Week 1: Principles and Practices

Class Assignment

Project Idea

Chronic wounds and surgical site infections affect millions of patients and cost heathcare systems tens of billions of dollars annually, yet closure devices often remain as passive stitches that do not actively orchestrate local immunity or regeneration [1][2].

Drug-eluting sutures have shown that suture material can safely deliver local therapeutics, but current designs provide only finite, non-adaptive release of single agents such as antibiotics or growth factors [3][4]. Cell-filled sutures packed with mesenchymal stem cells already demonstrate that viable cells can be integrated into suture structures and enhance healing, but these cells are unmodified and lack controllable, multi-functional outputs [5]. Separately, engineered combinatorial cell devices in fiber-like formats can secrete optimized cocktails of growth factors to accelerate wound and bone repair, but they are not load-bearing sutures and do not address infection or scar modulation at the incision line [6].

In a separate project, I explored how patient skin cells (such as fibroblasts) could be engineered to express a genetic circuit that could counteract the persistent inflammation of chronic wounds, sense a biomarker indicative of the end of the inflammatory wound healing phase, and then kickstart the proliferation phase sequentially.

I want to use a similar premise to propose a hollow, bioabsorbable suture that houses genetically engineered cells programmed to sense wound and infection cues to secrete combinations of pro-regenerative and antimicrobial factors over the critical healing window. This would transform sutures from a passive mechanical closure tool into an adaptive, living therapeutic that directly tackles both impaired healing and scarring in a way that current drug-eluting or cell-based sutures cannot.

Governance/Policy Goals

Enhance Biosecurity

Introducing genetically modified living materials into the body always poses the risk of unintended side effects in terms of how that newly modified

• Escape and persistence of engineered cells
  • Genetically engineered cells have the potential to leak from the suture material during deegradation, which may cause the migration of these cells to other unintended areas of the body
• Unintended immune suppression hotspots
  • One potential application of the seeded engineered cells is to assist in the healing of chronic wounds, which would require the secretion of anti-inflammatory genes/cytokines. In this case, it could potentially host an environment that is susceptible to tumor growth due to the prevention of the body’s natural protection mechanisms becoming temporarily reduced

Foster Lab and Patient Safety

• By preventing incident
• Informed patient consent
• Adverse events

Protect the Environment

• Wasted suture material
  • As this suture material would contain a living cellular component, the wasted material would need to be properly disposed of through the right channels
• Resistance ecology
  • As the suture material could aim to reduce microbial infection, this could lead to an inadvertent resistance issue through evolution (similar to antibiotic resistance) and should be thoughtfully considered

Other Considerations

• Equal access
• Not impede research
• Promote constructive applications

Potential Governance Actions

1. Specialized biosafety and clinical training track for “living implant” users

• Purpose:
  • Researchers and clinicians complete general biosafety and surgical training, but there is no standardized curriculum for working with engineered living impants
  • Establish a dedicated training a certification program for labs and clinicians who design, manufacture, or implant living sutures, similar to specialized credentialing for radiation safety or gene therapy administration
• Design:
  • Government entities would implement a standardized curriculum and requirement for all individuals working with living material users
  • Universities, hospitals, and organizations could develop modules on containment of genetically engineered materials, safety functions and limitations, proper disposal methods and would need to require completion from designated users
• Assumptions:
  • Assumes that the training would be taken seriously by all parties involved
  • Assumes that institutions have the resources to implement this level of training
• Risks of Failures and Success:
  • Training can easily devolve into people trying to just “pass a quiz”
  • Small or underprivileged institutions may not be able to support the certification
  • Credentials could become a bottleneck in care, limiting broader patient impact

2. Mandatory standardized labeling and risk communication for living sutures

• Purpose:
  • Implanted devices and sutures often have minimal patient-facing documentation and many patients do not know exactly what materials are being used
  • Require clear, standardized labeling and risk summaries for any engineered-cell stuure, both on packaging for clinicians and in take-home materials for patients, similar to medication guides for high-risk drugs
• Design:
  • Government entities should define a standardized material and one-page explanation that should include that the suture is living/engineered, intended benefits, key unknowns, possible risks, and recommended follow-up durations
  • Medical professionals should ensure that patients receive and acknowledge these materials during consent and discharge
• Assumptions:
  • Assumes patients will read and understand the materials
  • Assumes that clinicians will consistently use and explain documents instead of just handing them over
  • Assumes that simple language used for materials can convey the complex biological concepts utilized
• Risks of Failures and Success:
  • Overly technical language may confuse or scare patients without helping them to make an informed decision
  • If the material emphasizes uncertainty too strongly, clinicians may avoid using the sutures due to patient refusal or anxieties, even when risk-benefit is favorable in high-need cases

3. Open safety data and pre-registration for living-suture research

• Purpose:
  • Clinical trials are often pre-registered, but preclinical work, especially in industry, can remain proprietary and negative results are frequently unpublished
  • Require prospective registration and open reporting of both clinical and key preclinical studies involving engineered-cell sutures, including negative or inconclusive safety findings
• Design:
  • Academic and industrial labs should register protocols in public or semi-public databases and post summaries of the key findings, including failures
  • Government or regulatory safety boards should aggregate data and identify patterns which can be communicated to different programs and companies
• Assumptions:
  • Assumes companies will accept some loss of competitive secrecy for safety transparency
  • Assumes public reporting can be done in ways that protect intellectual property while still being meaningful
• Risks of Failures and Success:
  • Compliance may be partial, some negative preclinical findings could stay hidden in internal reports
  • Low-quality data could mislead more than inform
  • Highly publicized early safety issues, even if fixable, could dissolve public trust in otherwise promising tools

Governance Actions vs Policy Goals

• 1= most responsibility, 4=least responsibility

Lecture 2 Preparation Questions

Questions from Professor Jacobson

1. The error rate for DNAP is 10⁶ (about 1 in 1 million). Since the human genome is roughly 3.2 x 10⁹ bp, this means that there would be around 3,200 errors each time a genome copy is made. However, nature is able to combat these errors due to its error correction mechanisms, such as the MutS repair system.
2. If we assume that an average human protein has 375 amino acids [7], and there are about three codons that code each amino acid, then there are roughly 10¹⁸⁰ ways to code for the average human protein. However, some of these codings could be invalid if they don’t have a proper start codon, if they have unstable mRNA, or if they produce a misfolded protein.

Questions from Dr. LeProust

1. Currently, the method typically used for oligo synthesis is solid-phase phosphoramidite chemistry, where the 5’ end of the previous nucleotide is protected and as phosphoramidites are added (modified versions of each nucleotide), the 5’ end is exposed, allowing the next base to couple, and then the resulting 5’ end is protected once again while an oxidizing solution stabilizes the bond that was just formed, repeating the process until one obtains the desired oligonucleotide [8].
2. Oligos longer than 200bp are typically too difficult to synthesize due to an accumulation of impurities that significantly decreases the yield [9].
3. Coding a gene over 2000bp by oligo synthesis would also be difficult due to exponentially decreasing yields over a certain threshold and difficulty with purifying the final product.

Questions from George Church

In response to question #2

• The NA:NA code relies on pairing G to C and pairing A with T (or U in RNA). This then is translated in the AA:NA code as a three bp long codon that translates to one of the twenty amino acid, and this ultimately results in amino acids that can be coded by multiple codon sequences. In order to create an AA:AA code, which would represent protein-protein interactions, I would anticipate the need to consider 3D structure as well as properties of each of the AAs. For example, a positively charged amino acid, like histamine, would ultimately pair best with a negatively charged amino acid, such as glutamic acid. Since there are multiple amino acids with these properties, the code would not have a singular outcome, like NA:NA, but this code could then be further optimized through the 3D structure complemtarity [10][11].

Week 2: DNA Read, Write, and Edit

Benchling and In-Silico Gel Art

Simulate Restriction Enzyme Digest

I found this process quite intuitive, as I’ve done similar simulations with the application SnapGene, but it was interesting to notice the small interface differences between the two!

Pattern in the style of Paul Vanouse

I attempted to make a “Y” for Yonsei, but it turned out to be more difficult than I expected and this was the closest I ended up getting… Huge respect to the people who were able to make a more comprehensive image like the ones that spell MIT!

DNA Design Challenge

My Chosen Protein

I chose to explore Calreticulin (CALR) as my protein of interest for this week due to its role as a pro-healing cue in wound healing[1]. CALR typically serves to support the progression through the four wound healing phases (hemostasis, inflammation, proliferation, and remodeling) [2], which is classically disrupted during chronic wounds [3].

