Extensive Sex- and/or Hormone-Dependent Expression of Rat Housekeeping Genes

Abstract

Using a commercially available qRT-PCR array designed to measure the expression of 32 rat housekeeping genes, we found that the levels of 5 of the genes were sexually dimorphic, 22 genes were over-expressed and one under-expressed in multi-hormone deficient hypophysectomized rats of both sexes. Only 3 genes fulfilled the stability criteria determined by geNorm and NormFinder as suitable housekeeping genes. Normalizing qRT-PCR data with either of these 3 genes alone, the geometric means of any 2 of the genes, or even the geometric mean of all 3 genes, produced similar results. In contrast, application of unproven housekeeping genes could lead to erroneous conclusions, having found that IGF-1 mRNA levels could be calculated as either dramatically male- or female-predominant depending upon the choice of housekeeping gene. Accordingly, it is essential to validate the constancy of housekeeping genes under every experimental condition. (This research protocol was approved by the university’s Institutional Animal Care and Use Committee.)

INTRODUCTION

Quantitative reverse transcription-polymerase chain reaction [qRT-PCR] is a highly sensitive, accurate and rapid method to measure mRNA in biological material. To control for protocol induced errors between samples, results should be normalized. Normalization is accomplished by comparing a simultaneously measured reference gene, considered requisite for cell survival, [ie, housekeeping] and whose expression is thus believed to remain constant regardless of experimental conditions, to that of the gene of interest. Identifying a suitable housekeeping gene, however, has proven more difficult than would appear. Meta-analysis of 13,629 human gene array samples revealed only a nominal number of suitable housekeeping genes using qRT-PCR [1]. Since many genes of interest have low RNA expression and may respond to experimental conditions with only small but physiologically significant changes in their levels, even minor fluctuations in a housekeeping gene could result in erroneous findings. In this regard, the use of β–actin resulted in incorrect conclusions in a study of human asthma because it was the housekeeping gene and not the target gene that was affected by the treatment [2]. In the present study, we have examined the effects of sex and/or hormone deprivation on the expression of a commercially designed qRT-PCR array of 32 classified housekeeping rat genes¹.

METHODS

Animals were housed in the University of Pennsylvania Laboratory Animal Resources facility, under the supervision of certified laboratory animal medicine veterinarians, and were treated according to a research protocol approved by the university’s Institutional Animal Care and Use Committee. Male and female Sprague-Dawley rats [Crl:CD(SD)BR] were hypophysectomized by the vendor (Charles River Laboratories, Wilmington, MA) at 8 weeks of age and observed in our facility for 5 to 6 weeks. The effectiveness of the surgery was verified by the lack of weight gain over this period and the absence of pituitaries or fragments at necropsy. Hypophysectomy was chosen as a model to examine the effects of multiple hormone deficiencies on the expression of housekeeping genes with no intention of identifying which individual or cohort of the approximate dozen affected hormones were responsible for changes in gene expression.

Total RNA was isolated using Trizol^® reagent [Life Technologies, Carlsbad, CA; http://tools.invitrogen.com/content/sfs/manuals/trizol_reagent.pdf] and purified with the Qiagen RNeasy mini kit and treated with DNase in order to remove any trace of genomic DNA [Qiagen, Valencia, CA] according to the manufacturer’s protocol [www.qiagen.com/literature/protocols/RNeasyMini.aspx]. RNA concentration and purity were determined by UV spectrophotometry [A260/280 >1.8 & A260/230 >1.7] and integrity was verified by the intensities of 28S and 18S rRNA bands on a denaturing agarose gel. cDNA synthesis was performed using the High Capacity RNA-to-cDNA kit [Life Technologies] as per instructions with appropriate no-reverse transcription [– RT] and non-template controls. [http://www3.appliedbiosystems.com/cms/groups/mcb_support/documents/generaldocuments/cms_047249.pdf] Real time PCR was performed using the ABI StepOnePlus™ System [Life Technologies] on validated TaqMan^® Array Rat Endogenous Control, Fast 96-well plate containing 32 genes, plated in triplicate [Life Technologies] according to the vendor’s instructions [http://www3.appliedbiosystems.com/cms/groups/mcb_support/documents/generaldocuments/cms_053406.pdf]. IGF-1 mRNA was also measured by a TaqMan^® Assay (Rn00710306_m1). A combined study was designed and analyzed for comparative evaluation of the threshold cycle (Ct), each whole digit change in the Ct value indicating a 2-fold change in expression levels, for each gene across all the samples using the ABI StepOnePlus™ Real-time PCR software as per recommended protocol [http://www3.appliedbiosystems.com/cms/groups/mcb_support/documents/generaldocuments/cms_059679.pdf]. Ct values were subjected to two-way analysis of variance [α=0.05] demonstrating statistically significant interactions between groups, and differences were determined with t statistics and the Bonferroni procedure for multiple comparisons. Expression of those housekeeping genes found to be unaltered by sex and/or hormone deprivation were further evaluated for stability by both geNorm [3] and NormFinder [4] analyses.

