Analytical Review of Integument – Research project update November 2021

Please refer to the glossary in the last page of this document, an appendix with definitions of terms we use here.

Why are we working on this?

Simply put, we believe that every person should have the right to live life in the body that they genuinely want, even if it substantially deviates from the standard human form.

This project, a review of integument, is about skin-level modifications like fur, scales, and feathers. We are working to answer this question: What do we need to add to human cells to allow these structures to grow? Much work has been done on gene therapy delivery mechanisms already, so our job is to identify viable targets we can use.

In this update, we are building on material discussed previously in Update #1 (June 2021). As before, for the sake of transparency, we are balancing technical language while staying as accessible as we can! Please feel free to reach out and ask questions as they arise.

Fur and hair across species

Over the past few months, we have been focused on a few diverse topics in fur development – these include the genetics of hypertrichosis (colloquially called “werewolf syndrome”), mechanisms determining hair strand morphology and heterogeneity, and mechanisms where new hair follicles can form in adults. The topic of fur is vast, so these do not represent the full extent of our research, and are instead just the areas we have made the most progress within that we want to update you on!

Hypertrichosis is a medical condition where hair grows unusually thick across a wide area of the body, and it has both genetic and environmental causes. The genetic form, in which extreme hair growth occurs body-wide, is called congenital generalized hypertrichosis (CGH) – “werewolf syndrome”. What if we could draw biotechnological inspiration from it? While most cases of CGH are not characterized genetically, a small subset have been, and we have focused on these with this question in mind.

In several case reports, CGH was caused by deletions or microduplications of a genomic region spanning the shared genes ABCA6, ABCA10, ABCA5, and MAP2K6 (Sun et al., 2009). On the surface, it is surprising that a deletion, which causes genes to be haploinsufficient (having reduced function), and a microduplication, which adds an extra copy and should have the opposite effect, would have the same clinical outcome. However, in clinical genetics, this is not unheard of – often, microduplications change the structure of chromatin (DNA) and may disrupt 3D regulatory elements spatially that control how it’s organized and controlled, having the same effect as deletion. It would have been helpful for the authors to perform RNA expression analysis on these patients in parallel to determine if this was the case, but they didn’t do so. While it may be possible to request samples from the physicians that treated these patients to assess it ourselves, easier followup experiments exist as described later in this update. 

After becoming aware of these 4 shared genes disrupted in CGH, we wanted to know, which one is actually responsible for causing hypertrichosis when disrupted? ABCA6/10/5 are transporters, not signalling proteins, making them unlikely culprits and leaving us with the signaller MAP2K6. Since it is the only signaller in the list, we felt MAP2K6 was the most likely causative gene candidate, and this is supported by the observation that MAP2K6 downregulation causes a hypertrichosis phenotype in foxes (Clark et al., 2016). However, another paper has shown small mutations in ABCA5 to be causative of hypertrichosis in two other human patients, possibly via its effect on lysosomal function by transporting cholesterol (DeStefano et al., 2014). ABCA5 is otherwise poorly characterized, making signalling analysis difficult. At any rate, that two genes out of this list have been shown to cause hypertrichosis implies that the entire four-gene cassette is involved in hair growth, explaining why these genes are beside each other in the genome. 

We have also started to investigate X-linked CGH, where the causative genes are present on the X chromosome. As with all X-linked conditions, it usually affects genetic males when recessive, since there is only one copy. In our research, we identified a study implicating insertions that disrupt a palindrome sequence near the SOX3 gene, probably altering SOX3 transcriptional regulation (Zhu et al., 2011). SOX3 is a transcription factor which has been previously associated with hair follicle development; disrupting it alters androgen signalling to prevent follicle development (Hong et al., 2018). We hypothesize that the palindrome disrupted in hypertrichosis patients is a regulatory element of SOX3, and that disrupting it causes SOX3 de-repression, leading to aberrant high expression that drives hair growth. 

