Abstract

Colorectal cancer (CRC) is a prevalent and untreatable form of cancer, often caused by genetic, epigenetic, or lifestyle factors. Hypermethylation at specific CpG islands has been implicated in carcinogenesis. One gene, TLX1 (a transcription factor critical for many stages of development), stands out as a candidate to study aberrant methylation status in CRC. We hypothesized that in CRC, promoter CpG islands of TLX1 are hypermethylated. Here, we investigated the methylation status of TLX1 in CRC and the effects of decitabine treatment. We bisulfite converted DNA from HCT-116 cells, designed methylation-sensitive primers, and sequenced the amplified product. We found that decitabine significantly reduces methylation levels at individual CpG sites in the promoter region of TLX1. We also established that decitabine is effective in our HCT-116 cells using a clonogenic survival assay. Such studies, and future refinements, could help us establish biomarkers that could improve our understanding and potentially treatment of CRC. 

Keywords: colorectal cancer, DNA modifications, methylation, decitabine

Introduction

Colorectal cancer (CRC) is a highly prevalent and often untreatable disease. It is the third most common type of cancer worldwide, yet it is often caught only in its late stages [1]. Chemotherapy is not effective, and patients typically survive only 44 months after the completion of treatment [2]. Despite this pressing reality, there is still debate about the molecular mechanisms underlying this disease.  Genetic and epigenetic factors contribute to pathogenesis, but there is no consensus on a definitive molecular pathway [3]. Somatic mutations and epigenetic modifications could interact with unpredictable environmental and lifestyle factors. There is still active investigation to elucidate the molecular underpinnings of this devastating disease. 

Recently, DNA methylation (an epigenetic modification) has been heavily implicated in carcinogenesis, or at least as a biomarker for CRC. For example, one study found that hypermethylation at SFRP1 and SFRP2 could be detected in stool samples of future CRC patients [4]. Methylation is the attachment of methyl groups to cytosines of DNA, typically occurring on 5’-CG-3’ sequences [5]. Methylated regions of the genome (often CpG islands) are transcriptionally silent, suggesting that in CRC, this modification likely serves to silence tumor suppressor transcription or coincides with other carcinogenic pathways [5]. Three methyltransferase enzymes, DNMT1, DNMT3a, and DNMT3b, are responsible for methylating regions of human DNA [6]. Patterns of methylation vary within species, and over the lifetime of an organism, the distribution of methyl groups changes [7]. Thus, the dynamic nature of this epigenetic modification could influence the development of CRC, depending on the differential distribution of methylation throughout the genome. 

Methylation is a key epigenetic modification in cancer mechanisms. Multiple studies have shown that methylation status is altered in the DNA of CRC tissue samples. There is hypomethylation on the genome level but hypermethylation at CpG islands [8]. Genome-wide methylation analyses reveal that CpG island hypermethylation is common in many CRC patients, and methylation levels may inversely correlate with prognosis [9]. Higher methylation levels, especially at promoter sites, are associated with a worse prognosis. Hypermethylation may silence tumor suppressors, leading to uncontrolled cell proliferation. A double knockout of DNMT1 and DNMT3b sufficiently reduces this hypermethylation, and crucially, attenuates cell growth [10]. This connects hypermethylation with abnormal cell proliferation. Within the current decade, CpG island hypermethylation has taken a central role, and many genes have been shown to exhibit differentially methylated regions [11]. Specifically, these differentially-methylated regions are often hypermethylated at promoter sites [11].   

The gene TLX1 shows strong promoter hypermethylation at a CpG island near the 5’ end. A recent study reports that TLX1 hypermethylation was observed in all CRC tissue samples investigated and that the aforementioned DNMT1 and DNMT3b knockout significantly increased expression in CRC cell lines [11]. In a different study, 90% of CRC tumor samples exhibited hypermethylation of TLX1, compared to healthy control tissue, as indicated by a GWAS [12]. Through differential expression, TLX1 appears to have a functional role in CRC, and it could serve as a biomarker for the disease.  

TLX1 functions as a transcription factor in diverse scenarios, including embryonic development and spleen growth. Given its overarching role, it may plant the seeds for CRC early in development through aberrant epigenetic modifications [13]. TLX1 has been implicated in cancers other than CRC. For example, in leukemia, it associates with mutated NOTCH and MYC to allow abnormal transcription and aberrant cell growth [14]. The molecular mechanisms of TLX1 in CRC (and the types of epigenetic modifications that occur) remain open questions. In leukemia, TLX1 is overexpressed but in CRC, it is underexpressed, suggesting that different mechanisms may be at play. 

