The

The Veterinary Journal 168 (2004) 81–86

Veterinary Journalwww.elsevier.com/locate/tvjl

A high yield method for growing primary canine keratinocytes

Adrian Watson *, Claire Baker, Julie Bailey, Tim Fray, Peter Markwell

Waltham Centre for Pet Nutrition, Freeby Lane, Waltham-on-the-Wolds, Leicestershire LE14 4RT, UK

Accepted 3 June 2003

Abstract

From a small amount of starting material, a large quantity of canine keratinocytes can be generated for experimental purposes

using a refined method of explant culture to initiate the growth of basal cells with a high proliferative potential. The dividing ca-

pacity of cultures was promoted by a system selecting clonogenic cells onto an i3T3 feeder layer in combination with carefully

monitoring cell morphology and passaging to select out excessive numbers of differentiated keratinocytes. Levels of contaminating

dermal fibroblasts, which if left unchecked will overgrow keratinocytes, were kept to a minimum by a combination of careful explant

micro-dissection to remove dermis, eliminating explants with signs of fibroblast growth as well as using cholera toxin, EGF and i3T3

feeder layers. The advantage of the method described is that it does not rely on the provision of large quantities of starting material

thereby reducing the need for repeated tissue sampling, and passage numbers of five or six can be routinely achieved. This technique

can therefore be useful to experimenters who require a regular and reliable source of cells for their studies.

� 2003 Elsevier Ltd. All rights reserved.

Keywords: Culture; Explant; Protocol; Proliferation; Fibroblast

1. Introduction

The scope and detail of research conducted into the

biology of mammalian skin has grown enormously over

the last 20–30 years. In order to achieve this expansion,

a number of methods have been developed for obtaining

and culturing keratinocytes from the primary skin tissueof a variety of species (Fusenig and Worst, 1975; Hol-

brook and Hennings, 1983; Wilkinson et al., 1987;

Tenchini et al., 1992). The available methods can be

broadly divided into two groups. Firstly, those in which

the skin is enzymatically digested, typically using pro-

teolytic enzymes such as dispase and/or trypsin, in order

to selectively detach viable keratinocytes (Tenchini

et al., 1992). The extracted cells are subsequently cul-tured as an adherent single cell population either on

plastic alone or plastic plus an extracellular matrix

substratum (Woodley et al., 1988, 1990). Alternatively,

the cells are co-cultured with a feeder layer of murine

fibroblasts (Rheinwald and Green, 1975).

* Corresponding author. Tel.: +44-1664-415655; fax: +44-1664-

415440.

E-mail address: adrian.watson@eu.effem.com (A. Watson).

1090-0233/$ - see front matter � 2003 Elsevier Ltd. All rights reserved.

doi:10.1016/S1090-0233(03)00117-5

A second category of protocols exists whereby whole

skin is micro-dissected and keratinocytes grown out

from the explanted epidermal tissue fragments (Jepsen

et al., 1980). Both approaches have demonstrated cred-

itable levels of success but are also equally limited by

access to adequate amounts of good quality tissue. En-

zymatic digestion, in particular, requires relatively largepieces of skin (typically 2–4 cm2) in order to get the

necessary cell numbers for an effective seeding density

(Tenchini et al., 1992). Refinements of existing methods

can therefore be valuable for improving the efficacy of

primary culture with the aim of obtaining the maximum

number of cells from the minimum amount of starting

material (Tenchini et al., 1992; Leigh and Watt, 1994;

Daniels et al., 1996).Canine skin biology has attracted considerable atten-

tion from dermatologists due in no small measure to the

apparent predisposition of the dog to dermatological

problems. Coupled with this is the fact that many of these

conditions have human equivalents (Scott et al., 1980;

Muller et al., 1983). Owing to the prevalence of immune-

mediated skin diseases (Scheidt, 1988; Hillier andGriffen,

2001), much canine dermatological research has beenconcerned with immunological (mal/dys)function and

82 A. Watson et al. / The Veterinary Journal 168 (2004) 81–86

the skin associated lymphoid tissue (SALT). The effecthas been that, at least in part, the role of the keratinocyte

in canine skin health and disease has been under-

emphasised and consequently progress in developing in

vitro methods has been similarly neglected.