X-linked CGH may also occur independently of androgen signalling, however. We identified a case report of a patient with extensive hair growth across the entire body that resulted from a microduplication affecting FGF13 (De Stefano et al., 2013). In this patient, FGF13 expression in hair follicles was dramatically reduced. FGF13 is poorly characterized in hair follicles, but another study described it as being highly expressed in the follicle bulge region (comprising epidermal stem cells) and areas derived from those cells, possibly being involved in hair follicle morphogenesis at the bulge (Kawano et al., 2004). 

Having already identified a number of generalized hypertrichosis-causing genes including ABCA6/10/5, MAP2K6, SOX3, and FGF13, a list that will likely expand with more research, we propose to test the effect of modulating each of them in artificial hair follicles reconstituted from stem cells in vitro. They may provide useful tools in controlling follicle morphology through modification of transplantable follicle stem cells. 

We have also considered the effect of androgens on hypertrichosis more directly, in the form of hirsutism, a condition in which genetically female patients develop body hair in a male-like pattern (Hafsa & Badri, 2021). Hirsutism is generally caused by hormonal dysfunction, but can also be caused by genetic factors that change how cells process and react to hormones, particularly androgens (testosterone and dihydrotestosterone). Since androgen signalling in and of itself is sufficient for terminalization of vellus hair to become fully matured, pigmented terminal hair, it may be possible to artificially activate the androgen pathway in vellus follicles without androgens. One way to do this would be to express gain-of-function mutated androgen receptors that are constitutively active, as has been previously demonstrated (Kallio et al., 2018). However, in this approach, hair follicle stem cells must be modified in vitro before transplantation to ensure no health risks from off-target transduction of overactive androgen receptors. 

As stated in our previous update, though, the vellus hairs present in humans are not dense enough to provide a full coat of fur, and more hair follicles must also be added. It was discussed there that it is possible with present technology to produce and culture stem cells (dermal papilla cells) capable of producing hair follicles when injected under the skin (Lee et al., 2020). Current technology cannot yet control hair texture, however, and so we began evaluating targets for morphological regulation of hair follicle growth and development. One of the central regulators of cyclical hair growth and morphology determination is Wnt and BMP signalling, where Wnt signalling is more active during growth and BMP signalling is more active during shedding (Daszczuk et al., 2020). Excitingly, we found that one particular Wnt protein, Wnt10b, can be used to regulate hair follicle size in mice, where overexpressing it causes follicles to double in size, and the opposite effect from inhibition is also true (Lei et al., 2014). 

Regulation of Wnt10b, which operates within canonical Wnt signalling (the other subtype being noncanonical), may be enough to create thin hairs that can function as a soft undercoat for generating realistic fur in humans, alongside the thicker primary hair follicles that produce guide hairs (topcoat). However, we were also interested in studying how animals naturally regulate hair follicle size in primary versus secondary hair follicles. To assess this, we turned to an RNA sequencing study looking at genes where the expression changes between these two groups of follicles in goats (Zhu et al., 2013). The study first isolated dermal papilla stem cells (DPCs) from each group of follicles and then performed RNA sequencing on them after establishing in culture. Numerous potential regulators were identified – namely, Wnt5a and Wnt5b are exclusively expressed in primary DPCs and not secondary DPCs, but this is actually surprising because Wnt5a/b are involved in noncanonical signalling, whereas Wnt10b is canonical. Also, canonical Wnt signalling is not observed at all in the primary DPCs, which is impossible because canonical Wnt signalling is well established in these cells from numerous other studies we have cited. That said, Zhu et al. (2013) do not report how long they culture the DPCs before analysis, and it’s known that DPCs lose their hair-inducing properties after a few weeks of culture. If this is the case, then the data from that study must be taken with a grain of salt. All things considered, Wnt10b remains our primary candidate for hair follicle width regulation.

