griseus DNA (Cq = 28.15 ± 1.38) with the final primer set (RevMyco1∗ and FwMyco7). Due to sequence similarity in the 16S rRNA genes, efficient amplification was observed with DNA from other eukaryotic species (Cq = 19.96 ± 2.63), including species that should not give rise to cross-reaction according to various guidelines.^25,26,74 As highlighted by others, such cross-reaction can be tolerated. Indeed, some bacteria grow in anaerobic conditions and would not proliferate in mammalian cell cultures. Ambient bacteria would not proliferate either in cell cultures maintained in the presence of antibiotics. Also, because samples are tested for sterility, a positive mycoplasma test can be confidently attributed to mycoplasma contamination if the sterility test returns negative results.^41,44,72 While these are sound reasons, antibiotics tend not to be used in the biologics production phase and even not at all for cell and gene therapy products. Set aside the fact that sterility testing takes longer than PCR, the mycoplasma test accuracy should not rely on information provided by any other test, which is what pharmacopoeias expect. A second level of specificity was therefore added to our assay with the use of hydrolysis probes. Careful in silico design and experimentation resulted in a PCR assay with efficient detection on all relevant mycoplasma species (Cq = 25.18 ± 2.65) and no cross-reaction on DNA from the other species tested in this study. This specificity required, however, the use of two hydrolysis probes in a single PCR reaction (SMyco5bisLNA and SAcho1LNA) in order to accommodate the genetic variations observed in A. laidlawii sequences compared to those of other mycoplasma species. Nevertheless, this is a major achievement over hydrolysis-probe-based existing qPCR often relying on the use of many primers and probes.^49,50 Of note, while our method was tested on reference DNA and mycoplasma 10CFU standards, its specificity was confirmed on laboratory samples contaminated with bacteria that tested negative with our mycoplasma qPCR assay but generated amplicons that sequencing pointed at possible contamination with M. luteus.

Besides specificity, mycoplasma detection assays must comply to the LOD mentioned in guidelines. Some published work express LOD in mycoplasma target gene copy number/mL.^{43,49,51,70,71,72} While this is the most accurate way to document the LOD, given the accuracy of DNA quantification, LOD must nevertheless be documented in CFU/mL. This is a major challenge in technology development given the variability of the CFU/genome ratio depending on the mycoplasma culture preparation and growth conditions for colony numbering. It is therefore important to use state of the art microbiology methods to prepare mycoplasma colony standards, otherwise the determination of the LOD in CFU/mL can be biased. For this reason, the LOD of the method described in this study followed a two-step approach. First, spiking studies were performed with mycoplasma DNA reference standards, which allowed for the investigation of a wide range of DNA concentration and the determination of the LOD expressed in target gene copy number/mL. Then, 10 CFU mycoplasma reference standards prepared in specialized labs were analyzed, and the results expressed in genome copies were used to calculate an LOD expressed in CFU/mL. The use of CFU reference standards guarantees accuracy for LOD determination, which is below 10 CFU/mL for all mycoplasma organism tested with our method and complies with guidelines. While two primers and two FAM probes suffice to guarantee both specificity and sensitivity, our method has been complemented to include a set of reagents dedicated to quality control. First, sample treatment with UNGase, as suggested in the Japanese pharmacopoeia, has been added to the process in order to degrade ambient amplicons and prevent false-positive events. This is of particular importance in QC labs where PCR is used extensively. Second, the PCR master mix has been supplemented with a specific IPC containing the Cy5 hydrolysis probe target (see below). The Cy5 probe must generate Cq at the expected value in all samples, including the positive and negative controls, as a prerequisite to the interpretation of the mycoplasma PCR results. Third, a DPC recognized by both FAM probes and an additional HEX hydrolysis probe is added to the positive control and should generate signals in both channels for this sample only to confirm the integrity of the assay. Others have also included exogenous PCR targets for quality control purposes; however, these are often plasmid constructs or genomic DNA.^{41,42,43,44,45,49,50,71,72} Our method relies on the use of gBlocks instead which is a less complex matrix than genomic DNA that can be easily synthesized at the appropriate quality grade and, unlike plasmids, does not require GMO filing. Furthermore, it is not subjected to super coiling that might affect PCR efficiency. Such reagents may be referenced as a PCR kit component along with all other primers and probes at the manufacturer’s, in order to ease sourcing and operations when performing routine mycoplasma PCRs.

