Regulation of integrin and extracellular matrix genes by HNRNPL is necessary for epidermal renewal

Stratified epithelia such as the epidermis require coordinated regulation of stem and progenitor cell proliferation, survival, and differentiation to maintain homeostasis. Integrin-mediated anchorage of the basal layer stem cells of the epidermis to the underlying dermis through extracellular matrix (ECM) proteins is crucial for this process. It is currently unknown how the expression of these integrins and ECM genes are regulated. Here, we show that the RNA-binding protein (RBP) heterogeneous nuclear ribonucleoprotein L (HNRNPL) binds to these genes on chromatin to promote their expression. HNRNPL recruits RNA polymerase II (Pol II) to integrin/ECM genes and is required for stabilizing Pol II transcription through those genes. In the absence of HNRNPL, the basal layer of the epidermis where the stem cells reside prematurely differentiates and detaches from the underlying dermis due to diminished integrin/ECM expression. Our results demonstrate a critical role for RBPs on chromatin to maintain stem and progenitor cell fate by dictating the expression of specific classes of genes.


Introduction
Integrins are large, transmembrane spanning proteins that are essential for tissue organization, survival, homeostasis, differentiation, and signal transduction [1]. Integrins form heterodimers between 18 α and 8 β subunits that can form up to 24 different receptors [2,3]. Integrins are important for epithelial tissues such as the epidermis to communicate as well as attach to the extracellular matrix (ECM) [4]. The human epidermis acts as an essential barrier between the body and the outside environment. It is a stratified epithelium that requires a proper balance of cell proliferation and differentiation to maintain its homeostasis and integrity [5]. Stem and progenitor cells in the basal layer of the epidermis maintain a population of proliferative cells while also giving rise to differentiated progeny. Those cells entering the differentiation program undergo various stages of differentiation as they migrate outwards through the epidermis, where they will eventually form the protective outer layer known as the stratum corneum. In order for the basal layer stem and progenitor cells to stay undifferentiated, they must attach (S1A-S1D Fig). In contrast, loss of HNRNPL severely decreased cell numbers partly due to a large increase in cells undergoing apoptosis (Fig 1A-1C). HNRNPL knockdown cells also started expressing differentiation genes such as FLG, LOR, KRT1, IVL, LCE3D, ZNF750, GRHL3, and ABCA12 despite the cells being grown in proliferation conditions ( Fig 1D). These differentiation genes are critical to epidermal structure and function. Mutations in these genes lead to human genetic skin diseases such as epidermolytic hyperkeratosis (KRT1), ichthyosis vulgaris (FLG), seborrheic dermatitis (ZNF750), psoriasis (LCE3D), and harlequin ichthyosis (ABCA12) [20][21][22][23][24]. We also used retroviral delivery of short hairpin RNAs (shRNAs) to knockdown HNRNPL to complement our siRNA results. Depletion of HNRNPL using 2 distinct shRNAs (shRNAs A and B) targeting different regions of HNRNPL resulted in loss of cell number as well as up-regulation of epidermal differentiation genes (S1E-S1G Fig). To determine the impacts of HNRNPL loss in a tissue setting that allows cell-cell and cell basement membrane contact in a three-dimensional context, control (CTLi) and HNRNPL knockdown (HNRNPLi) primary human keratinocytes were seeded onto devitalized human dermis to regenerate human skin [17,18,[25][26][27]. In control tissue, late terminal differentiation proteins filaggrin (FLG) and loricrin (LOR) were properly expressed and localized to the outermost layer of the epidermis (stratum corneum) ( Fig 1E). Interestingly, loss of HNRNPL caused FLG and LOR to be expressed throughout the entire epidermis even in the basal layer where the stem and progenitor cells reside (Fig 1E and 1F). Apoptosis as assayed by the presence of cleaved caspase 3 was found throughout the HNRNPLi epidermis (S2A Fig). Large clumps of apoptotic cells were also detected at the top of the epidermis in HNRNPLi tissue (S2A Fig,  white arrows). To understand the kinetics of HNRNPL's role on epidermal growth and differentiation, a time course of human skin regeneration was performed. After only 1 day of seeding the control and HNRNPL knockdown keratinocytes onto human dermis, the HNRNPLi tissue already had robust expression of early differentiation protein keratin 10 (K10) (Fig 1G). This was in contrast to control tissue where K10 was just starting to be expressed ( Fig 1G). The basal layer of the HNRNPLi epidermis became progressively more differentiated, and, by day 5, there were few cells not expressing K10 (Fig 1G). This premature differentiation coincided with a significant loss of proliferative, Ki67 positive cells in the basal layer (Fig 1G and 1H). In human skin, HNRNPL was expressed in the nucleus in all layers of the epidermis (S2B Fig).
Similarly, HNRNPL protein levels did not change upon induction of epidermal differentiation (S2C Fig). These results suggest that HNRNPL is required to prevent premature differentiation and apoptosis while promoting self-renewal of the epidermis.

