Implementation and design of a whole-cell SELEX process for the development of aptamers specific to A. niger conidia
A whole-cell SELEX approach was applied to conidia of A. niger as illustrated in Fig. 1. Three main changes were introduced to a previously published protocol used to select aptamers against bacterial cells (Kolm et al., 2020): the overnight precipitation in ethanol to recover the bound-ssDNA after each SELEX round, an additional test-PCR (dnc2) for the determination of the number of cycles and the negative selection performed with A. tubingensis before sequencing. Compared to a previous study where aptamers against three different Aspergillus species were generated with a toggle approach (Seo et al., 2021), here we aimed at the in-vitro selection of species-specific aptamers. Advanced tools (qPCR and melting curves) were applied to quantify and monitor the diversity of the recovered ssDNA, while next-generation sequencing was used to identify potential aptamer candidates binding to fungal conidia.
In the first round, conidia of the two A. niger strains ATCC 1015 and CBS 554.65 were incubated with a total of 1015 molecules of the FAM-labelled ssDNA library. Upon incubation, the unbound sequences were removed by centrifugation and washing. The bound sequences were recovered by elution and subsequent precipitation in ethanol to remove the conidia from the recovered ssDNA before PCR amplification. This step was introduced due to the large amounts of PCR inhibitors present in A. niger conidia, among which melanin, that would otherwise strongly interfere with the amplification reaction (Eckhart et al., 2000; Fraczek et al., 2019; Yuan et al., 2023). After recovery, the precipitated ssDNA was subjected to two independent test-PCRs (dnc1 and dnc2) for the determination of the optimal number of cycles for enrichment PCR. This is crucial for the subsequent efficient ssDNA generation via lambda exonuclease. The optimal number of cycles is defined as the number of cycles at which the highest PCR product yield can be obtained without generating by-products (Wang et al., 2019). Recovered ssDNA was subsequently amplified by enrichment PCR with the selected number of cycles and single-stranded DNA generated from it with a lambda exonuclease enzyme, before being subjected to another round of in-vitro selection. In total, 3 independent whole-cell SELEX experiments were performed, each consisting of a total of 9 rounds. The selection conditions applied in each round are reported in Tables 1, 2 and 3. In general, the conditions were rendered more stringent over the rounds by decreasing the amount of input DNA, increasing the number of washes or adding competitors to the reaction. To increase the species-specificity of the enriched sequences, counter-selection was performed starting from round 4 until round 8. In this case, the recovered ssDNA was incubated first with A. tubingensis and the unbound sequences were then recovered by centrifugation and incubated with A. niger. This allowed to preferentially enrich sequences that only bind to A. niger and do not recognize conidia of its close relative A. tubingensis. In all three SELEX experiments, round 9 was performed in parallel against both target (A. niger; R9-N) and non-target species (A. tubingensis; R9-T) using aliquots of the ssDNA pool recovered after round 8. This allowed to assess unspecific binding to A. tubingensis and potential PCR bias introduced during the in-vitro selection. Additionally, a positive selection against A. niger with counter-selection against A. tubingensis was performed in SELEX-3 (R9 in Table 3).
Table 1
Selection conditions applied in SELEX-1.
Whole-cell SELEX-1 |
Round | ssDNA | Positive selection | Reaction volume (µL) | Competitors | Washes | Counter-selection |
| pmol | Final conc. (nM) | Target strains | Number of conidia | | | | |
1 | 1800 | 7200 | ATCC 1015, CBS 554.65 | 107 | 250 | - | 1 | - |
2 | 25 | 100 | ATCC 1015, CBS 554.65 | 107 | 250 | - | 1 | - |
3 | 10 | 100 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 2 | - |
4 | 10 | 100 | ATCC 1015, CBS 554.65 | 107 | 100 | - | 2 | A. tubingensis MA 3973 |
5 | 10 | 100 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 3 | A. tubingensis MA 3973 |
6 | 10 | 100 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 3 | A. tubingensis MA 3973 |
7 | 10 | 100 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 3 | A. tubingensis MA 3973 |
8 | 10 | 100 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 5 | A. tubingensis MA 3973 |
9-N | 10 | 100 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 6 | - |
9-T | 10 | 100 | A. tubingensis MA 3973 | 107 | 100 | BSA, 4sDNA | 6 | - |
Table 2
Selection conditions applied in SELEX-2.