Although the use of fluorescent probes ensures mycoplasma specificity at the detection level, our method leaves some flexibility at the amplification level. When coupled with bench-size sequencing devices, amplified material may be characterized to determine the nature of contamination not revealed by the method’s probes. Our method therefore offers detection and characterization capacity with limited laboratory footprint in terms of equipment.

As mentioned above, our method relies on the use of two primers and two fluorescent probes for the detection of mycoplasma. It also requires IPC and DPC gBlocks together with their specific probes. Although these controls may add complexity to the procedure, they ensure the validity of the results. Based on practice, process knowledge, and quality requirements, such controls may be omitted to prioritize operational simplicity. On the contrary, should this method be used in regimented conditions such as in GMP, we strongly recommend not to omit such controls. In addition, operators should be trained appropriately, and the method should be fully validated according to the European Medicines Agency guideline dedicated to analytical procedures validation and also described in scientific publications dedicated to this topic.^{41,42,47,48,59} The fact that qPCR is already present in QC labs, operator training for this method should not be a limiting factor for implementation. Regarding method validation, specificity, range, accuracy, and robustness should be demonstrated on product-specific substrates. Given the high attrition rate of investigational medicinal products from development to commercialization, a product-specific validation early on could be a waste of time. One strategy could consist in using the method described in this work as is, without product-specific validation, to mitigate risks during early-stage development. Product-specific validation could then be conducted during late-stage development so that it can be used to characterize GMP banks and unprocessed bulks as required in clinical dossiers, marketing application, and during the commercial phase.

Materials and methods

Overall strategy

We have developed a qPCR assay capable of discriminating the presence of microorganisms belonging to the mycoplasma genera from other microbes or eukaryotic cells, in cell culture samples. Our approach first consisted in aligning genomes from mycoplasma and other species to identify the most relevant DNA region to be targeted for amplification and detection by qPCR. DNA sequences within 16S rRNA genes were retained, with sufficient homology across mycoplasma species, yet distinct from similar sequences found in other bacteria and eukaryotic species. The best primer sets were then selected based on their ability to amplify this region by SYBR-Green-based qPCR. Next, hydrolysis probes were designed to ensure specific detection of mycoplasma-derived amplicons but not amplicons derived from bacteria or eukaryotic organisms. Finally, the LOD and LOQ of the method were evaluated on not only standard reference DNA but also standard mycoplasma samples of known CFU concentrations, and the method was applied on experimental samples.

Sequences

The mycoplasma-specific PCR method setup was based on the analysis of genomes of interest listed in Table S1. Species and strains were chosen based on recommendations listed in the European, US, and Japanese pharmacopoeia.^25,26,27 For M. hyorhinis, sequences from strains BTS7 and DBS1050 were used as recommended by the European pharmacopoeia. For the BTS7 strain, two genomes of reference are listed because of cross-referencing ambiguities between ATCC#, NCTC#, and NCBI ID and the absence of a clear link between all sources. Of note, these two sequences are identical. Similarly, the two sequences listed for E. coli strain B are identical within the region of interest of the 16S rRNA gene. Sequences of the 16S and 23S rRNA genes were retrieved from SILVA and NCBI websites (www.arb-silva.de and www.ncbi.nlm.nih.gov).

Sequence analysis was performed with Vector NTI, Snap Gene, and Geneious software. For primer thermodynamics calculation (Tm, primer dimer formation, and hairpin Tm), PrimerExpress 3.0, Geneious, and the IDT online software (http://eu.idtdna.com/calc/analyzer) were used with the following settings: PrimerExpress, standard settings; Geneious, Monovalent ions 50 mM, Divalent ions 3 mM, Primer 200 nM, dNTP 0.8 mM; IDT, parameter set, qPCR, Monovalent ions 50 mM, Divalent ions 3 mM, Primer 200 nM, dNTP 0.8 mM. Similar parameters were used for probe thermodynamics calculation.

Biological samples and standards

Lyophilized mycoplasma genomes, some of which being qPCR standards, listed in Table S1 were purchased from Minerva Biolabs (Berlin, Germany), Thermo Fisher (Waltham, MA), and Roche (Basel, Switzerland) and used in qPCR reactions to generate calibration curves. DNA samples from microbial species and H. sapiens (Table S1) were also acquired as lyophilized genomic DNA while C. griseus was extracted from CHO cells (Sanofi proprietary). DNA samples were diluted to reach a concentration of 4,000 and 3,000 genome copies/μL for mycoplasma and all other species, respectively. Equivalence to genome copies was calculated based on genome size and ploidy state, and samples were used for primer characterization and method specificity assessment. Standards in the form of lyophilized mycoplasma colonies with known CFU content were also purchased, resuspended in 100 μL of water, and used in qPCR reaction following heat and freeze cycles, to determine the method LOD and LOQ. The method was also challenged with U2OS cell cultures used in a specific research program⁷⁵ and cultures from a Sanofi proprietary CHO cell line derived from a CHO-K1 clone (ECACC, UK).

