Team:Manchester/Repair

UoM iGEM | Project Cutiful

DECAPEPTIDE

Manchester iGEM 2019 is proud to present its most recent development, integrated as a result of our human practices outreach at the Manchester Museum: repair. Not only a dye, we seek to present a product holistic in its approach to hair care. Cutiful is an alternative to damaging hair dyes – and goes further. We straighten and repair hair too.

Our Decapeptide Module…

In 7 bullet points:

1.

Cysteine-rich peptides- were researched (Human Practices, Act II: The Community Festival), and their ability to infiltrate the cortex and repair disulphide bridges was studied through literature.

2.

New Registry Part: PepG - we selected one specific decapeptide, PepG, which did not belong to the iGEM registry.

3.

Investigating Auto-secretion- possible auto-secretion, due to PepG’s small size, was checked.

4.

OmpA secretion signal- In case of lack of auto-secretion, an OmpA secretion signal (Type II secretion) was added to PepG and this sequence was synthesised, alongside PepG alone.

5.

His-tag for Qualitative Detection- Detection of PepG was from the start recognised as the most challenging part of the project. For this reason, two versions of each construct were synthesised, with and without a His-tag. This enabled the use of His-tag purification columns to determine the presence of PepG.

6.

Biomolecular probe for Quantitative Detection- As an alternative detection method in case the His-tag inhibited the secretion or expression of the peptides, a quantitative method was devised, relying on biomolecular probes to determine the relative proportion of sulphur present in a protein or peptide.

7.

Product Comparison: Hair Repair on the Market - Currently marketed hair-repair products were compared, and attributes directly relevant to ‘Cutiful’ were analysed. This enabled us to situate where, in regards to available products, ‘Cutiful’ situates.

Decapeptide – a play in four parts

Or: Mujtaba’s story. It involves everyone else, a little, but mostly burning gloves, Western blots, a lot more Chemistry than expected, and the assumption that everything was going to be straight forward.

Prologue

Anyone who has ever used a pair of straighteners can tell you (and that would happen to be about 45% of women): it comes with the lovely scent of burnt, reminiscent of the time something was forgotten in the oven. According to a 2014 Mintel report: over four in ten women own straighteners. Operating at temperatures of up to 235°C (455°F), the heat from straighteners is not only damaging to the user’s hair but also causes a significant number of burns in reported children cases. Straightening hair can also be done using relaxers: these weaken and break the disulphide bonds which shape hair, straightening it. Due to the strong alkalis used, repeated straightening risks not only hair damage but also chemical burns. The breaking of keratin’s disulphide bonds, through heat, chemical agents and/or sunlight, leads to hair weakening and gives the appearance of damage and brittleness, whilst making the hair fragile.

Cruz et al. highlights the use of particularly cysteine-rich peptides, short enough to infiltrate the hair cortex, which enable the formation of new disulphide bonds. These “decapeptides” have been shown to be able to both strengthen and straighten hair. Being protein based, decapeptides are ideal for Cutiful’s use of bacterial production and secretion.

Act I: Design

“I found a dimpled spider, fat and white…” – Robert Frost, “Design”

Due to the time constraint placed upon our project, we chose to focus on one specific decapeptide – PepG. We selected this peptide due its affinity for hair; high straightening efficiency at low concentrations and straightening properties based on the findings from Cruz et al..

In order for our decapeptides to be functional, they must be able to infiltrate the cortex, and therefore need to be secreted from bacterial cells. Due to the small size of the decapeptide (1.36kDa or 2.72kDa - as PepG exists primarily as dimers) we hypothesised the protein might auto-secrete. Furthermore, to test the efficiency of PepG secretion using a N-terminal secretion tag, we also created constructs with an N-terminal, type II, secB-dependent signal sequence, OmpA (BBa_K208003)

decapeptide-constructs
decapeptide-constructs

Fig.1 Decapeptide constructs: two of four variations tested. OmpA refers to the N-terminal type II secB dependent secretion signal. The LacUV5 promoter, lac operator and T1/T7 terminators were taken from the plasmid we transformed into, while the ribosome binding site (RBS), and coding region were constructed via primer overlap. The constructs were cloned into the plasmid through IIS overlaps generated by the primer overlap construction.

