This study demonstrates the utility of toxicogenomic approaches to detect evidence of testicular injury in sperm. The transcript changes initially identified by microarray analysis were confirmed using qRT-PCR arrays in two additional toxicant exposure paradigms. A time course experiment was performed to validate the alterations in sperm transcripts and to examine the time dependence of the HD effect. Over the time course of HD exposure and post-exposure recovery, we observed dynamic changes in sperm transcript content. Significant differences in steady state transcript levels for 12 of the 29 candidates were identified at various time points, with the most robust response occurring in the post-exposure recovery period. This was an unexpected finding because the microarray data suggested that the phenotypic and transcriptomic indicators of injury resolved 3 months after exposure cessation. The differences between the microarray and qRT-PCR recovery data are most likely due to differences in the sensitivity of the platforms, with the qRT-PCR arrays being more robust than the microarrays 
Heirarchical clustering suggested a time-dependence to the transcript changes. The temporal dependence may result from effects to distinct germ cell populations during spermatogenesis, with later changing transcripts altered in spermatogonia, which have a long maturation sequence to undergo, and earlier changing transcripts disrupted in spermatids or spermatocytes, which take less time to mature into sperm. Sertoli cells support developing germ cells by creating a nurturing microenvironment that facilitates spermatogenesis. Sertoli cell toxicants alter this microenvironment and disrupt spermatogenesis. Our data suggests that exposed sperm contain abnormal transcript content. It is known that the pool of RNAs in mature spermatozoa represent a significant proportion of the RNAs synthesized prior to transcriptional arrest 
, so it is possible that these transcripts are leftover cell survival signals during early spermatogenesis. It is also possible that the regulation of these transcripts was purposefully altered to prepare sperm for embryogenesis in a stressful environment. On the other hand, the Sertoli cells or epididymal epithelial cells may be transferring transcripts to the sperm directly, and toxicant exposure could affect these processes. It has been previously hypothesized that sperm can take up foreign mRNAs, and Clu
, a putative Sertoli cell mRNA, may be one of these transcripts 
. Future work will investigate whether the epididymal epithelial cells or Sertoli cells directly provide the developing sperm with transcripts, or if the germ cells alter their own transcript content during the early stages of spermatogenesis in response to environmental changes.
The panel of sperm mRNA indicators identified by HD exposure was tested with another Sertoli cell toxicant, CBZ. No changes in body or reproductive weights were observed after low dose sub-chronic exposure to CBZ, but histology (an increase in RSH) and serum analysis (a decrease in inhibin B) provided evidence of subtle testicular injury. The HD-generated qRT-PCR array transcript panel was predictive of testicular injury resulting from CBZ exposure, with 10 of the 29 transcripts altered.
Although HD and CBZ both target the same cell-type within the testis (the Sertoli cell), the qRT-PCR data identified toxicant-specific alterations in the sperm transcriptome (). This may be explained by the opposing actions of these two toxicants on the Sertoli cell microtubules, and highlights the potential of the qRT-PCR array panel to detect specific transcriptional signatures for different toxicants. HD promotes and stabilizes microtubule assembly by cross-linking tubulin, while CBZ inhibits microtubule assembly by binding the β-tubulin subunit of the αβ-tubulin heterodimer 
. Even though the toxicant mechanisms of actions differ, they ultimately produce similar phenotypic alterations in the seminiferous epithelium, including sloughing and RSH 
. In this study, RSH were increased 14.5-fold by HD and 3.4-fold by CBZ. Our results suggest that RSH is an appropriate phenotypic anchor for determining Sertoli cell toxicity due to low dose sub-chronic exposures. However, the comparison of the RSH for the two toxicants highlights the fact that the doses we selected for each toxicant were not equipotent. The differences in the severity of the injury may also explain why the two toxicants produced dissimilar transcript profiles. Additional low-dose HD experiments are now underway in our laboratory to test this hypothesis.
Venn diagram relating transcript profiles for HD and CBZ.
