- Top of page
- Materials and methods
We have identified and cataloged 54 genes that exhibit predominant expression in the striatum. Our hypothesis is that such mRNA molecules are likely to encode proteins that are preferentially associated with particular physiological processes intrinsic to striatal neurons, and therefore might contribute to the regional specificity of neurodegeneration observed in striatal disorders such as Huntington's disease (HD). Expression of these genes was measured simultaneously in the striatum of HD R6/1 transgenic mice using Affymetrix oligonucleotide arrays. We found a decrease in expression of 81% of striatum-enriched genes in HD transgenic mice. Changes in expression of genes associated with G-protein signaling and calcium homeostasis were highlighted. The most striking decrement was observed for a newly identified subunit of the sodium channel, beta 4, with dramatic decreases in expression beginning at 8 weeks of age. A subset of striatal genes was tested by real-time PCR in caudate samples from human HD patients. Similar alterations in expression were observed in human HD and the R6/1 model for the striatal genes tested. Expression of 15 of the striatum-enriched genes was measured in 6-hydroxydopamine-lesioned rats to determine their dependence on dopamine innervation. No changes in expression were observed for any of these genes. These findings demonstrate that mutant huntingtin protein causes selective deficits in the expression of mRNAs responsible for striatum-specific physiology and these may contribute to the regional specificity of degeneration observed in HD.
The striatum (consisting of the caudate nucleus and putamen) is the largest and major receptive component of the basal ganglia, a group of subcortical nuclei that include the substantia nigra, globus pallidus and the subthalamic nucleus. The basal ganglia act via multiple intrinsic and extrinsic circuits to control motor and cognitive functions (Parent 1990; Graybiel 1995). The striatum exhibits complex patterns of neurochemical and neuroanatomical connectivity and compartmentalization, and is critically involved in the generation of directed motor behaviors through highly specialized output pathways (Graybiel 1995). Proper dynamic regulation of cellular signaling in the striatum is the key to its specific functions. Disruption of this complex balance can have detrimental consequences, as observed in Huntington's disease (HD) and Parkinson's disease (PD), two devastating neurodegenerative disorders affecting intrinsic striatal neurons, and the dopaminergic input to the striatum, respectively.
HD is a progressive neurodegenerative disorder characterized by motor, psychiatric, and cognitive disturbances (Ross et al. 1997). HD is caused by an abnormal expansion of a CAG trinucleotide repeat region in exon 1 of the HD gene (IT15) resulting in an abnormally long stretch of glutamine residues in the encoded huntingtin protein (Huntington's Disease Collaborative Research Group 1993). Huntingtin, a 3144 amino acid protein with no strong homology to other known proteins, exhibits widespread expression in both brain and peripheral tissues (DiFiglia et al. 1995). A core pathological feature of HD is severe atrophy and neurodegeneration of the striatum, and to a lesser extent, cortex (Vonsattel et al. 1985). However, how the disease protein causes specific neurodegeneration in the striatum, despite its ubiquitous expression, is unknown. Several transgenic mouse models for HD have been generated with the mutant HD gene. The first to be developed were the R6/1 and R6/2 mice expressing exon 1 of the human HD gene with an expanded number of CAG trinucleotide repeats: 144 CAG repeats for the R6/2 line and 116 CAG repeats for the R6/1 line (Mangiarini et al. 1996). The R6/1 transgenic mice, which have a later age of motor impairment onset and progression of symptoms (> 20 weeks) compared with the R6/2 line (9–11 weeks) (Mangiarini et al. 1996), are a useful model to study adult-onset HD, which represents a majority of cases.
Our hypothesis is that mRNAs with restricted expression in the striatum are likely to encode proteins that are preferentially associated with particular physiological or behavioral processes in the striatum compared with molecules with more generalized patterns of expression. Hence, their functional roles might be especially relevant to pathological conditions of the striatum and may account for the regional specificity of neurodegeneration observed in HD. Therefore, we have identified and cataloged a list of 54 striatal-enriched genes and have examined their expression in rodent models of Huntington's and Parkinson's diseases. Simultaneous measurement of expression of all 54 genes in mice was possible with the use of Affymetrix oligonucleotide microarrays. We found a decrease in expression of 81% of striatum-enriched genes in HD transgenic mice and similar decreases when caudate from human HD patients was examined. In contrast, 6-hydroxydopamine (6-OHDA) lesioning of the nigrostriatal pathway in rats resulted in no differences in expression in any of the striatal-enriched genes we analyzed, indicating independence from dopamine innervation. These results demonstrate that the presence of the mutant huntingtin protein causes selective deficits in the expression of mRNAs responsible for striatum-specific physiology.
