The aim of this study was to explore the underlying mechanistic pathways which may explain the development of metabolic disease following in utero
exposure to iron deficiency in a well-established rat model. We have previously suggested that a limited number of gatekeeper processes could explain the common phenotype which manifests in offspring of both prenatal iron and protein restriction in two different strains of rat 
. However, the current study is much more specific and aimed to establish if additional, or alternative, diet-specific pathway responses occur with iron deficiency alone in RHL rats. Such responses may be associated with the reduced nephron endowment and increase in blood pressure previously found in this cohort 
compared to the offspring of iron-replete controls, although this study did not set out to establish definitive causality.
This study was well powered to confidently identify differences in gene and protein expression between groups, and in excess of 2500 embryonic genes were found to be differentially expressed with maternal iron deficiency compared to exposure to a control diet. This offered a much greater and varied pool of genes to work with than the original gatekeeper study, which only considered the 153 gene changes reflected by both diets, or both strains. Pathway analysis of microarray data allows this mass of data to be refined so that the specific processes
affected by maternal iron deficiency can be identified. In the previous gatekeeper study 
, aspects of cell cycle regulation represented four of the seven most significant pathways. It is noteworthy that the pathway most significantly impacted by iron deficiency in RHL embryos was the initiation of mitosis. This to some extent confirms the findings of our earlier study 
. Specific genes affected in this pathway included cdk1, cdk7, Cyclin B1, Cyclin B2 and Cyclin H.
Cdk1 and cyclin B form a complex initiating the onset of mitosis, shuttling back and forth between the nucleus and cytoplasm. Their gene expression, the initiation of mitosis and nucleocytoplasmic transport of cdk and cyclins were significantly affected by iron deficiency. Phosphorylation of the BH3-only protein BAD was the second most impacted process with iron deficiency. This is important during development when neuron numbers are controlled to ensure the correct architecture of the nervous system (Konishi et al, 2002). Iron deficiency specifically affected the WNT signalling pathway’s role in development. During embryogenesis, WNT proteins are involved in regulation of cell fate and patterning.
The two genes demonstrating the greatest up-regulation with iron deficiency were SOX4
and Map1b. SOX4
is a transcription factor involved in the regulation of embryonic development and determination of cell fate. The protein may function in the apoptosis pathway leading to cell death and tumorigenesis. Map1b
is involved in microtubule assembly, and plays an important role in development and function of the nervous system 
. The two most down-regulated genes with iron deficiency were Myo5c
is involved in transferrin trafficking, and therefore plays an essential role in iron uptake and the regulation of cell proliferation. It is also likely to power actin-based membrane trafficking in a number of tissues. S100 proteins are involved in the regulation of cellular processes such as cell cycle progression and differentiation. S100a6
may indirectly play a role in the reorganization of the actin cytoskeleton and in cell motility 
. It was not surprising that a number of other iron metabolism genes were also down-regulated, to a lesser extent (Table S1
; e.g. pirin, calreticulin, ferric-chelate reductase). Maternal iron deficiency may be expected to have a major impact upon all iron-regulated pathways in the embryo. However, we observed no change in expression of the main iron storage and transport proteins (transferrin, ferritin, transferrin receptor, hepcidin). This may indicate that the programming effects of maternal iron deficiency may not be solely or simply mediated by a gross reduction in iron supply to the developing embryo or fetus. Other mechanisms such as endocrine imbalance across the placenta and resetting of epigenetic marks are known to be involved in programming responses to undernutrition 
The iron requirements of the day 13 embryo are likely to be small in comparison to that of the mother and may be largely met, even in the face of maternal deficiency, which may explain why no change in expression of the transport proteins was observed. Previous work by the authors determined that during pregnancy the maintenance of iron stores are prioritised towards fetal needs at the expense of the mother. It was demonstrated that despite a significant decrease in maternal liver iron content from day 0 of pregnancy in iron-deficient dams, hematocrit (Hct) levels were maintained throughout the first half of pregnancy, falling by day 21 