Type II secretion is a two-step process where first the protein is trafficked to the periplasm and then secreted to the environment. This specific signal sequence is called “secB-dependent”: this appellation refers to the mechanism by which the protein is trafficked to the periplasm. When the signal sequence is transcribed by a ribosome, the secB protein recognises it and binds to the protein sequence. The now secB-bound protein is trafficked to secA at the inner membrane, where the protein is translocated into the periplasm. Once in the periplasm, the signal sequence is cleaved. The protein will either slip out to the environment through the outer membrane or be recognised and secreted by further secretion complexes (Figure 2).

SecB dependent secretion diagram

Fig.2 SecB dependent secretion diagram adapted from. After protein synthesis, SecB recognises the protein with signal sequence, binding to it and trafficking it to the inner membrane. Then, SecA ATPase translocates the protein across the membrane into the periplasm, where it can fold, have the signal sequence cleaved, and then be further secreted by a secreton: secretion machinery composed of 12-16 proteins.

We selected this secretion method for two reasons. First, the N-terminal signal sequence is cleaved in the periplasm, which leads us to believe we will be left with a peptide very similar to PepG (CBS signalP-5.0 [http://www.cbs.dtu.dk/services/SignalP/index.php]) . Since our decapeptide is so small, a secretion signal of 18-30 amino acids is likely to drastically alter function: cleavage is imperative. Secondly, the type II secB-dependent secretion pathway is naturally used for the majority of secreted proteins from bacteria and so is more likely to work with our decapeptides, at an acceptable secretion rate.

likelihood that the decapeptide gene will be secreted

Fig.3 Graph showing the likelihood that the decapeptide gene will be secreted. It also shows that the most likely secretion pathway is Sec dependent and cleaved by signal peptidase (CBS signalP-5.0)

As we designed the parts, we also kept in mind the detection, purification and quantification of our decapeptides. The size of our decapeptide causes two challenges:

1. They fall at the limits of sensitivity for most common detection and purification methods. Any results using these traditional means would likely be both inaccurate and unreliable.

2. Methods making an assumption as to the relative abundances of amino acids in proteins, which is most of the traditional detection methods, will fall apart on short chains like our 11 amino acid-long decapeptide, because of its particularly high cysteine content.

Therefore, to improve purification, we decided to use 5 kDa spin columns (traditionally used for concentrating protein) to remove larger proteins and isolate the smaller peptides, such as our decapeptide. To aid in the initial detection of our peptides we decided to make two further C-terminal His-tagged constructs of the sequences shown in Fig.1 i.e. PepG alone, and PepG with the OmpA secretion signal. Resulting in two more constructs (Figure 4):

Fig.4 Decapeptide constructs: constructs from Figure 1 with an added His-tag (His).

This makes a total of four constructs:

1. PepG alone, to check for auto-secretion

2. PepG with an N-terminal secretion tag: OmpA, to ensure at least one construct will secrete.

3. PepG with a C-terminal His-tag, for easy detection of auto-secretion (though the His-tag may hinder it, which is why we have kept PepG alone as a construct).

4. PepG with both OmpA and His-tag, for easy detection of Type II secretion.

Aside: The ‘Plexes

“What’s in a name?” – William Shakespeare, “The Tragedy of Romeo and Juliet”, Act II, Scene 2.

Integral to marketing Cutiful as a competitive and attractive product, we decided to compare our bacteria’s abilities and mode of action to hair care options already present on the market. We focused especially on that which gives Cutiful an edge over regular dyes: its repair module.

Present on the markets are many shampoos, conditioners and various cream products which claim to repair the ultrastructure of the hair. Particularly, in recent years, well-known brands have been coming out with lines boasting of containing ‘liquid keratin’ (Schwarzkopf), ‘amino-fix’ technologies (Knight and Wilson,), ‘PRO-V’ (Pantene), or other publicly known ‘healthy’ molecules and oils (Elvive, Dove, Garnier…). A campaign has been seen in advertisements to promote compounds able to close the shaft’s scales, often accompanied by clean-cut diagrams of rachitic hair undergoing transformation to a smoother surface.