- Top of page
- Materials and methods
We have compiled a list of 54 genes that exhibit predominant expression in the striatum (Table 1). We designated these genes as ‘striatal enriched’ as they exhibit predominant expression in the striatum and related areas (nucleus accumbens, olfactory tubercle and islands of Calleja), with some genes being expressed at lower levels in additional regions (usually cortex, hippocampus or thalamus). These represent sequences previously published from our laboratory, genes previously described in the literature and unpublished sequences we recently identified in a mouse brain anatomy (MBA) expression study using the method TOGA® (Sutcliffe et al. 2000) (Table 1).
To measure simultaneously the expression of these striatal genes in HD transgenic mice, we used Affymetrix oligonucleotide array analysis, focusing on those probe sets representing the striatum genes on our list. The Affymetrix mouse 430 2.0 chip, which contains 39 000 mouse nucleotide sequences, contained probe sets for all but two of the striatal enriched sequences on our list, one novel striatal sequence, MBA55 and Nolz-1. Arrays were hybridized with three biological replicates for each condition: striata from R6/1 and wild-type littermate controls at two different stages of illness: presymptomatic (10 weeks) and symptomatic (6 months). Using the Affymetrix detection call algorithm, 48 of the 52 striatum genes on the array were called as Present in two out of three of the experimental replicates. Class comparisons were performed in BRB-ArrayTools to identify specific genes differentially expressed between classes (see methods for details). Of these 48 genes, and to our surprise, 81% (n = 39) showed statistically significant differential expression in symptomatic HD transgenic mice compared to littermate controls. The likelihood that 39 out of 48 striatum-specific genes would be selectively affected by chance is 5 × 10−119 (see Methods). The expression of 38 of these genes was significantly decreased, and one gene significantly increased, in the striatum of HD transgenic mice (Table 1). When considering all genes screened on the array, 1.2% and 3.6% were found to be expressed at significantly altered levels at the p < 0.01 and p < 0.05 levels, respectively, and these data showed a normal distribution (Fig. 1). Comparing the expression fold-change ratios of the striatum genes to all genes expressed, HD transgenic mice differed in the distribution of striatal genes that were decreased over the 95% confidence level. The striatal gene group exhibited a pronounced decrease (shift to the left) in transcript expression in the HD transgenic mice (Fig. 1).
Figure 1. Gene expression differences in striatum of symptomatic Huntington's disease (HD) transgenic mice and wild-type littermate controls. All expressed genes were classified into signal intensity difference intervals (0.2 bins) according to their fold-change ratios. Transcripts in a ‘1’ bin had identical signal intensities. Positive values (to the right) on the x axis denote higher hybridization signal in HD mice, negative values (to the left) correspond to higher hybridization signal intensity in the wild-type mice. The y axis reports the percentage of expressed genes. White bars denote distribution of all expressed genes and shaded bars denote distribution of striatum-enriched genes. WT, wild-type.
Download figure to PowerPoint
Among the striatal genes, the novel subunit, beta 4, for the voltage-gated sodium channel showed the most striking difference in expression (7.75-fold decrease) in 6-month-old HD brains. Eight of these 39 genes have been reported in an earlier microarray experiment to be decreased in the striatum of symptomatic HD R6/2 mice (Luthi-Carter et al. 2000), which have an accelerated HD phenotype compared to the R6/1 strain. These are the adenosine A2a receptor, dopamine D1 and D2 receptors, dopamine and cAMP regulated phosphoprotein of 32 KDa (DARPP-32), enkephalin, adenylyl cyclase V, alpha-actinin and the retinoid × (RXR) gamma receptor. Only one gene, neurotensin, showed significantly elevated expression in symptomatic HD transgenic mice, while another gene encoding the islet activating factor 1 (ISL1) transcription factor was increased in presymptomatic transgenic mice. Other striatal genes, including diacylglycerol kinase, ephrin A4 and the 5HT6 receptor, showed higher levels of expression in presymptomatic or symptomatic transgenic mice, however, these differences were not statistically significant in this experiment. Other genes encoding proteins related to those striatum-enriched species, but whose expression is not limited to the striatum, were found to be significantly increased in the striatum of symptomatic mice. These include G-protein γ4 and γ3 subunits, regulator of G-protein signaling-19, Rho-guanine nucleotide exchange factor, protein phosphatase 6, 5-hydroxytryptamine receptor 2c (5HT2c), orphan G protein coupled receptor 22 (GPR22) and several cell adhesion molecules (data not shown). Besides the striatal-enriched genes, additional sequences found to be highly significantly decreased in expression were also related to G-protein and intracellular signaling (data not shown).