Whole-cell SELEX-2 |
Round | ssDNA | Positive selection | Reaction volume (µL) | Competitors | Washes | Counter-selection |
| pmol | Final conc. (nM) | Target strains | Number of conidia | | | | |
1 | 1800 | 7200 | ATCC 1015, CBS 554.65 | 107 | 250 | - | 1 | - |
2 | 25 | 100 | ATCC 1015, CBS 554.65 | 107 | 250 | - | 1 | - |
3 | 25 | 100 | ATCC 1015, CBS 554.65 | 107 | 250 | BSA, 4sDNA | 2 | - |
4 | 10 | 100 | ATCC 1015, CBS 554.65 | 107 | 100 | - | 2 | A. tubingensis MA 3973 |
5 | 8 | 100 | ATCC 1015, CBS 554.65 | 107 | 80 | BSA, 4sDNA | 3 | A. tubingensis MA 3973 |
6 | 8 | 100 | ATCC 1015, CBS 554.65 | 107 | 80 | BSA, 4sDNA | 3 | A. tubingensis MA 3973 |
7 | 8 | 100 | ATCC 1015, CBS 554.65 | 107 | 80 | BSA, 4sDNA | 3 | A. tubingensis MA 3973 |
8 | 8 | 80 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 6 | A. tubingensis MA 3973 |
9-N | 8 | 80 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 6 | - |
9-T | 8 | 80 | A. tubingensis MA 3973 | 107 | 100 | BSA, 4sDNA | 6 | - |
Table 3
Selection conditions applied in SELEX-3. *ssDNA applied to round 6 was derived from a mixture of dsDNA obtained from the amplification of the ssDNA recovered after round 5 and amplified from the already amplified ssDNA.
Whole-cell SELEX-3 |
Round | ssDNA | Positive selection | Reaction volume (µL) | Competitors | Washes | Counter-selection |
| pmol | Final conc. (nM) | Target strains | Number of conidia | | | | |
1 | 1800 | 7200 | ATCC 1015, CBS 554.65 | 107 | 250 | - | 1x | - |
2 | 25 | 100 | ATCC 1015, CBS 554.65 | 107 | 250 | BSA, 4sDNA | 1x | - |
3 | 23 | 92 | ATCC 1015, CBS 554.65 | 107 | 250 | BSA, 4sDNA | 2x | - |
4 | 23 | 92 | ATCC 1015, CBS 554.65 | 107 | 250 | BSA, 4sDNA | 2x | A. tubingensis MA 3973 |
5 | 9.2 | 92 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 3x | A. tubingensis MA 3973 |
6* | 9.2 | 92 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 3x | A. tubingensis MA 3973 |
7 | 9.2 | 92 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 3x | A. tubingensis MA 3973 |
8 | 9.2 | 92 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 5x | A. tubingensis MA 3973 |
9 | 9.2 | 92 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 5x | A. tubingensis MA 3973 |
9-N | 9.2 | 92 | ATCC 1015, CBS 554.65 | 107 | 100 | BSA, 4sDNA | 5x | - |
9-T | 9.2 | 92 | A. tubingensis MA 3973 | 107 | 100 | BSA, 4sDNA | 5x | - |
When using lambda exonuclease for the generation of ssDNA, it is important that only full-length double-stranded products (homoduplexes) are generated during enrichment PCR, as heteroduplexes cannot be efficiently digested by the enzyme. Due to the high sequence heterogeneity characterizing the utilized random ssDNA library, there is a risk of forming by-products (heteroduplexes) and introducing biases during the PCR (Tolle et al., 2014; Kohlberger and Gadermaier, 2022). A commonly used method to avoid formation of heteroduplexes is the determination of the optimal number of cycle prior enrichment PCR (Sefah et al., 2010). To this end, an aliquot of the recovered ssDNA was subjected to two test-PCRs: dnc1 (Fig. 2A) and dnc2 (Fig. 2B). The amplification profile of the ssDNA recovered after round 3 of SELEX-3 is illustrated in Fig. 2A as an example. The fluorescence signal increases over the amplification cycles, before reaching a maximum at cycle 18. The subsequent decrease in fluorescence, previously described as “hook effect”, corresponds to the formation of heteroduplexes (Warton et al., 2020). Heteroduplexes, which start to form after depletion of the primers, generally have a lower melting temperature than the full-length PCR products (homoduplexes), as they are composed of only partially complementary sequences. If their melting temperature is lower than the temperature at which the fluorescence signal is measured, they will be dissociated during the measurement. This is reflected in a lower fluorescent signal which causes the hook effect (Warton et al., 2020). In a previous study, this test-PCR only was sufficient to determine the optimal number of amplification cycles, corresponding to the number of cycles before the peak (Kolm et al., 2020). In this study, three reactions were performed in an additional test-PCR (Fig. 2B) with the three subsequent number of cycles before the peak (15, 16 and 17 in the example of Fig. 2A). This allowed to obtain an independent confirmation of the absence of heteroduplexes after the enrichment PCR. In the case reported in Fig. 2B, the cycle right before the peak (17) corresponded to the start of heteroduplex formation, visible as a shorter product on the gel, while after 16 cycles only the specific product (86 bp, homoduplexes) was visible. This led to the decision of using 16 cycles to perform amplification of the recovered ssDNA.
