Gene microarray analysis of human renal cell carcinoma

Abstract

Histone deacetylase (HDAC) inhibitor treatments can augment the anti-tumor effects of retinoids in renal cancer cells. We studied the effects of the HDAC inhibitor suberoylanilide hydroxamic acid (SAHA) and 13-cis retinoic acid (cRA) on two human renal cell carcinoma (RCC) lines. Cells were cultured in the presence of each drug for six days to determine the responses to monotherapy and to combination therapy. The proliferation of SKRC06 was inhibited with cRA treatment; the proliferation of SKRC39 was not. However, both RCC lines were sensitive to growth inhibition by dibutyryl cyclic AMP, with or without 13-cis RA. SAHA alone alone reduced cell proliferation in both cell lines. To identify the alterations in gene expression that correlate with the responsiveness to treatment, gene microarray analyses were performed. Several retinoid-regulated genes exhibited much higher mRNA levels in SKRC06 than in SKRC39, even in the absence of drugs; these included crabp2, rarγ and cyp26A1. Combination treatment of cells with both SAHA and cRA induced several transcripts with known anti-cancer/immunomodulatory effects, including dhrs9, gata3, il1β, phlda1, txk and vhl. Immunostaining confirmed the decreased expression of gata3 in human RCC specimens compared to normal kidney. Together, our results show that treatment of RCC with cRA and/or SAHA increases the expression of several genes and gene families that result in reduced cell proliferation.

Introduction

Renal cell carcinoma (RCC) arises from proximal renal tubular epithelium and accounts for the majority of primary kidney cancers in adults.¹ In the United States, deaths from renal cell carcinoma exceed 12,000 per year. More than 50,000 cases are diagnosed annually, and the incidence is rising.^2,3 Although most patients with early-stage lesions will be cured, metastatic disease is highly resistant to radiotherapy and conventional chemotherapy.⁴ Despite innovations in surgical techniques and the emergence of targeted therapies,^5-7 survival for patients with advanced stage disease is poor; more than 90% of patients with metastases will die within five years.⁴ For these reasons, novel therapies are required and a variety of new treatment regimens have been developed.⁸

A large body of clinical, pre-clinical and epidemiological evidence supports the use of retinoids in the treatment and/or prevention of cancer.^9,10 Retinoids are naturally occurring or synthetic compounds related to vitamin A (retinol). Retinoid action is generally mediated by two classes of nuclear receptors, retinoid X receptors (RXR-α, RXR-β and RXR-γ) and retinoic acid receptors (RAR-α, RAR-β and RAR-γ). RARs and RXRs belong to the nuclear hormone receptor superfamily.¹⁰ Members of this family act as ligand-inducible transcription factors that regulate the expression of target genes via binding to DNA response elements in gene promoters or enhancers.¹⁰ RAR-RXR heterodimers bind to these retinoic acid response elements (RAREs) and act to control transcription.¹¹ Retinoids have significant effects on many cell functions, including differentiation and apoptosis. They are currently widely used in the treatment of several cancers, most notably acute promyelocytic leukemia and neuroblastoma.^12-14

Several clinical trials have evaluated the use of retinoids, vitamin A, and related signaling molecules in combination with cytokines. A 1995 study noted a small but statistically significant improvement in progression-free and overall survival in RCC patients given a combination of 13-cis retinoic acid (cRA) and IFNα-2a (interferon-α-2a) compared to patients treated with monotherapy.¹⁵ Objective responses were also noted in bone metastases, a site known for resistance to therapy.¹⁵ A follow-up phase III study, headed by the same investigator, failed to show an increase in median overall survival with addition of cRA to IFN, although the progression-free survival and duration of response at two years were longer in the combination group when compared to the monotherapy (IFN alone) group.¹⁶ A subsequent trial by another group showed that the median overall survival in the combination therapy (IFN + cRA) arm was greater than in the arm with IFN alone (p = 0.048).¹⁷ Another study evaluating cRA, IFN and IL2 +/- 5FU or the antimetabolite capecitabine showed a three year overall survival of 29.7% in patients treated with triple therapy.¹⁸ A similar trial was recently conducted using a retinoid-containing four-drug regimen. Patients were administered cRA, IL-2, IFNα and capecitabine. The overall response rate was 53.6%; median progression-free survival and overall survival were 14.7 months and 27.8 months, respectively.¹⁹ Liposomal all-trans RA combined with interferon α also resulted in stable disease in patients with metastatic renal cell carcinoma.²⁰ Maintenance therapy with IL-2, IFNα and cRA is indicated by recent clinical research.²¹

Histone deacetylase inhibitors are a new class of anti-cancer drugs.^22-24 Histone deacetylases (HDACs) and histone acetyltransferases (HATs) are enzymes that influence transcription by selectively deacetylating or acetylating core histone proteins.^22-24 Acetylation of lysine residues at the amino terminus of histone proteins removes positive charges and therefore reduces affinity for the negatively charged DNA.^25,26 Deacetylation causes gene silencing by creating a compact chromatin state that prevents the transcription machinery from obtaining access to promoters.²⁶ HDACs are also involved in the reversible acetylation of non-histone proteins.^22,27 HATs are thought to function as tumor suppressors, and altered HDAC and/or HAT activities are present in many types of cancers.^27,28 HDAC inhibitors result in hyperacetylation of core histones.²⁹ These compounds have varied effects on malignant cells, including apoptosis, growth arrest and differentiation.^26,27

HDAC inhibitors have been shown to have anti-tumor effects in several malignancies, including, but not limited to, neuroblastoma,¹³ pancreatic cancer,³⁰ oral squamous cell carcinoma,³¹ ovarian carcinoma,³² medulloblastoma,³³ lung,³⁴ breast³⁵ and embryonal tumors.²⁹ A high level of HIF-1-α increases expression of vascular endothelial growth factor (VEGF) and platelet derived growth factor (PDGF). These molecules promote tumor angiogenesis and tumor growth.^36,37 Some HDAC inhibitors have been shown to inhibit HIF-1α (hypoxia-inducible factor) expression via a VHL-independent mechanism in RCC. Treatment of a VHL-deficient human RCC line with the HDAC inhibitor LAQ824 causes a dose-dependent inhibition of HIF-1-α.³⁸