In order to validate these findings, we performed real-time PCR on striatal RNA isolated from two independent sets of pooled wild-type and transgenic R6/1 mice at presymptomatic (8–10 weeks) and symptomatic time points (6 months old) for eight striatal sequences we recently isolated or previously identified: FoxP1, Bcl11b, sodium channel subunit beta 4 (SCNβ4), regulator of G-protein signaling-9 (RGS9), insulin receptor substrate protein-53 (IRSp53), copine V, G-protein γ7 subunit and MBA52 (Fig. 2). Relative abundance values correlated with the initial array data for both presymptomatic and symptomatic time points for all eight selected mRNAs.
Figure 2. Decrease of striatal transcripts levels in Huntington's disease (HD) transgenic mice. Real-time PCR amplification of the indicated genes from striatum of presymptomatic (10 weeks) symptomatic (6 months) wild-type (WT) and HD R6/1 transgenic mice was performed as described in Methods. Values are averages ± SEM of duplicate determinations from two independent experiments (n = 6 mice per condition). Amplification of the hypoxanthine guanine phosphoribosyl transferase (HPRT) housekeeping gene was used as an internal reference and the expression level from the age-matched wild-type was used as calibrator in the ΔCt method. Asterisks denote significant differences in gene expression as determined first by one-way anova and then Student's t-test (unpaired; two-tailed) to get exact p-values: *p < 0.05; **p < 0.001; ***p < 0.0001.
Download figure to PowerPoint
In situ hybridization analysis was performed on brains from different litters of symptomatic R6/1 mice with probes for RGS9, Bcl11b, SCNβ4, MBA51, MBA52, the ras homolog enriched in striatum (Rhes), adenylate cyclase V (ACV) and FoxP1. In all cases, decreases in expression in the striatum of symptomatic mice were confirmed (Figs 3 and 4). Oxysterol binding protein like-8 (OSBPL-8), Golf and IRSp53 also showed decreases in expression in HD mice (data not shown). In addition to the striatum, decreases for these genes were observed in other brain regions, such as cortex, thalamus and hippocampus. For striatal genes exhibiting decreases in expression in presymptomatic time points in the array experiment, we examined expression in 4 week-old HD mice by in situ hybridization to determine how early diminished expression could be detected. None of the gene showed substantial decrements at 4 weeks of age, suggesting that its expression levels are not affected directly by the presence of the transgene (data not shown).
Figure 3. In situ hybridization of striatal transcripts in symptomatic Huntington's disease (HD) transgenic mouse brain. In situ hybridization analysis was performed on free-floating coronal sections from control (wild-type) and symptomatic HD transgenic mice brains (n = 3–4 pairs of wild-type and HD transgenic mice). Antisense 35S-labeled riboprobes against the indicated striatal genes were used as described in Methods. CPu, caudate putamen; Cx, cortex; HD-Tg, Huntington's disease transgenic mouse brain; Hipp, hippocampus; Thal, thalamus; WT, wild-type.
Download figure to PowerPoint
Figure 4. High magnification views of striatal transcripts in symptomatic Huntington's disease (HD) transgenic mouse brain. Darkfield photomicrographs are shown for the indicated genes in wild-type (WT) and R6/1 mice. Antisense 35S-labeled riboprobes against the indicated striatal genes were used as described in Methods. Scale bar = 80 µm.