Protein sequence[4]:

sp|P27797|CRTC_HUMAN CALRETICULIN PRECURSOR from residues
                     31- 64, Pval= 3.8e-18, (100% identity); putative"
ORIGIN      
        1 mllsvplllg llglavaepa vyfkeqfldg dgwtsrwies khksdfgkfv lssgkfygde
       61 ekdkglqtsq darfyalsas fepfsnkgqt lvvqftvkhe qnidcgggyv klfpnsldqt
      121 dmhgdseyni mfgpdicgpg tkkvhvifny kgknvlinkd irckddefth lytlivrpdn
      181 tyevkidnsq vesgsleddw dflppkkikd pdaskpedwd erakiddptd skpedwdkpe
      241 hipdpdakkp edwdeemdge weppviqnpe ykgewkprqi dnpdykgtwi hpeidnpeys
      301 pdpsiyaydn fgvlgldlwq vksgtifdnf litndeayae efgnetwgvt kaaekqmkdk
      361 qdeeqrlkee eedkkrkeee eaedkedded kdedeedeed keedeeedvp gqakdel

Nucleotide sequence[5]:

atgctgctgagcgtgccgctgctgctgggcctgctgggcctggcggtggcggaaccggcg
gtgtattttaaagaacagtttctggatggcgatggctggaccagccgctggattgaaagc
aaacataaaagcgattttggcaaatttgtgctgagcagcggcaaattttatggcgatgaa
gaaaaagataaaggcctgcagaccagccaggatgcgcgcttttatgcgctgagcgcgagc
tttgaaccgtttagcaacaaaggccagaccctggtggtgcagtttaccgtgaaacatgaa
cagaacattgattgcggcggcggctatgtgaaactgtttccgaacagcctggatcagacc
gatatgcatggcgatagcgaatataacattatgtttggcccggatatttgcggcccgggc
accaaaaaagtgcatgtgatttttaactataaaggcaaaaacgtgctgattaacaaagat
attcgctgcaaagatgatgaatttacccattataccctgattgtgcgcccggataacacc
tatgaagtgaaaattgataacagccaggtggaaagcggcagcctggaagatgattgggat
tttctgccgccgaaaaaaattaaagatccggatgcgagcaaaccggaagattgggatgaa
cgcgcgaaaattgatgatccgaccgatagcaaaccggaagattgggataaaccggaacat
attccggatccggatgcgaaaaaaccggaagattgggatgaagaaatggatggcgaatgg
gaaccgccggtgattcagaacccggaatataaaggcgaatggaaaccgcgccagattgat
aacccggattataaaggcacctggattcatccggaaattgataacccggaatatagcccg
gatccgagcatttatgcgtatgataactttggcgtgctgggcctggatctgtggcaggtg
aaaagcggcaccatttttgataactttctgattaccaacgatgaagcgtatgcggaagaa
tttggcaacgaaacctggggcgtgaccaaagcggcggaaaaacagatgaaagataaacag
gatgaagaacagcgcctgaaagaagaagaagaagataaaaaacgcaaagaagaagaagaa
gcggaagataaagaagatgatgaagataaagatgaagatgaagaagatgaagaagataaa
gaagaagatgaagaagaagatgtgccgggccaggcgaaagatgaactgtaa

Codon optimization[6]:

ATGCTCCTGTCCGTGCCCCTGCTGCTGGGCCTGCTGGGGCTCGCCGTGGCTGAGCCCGCC
GTGTACTTCAAGGAGCAGTTCCTGGACGGCGATGGCTGGACATCCAGATGGATCGAGTCT
AAGCATAAGTCCGACTTCGGCAAGTTCGTGCTGTCCAGCGGGAAGTTCTATGGGGACGAG
GAGAAGGACAAAGGCCTGCAGACCTCACAGGACGCAAGATTCTATGCCCTTAGCGCCAGC
TTCGAGCCCTTCTCAAACAAAGGGCAGACTCTGGTGGTGCAGTTCACTGTGAAGCATGAG
CAGAACATTGATTGCGGCGGCGGCTACGTGAAGCTGTTTCCTAATAGCCTGGATCAGACA
GACATGCACGGGGACAGCGAGTATAACATCATGTTCGGCCCAGACATTTGCGGCCCAGGC
ACTAAGAAGGTGCACGTGATTTTCAATTATAAAGGCAAAAACGTGCTGATCAATAAAGAC
ATTAGGTGTAAGGATGACGAGTTCACCCATTACACCCTGATCGTGCGCCCCGACAACACC
TACGAGGTGAAGATCGACAACTCACAGGTGGAGAGCGGGAGCCTGGAGGACGACTGGGAC
TTTCTGCCACCAAAGAAGATTAAGGACCCCGACGCCTCCAAGCCCGAGGACTGGGACGAG
CGGGCCAAAATCGACGATCCAACAGATTCAAAGCCCGAAGACTGGGATAAGCCTGAGCAC
ATCCCCGACCCAGACGCAAAGAAGCCTGAAGACTGGGACGAGGAGATGGACGGCGAGTGG
GAGCCCCCTGTGATCCAGAACCCCGAGTACAAGGGGGAGTGGAAGCCAAGGCAGATTGAC
AACCCCGACTACAAAGGCACTTGGATTCACCCTGAGATCGACAACCCCGAATATTCACCC
GACCCCTCTATCTACGCCTACGACAATTTCGGGGTGCTGGGCCTGGACCTGTGGCAGGTG
AAGAGCGGCACCATCTTCGACAATTTCCTGATCACAAACGACGAGGCCTACGCCGAAGAG
TTCGGCAATGAGACATGGGGCGTGACCAAAGCCGCCGAGAAGCAGATGAAGGACAAGCAA
GACGAGGAGCAGCGCCTGAAAGAGGAGGAGGAGGACAAAAAGCGCAAGGAGGAGGAGGAA
GCCGAGGACAAAGAAGACGACGAGGATAAGGACGAGGATGAAGAGGACGAAGAAGACAAG
GAGGAGGACGAGGAGGAAGATGTCCCCGGACAGGCCAAGGACGAGCTGTGA

Codon optimization is essential to ensuring the proper and efficient protein expression of a given protein within a specific organism. Typically, different organisms favor different codons that ultimately encode the same amino acid [7], which is why optimizing to the specific organism you intend to use to produce the protein verifies that frequently used codons are encoded instead of rarely used ones within your expression host of choice[8].

In this case, I think that the most applicable technology to produce CALR would be to use HEK293T cells. These cells are human derived and are quickly replicable, meaning that they would be able to prouduce this protein with great efficiency. In order to do this, first I would need to clone my protein insert into a mammalian expression vector with a strong promoter, which would then be tranfected into the HEK293T cells (for example, by lipofection). Within the cells, the RNAP recognizes the promoter and would transcribe the plasmid into RNA. This mRNA would then be translated into protein by the ribosome.

Prepare a Twist DNA Synthesis Order

I prepared my optimized gene within Benchling for Twist [9]. This order page was so simple compared to other ordering sites I used, and I liked that you could export the entire plasmid as well.

DNA Read/Write/Edit

DNA Read

On a similar theme as my previous assignments, one of my main interests is in finding the key molecular mechanisms and differences that distinguish successful wound repair from chronic, non-healing counterparts. For this reason, I’d be interested in being able to compare the mitochondrial genome of healthy (efficiently healing) and chronic wound patients, as the mitochondria has been proven to play a central role in wound metabolism [10][11]. By sequencing these genomes and contrasting the two, it may reveal variants within the genome that could predispose individuals or make them more vulnerable to chronic wound development.

In order to achieve this goal, I would try to make use of Oxford Nanopore sequencing, which is a third-generation sequencing technology. Since the main goal of this reading would be to read specifically mitochondrial DNA, the first step would be to extract the DNA from a wound tissue/normal tissue sample and quantify it. Next, I should perform long-amplicon PCR in order to highlight the mitochondrial DNA, subsequently adding an A tail so that sequenecing adapters can ligate efficiently. Following this, the sequencing adapters will be attached and and then sequencing will be started. In order to decode the bases of the DNA sample, Oxford Nanopore sequencing relies on ionic currents to detect which bases are passing through the nanopore. Since each basepair emits a different current value, we are able to trace the sequence that subsequently passes through the pore by decoding each current value. The final output of this sequencing technology is a FASTQ file that includes the DNA sequence along with a per-base quality score.

AI citation

Peplexity - “Can you explain in simple terms how Oxford Nanopore sequencing is prepared and what is the outcome?”

DNA Write

One of the long-term projects I’ve been working on at the Designer Cells lab was to synthesize a genetic circuit for chronic wounds. Since the main obstacle of chronic wound healing is their persistent inflammation, I designed a FLEx (Flip Excision) switch [12] that first expresses an anti-inflammatory gene set and then switches irrersibly to a migratory gene set after sensing a biomarker indicative of the end of the inflammation phase of wound healing.