We describe here a refined method by which a high

yield of canine keratinocytes can be obtained from a

small starting quantity of normal canine skin tissue. We

also demonstrate that the cells generated by this methodcan be maintained for multiple passages in vitro. The

method illustrates the ability, through refinement of

current protocols, to maximise the experimental yield

from rare and valuable research material.

2. Materials and methods

2.1. Ethical standards statement

All animals were fed commercially available, com-

plete diets, throughout the study period at energy levels

to maintain adult body weight. The animals were

housed at the Waltham Centre for Pet Nutrition

(Leicestershire, UK) where they were housed in pur-

pose-built, environmentally enriched facilities (Love-ridge, 1994) and treated in accordance with the Centre�sresearch ethics and UK Home Office Regulations. The

dog breeds used for these studies were Labrador re-

trievers and Beagles. Both male and female dogs were

used between the ages of two and 10 years.

2.2. Materials

All materials for tissue culture were obtained from

Invitrogen with the exception of adenine, insulin,

hydrocortisone, EGF, cholera toxin and EDTA (Sigma–

Aldrich). All culture plastics were obtained fromCorning

Costar. Materials for TEM processing were from

Sigma–Aldrich.

2.3. Obtaining skin samples

Four millimetre biopsies were taken from the flank of

dogs according to the method described by Lloyd

(1985). Areas around the biopsy site were clipped to

remove hair prior to the procedure. Lignocaine was used

as the local anaesthetic on all occasions (0.5 mL Lignol,

Arnolds). Once removed, skin was immediately placed

into a universal bottle containing ice-cold Green�s media(10 mL) (Green, 1978).

2.4. Primary culture of canine keratinocytes from explant

The skin samples were dissected immediately under a

binocular microscope while submerged in cold phos-

phate buffered saline (PBS). Subcutaneous fat and as

much dermis as possible were removed without dam-aging the epidermis. Any remaining dermis was then

carefully trimmed away. The remaining epidermal tissue

was washed in PBS containing antibiotic mix (2.5 lg/mL

amphotericin, 100 U/mL penicillin, and 0.1 lg/mL

streptomycin) and trimmed into pieces of 0.5–1 mm2.

Tissue fragments were then placed in ice-cold Green�smedium consisting of DMEM/HAMS F12 mix, 30 mM

NaHCO3, 2 mM glutamine, 10% fetal bovine serum, 2.5lg/mL amphotericin, 100 U/mL penicillin, 0.1 l/mL

streptomycin, 25 lg/mL adenine, 5 lg/mL insulin, 0.4

ng/mL hydrocortisone, 10 ng/mL epidermal growth

factor (EGF) and 8 ng/mL cholera toxin (Green, 1978).

Each individual 1 mm2 piece of tissue was placed into a

separate well of a 24-well plate and allowed to adhere to

the plastic for approximately 5 min. Green�s media was

slowly added to each well (0.5 mL) and the explantsincubated at 37 �C/5% CO2. Cell outgrowth from the

explants was monitored daily.

2.5. Subculture of keratinocytes

Once outgrowth of keratinocytes was observed in

around one-third to one-half of all explants, the cells

were transplanted. Explants were lifted before therewere significant signs of cell differentiation as this was

found to make dispersal of individual keratinocytes

problematic. To move the explant grown cells, Green�smedia was first removed, the explant washed with 0.5

mL PBS. The PBS was removed and trypsin/versene

added (0.5 mL, 0.25% trypsin in versene, pH 7.3; 37 �Cfor 10 min). The cells and explant tissue were removed

by gentle agitation with a pipette-tip and the solutionadded to 5 mL of Green�s medium. Cells were then

sedimented at 700g for 3 min, resuspended in 1 mL

Green�s medium and plated onto a i3T3 (irradiated

mouse embryonic) fibroblast feeder layer (106 i3T3s per

T75 flask) plus 9 mL of Green�s medium. Approxi-

mately 15–20 successful explant cultures were trans-

ferred per T75 flask. The explant tissue itself was also

transferred to the feeder layer as this was found togenerate more colonies with no significant increase in

fibroblast contamination.

