Keratin intracellular concentration revisited: Implications for keratin function in surface epithelia

TO THE EDITOR:

In 1978 Sun and Green carried out a densitometry-based analysis of polyacrylamide gel-electrophoresed samples and reported that keratin proteins account for 25~35% of total cellular proteins in human epidermal keratinocytes serially passaged in primary culture. This represents an astounding figure with very significant implications for the structural support role of keratin intermediate filaments (IFs) in epidermis, and its regulation by post-translational modifications and/or interaction with associated proteins. We report here on an effort to investigate this issue further, using a distinct method of keratin quantitation in sorted but otherwise native cell populations, and taking advantage of the significant progress made in our understanding of the structure of keratin filaments and differential regulation of keratin expression in epidermis.

Freshly isolated keratinocytes were obtained directly from 2-day old C57Bl/6 newborn mice, labeled with fluorochrome-conjugated antibodies to the surface marker integrin beta 1, and then sorted by flow cytometry according to expression level of this antigen (Jones and Watt, 1993). As an essential component of cell-surface receptor for extracellular matrix ligands (Hynes 1987), this integrin is primarily expressed in basal layer keratinocytes and only poorly so in differentiating keratinocytes in the suprabasal layers of epidermis. We collected cells that express intermediate levels of integrin beta 1 at their surface (Figure 1a). Such integrin β-1 “dim” cells (Jones et al., 1995) consist of basal cells with transit amplifying proliferation status and represent the majority of basal layer keratinocytes. The less abundant cells that express surface integrin β-1 at a higher level (designated “bright”) are enriched in epidermal stem cells (Jones et al., 1995). Ultrastructural analysis shows that the sorted integrin β-1-dim cells are homogeneous, with a round shape and a smooth surface, and features a round and centrally-located nucleus surrounded by thick bundles of keratin filaments in the cytoplasm (Figure 1b). The appearance of keratin filaments, in particular, is very reminiscent of those seen in basal layer keratinocytes in situ (Coulombe et al., 1989).

Figure 1.

Isolation and characterization of epidermal basal keratinocytes. (a) Keratinocytes harvested from C57Bl/6 P2 mouse skin were stained with fluorochrome-conjugated antibody to the surface protein integrin β-1 and sorted using flow cytometry. An integrin β-1-dim status reflects a transit-amplifying population of basal keratinocytes, whereas an integrin β-1-bright status denotes a subpopulation enriched in stem cells. (b) Transmission electron micrograph of a representative sorted basal (integrin β-1-dim) keratinocyte. N = nucleus, Krt IFs = keratin IFs. Bar = 1µm. (c) Immunofluorescence micrographs of P2 skin cross-sections highlight the occurrence of K17-positive basal cells in the hair follicle-proximal interfollicular epidermis (see arrows). bc = basal cell, hf = hair follicle. Bar = 10 µm. (d) Western blots show the presence of K17 but not K16 in FACS-sorted basal cells, given the protocol used (see a). “Std” refers to relevant purified keratin standard.

Sorted basal cells were lysed in urea lysis buffer and the resulting protein extracts were analyzed to measure the concentrations of K5, K14 and K15, the three major keratins in the basal progenitor cells of epidermis. Serially diluted aliquots of native protein lysates were resolved by polyacrylamide gel electrophoresis, and K5, K14 and K15 antigens were quantitated by infrared Western blot analysis (Li-Cor Biosciences, Lincoln, NE). Calibration curves were established using purified recombinant forms of human K5, human K14, and mouse K15 as standards, taking advantage of the linear relationship between Western signal intensity and protein concentration (Figure 2a–c). Relevant experimental details are given in Supplementary Materials.

Figure 2.