HDAC inhibitors also have the potential for additive/synergistic effects when used with other anti-cancer agents.³⁹ Combination therapy with an HDAC inhibitor and a retinoid has been studied in several tumor types including leukemia,^40,41 cervical cancer,⁴² melanoma,⁴³ breast carcinoma,^44,45 prostate cancer,^14,46 RCC⁴⁷ and head and neck squamous carcinoma cells.⁴⁸ In the RCC study, the combination of all-trans retinoic acid and the HDAC inhibitor trichostatin A was found to cause an additive inhibition of cell proliferation in culture and in a mouse xenograft model.⁴⁷ Combination therapy with 13-cis RA and SAHA for advanced, refractory RCC is currently being evaluated in an open Phase I/II clinical trial. It is a multicenter, phase I, dose-escalation study of 13-cis RA, followed by a multicenter, phase II, prospective, non-randomized study (www.clinicaltrials.gov).

To elucidate the alterations in gene expression caused by the drug combination of SAHA and cRA which result in inhibition of cell proliferation, we investigated the responses of two human RCC cell lines to cRA and SAHA. One cell line, SKRC06, which was derived from a primary renal tumor,⁴⁹ is sensitive to growth inhibition by retinoids.⁵⁰ The second line, SKRC39, was derived from a soft tissue metastasis of RCC⁴⁹ and is retinoid-resistant.⁵⁰ Gene microarray analysis was performed to determine the alterations in gene expression that are associated with responsiveness to treatment.

Results

Responsiveness of SKRC39 and SKRC06 cells lines cRA and SAHA

To determine if SKRC39 cells were sensitive to growth inhibition by SAHA and/or CRA, cells were grown in the presence of SAHA (0.5 μM), 1 μM cRA, or both for six days. As shown in Figure 1A and B, SAHA significantly inhibited cell growth in SKRC 39 [55% of control (p < 0.001; 95% CI 23.02 to 66.98)] and SKRC06 cells [66% of control (p < 0.001; 95% CI 17.59 to 51.41)], whereas cRA only inhibited growth of SKRC06 cells [44% of control (p < 0.001; 95% CI 39.09 to 72.91)]. Combination treatment with cRA and SAHA resulted in increased growth inhibition (Fig. 1). [44% of control (p < 0.001; 95% CI 39.09 to 72.91)].

Figure 1.

Effect of 13-cis retinoic acid and/or SAHA on the proliferation of SKRC06 and SKRC39. Cells were cultured in triplicate wells for six days in the presence of the indicated drug(s). Drugs and medium were changed and replaced on day three. On day six, cells were counted and the average number of cells was compared to the average number of cells in untreated wells. Data is represented as number of cells in treated wells/number of cells in control wells × 100%. All data are from three independent experiments, and each treatment within an experiment was performed in triplicate. Cells were treated with cRA (1 μM) alone, SAHA (0.5 μM), or cRA (1 μM) + SAHA (0.5 μM). (A) upper panel SKRC06, (B) lower panel SKRC39. *p < 0.05 relative to either drug alone.

cRA and cyclic AMP

Cyclic AMP can enhance the actions of retinoids in many different tumor cell types.^58-62 Moreover, in a mouse model of RA resistant acute promyelocytic leukemia, infusions of 8-chloro-cyclic adenosine monophosphate resulted in enhanced RA induced differentiation.⁵⁸ Thus, we investigated the responsiveness of RCC cells to treatment with the differentiation agent/cAMP analog, dibutyryl cAMP (dbcAMP). Cells were plated in triplicate with either no drug, cRA (1 μM), dbcAMP (250 μM) + IBMX (100 μM) or with all three drugs in combination. IBMX (isobutylmethylxanthine) is a general cAMP phosphodiesterase inhibitor. The number of cells in dbcAMP + IBMX treated wells was 10% of the number of cells in control wells for both cell lines (p < 0.001 for both cell lines) (Fig. 2). Addition of cRA did not further increase the growth inhibition.

Figure 2.

Effect of 13-cis retinoic acid and/or dcAMP/IBMX on the proliferation of SKRC06 and SKRC39. Cells were cultured in triplicate wells for six days in the presence of the indicated drug (s). Drugs and medium were changed and replaced on day three. On day six, cells were counted and the average number of cells was compared to the average number of cells in untreated wells. Data are represented as number of cells in treated wells/number of cells in control wells × 100%. All data are from three independent experiments, and each treatment within an experiment was performed in triplicate. SKRC06 (A) and SKRC39 (B) cells were cultured for six days in the presence of: cRA (1 μM) alone; or dibutyryl cAMP (dbcAMP) (250 μM) + 3-isobutyl-1-methylxanthine (IBMX) (100 μM); cRA (1 μM) or + dbcAMP (250 μM) + IBMX (100 μM).

Retinoid-related gene expression

SKRC06 cells are sensitive to inhibition of proliferation by retinoids. Analysis of the expression values of the retinoic acid receptors (RARs) in the two cell lines showed that RARα mRNA levels were similar, RARβ mRNA levels were higher in SKRC06 cells and that RARβ mRNA was increased by 1.5-fold in SKRC06 with combination treatment (Fig. 3B) as compared to untreated SKRC06 cells. (This value is not isoform specific). RARγ mRNA levels differed more between the lines than the other RARs. Levels of RARγ mRNA in SKRC06 were 10-fold greater than in SKRC39 (Fig. 3C). RXRα mRNA levels were higher in SKRC06 by an average of approximately two-fold (Fig. 3D). RXRβ and RXRγ mRNA levels were not detectable.

Figure 3.

mRNA levels of genes at 24 hours after treatment with 1 μM cRA: microarray data, 24 hours. (A) levels of RARα mRNA in both cell lines; (B) levels of RARβ mRNA in both cell lines; (C) RARγ mRNA levels in both cell lines; (D) RXRα mRNA levels in both cell lines; (E) crabp2 mRNA levels in both cell lines; (F) Northern blot analysis of crabp2 mRNA in SKRC06 and SKRC39, 24 hour treatment; (G) lrat mRNA levels in both cell lines; (H) cyp26A1 mRNA levels; (I) cyp26B1 mRNA levels in both cell lines. The y-axis, arbitrary units; the bars indicate standard deviation. All experiments were performed three times starting with the cultured cells.