HNRNPL promotes expression of hemidesmosome and extracellular matrix genes while suppressing differentiation genes
To determine the genes that HNRNPL is regulating, we performed RNA-seq on CTLi and HNRNPLi primary human keratinocytes. A total of 1,427 genes (p-value < 0.05 and �2-fold change) decreased in expression upon HNRNPL knockdown, which were enriched in genes involved in ECM organization and hemidesmosome assembly (Fig 2A and 2B, S1 Table). Moreover, 1,506 genes were up-regulated upon HNRNPL depletion, which were enriched in gene ontology (GO) terms such as skin development and keratinocyte differentiation (Fig 2A  and 2C, S1 Table). A comparison with the differentiation gene expression signature (genes that changed upon epidermal differentiation obtained from a previous publication [28]) showed that close to 43% of the HNRNPL gene signature overlapped ( Fig 2D). This suggests that HNRNPL is critical to prevent premature unraveling of the epidermal differentiation gene expression program. Notably, 2 critical gene classes were down-regulated in HNRNPLi cells including ECM organization and hemidesmosome assembly genes ( Fig 2E, S2 Table). This comprised of 69 genes including collagens, integrins, and laminins ( Fig 2E, S2 Table). Attachment of the basal layer of the epidermis to the dermis depends on those proteins and without it; the cells would prematurely differentiate as well as result in anoikis [6,[29][30][31]. Thus, downregulation of these critical attachment genes could explain the large increase in apoptotic cells as well as the premature differentiation phenotype of HNRNPL-depleted cells. Validation of these results by reverse transcription quantitative PCR (RT-qPCR) showed that HNRNPL loss resulted in decreased expression of integrins (ITGA5, ITGAV, ITGA6, ITGB1, ITGB4, and  ITGB6), laminins (LAMA3, LAMC2, and LAMB3), and fibronectin1 (FN1) (Fig 2F). Importantly, protein levels of β4 (ITGB4), α5 (ITGA5), β1 (ITGB1), and αV (ITGAV) integrins were also diminished upon HNRNPL loss ( Fig 2G). In tissue, HNRNPL knockdown resulted in loss  of β4 integrin (ITGB4), which is necessary for the attachment to the underlying dermis ( Fig  2H). Two days after seeding HNRNPLi keratinocytes onto dermis, there were local areas of detachment that progressed to complete detachment by day 4 (Fig 2I). This loss of epidermal attachment to the underlying dermis is similar to the clinical phenotypes seen in epidermolysis bullosa patients.