Exposure to low dose HD induced alterations in sperm transcripts associated with stress response and apoptosis inhibition, possibly reflecting adaptive mechanisms invoked to maintain homeostasis in the developing germ cells. For example, SOD3 is an extracellular superoxide dismutase responsible for the removal of reactive oxygen species, and recent studies have suggested that SOD3 has an important role in regulating cellular signaling networks that reduce the development of injury and apoptosis 
. In addition, DNAJB4 is a member of the DNAJ/HSP40 protein family, which participate in many cellular processes, including protein translation, folding, unfolding, translocation, and degradation; these proteins stimulate ATPase activation of heat shock 70 chaperone proteins to protect cells from stress 
. This hypothesis is consistent with our histopathology results that showed no changes in GC apoptosis after the 3-month exposure to HD.
Low dose CBZ exposure altered sperm transcripts associated with cell junctions and apoptosis. At higher doses, CBZ induces germ cell apoptosis in the testis and increases sloughing, which is the premature release of the germ cell from the Sertoli cell crypts. ABI2 is important for dynamic actin cytoskeleton remodeling at adherens junctions, which are important for Sertoli cell-germ cell interactions, including the movement of germ cells from the basement membrane to the lumen of the seminiferous tubules during spermatogenesis 
. BAG1 plays many roles in promoting cell survival and can interact with proteosomes and heat shock proteins to prevent cell death 
. Our data showed an induction of these two transcripts, suggesting these mRNAs may reflect a germ cell response to prevent injury and apoptosis.
The qRT-PCR studies identified 4 transcripts (Clu
, and Ift81
) that were found in sperm from both HD and CBZ exposed rats associated with functions important for stress response and spermatogenesis. CLU is a glycoprotein involved in many biological processes, including protecting cells from injury, mediating apoptosis, and influencing the differentiation and maturation of germ cells 
mRNA has previously been detected by qRT-PCR in porcine spermatozoa, and sperm are hypothesized to deliver the Clu
mRNA transcript to the oocyte to support its subsequent development after fertilization 
. PTGDS catalyzes conversion of prostaglandin H2 to prostaglandin D2, a major prostaglandin that regulates many bodily functions including sleep, body temperature, hormone release, and odor responses 
mRNA has been measured in Sertoli cells and germ cells, and in the epididymis 
, where the protein product is an important component in the seminal fluid and may have a role in fertilization 
. SIL1 is an adenine nucleotide exchange factor for the heat-shock protein 70 member HSPA5/BiP 
. Defects in this gene have been implicated in Marinesco–Sjögren syndrome, which commonly presents with hypogonadism 
. Heat-shock proteins are important for proper gametogenesis and embryogenesis, and induction of heat-shock protein-related mRNAs in sperm due to environmental stress could benefit the embryo after fertilization 
. IFT81, also known as CDV-1, is necessary for the assembly and maintenance of eukaryotic cilia and flagella 
mRNA is predominantly expressed in the testis with expression increasing with male sex maturation and onset of spermatogenesis 
. In addition, Cdv-1R
mRNA has been localized to the epididymis and may play an important role in sperm maturation 
. Interestingly, the direction of change for all 4 transcripts differed between the two toxicants, and this may be due to the opposing actions of HD and CBZ on the microtubules within the Sertoli cell. Future research in our laboratory will further characterize these mRNA transcripts using comprehensive dose response studies and additional testicular toxicants.
Eighteen of the 29 sperm transcripts identified by microarrays as indicators of testicular toxicity were verified by subsequent qRT-PCR array analysis. We were unable to observe statistically significant changes for the other 11 sperm transcripts identified by microarray, and this may be due to the smaller sample size for the qRT-PCR studies (n
4–9 compared to the ~18–20). The 11 transcripts that did not change on the qRT-PCR array could also have been false positives on the microarray.
We have used a novel molecular based approach to identify sperm transcript changes after exposure to well characterized testicular toxicants. Given the novelty of this observation, we do not yet know whether these transcript alterations are shared manifestations of testicular injury in general across multiple species, or are chemical-specific in the rat. This study provides a proof-of-principle for the development of sperm indicators of testicular injury and suggests that measuring sperm mRNAs is a promising approach for screening toxicant-induced testicular injury. These data build upon our existing knowledge of testicular toxicity in animal models and develops the foundation required to extrapolate these observations to additional testicular toxicants and across species.