. Fetal liver iron and Hct levels measured at day 21.5 mirrored maternal concentration at the same point of gestation, i.e. they were decreased with iron-deficiency. At day 21.5 of gestation, placental and maternal liver transferrin receptor (TfR) expression was increased with iron-deficiency. An elevation in placental TfR was also found in iron-deficient day 20 placenta (FC: 3.21, P<0.03; unpublished data). Fetal liver TfR expression was unchanged by maternal iron deficiency throughout pregnancy 
, while TfR2 expression was decreased. In this study, day 13 embryos were too early in development to measure equivalent changes. Fetal transferrin expression is low in early-mid gestation and does not begin to increase until around day 18 
The proteomics analysis identified only a limited number of proteins which were differentially expressed with prenatal iron restriction. This is partially due to sensitivity and methodology issues, as the separation of the proteins are limited by the pI range and size of gel. It may also be due to post translational modifications allowing a range of spots for a protein, therefore diluting the potentially significant changes in expression. Despite these limitations, there was a remarkable similarity between the microarray and proteomic results in terms of the processes and pathways affected. These proteins could be broadly categorised by function, including cytoskeletal remodelling, cell proliferation and the proteasome complex, which were also identified by the earlier gatekeeper protein analysis 
. However, the actual proteins with significantly differing expression altered between the two studies. For example, actin-related protein 3 and tubulin α-1 chain were key gatekeeper proteins associated with cytoskeletal functions, whereas ADP-ribosylation factor-like 3, dihydropyrimidinase-related protein 2 and chaperonin containing TCP1 were identified in this role with iron deficiency. Only SUG1, a subunit of the proteasome complex, featured as a significantly affected protein in both studies. This emphasises the importance of the complex, which is related to metabolic regulation and cell cycle progression, in nutritional studies.
A small number of proteins were also significantly differentially regulated in the same direction at the gene level as shown by the microarray. Further processes which the proteomics highlighted in this study were protein folding, unfolding and transport. Failure of normal folding, accumulation of denatured proteins or failure of the proteolytic machinery of a cell can lead to a build up of potentially damaging polypeptides which could cause cellular dysfunction or trigger apoptosis. The proteasome complex and molecular chaperones function together as a quality-control system to selectively eliminate abnormal proteins. A number of chaperone proteins that were down-regulated following exposure to a prenatal FeD diet are involved in folding of cytoskeletal components (chaperonin) and other targets of interest such as p53 (nucleophosmin).
As this study used whole embryos, the gene and protein changes noted are expressed within a heterogenous cell population. Therefore a limitation of this study is that it cannot be concluded that impacted processes are related specifically to development of any specific organs, systems or tissues. Further work will be needed to isolate the location of the key genes and proteins affected by iron deficiency in tissues of interest, such as the kidney. The dilution effect conferred by using whole embryos may be the reason for the modest fold-change values found in this study, and for generalised processes such as cell division being the most highly impacted by the dietary insult, rather than tissue-specific effects. As considered in our previous study 
, criteria concerning the validation of microarray techniques are still being addressed by investigators in the field 
. However, we are satisfied that this study was adequately powered and controlled to reveal only robust results. Importantly these findings corroborate earlier observations as well as changes in protein expression in the current study.
Iron deficiency is the greatest micronutrient deficiency among humans, impacting on pregnant women in both developed and developing countries. Iron is essential for a variety of metabolic processes, and in the embryo clearly plays a key role in cellular proliferation and regulating cell-cycle proteins. This thorough study has shown that maternal iron deficiency impacts significantly on genes and proteins which regulate cell proliferation, differentiation and apoptosis in the embryo. These findings may provide important indicators of the primary mechanisms which link fetal exposure to maternal undernutrition to the development of cardiovascular, renal and metabolic disorders later in life.