Monitoring the amount and the diversity of the recovered ssDNA during the SELEX process
Advanced tools were applied to monitor the in-vitro selection process allowing to precisely quantify the amount by qPCR (Fig. 3) and measure changes in the diversity by melting curves of the recovered ssDNA after each round (Fig. 4). Round 1 was excluded from these analyses as all of the recovered ssDNA was amplified for further processing.
Quantification of the recovered ssDNA was performed by qPCR. With this method, absolute recovered DNA quantities can be determined with high sensitivity (Avci-Adali et al., 2013). Differences in the amount of recovered ssDNA could be observed between different rounds, with concentrations ranging from 107 to 109 molecules/µL.
The increase in the amount of bound DNA was reported in literature as an indicator of successful sequence enrichment (Kohlberger and Gadermaier, 2022). However, the amount of recovered DNA does not only depend on the enrichment of certain sequences but also on the selection conditions applied at each round and on the specificity and accessibility of conidia surface targets to the binder sequences present in the ssDNA pool. The decrease of recovered DNA measured at round 4 of SELEX-1 and SELEX-2 might be due to the counter-selection, which was applied starting from this round. However, a similar effect is not visible in SELEX-3, indicating that it is likely a combination of factors, rather than one factor only, to contribute to the number of bound sequences. Moreover, although all the three SELEX experiments were initiated with the same ssDNA library, the sequences randomly present in each aliquot were not the same and most likely led to different enrichment patterns. Round 5 of SELEX-3 yielded a very low amount of DNA (Fig. 3A) and in order to obtain enough DNA to continue with round 6 of the SELEX process, additional DNA was obtained by the dilution and further amplification of the already amplified PCR product, introducing a bias in the selection.
When the same ssDNA pool was applied to A. niger (R9-N) or to A. tubingensis (R9-T), higher concentrations of recovered ssDNA were measured for R9-N in all SELEX experiments. This suggests successful enrichment of A. niger-specific sequences.
A more effective method to monitor sequence enrichment is based on the analysis of the melting curves, performed on the recovered DNA after PCR amplification (Vanbrabant et al., 2014; Kolm et al., 2020). Melting curves of the three SELEX experiments are reported in Fig. 4. Melting curves allow to monitor the formation of homoduplexes, derived from the annealing of two complementary strands of a PCR products. At the beginning of selection, homoduplexes are rare, as most of the sequences are unique. However, if sequence enrichment is successful, homoduplex formation can be observed as a distinct melting peak at around 82°C in the melting profile. Distinct melting peaks started to appear in rounds 7 of SELEX-1 and SELEX-2 and in round 5 of SELEX-3, indicating a decrease in sequence diversity and the appearance of enriched sequences. Melting peaks increased further in subsequent rounds, suggesting further enrichment. However, while this increase appeared gradual in SELEX-1 and SELEX-2, it was abrupt between rounds 5 and 6 of SELEX-3. This is most likely due to the PCR bias introduced in this experiment which led to the loss of sequences present in low abundance while those present in higher copies had a higher chance to be amplified and carried over to the next round. Different peak shapes correspond to changes in nucleotide composition of the analyzed pool (Kolm et al., 2020). Based on the evolution of the melting peaks, the selection was stopped after nine rounds. In SELEX-3, the highest melting peak was reached at round 8, suggesting a loss of potential binders at round 9. Interestingly, melting peaks of round 9-T (selection against A. tubingensis in round 9) were higher than those of round 9-N.