Internal and discriminatory positive control gBlocks

Two gBlocks were synthesized with structures described in Figure S1. The IPC (internal positive control) is added to all samples and contains a sequence derived from the TraA gene (E. coli) specifically recognized by the Cy5-labeled StraA probe (5′ CGCAGTGATCCCGGCTGCTG 3′, synthesized with an IabFQ quencher). This sequence, which is flanked by sequences targeted by the mycoplasma PCR primers, generates positive signals in the Cy5 channel in all samples. It serves as PCR positive control and reports the presence of potential PCR inhibitors. The DPC construct is used in place of the test article in a separate reaction. It contains a sequence derived from a modified sequence of the AtPSY gene (Arabidopsis thaliana, RefSeq: GCF_000001735.4) specifically recognized by the HEX-labeled SAtPSY probe (5′ CCGACTCGCAGAACCGGAACGA 3′, synthesized with an IabFQ quencher) and DNA sequences specifically recognized by the FAM-labeled mycoplasma-specific probes, flanked by sequences targeting the mycoplasma PCR primers. This construct generates positive signal in both the mycoplasma probe FAM channel, which confirms the presence of mycoplasma-specific reagents in the PCR mix, and the SAtPSY probe HEX channel. Any HEX signal detected in test articles would demonstrate sample contamination with this construct. The TraA and AtPSY genes were chosen because of the phylogenetic distance between the A. thaliana and E. coli organisms and the cells that will be characterized with this method (mostly human and murine cells). Of note, the IPC and DPC have a design that fits other microorganism detection (not described in this study). Should IPC and DPC be designed for mycoplasma detection only, their structure should at least retain the sequences underlined in Figure S1. Caution should be given to the addition of extra sequences, however, to make sure amplicons generated from the IPC/DPC can be distinguished from amplicons generated in case of test article contamination with mycoplasma. Both constructs should also have comparable length to ease copy number calculation from masses. Finally, the IPC/DPC sizes and DNA extraction/purification methods should be compatible, particularly if such methods have purification yields that vary depending on DNA molecule size.

qPCR reactions

For SYBR-Green-based PCR product detection, each reaction mix consisted in 10 μL of Luna Universal qPCR Master Mix (NEB), 250 nM of each forward and reverse primers, and 5 μL of test article, in a final volume of 20 μL. Cycling was run as follows: 1 cycle of 90°C for 1 min, 45 cycles of 90°C for 15 s and 60°C for 30 s, and 1 cycle of 60°C–95°C by increment of 0.5°C to generate the melting curve. For hydrolysis-probe-based PCR, each reaction mix consisted in 10 μL of Luna Universal Probe qPCR Master Mix, 400 nM of each forward and reverse primers, 200 nM of FAM-labeled mycoplasma-specific hydrolysis probe, and 5 μL of test article, in a final volume of 20 μL. Cycling was run as follows: 1 cycle of 90°C for 1 min and 50 cycles of 90°C for 15 s and 60°C for 30 s. In cases where PCRs were performed with two mycoplasma-specific probes, the second probe was also added at a final concentration of 200 nM. For final use and method qualification, the PCR mix described above was supplemented with 0.2 U of UNGase (Cod UNG, ArcticZymes), 200 nM of each SAtPSY and StraA probes, and 200 copies of the IPC and the PCR performed on cell suspension (1 × 10⁶ cells) following DNA extraction (MVP DNA extraction kit used on the KingFisher Flex 96 DNA extraction automate). A PCR is also performed with 2,000 copies of the DPC in place of the test article. Cycling conditions were as follows: 1 cycle of 40°C for 2 min for UNG activation, 1 cycle of 90°C for 1 min, and 45 cycles of 90°C for 15 s and 60°C for 30 s. The Cq thresholds were set through the equipment automatic setup. In case of ambiguity, thresholds were reexamined and set manually to avoid false positive.

All primers, probes, and gBlocks were purchased from IDT. Quantitative PCR reactions were performed on BioRad CFX96 cycler and analyzed with the BioRad CFX Maestro v 2.3 software with “Target Names”: FAM, HEX, and Cy5 activated.