With the aim to relate Cutiful to products currently on the market, two hair care options were selected. Their lists of ingredients were perused and analysed, in order to work out the most likely compound responsible for the ‘repair’ aspect of their advertising, and the methods through which these compounds could achieve this was reasoned.

Product A was found online in a post by the Lab Muffin (https://labmuffin.com). Michelle, the blogger of Lab Muffin, describes product A to be part of a line of “hair repair, ‘bond building’ treatments” (Lab Muffin, 2018), popular amongst those who dye and therefore bleach their hair regularly. After going through the composition of product A, she states bis-aminopropyl diglycol dimaleate to be the most likely binding agent (Fig. 5), and compares what product A claims to do to results experienced by users.

structure

Fig. 5 Structure of bis-aminopropyl diglycol dimaleate, (Michelle, 2018)

Michelle explains the origin of damage in hair dyeing processes (project description), theorises a mode of action for the binding agent and highlights the lack of scientific trials and experiments which could back up Product A’s claim. She also goes on to describe further products belonging to the same range, and other possible applications of Product A.

Scene 1: Mechanism of action

When bleaching and perming hair, the disulphide bridges formed between two cysteine residues are broken to generate two free thiol groups. The pKa of these is about 8.5, meaning that at pH = 8.5, 50% of cysteine residues have a thiol group and 50% exist as counter-ion thiolate:

Fig. 6 structure of thiol and counter-ion thiolate. They exist in a 1:1 ratio at pH=8.5, (Poole, 2015)

It is the breaking of the disulphide bonds in keratin which straightens and damages the hair.

According to the patent (United States Patent No. 15/182,795, 2016), Product A works by forming covalent bonds between the free thiol groups and the maleate moieties on what they refer to as the binding agent. The binding agent exists in a multitude of forms, though they all have in common the two maleate moieties at each end, ionically bonded to a cation of overall charge 2+ (which can be Ca2+, or any chain of backbone composition CxH(2x+3y)NyOz as above, where 2 ≤ x+y+z ≤ 25.) In fig. 7, the cationic chain is “bis-aminopropyl diglycol”, formula: C10H26N2O3 . Through a thiol-ene Michael addition, the thiols reacts with the double bond on the maleate: this forms a covalent bond between the maleate moiety and the keratin polymer (K on diagram):

addition-rxn

Fig. 7 Base/nucleophile catalysed Michael Thiol-ene addition, where keratin polymer is labelled K. Cationic chain bound to the carboxylic moiety not shown.

This is believed to be the mode of action through which product A claims to repair hair. This reaction happens at both ends of the binding agent, resulting in a bridge between keratin polymers.

In her blog (Lab Muffin, 2018), Michelle points out that when in water the ionic COO- +NH3 linkage may weaken and break. Ammonium ions are known to be quite readily soluble in aqueous media, though the molecule in question has a long hydrophobic chain which would hinder solvation. It is possible that the binding agent will separate into free cationic chains and anionic maleate-capped keratin over time. This would prevent the reformation of natural disulfide bridges, since the free thiols would no longer exist: making continued use of product A necessary, lest the hair reverts back to its damaged state. (Though, because of the maleate capping the ends, only product A, or variations thereof, would be able to give it back shine and strength: making its continued use a necessity.)

Furthermore, it is of note that some Michael additions are in fact reversible reactions: if this were the case, then product A would exist in the hair as a mixture of bound and unbound agent (the extent of which is given by the equilibrium constant, specific to this Michael addition). This would however solve the possible exclusivity issue arising from the use of product A, since the free thiols would re-form.