We also examined the expression of retinoid-related genes. The crabp2 mRNA levels in SKRC06 were more than ten-fold higher than the levels in SKRC39 by microarray (Fig. 3E). None of the treatments (13-cis RA, SAHA, or 13-cis RA + SAHA) altered crabp2 mRNA in SK-RC 06 cells. Northern blot analysis confirmed the higher crabp2 mRNA levels in SKRC-06 cells (Fig. 3F).

Lrat mRNA was upregulated in SKRC06 cell samples treated with cRA by two-fold, and by three-fold in samples which were treated with a combination of cRA and the HDAC inhibitor, SAHA. There was no change in the SKRC39 lrat mRNA levels regardless of treatment; basal LRAT mRNA levels in SKRC39 control were similar to basal levels in the SKRC06 control samples (Fig. 3G).

The mRNA levels of two key retinoid-metabolizing enzymes were also evaluated. Cyp26A1 (cytochrome P450, family 26, subfamily A, polypeptide 1) and cyp26B1 transcripts (cytochrome P450, family 26, subfamily B, polypeptide 1) were increased in the SKRC06 cells in response to cRA. Also, the cyp26A1 basal mRNA level was higher in untreated SKRC06 cells than in untreated retinoid-resistant SKRC39 cells (Fig. 3H). Similar results were obtained by Northern analysis and semi-quantitative PCR for cyp26A1 (not shown), and by microarray for cyp26B1 (Fig. 3I).

Gene expression in SKRC06 and SKRC39 cells treated with cRA (1 μM)

Nineteen genes show statistically significant changes (and at least three fold upregulation or downregulation) in cRA-treated SKRC06 cells (Table 1). The five most highly upregulated genes were, in order: cyp26B1, trim31, cyp26A1, cdh5 and spp1.

Table 1.

Transcripts with at least 3-fold higher levels in cells treated for 24 hours with cRA alone relative to untreated cells

(A) SKRC06 treated with cRA: transcripts changed by at least 3 fold
NCBI ID	Gene	Fold change + cRA/con	p value	Description
NM 003670	bhlbh2	3.8	0.000299	Basic heli-loop-helix domain containing, class b, 2
NM 001795	cdh5	14.8	0.000193	Cadherin 5, type 2. VE-cadherin (vascular epithelium)
NM 001807	cel	3.8	1.48e-06	Carboxyl ester lipase
NM 000783	cyp26a1	40.1	0.000198	Cytochrome p450, family 26, subfamily a, polypeptide 1
NM 019885	cyp26b1	118.6	4.13e-06	Cytochrome p450, family 26, subfamily b, polypeptide 1
NM 001465	fyb	3.6	0.00133	Fyn binding protein
NM 004751	gcnt3	4.8	0.000245	Glucosaminyl transferase, mucin-type
NM 006819	hop	0.236	0.00284	Homeodomain-only protein
NM 000598	igfbp3	3.5	0.0294	Insulin-like growth factor binding protein 3
NM 000422	krt17	0.2	0.000306	Keratin 17
NM 005564	lcn2	4.3	0.000186	Lipocalin 2 (oncogene 24p3)
NM 000582	spp1	16.9	0.00376	Secreted phosphoprotein 1
NM 005416	sprr3	0.331	0.000193	Small proline rich protein 1b
NM 001038	scnn1a	4.4	0.000245	Sodium channel, nonvoltage-gated 1 alpha
NM 003246	thbs1	0.288	0.0403	Thrombospondin 1
NM 014428	tjp3	3.4	0.0429	Tight junction protein 3 (zona occuldens)
NM 006074	trim22	3.5	0.00425	Tripartite motif-containing 22
NM 007028	trim31	52.8	8.64e-07	Tripartite motif-containing 31
NM 003810	tnfsf10	4.2	0.0138	Tumor necrosis factor (ligand) superfamily, member 10

(B) SKRC39 treated with cRA: transcripts changed by at least 3 fold
NCBI ID	Gene	Fold change	p value	Description
NM004753	Dhrs3	3.3	0.00686	Dehydrogenase reductase
NM005564	Lcn2	3.1	4.26e-5	Lipocalin 3 (oncogene 24p3)

(A) Transcripts for SKRC06 are listed in alphabetical order. The fold-change, p value and gene description are shown. (B) Transcripts for SKRC39 are listed in alphabetical order. The fold-change, p value and gene description are shown.

The function of trim31, tripartite motif-containing 31, is unknown.⁶³ Cdh5 is a member of the cadherin superfamily. Cadherin loss is implicated in several malignancies.⁶⁴ Spp1, secreted phosphoprotein 1, is a target gene of the TP53 tumor suppressor.⁶⁵ The expression patterns of the three genes with the highest fold increases in the presence of cRA were confirmed by semi-quantitative RT-PCR: cyp26B1, trim31 and cyp26A1 (Fig. 4A and B). Functional analysis of the differentially expressed genes revealed that the five most highly enriched gene clusters in SKRC06 cells treated with cRA for 24 hours were: localization; protein binding; cell fraction; cell adhesion; and apoptosis/cell death.

Figure 4.

(A and B) Semi-quantitative RT-PCR analyses of the genes cyp26A1, cyp26B1, and trim31 for validation of microarray. 24-hour treatment samples. A representative experiment for each gene is shown. A housekeeping gene, gapdh, was used as an internal control. A sample without reverse transcription, (NO RT) was used as a negative control in the PCR reactions (not shown). These experiments were repeated three times with different RNA preparations and similar results were obtained. Data are shown for (A) cyp26a1 and cyp26b1; and (B) for trim31.

