Experiments

miR-E-based shRNA design strategy

To target HDAC6, we designed a short hairpin RNA (shRNA) embedded within the optimized miR-E backbone2, following the design principles outlined in “Guidelines for the Optimal Design of miRNA-Based shRNAs” and “An Optimized microRNA Backbone for Effective Single-Copy RNAi”.

The target sequence (CTCACTGATCAGGCCATATTT) was selected using the Broad Institute's GPP Web Portal, based on its strong predicted efficacy, receiving a Sequence-Dependent RNAi (SDR) score of 100% , with an intrinsic score of 15.000 and an adjusted score of 21.000 , quantitative metrics reflecting on-target potency and design quality.

As miRNA-based shRNAs typically require a 22-nt guide strand for optimal Dicer processing, we performed a BLAST alignment against the HDAC6 mRNA using NCBI tools to identify the upstream nucleotide, which was a cytosine. This cytosine was added to the 5′ end of the guide to complete the final 22-nt antisense strand (AAATATGGCCTGATCAGTGAGG), which was then embedded into the 3′ arm of the miR-E scaffold.

The entire hairpin structure was manually constructed to conform to the optimized miR-E design, and crucial structural corrections were made by Dr. Xavier Bofill De Ros, co-author of the shRNA design guidelines, including the removal of an extra nucleotide in the loop and the addition of a missing base at the lower stem. These adjustments allowed the construct to fold identically to validated miR-E backbones.

Final validation was performed using RNAstructure, RNAfold, and SplashRNA to confirm correct folding and targeting efficacy.

To enable the insertion of our sequence into the transfer plasmid, we designed custom flanking sites. This allowed us to generate the final nucleotide sequence as follows:

5’TGTTTGAATGAGGCTTCAGTACTTTACAGAATCGTTGCCTGCACATCTTGGAAACACTTGCTGGGATTACTTCGACTTCTTAACCCAACAGAAGGCTCGAGAAGGTATATTGCTGTTGACAGTGAGCG ACTCACTGATCAGGCCATATTTTAGTGAAGCCACAGATGTAAAATATGGCCTGATCAGTGAGG TGCCTACTGCCTCGGACTTCAAGGGGCTAGAATTCGAGCAATTATCTTGTTTACTAAAACTGAATACCTTGCTATCTCTTTGATACATTTTTACAAAGCTGAATTAAAATGGTATAAATTAAATCACTTT 3’

Why miRE-based shRNA…

Morphe Project

Curious how miR-E achieves all this? Dive deeper into the science

▼

Accurate guide strand selection

Precise cleavage by Dicer ensures that the correct strand, the guide strand, is incorporated into the RNA-induced silencing complex (RISC), minimizing off-target effects. Deep sequencing shows that miR-E constructs are processed accurately over 82% of the time, compared to just 25–70% with conventional shRNA designs. This high fidelity enhances both the specificity and reliability of gene silencing1.

Lower expression, reduced cytotoxicity

Although miR-E shRNAs are expressed at lower levels than traditional pol III–driven constructs (such as those using U6 or H1 promoters), their improved processing leads to comparable or superior gene knockdown. The lower expression level helps to avoid off-target effects, immune activation, and cellular stress, making the system especially advantageous for in vivo experiments and therapeutic applications, where minimizing toxicity is critical1,2.

Tissue-specific and inducible expression

A key limitation of traditional shRNA systems is their use of constitutive RNA polymerase III promoters, which cannot be easily regulated. In contrast, miR-E constructs are compatible with RNA polymerase II (pol II) promoters, which allow for spatial and temporal control of expression. Researchers can drive shRNA expression in specific tissues, at certain developmental stages, or in response to drugs. This makes the system highly versatile for experimental setups as well as translational applications, that aim to move from basic science toward clinical or real-world use1-3.

Multiplexed gene regulation

The miR-E system also supports the expression of multiple shRNAs from a single transcript, enabling the simultaneous targeting of several genes. Thanks to optimized processing, each shRNA is efficiently released and functional. This makes miR-E constructs ideal for coordinated regulation of gene networks, synthetic biology circuits, and combinatorial strategies, approaches where multiple targets are modulated together to better study complex pathways or enhance therapeutic outcomes1.