2.6. Maintaining keratinocytes in culture

Once the flask of primary cells had reached approx-

imately 80% confluency any remaining i3T3s were re-

moved using EDTA (5 mL of 0.02%, 5 min) and thekeratinocytes passaged with trypsin/versene (37 �C, 10min). Cells were seeded at 106 per T75 (approximately

1.5� l04/cm2) onto plastic alone and the media changed

after 24 h to remove dead cells. Subsequent media

changes were every 3–4 days. In cultures observed to

have greater than 20% larger proliferative cells (Staiano-

Coico et al., 1986) (Fig. 3B), a passage was performed

A. Watson et al. / The Veterinary Journal 168 (2004) 81–86 83

automatically, with the aim of selecting the more ad-herent basal cells into the next passage.

2.7. Morphological analysis

Live keratinocyte cell cultures were photographed

using an Olympus SC35 SLR Camera mounted on an

Olympus CK2 Inverted Phase Contrast microscope.

Cells for transmission microscopy were processed asfollows: cells were grown on polycarbonate membranes.

The membrane and cells were then washed twice in 0.1

M piperazine-N,N0-bis(2-ethanesulphonic acid) buffer,

pH 7.3, and then fixed in 3% glutaraldehyde/0.3% hy-

drogen peroxide at 4 �C for 1 h. Following secondary

fixation in 1% osmium tetroxide for 1 h at room tem-

perature, the cell monolayer was embedded in araldite

epoxy resin. Ultrathin sections (150 nm) were counter-stained using uranyl acetate and lead citrate and sub-

sequently viewed under a Philips CM100 transmission

electron microscope.

3. Results

3.1. Explant culture of skin tissue

Initial experiments resulted in significant numbers of

explants becoming contaminated with fibroblast out-

growth from the dermal component of the tissue; up to

50% of explants where outgrowth was observed. In order

to reduce the level of fibroblast contamination, tissue

samples were fine dissected to remove as much of the

dermal material as possible without damaging the epi-

Fig. 1. Light micrographs of dissected and explanted canine skin fragments

defined keratinocytes can be seen with few sign of differentiation. (B) Outgr

distinguish individual cells due to differentiation of cells in many areas prox

dermal component. Generally this resulted in removal of90% of the total tissues wet weight. Explants where there

was evidence of fibroblast growth were not transplanted

in order to reduce contamination of cultures. Explant

orientation of the remaining largely epidermal tissue was

not found to significantly influence the success of kerat-

inocyte outgrowth. Cell growth was seen on average for

40% of explants following seven days of culture

(Fig. 1A), based on 24 experiments (range 20–80%). Cellsspread out initially as a monolayer, although latterly

there was a tendency for layering to occur in proximity to

the explant, presumably due to differentiation and con-

comitant stratification. If the layering spread beyond an

area proximal to the explanted tissue, removal and

transferring of cells was less successful (Fig. 1B).

Therefore, explant grown cells were transferred to feeder

layer as soon as evidence of differentiation was seen. Itwas noticed that this approach also resulted in greater

proiliferative potential of keratinocytes.

3.2. Transplantation and growth of primary keratinocytes

Cells transferred onto an irradiated-3T3 fibroblast

feeder layer would normally start to form identifiable

colonies following 5–8 days in culture (Fig. 2A). Cells

from two 4 mm diameter biopsies resulted in the for-

mation of between 30 and 50 colonies of keratinocytes in

a 75 cm2 flask. The colonies would typically take a

further 8–12 days to reach near confluency (Fig. 2B).Very little differentiation was observed during this time,

as judged by morphological appearance (Fig. 3A) and

absence of involucrin immunostaining of cells (not

. (A) Outgrowth of cells from explant after seven days. Small, clearly

owth after seven days, however on this occasion it is more difficult to

imal to the explant (arrows). Scale bars 300 lm.

Fig. 2. Light micrographs showing growth of keratinocytes at first subculture following (A) (seven days) and (B) (18 days) of culture. Colonies

initially grow as focal group of cells that progressively displace surrounding i3T3s (A). Colonies continue to expand, eventually merging and dis-

placing virtually all i3T3s (B). Cells very rarely demonstrate any signs of differentiation at this stage. Scale bars 150 lm.

Fig. 3. Light micrographs of keratinocytes at passage 4. (A) Typical cobbled appearance of undifferentiated cells. (B) Culture with a combination of

small cells in addition to larger cells in the early stages of maturation (arrows). The larger cells have a limited proliferative potential and are selected

out through passage. Scale bars 50 lm.