In contrast, only two genes (lcn2 (lipocalin 2) and dhrs3 (dehydrogenase/reductase member 3) exhibited greater than three fold changes in mRNA levels in cRA-treated SKRC39 cells (Table 1B). The dhrs3 gene was upregulated by 3.three fold, and lcn2 was upregulated by 3.1-fold. The gene lcn2, which is also known as oncogene 24p3, also appeared in the analysis of the SKRC06 cells as a gene upregulated 4.three fold by cRA. Lcn2 is known to be upregulated by the retinoid 4-HPR and is involved in regulation of cell proliferation.^66,67 The function of dhrs3 is not well characterized; a family member, however, is known to be retinoid-inducible and downregulated in some neuroblastoma cells.⁶⁸ As the SKRC39 line is otherwise not responsive to cRA, there were insufficient genes available to perform cluster analysis.

Gene expression in SKRC39 and SKRC06 treated with the HDAC inhibitor SAHA

To evaluate the response of each cell line to treatment with 0.5 μM SAHA for 24 hours, we identified the transcripts with statistically significant alterations (at least three fold) for each line (Table 2A for SKRC06 and 2B for SKRC39).

Table 2.

Transcripts with at least three fold higher levels in cells treated with SAHA alone vs. untreated cells

(A) Transcripts changed by at least 3 fold after treatment with SAHA (alphabetical order): SKRC06
Gene	Fold change SAHA/con	p value	Description
Akap2	3.127	0.0123	A KINASE ANCHOR PROTEIN 2
Atf5	0.31	0.00634	ACTIVATING TRANSCRIPTION FACTOR 5
Adcy7	0.312	0.00338	ADENYLATE CYCLASE 7
Akrc1c3	3.996	0.00391	ALDO-KETO REDUCTASE FAMILY 1, MEMBER C3
Ccl20	10.39	5.2e-5	CHEMOKINE LIGAND 20
Cyp1b1	0.328	0.00971	CYTOCHROME-450, FAMILY 1, SUBFAMILY B, POLYPEPTIDE 1
Ef1a	4.144	0.000342	EUKARYOTIC TRANSLATION ELONGATION FACTOR1, α 2
Igsf4	3.076	0.00873	IMMUNOGLOBULIN SUPERFAMILY, MEMBER 4
Il1 α	5.224	4.59e-6	INTERLEUKIN1, α
Mbnl3	0.273	0.00851	MUSCLEBLIND-LIKE 3 (DROSOPHILA)
Phf11	0.287	5.87e-5	PHD FINGER PROTEIN 11
Plekhc1	3.922	0.00162	PLECKSTRIN HOMOLOGY DOMAIN CONTAINING, ,FAMILY C
Ryr1	0.191	0.0457	RYANODINE RECEPTOR 1
Serpinf1	0.294	0.000996	SERPIN PEPTIDASE INHIBITOR, CLADE F
Slc1a1	3.518	0.00176	SOLUTE CARRIER FAMILY 1
Srpx	7.497	9.12e-5	SUSHI-REPEAT-CONTAINING PROTEIN, X-LINKED
Trim22	0.264	0.00504	TRIPARTITE MOTIF CONTAINING 22
At3	3.004	0.00259	TUBULIN, α 3
Znf518	0.33	0.028	ZINC FINGER PROTEIN 518

(B) Transcripts changed by at least 3 fold after treatment with SAHA: SKRC39
Gene	Fold change SAHA/con	p value	Description
ube2l6	0.211	0.0411	UBIQUITIN-CONJUGATING ENZYME E2L 6
ugt	0.278	0.0381	UDP GLUCORONOSYL 1 FAMILY, POLYPEPTIDE A10
kcns3	0.139	0.0374	POTASSIUM VOLTAGE-GATED CHANNEL
ugt1a8	0.323	0.0282	UDP GLUCORONYLTRANSFERASE 1, FAMILY
hpse	0.333	0.0243	HEPARANASE
rabac1	0.313	0.0217	RAB ACCEPTOR 1 (PRENYLATED)
sc[e[1	0.170	0.0191	SERINE CARBOXYPEPTIDASE
stom	0.289	0.0119	STOMATIN
eno2	0.174	0.0114	ENOLASE 2
fgf18	3.752	0.0112	FIBROBLAST GROWTH FACTOR 18
nef1	0.134	0.0112	NEUROFILAMENT, LIGHT POLYPEPTIDE 68KDA
epas1	0.2	0.0112	ENDOTHELIAL PAS DOMAIN PROTEIN 1
gucy1b3	0.317	0.0112	GUANYLATE CYCLASE, SOLUBLE, β 3
nedd9	0.307	0.0112	NEURAL PRECURSOR CELL EXPRESSED
gm2a	0.291	0.0112	GM2 GANGLIOSIDE ACTIVATOR
adrb2	0.191	0.00901	ADRENERGIC, β 2, RECEPTOR, SURFACE
itpnc1	0.226	0.00901	PHOPHATIDLINOSITOL TRANSFER PROTEIN, CYTOPLASMIC
akr1c3	0.32	0.00901	ALDO-KETO REDUCTASE FAMILY 1, MEMBER C3
at3	0.038	0.00721	TUBULIN, α 3
gsn	0.148	0.00632	GELSOLIN (AMYLOIDOSIS, FINNISH TYPE)
procr	0.194	0.00439	PROTEIN RECEPTOR, ENDOTHELIAL
c14orf78	0.299	0.00412	CHROMOSOME 14 OPEN READING FRAME 78
lmcd1	0.22	0.00412	LIM AND CYSTEINE-RICH DOMAINS 1
man1a1	0.264	0.00368	MANNOSIDASE, α, CLASS 1A, MEMBER 1
ccl20	0.123	0.000231	CHEMOKINE (C-C MOTIF) LIGAND 20

(A) Genes with three sfold or greater changes in mRNA levels in SKRC06 are listed in alphabetical order. The fold-change, p value and gene description are shown. (B) Results for SKRC39; genes are listed by decreasing p value. The greatest fold change for a gene with increased mRNA levels after SAHA addition was fgf18.

We used DAVID⁶⁹ to identify the biological processes most associated with SAHA treatment in the two cells lines tested. Among the top five, the major biological category enriched in SKRC06 cells during treatment with 0.5 μM SAHA was a cluster representing inflammation. Two other important gene clusters were highly enriched in SKRC06: apoptosis/cell death and regulation of progression through the cell cycle.