Our plasmids

Due to limited expertise in neuronal systems, we opted to work with non-neuronal cells, which in turn influenced our plasmid design strategy.

Therapeutic plasmid (K562 experiments)

This plasmid was primarily designed to evaluate the efficiency of our miRE-based shRNA in knocking down HDAC6, and secondarily to demonstrate the correlation and rescue of the insulin signaling pathway upon knockdown in K562 PA-induced insulin-resistant cells.

In K562 cells, the use of condition-specific promoters (AgRP and FoxO1), as originally planned, was not feasible. As a result, we altered our therapeutic design, demonstrating the effectiveness of the shRNA molecule independently of the AND gate system (see project description).

Morphe Project

Want the full breakdown of each element?

▼

EF1α core promoter

The human EF1α promoter is widely used for driving strong and stable gene expression across various mammalian cell lines, including HEK293T and K562. Due to its mammalian origin, it is less prone to epigenetic silencing, making it ideal for sustained expression. As a Pol II promoter, it is also compatible with miRNA-based shRNA (miRE) expression.

Kozak sequence

Positioned just upstream of the start codon, the Kozak sequence improves translation efficiency by ensuring that the ribosome recognizes the correct initiation site. Its presence is crucial for consistent expression of the EGFP reporter.

EGFP

EGFP was placed upstream of the miRE-based shRNA to act as a visual reporter. Because both are transcribed from the same promoter, the presence of EGFP fluorescence indirectly confirmed successful transcription of the shRNA as well. We selected EGFP for its compact size, strong signal, and widespread use in mammalian cell imaging.

WPRE

The Woodchuck Hepatitis Virus Posttranscriptional Regulatory Element improves gene expression by enhancing mRNA stability and export from the nucleus. It does not interfere with the promoter itself but acts at the post-transcriptional level. WPRE is a standard element in lentiviral vectors and gene therapy applications.

hPGK promoter

The human phosphoglycerate kinase (hPGK) promoter drives constitutive, moderate expression of the puromycin resistance gene. Its endogenous origin makes it less susceptible to silencing, ensuring reliable selection in both HEK293T and K562 cells over time.

Puromycin

Puromycin is a fast-acting antibiotic that selects cells expressing the pac gene, which encodes puromycin N-acetyltransferase. It causes premature termination of protein synthesis, allowing us to efficiently eliminate non-transduced cells. Given the time constraints of our project, puromycin was ideal for rapid selection.

Logic gate & fail-safe plasmids (HEK293T experiments)

Proof-of-concept circuits

Before finalizing our AND gate strategy, we tested two candidate proof-of-concept (POC) circuits to model how independent inputs (like AgRP and FoxO1) could combine into a logical output.

Morphe Project

Curious about how each system worked in detail?

▼

Option A – Visual dual-input system (Selected for POC)

Upon doxycycline addition, rtTA was activated, leading to expression of TagBFP2 (blue). Cumate addition inactivates CymR, allowing expression of mCherry (red). When both inducers are present, cells co-express TagBFP2 and mCherry, resulting in purple fluorescence—a fully visual proxy for AND logic. Gal4 and VP16 were excluded, as their function as split activators is already well validated in the literature. Instead, we focused on a simplified, modular, and visual design.

Option B – compact genetic AND gate (Not selected)

Upon cumate addition, CymR is released from the CuO operator, allowing rtTA to be expressed; however, without doxycycline in the media, rtTA cannot activate TetO, so TagBFP2 is not produced. When only doxycycline is added, rtTA is not expressed, and TagBFP2 remains off. Only when both cumate and doxycycline are present does CymR release rtTA and doxycycline binds to rtTA, forming an active complex that binds TetO and drives TagBFP2 expression.

Morphe Project

While this system functions as a true genetic AND gate, it produces only a single output, offering no internal control and making negative results difficult to interpret. It also requires molecular assays, such as qPCR, for validation, which were not feasible within our budget and timeline. Furthermore, it is more prone to failure and less visually informative for a proof-of-concept study.