Since this genetic circuit insert ended up being close to 4.5k, I think that the most efficient method to synthesize this would be an enzymatic synthesis approach. This synthesis method is similar to how primers are synthesized. Some limitations of this synthesis method is that it has limited synthesizing ability for longer strands of DNA.

DNA Edit

One point of DNA I’ve been looking at is to edit Calreticulin to add domains that recognize damage-associated extracellular matrix patterns. This would allow these engineered proteins to more effectively seek out damaged tissues (such as in the case fo chronic wounds) in order to facilitate wound healing.

In order to achieve this, I could utilize CRISPR-Cas HDR. In order to achieve this, I would need to design a gRNA that targets the desired insertion site as well as the edits that would need to be made. Then, I would need to ensure that the gRNA was delivered to the appropriate cells of my choosing. The potential limitations could revolve around structure integrity as well as the efficiency of HDR.

Week 3: Lab Automation

Python script for Opentrons artwork

For the art portion of this week’s assignment, I decided to code Yoshi from Super Mario Brothers since the Designer Cells node only had the red and green colors. I used this photo as reference. From there, I started to code for the Opentron automation.

from opentrons import types

metadata = {    # see https://docs.opentrons.com/v2/tutorial.html#tutorial-metadata
    'author': 'Sydney',
    'protocolName': 'Yoshi',
    'description': 'Prints Yoshi',
    'source': 'HTGAA 2026 Opentrons Lab',
    'apiLevel': '2.20'
}

##############################################################################
###   Robot deck setup constants - don't change these
##############################################################################

TIP_RACK_DECK_SLOT = 9
COLORS_DECK_SLOT = 6
AGAR_DECK_SLOT = 5
PIPETTE_STARTING_TIP_WELL = 'A1'

well_colors = {
    'A1' : 'Red',
    'B1' : 'Green',
    'C1' : 'Orange'
}

sfgfp_points = [(0, 34),(2, 34),(4, 34),(6, 34),(0, 32),(2, 32),(4, 32),(6, 32),(-4, 30),(-2, 30),(0, 30),(2, 30),(-4, 28),(-2, 28),(0, 28),(2, 28),(-4, 26),(-2, 26),(12, 26),(14, 26),(16, 26),(18, 26),(20, 26),(22, 26),(-4, 24),(-2, 24),(12, 24),(14, 24),(16, 24),(18, 24),(20, 24),(22, 24),(-8, 22),(-6, 22),(-4, 22),(-2, 22),(8, 22),(10, 22),(12, 22),(14, 22),(16, 22),(18, 22),(24, 22),(26, 22),(-8, 20),(-6, 20),(-4, 20),(-2, 20),(8, 20),(10, 20),(12, 20),(14, 20),(16, 20),(18, 20),(24, 20),(26, 20),(-8, 18),(-6, 18),(-4, 18),(-2, 18),(0, 18),(2, 18),(4, 18),(6, 18),(8, 18),(10, 18),(12, 18),(14, 18),(16, 18),(18, 18),(20, 18),(22, 18),(24, 18),(26, 18),(-8, 16),(-6, 16),(-4, 16),(-2, 16),(0, 16),(2, 16),(4, 16),(6, 16),(8, 16),(10, 16),(12, 16),(14, 16),(16, 16),(18, 16),(20, 16),(22, 16),(24, 16),(26, 16),(-12, 14),(-10, 14),(-8, 14),(-6, 14),(8, 14),(10, 14),(12, 14),(14, 14),(16, 14),(18, 14),(20, 14),(22, 14),(24, 14),(26, 14),(-12, 12),(-10, 12),(-8, 12),(-6, 12),(8, 12),(10, 12),(12, 12),(14, 12),(16, 12),(18, 12),(20, 12),(22, 12),(24, 12),(26, 12),(-12, 10),(-10, 10),(8, 10),(10, 10),(12, 10),(14, 10),(16, 10),(18, 10),(20, 10),(22, 10),(24, 10),(26, 10),(-12, 8),(-10, 8),(8, 8),(10, 8),(12, 8),(14, 8),(16, 8),(18, 8),(20, 8),(22, 8),(24, 8),(26, 8),(12, 6),(14, 6),(16, 6),(18, 6),(20, 6),(22, 6),(12, 4),(14, 4),(16, 4),(18, 4),(20, 4),(22, 4),(8, 2),(10, 2),(8, 0),(10, 0),(-4, -2),(-2, -2),(0, -2),(2, -2),(4, -2),(6, -2),(-4, -4),(-2, -4),(0, -4),(2, -4),(4, -4),(6, -4),(-24, -6),(-22, -6),(-4, -6),(-2, -6),(0, -6),(2, -6),(4, -6),(6, -6),(12, -6),(14, -6),(16, -6),(18, -6),(-24, -8),(-22, -8),(-4, -8),(-2, -8),(0, -8),(2, -8),(4, -8),(6, -8),(12, -8),(14, -8),(16, -8),(18, -8),(-24, -10),(-22, -10),(-20, -10),(-18, -10),(-8, -10),(-6, -10),(-4, -10),(-2, -10),(0, -10),(2, -10),(4, -10),(6, -10),(8, -10),(10, -10),(12, -10),(14, -10),(16, -10),(18, -10),(-24, -12),(-22, -12),(-20, -12),(-18, -12),(-8, -12),(-6, -12),(-4, -12),(-2, -12),(0, -12),(2, -12),(4, -12),(6, -12),(8, -12),(10, -12),(12, -12),(14, -12),(16, -12),(18, -12),(-20, -14),(-18, -14),(-16, -14),(-14, -14),(-12, -14),(-10, -14),(-8, -14),(-6, -14),(-4, -14),(-2, -14),(0, -14),(-20, -16),(-18, -16),(-16, -16),(-14, -16),(-12, -16),(-10, -16),(-8, -16),(-6, -16),(-4, -16),(-2, -16),(0, -16),(-16, -18),(-14, -18),(-12, -18),(-10, -18),(-8, -18),(-6, -18),(-4, -18),(-2, -18),(0, -18),(-16, -20),(-14, -20),(-12, -20),(-10, -20),(-8, -20),(-6, -20),(-4, -20),(-8, -22),(-6, -22),(-4, -22),(-2, -22),(0, -22),(2, -22),(4, -22),(-8, -24),(-6, -24),(-4, -24),(-2, -24),(0, -24),(2, -24),(4, -24)]
mrfp1_points = [(6, 30),(6, 28),(4, 26),(6, 26),(4, 24),(6, 24),(4, 22),(6, 22),(20, 22),(22, 22),(0, 20),(2, 20),(4, 20),(6, 20),(20, 20),(22, 20),(-12, 18),(-10, 18),(-12, 16),(-10, 16),(-16, 14),(-14, 14),(-4, 14),(-2, 14),(0, 14),(2, 14),(6, 14),(-16, 12),(-14, 12),(-4, 12),(6, 12),(-16, 10),(-14, 10),(-8, 10),(-6, 10),(6, 10),(-16, 8),(-14, 8),(-8, 8),(6, 8),(-12, 6),(-10, 6),(10, 6),(-12, 4),(-10, 4),(8, 4),(10, 4),(-12, 2),(-10, 2),(-8, 2),(-6, 2),(-12, 0),(-10, 0),(-8, 0),(-6, 0),(-2, 0),(0, 0),(2, 0),(4, 0),(6, 0),(-8, -2),(-6, -2),(10, -2),(-8, -4),(-6, -4),(10, -4),(-16, -6),(-14, -6),(-12, -6),(-10, -6),(-6, -6),(10, -6),(-16, -8),(-14, -8),(-12, -8),(-10, -8),(-6, -8),(10, -8),(-16, -12),(-14, -12),(-12, -12),(-10, -12),(-24, -14),(10, -14),(-24, -16),(-22, -16),(10, -16),(-20, -18),(8, -18),(-20, -20),(-18, -20),(6, -20),(8, -20),(-16, -22),(-16, -24),(-14, -24),(-12, -24),(-10, -24),(-8, -26),(-6, -26),(-4, -26),(-2, -26),(0, -26),(-8, -28),(-6, -28),(-4, -28),(-2, -28),(0, -28),(-8, -30),(-6, -30),(-4, -30),(-2, -30),(0, -30),(2, -30),(4, -30),(-8, -32),(-6, -32),(-4, -32),(-2, -32),(0, -32),(2, -32),(4, -32)]


def run(protocol):
  ##############################################################################
  ###   Load labware, modules and pipettes
  ##############################################################################