HNRNPL binds to hemidesmosome and cell-substrate junction assembly genes and promotes their expression
Since HNRNPL was one of the RBPs found to bind chromatin in a large-scale analysis [16] of 45 RBPs in K562 cells and 58 RBPs in HepG2 cells, we performed chromatin immunoprecipitation sequencing (ChIP-Seq) on HNRNPL in proliferating primary human keratinocytes to determine the genes that it may be directly regulating. HNRNPL bound to 2,843 peaks, with approximately 68% of the bound sites mapping back to genic regions (5 0 UTR, promoter, intron, exon, TTS, 3 0 UTR) ( Fig 3A). The 2,843 peaks mapped back to 2,332 genes that were  Table). A de novo motif search of the peaks that HNRNPL bound using HOMER [32] showed enrichment for transcription factors such as RUNX2, TWIST2, FOXH1, and SOX9 (S3A Fig). Of import, RUNX2 and SOX9 have been shown to be necessary for skin self-renewal and growth, suggesting that HNRNPL may be cooperating with these factors to maintain the stem cell state [33][34][35]. Since the vast majority of HNRNPL binding occurred within genic regions, we visualized how HNRNPL bound to these regions. HNRNPL binding peaked at the transcription start site (TSS) and was bound throughout the genic regions with a second peak near the transcription end site (TES) (Fig 3C). This binding was reminiscent of the RNA Pol II binding we had previously published [17,19,36] in human keratinocytes, suggesting that HNRNPL may be regulating transcription ( Fig 3C). Moreover, 10.4% of the genes up-regulated upon HNRNPL depletion were bound by HNRNPL, which were enriched in GO terms such as regulation of transcription (Fig 3D and 3E). A total of 216 (15.1%) of the genes down-regulated upon HNRNPL loss were also bound by HNRNPL and were enriched for cellsubstrate assembly, hemidesmosome assembly, and ECM organization genes (Fig 3D and 3F). This included genes such as ITGB4, LAMA3, and LAMC2, which are critical for the attachment of basal layer stem and progenitor cells to the underlying dermis ( Fig 3G and 3I). A zoomed out view of the ITGB4 genomic region with its proximal genes UNK and SAP30BP also confirmed the specificity of HNRNPL binding to ITGB4 (S3B Fig). To test the specificity of HNRNPL binding to integrin/ECM genes, we performed ChIP-qPCR to determine if HNRNPL knockdown could reduce the binding to those regions. ChIP-qPCR showed that

HNRNPL is necessary for Pol II binding and transcription of hemidesmosome and cell-substrate junction assembly genes
Because HNRNPL's binding on chromatin resembled Pol II, we wanted to determine if HNRNPL associated with Pol II. Immunoprecipitation (IP) of HNRNPL showed that it could associate with Pol II even in the absence of RNA (Fig 4A). These results suggest that HNRNPL may potentially be regulating transcription of hemidesmosome and ECM genes in 2 main ways: (1) HNRNPL may be promoting the transcription elongation of HNRNPL bound genes. This is a possibility since HNRNPL has been previously found to interact with transcription elongation factors such as P-TEFb [37]. peaks upon HNRNPL knockdown (Fig 4D, S4 Table). These 495 genes were enriched for GO terms such as cell-substrate junction and hemidesmosome assembly genes (Fig 4E, S4 Table).
Notably, 51% (111/216) of the HNRNPL bound genes that are down-regulated upon HNRNPL knockdown also lost Pol II peaks (S4B Fig). These 111 genes were also highly enriched for GO terms such as ECM organization, cell-substrate junction assembly, and hemidesmosome assembly (S4C Fig). These HNRNPL bound genes include ITGB4, LAMC2, ITGB1, ITGA3, ITGA2, LAMA3, and FN1 where HNRNPL loss led to depletion of Pol II from both the gene body and TSS (Fig 4F-4I, S4D-S4F Fig). Validation of these results by ChIP-qPCR also showed that HNRNPL knockdown led to a significant loss of Pol II binding to the TSS of ITGB4, ITGB1, ITGA3, FN1, and LAMB3 (S4G Fig). Since HNRNPL could potentially act as a repressor at its bound genes, we overlapped the genes that gained Pol II peaks upon HNRNPL knockdown with its bound genes. A total of 430 of those genes overlapped, which were enriched for GO terms such as peptidyl-tyrosine phosphorylation and positive regulation of transcription (S4H and S4I Fig). While it is possible that HNRNPL is acting as a repressor at those genes, it is unlikely that those genes are a major contributor to the observed phenotype. Altogether, our data suggest that HNRNPL is necessary to promote the stability of Pol II transcription through hemidesmosome and ECM genes.