The same analysis was performed for SKRC39. None of the highly enriched gene clusters in SKRC06 was enriched in SKRC39. In SKRC39, the most enriched clusters were related to metabolism, enzyme activity and cell structure.

A similar analysis was performed using Ingenuity Pathway Analysis. For SKRC06, the most highly represented pathways in the gene list were immune response and metabolism; the second highest score was for the pathway that included growth and proliferation of cells. For SKRC39, the best fit for the data was the pathway for development, signaling and connective tissue disorders. The second best fit was inflammation and cell growth/proliferation.

The mRNA expression in SKRC06 with combined treatment: cRA and SAHA

To identify genes which showed altered mRNA levels (at least three fold) only in samples treated with both SAHA and cRA, we compared the genes expressed in control samples to genes expressed in SAHA⁺ RA-treated SKRC06 samples and found that 72 transcripts differed by more than three fold.

Functional annotation clustering was performed using DAVID. The top five enriched clusters were protein sequencing/signaling, membrane functions, cell response, neurogenesis and apoptosis. To identify the genes which exhibited alterations in mRNA levels in the RA sensitive SKRC06 cells only in the combination treatment, we evaluated transcripts which were upregulated or downregulated by at least three fold in the combination-drug treated samples. We then eliminated genes which were upregulated or downregulated by at least three fold in retinoid-treated samples or in HDAC inhibitor-treated samples. The resulting gene list (Table 3) includes one downregulated transcript, atf5, and nine upregulated transcripts, including dhrs9, gata3, ceacam5, il1β and phlda1.

Table 3.

Transcripts changed >3-fold with treatment with cRA plus SAHA in the SKRC06 line

Gene	Description	Fold change (cRA + SAHA/control)
dhrs9	retinol-metabolizing enzyme	7.9
gata3	GATA binding protein 3	4.4
ceacam5	Adhesion molecule	4.1
Il1 β	Interleukin 1 β	4.0
pcdh1	Protocadherin family member	3.4
scap2	Immune activation factor	3.1
phlda1	Apoptotic marker carcinoma	3.1
artn	neurotophic factor	3.0
abcb1	Multidrug transporter	3.0
Atf5	Activating transcription factor 5	0.3

Genes with at least three fold higher mRNA levels in SKRC06 cells after treatment with a combination of both cRA (1 μM) and SAHA (0.5 μM). Genes are listed in order of fold-change. Gene descriptions are indicated.

Ingenuity pathway analysis was performed on this subset of genes. The most commonly appearing biological functions were cell death and cell proliferation/growth. Microarray results were confirmed using independent semi-quantitative RT-PCR experiments for selected genes: dhrs9, gata3, ceacam5, il1β and phlda1 (Table 4 for primers; Fig. 5A). Txk, a gene which regulates the expression of interferon γ, was upregulated with combined treatment in SKRC06 by five-fold compared to control. It was also upregulated by two-fold by cRA alone.

Table 4.

Primers designed for semiquantitative PCR

	Forward	Reverse	Size (bp)
ceacam5	5' CCT CAC TCT ACT CAG TGT CAC AAG 3'	5' CAG CCA AGA ATA CTG TGC AGG TGG 3'	220
IL1 β	5'GAT AAC GAG GCT TAT GTG CAC GAT G 3'	5' CAA CAC GCA GGA CAG GTA CAG ATT C 3'	243
cyp26A	5' GGC TGC AGC TGT TGA TCG AGC ACT C 3'	5' CAA CTT GTT GTC TTG ATT GCT CTT GC 3'	223
cyp26B	5' CCA ACA CGG TGT CCA ATT CCA TTG 3'	5' CTC GTG GCT GAA GAT CTT GGA GAA G 3'	77
txk	5' GAA CAC AGA TAA GCC TGA GCA CAG 3'	5' GAC TTG GAT CTT CTC TTC AGC AAC 3'	175
trim31	5' GAA GAA CGC AAT CAG GTT CAA CTC 3'	5' GAC TTG GAT CTG CTC TTG AAT CTG 3'	256

The forward and reverse sequences and expected product sizes are shown.

Figure 5.

Semi-quantitative RT-PCR analyses of selected genes cyp26A1, cyp26B1, il1β, gata3, phlda1, ceacam5, txk and gapdh for validation of microarray results: 24-hour treatment samples. A representative experiment for each gene is shown. Gapdh was used as an internal control. A sample without reverse transcription, (NO RT), was used as a negative control in the PCR reactions. These experiments were repeated three times with different RNA preparations and similar results were obtained. (A) PCR results for il1ß, gata3, phlda1, ceacam5, txk, and gapdh are shown. (B) Gata3 mRNA levels in SKRC06 and SKRC39 relative to HK-2 cells; (C) vhl3 mRNA levels in HK-2, SKRC06 and SKRC39 cells.

Changes in gata3 mRNA levels in SKRC06 and SKRC39 with combined cRA and SAHA

In SKRC06, but not SKRC39, by microarray analysis, gata3 mRNA is increases by 3.9-fold (p < 0.01; 95% CI -4.651 to -1.181) after combination treatment with SAHA and cRA for 24 hours (Table 3). 13-Cis RA monotherapy increased gata3 mRNA levels by two-fold in SKRC06 cells, which was not statistically significant. Gata3 basal mRNA levels were similar in untreated SKRC06 as in untreated SKRC39. In SKRC39 cells there were no significant changes in gata3 mRNA levels with any of the treatments.

Semi-quantitative RT-PCR confirmed the microarray findings (Fig. 5B). HK-2 cells, which are derived from normal kidney, were used to compare the levels of gata3 mRNA. By PCR, gata3 mRNA was not detectable in SKRC39. In SKRC06, however, gata3 mRNA was RA-inducible (Fig. 5B). Microarray data showed that mRNA levels were similar to those in the HK-2 normal human kidney epithelial cells. By semi-quantitative RT-PCR, vhl mRNA was lower in both SKRC06 and SKRC39 than in the normal kidney cell line HK-2 (Fig. 5C).