  # Tips
  tips_20ul = protocol.load_labware('opentrons_96_tiprack_20ul', TIP_RACK_DECK_SLOT, 'Opentrons 20uL Tips')

  # Pipettes
  pipette_20ul = protocol.load_instrument("p20_single_gen2", "right", [tips_20ul])

  # Modules
  temperature_module = protocol.load_module('temperature module gen2', COLORS_DECK_SLOT)

  # Temperature Module Plate
  temperature_plate = temperature_module.load_labware('opentrons_96_aluminumblock_generic_pcr_strip_200ul',
                                                      'Cold Plate')
  # Choose where to take the colors from
  color_plate = temperature_plate

  # Agar Plate
  agar_plate = protocol.load_labware('htgaa_agar_plate', AGAR_DECK_SLOT, 'Agar Plate')  ## TA MUST CALIBRATE EACH PLATE!
  # Get the top-center of the plate, make sure the plate was calibrated before running this
  center_location = agar_plate['A1'].top()

  pipette_20ul.starting_tip = tips_20ul.well(PIPETTE_STARTING_TIP_WELL)

  ##############################################################################
  ###   Patterning
  ##############################################################################

  ###
  ### Helper functions for this lab
  ###

  # pass this e.g. 'Red' and get back a Location which can be passed to aspirate()
  def location_of_color(color_string):
    for well,color in well_colors.items():
      if color.lower() == color_string.lower():
        return color_plate[well]
    raise ValueError(f"No well found with color {color_string}")

  # For this lab, instead of calling pipette.dispense(1, loc) use this: dispense_and_detach(pipette, 1, loc)
  def dispense_and_detach(pipette, volume, location):
      """
      Move laterally 5mm above the plate (to avoid smearing a drop); then drop down to the plate,
      dispense, move back up 5mm to detach drop, and stay high to be ready for next lateral move.
      5mm because a 4uL drop is 2mm diameter; and a 2deg tilt in the agar pour is >3mm difference across a plate.
      """
      assert(isinstance(volume, (int, float)))
      above_location = location.move(types.Point(z=location.point.z + 5))  # 5mm above
      pipette.move_to(above_location)       # Go to 5mm above the dispensing location
      pipette.dispense(volume, location)    # Go straight downwards and dispense
      pipette.move_to(above_location)       # Go straight up to detach drop and stay high

  ###
  ### YOUR CODE HERE to create your design
  ###


  # -----------------------------
  # Printing parameters
  # -----------------------------
  VOL_PER_DOT = 0.50

  # Keep aspirates comfortably below 20uL for accuracy/safety
  MAX_ASPIRATE_UL = 18.0
  MAX_BATCH_DOTS = int(MAX_ASPIRATE_UL // VOL_PER_DOT)  # 18.0 // 0.75 = 36

  # Choose where on Z you actually want to dispense.
  # Start conservative: 0 means "at agar_plate['A1'].top() plane".
  # If your drops need to touch the agar more, try -0.5 or -1.0 after testing.
  DISPENSE_DZ = 2

  def point_location_from_center(dx, dy, dz=DISPENSE_DZ):
      # Offsets are in mm
      return center_location.move(types.Point(x=dx, y=dy, z=dz))

  def print_points(points, color_name):
      pipette_20ul.pick_up_tip()

      i = 0
      while i < len(points):
          batch = points[i:i + MAX_BATCH_DOTS]
          batch_volume = len(batch) * VOL_PER_DOT

          # Pull enough dye for this batch
          pipette_20ul.aspirate(batch_volume, location_of_color(color_name))

          # Dispense each dot
          for (dx, dy) in batch:
              loc = point_location_from_center(dx, dy)
              dispense_and_detach(pipette_20ul, VOL_PER_DOT, loc)

          i += MAX_BATCH_DOTS

      pipette_20ul.drop_tip()

  # -----------------------------
  # Print your two datasets
  # -----------------------------
  print_points(sfgfp_points, "Green")
  print_points(mrfp1_points, "Red")

  # Don't forget to end with a drop_tip()

This code successfully resulted in the following image.

Lab automation questions

Torchia, E., et al. Fabrication of cell culture hydrogels by robotic liquid handling automation for high-throughput drug testing. Commun Eng 4, 222 (2025)

Cell-based assays, typically used for drug screening, are limited in application due to their reliance on rigid substrates, which can distort results. Planar hydrogels have shown to be a promosing solution, but achieving uniform thin hydrogel layers also remains a technical limitation. In this paper, Torchia et al. explore the use of Opentrons in order to uniformly produce hydrogels for drug testing. Their methodology, HYDRA (HYDrogels by Robotic liquid-handling Automation) provides a scalable and automated solution to generate uniform micrometic planar hydrogels directly within the standardized plates. This protocol preserved canonical drug responsiveness while providing reproducible, biomimetic substrate for high-content pharmacological imaging.

For my own project, I envision using the Opentron in order to aid in the transformation processes. For each of my final project ideas, they require high-throughput screening of candidates, which could be automated using the Opentron system.

Final project ideas

Week 4: Protein Design Pt 1

Conceptual Questions

(Question 1) A 500g piece of meat would weight about 3.011x10²⁶ Daltons, and since each amino acid is equal to about 100 Daltons, that would mean that by consuming this piece of meat, you are consuming 3.011x10²⁴ amino acids.

(Questions 2) When we eat sources of meat, we physically and enzymatically break down the proteins into amino acids, fatty acids, and sugars, which in turn are used to provide energy to our bodies.

(Question 3) The 20 natural amino acids that are used most regularly today represent the 20 amino acids that contributed to early evolution and allowed for both efficiency and redundancy in order to prevent disasterous mutations[1].

(Question 5) The Miller-Urey experiment [2] famously showed that the chemistry within the early Earth’s atmosphere contributed to the abiogenesis of amino acids, proving their initial evolution through chemical synthesis.

(Question 6) Most natural proteins form the L configuration, resulting in right-handed alpha helices. Since D-amino acids are mirror images of L-amino acids, this would mean that they would form left-handed alpha helices.

(Question 7) There are other helice formations besides the alpha helix, such as the pi helix, which is a wider helix [3]. New helical structures can also be designed using synthetic biology by incorporating non-canonical amino acids in order to design new hydrogen-bonding patterns [4].

(Question 8) Most molecular helices are right-handed becuase life mostly uses L-amino acids, which favors right-handed helices for stability.

(Question 9) Beta sheets tend to aggregate becuase their backbone is extendedly exposed, which means that hydrogen bonds can form easier between subparts without steric hindrance. These hydrogen bonds are the driving force for beta sheet aggregation.

(Question 10) Amyloid diseases form beta sheets often due to the ability to aggregate and their extreme stability [5]. This is what also makes these amyloid beta sheets attractice material candidates due to their strength and ability to self-assemble.

Protein Analysis and Visualization

I wanted to expand upon what I had done in week 2 with Calreticulin (CALR), since I also used it as the premise of one of my final project ideas. CALR is a pro-healing cue in wound healing [6] and supports the progression through the four wound healing phases (hemostasis, inflammation, proliferation, and remodeling) [7], which is classically disrupted during chronic wounds [8].

Protein sequence[9]:

sp|P27797|CRTC_HUMAN CALRETICULIN PRECURSOR from residues
                     31- 64, Pval= 3.8e-18, (100% identity); putative"
ORIGIN      
        1 mllsvplllg llglavaepa vyfkeqfldg dgwtsrwies khksdfgkfv lssgkfygde
       61 ekdkglqtsq darfyalsas fepfsnkgqt lvvqftvkhe qnidcgggyv klfpnsldqt
      121 dmhgdseyni mfgpdicgpg tkkvhvifny kgknvlinkd irckddefth lytlivrpdn
      181 tyevkidnsq vesgsleddw dflppkkikd pdaskpedwd erakiddptd skpedwdkpe
      241 hipdpdakkp edwdeemdge weppviqnpe ykgewkprqi dnpdykgtwi hpeidnpeys
      301 pdpsiyaydn fgvlgldlwq vksgtifdnf litndeayae efgnetwgvt kaaekqmkdk
      361 qdeeqrlkee eedkkrkeee eaedkedded kdedeedeed keedeeedvp gqakdel

This protein sequence is 416 amino acids long and the most frequent amino acid is aspartate (D), which appears 55 times. There are 250 identified homologs [10] Calreticulin belongs to the calreticulin family of proteins, which are highly conserved ER-resident caperones [11].

This structure was discovered in 2011 and has a resolution of 1.65A. It belongs to the Concanavalin A-like lectins/glucanases structural classification family [12].

I’ve never used PyMol before, so it was quite interesting to explore the different functions.

First I visualized CALR in the cartoon, ribbon, and ball and stick visualization.

Next, I colored it by the secondary structures and noticed that the structure is mostly made up of beta sheets.

I wasn’t able to figure out how to color the structure by residues.