NGS data analysis and selection of aptamer candidates
Sequencing data were processed with the previously developed Aptaflow script (Kolm et al., 2020). Graphs showing the sequence enrichment in recovered ssDNA pools over the subsequent SELEX rounds (Fig. 5) and a list of the 1,000 most enriched sequences for each round were generated. The total count of individual sequences, representing sequence enrichment, increased during the subsequent rounds in all three performed SELEX experiments, reaching a peak at round 9 in SELEX-1 and SELEX-2 and at round 8 in SELEX-3. The sequencing data confirmed the changes in diversity observed in the melting curves, highlighting the power of combining these two techniques to determine how many SELEX rounds to perform and which rounds to sequence. As already observed in the melting curves, the increased count of enriched sequences in round 9-T (negative selection, performed with A. tubingensis instead of A. niger) might indicate that the incubation of the aptamers with A. tubingensis after selection with A. niger led to a loss of diversity of specific enriched sequences. Only a few sequences were retained by A. tubingensis conidia and these had a higher chance to be amplified at higher rates. This phenomenon is reflected as an apparent increase in sequence enrichment, similar to what observed in round 6 of SELEX-3, which, however, does not correspond to an increase of binder molecules. Therefore, performing negative selection and subsequent analysis of the sequences enriched in such an unspecific round can be a valuable strategy to more easily identify potential binders enriched in the target rounds as well as to remove unspecific sequences from the potential binding candidates.
The first selected aptamer candidates were identified from round 8 of SELEX-3, as melting curve analyses, as well as the sequencing data, showed the highest enrichment during this round. Additional 8 potential aptamer candidates were identified from rounds 9 of SELEX-1 and SELEX-2. Selection was performed on sequences belonging to different clusters, by ranking them based on their prevalence at the selected round and their appearance in earlier rounds. Sequences AN03-R8-AN435, AN03-R9-N-AN070, AN01-R9-095, AN01-R9-105, AN01-R9-115, AN02-R9-099 and AN02-R9-185 were selected because present with higher reads in the positive selection round (R9-N) than in the negative selection round (R9-T). AN03-R8-AN156 was selected as negative control as it showed higher read counts in round 9-T than in any of the rounds performed with A. niger conidia as targets. The minimum free energy and the secondary structure of the selected sequences were predicted using the online tool RNAFold 2.5.1. All 18 selected aptamer candidates with their characteristics and reason for selection are listed in Supplementary Table 1.
Aptamer identification and impact of the FAM label on the aptamer binding
The selected aptamer candidates were screened for their capability to bind to A. niger conidia by performing binding assays and subsequent quantification of the recovered ssDNA by qPCR.
In a first screening experiment, ten of the selected candidates were ordered with a FAM-label at the 5´end (Fig. 6). The quantified DNA was compared to three negative controls: the starting ssDNA library, a labelled random negative control (BA-NC-1) and the AN03-R8-AN156 sequence.