Gata3 protein expression in normal renal tissue and in renal cell carcinoma

To investigate further the levels of gata3 protein in clear cell renal cell carcinoma, we used immunohistochemistry to assess human tumor samples (Fig. 6). Gata3 antibody staining was positive in all normal kidney sections (20 cases were examined in total), including normal tissue adjacent to malignant tissue. Gata3 antibody staining was absent in all of the malignant renal specimens evaluated. These data indicate that a reduction in f GATA3 protein expression occurs in renal cell carcinoma. The fact that cRA and the combination of cRA and SAHA increased Gata3 mRNA levels (Fig. 5B) in an RA sensitive RCC cell line suggests that Gata3 may be a marker of RA-responsiveness. These Gata3 protein IHC data are similar to the microarray data using the RCC cell lines.

Figure 6.

Gata3 staining of human RCC. Normal and malignant human kidney sections were stained. Gata3 protein staining in a normal human kidney section is shown (A) (200X). The absence of staining for gata3 protein in a clear cell renal cell carcinoma section is demonstrated (B) (200X). This is an example; twenty different patient samples were examined, and similar staining patterns were seen.

Discussion

Inhibition of proliferation in response to 13-cis retinoic acid plus SAHA or dibutyryl cyclic AMP

SKRC06 cells were growth inhibited in response to cRA (1 μM) and/or SAHA (0.5 μM), while SKRC39 cells were resistant to proliferation inhibition by cRA (1 μM) but were sensitive to inhibition of proliferation by SAHA (0.5 μM).

Proliferation of both SKRC39 and SKRC06 cells was inhibited when treated with a combination of dibutyryl cyclic AMP and a phosphodiesterase inhibitor (Fig. 2A and B). Adenosine 3',5'-cyclic monophosphate (cAMP) is a second messenger with pro-apoptotic effects in some cells, and analogs of cyclic AMP are known to induce differentiation and apoptosis in several tumor types, including breast,⁷⁰ colon,⁷⁰ epidermoid,⁷¹ ovarian,⁷² pancreatic cancer⁷³ and retinoblastoma.⁷⁴ While we did not perform microarray analyses on dibutyryl cyclic AMP treated RCC lines, these data indicate that cyclic AMP analogs could be efficacious in RCC treatment.

Difference in responsiveness to cRA of SKRC06 and SKRC39

We found that the basal mRNA expression levels of several retinoid-related genes were higher in the cRA-responsive line, SKRC06, than in SKRC39. Crabp2 mRNA levels in the SKRC06 cell line were 10-fold higher as compared to the retinoid resistant SKRC39 cell line (Fig. 3). Cellular retinoic acid binding proteins are specific carrier proteins for all-trans retinoic acid, a member of the vitamin A (retinol) family of signaling molecules. The crabp2 protein can enhance the transcriptional activity of retinoic acid receptors and can transport RA from the cytosol to the nucleus, where it associates with RARs.⁷⁵ It has been suggested that crabp2 increases responsiveness to retinoid treatment,⁷⁶ though not all of the data in the literature are in agreement.⁷⁷ Therefore, the higher levels of this transcript in the SKRC06 line may play a role in the sensitivity of this cell line to inhibition of proliferation by cRA.

Our microarray results also reveal a 10-fold higher RARγ mRNA level in untreated SKRC06 as compared to untreated SKRC39 (Fig. 3). Overexpression of RARγ increases responsiveness to RA in SHJ-SY5Y, a neuroblastoma cell line.⁷⁸ RARγ is an important effector of the differentiation functions of RA.^79-84 In addition, activation of RARγ is known to be required for induction of apoptosis by retinoids in pancreatic adenocarcinoma cells.⁸⁵ It has recently been shown that RARγ deficiency can induce a myeloproliferative syndrome in mice.⁸⁶ Therefore, RARγ may contribute to the sensitivity of the SKRC06 line to inhibition of proliferation by cRA, and our data suggest that RARγ agonists may be a useful therapy for RCC.

SKRC06 and SKRC39 growth inhibition by SAHA

Epigenetic changes contribute to the pathogenesis and/or progression of several forms of cancer.⁸⁷ Several solid tumors have demonstrated responsiveness to HDAC inhibitor monotherapy and combination therapy.²⁵ In RCC, several HDAC inhibitors were shown to enhance TNF-related apoptosis-inducing ligand (TRAIL)/Apo-2L induced apoptosis.⁸⁸ The HDAC inhibitor MS-275 reversed retinoid resistance in RCC via histone acetylation and reexpression of RARβ2.⁸⁹ We showed previously that the HDAC inhibitor trichostatin A inhibits the proliferation of human renal cell carcinoma and increases RARβ2 mRNA levels in a RCC xenograft model in immunocompromised mice.⁴⁷ Furthermore, SAHA, the only FDA-approved HDAC inhibitor, is being evaluated in an open clinical trial as combination therapy for resistant, advanced RCC (www.clinicaltrials.gov).

Several genes identified in our microarray analysis were previously shown to be involved in HDAC-related and/or retinoid-related cell functions. These include bhlbhl2, which was upregulated in cRA treated SKRC06 cells (Table 1A). This gene is associated with growth arrest in fibroblasts via HDAC dependent and independent mechanisms after treatment with a retinoid and the HDAC inhibitor trichostatin A.⁹⁰ cRA also elevates tnfsf10 mRNA in SKRC06 cells (Table 1A); tnfsf10 causes apoptosis in some types of cancer cells.⁹¹ SAHA treatment also induced ccl20 (chemokine (C-C motif) ligand 20) mRNA expression by ~20-fold in SKRC06 cells (Table 2A). Ccl20 has been shown to be important in anti-cancer immunity.⁹² Microarray analysis of esophageal cancer cells revealed that the HDAC inhibitor FK228 alters transcripts involved in metabolism, signal transduction and regulation of cell growth.⁹³ This pattern is similar to what we found for RCC (Table 2).