In this study, we use bisulfite conversion to compare the methylation status of TLX1 isolated from WT HCT-116 cells to the same gene amplified from decitabine-treated DNA. DNA from cells with the double DNMT1 and DNMT3b knockout (unmethylated) and artificially-methylated DNA are used as controls. Multiple model systems and approaches have been utilized to study colorectal cancer. The HCT-116 cell line has emerged as a powerful tool in the field. HCT-116 cells belong to a line derived from human colon carcinoma, and they have served as a platform for investigating methylation in CRC. HCT116 cells have taught us that in CRC, methylation increases at specific CpG-rich sites [11]. This cell line has shown us that a knockout of both DNMT1 and DNMT3b is needed to abolish the increase in methylation associated with cancer [10]. Interestingly, HCT-116 cells have been shown to respond to decitabine (5-aza-2’-deoxycitidine), an anticancer drug that prevents DNMT1 from methylating CpG islands [15]. 

We will amplify TLX1 using PCR primers evenly amplifying methylated and unmethylated regions of the gene. Based on the available evidence, we hypothesize that CpG islands in TLX1 will be hypermethylated in HCT-116 cells, a phenomenon reversible by decitabine. In our paradigm, methylated cytosines will remain cytosines in the resulting sequence, while unmethylated ones will be converted to thymines.

Results

Bioinformatics Search

We first used bioinformatics to analyze the methylation patterns and transcriptional properties of TLX1 in CRC. Figure 1a summarizes these results. We found that in HCT-116 cells, which model colorectal cancer, there are multiple CpG sites that are hypermethylated, as shown by reduced representation bisulfite analysis (RRBS) from ENCODE. Six of these sites fall within the promoter region of TLX1 (in Figure 1a, red lines indicate hypermethylation, while the promoter is boxed). Crucially, hypermethylation occurred in HCT-116 cells but not in NT2 cells (green lines instead of red ones). This indicates that CpG hypermethylation is specific to colorectal cancer. Our search revealed that normally, eight transcription factors (EZH2, CTPB2, ZNF263, PolR2A, Suz12, FoxA1, CTCF, and PHF8) bind to the promoter region of TLX1. However, in HCT-116 cells, PolR2 and CTCF do not bind. The loss of TF binding strongly suggests that in this cell-line, promoter hypermethylation correlates with transcriptional silencing. More interestingly, in many cell lines (such as HCT-116, HEK293, and HeLa), the lack of transcription factor binding correlates with hypermethylation. To further analyze this correlation, we searched for DNAseI hypersensitivity in our region of interest. According to UCSCS Genome Project data, HCT-116 cells do not exhibit this characteristic, suggesting that chromatin is in a closed state. Conversely, in cell lines where hypermethylation does not occur (such as HEPG2 and H1-hESC), there is hypersensitivity to DNAseI, indicating transcriptionally-active chromatin. To determine whether various epigenetic modifications (apart from DNA methylation) could account for these transcriptional changes in TLX1, we observed the methylation and acetylation patterns of histones in various cell lines, again using the UCSCS Genome Project. For HCT-116 cells, no acetylation of H3K27 was observed, but there was some methylation H3K4. Since this was only mono-methylation, we concluded that it is not enough to fully silence transcription of TLX1. Taken together, these data suggest that the drastic decrease in TLX1 expression in CRC may be due to promoter hypermethylation. This makes TLX1 a good candidate for further low-throughput exploration of the role of methylation in CRC. Supplementary Table 1 summarizes the full results of the UCSCS Genome Project search. 

To investigate the promoter of TLX1 on a low-throughput level, we designed PCR primers using MethPrimer to amplify that region of the human genome. The sequence of the amplified region is presented in Figure 1b (primer binding sites in green; CpG sites in red). The resulting amplicon is 317 base pairs long and encompasses 31 CpG sites, all of which are targets of potential hypermethylation. While the UCSCS Genome Project reports that only 6 of these sites are hypermethylated, it is important to remember that these data utilized high-throughput techniques whose site-level resolutions are limited. In this experiment, we used genomic DNA from HCT-116 cells treated with DMSO or the anti-cancer drug decitabine to compare methylation levels on the single-nucleotide level in the promoter region of TLX1. As controls, we used DNA from DKO cells (methylation-deficient) as well as fully methylated (mCpG) HCT-116 genomic DNA. Before PCR amplification, all DNA was bisulfite-converted. 