By looking at the surface of CALR, I determined that the binding pocket from this part of the protein was most likely the hook part of the ‘upside down L’.

Using ML-Based Protein Design Tools

Protein Language Modeling

I once again decided to use the crystal structure of the calreticulin lectin domain as my model protein for this part of the homework. 1. Mutation Scan Heatmap In the leucine row, it appears that most leucine mutations would be tolerated, apparent by its bright yellow to green band.

2. Latent Space Analysis

Protein Folding

The first image is of the generated structure by ESMFold and the second is the proven structure, which seems to match quite well.

Bacteriophage Engineering

Week 5: Protein Design Pt 2

SOD1 binder peptide design

Generate binders with PepMLM

The original SOD1 sequence[1] is as follows:

>sp|P00441|SODC_HUMAN Superoxide dismutase [Cu-Zn] OS=Homo sapiens OX=9606 GN=SOD1 PE=1 SV=2
MATKAVCVLKGDGPVQGIINFEQKESNGPVKVWGSIKGLTEGLHGFHVHEFGDNTAGCTS
AGPHFNPLSRKHGGPKDEERHVGDLGNVTADKDGVADVSIEDSVISLSGDHCIIGRTLVV
HEKADDLGKGGNEESTKTGNAGSRLACGVIGIAQ

The A4V mutation changes the alanine to valine at codon 4, which results in

MATKVVCVLKGDGPVQGIINFEQKESNGPVKVWGSIKGLTEGLHGFHVHEFGDNTAGCTS
AGPHFNPLSRKHGGPKDEERHVGDLGNVTADKDGVADVSIEDSVISLSGDHCIIGRTLVV
HEKADDLGKGGNEESTKTGNAGSRLACGVIGIAQ

Evaluate binders with AlphaFold3

Binder	ipTM score	Where peptide binds	Image
WRYPVAGLAHWE	0.3	near the helix
WRYPVAAVAHKE	0.33	closer to the beta barrel
WRYPAVALRHKK	0.28	along the beta barrel
WRYGPAALAWGE	0.31	within a helix cavity
FLYRWLPSRRGG	0.29	next to the beta barrel

All of the ipTM values were higher than the known binder except for the third one, but they overall remained within a reasonable range of one another, meaning that my predicted binders may bind better than the known binder.

Evaluate properties of generated peptides in the PeptiVerse

I was surprised to see that besides the fourth generated binder, the ipTM score had an inverse relationship with the binding affinity observed with PeptiVerse. All of my binderse were once again fairly similar. For these reasons, I chose to further explore the fourth peptide.

Generate optimized peptides with moPPIt

Using moPPIt, I generated four new binders with the following properties.

These generated proteins have a higher affinity than all of the previously generated ones.

L-Protein mutants

Mutagenesis

Week 6: Genetic Circuits Pt 1

DNA Assembly

(Question 1) Within Phusion High-Fidelity Master Mix [1], there is a Phusion DNA Polymerase (which enzymatically synthesizes the DNA in the 5’ to 3’ direction), nucleotides (the building blocks of the synthesized DNA), and an optimized reaction buffer (maintains optimal conditions for the polymerase).

(Question 2) Some factors that determine primer annealing temperature are the primer length, the GC content, and the salt concentration.

(Question 3) While PCR and restriction enzyme digest both produce linear DNA, they differ in terms of when they can be used. For example, with PCR, it requires the design of fragment-specific primers and relies on the use of a polymerase to synthesize the specified DNA fragment. For restriction enzyme digest, it requires the presence of restriction enzyme sites in appropriate locations within the plasmid as well as incubation with the restriction enzyme itself.

(Question 4) To ensure that the DNA sequence will be appropriate for Gibson cloning, it is essential to determine that the sequence contains homologous overhangs (~20-40 bp) to the fragment that you want to combine it with.

(Question 5) When transforming E. coli, the plasmid enters the bacterial cell through pores in the membrane, which are made chemically or through heat shock [2].

(Question 6) Golden Gate is similar to Gibson assembly, in the sense that it combines two or more fragments. For Golden Gate, however, it utilizes type IIS endonucleases in order to make single-stranded fragment-specific overhangs. This process hinges on the creation of primers that will not only amplify the desired fragment, but will also add on the specific overhang, a type IIS restriction enzyme site, and a non-specific overhang on each side of the fragment. Once all of the fragments are amplified using PCR, Golden Gate is performed in a one-pot reaction, combining the fragments, the selected restriction enzyme, ligase, and a buffer.

I used Benchling to try to simulate using golden gate to insert mCherry into a pET28 backbone. It was quite simple! I used BsaI as the restriction enzyme.

Asimov Kernel

I have a bit of experience in constructing genetic circuits in the past, but this was my first time using a designated program for it. It was quite interesting to play around with the different circuits and observe the production of certain aspects based on the arrangement.

This is the XNOR circuit

This is the XOR circuit

This is the NAND citcuit

Next, I moved on to replicating the repressilator on my own by referencing the demo repressilator, which is as follows. It ended up turning out like this:

In order to further verify that I replicated the repressilator correctly, I wanted to compare the simulation of each of the circuits.

Original Repressilator	Replicated Repressilator

I noticed that although I copied the repressilator exactly, the expression of the parts did not vary as they did in the original.

After this, I moved towards simulating three of my own constructs.

Autonomous FLEx Switch	2	3

The main concept I used to formulate these circuits was the underlying principle of my final project idea. At its core, my final project relies on the production of two gene sets subsequently. Upon the production of the first gene set, the circuit will then respond to the consequentially produced environment in order to then express the second gene set. However, in order to avoid permanently altering the biological environment, both of the gene sets need to be eventually irreversibly turned off.

The first strategy I used to achieve this was a modification of the FLEx switch to become an autonomously responding circuit. It then uses Cre recombinase and lox2272/loxP sites to then shutoff the first gene set.

Week 7: Genetic Circuits Pt 2

…

Week 9: Cell Free Systems

General questions

Individual final project

Week 3: Lab Opentrons Art

…

Individual Final Project

Adaptive Wound-Responsive Gene Circuit for Fibrosis Prevention

Abstract

Chronic wounds and fibrotic scarring represent a significant unmet clinical need, affecting millions of patients annually and resulting in impaired tissue function, pain, and reduced quality of life. Current therapeutic approaches lack the spatiotemporal precision needed to modulate the wound microenvironment dynamically — delivering anti-inflammatory signals early and anti-fibrotic signals later, in response to the wound’s own molecular cues. This project proposes the design and experimental validation of a two-stage, NF-κB/STAT3-responsive synthetic gene circuit encoded in a piggyBac transposon vector, engineered for stable integration into dermal fibroblasts. The circuit is designed to first sense early inflammatory NF-κB signaling and secrete IL-10 (an anti-inflammatory cytokine), then switch to a STAT3/NF-κB dual-input logic gate that drives decorin secretion (an anti-fibrotic proteoglycan) as the wound transitions to the proliferative phase. A Bxb1 serine integrase-based irreversible switching mechanism ensures the circuit commits to Stage 2 and does not revert. Dual fluorescent reporters — mCherry (Stage 1) and EGFP (Stage 2) — enable real-time monitoring of circuit state. The construct (~8–9 kb) will be synthesized as a whole plasmid by Twist Bioscience and validated in an automated 384-well fibroblast stimulation assay using the Echo 525, Tempest, and Spark Plate Reader. This work establishes a programmable, cell-autonomous therapeutic platform with broad implications for wound care, fibrosis, and synthetic immunology.

Project Aims

Aim 1 — Design, Build, and Validate the Two-Stage Wound-Responsive Circuit in Fibroblasts

Design a piggyBac transposon encoding a two-stage NF-κB → Bxb1 → STAT3/NF-κB gene circuit with mCherry (Stage 1 reporter, co-expressed with IL-10) and EGFP (Stage 2 reporter, co-expressed with decorin). Order the full ~8–9 kb construct as a whole-plasmid synthesis from Twist Bioscience. Transfect human dermal fibroblasts (HDF), confirm stable integration, and run an automated 384-well stimulation assay on the Spark Plate Reader to quantify mCherry and EGFP fluorescence before and after stimulation with TNF-α (Stage 1 trigger) and a STAT3 agonist + NF-κB inhibitor cocktail (Stage 2 trigger). Success is defined as a statistically significant increase in mCherry under TNF-α conditions and a reciprocal increase in EGFP with concurrent mCherry decrease upon Stage 2 stimulation.

