We used an Fntb conditional knockout allele, Fntbfl (M. Liu et al., submitted for publication; Fig. A and B), and a keratin 14–Cre (KCre) transgene to produce mice lacking FTase in stratified epithelial cells. Keratinocyte-specific Fntb knockout mice (Fntbfl/flKCre+, designated FntbΔ/Δ) were born at the expected Mendelian frequency. Skin keratinocytes from newborn FntbΔ/Δ mice contained only trace levels of Fntb transcripts, as judged by quantitative polymerase chain reaction (PCR) (2.5 ± 0.3% of levels in wild-type keratinocytes; n = 2 wild-type and n = 5 FntbΔ/Δ keratinocyte preparations). Consistent with those results, the β-subunit of FTase (FNTB) was undetectable in keratinocyte extracts, as judged by western blots with an FNTB-specific antibody (Fig. C).
Figure 1. Inactivating FTase in keratinocytes. (A) Wild-type and conditional knockout alleles for Fntb (Fntb+ and Fntbfl, respectively). LoxP sites, indicated by a triangle, are located upstream of the Fntb promoter and in intron 1. (B) PCR-amplified fragments (more ...)
The absence of FTase was accompanied by reduced farnesylation of keratinocyte proteins. Western blots of FntbΔ/Δ keratinocyte extracts with a DNAJA1-specific antibody revealed that the majority of the DNAJA1 was non-farnesylated (Fig. D and E). Also, FntbΔ/Δ keratinocytes exhibited a striking accumulation of prelamin A. A prelamin A band was obvious on western blots with antibodies against prelamin A or mature lamin A (top and middle panels, respectively, of Fig. F). In western blots with the lamin A antibody, the mature lamin A band was detectable, but was less intense than the prelamin A band. The prelamin A in FntbΔ/Δ keratinocytes did not appear to be farnesylated, as judged by its retarded electrophoretic mobility (Fig. G). The prelamin A in FntbΔ/Δ keratinocytes co-migrated with the non-farnesylated prelamin A in FTI-treated wild-type cells, and the prelamin A in FntbΔ/Δ keratinocytes migrated slightly more slowly than the farnesyl-prelamin A in Zmpste24−/− keratinocytes (ZMPSTE24 is the enzyme required for the conversion of farnesyl-prelamin A to mature lamin A) (Fig. G). As expected, HRAS was found in the S100 (cytosolic) fraction in FntbΔ/Δ keratinocytes, but was located in the P100 (membrane) fraction of wild-type cells (Fig. H).
We suspected that we might be able to observe an accumulation of prelamin A in the skin of FntbΔ/Δ
mice by immunohistochemistry. In early immunohistochemistry experiments with a prelamin A antibody developed by Fong et al
) and several commercial antibodies, we were disappointed by a high level of background staining. For that reason, rats were immunized with a peptide corresponding to the last 18 amino acids of mouse prelamin A. A rat with a strong immune response was used to develop a prelamin A-specific monoclonal antibody (antibody 7G11). Antibody 7G11 bound avidly to the non-farnesylated prelamin A in FTI-treated fibroblasts, but did not bind to lamin A or lamin C (Fig. A and B). In immunohistochemistry studies, prelamin A staining with antibody 7G11 was undetectable in wild-type skin keratinocytes (Fig. C), but was intense in keratinocytes from FntbΔ/Δ
mice (Fig. D). As expected, prelamin A in FntbΔ/Δ
skin was confined to keratinocytes expressing K14 and was undetectable in the dermal fibroblasts (Fig. D). Prelamin A was undetectable in the skin keratinocytes of mice lacking GGTase-I (Pggt1bΔ/Δ
) (Fig. E), but was found in the skin keratinocytes of mice lacking both FTase and GGTase-I (FntbΔ/ΔPggt1bΔ/Δ
) (Fig. F). As expected, prelamin A accumulated in both keratinocytes and dermal fibroblasts of Zmpste24−/−
mice, but was confined to keratinocytes in FntbΔ/Δ
mice (Supplementary Material, Fig. S1
). Interestingly, higher power images revealed that most of the prelamin A in FntbΔ/Δ
keratinocytes was located at the nuclear rim (Fig. G–J).
Figure 2. Experiments with a new rat monoclonal antibody against prelamin A, 7G11, showing that prelamin A accumulates in FntbΔ/Δ keratinocytes. (A) Schematic depicting prelamin A, lamin A and lamin C and their recognition by the indicated antibodies. (more ...)