After considering both options, we proceeded with Option A and used it as the basis for our experimental design. This choice set the direction for the plasmids we constructed and determined the overall workflow of the project.

These two plasmids demonstrate our AND gate logic system and to validate the functionality of our fail-safe mechanism based on hsa-miR-7-5p.

Morphe Project
Morphe Project

For the AND gate, we built on the validated GAL4-VP16 system, replacing its DNA-binding and activation domains with two fluorescent proteins: mCherry and TagBFP2. To enhance biosafety and achieve precise control, we employed inducible promoters-Cumate and Doxycycline-(instead of the AgRP and Foxo1, see project description) which function independently of cell type.

Key point: Only when both inducers (cumate and doxycycline) were present, did cells co-express TagBFP2 and mCherry, thereby recapitulating AND gate behavior.

Fail-safe: As HEK293T cells lack endogenous miR-7-5p, we introduced a synthetic mimic. In its presence, cumate-induced mCherry expression was suppressed, demonstrating effective disruption of the AND gate output via RNA interference.

Want the full breakdown of how each element works?

▼

Tet-On System: TRE (TetOx7) and rtTA-M2

The TetOx7 sequence contains seven tetracycline operator (TetO) sites, forming TRE (Tetracycline Response Element). This promoter is activated only in the presence of doxycycline, which binds to the rtTA-M2 transcriptional activator. The M2 version offers reduced leakiness and improved sensitivity, resulting in tight and reliable inducible expression of TagBFP2.

Cumate system: 2xCuO and CymR

The cumate-inducible system functions similarly to the Tet-On system but does not rely on antibiotics. The CuO operator sites bind the CymR repressor protein in the absence of cumate, blocking transcription. When cumate is added, CymR undergoes a conformational change and dissociates from CuO, enabling transcription. This system regulated mCherry expression in our design.

TagBFP2 & mCherry

We used TagBFP2 and mCherry as fluorescent reporters representing two separate inputs. TagBFP2 was expressed only upon doxycycline induction (via TRE), while mCherry was controlled by the cumate system. Only when both inducers were present did we observe simultaneous expression of both fluorophores, thereby demonstrating the successful implementation of our synthetic AND gate.

YB_TATA core promoter

The YB_TATA promoter is a minimal core promoter (25 bp) known for low basal expression and high inducibility, ideal for synthetic constructs with limited plasmid space. It ensures minimal leakiness and strong activation when paired with upstream response elements, in our case TetO5.

Minimal CMV

This truncated version of the cytomegalovirus (CMV) promoter is used to drive CymR expression. It offers strong, constitutive transcription across various mammalian cell types, ensuring consistent repressor levels.

4x hsa-miR-7-5p target sequences

To implement a fail-safe mechanism, we included four tandem miR-7-5p target sites downstream of mCherry. This allows post-transcriptional silencing in the presence of a synthetic miR-7-5p mimic, even when transcription is actived by cumate. The decision to use four sites is supported by literature, which indicates that fewer sites are ineffective, while more do not significantly enhance repression6.

In a broader context, miR-7-5p is upregulated in AgRP neurons during fasting7. In our therapeutic design, it targets VP16AD, preventing shRNA production under normal physiological conditions. This acts as a third biosafety layer, ensuring activation only occurs during pathological states like insulin or leptin resistance, thereby preventing unintended gene silencing outside the intended therapeutic context.

Blasticidin (Bsd)

Blasticidin is an antibiotic used in molecular biology for the selection of genetically modified cells carrying the blasticidin resistance gene (bsd). It acts by inhibiting protein synthesis, enabling efficient elimination of non-transfected cells in both prokaryotic and eukaryotic systems.

Cell cultures

To align with our team’s expertise and available resources, we selected HEK293T and K562 cell lines, well-established models known for their high transduction efficiency, for initial validation.

Morphe Project

Want the full breakdown of how each element works?

▼

HEK293T cells

HEK293T is an adherent cell line derived from Human Embryonic Kidney (HEK293) cells and engineered to express the SV40 large T antigen, which enhances both transfection and transduction efficiency. These cells are widely used in molecular biology and gene therapy studies due to their robust growth, ease of handling, and high reproducibility. They were selected for our proof-of-concept experiments, primarily because our lab had prior experience culturing them, and they are readily accessible.