The ssDNA recovered after most of the binding assays was similar to the amount of ssDNA recovered after incubation of A. niger conidia with either the ssDNA library, the negative control BA-NC-1 or the negative control AN03-R8-AN156. This might be due to the PCR bias introduced in round 6 of SELEX-3, which most likely led to the enrichment of sequences which are more easily amplified but might not bind to the conidia and the concomitant loss of potential binders. However, two candidates, AN03-R8-AN435 and AN03-R9-N-AN070, showed significantly higher recovery rates compared to the negative controls, indicating that they can bind to the target conidia. As a high background could be measured for the negative controls (ssDNA library, BA-NC-1 and AN03-R8-AN156), a second screening was performed with unlabeled aptamer candidates to determine if the FAM-label had an effect on the binding process. To this end, eight aptamer candidates identified in SELEX-1 and SELEX-2 were tested in their unlabeled version and compared to the unlabeled BA-NC-1 (Fig. 7). Additionally, labelled and unlabeled versions of aptamer candidates AN01-R9-006, AN01-R9-115, AN02-R9-099 and AN02-R9-185 and of the negative control BA-NC-1 were compared (Fig. 7). Candidates AN01-R9-006, AN01-R9-115, AN02-R9-099 and AN02-R9-185 showed higher recovery rates than the negative control in both versions (labelled and unlabeled). Interestingly, labelled aptamers were associated with higher recovery rates than unlabeled ones. We confirmed that this was not an artifact due to the interference of the FAM fluorescence during the qPCR, but it rather derived from higher binding of the FAM-labelled sequences to conidia of A. niger than the unlabeled counterparts. This suggests that the FAM fluorophore itself interacts with the target cells to a certain extent. Based on the measured recovery rates, aptamers AN03-R8-AN435, AN03-R9-N-AN070, AN01-R9-006, AN01-R9-115, AN02-R9-099 and AN02-R9-185 were selected for further characterization.
Aptamer specificity to other Aspergillus species
To determine whether the selected aptamers can bind to A. niger in a species-specific manner, binding assays were performed with other two Aspergillus species, A. tubingensis and A. nidulans. Based on the qPCR results, all the selected aptamers showed to be species-specific for A. niger (Fig. 8A). DNA recovered after incubation with A. niger increased from 2.5 to 17-fold when compared to A. tubingensis and from 7 to 500-fold when compared to A. nidulans. Interestingly, the negative control (BA-NC-1) seems to bind preferentially to the conidia of A. niger than to those of the other two fungal species. This sequence was not present in the sequencing data but it was randomly generated and it is possible that it binds to a certain extent to the conidia of A. niger.
To confirm successful and species-specific binding, fluorescent measurements were performed on the eluted samples upon binding with three selected aptamers (AN02-R9-185, AN01-R9-006 and AN02-R9-099) (Fig. 8B). Additionally, to avoid the introduction of a bias due to the selection of the random sequence, other two negative controls differing in the unique internal 40 bp region (BA-NC-2 and BA-NC-3) were measured in parallel (Fig. 8B).
Fluorescent measurements confirmed the species-specific binding of the selected aptamers to A. niger conidia. Furthermore, different fluorescent values could be measured when comparing the three different negative controls. The first selected negative control (BA-NC-1) showed the highest binding to A. niger conidia. BA-NC-2 did not bind at all to target conidia and BA-NC-3 only slightly. Therefore, randomly selected sequences have the potential to bind to a certain extent to the target cells. These results highlight the importance of choosing a suitable negative control and suggest that using multiple negative controls should be preferred.
A. niger and A. tubingensis are phylogenetically closely related, belonging both to the section Nigri of the genus Aspergillus (Visagie et al., 2024). Due to their highly similar phenotype, they can be hardly distinguished based on classical morphological criteria and the use of molecular analyses is crucial for their differentiation (Samson et al., 2007; Susca et al., 2007). The capability of the DNA aptamers developed in this study to distinguish between these closely related species is of high relevance and indicates that these fungi might substantially differ in their surface proteome. These results could open the way to new strategies in the identification and characterization of closely related Aspergillus species.
Aptamer binding affinity
The binding affinity of the aptamers AN01-R9-006, AN02-R9-099 and AN02-R9-185 was determined by incubating the A. niger conidia with different concentrations of the corresponding aptamer. The binding curves were obtained by measuring fluorescence after elution (Fig. 9).
The binding curve of aptamer AN02-R9-099 did not show saturation (data not shown), indicating that this aptamer might bind non-specifically to the conidia of A. niger (Henri et al., 2019). KD values were calculated for aptamers AN01-R9-006 and AN02-R9-185. AN01-R9-006 showed a KD of 58.97 nM (95% confidence interval 42.89–81.03 nM). AN02-R9-185 showed a KD of 138.71 nM (95% confidence interval 79.65–255.51 nM). The measured equilibrium dissociation constants are in the range of aptamers previously developed against fungal conidia (Seo et al., 2021) and indicate specific binding on the conidial surface with high affinity. The binding affinity curves indicate that the aptamers interact in a concentration-dependent manner with the A. niger conidia.