FntbΔ/Δmice could be distinguished from littermates on post-natal day 6 (P6) by alopecia (hair normally becomes visible at P6 when the hair shafts emerge from follicles) (Fig. A). The alopecia was virtually complete and persisted throughout life (Fig. A). FntbΔ/Δmice were smaller than unaffected littermates, but appeared to be healthy and survived normally (some FntbΔ/Δmice were maintained in the colony for >300 days).
Figure 3. Alopecia in keratinocyte-specific Fntb knockout mice (FntbΔ/Δ). (A) Alopecia in FntbΔ/Δ mice was detectable by close visual inspection on post-natal day 6 (P6) but was obvious at later time points (P17, P40, P65). (B) Hematoxylin (more ...)
Hematoxylin and eosin-stained sections of FntbΔ/Δskin revealed that the hair follicles in post-natal FntbΔ/Δmice were stunted in appearance (both shorter and narrower) (Fig. B). Some follicles from FntbΔ/Δmice contained hair shafts, but most of the hair shafts were dysmorphic and improperly angled and few penetrated the surface of the skin (Fig. B). The hair follicles appeared normal at embryonic day 17.5 (E17.5), but stunting of hair follicles was easily apparent at all post-natal time points (Fig. A).
Figure 4. Analysis of alopecia in keratinocyte-specific Fntb knockout mice (FntbΔ/Δ). (A) Hematoxylin and eosin-stained sections of skin from Fntb+/+ and FntbΔ/Δ mice at different time points. At P120, few hair follicles remained (more ...)
Hair follicle stem cells express Sox9 and can first be detected during epidermal development by immunohistochemistry with Sox9-specific antibodies (44
). Interestingly, Sox9 staining was normal in FntbΔ/Δ
mice, suggesting that Fntb
deficiency has little effect on the formation of the stem cell compartment (Fig. B). The matrix, located at the base of the hair follicle, contains proliferating cells that differentiate to form hair shafts. Immunohistochemical staining for the matrix marker CDP (CCAAT displacement protein) was normal in FntbΔ/Δ
mice, although the size of the matrix was smaller than that in wild-type mice (Fig. C). Staining for keratin 31, a marker of the hair cortex, was detectable in FntbΔ/Δ
mice, but the staining was less intense (Fig. D). Electron microscopy of skin from FntbΔ/Δ
mice revealed that all of the appropriate layers of the hair shaft were generated but the hair shafts were smaller and not as straight as those of wild-type mice (Fig. A).
Figure 5. Electron microscopy of skin in Fntb+/+ and FntbΔ/Δ mice. (A) Electron micrographs of hair shafts, revealing that all layers of the hair shaft are present in FntbΔ/Δ mice, but the hair shafts were smaller than those in (more ...)
There were no obvious morphological defects in the interfollicular epidermis in FntbΔ/Δ
mice, as judged by routine histology (Figs B and A), and staining for markers of epidermal differentiation did not uncover any abnormalities (Supplementary Material, Fig. S2
). Also, electron microscopy showed that all layers of the epidermis were intact (Fig. B). The absence of pathology in the interfollicular epidermis, either by light or electron microscopy, led us to predict that skin barrier function in FntbΔ/Δ
mice would be normal. Indeed, barrier function assays in E19.5 FntbΔ/Δ
mice uncovered no abnormalities (Supplementary Material, Fig. S3A
We examined immunohistochemical markers for cell division and apoptosis in the skin of FntbΔ/Δ and control mice. Staining for the phosphorylated form of histone H3 (phosphoH3), a marker of cell mitosis, appeared to be normal in FntbΔ/Δ mice (Fig. A). Staining for another marker of dividing cells, Ki67, was also normal in FntbΔ/Δ skin (Fig. B). To identify apoptotic cells, skin was stained with an antibody recognizing the cleaved form of caspase 3. Apoptotic cells are normally found in hair follicles only during the catagen phase of the hair cycle, which begins around P15, but apoptosis is normally absent in P1–P8 skin. In FntbΔ/Δ mice, apoptotic cells were detected in the hair follicle at P1 and were even more numerous at P4, P6 and P8 (Fig. C), providing a plausible explanation for the stunted morphology of the hair follicles. To explore this issue further, we examined the hair follicles of FntbΔ/Δ and control mice by electron microscopy. Numerous apoptotic cells were identified in the hair follicle matrix of FntbΔ/Δ mice at P6, whereas none was found in control mice (Fig. C).