Comparison with product B, a product with similar claims to A which appeared on the market following the latter’s success, shows the most probable binding agent to be lauridimonium hydroxypropyl hydrolysed keratin. An analogue of this molecule is found in the patent for a third popular brand of hair care products, which states that their “leave-on hair creme comprises one or more cationic polymers selected from the group consisting of […] lauryldimonium hydroxypropyl hydrolyzed soy protein” (United States Patent No. 15/665,147, 2017)). We suspect the structure of lauridimonium hydroxypropyl hydrolysed keratin to be:

probable-structure

Fig. 8 Probable structure of lauridimonium hydroxypropyl hydrolysed keratin, where K represents the keratin polymer. It is very likely that more than one lauridimonium hydroxypropyl chain will be attached to the same keratin molecule.

Product B was bought in a high-street shop and tried on virgin hair (hair that has never been dyed). A silkier feel was noticed, but little repair on long, damaged ends – this is probably because the hair was not chemically damaged, but rather through time and mechanical stress. Lacking a double bond in the proposed structure above, product B cannot bind to hair through Thiol-ene Michael addition. The elucidated structure of product B is uncertain, however it is clear that it does not possess the same symmetrical structure which enables Product A to create bridges between keratin polymers. We postulate that product B does not reform analogues of the natural hair structure (as product A does) but that it aims to give the appearance of shine and strength through the well-documented use of cationic polymers as coating agents. (United States Patent No. 6/755,201, 2002)

PepG, the decapeptide produced by Cutiful, is significantly larger than the active ingredients in product A and B. It is a sequence of eleven amino acids, MCQCSCCKPYCS, which has primary structure:

Primary-structure

Fig.9 Primary structure of PepG, in yellow are sulphur containing residues– all but the first have the ability to form disulphide bonds.

PepG’s particularity is to be extremely cysteine-rich, and it is not risible to suggest that it behave as product A does in binding to the free thiols in keratin, and bridging two polymers thus. PepG, unlike product A, has free thiols group itself – we can expect PepG to re-form disulphide bridges that way. This means that the bonds will break if exposed to hair dyeing agents (use of peroxides and bleach to remove natural colour, for example). However, unlike product A (whose effects patents states to last at least two months, though advertising brands it as dissipating “in a few weeks to a month”), PepG should not need constant re-applying. It is probable that product A will wear off due to shampooing: the aqueous environment it will be subjected to, especially if one does not supplement the use of product A with the four other products the brand markets alongside it, will likely cause some of the ionic bonds to break. Without repeated use of the product (as, of course, recommended by the brand) the effects of product A will wear off over time: an issue we do not anticipate with Cutiful as it will truly re-form disulphide bridges, naturally present in the hair and proven to be stable under everyday-life conditions.

Product A contains a long organic chain of fourteen atoms, which comes to replace a single bond between two polymers. The size difference between the linkers is noticeable, and may introduce space for foreign compounds to occupy inside the hair’s structure. One might expect the size of the bridging agent to increase the water content of the hair, making it appear softer and shinier, and it may also make hair more permeable to natural oils, or colour during dyeing. However, this also implies that smell will be retained by hair more efficiently – especially the heavier, oilier ones. Increased porosity is a behaviour we anticipate to also encounter with the use of PepG. Whilst the length of the peptide will be reduced from what the Natta projection suggests, it will still create gaps in the fibre structure. The increased permeability to smell will be expected, however, to enhance another of Cutiful’s modules: fragrance production. Since the odorant concentration in the vicinity of the hair will be at its highest, a large proportion of the molecules being uptaken will be fragrance compounds: limonene or vanillin. We therefore expect hair treated with Cutiful to be able to act akin to a reservoir, helping to enhance the fragrance our E. coli secrete.

Scene 2: Summary

• Some products, like product B, act as coating agents, which do not reconnect the keratin chains. These show less effectiveness: though they help reform the natural oily coating of hair, they do not repair its ultrastructure.

• Some products, like product A, act as bridging agents. These chemically reconnect the keratin strands, but they introduce a new weakness in the form of the ionic COO- +NH3 bond, potentially unstable under aqueous conditions.