Different HDAC inhibitors have differing specificity. Touma et al. showed that a very low dose (2 ng/ml) of the HDAC inhibitor trichostatin A (TSA) plus all-trans RA (1 μM) inhibited proliferation of both SKRC39 and SKRC45, a human RCC line with intermediate sensitivity to retinoid. Furthermore, all-trans RA exerts differentiating effects at significantly lower doses than cRA.⁹⁴

Combination therapy with SAHA plus 13-cis RA induces a unique set of transcripts that includes several genes known to contribute to anti-tumor effects

We identified several genes which showed increased transcript levels in SKRC06 cells treated with both drugs, but which were not significantly increased in control or in samples treated with one drug only. One of these, Dhrs9 (Table 3), is a retinol metabolizing enzyme that is deleted or mutated in several malignancies.⁹⁵ In SKRC06 cells, dhrs9 transcripts were increased in SKRC06 cells treated with both cRA and SAHA (Table 3). This pattern was not seen in the SKRC39 line. Expression of this gene is reduced in colon carcinomas with APC mutations.⁹⁶ APC deficiency has also been associated with an increased tendency to develop RCC in mice.⁹⁷

Gata3 is a transcription factor that is involved in the regulation of proliferation and differentiation of many types of epithelial cells, including breast epithelial cells⁹⁸ and has been shown to maintain breast epithelial luminal differentiation.⁹⁹ Mehra et al. have shown that in human breast carcinoma, low gata3 expression is associated with higher histologic grade, metastasis, larger tumor size and negative estrogen receptor α expression. Clinically, breast tumors with low levels of gata3 were associated with a worse outcome.¹⁰⁰

In RCC, gata3 protein levels are reduced compared to those in normal kidney.¹⁰¹ Gata3 is an important regulator of kidney morphogenesis¹⁰² and haploinsufficiency results in a distinct clinical syndrome that includes renal dysplasia.¹⁰³ We show that Gata3 protein is decreased in human renal cell carcinoma specimens compared to normal kidney (Fig. 6). The increase in Gata3 transcripts in response to cRA plus SAHA in SKRC06 cells may enhance or correlate with the differentiation of these carcinoma cells (Fig. 5).

One gene, atf5, was consistently downregulated with the combination of HDAC inhibition and retinoid treatment in SKRC06. Atf5 is known to be widely expressed in carcinomas.¹⁰⁴ Furthermore, lowering of Atf5 expression levels increases apoptosis in breast cancer cells.¹⁰⁴

In conclusion, the use of SAHA and 13-cis retinoic acid increases the expression of several genes, gene pathways and gene clusters which have important anti-cancer effects. As the current therapies for advanced RCC are not curative, new treatment regimens are urgently needed. The use of a histone deacetylase inhibitor in conjunction with a retinoid is currently being evaluated in an open Phase I/II clinical trial. Furthermore, a continued exploration of the microarray data will clarify the importance of the genes which undergo alterations in expression in response to cRA. Overexpression and/or knockout of key transcripts such as gata3, dhrs9 and trim31 may further elucidate the potential roles of these genes in cell growth inhibition and differentiation.

Materials and Methods

Cell culture and drug preparation

SKRC39 and SKRC06 (Tumor Cell Bank, Memorial Sloan Kettering Cancer Center, New York, NY) were maintained in DME medium (MP Biomedicals; Cat. No. 1033122) containing 10% ES Qualified Fetal Bovine Serum (Gibco, Invitrogen; Cat. No. 10439-024) and penicillin (10,000 U/ml)/streptomycin (10,000 U/ml)/glutamine (29.2 mg/ ml) (Gibco, Invitrogen; Cat. No. 10378-016) in a humidified tissue culture incubator at 37°C in a 10% CO₂ atmosphere. SKRC06 is derived from a RCC lesion with both clear cell and granular features. SKRC39 is derived from a clear cell renal carcinoma. Both are from primary tumors.^51,52 SKRC39 has wt VHL. HK2 cells were cultured in Epilife medium and Epilife Defined Growth Supplement (EDGS) (Invitrogen; Cat. Nos. M-EPI-500-CA, S-012-5). Medium was replaced every 2–3 days. For subculturing, SKRC06 cells were trypsinized with 0.25% Trypsin EDTA (Gibco, Invitrogen; Cat. No. 25200-056); SKRC39 cells were trypsinized with a trypsin reagent pack (Clonetics/BioWhittaker; Cat. No. CC-5034). SAHA was obtained from Aton Pharmaceuticals and diluted in 100% DMSO to a final concentration of 10 mM. SAHA was further diluted in medium to obtain a final concentration of 0.5 μM.⁵³ The phosphodiesterase inhibitor IBMX (3-isobutyl-1-methylxanthine) was obtained from Sigma-Aldrich (St. Louis, Mo.) Drug was diluted to a final concentration of 1 × 10^-6 M. All-trans retinoic acid (Product No. R-2625) and 13-cis retinoic acid (Product No. R-3255) (isotretinoin) were obtained from Sigma Aldrich, St. Louis, Mo. Both retinoids were prepared in 100% ethanol and diluted in medium to obtain a final concentration of 1 × 10^-6 M. All experiments containing light-sensitive compounds were performed in minimal light.

Cell proliferation measurements

Cells (1 × 10³ to 1 × 10⁵) were seeded in multiwell plates, depending on the experiment. Triplicate wells were plated for each time point and for each drug treatment. Viable cells were determined for each time point by direct cell counting in a particle counter (Z1 Particle Counter, Coulter-Beckman Pharmaceuticals, Fullerton, CA) and average relative cell proliferation (compared to untreated control cells) was plotted with standard error.

