Proximal tubular cholesterol loading after mitochondrial, but not glycolytic, bl - 《American Journal of Physiology》美国生理学杂志干细胞之家



免疫细胞治疗专区	欢迎关注干细胞微信公众号

12 3 4 5 6 7 8 9 10 ... 24 下一页

返回列表

查看: 488052\|回复: 231	go Proximal tubular cholesterol loading after mitochondrial, but not glycolytic, bl [复制链接]

轻羽

新手上路

Rank: 1

积分: 0
威望: 0
包包: 0

楼主

发表于 2009-4-21 13:49 |只看该作者 |倒序浏览 |打印

作者：Richard A. Zager, Ali C. M. Johnson,  Sherry Y. Hanson作者单位：Fred Hutchinson Cancer Research Center and the University of Washington, Seattle, Washington 98109 5 w4 r' T  W) e5 ^2 w0 y6 {- W, Z6 K

   , x/ i' Z4 e' h1 M& A. f0 s: w# P


         7 J- B3 I- }9 ?! t6 K1 C! I
         * l+ Y# U2 B/ f: s
         - `0 o4 S! ~# p9 T. M' c$ I
         4 U. T5 w; l3 `
                  . F. e3 w% N0 j

      * Q/ v! B2 u. n0 g) _
      ) G. W4 u5 Q; ]$ q
      【摘要】
   Diverse forms of injury cause proximal tubular cholesterol accumulation. However, underlying mechanisms in general, and those involved with ATP depletion injury in particular, remain poorly defined. To help elucidate this issue, cholesterol homeostasis and its determinants were assessed after partial ATP depletion states. Serum-exposed HK-2 cells were subjected to mild ATP depletion, induced by mitochondrial inhibition (antimycin A; AA) or glycolytic blockade (2-deoxyglucose; DG). Four or 18 h later, cell cholesterol levels, hydroxymethylglutaryl (HMG)-CoA reductase (HMGCR), the LDL receptor (LDL-R), and ABCA1/SR-B1 cholesterol transporters were assessed. AA and DG each induced mild, largely sublethal ATP depletion injury. Each also caused significant HMGCR increments and SR-B1 decrements and left ABCA1 intact. In contrast, only AA increased the LDL-R, and only AA evoked a cholesterol-loading state ( 25% ). One-half of this increase was statin inhibitable, and one-half could be blocked by serum deletion, implying that both synthetic and nonsynthetic (e.g., LDL-R transport) pathways were involved. The AA-induced HMGCR and LDL-R protein changes were paralleled by their mRNAs, suggesting the presence of altered transcriptional events. We conclude that 1 ) sublethal ATP depletion, whether induced by mitochondrial or glycolytic blockade, can upregulate HMGCR and decrease SR-B1, and these changes represent a previously unrecognized ATP depletion phenotype; 2 ) mitochondrial blockade can also upregulate the LDL-R and evoke a cholesterol-loading state; 3 ) the latter likely occurs via synthetic and transport pathways; and 4 ) the mitochondrion may be a critical, and previously unrecognized, determinant of postinjury cell cholesterol homeostasis, potentially by impacting the LDL-R.
      【关键词】 antimycin A deoxyglucose LDL receptor SRB ABCA
   CELLULAR STRESS EVOKES A PLETHORA of responses that appear aimed at protecting against subsequent bouts of tissue damage. Most attention in this area has focused on stress-induced cytoprotective proteins, such as heat shock proteins, heme oxygenase-1, and ferritin ( 1, 4, 9, 12, 18 ). However, alterations in lipid expression may also impart cellular resistance to further attack (e.g., unsaturated fatty acids, sphingosine, lysophosphatidic acid; Refs. 5, 10, 21 ). Perhaps the most consistent cytoprotective lipid yet identified is cholesterol. This conclusion is based on a series of experiments that demonstrate that diverse forms of renal injury (ischemia, toxins, oxidative stress, sepsis, heat shock, hyperosmolality; immunological injury, urinary tract obstruction) evoke increases in proximal tubular cholesterol content ( 19, 22, 24 - 30 ). That these cholesterol increases are critical to cellular resistance to injury is indicated by a series of observations that demonstrate that 1 ) either reversing postinjury cholesterol increments, or preventing them (with statin or zaragozic acid therapy), cancels the cytoresistant state ( 11, 16, 22, 24, 29, 30 ); and 2 ) alterations of cholesterol homeostasis within normal cells (e.g., via oxidation; deesterification; interference with normal P-glycoprotein-mediated cholesterol cycling) cause cellular ATP depletion and lethal cell damage ( 19, 20 ). Of interest, chemotherapeutic injury imposed on malignant myeloid cells also induces cholesterol accumulation and a cytoresistant state ( 11, 16 ). That essentially identical findings have been observed in proximal tubules and myeloid cells indicates the potential generalized nature of this postinjury cholesterol overload state." B6 R! [! M* p4 ?7 d7 v6 _
( m, O1 q. p% x9 ?+ D
Despite the apparent uniformity of postinjury cholesterol accumulation, the mechanisms by which it occurs may vary according to the specific cell type involved and the type of injury sustained. The most extensive evaluation of this issue has been conducted with Fe-mediated oxidative stress. Based on a series of in vivo (glycerol-induced acute renal failure) and in vitro (ferrous ammonium sulfate-mediated oxidative stress) investigations from this laboratory ( 24, 28, 30 ), it has been hypothesized that multiple defects in cholesterol homeostasis may coexist. These include the following: 1 ) increased cholesterol synthesis, based on findings of increased hydroxymethylglutaryl (HMG)-CoA reductase (HMGCR) protein levels and activity ( 24 ); 2 ) increased LDL receptor (LDL-R) expression ( 28 ); and 3 ) Fe-induced reductions in ABCA1 and SR-B1 ( 28 ). The former can efflux free (i.e., unesterified) cholesterol from cells, whereas the latter may evoke bidirectional cholesterol transport (increasing cholesteryl ester uptake, FC efflux; Refs. 8, 28 ). Thus changes in their expression would be expected to alter cell cholesterol content.
! ^" Y8 j/ n* D# @7 z: M! v2 q9 {
It remains unknown whether these Fe-mediated changes are relevant to other injury-induced cholesterol overload states. Therefore, the present study was undertaken to ascertain potential mechanisms for cholesterol accumulation in a second form of cellular injury: that evoked by partial ATP depletion. The relevance of this issue is indicated by prior observations that in vivo renal ischemia-reperfusion injury induces renal cortical cholesterol overload ( 22 ). Given the heterogeneity of cell types within the renal cortex, the present study focused on events occurring within cultured renal proximal tubular (HK-2) cells to gain cell-type-specific data. With the use of this approach, the following questions were addressed. 1 ) Can mild, sublethal ATP depletion trigger an increase cell cholesterol content? 2 ) If so, does the mechanism by which ATP depletion occurs (e.g., decreased mitochondrial ATP production vs. decreased glycolysis) impact the development of a cholesterol overload state? 3 ) Does ATP depletion injury increase cell cholesterol predominantly via increased synthesis (e.g., as previously noted with Fe-mediated injury; Ref. 29 ), or might other mechanism(s) also be involved? 4 ) If increased cholesterol synthesis results from partial ATP depletion, does increased protein prenylation, a potential by product of increased HMGCR activity/mevalonate pathway flux ( 30 ), also result? 5 ) Does partial ATP depletion cause reductions in key cholesterol transport proteins (ABCA1, SR-B1) in a fashion analogous to that which follows Fe-mediated oxidative stress?