• Cutiful offers an alternative, PepG, which will reconnect keratin polymers together, using the same type of bonds which is naturally found in hair: disulphide bridges. This has the advantage over product A of potentially lasting longer when exposed to routine conditions.

• Bridging agents, like product A and PepG, may increase the permeability of the hair by introducing gaps in the structure. We expect these gaps to be occupied mainly by water and natural oils, which will give the hair a silkier finish and smoother feel.

• We also expect odorants to diffuse into the hair, thus increasing the intensity of the perceived smell in the headspace. In the case of Cutiful, we expect the main odorant uptaken to be our fragrance molecules: vanillin and limonene.

Act II: Experimental and Results

“I love to doubt as well as know” – Dante, “Inferno”, Canto XI 93: "non men che saver, dubbiar m'aggrata.”

As mentioned above, we are using PepG, a decapeptide that is capable of strengthening and repairing hair. To aid secretion and/or detection we added to our constructs both/either:

• SecB-dependent N-terminal secretion tag

• C-terminal His-tag.

Ultimately, we wanted to produce four different constructs:

1. PepG by itself (in diagrams, D – for Decapeptide)

2. PepG with an N-terminal secretion tag (in diagrams, ND)

3. PepG with a C-terminal His-tag (in diagrams, DH)

4. and PepG with both the secretion and His-tags (in diagrams, NDH).

By the end of the cloning phase, we had cloned these constructs into pBbE5k plasmid vector in E. coli DH5α. The pBbE5k plasmid vector was chosen because:

1. It has a ColE1 origin of replication, which has a copy number of 23-30, thus maximal protein expression. Moreover, the small size of our constructs will place less metabolic strain on the E. coli.

2. The vector contain a LacUV5 promoter, which is slightly leaky. But given the low probability of cytotoxicity of our protein and easy induction of the construct, this promoter was suitable.

3. This vector also has a kanamycin resistance gene, which will allow selection. We ensured that each of our sub-projects (i.e. fragrance, colour and repair) used different selection markers; in the hopes that all three could be integrated into one host.

overall-cloning-process

Fig.10 This figure shows our overall cloning process. To the left there is a schematic of the primer annealing step, where multiple oligonucleotides anneal to form our constructs. To the left is a schematic of the inverse PCR of pBbE5k-RFP, where the RFP is removed and replaced by BsaI sites, which are later cleaved in Type IIS ligation reaction.

Given the small size of the decapeptide constructs we decided that it would be more efficient to construct the genes ourselves using primer annealing rather than ordering the full constructs. This primer annealing process was based on a similar experiment carried out in yeast (strain: VL6-48N), where a full gene was constructed using overlapping oligonucleotides in vivo (Gibson, 2009). We ordered several overlapping primers that, when annealed, would make our four different constructs. The annealed constructs had 4 bp overhangs, which would allow easy ligation into the plasmid.

Once the primer annealing reaction had been carried out, the resulting products were run on a gel (Fig.11). For D and DH, we can see one distinct band, where most of the primers have correctly annealed to one another. For the larger constructs i.e. NDH and ND, we can see two distinct bands, where a proportion of the primers have not fully annealed to make the full construct. It is interesting to note that the smaller the gene construct, the more successful the annealing reaction.

annealed-primer-constructs

Fig.11 gel electrophoresis was carried out on the annealed primer constructs to see the efficiency of the primer annealing process, where N: OmpA; D: PepG; H: His Tag.

In parallel with the primer annealing process we carried out inverse PCR of the pBbE5k plasmid vector. The primers used were designed to simultaneously remove the RFP gene as well as integrate BsaI site into the plasmid.

PCR-products

Fig.12 PCR products of pBbE5k on a gel, after amplification removed RFP and integrated BsaI sites. The expected band size of ~3.5 kb has been observed.

Once we had both the annealed primers and the amplified vector, we carried out a variation of Type IIS cloning (i.e. a one pot assembly using BsaI and T4 DNA ligase according to the golden gate procedure). This was a ligation reaction where the BsaI enzyme cleaves the BsaI site in the PCR product to produce overhangs, which were complementary to the overhangs on all of the constructs. The overhangs of the construct and the overhangs of the vector then anneal and are ligated together by T4 DNA ligase.