RESULTS

Five housekeeping genes measured in rat liver expressed sexually dimorphic levels [Table 1]. 18S RNA as well as Tfrc and Ubc mRNA levels were greater in males whereas Arbp and Cdkn1a mRNA levels were greater in females. Expression levels of 23 housekeeping genes were altered by hypophysectomy in females while expression of 22 of the same genes were similarly affected by hypophysectomy in males [Table 1]. Eight housekeeping genes [Gadd45, Gapdh, Gusb, Hmbs, PES1, Ppia, Ppib, Psmc4] were unchanged by sex or hypophysectomy. Whereas MT-ATP mRNA concentrations appeared to be sex- and hypophysectomy dependent, the standard deviations for the intact animal groups remained unusually large after several determinations resulting in nonsignificant statistical comparisons. With the sole exception of 18S, all of the housekeeping genes responsive to hypophysectomy were over-expressed; more often than not at 2 or more-fold higher concentrations than in the same sex intact rats [Table 1]. Although it is not clear why hypophysectomy resulted in the over-expression of 21 to 22 housekeeping genes, it is possible that the affected genes are normally down regulated by one or more of the now deficient hormones in the pituitary ablated rats.

Table 1.

Sex- and Hormone-Dependent Expression of Rat Hepatic Housekeeping Genes

Gene symbol		Intact
		Male	Female
18S		11.35 ±0.51	12.15 ±0.29*
Arbp		26.35 ±0.26	25.69 ±0.21*
Cdkn1a		34.16 ±0.38	31.65 ±0.31**
Tfrc		29.49 ±0.36	30.37 ±0.33*
Ubc		22.01 ±0.35	22.94 ±0.38*
	Male		Female
	Intact	Hypox	Intact	Hypox
18S	11.35 ±0.51	13.19 ±0.26**	12.15 ±0.29	13.70 ±0.33**
Abl1	30.22 ±0.43	29.34 ±0.24*	30.49 ±0.29	29.48 ±0.16**
Actb	24.64 ±0.26	23.81 ±0.22**	24.66 ±0.28	23.84 ±0.17**
Arbp	26.35 ±0.25	25.22 ±0.32**	25.69 ±0.21	25.15 ±0.22*
B2m	22.70 ±0.39	21.45 ±0.21**	22.61 ±0.16	21.74 ±0.31**
Casc3	30.05 ±0.34	29.06 ±0.15**	30.20 ±0.27	29.25 ±0.23**
Cdkn1a	34.16 ±0.38	31.09 ±0.32**	31.65 ±0.31	28.60 ±0.29**
Cdkn1b	30.72 ±0.20	29.08 ±0.21**	30.45 ±0.39	29.11 ±0.21**
Eif2b1	30.01 ±0.40	31.07 ±0.17**	31.34 ±0.10	29.95 ±0.11*
Elf1	30.49 ±0.48	28.92 ±0.31**	30.47 ±0.38	29.08 ±0.29**
Hprt1	27.89 ±0.28	26.69 ±0.23**	27.83 ±0.32	26.50 ±0.23**
Mrpl19	30.91 ±0.40	29.80 ±0.21**	30.80 ±0.35	29.80 ±0.18**
Pgk1	27.12 ±0.38	26.04 ±0.20**	26.80 ±0.31	25.90 ±0.36*
Pop4	32.00 ±0.29	31.22 ±0.27*	32.26 ±0.33	31.57 ±0.17*
Pum1	28.57 ±0.36	27.81 ±0.20*	28.65 ±0.36	27.61 ±0.22**
Rpl30	24.46 ±0.29	23.78 ±0.18*	24.31 ±0.12	23.73 ±0.23**
Rpl37a	33.80 ±0.36	32.87 ±0.28*	33.70 ±0.24	33.16 ±0.21*
Rplp2	25.09 ±0.40	23.87 ±0.21**	24.96 ±0.20	23.83 ±0.26**
Rps17	24.78 ±0.39	23.82 ±0.30*	24.64 ±0.25	23.87 ±0.18**
Tbp	32.54 ±0.38	31.12 ±0.21**	32.54 ±0.25	31.25 ±0.18**
Tfrc	29.49 ±0.36	28.76 ±0.22*	30.37 ±0.33	28.77 ±0.22**
Ubc	22.01 ±0.35	21.84± 0.13NS	22.94 ±0.38	21.63 ±0.22**
Ywhaz	28.59 ±0.27	27.53 ±0.28**	28.71 ±0.22	27.59 ±0.21**