LOD and LOQ determination

DNA standards were serially diluted as described in the text and used in qPCR to establish a unique standard curve for each standard tested, with the following equation, where CN is the target genome copy number per reaction, Cq is the qPCR result, and a and b are the slope and the y-intercept, respectively:

$$Cq=a·{\mathit{Log}}_{10}\left(CN\right)+b$$ (Equation 1)

This set of data was also used to determine the method LOQ expressed in copy number per reaction (LOQ_CN/reaction) by using the lowest standard DNA dilution giving measurable Cq (Cq_lowest) and Cq SD < 5%. This value was calculated as follows with Equation 1:

$${LOQ}_{\raisebox{1ex}{$CN$}\!\left/ \!\raisebox{-1ex}{$reaction$}\right.}={10}^{\frac{{Cq}_{lowest}-b}{a}}$$ (Equation 2)

The qPCR was performed in the same run on the 10 CFU mycoplasma standards and returned standard specific values expressed as Cq_{10CFU standard}. This value was used in Equation 1 to calculate the corresponding 10 CFU standard gene copy number (CN_{10 CFU standard}) as follows:

$${CN}_{10CFUstandard}={10}^{\frac{{Cq}_{10CFUstandard}-b}{a}}$$ (Equation 3)

The LOQ expressed in CFU/reaction (LOQ_CFU/reaction) was then calculated from the CN_{10 CFU standard} by taking into account the known CFU number in the diluted standards (5 μL of the 10 CFU standard sample resuspended in 100 μL, which represents 0.5 CFU/reaction) and the LOQ expressed in copy number per reaction (LOQ_CN/reaction) calculated in Equation 2, as follows:

$${LOQ}_{\raisebox{1ex}{$CFU$}\!\left/ \!\raisebox{-1ex}{$reaction$}\right.}=0.5·\frac{{LOQ}_{{\displaystyle \raisebox{1ex}{$CN$}\!\left/ \!\raisebox{-1ex}{$reaction$}\right.}}}{{CN}_{10CFUstandard}}$$ (Equation 4)

The LOQ expressed in CFU/mL (LOQ_CFU/mL) was further calculated by taking into account a maximum sample volume of 5 μL as follows:

$${LOQ}_{\raisebox{1ex}{$CFU$}\!\left/ \!\raisebox{-1ex}{$mL$}\right.}=\frac{{LOQ}_{{\displaystyle \raisebox{1ex}{$CFU$}\!\left/ \!\raisebox{-1ex}{$reaction$}\right.}}}{{5.10}^{-3}}$$ (Equation 5)

Finally, the LOD expressed in CFU/mL (LOD_CFU/mL) was calculated based on the most diluted experimental point with measurable Cq, which is one 10-fold dilution factor below the LOQ:

$${LOD}_{\raisebox{1ex}{$CFU$}\!\left/ \!\raisebox{-1ex}{$mL$}\right.}=\frac{{LOQ}_{{\displaystyle \raisebox{1ex}{$CFU$}\!\left/ \!\raisebox{-1ex}{$mL$}\right.}}}{10}$$ (Equation 6)

Of note, because the experiments are performed on qPCR standard DNA and 10CFU standard samples for each microorganism, this calculation method does not require to take into account 16S rRNA gene duplication across mycoplasma species.

Sequencing

Amplicon sequencing was performed on the MinION device (Oxford Nanopore Technologies, Oxford, UK) connected to a Linux platform, and live basecalling was performed by an Nvidia RTX8000 GPU. Libraries were prepared alternatively with the Rapid Barcoding Sequencing Kit (SQK-RBK004) or the Ligation Sequencing Kit (SQK-LSK109) + Native Barcoding Kit (EXP-NBD104) according to the instructions of the manufacturer. Seventy-five microliters of the barcoded sample libraries were then loaded on two different R9 MinION flow cells and sequenced for 10 h with MinKNOW v21.02.1 (standard settings). At the end of the run, sequences were blasted against the 16S rRNA gene library of the organisms listed in Table S1 with the Geneious Prime v2022 software to identify the species from which amplicons derived.

Data availability

All raw data can be made available upon request.

Acknowledgments

We acknowledge Christine DeMaria and Henry Lin at Sanofi for critical review of the manuscript.

Author contributions

S.D.S. designed the experiments. S.D.S., E.L., and B.P. conducted the experiments. S.D.S. and S.B. analyzed the data. S.B. wrote the paper.

Declaration of interests

S.D.S., E.L., and S.B. are full-time employees of Sanofi and may hold shares and/or stock options in the company. Sanofi has filed a patent application (PAT23301 WO 2025/114394 A1) covering the method reported here. S.D.S. is inventor.