Download figure to PowerPoint
To determine if the changes we observed in the R6/1 mouse model mimic the pathology in human HD, we performed real-time PCR on postmortem caudate samples from HD patients and controls with eight striatal-enriched genes that showed changes in the mouse transgenic model: SCNβ4, FoxP1, RGS9, Kv channel-interacting protein 2 (KCNIP1) (homologous to mouse KCNIP2), OSBPL-8, G-protein γ7 subunit, RAS, guanyl releasing protein 2 (CalDAG GEF) and neurotensin (Fig. 5). Alterations in expression matching those measured in transgenic mice were observed for all eight genes and the differences in expression were statistically significant for four of these, despite the n of only two. The increase in neurotensin expression is important, because despite the severe loss of medium spiny neurons in the caudate of HD patients (grade 3), there is no decrease in expression of this neuronally expressed gene.
Figure 5. Alterations of striatal-enriched transcripts in postmortem caudate of Huntington's disease (HD) subjects and controls. Real-time PCR amplification of the indicated genes was performed on human postmortem caudate samples as described in Methods. Values are averages ± SEM of duplicate determinations from two independent experiments. HD brains were a pathological grade 3. Amplification of the beta-2-microglobulin housekeeping gene was used as an internal reference and the expression level in the control brain was used as calibrator in the ΔCt method. Asterisks denote significant differences in gene expression as determined first by one-way anova and then Student's t-test (unpaired; two-tailed) to get exact p-values: *p < 0.05; **p < 0.001; ***p < 0.0001.
Download figure to PowerPoint
We next compared the expression of selected striatal genes in a rodent model of PD. Although there is no spontaneous occurrence of parkinsonism in animals, several experimental animal models of PD have been developed to mimic clinical features of this disorder in animals, with the first and most frequently used being the 6-OHDA-lesioning model (Mendez and Finn 1975; Mokry 1995). Because simultaneous measurements of the 54 striatal genes was not possible in rats (Affymetrix whole genome arrays for rat were not available at the time of the study), we investigated the expression of randomly selected striatal genes using in situ hybridization. Rats received unilateral injections of 6-OHDA resulting in a loss of dopaminergic input to the striatum with neuronal death occurring in a biphasic manner: an initial phase beginning 12 h postinjection to about 7 days postinjection in which most of the cell loss occurs, and a prolonged phase that extends ∼1 month. Tyrosine hydroxylase staining confirmed a decrease in dopamine content in the striatum and substantia nigra on the ipsilateral side of the injected toxin at both 7 and 28 days postinjection (Fig. 6). In situ hybridization was performed on brains from rats at 7 days and 1 month postinjection. No changes in the ipsilateral vs. contralateral sides of the injection were observed for any of the 15 striatal genes we tested at either 7 days or 1 month postinjection (Fig. 6; Table 2). Representative images for four striatal genes from rats at 7 day postinjection are shown in Fig. 6. Previously published studies have shown that expression of the genes for DARPP-32 and the adenosine A2a receptor are unchanged by destruction of the nigrostriatal pathway (Ehrlich et al. 1990; Kaelin-Lang et al. 2000); however, elevations in the expression of neurotransmitter and neuropeptides receptors in the striatum has been reported (see Table 2). Striatal neuropeptides, enkephalin and neurotensin, and the striatal enriched phosphatase, STEP61, have also been reported to be altered in the 6-OHDA model (Smith et al. 1993; Hanson and Keefe 1999; Napolitano et al. 2002). These results indicate that dopaminergic striatal denervation does not modify the expression of a majority of striatal-enriched genes.
Figure 6. 6-Hydroxydopamine (6-OHDA) treatment results in a decrease in tyrosine hydroxylase (TH) staining in the striatum (a vs. b) and no change in mRNA expression of select striatal genes (c–f). Immunohistochemistry was performed on free-floating rat brain sections from rats 7 days postinjection of 6-OHDA into the medial forebrain bundle using an anti-TH antibody as described in Methods. In situ hybridization images are shown for the regulator of G-protein signaling 9 (RGS9), sodium channel beta 4 subunit (SCNβ4), ras homolog enriched in striatum (Rhes) and adenylyl cyclase V (ACV) at 7 days postinjection. Arrows designate side of unilateral 6-OHDA lesion.