84 A. Watson et al. / The Veterinary Journal 168 (2004) 81–86

shown). Seeding cells from explants on to plastic in the

absence of i3T3s was found to result in reduced colony

formation and a greater frequency of fibroblast out-

growth.

Subsequent passage of the cells on to further i3T3

feeder layers was not found to significantly improve

proliferation or deter differentiation of keratinocytes.Continued use of i3T3s was not found to influence

dermal fibroblast growth. Cells were therefore passaged

from this point on onto culture plastic alone. Contam-

ination with fibroblasts was seen infrequently in the

primary cultures, presumably due to the low level of

initial contamination as well as the incorporation of

cholera toxin and EGF into the media at all times.

Reducing the concentration of antibiotics at thisstage was found to result in unacceptable levels of

bacterial contamination. Population doubling times

for the keratinocytes were estimated at 4–6 days on

plastic.

The efficacy of passage was improved by conducting

this procedure at the appropriate time. Conventionally,

passage would be performed as the cells neared conflu-

ence. However, passaging the canine keratinocytes when

the proportion of partially differentiated cells (Fig. 3B)increased beyond 15% of the total served to eliminate up

to 90% of these cells from the subsequent passage,

thereby enriching the clonogenic cells. Commonly, in

this way, cells could be passaged up to six times. After

this stage the cells were liable to excessive differentiation

with a concomitant high level of cells loss.

Overgrowth with dermal fibroblasts was seen in less

than 15% of cultures using the protocol described.The total number of cells that could be generated

using this method was not definitively tested due to the

Fig. 4. Transmission electronmicrograph showing a monolayer of ke-

ratinocytes at passage 4 grown on polycarbonate filters. The image

shows large active nuclei surrounded by numerous mitochondria. Cell–

cell contacts can be clearly seen in the form of desmosomes (arrow-

heads). Scale bar 5 lm.

A. Watson et al. / The Veterinary Journal 168 (2004) 81–86 85

requirement of cells for experimental use. The primary

flask (plus i3T3s) was split when near confluent resulting

in around 2.5–3 million total cells. Subsequent passages

yielded on average 1.5 million cells per flask, diminish-

ing to 1 million by passages 5 and 6. Using the protocol

described it was generally possible to generate up to 80–

90 million cells in total from a starting point of two 4mm diameter skin samples.

3.3. Morphological characterisation of cultured primary

keratinocytes

The morphology of the cells under phase contrast mi-

croscopy demonstrated the classical cobbled hexagonal

appearance of undifferentiated keratinocytes (Fig. 3A).The larger, partially differentiated keratinocytes de-

scribed previously can be seen in Fig. 3B. Ultrastructural

observation of the keratinocyte monolayer clearly

showed discrete undifferentiated cells with undegraded

nucleus and a rich supply of intact perinuclear mito-

chondria (Fig. 4). Adjacent cells are connected via des-

mosomes formed between the numerous small processes

projecting from the cell membranes.

4. Discussion

Obtaining primary keratinocytes via proteolytic di-

gestion of skin tissue has one overriding drawback,

namely the necessity for sufficient tissue for physical

manipulation. Such a constraint can pose significantproblems if a regular and reliable source of cells for ex-

perimental work is required. Therefore, the development

and refinement of a method by which cells can be gener-

ated using smaller amounts of starting material is of

considerable value to the researcher. What we describe

here is a reliable method of obtaining a population of

undifferentiated canine keratinocytes from a relatively

small amount of starting material. The biopsies are of aminimal size (4 mm diameter) and can, by a series of fine-

dissection and sequential culture steps, be used to gener-

ate many millions of cells from multiple passages.

Two problems associated with previous skin explant

protocols have been addressed in the new protocol.

These are the high frequency of fibroblast overgrowth

and the tendency for keratinocytes to enter into differ-

entiation prematurely, thereby limiting their use (Leighand Watt, 1994). To address the former problem a

combination of meticulous removal of dermal tissue via

dissection, selecting fibroblast free explant growths, us-

ing i3T3 feeder layers and including cholera toxin/EGF

routinely reduced contamination to 15% of cultures in

the first 3–4 passages. It should be noted that limited

success was achieved through removing fibroblasts with

either EDTA or trypsin once the fibroblast contamina-tion was established. Secondly, the readiness with which

cultured keratinocytes enter into their differentiation

pathway is a major limitation on the number of cells

which can be derived from a piece of starting material.