Microarray analysis

SKRC39 or SKRC06 cells (1 × 10⁶) were plated on 10 cm tissue culture plates (Falcon, Becton-Dickinson), and treated for 24 hours in one of the following four conditions: no drug; cRA 1 μM; SAHA 0.5 μM; or cRA 1 μM + SAHA 0.5 μM. Total cellular RNA was isolated with Trizol reagent (Invitrogen, Life Technologies, Inc.; Cat. No. 15596-018). RNA purification and subsequent steps were performed by the Genomics Core Lab at Memorial Sloan Kettering Cancer Center, New York, NY. Total RNA was purified using the Qiagen Rnaeasy kit (Valencia, CA). RNA quality was assessed by ethidium bromide agarose gel electrophoresis. cDNA was then synthesized in the presence of oligo(dT)24-T7 from Genset Corp., (La Jolla, CA). cRNA was prepared using biotinylated UTP and CTP and hybridized to Affymetrix GeneChip HG U133A_2 oligonucleotide arrays (Affymetrix Inc., Santa Clara, CA). Fluorescence was measured by laser confocal scanner (Agilent, Palo Alto, CA) and converted to signal intensity by means of Affymetrix Microarray Suite v4.0 software. Raw data files were uploaded into the NCBI GEO database. Results were imported into the software program GeneSpring GX 7.3.1 (Redwood, CA). Each indicated sample represents three independent trials. All genes were analyzed individually, although there are multiple probe sets for several genes. Probes sets with low expression (less than 100 arbitrary units of raw data) were filtered out. This filtering step eliminates genes with background expression levels that skew fold-change determinations. Probe sets that passed the first filter were further filtered to identify genes with at least a three fold difference between the samples being compared. Furthermore, probe sets that had different (p < 0.05) means (log ratio) based on the student's t-test were considered to be differentially expressed. Three independent experiments, beginning with cell culture and drug treatment, were performed.

An examination of the genes which have expression levels that differ by more than 10-fold between the two lines was performed using Ingenuity Pathway Analysis, a web-based (www.ingenuity.com) mammalian biology knowledgebase and analysis tool used for drug discovery and genomic data analysis. Functional analysis of differentially expressed genes was performed using DAVID Dennis 2003. DAVID is an acronym for Database for annotation, Visualization, and Integrated Discovery. This tool is an online program (www.david.niaid.nih.gov) that analyzes and annotates genome-scale data sets based on gene ontology. It is capable of identifying categories within results that are over-represented in the gene lists relative to the representation within the genome. The functional annotation clustering tool displays similar annotations together in clusters and DAVID calculates the chances of over-representation of the clusters using the Fisher Exact test.

Northern blots

SKRC39 or SKRC06 cells (1 × 10⁶) were plated on 10 cm tissue culture plates (Falcon, Becton-Dickinson), and treated with drugs as indicated. The cells were harvested with Trizol reagent (Invitrogen, Life Technologies, Inc.; Cat. No. 15596-018) and total cellular RNA was isolated. RNA was fractionated by size on 1% agarose/2.2 M formaldehyde gels, transferred to nylon filters (Hybond N; Amerisham Biosciences) by blotting, and cross-linked via a UV Stratalinker 1800. Filters were hybridized to [³²P] radiolabeled cDNA probes using a random primer labeling kit (Boehringer Mannheim, Indianapolis, IN) according to the manufacturer's directions in hybridization buffer containing 5X SSC, 1 mM EDTA, 10 mM phosphate buffer, 10X Denhardts solution, 0.5% SDS, 50% formamide (Fluka, cat. No. 47670), 10% dextran sulfate (Sigma, D-6001) and 1% denatured salmon sperm (Sigma, Cat. No. D1626). To remove unhybridized probe, blots were washed in low stringency wash (2X SCC, 0.1% SDS) and high stringency wash (0.2X SSC, 0.1% SDS) at 42°C and 55°C until the background was clear. Hybridized filters were subsequently exposed to Kodak BioMax films or quantitated using a phosphorimager. Northern experiments were performed three times independently, starting with fresh cells.

Semi-quantitative reverse transcription-polymerase chain reaction

The results of microarray analysis were validated with semi-quantitative RT-PCR. For RT-PCR analysis, 1 × 10⁶ SKRC39 or SKRC06 cells were plated on 10 cm tissue culture plates (Falcon, Becton-Dickinson) and treated with drugs as indicated. The cells were harvested with Trizol reagent (Invitrogen, Life Technologies, Inc.; Cat. No.15596-018) and total cellular RNA was isolated. For cDNA synthesis, total RNA (5 μg) was reverse-transcribed with Superscript II (Invitrogen; Cat. No. 18064-014) and primed with 1 μg of random hexanucleotides in a 20 μl reaction containing 10% 0.1 M DTT (Invitrogen; Part No. Y00147), 20% 5X first strand buffer (Invitrogen; Part No. Y00146), 5% RNAsin (Promega; Part No. N211A) and 5% 10 mM dNTPs (Invitrogen). The primer sequences are listed in Table 4. Additional primers were selected from published sequences: PHLDA1;⁵⁴ GATA3;⁵⁵ VHL and VEGF⁵⁶ and GAPDH.⁵⁷ Primers created in our lab were designed to span two exons. Expression of GAPDH was used for the normalization of amplification signals in different samples. The images were obtained and quantitated with a gel camera (Alpha Innotech).

Immunohistochemistry

Immunohistochemistry was performed on human patient specimens by the Department of Pathology, the New York Presbyterian Hospital, Weill Cornell Medical College of Cornell University. Briefly, tissue sections from patients with clear cell RCC (20 cases total) were subjected to immunohistochemistry. First, endogenous peroxidase activity was blocked with 3% hydrogen peroxide. Antigen retrieval was carried out in citrate buffer (10 mM, pH 6) for 15 minutes at 100°C in a microwave oven. A primary rabbit polyclonal antibody specific for GATA-3 (Abcam, Cambridge, UK) at 1:100 dilution was applied for one hour at room temperature. A subsequent reaction was performed with biotin-free HRP enzyme labeled polymer of EnVision plus a detection system (DAKO, Carpinteria, CA). A positive reaction was visualized with diaminobenzidine, solution followed by counterstaining with hematoxylin.

Statistical analysis

Data are expressed as mean +/- SD. One-way analysis of variance (ANOVA) was used to determine statistically significant differences among multiple groups. Cell proliferation assays were analyzed using a one-way ANOVA. A value of p < 0.05 was considered significant. The false discovery rate was determined for each set of genes.

Abbreviations

cAMP

adenosine 3', 5'-cyclic monophosphate

CI

confidence interval

cRA

13-cis retinoic acid

HDAC

histone deacetylase

IBMX

isobutylmethylxanthine

IHC

immunohistochemistry

RA

all-trans retinoic acid

RAR

retinoic acid receptor

RARE

retinoic acid response element

RCC

renal cell carcinoma

RXR

retinoid X receptor

SAHA

suberoylamide hydroxamic acid

TSA

trichostatin A