Download figure to PowerPoint
Table 2. Comparisons of expression of striatal genes in rodent models for Huntington's and Parkinson's diseases
|Gene name||HD R6/1 mice||PD 6-OHDA rats|
|Adenylate cyclase V||NC|
|B-cell rag ass'd protein||NC||NC|
|BTE binding protein||NC|
- Top of page
- Materials and methods
Several studies have investigated gene expression in HD mouse models using microarray analysis and have discovered a wide variety of genes altered in their expression (Luthi-Carter et al. 2000, 2002a,b; Chan et al. 2002). Our approach was to focus specifically on those genes with enriched expression in the striatum, as we have previously hypothesized that these are among the most relevant candidates for involvement in neurological disorders that selectively involve the striatum and may provide a mechanism for the specific neurodegeneration observed in this brain region in HD (Usui et al. 1994; Watson et al. 1994; Thomas et al. 1998, 2001; Falk et al. 1999). Combining our own findings with those in the literature, we have cataloged 54 genes deemed ‘striatum-enriched’. In this study, we demonstrate that a large majority (81%) of the mRNA products of these genes are decreased in their levels in HD R6/1 transgenic mice, indicating a selective dysfunction of their transcription or stability within the striatum. An important issue is how well the results from the animal model correlate to the human disease. We found that alterations in expression of eight genes observed in the R6/1 HD mouse were also observed in postmortem caudate samples from HD patients. As the availability of high quality human HD caudate samples is very limited, we consider that the results presented here, despite being derived from only two subjects, are relevant findings. They suggest not only that striatal gene expression deficits are important to disease pathology, but also that the R6/1 transgenic mouse replicates the molecular aspects of HD pathology well.
Co-dysregulation of a large number of genes normally coregulated, suggests that one or more yet unknown transactivating factors are responsible for global transcriptional deficits in the striatum, although coordinated regulation of mRNA stability has not been ruled out. Transgenic R6/1 mice show no massive cell death, neuronal degeneration or reactive gliosis (Mangiarini et al. 1996; Hansson et al. 1999; Naver et al. 2003), hence the generalized decrease in striatal function implicated in this study occurs in the absence of neuronal degeneration and in many instances, prior to the onset of motor symptoms. We did not observe a decrease in the expression of the neuronal marker, neurofilament, or increases in the glial markers, glial fibrillary acidic protein and myelin basic protein, in our overall array study. This is consistent with postmortem studies in humans (Vonsattel et al. 1985; Myers et al. 1988) that suggest that both motor and cognitive symptoms appear in the absence of neuronal cell loss, suggesting that impairment in cognition is likely to be caused by a cellular dysfunction rather than a consequence of neuronal cell death. The data presented here indicate that the physiological deficit in HD is a result of diminished ability of striatal neurons to perform tasks for which they are uniquely programmed.
Previous studies identifying gene expression differences in HD mice have also reported changes in the expression of some striatal-enriched species. Early studies showed decreases in adenosine A2a, dopamine D1 and D2 receptor mRNA expression and levels of radioligand binding in the R6/1, R6/2 and R6/5 HD transgenic mouse lines (Cha et al. 1998, 1999). One of the first oligonucleotide microarray studies performed on the HD transgenic R6/2 mice demonstrated a reduction in the expression of some genes involved in striatal signaling pathways (Luthi-Carter et al. 2000). Of the 6000 genes screened on that array and ∼70 genes altered in the transgenic model, only eight had specific or enriched expression in the striatum: the adenosine A2a receptor, the dopamine D2 receptor, DARPP-32, enkephalin, adenylyl cyclase V, PDE1B1, alpha-actinin and the RXR gamma receptor. Five of these genes had similar altered expression in a second mouse model of HD, the N171-82Q transgenic line (Luthi-Carter et al. 2000). Candidate gene expression studies have shown decreases in the mRNA and/or protein expression of DARPP-32, ARPP-16, ARPP-21, enkephalin, PDE1B and PDE10A in R6/1 and/or R6/2 mice (Bibb et al. 2000; Menalled et al. 2000; van Dellen et al. 2000; Hebb et al. 2004). Importantly, similar changes in dopamine D1 and D2 receptors, enkephalin and substance P have also been observed in the striatum of Huntington's patients (Richfield et al. 1995; Augood et al. 1996, 1997). Our studies reveal the extent of such alterations.