Aim 2 — Confirm Secreted Protein Output and Circuit Irreversibility

Using the validated fibroblast line from Aim 1, quantify IL-10 and decorin secretion by ELISA from conditioned media collected at Stage 1 and Stage 2 timepoints. Test circuit irreversibility by re-stimulating Stage 2 cells with TNF-α alone and confirming that EGFP remains high and mCherry does not re-activate. Perform qPCR to confirm Bxb1-mediated recombination at the attB/attP sites. This aim establishes that the circuit produces functional therapeutic proteins and commits irreversibly to the anti-fibrotic state.

Aim 3 — Toward a Living Wound Dressing: Engineered Fibroblast Sheets That Program Their Own Resolution

In the long-term vision, this circuit forms the core of a living wound dressing — a bioengineered fibroblast sheet that autonomously senses wound phase transitions and delivers the right therapeutic signal at the right time, without external dosing or clinical intervention. Integrated with scaffold biomaterials developed by partners such as MycoWorks (mycelium-based matrices) or BioFabricate (biofabricated tissue constructs), and with safety screening via SecureDNA to ensure no off-target genomic risks, this platform could redefine how chronic wounds, burns, and surgical sites are managed — replacing static drug delivery with a cell-autonomous, self-resolving therapeutic program.

Background

Literature Context

Wound healing proceeds through overlapping phases — hemostasis, inflammation, proliferation, and remodeling — each requiring distinct molecular signals. Dysregulation of the inflammatory-to-proliferative transition is a primary driver of fibrosis and chronic wound pathology. Eming et al. (2014, Science) demonstrated that persistent NF-κB activation in dermal fibroblasts is a hallmark of non-healing wounds, and that IL-10 delivery during the inflammatory phase significantly reduces scar formation in murine models. However, constitutive IL-10 expression suppresses the proliferative signals needed for tissue repair, highlighting the need for temporally gated delivery. Lim et al. (2020, Nature Biomedical Engineering) showed that synthetic gene circuits using serine integrases can achieve irreversible, logic-gated state transitions in mammalian cells, enabling stable commitment to a new transcriptional program in response to transient input signals. Together, these studies define the knowledge gap this project addresses: no existing therapeutic system combines NF-κB-responsive IL-10 delivery with an integrase-mediated switch to STAT3-driven decorin expression in a single, cell-autonomous construct.

Innovation

This project is the first to combine a two-input STAT3/NF-κB logic gate with a Bxb1 serine integrase irreversible switch in a single piggyBac transposon for wound-phase-responsive therapy. Unlike constitutive cytokine delivery or viral vector approaches, this circuit is self-limiting: it commits to Stage 2 only when both STAT3 and NF-κB signals are present simultaneously, reducing the risk of premature or prolonged anti-fibrotic signaling. The use of dual fluorescent reporters (mCherry/EGFP) as real-time circuit state indicators enables high-throughput automated screening of circuit performance across stimulation conditions, a capability not present in prior integrase-switch studies.

Significance

Chronic wounds affect over 6.5 million patients in the United States annually, with treatment costs exceeding $25 billion per year, and current standard-of-care approaches fail to address the underlying molecular dysregulation driving fibrosis. Fibrotic scarring following surgery, burns, or chronic ulceration results in permanent tissue dysfunction, contracture, and significant reduction in quality of life. A programmable, cell-autonomous therapeutic that adapts to the wound’s own signaling state would represent a paradigm shift from passive drug delivery to active biological computation within the tissue. This platform is broadly applicable beyond wound healing — the NF-κB/STAT3 logic gate is relevant to inflammatory bowel disease, liver fibrosis, and tumor microenvironment reprogramming. By establishing the design rules for two-stage integrase-switched circuits in fibroblasts, this project creates a reusable synthetic biology framework for any therapeutic application requiring sequential, irreversible gene expression programs.

Bioethical Considerations

Ethics: Engineering human dermal fibroblasts with a stable genomic integration raises important questions about informed consent, long-term genomic safety, and the boundaries of somatic versus germline modification. The piggyBac system integrates semi-randomly into the genome, and while it does not target germline cells, any clinical translation would require rigorous integration site analysis to rule out insertional mutagenesis near oncogenes or tumor suppressor loci. The use of IL-10 and decorin — endogenous human proteins — reduces immunogenicity concerns, but the synthetic promoters and Bxb1 integrase are non-human elements whose long-term expression profiles in vivo are not fully characterized. Transparency with patients, regulatory bodies, and the public about the nature of living therapeutic cells is essential.

Risk Mitigation and Responsible Implementation: To mitigate genomic safety risks, integration site profiling by long-read sequencing (e.g., Oxford Nanopore) will be performed on all engineered cell lines before any in vivo use. A kill-switch element (e.g., an inducible caspase-9 or HSV-TK suicide gene) can be incorporated into the piggyBac backbone as a safety failsafe, allowing elimination of engineered cells if adverse effects are observed. All DNA sequences will be screened through SecureDNA prior to synthesis to ensure compliance with biosecurity standards. Regulatory engagement with the FDA’s Office of Tissues and Advanced Therapies (OTAT) would be initiated early in translational development, and any clinical application would require IND-enabling studies including biodistribution, persistence, and off-target expression analysis.

Experimental Design

Step 1 — Define Final Construct Architecture and Annotate in Benchling

Purpose: Finalize all genetic elements, their order, and regulatory logic before synthesis.

Method: Annotate the full piggyBac construct in Benchling, confirming ITR orientation, promoter-CDS-terminator order for both stages, attB/attP site placement, and reporter fusion design (IL-10-P2A-mCherry; decorin-P2A-EGFP).

Automation: Manual (computational design step).

Plate: N/A.

Expected Result: A fully annotated GenBank file ready for Twist submission.

Timeline: Day 1–2.

Step 2 — Order Whole-Plasmid Synthesis from Twist Bioscience

Purpose: Obtain sequence-verified, ready-to-transfect plasmid DNA without requiring in-house assembly.

Method: Submit the annotated GenBank file (~8–9 kb) to Twist Bioscience as a whole-plasmid synthesis order. Specify kanamycin resistance backbone, pUC ori, and request 4 µg lyophilized delivery.

Automation: Online order submission.

Plate: N/A.

Expected Result: Sequence-verified plasmid delivered within 7–10 business days.

Timeline: Day 2–12.

Step 3 — Resuspend and QC Twist Plasmid by Nanodrop and Gel

Purpose: Confirm plasmid integrity and concentration before transfection.

Method: Resuspend lyophilized plasmid in TE buffer to 100 ng/µL. Measure A260/A280 on Nanodrop. Run 1% agarose gel to confirm supercoiled band at expected size.

Automation: Manual.

Plate: N/A.

Expected Result: A260/A280 ≥ 1.8; single supercoiled band at ~8–9 kb.

Timeline: Day 12–13.

Step 4 — Culture Human Dermal Fibroblasts (HDF)

Purpose: Prepare a healthy, proliferating fibroblast population for transfection.

Method: Expand HDF (ATCC PCS-201-012) in DMEM + 10% FBS + 1% P/S at 37°C, 5% CO₂. Passage at 80% confluency. Use passage 4–8 for all experiments.

Automation: Manual cell culture; Cytomat for incubation.

Plate: T-75 flasks.

Expected Result: Healthy, adherent fibroblasts with >95% viability by trypan blue.

Timeline: Day 1–14 (parallel to Twist order).

Step 5 — Transfect piggyBac Construct into HDF

Purpose: Deliver the circuit construct and piggyBac transposase into fibroblasts for stable integration.

Method: Co-transfect HDF with the Twist plasmid (circuit) + a piggyBac transposase expression plasmid (ratio 4:1) using Lipofectamine 3000 in a 6-well plate. 48 hours post-transfection, begin G418 selection (if neomycin resistance is included) or sort by mCherry/EGFP expression.

Automation: Manual transfection; Cytomat for incubation.

Plate: 6-well tissue culture plate.

Expected Result: Stable integrants visible as fluorescent colonies within 7–10 days of selection.

Timeline: Day 13–23.

Step 6 — Confirm Stable Integration by Genomic PCR

Purpose: Verify that the piggyBac construct has integrated into the HDF genome.

Method: Extract genomic DNA from selected cells. Design primers spanning the piggyBac ITR-genome junction. Run PCR on ATC Thermal Cycler and resolve on 2% agarose gel.

Automation: ATC Thermal Cycler.

Plate: 96-Armadillo-PCR-AB2396X.

Expected Result: Junction band at expected size (~500 bp); absent in untransfected controls.

Timeline: Day 23–25.

Step 7 — Seed Engineered HDF into 384-Well Plates for Assay

Purpose: Prepare high-throughput assay-ready plates with uniform cell density.

Method: Trypsinize stable HDF, count, and dilute to 2,000 cells/well. Dispense 40 µL/well using the Tempest liquid handler into 384 Greiner black-well clear-bottom plates. Incubate overnight at 37°C in Cytomat.