The expression of rat housekeeping genes were determined by qRT-PCR [TaqMan array] in liver of intact [n=5] and hypophysectomize, Hypox, [n=3] adult male and female rats. Threshold cycle values are presented as mean ±sd.

^*P<0.05,

^**P<0.01 [when comparing intact male to intact female rats [upper portion of the table] or when comparing intact to Hypox rats of the same sex. NS, not significant.

Further stability assessment of the 8 sex and hormone-independent genes by geNorm and NormFinder indicated both agreements and variations in their order of ranking according to the type of analysis used and the groups compared. The geNorm algorithm calculates a stability value (M); the lower the M value, the more stable the gene expression, with less than optimal M values requiring the inclusion of more than one housekeeping gene in a qRT-PCR study [3]. Accordingly, when comparing intact males to hypophysectomized males, only Psmc4, Ppia and Gusb met the stability value criteria [Fig. 1]. Comparing intact females to hypophysectomized females, Hmbs, Psmc4, Ppia, PES1 and Gapdh had acceptable M values. When we compared all males (ie, both intact and hypophysectomized) to all females, Psmc4, Ppia, Hmbs, PES1 and Gapdh met the stability value criteria. However, when we determined the stability values by comparing intact males with intact females, only Eif2b1, Psmc4, Ywhaz, Ppia and Actb had satisfactory M values.

Fig. 1.

geNorm analysis of the candidate reference genes. Results are presented according to the output file of the geNorm program [3]. Average expression stability values (M) of the 8 most stable housekeeping genes for each treatment comparison and their associated ranking from least to most stable expression are presented. Hypox, hypophysectomy.

Whereas, the numerical rankings of gene stability by NormFinder [4] varied somewhat according to treatment comparisons, Psmc4, Ppia, PES1 and Ppib performed best across all groups [Table 2]. Furthermore, Psmc4, Ppia and PES1, in descending order of stability, were found to be the most stable housekeeping genes in all comparison groups when evaluated by both geNorm and NormFinder.

Table 2.

Candidate reference genes for normalization of qRT-PCR listed according to their expression stability calculated by the NormFinder program [4].

	Intact males vs Intact females		Intact males vs Hypoxa males		Intact females vs Hypoxa females		All males vs All females
Ranking Order	Gene Name	Stability Value	Gene Name	Stability Value	Gene Name	Stability Value	Gene Name	Stability Value
1	Cdkn1b	0.063	Ppia	0.035	Hmbs	0.047	Gusb	0.045
2	Psmc4	0.064	Psmc4	0.045	Psmc4	0.106	Psmc4	0.052
3	Ppib	0.066	Gusb	0.054	Gusb	0.113	Ppia	0.053
4	B2m	0.066	PES1	0.058	Ppia	0.115	Hmbs	0.056
5	PES1	0.068	Gapdh	0.064	Gapdh	0.120	Gapdh	0.083
6	Ppia	0.076	Hmbs	0.072	Ppib	0.132	PES1	0.084
7	Hprt1	0.077	Ppib	0.075	PES1	0.133	Ppib	0.085
8	Pum1	0.079	Gadd45a	0.164	Gadd45a	0.173	Gadd45a	0.165

^aHypox, hypophysectomized

The stability of the 3 “best” choice housekeeping genes has been graphically characterized by their threshold cycles [Fig. 2]. Accordingly, hepatic expression levels of Psmc4, Ppia and PES1 were found to be extremely stable in, and between intact males and females as well as hypophysectomized males and females. The variation of each housekeeping gene between the 4 treatment groups was further analyzed by calculating the standard deviation as a percentage of its mean. In all cases, the standard deviation for each gene varied about 1% or less between all treatment groups.