TLX1 Primer Confirmation 

To test our primers for TLX1, we ran a temperature-gradient PCR using a mixture of unmethylated and methylated DNA templates. The results are reported in Figure 1c. We found that the primers efficiently amplify the mixture of DNA (the bands are as thick as for the DapK control primers). Lower annealing temperatures (52, 56, or 60 ℃) seemed more efficient. To find the optimum annealing temperature (one at which the primers evenly amplify both DKO and mCpG DNA), we ran qPCR followed by a high-resolution melt-curve analysis (Figure 1d). While all three annealing temperatures tested seemed efficient, amplification was most even at 52 ℃. As a result, this temperature was chosen for the rest of the analysis. 

Bisulfite Treatment and Decitabine Treatment Confirmation

We next PCR-amplified the promoter region of interest on TLX1 using unmethylated (DKO) DNA and fully methylated (mCpG) DNA separately as templates. The purpose of this was to determine whether bisulfite conversion was effective. We found that for our region of TLX1, the DKO DNA (M = 33.59%, SD = 31.38%) had a lower methylation percentage than the same region amplified from mCpG DNA (M = 60.39%, SD = 38.00%): t (24) = - 3.526, p = 0.00086. We concluded that bisulfite treatment preserved most of the methylated cytosines in the mCpG template, while it converted the unmethylated cytosines to thymines in the DKO template. These data are displayed graphically in Figure 2. 

To investigate whether decitabine treatment was effective, we performed a clonogenic survival assay using a crystal violet stain of HCT-116 colonies. HCT-116 cells were treated with 0 µmol (vehicle), 0.1 µmol, 0.25 µmol, and 1 µmol decitabine for 48 hours with one media change, and the number of resulting colonies in each condition was counted after 14 days of growth. There was a statistically-significant effect of decitabine for all three drug concentrations tested, as revealed by a one-way ANOVA: F (3, 16) = 94.2, p < .0001. Tukey’s HSD post-hoc test revealed that relative to the 0 µmol condition (M = 169.3, SD = 37.2), there were fewer colonies in the 0.1 µmol decitabine condition (M = 11, SD = 8.9), the 0.25 µmol (M = 0, SD = 0), and the 1 µm (M = 0, SD = 0). These results are displayed graphically in Figure 3. They suggest that at all concentrations, decitabine was effective in killing off CRC cells. 

To confirm that decitabine effectively demethylated our genomic DNA, we performed an HpaII digest. This restriction enzyme cuts only unmethylated DNA. These data are summarized in Figure 4. There was more smearing on the gel upon treatment with both decitabine and HpaII, showing that decitabine demethylated the genomic DNA. In the control conditions, HpaII cut up the DKO unmethylated DNA, while it had no effect on the fully methylated DNA. 

Effect of Decitabine on TLX1 Methylation 

Finally, we set out to determine whether or not decitabine was effective in attenuating methylation levels of TLX1. To test this hypothesis, we PCR-amplified TLX1 from a genomic template coming from HCT-116 cells treated either with DMSO or decitabine. Methylation levels at TLX1 decreased significantly when the HCT-116 cells were treated with decitabine (M = 43.38%, SD = 44.52%) compared to DMSO (M = 80.26%, SD = 22.91%): t (24) = 3.6237, p = .0005. These data, displayed in Figure 5, suggest that decitabine may work by reducing hypermethylation levels at TLX1. Figure 6 summarizes the overall methylation trends across all conditions (a) and depicts the sites at TLX1 at which decitabine led to at least 10% less methylation compared to DMSO (b). Supplementary Figure 1 shows the raw data obtained from PCR and sequencing. 