Figure 6. Fntb deficiency in the epidermis does not affect proliferation in hair follicles but leads to increased numbers of apoptotic cells. (A) Immunohistochemical staining of skin with an antibody against phosphorylated histone-H3 (ph-H3, green), a marker of (more ...)
Scattered apoptotic cells were also found in the interfollicular epidermis of FntbΔ/Δ mice at P4, P6 and P8 (Fig. D). Why apoptosis was associated with clear-cut pathology in hair but less so in interfollicular epidermis is unknown, but one possibility is that dying cells are replaced more effectively in the interfollicular epidermis. Consistent with this notion, hair follicles had largely disappeared by P120, whereas the interfollicular epidermis appeared to be normal at that time point (Fig. A).
Inactivating Fntb in mouse embryonic fibroblasts abolishes their ability to proliferate (M. Liu et al., submitted for publication). To determine whether Fntb deficiency in keratinocytes causes a similar proliferation abnormality, equal numbers of keratinocytes from P1 FntbΔ/Δ and Fntb+/Δ mice were seeded into culture dishes. Fntb+/Δ keratinocytes proliferated in a robust fashion, whereas FntbΔ/Δ keratinocyte cultures were sparse and the cells had a flattened morphology (Fig. A). Cell counting confirmed a marked proliferation defect in FntbΔ/Δ keratinocytes (Fig. B and C).
Figure 7. Fntb deficiency abolishes the ability of keratinocytes to proliferate. (A) Phase contrast images of keratinocytes from P1 Fntb+/Δ and FntbΔ/Δ mice after 5 or 8 days in culture. (B, C) Quantitative analysis of keratinocyte growth. (more ...)
The same experimental approach was used to assess the importance of GGTase-I in stratified epithelium. A conditional knockout allele for Pggt1b, Pggt1bfl
) and the KCre
transgene was used to produce keratinocyte-specific Pggt1b
knockout mice (Pggt1bfl/flKCre+
, designated Pggt1bΔ/Δ
). The Cre-mediated recombination event excised exon 7 from Pggt1b
, eliminating the catalytic site of the enzyme and yielding a frameshift (41
keratinocytes had negligible levels of Pggt1b
transcripts, as judged by quantitative PCR (0.8 ± 0.2% of those in wild-type mice; n
= 2 wild-type and n
= 4 Pggt1bΔ/Δ
keratinocyte preparations). The β-subunit of GGTase-I was undetectable in keratinocytes by western blotting (Fig. A). As expected, non-prenylated RAP1 (a protein that is normally geranylgeranylated) accumulated in Pggt1bΔ/Δ
keratinocytes (Fig. B).
Figure 8. Inactivating GGTase-I in skin keratinocytes. (A) Western blot of keratinocytes from Pggt1b+/+and Pggt1bΔ/Δ mice with an antibody against the β-subunit of GGTase-I. Actin was used as a loading control. (B) Western blot of keratinocyte (more ...) Pggt1bΔ/Δ
mice survived development and were born at the expected Mendelian ratios, but died within a few hours of birth. At the time that the mice died, they invariably had little milk in their stomach. Histological studies of E19.5 Pggt1bΔ/Δ
embryos revealed stunted hair follicles, but the interfollicular epidermis appeared normal (Fig. C). The skin barrier was intact in Pggt1bΔ/Δ
mice (Supplementary Material, Fig. S3B
). Even though no abnormalities were identified in the interfollicular epidermis of Pggt1bΔ/Δ
epidermal keratinocytes did not proliferate in culture (Fig. D).
Fntbfl/flPggt1bfl/fl and Fntbfl/+Pggt1bfl/+KCre+ mice were intercrossed with the goal of generating FntbΔ/ΔPggt1bΔ/Δ embryos. A few FntbΔ/ΔPggt1bΔ/Δ embryos were identified. At E17.5, FntbΔ/ΔPggt1bΔ/Δ embryos had fewer hair follicles, and the interfollicular epidermis was slightly thinner compared with skin from mice lacking only Fntb or only Pggt1b (Fig. A and B). Immunohistochemical studies of E17.5 FntbΔ/ΔPggt1bΔ/Δ skin revealed more apoptotic cells than in embryos lacking only Fntb or only Pggt1b (Fig. C).
Figure 9. Inactivating both FTase and GGTase-I results in fewer hair follicles and a thinner interfollicular epidermis at E17.5. (A) Hematoxylin and eosin-stained sections of skin from Fntb+/+Pggt1b+/+, Fntb+/ΔPggt1bΔ/Δ, FntbΔ/Δ (more ...)