While neuron-like cell lines such as SH-SY5Y would have been more biologically relevant to our therapeutic target, our team lacked the expertise and resources to culture neuronal models. In contrast, HEK293T cells are extensively characterized in literature, allowing for reliable troubleshooting and benchmarking of experimental results. This helped ensure that any unexpected outcomes could be attributed more confidently to our constructs rather than the cellular model itself.

To maintain consistency and cell health, we followed standard protocols for freezing, thawing, seeding, and passaging, which are thoroughly documented in our lab notebook.

K562 cells

K562 is a non-adherent, suspension cell line derived from a patient in blast crisis of chronic myelogenous leukemia (CML). It is a commonly used model in hematopoietic and molecular biology research due to its stable growth, high lentiviral transduction efficiency, and ease of maintenance.

K562 cells were specifically chosen for the shRNA-mediated knockdown of HDAC6. Literature indicates that K562 cells express high endogenous levels of HDAC6 mRNA and moderate protein levels, providing a suitable model for assessing knockdown efficiency both at the transcript and protein level. Their compatibility with lentiviral systems and biological relevance to our RNAi target made them an ideal model for validating our miRE-based shRNA.

Lentiviral production and delivery

Morphe Project

Transfection

Lentivirus production began with transient co-transfection of HEK293T cells using the standard three-plasmid system:

  1. Envelope plasmid: p23 (encodes VSV-G protein)
  2. Packaging plasmid (p202): Encodes Gag, Pol, Rev, Tat
  3. Transfer plasmid.

This protocol was performed using established lipid-based transfection reagents and optimized for high-yield viral particle production.

Virus harvest

72 hours post-transfection, the viral supernatant was collected and filtered through a 0.45 μm PES membrane to remove cell debris while retaining lentiviral particles.

Following harvest, we quantified the viral titer (method to be specified) to confirm successful virus production before proceeding with cell transduction experiments.

Lentivirus surface modification

Our therapeutic design includes targeted delivery of lentiviral vectors to hypothalamic AgRP neurons via intranasal administration. To enable cell-type specificity, we proposed modifying the viral surface post-harvest using a click chemistry reaction8. This strategy involves conjugating a leptin-derived peptide to the envelope, facilitating receptor-mediated endocytosis through interaction with the LepRb receptor, which is highly expressed on AgRP neurons9,10.

Due to the absence of neuronal cultures and the lack of LepRb expression in our available cell lines, combined with limited experimental time, we were unable to evaluate this step in vitro. Furthermore, the click chemistry process under consideration is relatively straightforward, and modeling it would not provide novel insights. Specifically, obtaining binding rate constants for this reaction does not contribute meaningful information relevant to downstream analysis or impact the drug delivery mechanism itself. Additionally, constructing differential equations to describe this system is challenging due to uncertainties and lack of precise kinetic parameters. Therefore, investing effort into modeling this aspect is unlikely to advance the project significantly.

Choice of biomarker: LepRb receptor – 1st layer of safety

An essential requirement for gene therapy is minimizing off-target effects. To achieve tissue specificity, we needed a receptor highly expressed in target cells but absent in peripheral tissues. We conducted ROC-based expression analyses across multiple brain regions and highly perfused organs (e.g. liver, heart, lungs, kidneys, skin) and consulted domain experts and our PIs.

Based on bioinformatics data, literature data and expert feedback, we selected the long isoform of the leptin receptor (LepRb) as the most appropriate cell surface biomarker for our targeted delivery strategy. Its restricted expression in AgRP neurons provides the foundation for our system’s first biosafety layer.

Morphe Project

Lentiviral transduction and experimental validation

Lentiviral transduction was performed in K562 cells, while HEK293T cells were subjected to transient co-transfection of the lentiviral plasmids, without proceeding to viral particle production. For K562 cells, the viral supernatant was collected and used to infect the cells in the presence of polybrene, which enhances viral entry and infection efficiency. After one week, selection antibiotics were applied to both HEK293T and K562 cells to enrich for successfully transduced populations, resulting in stable cell lines expressing the gene of interest.