Discussion

In these experiments, we aimed to investigate methylation of the TLX1 promoter region on a low-throughput level. Using the UCSC Genome Browser, we confirmed that TLX1 is differentially-methylated in HCT-116 cells, and that this may relate to transcriptional silencing (Figure 1). These bioinformatics analyses were consistent with the data on TLX1 presented in earlier research. Previously, it was known that when methylation is abolished by knocking out key methyltransferases, TLX1 expression in CRC lines increases dramatically [11]. Our UCSC-genome search corroborated this trend from the opposite angle – in the presence of hypermethylation, key TLX1 transcription factors fail to bind, and the decrease in expression cannot be ascribed solely to histone modifications. We concluded that TLX1 hypermethylation is worth studying in more detail. Next, we confirmed that the anticancer drug decitabine effectively kills off HCT-116 cells in culture (Figure 3). These findings fit with the trends in the literature. A recent study showed that in HCT-116 cells, decitabine is taken up by the ENT1 transporter, and its effects could be attenuated by applying NBMPR, an ENT1 inhibitor Decitabine has been previously shown to causally impact [16]. Additionally, Rhee et al. (2002) demonstrated that decitabine significantly inhibits the growth of HCT-116 cells. Our lab was also able to show that decitabine is effective in reducing the colony formation of CRC cells. Interestingly, decitabine has been shown to decrease DNA methylation levels, making it a good anti-cancer agent for this study [17].

Our finding that methylation in the promoter region (Figure 1a) of TLX1 decreased (Figure 4) following decitabine treatment supports our hypothesis that TLX1 epigenetic modification is affected by decitabine. This result is consistent with the anti-cancer drug’s demethylating effects and supports the previous finding that TLX1 is hypermethylated. The stark difference in methylation between our two control conditions (DKO and mCpG) was completely expected. DKO and mCpG represent opposite ends of the methylation spectrum. Not all of the CpG sites in the mCpG condition were fully methylated. This indicates that artificial methylation of DNA is not equally effective throughout the genome. Regardless, the DKO vs. mCpG comparison allowed us to establish valid controls for the rest of the study. Our results indicate that decitabine lowered methylation levels of TLX1 (relative to the DMSO control) by a statistically-significant amount (Figure 4). This finding supported the hypothesis that TLX1 methylation would be sensitive to decitabine treatment. This finding is consistent with the data from Simmer et al., 2012. The aforementioned researchers found that TLX1 is hypermethylated, and that globally removing this modification increases the expression of the gene more than 32-fold. We found that treatment with decitabine (an anti-cancer drug) is sufficient to both reduce the growth of HCT-116 cells and reduce methylation levels at the TLX1 promoter. In addition, since our analysis was at a low-throughput level, we were able to provide single-nucleotide resolution data on methylation changes in response to decitabine. Our data extend those previously published in ENCODE, as they include hypermethylated CpG sites in addition to those listed in the database. 

The sequencing data were very noisy, and alignment with the expected amplicon was difficult. The chromatograms for the DKO and mCpG condition had low signals, so our confidence in that data (Figure 2) is lower. The quality of the DMSO and decitabine was higher, giving us more confidence in these analyses (Figure 4). In many cases, it was not possible to match the expected nucleotide from the known sequence to those obtained from PCR. Several factors could account for this. After amplification and purification, the DNA in all conditions was not highly concentrated. This may have made it difficult for the sequencing facility to accurately read the amplified DNA. Another possibility is that the primers may have exhibited some non-specific binding (they contain many T and A nucleotides). A mixture of amplicons would have made it difficult for the sequencing facility to give accurate reads of the main region of interest. Regardless of the source, the noise in the data decrease our confidence in the results. This noise could be alleviated through the use of a nested sequencing primer, to increase the specificity of our PCR. Another surprise was that the fully methylated condition has a qualitatively lower percent methylation (60.39%) compared to the DMSO-treated condition (80.27%). One explanation for this trend is that, as stated above, artificial methylation in the mCpG condition was not entirely effective. This could combine with extreme hypermethylation at TLX1 (supported by Simmer et al., 2018 and likely to be evident in the DMSO condition) to produce the surprising trend. 

Previous research shows that DKO HCT-116 cells grow much more slowly than their counterparts, implying that removing methyltransferases could attenuate cancer properties. The question of whether or not methylation at TLX1 could explain abnormal growth remains open. The protein product of TLX1 is important in development of embryonic tissue and in the growth of the spleen. While the protein seems to not have a defined role in the colon, recent research has implicated it in many cancers, including acute T-cell lymphoblastic leukemia [18]. This broad involvement suggests that TLX1 could be an unknown driver of oncogenesis if its function is silenced by methylation. It is logical to predict that removing hypermethylation of TLX1 could reduce HCT-116 cells’ cancerous properties. 