Discussion
HNRNPL has been shown to be involved in most aspects of RNA regulation including RNA stability, splicing, as well as association with long noncoding RNAs (lncRNAs) to control gene expression [37][38][39][40]. It has been mostly studied in the context of cancer where it is required for tumor growth and metastasis [41]. In this context, HNRNPL interacts with various lncRNAs (specific lncRNAs dependent on tumor type) to promote tumorigenesis [42][43][44][45][46]. For example, the complex of lncRNA, DSCAM-AS1, and HNRNPL are necessary for breast cancer progression and tamoxifen resistance [42]. In hepatocellular carcinoma, the lncRNA SPRY4-IT1 associates with HNRNPL to promote growth and metastasis through the tumor necrosis factor (TNF) pathway [43]. Intriguingly, very little is known about the role of HNRNPL during normal tissue homeostasis as well as how it impacts gene expression by binding chromatin. In a large-scale study, HNRNPL was found to bind chromatin, but not much characterization was done on how it impacted gene expression [16].
Our results suggest HNRNPL is necessary to sustain epidermal stem and progenitor cells in the basal layer of the epidermis. In its absence, primary human keratinocytes undergo apoptosis, stop proliferating, and prematurely differentiate. In regenerated human skin, HNRNPL loss leads to terminal differentiation proteins being expressed in the basal layer of the epidermis, suggesting that the stem cell compartment has been lost. Apoptosis could be observed throughout the epidermis. In addition, the epidermis could no longer attach to the underlying dermis due to loss of expression of hemidesmosome/ECM genes. This phenotype resembled what is seen in epidermolysis bullosa patients who have mutations in integrin, laminin, or collagen. Integrin and ECM genes has been extensively studied; however, there is limited understanding on how these crucial genes are regulated on the transcriptional level [47,48]. Surprisingly, our HNRNPL ChIP-Seq data showed that HNRNPL bound genes were significantly enriched for hemidesmosome and cell-substrate junction assembly GO terms. This includes genes such as ITGB4 and LAMC2 that are crucial for the attachment of the basal layer to the dermis. It should be noted that HNRNPL binding to chromatin is weaker than other factors that have been traditionally characterized such as transcription or epigenetic factors. This weaker association was also observed by Xiao and colleagues when they mapped the DNA binding of a large number of RBPs [16]. Interestingly, HNRNPL binding to chromatin resembled that of Pol II and the 2 proteins also associate. This suggests that HNRNPL may be crucial for Pol II-mediated transcription on HNRNPL bound genes. Supporting this, knockdown of HNRNPL substantially decreased Pol II binding across HNRNPL bound genes that were enriched for hemidesmosome and ECM genes. Importantly, Pol II was only minimally impacted on genes not bound by HNRNPL.
Our data suggest that HNRNPL indirectly regulates differentiation and apoptosis since HNRNPL did not bind to these gene classes. Genes up-regulated (or gained Pol II peaks) upon HNRNPL depletion and were bound by HNRNPL were not enriched for differentiation or apoptosis genes. Instead, these genes were involved in the regulation of transcription from RNA Pol II promoter. Thus, HNRNPL indirectly prevents premature differentiation and anoikis by promoting the expression of hemidesmosome/ECM genes. This, in turn, allows attachment of the basal layer to the underlying dermis [6,49,50]. Previous studies have shown that keratinocytes grown in suspension without anchorage to the ECM spontaneously differentiate as it is mimicking the process of delamination from the basal layer [6,29,30]. A prior study showed that HNRNPL binds to muscle differentiation genes and activates their expression [51]. This is in contrast to our study where HNRNPL does not bind to differentiation genes in keratinocytes. HNRNPL is also necessary for maintaining the undifferentiated state in keratinocytes rather than promoting differentiation as shown in the muscle study. It is currently unclear whether the differences are due to cell type-specific differences (keratinocytes versus myoblasts) or the primary cells that were used in this study as compared to the transformed muscle cell line used in the previous study.
Since HNRNPL can bind to RNA, it is also possible that HNRNPL binds to RNA to control epidermal growth and differentiation. For example, HNRNPL could potentially control the mRNA stability or translation of skin master transcription factors such as P63 to promote expression of hemidesmosome/ECM genes. However, the major phenotypes of premature differentiation, apoptosis, and detachment of the epidermis from the dermis can be explained through transcriptional control of hemidesmosome and ECM genes through HNRNPL. Our model suggests that lineage-specific transcription factors open the chromatin to hemidesmosome/ECM genes, which then allows HNRNPL to maintain Pol II transcription at those genes. This is supported by our findings that HNRNPL bound genes were enriched with SOX9 and RUNX2 binding motifs. With HNRNPL being expressed throughout the epidermis, basal layer-specific transcription factors such as SOX9 [34,35] could potentially recruit HNRNPL to target genes to achieve layer-specific gene expression. In summary, we have elucidated how HNRNPL maintains epidermal renewal by promoting the expression of adhesion genes through RNA Pol II, which then allows attachment of the epidermis to the dermis.