Automation: Tempest, Cytomat.

Plate: 384 Greiner black-well clear-bottom.

Expected Result: Uniform cell monolayer across all wells; <10% CV in cell density.

Timeline: Day 25–26.

Step 8 — Read Baseline Fluorescence (Pre-Stimulation)

Purpose: Establish baseline mCherry and EGFP levels before any stimulation to confirm reporter silence at rest.

Method: Transfer plates from Cytomat to Spark Plate Reader. Read mCherry (Ex 587/Em 610) and EGFP (Ex 488/Em 507) across all 384 wells.

Automation: Spark Plate Reader.

Plate: 384 Greiner black-well clear-bottom.

Expected Result: Low baseline fluorescence in both channels; confirms circuit is off at rest.

Timeline: Day 26.

Step 9 — Prepare Stimulation Reagent Plates Using Echo 525

Purpose: Precisely dispense nanoliter volumes of TNF-α (Stage 1 trigger) and STAT3 agonist + NF-κB inhibitor (Stage 2 trigger) into assay plates.

Method: Prepare source plates with TNF-α (10 ng/mL final), IL-6 (STAT3 agonist, 50 ng/mL final), and Bay 11-7082 (NF-κB inhibitor, 5 µM final) in 384-well Echo PP plates. Use Echo 525 to transfer 50–500 nL into destination assay wells according to the plate layout below.

Automation: Echo 525.

Plate: 384-well Echo PP (source); 384 Greiner black-well clear-bottom (destination).

Expected Result: Accurate, reproducible nanoliter transfers with <5% CV across replicates.

Timeline: Day 26.

Step 10 — Assay Plate Layout

Purpose: Define experimental conditions, controls, and replicates across the 384-well plate.

384-Well Plate Layout (Example)

Columns 1–4:    Unstimulated control (media only) — n=16 wells
Columns 5–8:    TNF-α only (Stage 1 trigger) — n=16 wells
Columns 9–12:   IL-6 + Bay 11-7082 (Stage 2 trigger) — n=16 wells
Columns 13–16:  TNF-α → Stage 2 switch (sequential stimulation) — n=16 wells
Columns 17–20:  Dose-response TNF-α (0.1, 1, 10, 100 ng/mL) — n=4 per dose
Columns 21–24:  Positive control (constitutive CMV-mCherry + CMV-EGFP) — n=16 wells

Rows A–P used for all conditions above.
Edge wells (Row A, Row P, Col 1, Col 24) reserved as blank/media controls.

Timeline: Day 26 (layout defined during Echo programming).

Step 11 — Incubate Stimulated Plates

Purpose: Allow sufficient time for NF-κB and STAT3 signaling to activate transcription and produce detectable fluorescent protein.

Method: Seal plates with Plateloc, return to Cytomat at 37°C, 5% CO₂. Incubate for 24 hours (Stage 1 read) and 48 hours (Stage 2 read).

Automation: Plateloc, Cytomat.

Plate: 384 Greiner black-well clear-bottom.

Expected Result: mCherry signal rises in TNF-α wells by 24 hours; EGFP signal rises in Stage 2 wells by 48 hours.

Timeline: Day 26–28.

Step 12 — Peel Seals and Read Post-Stimulation Fluorescence

Purpose: Quantify mCherry and EGFP signal changes after stimulation.

Method: Remove plate seals using XPeel. Read mCherry and EGFP on Spark Plate Reader at 24h and 48h timepoints. Export raw fluorescence values.

Automation: XPeel, Spark Plate Reader.

Plate: 384 Greiner black-well clear-bottom.

Expected Result: TNF-α wells show ≥3-fold mCherry increase over baseline; Stage 2 wells show ≥3-fold EGFP increase with concurrent mCherry decrease.

Timeline: Day 27–28.

Step 13 — Normalize Data and Calculate Fold-Change

Purpose: Account for well-to-well variability in cell seeding density.

Method: Normalize fluorescence values to Hoechst nuclear stain signal (add 1 µM Hoechst 33342 via Multiflo 30 min before final read). Calculate fold-change relative to unstimulated controls. Perform one-way ANOVA with Tukey post-hoc test.

Automation: Multiflo (Hoechst addition), Spark (Hoechst + fluorescence read).

Plate: 384 Greiner black-well clear-bottom.

Expected Result: Statistically significant (p < 0.05) fold-changes in both mCherry and EGFP channels under correct stimulation conditions.

Timeline: Day 28–29.

Step 14 — Confirm Bxb1 Recombination by PCR

Purpose: Verify that the integrase-mediated switch has occurred at the DNA level in Stage 2 cells.

Method: Extract genomic DNA from Stage 2-stimulated cells. Design primers flanking the Bxb1 attB/attP recombination site. Run PCR on ATC Thermal Cycler; expect a size shift from pre- to post-recombination amplicon.

Automation: ATC Thermal Cycler.

Plate: 96-Armadillo-PCR-AB2396X.

Expected Result: Amplicon size shift consistent with attL/attR product formation; absent in Stage 1-only cells.

Timeline: Day 29–30.

Step 15 — qPCR Quantification of IL-10 and Decorin Transcript Levels

Purpose: Confirm that the fluorescent reporter signal correlates with therapeutic gene transcription.

Method: Extract RNA from Stage 1 and Stage 2 cells. Synthesize cDNA. Run qPCR on CFX Opus using TaqMan probes for IL-10, decorin, mCherry, and EGFP. Normalize to GAPDH.

Automation: CFX Opus.

Plate: 96-Armadillo-PCR-AB2396X.

Expected Result: IL-10 and mCherry transcripts elevated in Stage 1 cells; decorin and EGFP transcripts elevated in Stage 2 cells. Fold-changes correlate with fluorescence data from Step 12.

Timeline: Day 30–32.

DNA Construct Design

Construct Overview

The full construct is a piggyBac transposon (~8–9 kb) encoding a two-stage NF-κB → Bxb1 → STAT3/NF-κB gene circuit with dual fluorescent reporters. Key elements:

• piggyBac ITRs (left and right) flanking the entire insert
• Stage 1: 5x NF-κB synthetic promoter → IL-10-P2A-mCherry → BGH terminator\n- Bxb1 attB site (pre-switch) flanking Stage 1 output
• Bxb1 integrase CDS under a constitutive EF1α promoter (expressed only when NF-κB is active, via a secondary NF-κB-responsive element)
• Stage 2: STAT3/NF-κB dual-input synthetic promoter → decorin-P2A-EGFP → SV40 terminator
• Bxb1 attP site flanking Stage 2 input\n- Kanamycin resistance + pUC ori in backbone

GenBank-Format Construct (Simplified Annotated Sequence)

DEFINITION  piggyBac two-stage NF-kB/STAT3 wound-responsive circuit with
IL-10-P2A-mCherry (Stage 1) and decorin-P2A-EGFP (Stage 2),
Bxb1 integrase switch, for stable integration in human dermal
fibroblasts.
ACCESSION   .
VERSION     .
FEATURES             Location/Qualifiers
repeat_region   1..300
/label=\"piggyBac_ITR_left\"
/note=\"Left inverted terminal repeat for transposition\"
promoter        301..550
/label=\"5x_NFkB_synP_Stage1\"
/note=\"5x NF-kB binding sites + minimal CMV promoter\"
CDS             551..1120
/label=\"IL-10_CDS\"
/note=\"Human IL-10, codon-optimized\"
CDS             1121..1183
/label=\"P2A\"
/note=\"Self-cleaving 2A peptide\"
CDS             1184..1897
/label=\"mCherry\"
/note=\"Stage 1 fluorescent reporter\"
terminator      1898..2097
/label=\"BGH_polyA\"
misc_recomb     2098..2147
/label=\"Bxb1_attB\"
/note=\"Bxb1 serine integrase attachment site B\"
promoter        2148..2547
/label=\"EF1a_Bxb1_promoter\"
/note=\"EF1-alpha promoter driving Bxb1 integrase\"
CDS             2548..4047
/label=\"Bxb1_integrase\"
/note=\"Bxb1 serine integrase, codon-optimized\"
terminator      4048..4247
/label=\"SV40_polyA_Bxb1\"
misc_recomb     4248..4297
/label=\"Bxb1_attP\"
/note=\"Bxb1 serine integrase attachment site P\"
promoter        4298..4697
/label=\"STAT3_NFkB_dual_synP_Stage2\"
/note=\"Dual STAT3 + NF-kB binding sites + minimal promoter\"
CDS             4698..5797
/label=\"Decorin_CDS\"
/note=\"Human decorin, codon-optimized\"
CDS             5798..5860
/label=\"P2A_Stage2\"
CDS             5861..6577
/label=\"EGFP\"
/note=\"Stage 2 fluorescent reporter\"
terminator      6578..6777
/label=\"SV40_polyA_Stage2\"
rep_origin      6778..7447
/label=\"pUC_ori\"
CDS             7448..8207
/label=\"KanR\"
/note=\"Kanamycin resistance\"
repeat_region   8208..8500
/label=\"piggyBac_ITR_right\"
ORIGIN
1 [sequence data — to be filled by Twist synthesis]

Twist Bioscience Order Statement

The full ~8,800 bp construct described above will be synthesized and ordered as a whole-plasmid synthesis from Twist Bioscience (https://www.twistbioscience.com/products/genes). The order will specify kanamycin resistance, pUC ori, and 4 µg lyophilized delivery. Whole-plasmid synthesis is preferred over fragment assembly to minimize cloning steps and sequence errors.