It also seems likely that the slowdown in growth attributed to decitabine is due to TLX1 hypermethylation reversal. The effect of decitabine on methylation was statistically significant, showing a reduction in this epigenetic modification in response to the drug. Simmer et al. also report a slow-down in growth when methylation is removed via a DKO, and one of their top methylation hits is TLX1. Taken together, these findings point to TLX1 hypermethylation as a potential driver for carcinogenesis. Despite this promising direction, it is important to remember that decitabine likely interacts with gene targets other than TLX1. Given the complexity of colorectal cancer and of the mechanisms of cell division, it is highly improbable that one gene alone would drive a drug effect. It is much more likely that multiple methylation sites on different genetic factors are targeted simultaneously to reverse the cancerous properties of the cells. Decitabine may even target other mechanisms besides epigenetic modification. As the drug is incorporated into the DNA instead of cytosine, the cell may recognize it as a DNA lesion. Since this lesion is not one of the natural DNA damage scenarios that the cell is used to encountering, it may not be able to trigger mismatch repair, base excision repair, or nucleotide excision repair to overcome the stall in replication. Ultimately, the collapse of the replication fork will be fatal for the cell. Considering alternative mechanisms is important, as it may provide novel research directions supplementing the findings on epigenetic modifications.

In the future, to assess the effects of methylation in CRC more directly, one could replicate these experiments using DNA from CRC patients. A biopsy of the affected colorectal tissue (and a matched control) could be taken, and the same PCRs could be set up. Due to naturally occurring variations in the human DNA sequence, it may be difficult to obtain consistent primer binding with participants carrying SNPs or mutations. There are also several options to investigate whether methylation has a causative effect on cancer growth. One approach is to selectively demethylate DNA at the target region of interest. For instance, the growth of WT HCT-116 cells could be compared to the growth of cells that have been treated with TDG (thymine DNA glycosylase), an enzyme that selectively demethylates a desired sequence [19]. A decrease in growth would indicate that methylation is responsible for abnormal proliferation. One could also consider a conditional knockout of methyltransferase enzymes. This approach would allow for the use of vertebrate model organisms and could validate methylation as a causative factor in CRC (beyond the solid foundation already established by cell lines). 

Overall, our research suggests that decitabine is effective in killing CRC cells, it does adequately reduce methylation levels at a selected gene of interest. Our data suggest that TLX1 hypermethylation may serve as a biomarker for CRC. Additionally, as removing hypermethylation attenuates HCT-116 cell growth, this modification may also serve as a medical target for the disease. The pressing need to find more biomarkers for CRC remains. Using low-throughput molecular approaches is likely to reveal differences that can not only predict CRC development but also explain the mechanism behind this devastating disease.   

STAR Methods

Materials and Equipment

Bisulfite conversion

We first isolated genomic DNA from HCT-116 cells cultured in McCoy’s complete media, supplemented either with DMSO or 0.25 µmol decitabine in DMSO. To do this, we followed the protocol supplied in the PureLink Genomic DNA Isolation Kit from Thermo-Fisher. Subsequently, we had four sets of HCT-116 DNA to work with: DKO, mCpG, DMSO- and decitabine-treated. These four DNA samples were then subjected to a bisulfite conversion, in which all unmethylated cytosines become thymines. We started with 400 ng of DNA in each condition. We then added 5 uL of M-dilution buffer, topped off to 50 uL, and incubated the samples at 37 ℃. We then added 100 µL of CT Conversion Reagent and incubated at 50 ℃ overnight. On the following day, a spin column was prepared with 400 µL of M-Binding Buffer, and the sample was pipetted directly into the spin column. The column was centrifuged, and the flow-through was discarded. We then washed with 100 µL of M-Wash Buffer and 200 µL of M-Desulphonation buffer. After a 15-minute incubation at room temperature, we washed twice with 200 µL of M-Wash Buffer, and we finally eluted the DNA with 10 µL of M-Elution Buffer. 

Bioinformatics

We used pubicly available software for our bioinformatics analyses. Methylation levels, DNAseI hypersensitivity, transcription-factor binding, and histone modifications at TLX-1 were observed using the UCSC Genome Project’s database. Multiple cell lines, including HCT-116, NT2, and HeLa cells, were investigated. The histone modification data (on acetylation and methylation of H3) was uploaded onto the UCSC Genome Project site from the ENCODE Project. The sequence of the target region of TLX1 was also obtained from the UCSC Genome Project site. It was then uploaded to MethPrimer to design methylation-sensitive primers for the promoter region of the gene. Note that the primers were designed against the fully bisulfite-converted DNA template. Finally, EMBOSS-Needle was used for all sequence alignments to the expected amplicon. 