Cell culture
Primary human epidermal keratinocytes (derived from neonatal foreskin) were purchased from Life Technologies (Carlsbad, California, USA) (C0015C) and cultured in EpiLife medium (Life Technologies: MEPI500CA). This media was supplemented with penicillin/streptomycin (HyClone SV30010) and human keratinocyte growth supplement (Life Technologies: S0015). Cells were induced to differentiate in full confluence and 1.  -013245-01-0005). The retroviral constructs (3 ug) expressing shRNAs were transfected using Fugene 6 (Roche, Basel, Switzerland) into amphotrophic phoenix cells to knockdown genes. Viral supernatants were collected 48 hours posttransfection and used to infect primary human keratinocytes. Cells were incubated in the viral supernatants and centrifuged at 1,000 reads per million (rpm) for 1 hour with the addition of polybrene (5 ug/ml). Cells were transduced on 2 consecutive days with the shRNA retroviral constructs and then selected using puromycin (2 ug/ml).

Regenerated human epidermis
Human dermis obtained from the New York Firefighters Skin Bank was devitalized, cut, and placed upon an organotypic cassette. The bottom of the dermis was coated in matrigel, and the cassette was placed in keratinocyte growth medium, with the bottom of the dermis contacting the media and the top at the air interface. A total of 1 million keratinocytes were then seeded on the top of the dermis and allowed to regenerate and stratify for 4 days. The regenerated human skin were then harvested and embedded in OCT (Optimal Cutting Temperature) for sectioning and staining. Additional details for regenerating human skin have been previously described [25,36,[52][53][54].

Immunofluorescent staining
Sectioned tissue-derived regenerated human skin or normal adult human skin was fixed with 10% formalin solution (Sigma (St. Louis, MO) HT5012) for 12 minutes. Sections were blocked (PBS, 2% bovine serum albumin, 2.5% normal goat serum, and 0.3% Triton X-100) for 30 minutes. Primary antibodies were then added to blocking buffer and put onto sections for 1 hour. The following antibodies were used at the following concentrations:

Hematoxylin-eosin staining
Sectioned tissue derived from regenerated human skin was fixed with 10% formalin solution (Sigma HT5012) for 12 minutes. Sections were then dipped in 0.25% Triton X-100 in PBS for 5 minutes. Hemotoxylin (Vector H-3401) staining was performed for 8 minutes, rinsed in water, and then dipped in acid alcohol (1% HCL in 70% ethanol). After subsequent rinsing, the sections were dipped in 0.2% ammonia water for 1 minute, rinsed again, and then dipped in 95% ethanol. Eosin (Richard-Allan Scientific (Kalamazoo, MI) 71304) staining was performed for 30 seconds followed by 95% ethanol rinsing for 1 minute. Sections were then put into 100% ethanol for 4 minutes followed by 2 minutes in Xylene.

RNA extraction and analysis by RT-qPCR
The GeneJET RNA purification kit (Thermo Fisher Scientific: K0732) was used to extract RNA from cultured keratinocytes. RNA concentration for each sample was measured by nanodrop, and 1 μg of RNA was used for generating cDNA. This cDNA was generated using the Maxima cDNA Synthesis Kit (Thermo Fisher Scientific: K1642). qPCR was performed using this cDNA on the Bio-Rad (Hercules, CA) CFX96 real-time system. The housekeeping gene L32 was used for normalization of signal. Primer sequences for all genes tested are listed in S1 Data.

Apoptosis assay
Four days after transfection, control and HNRNPL knockdown cells were stained with Annexin V conjugated to Alexa Fluor 488 (Life Technologies: A13201) and analyzed using the Guava flow cytometer (Millipore, Burlington, MA) according to manufacturer's instructions and as previously described [17].