After the ligation reaction, the constructs were transformed into E. coli DH5α and grown on kanamycin plates. We were fairly certain the ligation reaction had worked, given the number of colonies on the plate. In our positive control, the bacteria were transformed with pBbe5k-RFP, the original vector, to make sure that the bacteria were competent. For our negative control, the bacteria were transformed with the linear PCR products from the initial pBbE5k-RFP PCR amplification. After transformation, the negative control plate contained more colonies than the DH construct plate. We carried out a colony PCR (Fig. 13) to find out what happened…

we used the following primers:

1. ColPCR (upstream): ATGCGTCCGGCGTAGA

2. ColPCR (downstream): ATTACCGCCTTTGAGTGAGC

All the constructs came out positive for colony PCR i.e. we saw the expected band weight for all of the constructs. We also carried out colony PCR on the colonies on the negative control plate and found that the bands were the same as the positive control plate. This shows that the construct within the vector was RFP, which suggests that the DnpI digestion of the original PCR amplification was not efficient, thus the template plasmid (pBbE5k-RFP) was not removed and was transformed into the ‘negative’ control bacteria.

gel-colony-PCR

Fig.13 Gel from colony PCR, showing presence of desired constructs in all the chosen colonies We knew this given the fact that the bands are roughly ~450 bp (varies between constructs); had the ligation ont worked the linear vector was incapable of undergoing PCR with the chosen primers..

We then carried out a plasmid purification to extract the plasmid from the bacteria. A master plate was made during colony PCR and those colonies were used to inoculate LB. The resulting culture was then used to make purified plasmids, using the Qiagen Plasmid prep kit.

We also carried out a restriction digestion to see if our construct was really in the plasmid. We found two restriction enzymes that were unique cutters in our full construct. We found that NcoI cut our construct in the kanamycin resistance gene, whereas HindIII cut our construct in the stop codon. So, if our plasmid did contain the gene of interest, we would see two bands where each restriction enzyme has cleaved the plasmid once. This is exactly what we saw; each construct was cleaved twice to produce two fragments of DNA. The negative control has another band ~10kd, this is the nicked conformation of the plasmid so appears to be heavier only because it travels through the gel slower.

restriction-digestion

Fig.14 Gel showing the products of restriction digestion

Once we were confident that our plasmid contained our decapeptide constructs we sent the plasmids for sequencing to Eurofins. Along with our plasmid we also sent the primers we used for colony PCR, as we knew these bound to the vector and also amplify the region within which the construct was. All of our samples had no mutations in the gene of interest; however, the DH construct sequencing results were not quite what we expected.

Fig.15 Sequencing results for the Decapeptide alone construct. This figure shows the full alignment (right) as well as the description of the alignment in terms of number of mutations, deletions etc

Fig. 16 Sequencing results for the N terminal secretion and His-tagged decapeptide construct. This figure shows the full alignment (right) as well as the description of the alignment in terms of number of mutations, deletions etc

Fig. 17 Sequencing results for the N terminal secretion tagged decapeptide construct. This figure shows the full alignment (right) as well as the description of the alignment in terms of number of mutations, deletions etc

There do seem to be ‘mutations’ at the beginning and at the end of each of the above constructs, but we do not consider these as mutations because of the drawbacks of matrix-based fluorescence sequencing. Fluorescence based Sanger sequencing relies on DNA fragments that have a fluorescence probe being run on a gel/matrix. The fragments separate depending on size, depending on where the integration of the dideoxy ribonucleic acid had terminated DNA replication. The matrix is optimised to separate DNA fragments of a particular weight; so, the smaller and larger bands, which correspond to the beginning and the end of the sequence, are not properly resolved, resulting in incorrect reads.