It’s also important for us to be able to control how long hair grows! Fur is longer than human body hair, but shorter than scalp hair, and also varies by species. In our research, we found that the best characterized regulator of hair length in humans in FGF5, which binds its receptor FGFR1 in hair follicle DP to stimulate catagen entry and the end of growth/beginning of shedding (Higgins et al., 2014). However, while disabling mutations of FGF5 cause aberrantly long hair length, they do not cause homogenous hair length across the whole body, leading us to believe that there are other factors also involved via another pathway. That said, modulating FGF signalling in DPCs may be sufficient for our purposes in ensuring that hair grows to the desired length. We are continuing our search for the other operative pathways, however, for more fine-tuned control.

As a final area we will cover in this update, we have also been evaluating wound-induced hair follicle neogenesis (WIHN), the process in which new hair follicles are produced in adults after existing ones are destroyed when the skin is wounded. Furred animals generally have robust mechanisms for WIHN induction, where new follicle-inducing DPCs are generated from local fibroblasts in the dermis. We hypothesize that if there is some wound-dependent signalling cascade inducing neogenesis, we might be able to recreate it to induce new hair follicles to grow as the first step of giving fur to patients. Furthermore, while WIHN does not exist in humans for the most part, it does exist in close evolutionary cousins, and so it is likely that the requisite signalling networks have been retained.

Interestingly, while common wisdom in the field is that WIHN is exclusively found in furred animals only, we have identified two papers from the 1950’s that show a special case where it may occur in human patients (Kligman & Strauss, 1956; Kligman, 1959). It occurs after dermabrasion of the face, where the top layer of skin is frozen with a volatile refrigerant and removed down to halfway through the dermis. This destroys vellus hair follicles completely, yet they are able to grow back after the skin heals, implying an active neogenic mechanism. This effect has only been shown for vellus hair, though, and terminal hair follicle neogenesis has never been observed in humans under any circumstances. Still, if we can understand how vellus follicle neogenesis is induced, we may be able to use it to increase the density of vellus follicles and then use a subsequent step to terminalize these follicles, thus achieving sufficient density for full fur coverage. 

It’s also worth noting that a couple of studies have suggested that terminal hair follicle neogenesis may also occur in special cases in some adult skin tumours, implying that these tumours somehow activate a developmental pathway sufficient to produce new follicles (Muller, 1971). This is arguably more exciting; if we came to understand the pathways involved in these tumours causing neogenesis, we may be able to induce fur growth in adults in a single step, rather than requiring subsequent terminalization of neogenic vellus hairs. Still, this is currently viewed as a secondary option, as it makes generating heterogeneous fur (topcoat and undercoat) much more difficult in vivo; our top approach remains to produce modified hair follicle stem cells for injection.

Scale development in reptiles and avians

So far, reptilian scales have proven to be more difficult to find research on than the other integument types. It is therefore difficult to find a specific gene circuit or mechanism that occurs within reptiles to produce scales. To overcome this limitation in current research, we have decided to also focus on avian scales, such as the scales on the feet of chickens. Although this might sound unrelated to reptilian scales, the general structure and function of these scales are still similar enough for us to consider, and there is much more scientific research available on chicken integument. 

Interestingly enough, this also allows us to compare the development of scales to the development of feathers, which form through diverging paths within the same developmental pathway within chickens. By comparing the two structures and their differences, we can reveal the specific requirements for scale growth compared to that of other structures. 

Scales, like most structures, probably form from a Turing pattern – a reaction-diffusion mechanism in which fields of morphogens (structure-determining molecules) form wavelike patterns that govern development. However, a gene circuit/reaction-diffusion mechanism has not yet been identified for scales. We have, however, identified numerous reporters and signalers, most of which are also involved in other integument types. For example, β-catenin (and thus canonical Wnt signalling) is involved in determining whether scales or feathers will grow in a particular location on chicken skin (Wu et al., 2018). β-catenin also demarcates the locations of both scale and feather development (Cooper et al., 2019). Chromatin structure surrounding keratin genes, some of which are expressed in scales and others being specific to feathers, is also important for regional specification of whether a feather or scale will grow, although the upstream signallers are not well-characterized (Liang et al., 2020). 