Western blotting
IP samples or 20 to 80 ug of cell lysates were loaded into 4% to 12% Bis-Tris (Thermo Fisher Scientific: NW04122BOX) or 3% to 8% Tris-acetate (Thermo Fisher Scientific: EA03752BOX) and transferred to PVDF membranes. Membranes were blocked in 5% BSA in TBS. Membranes were exposed to primary antibodies in blocking buffer overnight at 4 degrees. The following primary antibodies were used: ITGB4 (Cell Signaling Technology: 14803S)

RNA sequencing and bioinformatic analysis
Control or HNRNPLi cells were harvested 4 days after siRNA transfection. Independent biological duplicates were obtained for both CTLi and HNRNPLi, and total RNA was isolated using the GeneJET RNA purification kit (Thermo Fisher Scientific: K0732) and quantified by Nanodrop. RNA-seq was performed using the Illumina (San Diego, CA) NovaSeq 6000. RNAseq libraries were prepared with NEBNext Ultra II Library Prep Kit (Illumina (San Diego, CA): E7760) then multiplexed, and approximately 40 million reads per sample were obtained. Pair-end reads were aligned to the GENCODE v19 transcriptome hg19 using TopHat2 with default settings [55]. Differential expression among samples was calculated using ANOVA from the Partek Genomic Suite (Partek, St. Louis, MO). Analysis of the read count distribution indicated that a threshold of 10 reads per gene generally separated expressed from unexpressed genes, so all genes with fewer than 10 reads were excluded from ANOVA analysis. Gene lists for significantly up-regulated or down-regulated genes were created using p-value < 0.05 and �2-fold change. Enriched GO terms for RNA-seq differentially expressed gene sets were identified using Enrichr [56,57]. Heatmaps for the RNA-seq data were generated using Partek Genomic Suite (http://www.partek.com/partek-genomics-suite/).

ChIP sequencing and bioinformatic analysis
A total of 10 million cells and 5 μg of antibody were used for each antibody pulldown experiment for ChIP [17,27,54]. ChIP was performed using the following antibodies: HNRNPL (Bethyl: A303-895A), RNA Pol II (Active Motif: 91151), mouse IgG (Abcam: AB18413), and rabbit IgG (Millipore: 12-370). Cells for the RNA Pol II ChIP-qPCR or ChIP-Seq were fixed at a final concentration of 1% formaldehyde. Cells for the HNRNPL ChIP-qPCR or ChIP-Seq were fixed in both formaldehyde (1% final concentration, Thermo Fisher Scientific: 28908) and disuccinimidyl glutarate (2 mM final concentration, Thermo Fisher Scientific: 20593). qPCR results are represented as a percentage of input DNA. For ChIP-Seq, the ChIP DNA library was prepared using the TrueSeq DNA sample prep kit (Illumina). Sequencing was done on the HiSeq 4000 System (Illumina) using single 1 × 75 reads at the Institute for Genomic Medicine Core, UCSD. The ChIP-seq reads were processed by the ENCODE Transcription Factor and Histone ChIP-Seq processing pipeline (https://github.com/ENCODE-DCC/ chip-seq-pipeline2) on our local workstation. The reads were first trimmed based on quality score before alignment to reference hg19; upon alignment and deduplication, the peak calling was then carried out by MACS2.2.4 with a cutoff q-value of 0.05 [58,59]. The heatmaps and metagene plots for the ChIP-Seq data were generated using ngs.plot [60]. Gene tracks were visualized using UCSC genome browser along with annotation tracks. The differential peaks between CTLi and HNRNPLi Pol II ChIP-Seq were determined using diffReps [61].

Co-immunoprecipitation
Primary human keratinocytes were harvested in IP lysis buffer (25 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% NP-40 and 5% glycerol) and sheared with a syringe. Moreover, 5 ug of HNRNPL (Bethyl: A303-895A) or control rabbit IgG (Millipore: 12-370) was conjugated to 50 ul of Protein G dynabeads (Life Technologies: 10004D) for 30 minutes at room temperature. Lysis buffer was diluted to 1 ml per sample and was added to the antibody conjugated beads and incubated overnight at 4 degrees. The following day, the beads were washed with IP lysis buffer and boiled in RIPA buffer (25 mM Tris-HCl (pH 7.6), 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) supplemented with NuPAGE LDS Sample Buffer (Life Technologies: NP0008) to elute. Samples were then loaded for western blotting.