Fig. 18 Sequencing results for the His-tagged decapeptide construct. This figure shows the full alignment (right) as well as the description of the alignment in terms of number of mutations, deletions etc

Our DH construct gave us some trouble as at first it seemed that the construct was wildly mutated. However, further analysis of the sequencing results seemed to suggest another possibility…

We found that our DH sample had been contaminated for the following reasons:

1. There were overlapping peaks on the sequencing result, suggesting that there are two fragments of DNA of the same size but with a different fluorescently probed Dideoxy ribonucleic acid; this is only possible if there are two different DNA templates in the sample. When the algorithm detects overlapping peaks it just chooses the stronger signal, leading to an unreliable read.

restriction-digestion

Fig.19 Overlapping peaks on DH sample sequence, suggesting two different DNA templates are present in the same sample.

2. We also predicted that our sample had been contaminated with the D construct as the overlapping peaks only occurred at the His-tag gene, which also happened to be where the DH and D constructs diverge.

3. At each base pair the overlapping peaks were taken into account to generate an ambiguous DNA base, for example if there were two overlapping peaks, one showing A and the other T, then W was assigned to that base pair. This ambiguous DNA sequence aligned with the original DH sequence, showing that the DH construct was present.

restriction-digestion

Fig.20 Comparison of the Ambiguous sequence with the expected DH sequence, with nucleotide code table from IUPAC as reference, showing the presence of our desired DH construct.

Act III: Expression

“I am not even going to try” – Mujtaba Ansari, “The Tragedy of what is his life”, Act II, Scene 2.

Once we were confident that we had the correct sequence we transformed the constructs into BL21(DE3) E. coli. The colonies on the plate were grown up in a preculture and then a working culture, which was then induced using 1 mM IPTG at 0D600 = 0.6. From the working culture we prepared cell media, whole cell extract (by using boiling lysis), crude cell extract (cell lysate from sonication) and cell debris (pellet from centrifugation after sonication). All these samples were prepared, though only one of them would contain our decapeptides: PepG is novel and we were not sure where it would be found. As some of our constructs had N-terminal secretion tags, our decapeptide could have been, for example, both secreted into the media and found inside inclusion bodies.

We prepared our sample by incubating our working cultures with 1 mM IPTG overnight and then pelleting the culture the following day. The pellets were then resuspended and sonicated at 40% power for 2 minutes. The sonicated samples were then centrifuged to separate the soluble and insoluble fractions. At first, we thought that we could concentrate our samples by using spin columns, but because our decapeptide actually goes through the spin column, it is not concentrated. Therefore, we decided to use a vacufuge to concentrate our sample.

We then tried to carry out a dot blot, where hourly samples (from the working culture) are pipetted directly onto a PVDF membrane and allowed to bind. We then washed the membrane with antibodies to visualize the his-tagged proteins. Unfortunately, the proteins could not be seen on the membrane, neither could the positive control. This showed that there was a problem with our methodology, which did not allow us to see the proteins. We decided that it would be better to separate the proteins on an SDS gel and then try to visualise because given the size of our peptide, larger proteins may interfere with any visualisation method.

We prepared a peptide specific SDS loading buffer (200 mM Tris-HCl, pH 6.8, 40% glycerol, 2% SDS and 0.04% Coomassie Blue G-250 - adapted from Bio-Rad Tricine Sample Buffer for Protein Gels, 30 ml #1610739 ). Our preliminary tests showed that for some reason the Coomassie blue stain was interacting with the peptides, causing peculiar bands and potentially disrupting the interaction of the antibodies with the peptide during the western blot. We adapted this loading buffer and made another batch, which contained no dye. We also used different kinds of gels; we first used 15% tris-glycine gels and thereafter started using 12% tris-glycine gels. We found that there was good separation of proteins with both of these gels when run at 150V for 25 minutes.

BL21(DE3) E. coli were cultured again and then we prepared crude cell extract, cell pellet, whole cell extract and cell media samples. Our samples were then mixed with equal volumes of SDS loading buffer and then boiled for 15 minutes. Then the samples were loaded onto a 12% gel and run at 150V for 45 minutes, rather than the usual 300V which we found to not separate the proteins and cause the gel to ‘smile’.