Fig. 2.

Data presenting the threshold cycles (Ct) of the three most stable hepatic housekeeping genes [Psmc4, Ppia and PES1] in intact (n=5) or hypophysectomized, hypox (n=3) male and female rats. Values are means± sd.

To demonstrate the importance of choosing a stable housekeeping gene, we measured IGF-1 in rat liver by qRT-PCR and normalized the results with Psmc4, a sex-independent gene; Tfrc, a male-predominant gene; and Arbp, a female-predominant gene. We have expressed the relative levels of IGF-1 mRNA in female liver as a percent of IGF-1 mRNA measured in male liver arbitrarily designated as 100% for all comparisons. Using Psmc4, IGF-1 mRNA in females was 64±4.7% (P<0.01); using Tfrc, IGF-1 mRNA was 131±7.5% (P<0.01) in female liver; and IGF-1 mRNA normalized with Arbp was 48±7.1% (P<0.01) in female liver.

It has been suggested that using the geometric means of 2 or 3 of the most stable housekeeping genes rather than choosing just one of these genes would be a more accurate method to normalize qRT-PCR data [3, 4]. Again, arbitrarily designating male hepatic IGF-1 mRNA as 100%, we calculated IGF-1 mRNA levels in female liver to be 64± 4.7% (P<0.01), 72± 8.2% (P<0.05) and 69± 5.1% (P<0.01) when normalizing with either Psmc4, Ppia or PES1, respectively. Expression levels of IGF-1 mRNA in female rat liver were 67± 2.5% (P<0.01), 71± 3.4% (P<0.01) and 69± 3.4% (P<0.01) when normalized against the geometric means of Psmc4 and PES1, Ppia and PES1, and Psmc4 and Ppia, respectively. Lastly, when the geometric mean of all 3 stable housekeeping genes [Psmc4, Ppia and PES1] was used, the female level of IGF-1 mRNA was 61± 3.1% (P<0.01).

DISCUSSION AND CONCLUSION

The validity of many previously acceptable housekeeping genes has been questioned in studies using a wide variety of different tissues under varying experimental conditions. Studies using qRT-PCR to examine gene expression in fibroblasts exposed to serum factors [5], leukocytes from bacteria infected individuals [6], liver exposed to carcinogens [7], retina exposed to hyperoxia [8], brain following dietary restriction [9] or ischemia [10] and virus infected cell lines [11] have all reported inductions, often up to 20- to 30-fold of many, if not nearly all, of the supposedly invariable housekeeping genes in the affected tissues. Here, we add to these finding by demonstrating sex- and/or multi-hormone-regulated expression of 23 of 32 housekeeping genes commercially identified in a qRT-PCR rat array endogenous control kit. Further analyses of gene stability by geNorm and NormFinder identified only 3, uncommonly used genes [Psmc4, Ppia and PES1] whose stability values rendered them suitable housekeeping genes in sex and/or hormone-dependent rat studies. Normalizing data with either of these 3 genes alone, or any combinations of their geometric means, produced similar results, indicating that the identifications of just one acceptably stable housekeeping gene would be sufficient for normalizing qRT-PCR data.

Whereas housekeeping genes are defined by there invariant presence, it does not necessarily mean that their expression is unregulated. In fact, not a single housekeeping gene has been accorded the designation as the universal or gold standard. The identification of a housekeeping gene[s] is likely to vary depending upon methodologies, detection dyes (unpublished observation) as well as the particular experimental protocol and the physiologic state of the subjects, i.e., sex, age, weight. In the case of methodologies, for example, 18S RNA has been consistently used as a sex-independent housekeeping gene in semi-quantitative northern blotting [12–14]. Perhaps, more illustrative, is our own previous observation of greater expression levels of 18S RNA as well as Gapdh in female rats of similar age, strain and treatment as reported here [15]. This seeming contradiction, however, can be explained by our earlier use of the much less sensitive semi-quantitative PCR. However, when the experimental protocol induces more substantial effects, such as those produced by hypophysectomy, both methods (ie, semi-quantitative and qRT-PCR) show similar reductions in 18S RNA levels. Accordingly, a validation for each situation, on an individual basis, is absolutely necessary.