Statistics
Statistical analyses were performed using GraphPad Prism 9. Histogram data are presented as the mean ± SD, and significant differences between 2 samples were determined by Student t tests. For groups of 3 or more, 1-way ANOVA with Tukey was used. Significant changes were defined as p � 0.05.
Supporting information S1 Fig. Knockdown of HNRNP genes shows that only HNRNPL is necessary for epidermal growth and blockade of premature differentiation. (A) Primary human keratinocytes were knocked down with control (CTLi), HNRNPFi, HNRNPH1i, or HNRNPH2i siRNAs, and the remaining mRNA levels of each gene was measured using RT-qPCR. qPCR results were normalized to L32 levels. (B) CTLi, HNRNPFi, HNRNPH1i, or HNRNPH2i cells were seeded at 150,000 cells and counted 4 days later. Cell number was reported as a percentage of CTLi levels. (C, D) RT-qPCR for IVL (C) and FLG (D) mRNA expression in CTLi, HNRNPFi, HNRNPH1i, or HNRNPH2i cells after 4 days of knockdown. qPCR results were normalized to L32 levels. (E) Primary human keratinocytes were knocked down with control (CTL shRNA) or HNRNPL (HNRNPL shRNA-A or shRNA-B) shRNAs delivered by retroviruses. HNRNPL shRNAs A and B are distinct shRNAs that target 2 different regions of the gene. Knockdown was measured using RT-qPCR. qPCR results were normalized to L32 levels. (F) CTL shRNA, HNRNPL shRNA-A, and HNRNPL shRNA-B cells were seeded at 150,000 cells and counted 6 days later. Cell number was reported as a percentage of CTLi levels. (G) RT-qPCR for epidermal differentiation gene expression in CTL shRNA, HNRNPL shRNA-A, and HNRNPL shRNA-B cells after 6 days of knockdown. qPCR results were normalized to L32 levels. N = 3 independent experiments for the entire figure. Mean values are shown with error bars = SD. � p < 0.05, �� p < 0.01, ��� p < 0.001, ���� p < 0.0001 (t test for A-D and 1-way ANOVA with Tukey for E-G). Primary data for this figure can be found in S1 Data. HNRNPL, heterogeneous nuclear ribonucleoprotein L; RT-qPCR, reverse transcription quantitative PCR; shRNA, short hairpin RNA; siRNA, small interfering RNA. Mean values are shown with error bars = SD. � p < 0.05, �� p < 0.01, ��� p < 0.001, ���� p < 0.0001 (t test was used and each sample compared to CTLi). Primary data for this figure can be found in S1 Data. ChIP-Seq, chromatin immunoprecipitation sequencing; HNRNPL, heterogeneous nuclear ribonucleoprotein L; IGG, immunoglobulin; RPM, reads per million; TSS, transcription start site.  Table. RNA-seq of HNRNPL knockdown primary human keratinocytes. Primary human keratinocytes were grown in proliferation conditions and knocked down for control (CTLi) or HNRNPL (HNRNPLi). Gene lists for significantly up-regulated or down-regulated genes were created using p-value < 0.05 and �2-fold change. HNRNPL, heterogeneous nuclear ribonucleoprotein L; RNA-seq, RNA sequencing. (XLSX) S2 Table. ECM and hemidesmosome genes down-regulated upon HNRNPL depletion. List of the 69 ECM and hemidesmosome genes that are decreased in expression upon HNRNPL knockdown. ECM, extracellular matrix; HNRNPL, heterogeneous nuclear ribonucleoprotein L. (XLSX) S3 Table. ChIP-Seq of HNRNPL in proliferating keratinocytes. List of the genes bound by HNRNPL including the coordinates where the peaks were found. ChIP-Seq, chromatin immunoprecipitation sequencing; HNRNPL, heterogeneous nuclear ribonucleoprotein L. (XLSX) S4