Fig.21 Showing the resulting SDS gel (left) and western blot (right), where no evidence of decapeptide can be seen; there is evidence of the positive control, so the western blot has worked..

Unfortunately, we could not find any evidence that our decapeptide has been expressed. After some reflection with the supervisors we found that the volumes and dilutions we had been doing were too great, resulting in a lower concentration of all proteins. We redesigned our protocols to make ensure that any potential his-tagged protein is as concentrated as possible. We achieved this by decreasing the level of dilution in our samples; before we were growing 500 ml cultures and only using 50 ml-worth of culture, in our improved protocol we were growing 100 ml cultures and using all of it. From our renewed protocol we did find some evidence that the largest construct i.e. NDH was being expressed, but this was inconclusive as we were unable to not repeat these results.

Fig.22 Showing the resulting SDS gel (right), which shows little over-expression of the decapeptide, and western blot (left), which shows inconclusive evidence of the expression of PepG.

As mentioned by Dr A. Cavaco-Paulo, who was a co-author on the paper we based our decapeptide on, the machinery in E. coli is not able to express such small proteins in isolation. Further research showed that for small peptides to be expressed in E. coli, the peptide needs to be part of a larger construct. For example, (Chen et al., 2016) expressed atrial natriuretic peptide by making a construct with multiple ANPs in tandem with lysine residues in between, which were then cleaved apart by Endoproteinase Lys-C. We also considered other ways to make our DNA constructs better; as we could not determine whether the E. coli not being able to express pPeptides was a transcriptional problem or a translational one, we decided to make multiple constructs that would take into consideration both problems.

[sfGFP fusion protein to decapeptide gene]. This construct is of the decapeptide genes (NDH) fused to sfGFP. The idea being that the original plasmid vector was expressing RFP, if we were to use this construct, we would easily be able to see if the construct has been expressed if the colonies fluoresce green. In-between the sfGFP and the decapeptide construct there is TEV protease site that will allow easy cleavage of the decapeptide away from the sfGFP; there is also another TEV protease site in-between the His-tag and the decapeptide gene that will remove the His-tag, without needing to add another amino acid. Moreover, The His-tag will allow purification of this protein, as well as detection (e.g. through western blot)

[Maltose binding protein fusion to decapeptide gene]. This construct uses the same idea as the construct as above i.e. the decapeptide gene is fused to a larger protein, but this time the decapeptide gene is fused to maltose binding protein (MBP). This protein means that we do not need the His-tag proteins anymore and can purify the decapeptide using the MBP. Moreover, it is also possible to carry out a western blot using the MBP as the antibody target.

[Poly-RBS construct]. This is another approach to expressing peptides in bacteria. This approach assumes that E. coli can express/translate small peptides. In this construct there are multiple RBS sites that will allow multiple ribosomes to bind to the 1kb long mRNA molecule, producing a polysome. This should also increase the amount of peptide that is produced. We have taken into account the size of the ribosome (~21nm) and made sure that there is enough space between each RBS to allow multiple ribosomes to bind. We have also made the mRNA molecule 1kb long, but this can be extended much more.

Act IV: Epilogue

“Finally, what is reason?” – Samuel T. Coleridge, “Reason”

From the time we spent in the labs we found that cloning small peptides into a plasmid vector was surprisingly easy. We found that we can easily make small constructs of DNA ourselves using primer annealing. We also found that Type IIS cloning can work for the ligation of small DNA construct into a larger plasmid vector.

Critically, we found that in our hands, E. coli cannot were unable to express such small peptide molecules and to express a gene that small it must be fused to a larger construct. We were unable to determine as to why such small proteins are not expressed; whether there is a transcriptional limitation or a translational one in E. coli. Nevertheless, the expression of small peptides requires the peptide to be fused to a larger, well-expressed protein and then later cleaved off (if one prefers the peptide to be by itself).

We hope that our work on decapeptides will be a foundation for other iGEM teams to expand upon in the future, and that this log of our experiments will guide them in the construct design and modification of standard laboratory techniques for measuring protein expression.