Average keratin concentration and number of keratin filaments in basal layer keratinocytes of epidermis. The keratin fraction of total cell proteins in sorted basal keratinocytes was determined using quantitative infrared Western blot analysis. (a,b,c) Western blots show a linear relationship between keratin standard concentration signal intensity for K5 (a), K14 (b) and K15 (c). (d) The keratin fraction of total cell proteins, and the number of keratin monomers in sorted basal cells, were calculated from these calibration curves. Data were presented as average ± standard deviation (s.d.) from eight (for K5) or six (for K14, K15) independent experiments. (e) Estimation of keratin amount/concentration and number of keratin filaments in epidermal basal keratinocytes. Additional data are given in Tables S1–3.

The quantitative data obtained is related in part in Figure 2d and e, and otherwise in toto in Tables S1–3. We estimate that the average sorted basal keratinocyte contains 36.81±4.34 pg of total protein, of which 4.91±1.07 pg, 2.18±0.80 pg, and 0.98±0.01 pg respectively represent K5, K14 and K15 (Table S3). This yields a 1.27:1 molar ratio for type II vs. type I keratin proteins, rather than the expected 1:1 ratio (Fuchs et al., 1987). This difference could be due to the presence of K17 (but neither K6 nor K16) in a small subset of sorted basal keratinocytes originating from the hair follicle-proximal interfollicular epidermis and the upper infundibulum (McGowan and Coulombe, 1998; cf. Figure 1c,d). Alternatively, it is formally possible that the antibodies we used have different affinities for mouse and human keratins (if so, the difference would be expected to be small; cf. Supplementary Materials), or that there is a small imbalance in the keratin I/II molar ratio when considering the total pool, instead of the polymerized pool, of keratins (see Lu et al., 2005).

We selected the data collected for K5 as a reference, and assumed a 1:1 molar ratio per keratinocyte, for further calculations. Assuming 700 keratin monomers per micrometer length of filament (Herrmann and Aebi, 1998), a 4 micron average filament length (see Supplementary Materials), and polymerization to a 98% extent based on a previous study (Bernot et al., 2005; cf. Table S1), we estimate that the average basal keratinocyte contains 33,179 ± 7,214 keratin filaments (Fig. 2e). This estimate obviously depends on assumptions made about the length of individual filaments (Fig. 2e). Even when factoring modest errors made in either the assumptions or the data obtained, it appears that there are tens of thousands of keratin filaments in basal keratinocytes of epidermis.

In our hands, keratins account for 17~27% of total cellular proteins (Figure 2d–e) in newborn mouse keratinocytes (i.e., up to 13.4%, 5.9%, and 2.8% for K5, K14, and K15, respectively), a finding that is close to what was reported by Sun and Green (1978) for human keratinocytes more than 30 years ago. Using estimates of cell size, cell volume (and cytoplasmic volume) derived from measurements made on transmission electron micrographs of sorted cells (Table S2), we calculate that the total concentration of these three keratins in the average non-stem basal keratinocyte is ~40µg/µL or 520µM (Table S3). By comparison, the total actin concentration ranges between 25 and 200µM in various cell types (Pollard et al., 2000), and is reportedly up to 900µM in skeletal muscle cells (Jaeger et al., 2009). The latter figures convey that the concentration of keratin in basal keratinocytes approximates that of actin in muscle tissue. Further, our assumptions and measurements together yield a total protein concentration of ~180µg/µL in sorted keratinocytes (Table S3), a figure that is consistent with previous reported values for mammalian cells (50–400µg/µL; Schnell and Turner, 2004). We note that while the soluble pool represents only 2% of total keratin proteins in basal keratinocytes (Bernot et al., 2005), the corresponding number of protein monomers (~1.9 million, Figure 2e) and concentration (~10µM, Table S3) is large relative to the pool of most other cellular proteins. This sizable soluble pool is presumably available to sustain the remodeling of keratin filaments under steady-state conditions, and/or to fulfill non-structural roles in the cell.

These quantitative figures are essential to a deeper understanding of keratin organization and function, and their regulation, in epidermis and related surface epithelia.