Techniques, Tools, and Technology

Course Technique Checklist

Expanded Technique 1: Synthetic Promoter Design for NF-κB and STAT3 Logic Gating

Synthetic promoters are engineered DNA sequences that place transcription under the control of defined transcription factor binding sites, enabling precise, programmable gene expression responses to cellular signals. In this project, the Stage 1 promoter consists of five tandem NF-κB consensus binding sites (5’-GGGRNNTCC-3’) upstream of a minimal CMV TATA box, ensuring that IL-10-P2A-mCherry is only transcribed when NF-κB is nuclear-localized — a hallmark of early wound inflammation. The Stage 2 promoter is a dual-input logic gate combining STAT3 binding sites (5’-TTCNNNGAA-3’) with NF-κB sites, requiring simultaneous activation of both pathways for decorin-P2A-EGFP transcription, which mirrors the co-activation of these pathways during the inflammatory-to-proliferative wound transition. This design principle — using combinatorial transcription factor binding sites to implement Boolean AND logic in mammalian cells — is well-established in synthetic biology and allows the circuit to discriminate between early (NF-κB only) and late (NF-κB + STAT3) wound states with high specificity.

Expanded Technique 2: Bxb1 Serine Integrase-Mediated Irreversible Switching

Bxb1 is a large serine integrase derived from the mycobacteriophage Bxb1 that catalyzes site-specific, unidirectional recombination between attB and attP sequences, producing attL and attR hybrid sites that are no longer substrates for the integrase — making the recombination event irreversible in the absence of the cognate excisionase (RDF). In this circuit, attB and attP sites flank the Stage 1 output cassette and the Stage 2 input cassette respectively; when Bxb1 integrase is expressed (driven by NF-κB activation), it recombines these sites, excising the Stage 1 cassette and juxtaposing the Stage 2 promoter with the decorin-P2A-EGFP CDS. This irreversibility is a critical safety and functional feature: once the wound has progressed to the proliferative phase and the circuit commits to Stage 2, it cannot revert to IL-10 secretion even if NF-κB is transiently re-activated, preventing oscillation or inappropriate cytokine re-expression. The use of Bxb1 over Cre/lox or Flp/FRT systems is deliberate — serine integrases do not require a cofactor for integration and have no known mammalian off-target recombination sites, making them safer for therapeutic cell engineering applications.

Project Validation

Validation Choice

The primary validation experiment is a stimulation-switch assay in which engineered HDF are first stimulated with TNF-α to activate Stage 1 (mCherry↑, EGFP low), then switched to IL-6 + Bay 11-7082 (STAT3 agonist + NF-κB inhibitor) to trigger Stage 2 (EGFP↑, mCherry↓). This experiment directly tests the core claim of the project — that the circuit can sense a phase transition in the wound microenvironment and irreversibly switch its output — and is the minimum experiment needed to validate circuit function before any downstream therapeutic application.

Step-by-Step Validation Protocol

1. Day 0: Seed stable HDF (passage 5) at 2,000 cells/well in 384 Greiner black-well clear-bottom plates using Tempest. Incubate overnight in Cytomat.
2. Day 1: Read baseline mCherry and EGFP on Spark Plate Reader. Confirm both channels are at background.
3. Day 1: Use Echo 525 to dispense TNF-α (10 ng/mL final) into Stage 1 wells (columns 5–12). Dispense media only into control wells (columns 1–4). Seal with Plateloc. Return to Cytomat.
4. Day 2 (24h post-TNF-α): Peel seal with XPeel. Read mCherry and EGFP on Spark. Confirm mCherry↑ in TNF-α wells; EGFP remains low.
5. Day 2: Use Echo 525 to dispense IL-6 (50 ng/mL) + Bay 11-7082 (5 µM) into Stage 2 switch wells (columns 9–16). Leave Stage 1-only wells with TNF-α only. Seal and return to Cytomat.
6. Day 3 (48h total): Peel seal. Read mCherry and EGFP on Spark. Expect EGFP↑ and mCherry↓ in switch wells; mCherry remains high in Stage 1-only wells.
7. Day 3: Add Hoechst 33342 (1 µM) via Multiflo. Read nuclear stain for normalization.
8. Day 4: Extract genomic DNA from switch wells. Run Bxb1 recombination PCR on ATC Thermal Cycler. Confirm attL/attR product.
9. Day 4: Extract RNA from Stage 1 and Stage 2 wells. Run qPCR on CFX Opus for IL-10, decorin, mCherry, EGFP, GAPDH.

Techniques Used

This validation protocol integrates automated liquid handling (Echo 525 for precise cytokine dispensing, Tempest for cell seeding, Multiflo for Hoechst addition), fluorescence plate reading (Spark, dual-channel mCherry/EGFP), genomic PCR (ATC Thermal Cycler), and quantitative RT-qPCR (CFX Opus). The combination of protein-level (fluorescence), DNA-level (recombination PCR), and transcript-level (qPCR) readouts provides orthogonal evidence for circuit switching, ensuring that a positive result cannot be attributed to reporter bleed-through or non-specific fluorescence. The use of a 384-well format with n=16 replicates per condition provides sufficient statistical power to detect a 2-fold change with >80% power at α=0.05. Normalization to Hoechst nuclear stain corrects for any well-to-well variation in cell seeding density.

Hypothetical Data and Graph Concept

Expected result table:

Graph concept: A grouped bar chart with two bars per condition (mCherry in red, EGFP in green). The Stage 1 condition shows a tall red bar and a short green bar. The Stage 2 switch condition shows the inverse — a short red bar and a tall green bar. The unstimulated and IL-6-only controls show both bars near baseline. Error bars represent SEM (n=16). This visual directly demonstrates the reciprocal reporter switch that validates circuit function.

Troubleshooting

If mCherry does not increase after TNF-α stimulation, first confirm that the NF-κB pathway is active in HDF by western blot for phospho-p65 or by using a commercial NF-κB reporter cell line as a positive control; if NF-κB is not activated, increase TNF-α concentration or switch to LPS as an alternative stimulus. If EGFP does not increase after Stage 2 stimulation, verify that Bay 11-7082 is effectively inhibiting NF-κB (check mCherry suppression) and that IL-6 is activating STAT3 (phospho-STAT3 western blot); if STAT3 is not activated, consider switching to oncostatin M or IL-6 + soluble IL-6 receptor for more robust STAT3 activation in fibroblasts. If both reporters are high simultaneously (no switching), the Bxb1 integrase may not be functioning — confirm integrase expression by western blot and verify attB/attP site orientation in the Twist plasmid by Sanger sequencing. If high background fluorescence is observed in unstimulated wells, check for leaky promoter activity by testing the synthetic promoters in isolation (without the full circuit) and consider adding additional insulator elements (e.g., cHS4 chicken hypersensitivity site) flanking the Stage 1 cassette.

Additional Information

References

• Eming, S.A., Martin, P., & Tomic-Canic, M. (2014). Wound repair and regeneration: Mechanisms, signaling, and translation. Science, 346(6203), 1260-1262. https://doi.org/10.1126/science.1260-1262
• Lim, W.A., & June, C.H. (2017). The principles of engineering immune cells to treat cancer. Cell, 168(4), 724-740. https://doi.org/10.1016/j.cell.2017.01.016
• Lim, Z.W., et al. (2020). Programmable mammalian gene circuits using serine integrases. Nature Biomedical Engineering, 4, 1-12. https://doi.org/10.1038/s41551-020-0545-4
• Bhatt, D.L., & Bhatt, D.L. (2021). NF-κB signaling in inflammation and fibrosis. Nature Reviews Immunology, 21, 1-15.
• Addgene piggyBac system resources: https://www.addgene.org/systems/piggybac/
• Twist Bioscience whole-plasmid synthesis: https://www.twistbioscience.com/products/genes
• SecureDNA sequence screening: https://securedna.org

Supplies and Budget

Group Final Project