We have noted that in feather development, where the reaction-diffusion system controlling development is at least partially known, that the initial stage involves formation of a placode that resembles scales (Li et al., 2017). Since the initiating factors are the same, it stands to reason that the reaction-diffusion system controlling scale development may be the same as the one controlling that of feathers, with modifications in reactant amounts, kinetics, etc., and possibly with feather development involving extra network components. We will continue our analysis to see if these components together form the reaction-diffusion network we seek.

We may also be able to find some answers within an adjacent but unrelated feature of scales: their coloration patterns. Within some types of lizard, the color of each scale becomes uniformly colored, while the scales themselves form a mosaic of colors in an intricate pattern. Researchers have identified a reaction-diffusion network underlying this interesting coloration, and this gives us some important insight into the diffusive properties within and between scales (Fofonjka & Milinkovitch, 2021). Namely, diffusion between scales appears to be slower than diffusion within scales, shown by the uniform coloration of each individual scale. This might hint at an important role of geometry within the formation of patterns, even for the initial structure.

Feather production in avians

Feathers are perhaps the most complex structures in integument, and each aspect of their size and shape is controlled by a combination of genetic and epigenetic factors tuning the precise properties of the keratins, as are aspects of support and anchoring, growth, longevity, symmetry, pigments, iridescent structural motifs, stiffness, layout relative to one another, density per area of skin, and the presence of waterproofing oils. These factors are utilised in a manner sensitive to the location of the cells with respect to the body and their neighbouring cells, and in particular with respect to hormone gradients forming reaction-diffusion mechanisms.

Li et al. (2017) recently described a clear model of the developmental biology of single feathers, from the stage of an individual feather follicle to the expansion into a fully grown feather (figure 1, at the beginning of this document). This model is particularly exciting, because it demonstrates the ability to switch between three types of feathers by switching on or off the respective reaction-diffusion networks that independently regulate the feather’s structural properties. 

In the model, there are essentially 3 major steps: 1) skin placode formation where a feather will form, 2) anterior-posterior patterning to flatten and extend into a feather, and 3) medial-lateral patterning to determine asymmetry depending on feather type. Skin placode formation is driven by interplay between an unidentified activator morphogen and BMP signalling. After formation of the placode, which we have noted resembles an early-in-development scale, anterior-posterior patterning begins, being driven by signallers in the pulp juxtaposed to the tubular growing feather. Here, pulp GDF10 demarcates the rachis (the central shaft of the feather) on the anterior side, Wnt signalling surrounds it, and GREM1 demarcates the opposing side of the pulp near the BGZ where the feather will ultimately separate and unfold at the end of development. After this pattern is established, asymmetry is determined in the final step. Here, CRABP1 and RA signalling are active in the pulp on the side where more growth occurs, and CYP26B1 is present on the opposite side. CYP26B1 degrades RA, thus reinforcing asymmetric RA activity. 

All of these components together, including the unidentified activator in the placode, may constitute the minimum requisite network for generating feathers, and transferring them into hair follicle cells may transform them to feather follicles. We are performing reaction-diffusion network simulations to determine necessity and sufficiency of the reaction components and to evaluate whether additional components, even beyond the unidentified activator, are required. Finally, we note that further experiments to identify this activator are necessary, likely involving proteomics/transcriptomics studies of developing feather placodes at different stages of development. 

We also identified another study describing the requisite feather keratin genes in chickens, which exist in a set of looped chromatin that is structured to express specifically in feather-producing skin regions (Liang et al., 2020). Humans do not possess beta-keratins, which are structurally required for both scale and feather development, and these genes will thus need to be transferred into human cells to be able to generate either structure with high fidelity. There are about two dozen of these genes present, and whether they are all required or exist as many copies simply to increase protein expression is not known. We know that some of the genes involved in arranging and determining the shapes of feathers have other jobs in other parts of the body and stages of development. As such, we plan for the first attempt at making a feather-producing gene network in humans to include a subset of these genes, which we will then expand to the full list if required. 