We have previously noted that among the set of striatal-enriched transcripts, those related to G-protein signaling are particularly prominent (Usui et al. 1994; Thomas et al. 1998). Alterations in G protein-related striatal genes found in this study in HD include RAS, guanyl releasing protein 2 (also known as CalDAG GEF), guanine nucleotide binding protein, alpha olfactory type (Golf), γ7 subunit, RGS9, ACV, phosphodiesterase 1 B (PDE1B), phosphodiesterase 10A (PDE10A), DARPP32 61, protein phosphatase 1, regulatory subunit 16B (PPP1R16B), neuronal guanine nucleotide exchange factors (nGEF), Dopamine receptor regulating factor (DRRF), Rhes and striatal enriched phosphatase (STEP61). Of these, RGS9, ACV, PDE1B, PDE10A, CalDAG GEF, DARPP-32 were significantly decreased as early as 8–10 weeks of age in our Affymetrix analysis, far prior to the onset of motor symptoms, which do not occur until after 26 weeks of age in our colony, suggesting importance to the pathogenesis of the illness.
This study highlights another area of striatal dysfunction in HD: calcium homeostasis. Several genes related to calcium signaling were altered in symptomatic HD mice, including copine V, striatin, SCNβ4 and alpha-actinin 2, with the latter three showing diminished expression levels in presymptomatic mice. The potassium channel-interacting protein 2 (KCNIP2) may also be related, as potassium channels are key determinants of neuronal excitability. The gene demonstrating the highest level of expression changes was a novel subunit of the sodium channel, SCNβ4. It exhibits abundant expression in control striatum and an eight-fold decrease in expression in symptomatic R6/1 mice. Expression of SCNβ4 was not changed in mice at 4 weeks of age; however, significantly decreased expression was detected in mice 8–10 weeks of age, again prior to the onset of symptoms. Voltage-gated sodium channels are composed of pore-forming alpha subunits and auxiliary beta subunits that modify channel function. Sodium levels in the cell are directly related to intracellular calcium levels by means of the sodium–calcium exchanger, an ion transporter that exchanges Na+ and Ca2+ depending on membrane potential and transmembrane ion gradients (Blaustein and Lederer 1999). Therefore, alterations in expression or function of sodium channels can have dramatic consequences to intracellular calcium accumulation. These findings are consistent with previous studies reporting dysfunction in calcium signaling/currents in HD transgenic mice based on gene expression abnormalities and biological assessments (Luthi-Carter et al. 2000; Panov et al. 2002, 2003). Of direct relevance, a recent finding demonstrates that huntingtin-induced disturbances in calcium signaling leads to apoptosis of medium spiny neurons (Tang et al. 2005).
Major inputs to the striatum include the glutamatergic corticostriatal pathway and the dopaminergic nigrostriatal pathway. The substantia nigra (pars compacta) functions most prominently via massive projections to the striatum, thereby constituting the dopaminergic input to this region. The irreversible loss of dopamine-mediated control of striatal function is thought to cause a complex rearrangement of neuronal activity which involves specific dopamine-regulated cellular functions (Napolitano et al. 2002). A close interaction between the dopaminergic and glutamatergic systems exists in the striatum (Hersch et al. 1995; Starr 1995) and complex expression changes of glutamate receptor subunits have been reported in response to 6-OHDA (Ulas and Cotman 1996; Dunah et al. 2000). Accordingly, an array screen of 1756 genes showed that 6-OHDA treatment alters the expression of genes involved in several cellular functions, primarily alterations in genes related to glutamatergic neurotransmission (Napolitano et al. 2002). Only a handful of genes previously identified, however, show enriched striatal expression (see Table 2). In this study, none of the striatal-enriched genes we tested were altered in expression in 6-OHDA-treated rats, indicating that their expression is independent of dopaminergic innervation. Consistent with these findings, previous studies have also demonstrated no change in expression of striatal genes, DARPP-32 and adenosine A2a receptors, in 6-OHDA-lesioned rats (Ehrlich et al. 1990; Kaelin-Lang et al. 2000). This suggests that 6-OHDA treatment, which disrupts dopaminergic input to the striatum, does not affect the intrinsic functions of striatal neurons.
In contrast to PD, in Huntington's disease, although there is normal dopamine innervation of the striatum, the striatal neurons lose their intrinsic functionality and ultimately suffer selective neuronal death. The deficits may be initiated by deficient transcriptional regulation of a set of genes responsible for intrinsic striatal neuronal function. This may account for the selective striatal pathology caused by the widely expressed polyQ-expanded huntingtin protein.