Although it has become well established that conditions

need to be carefully controlled in order to facilitate

complete maturation (Noel-Hudson et al., 1995), the

initiating events for differentiation appear to rely asmuch on inhibitory as stimulatory mechanisms (Pillai

et al., 1990; Tennenbaum et al., 1996).

Methods commonly employed to prevent excessive

levels of keratinocyte differentiation include using ap-

propriate substratum matrices which promote prolifer-

ation, such as laminin, maintaining a low calcium (<0.1

mM) and using fetal bovine serum to provide a high

concentration of proliferative signalling molecules(Boyce and Ham, 1983; Iwazaki et al., 1998). In addi-

tion, the use of EGF and cholera toxin has been re-

ported to be successful in stimulating keratinocyte

growth rate (Rheinwald and Green, 1977; Green, 1978).

The problem of lack of proliferative potential in cells

derived from explant culture is all the greater as the

proportion of clonogenic cells present in whole skin is

far less than for a population of cells obtained by se-lectively removing the basal cells from epidermal sheets

via digestion (Watt, 1988; Tenchini et al., 1992). The

level of cell expansion achieved here was therefore en-

couraging.

Although neither low calcium media nor specific

substrata were employed, passages of up to 6 were regu-

larly achieved through employing a combination of

86 A. Watson et al. / The Veterinary Journal 168 (2004) 81–86

approaches. Important among these was found to betransplanting the cells from explant growth before sig-

nificant differentiation was observed. It is possible that

areas of differentiation, which cannot then be properly

disaggregated, may trap clonogenic cells with high pro-

liferationpotential and therefore reduce the proportion of

these being subcultured. The fact that up to 50 colonies

could be observed within a week of initiating i3T3 feeder

layer cultures indicates a good transfer rate of prolifera-tive cells. If explants were allowed to grow out further,

thereby generating more primary cells, but at the expense

of increased levels of differentiation, the colony number

was significantly less with, on occasions, no colonies at all

being formed. Evenwith a good population of clonogenic

cells, there is still a tendency for these cells to enter into

differentiation, first forming the larger rapidly dividing

cells described by Staiano-Coico et al. (1986) and thensubsequently becoming non-proliferative. By monitoring

the morphological changes of the cells in culture it was

possible to identify cultures where the proportion of lar-

ger cells had increased (>20%, Fig. 3A versus B). Pas-

saging at or before this stage ensured cultures did not

become overrun with cells in a stage of incipient differ-

entiation. Thus, slower proliferating, basal cells were re-

tained as the predominant type. The �strategic passage�approach to culturing keratinocytes is at variance with

many previously described protocols where the conven-

tion has been to leave the cells until near or at confluency

(Wilkinson et al., 1987; Tenchini et al., 1992). In our ex-

perience, for canine cells, this resulted in a decreased level

of overall cell generation.

In conclusion, it is possible to obtain large numbers

of viable canine keratinocytes from small pieces of freshskin tissue by a combined approach of promoting the

proliferative potential of the cell population and limiting

the influence of unwanted contaminating cell types,

namely dermal fibroblasts. The method described could

be extremely valuable to researchers for whom the

supply of primary skin tissue is limited.

References

Boyce, S.T., Ham, R., 1983. Calcium-regulated differentiated of

normal human epidermal keratinocytes in chemically defined

clonal culture and serum-free serial culture. Journal of Investigative

Dermatology 81 (Suppl. 1), 33s–40s.

Daniels, J.T., Kearney, J.N., Ingham, E., 1996. Human keratinocyte

isolation and cell culture: a survey of current practices in the

United Kingdom. Burns 22, 35–39.

Fusenig, N.E., Worst, P.K., 1975. Mouse epidermal cell cultures II.

Isolation characterisation and cultivation of epidermal cells from

perinatal mouse skin. Experimental Cell Research 93, 443–457.

Green, H., 1978. Cyclic AMP in relation to proliferation of epidermal

cell. A new view. Cell 15, 801.

Hillier, A., Griffen, C.E., 2001. The AVCD task force on canine atopic

dermatitis (I): incidence and prevalence. Veterinary Immunology

and Immunopathology 81, 147–151.

Holbrook, K.A., Hennings, H., 1983. Phenotypic expression of

epidermal cells in vitro. A review. Journal of Investigative

Dermatology 81 (Suppl.), 11s–24s.