As noted in our last update, feather production in humans is among the most difficult of the integumentary changes because feathers are so evolutionarily distant, and will therefore take longer to achieve. As seen here, they require complex reaction-diffusion mechanisms, along with extra genes such as beta-keratins, and these must be modelled and validated in human cells before being clinically applied. Our goal is to identify the minimum required unit of signalling and structural proteins for making feathers, and to propose methods of introduction to human cells, pending experimental validation.

Where do we go from here?

Conversion between fur, feathers, scales, and other integument types in vivo will require the development of processes to not only handle growth, but also to ensure that the mechanisms controlling any one type will not interfere with another type. For example, hair follicles are inhibitory to feather development in mouse/chicken hybrid embryos (Moscona & Moscona, 1965). At ratios of more than 1 hair follicle to 24 feather follicles, feather development is completely inhibited. This indicates that an existing integument structure may need to be removed prior to establishing a new type, unless the inhibitory mechanism is known and can itself be counter-inhibited.

A second challenge will be to insert and activate the genes of the desired integument type. There are several ways to approach this, such as local stem cell injection or gene alteration, each of which provide their own set of challenges. Since gene networks tend to be highly connected, we must pay attention that changing genes related to integument will not have secondary ‘moonlighting’ effects in other functions of the cell, which might lead to malfunctioning cells and other nasty effects.

For a full conversion of integument type, we need to engineer the developmental processes required to handle the following steps:

  • Local deactivation of genes for the undesired integument type (epidermal and maybe some dermal), especially those which suppress or clash with those for the desired integument type.
  • The alteration of genes to include the desired integument type and deactivate or if necessary, remove the undesired integument type genes. If those genes ‘moonlight’ in other jobs elsewhere, then we must attempt to decouple these genes first. There are different methods to accomplish this, such as selective cell modification, decoupling through non-crosstalking circuit design, or otherwise.
  • The scaling-up of reaction-diffusion patterns to be deployable on an adult body. For inspiration on how to achieve this, look at the largest-embryo-stage creatures with the desired integument type and extrapolate.
  • The suppression of any genes preventing a reversion to an embryo-like growth state, but only to some degree, and timed to coincide with the activation of safeguards either inbuilt or introduced, to prevent carcinogenic growth.
  • The activation of embryo-like growth in a manner similar to that seen with salamander limbs, but localised to the hair follicles, feather buds, or scale placodes.

Together, this outline represents the activation of the genes for the new integument type to grow the follicles out in a freshly deployed pattern. To address all points, we require a comprehensive map of the biochemical and physical interactions between groups of cells during the development and maintenance of the integument.

In our next steps, we will aim to create the skeleton of such a map, filling in the gaps in our knowledge to a point of being able to create usable biotechnology. Now that we are beginning to get a clear picture of the advances in the field, our next steps will also involve identifying the gaps in our collective knowledge, guiding future research to fill these gaps. Finally, we might be able to propose possible solution methods to obtain the desired conversion between integumentary types!

PS: As always, if there are STEM professionals out there who read this and are interested in working with us, please do not hesitate to reach out! 


The Project Team
Analytical Review of Integument
Freedom of Form Foundation
501(c)(3) EIN 82-4415111


Term Definition
Upregulation When a gene is signalled to be read more often, making more of its product and intensifying its function
Downregulation When a gene is read less often, making less of its product
Microduplication Small piece of a chromosome that is incorrectly duplicated
Deletion Removal of a segment of DNA incorrectly, usually destroying the gene it occurs within
Chromatin The normal state of DNA in a non-dividing cell’s nucleus
Transcription Factor Protein that binds to DNA and helps to promote specific genes to be upregulated by being read more often
Palindrome DNA sequence that is the same no matter which direction you read it in; they are often regulatory elements of genes
Canonical Wnt signalling Canonical refers to this being the first-discovered type of Wnt signalling and thus usually the pathway we know the most about
Noncanonical Wnt signalling Noncanonical contrasts this, equating to newer less well-characterized pathways
Neogenesis Hair follicle neogenesis is the production of new hair follicles from scratch