Note: Our original plan included lentiviral packaging and transduction to generate stable cell lines for long-term testing. However, delays in the delivery of HEK293T cells from the supplier compressed our experimental window. Given these time constraints, and since our goal was to characterize short-term inducible responses within a week rather than assess long-term integration over a month, we deliberately chose to perform transient plasmid transfections instead. While this strategy enabled rapid validation of circuit function, it comes with known limitations: transient expression is heterogeneous across cells, declines over time, and does not fully replicate stable genomic integration. Therefore, the data presented here should be interpreted as proof-of-principle results, with stable integration left for future work.

K562 experiments – HDAC6 knockdown

Morphe Project

In K562 cells, transduction efficiency was evaluated by monitoring EGFP expression via fluorescence microscopy, confirming transcription from the same promoter driving the miRE-based shRNA. To assess knockdown efficiency, we performed mechanistic assays involving: RT-qPCR to quantify HDAC6 mRNA levels and flow cytometry to measure HDAC6 protein expression, confirming successful knockdown at both the transcript and protein levels.

Next, we performed a functional assay to investigate the relationship between HDAC6 and insulin signaling. Given that obese patients with insulin resistance often exhibit elevated HDAC6 levels, we hypothesized that HDAC6 knockdown could restore insulin sensitivity.

To test this, we carried out immunofluorescence microscopy (IF) to examine the subcellular localization of phosphorylated Foxo1 (pFoxo1), a downstream effector of the PI3K/AKT pathway, as a readout of insulin pathway activation.

For this assay, K562 cells were treated with palmitic acid (PA) overnight to induce insulin resistance, serum-starved for 4 hours, and subsequently stimulated with insulin for 15 minutes to activate PI3K/AKT signaling. Control (non-transduced) and HDAC6 knockdown cells were processed in parallel.

We expect that comparative analysis of pFoxO1 localization and signal intensity will reveal that HDAC6 knockdown restores insulin signaling, supporting our hypothesis that HDAC6 contributes to insulin resistance and that its reduction could enhance insulin sensitivity in obese patients.

HEK293T experiments – Logic gate and fail-safe validation

To experimentally validate our circuit architecture, we planned a series of transduction and induction assays in HEK293T cells. Following transient co-transfection of lentiviral plasmids and antibiotic selection, only successfully transduced cells were maintained to ensure stable construct integration.

To assess the functionality of our dual-inducible logic gate, cells were exposed to doxycycline and cumate individually and in combination11. This setup allows us to evaluate whether each inducer specifically activates its corresponding module and whether co-stimulation leads to concurrent activation (magenta colour), as predicted for our AND gate design.

To examine the robustness of our fail-safe mechanism, we designed an additional condition in which cells are co-transfected with a synthetic hsa-miR-7-5p mimic. This step aims to test whether the safety circuit can suppress downstream expression in the presence of the miRNA, even when the corresponding inducer is present-verifying post-transcriptional control at the design level.

References

▼
  1. An Optimized microRNA Backbone for Effective Single-Copy RNAi. PubMed
  2. Optimization and comparison of knockdown efficacy between polymerase II expressed shRNA and artificial miRNA targeting luciferase and Apolipoprotein B100. PubMed
  3. Guidelines for the optimal design of miRNA-based shRNAs. PubMed
  4. Creating an miR30-Based shRNA Vector. PubMed
  5. Quantitative Analyses of Core Promoters Enable Precise Engineering of Regulated Gene Expression in Mammalian Cells. PubMed
  6. A mixed antagonistic/synergistic miRNA repression model enables accurate predictions of multi-input miRNA sensor activity. PMC
  7. Fasting-induced miR-7a-5p in AgRP neurons regulates food intake. PubMed
  8. Lipid-based nano-carriers for the delivery of anti-obesity natural compounds: advances in targeted delivery and precision therapeutics. PMC
  9. Novel mechanisms involved in leptin sensitization in obesity. ScienceDirect
  10. Leptin and Obesity: Role and Clinical Implication. PMC
  11. New dual inducible cellular model to investigate temporal control of oncogenic cooperating genes. Nature