Primer Checksd

The primers were checked via PCR. A mix of fully methylated (mCpG) and completely unmethylated (DKO) DNA was used as the template. The PCR was set up as follows for each reaction: 4 µl epiTaq buffer, 13.5 µl H2O, 0.4 µl dNTPs, 1 µl DKO/mCpG DNA mix, 0.5 µl forward primer, 0.5 µl reverse primer, and 0.111 µl epiTaq polymerase. The denaturing temperature was 95 degrees C. Four annealing temperatures were tested: 52, 56, 60, and 64 degrees C. The extension temperature was 78 degrees C. As controls, the same reactions were set up, but with DapKI forward and reverse primers (at an annealing temperature of 59 degrees C). The cycle was looped 39 times. A final 5-minute extension at 68 degrees C was performed. The results were visualized using gel electrophoresis. 

To make sure that the primers evenly amplified both methylated and unmethylated DNA, similar PCRs were set up. The above recipe was followed, but 1 µl EvaGreen dye was also pipetted so that qPCR could be performed. This dye fluoresces when the DNA is double stranded, and it loses fluorescence when the DNA denatures. Here, the three most successful annealing temperatures (52, 56, and 60 degrees C) were tested with a mixed mCpG and DKO template. As crucial controls, we used the same PCR paradigm, but with pure templates of either mCpG or DKO DNA. After qPCR, we ran a high-resolution melt-curve analysis. We took the curves from qPCR and transformed them by taking the negative derivative of fluorescence. This gave peaks at the melting temperatures of the DNA amplified from both the mCpG and DKO templates. For methylation quantification, we chose the temperature that gave the most even peaks. 

Methylation Quantification

To quantify methylation levels in all four of our samples (mCpG, DKO, DMSO-treated, and decitabine-treated), we once again ran PCR according to the following recipe: 4 µl epiTaq buffer, 13.5 µl H2O, 0.4 µl dNTPs, 1 µl template DNA, 0.5 µl forward primer, 0.5 µl reverse primer, and 0.111 µl epiTaq polymerase. The template was either pure DKO, mCpG, DMSO-treated, or decitabine-treated DNA. For TLX1, the most appropriate annealing temperature was 52 degrees C. The PCR products were then purified using a DNA Clean and Concentrator kit provided by ZYMO Research. The manufacturer’s suggested protocol was followed. We then sent the amplified and purified DNA to the University of Chicago for sequencing (using the reverse primer). Once the sequences came back, we analyzed the results by first aligning the received sequence with the expected amplicon in EMBOSS-Needle. Then, for each CpG site, we compared the G and A signal strengths. Since the reverse primer was sent, this corresponds to C conversion to T (which only occurs when CpG sites are not methylated). Thus, for each CpG site, we derived a percent methylation statistic from all four conditions (DKO, mCpG, DMSO, and decitabine). We then compared DKO and mCpG methylation using a Student’s t-test. We did the same for the DMSO- and decitabine-treated conditions. 

Clonogenic Survival Assay

To test whether decitabine had an effect on HCT-116 cell colony formation, we cultured HCT-116 cells in the presence of DMSO, 0.1 µmol decitabine, 0.25 µmol decitabine, or 1 µmol decitabine in 6-well dishes. Cells were plated at a density of 200 cells per well. Decitabine was removed, and the cells were allowed to grow for 12 days. We then aspirated the media and rinsed each well with 2 mL of PBS. The PBS wash was then removed. A crystal violet mixture (containing 6% glutaraldehyde for fixation and 0.5% crystal violet for staining) was prepared. Each well was treated with 2 mL of the mixture. After a 1 hour incubation, the mixture was removed, and the plates were immersed in running water for about an hour. They were left to dry for a week. Subsequently, number of colonies in each of the four conditions was counted. Results were analyzed using a one-way ANOVA followed by Tukey’s HSD post-hoc test. 

Acknowledgements

We would like to thank Dr. William Conrad for his incredible support and dedication throughout this research. This project would not have been possible without his patient guidance and expertise. We would like to acknowledge the efforts of Rebecca Shoup, the lab peer teacher, and Brett Palmero, a Biol322 class member, for their help with setting up the reactions and preparing reagents. We would also like to thank Samuel Gascoigne, for his willingness to provide raw data for EN1 while the TLX1 samples were still being sequenced at the university of Chicago. Finally, thank you to all students from Biol322 for their friendship and support.