Iwazaki, T., Obata, H., Shimizu, M., 1998. Expression of basement

membrane macromolecules and integrin receptors by keratinocytes

during canine wound healing. In: Kwochka, K.W., Willemse, T.,

von Tscharner, C. (Eds.), Advances in Veterinary Dermatology,

third ed. Butterworth Heinemann, Oxford, pp. 339–354.

Jepsen, A., MacCallum, O.K., Lillie, J.H., 1980. Fine structure of

subcultivated stratified squamous epithelium. Experimental Cell

Research 125, 141–152.

Leigh, I.M., Watt, M.W., 1994. The culture of human epidermal

keratinocytes. In: Leigh, I.M., Lane, E.B., Watt, P.M. (Eds.), The

Keratinocyte Handbook. Cambridge University Press, England,

pp. 43–47.

Lloyd, D.H., 1985. Diagnostic methods in Dermatology. British

Veterinary Journal 141, 463–471.

Loveridge, G., 1994. Provision of environmentally enriched housing

for dogs. Animal Technology 45, 1–19.

Muller, G.H., Kirk, D.W., Scott, D.W., 1983. In: Comparative

Dermatology, Small Animal Dermatology. WB Saunders, Phili-

delphia, pp. 785–817.

Noel-Hudson, M.S., Dusser, I., Collober, I., Muriel, M.P., Bonte, F.,

Meybeck, A., Font, J., Wepierre, J., 1995. Human epidermis

reconstructed on synthetic membrane: influence of experimental

conditions on terminal differentiation. In vitro Cellular and

Developmental Biology Animal 31, 508–515.

Pillai, S., Bikle, D.D., Mancianti, M.L., Cline, P., Hincenbergs, M.,

1990. Calcium regulation of growth and differentiation of normal

human keratinocytes: modulation of differentiation competence by

stages of growth and extracellular calcium. Journal of Cell

Physiology 143, 294–302.

Rheinwald, J.G., Green, H., 1975. Serial cultivation of strains of

human epidermal keratinocytes: the formation of keratinizing

colonies from single cells. Cell 6, 331.

Rheinwald, J.G., Green, H., 1977. Epidermal growth factor and the

multiplication of cultured human epidermal keratinocytes. Nature

265, 421.

Scheidt, V.J., 1988. Flea allergic dermatitis. Veterinary Clinics of

North America: Small Animal Practice 18, 1023–1042.

Scott, D.W., Wolfe, M.J., Smith, C.A., Lewis, R.M., 1980. The

comparative pathology of non-viral bullous skin diseases in

domestic animals. Veterinary Pathology 17, 257–281.

Staiano-Coico, L., Higgins, P.J., Darzynkiewicz, Z., Kimmel, M.,

Gottlieb, A.B., Pagan-Charry, I., Madden, M.R., Finkelstein, J.L.,

Hefton, J.M., 1986. Human keratinocyte culture: identification and

staging of epidermal subpopulations. Journal of Clinical Investi-

gation 77, 396–404.

Tenchini, M.L., Ranzati, C., Malcavati, M., 1992. Culture techniques

for human keratinocytes. Burns 18 (Suppl. 1), S11–S15.

Tennenbaum, T., Li, L., Belanger, A.J., DeLuca, L.M., Yuspa, S.H.,

1996. Selective changes in laminin adhesion and alpha-6 beta-4

integrin regulation are associated with the initial steps in kerati-

nocyte maturation. Cell Growth and Differentiation 7, 615–628.

Watt, F.M., 1988. Epidermal stem cells in culture. Journal of Cell

Science 10 (Suppl.), 85.

Wilkinson, J.E., Smith, C., Suter, M., Lewis, R.M., 1987. Long-term

cultivation of canine keratinocytes. Journal of Investigative Der-

matology 88, 202–206.

Woodley, D.T., Bachman, P.M., O�Keefe, E.J., 1988. Laminin inhibits

human keratinocyte migration. Journal of Cell Physiology 136,

140–146.

Woodley, D.T., Wynn, K.C., O�Keefe, E.J., 1990. Type IV collagen

and fibronextin enhance human keratinocyte thymidine incorpo-

ration and spreading in the absence of soluble growth factors.

Journal of Investigative Dermatology 94, 139–143.