METHODS

Cell culture methods. HK-2 cells, an immortalized human proximal tubular cell line established from normal human kidney ( 14 ), were used for all experiments. They were cultured in T75 flasks (Costar, Cambridge, MA) with keratinocyte serum-free medium (GIBCO Life Technologies, Grand Island, NY) to which were added 1 mM glutamine, 5 ng/ml epidermal growth factor, 40 µg/ml bovine pituitary extract, 25 U/ml penicillin, and 25 µg/ml streptomycin ( 14 ). The cells were passaged by trypsinization and then reseeded on 1 ) additional T75 flasks (for further passage or for experiments that called for cell protein or adenine nucleotide analysis); 2 ) T25 flasks (for total cholesterol determinations); or 3 ) 24-well Costar plates for assessments of cellular injury/viability. The cells were cultured for 1-3 days after passage to achieve near confluence before the conduct of specific experiments, as described below.! R$ r# w: _/ t* V* V
. r1 g8 l! e1 U
Antimycin A-induced mitochondrial inhibition: effect on cell cholesterol levels. Twelve T25 flasks were divided into two equal groups, as follows: 1 ) incubation with antimycin A (AA; 7.5 µM; dissolved in ethanol, final concentration, 0.1%; Ref. 6 ); or 2 ) incubation with the AA carrier (ethanol). Immediately before AA or carrier addition, heat-inactivated (56°C x 20 min) normal mouse serum (100-113, Gemini Bio Products) was added to a final concentration of 2.5%. The cells were incubated under these conditions for 18 h. The flasks were then rinsed twice with HBSS (HBSS Ca 2  /Mg ) to remove the serum and its contaminating cholesterol. The cells were recovered by scraping the flasks with a cell scraper and then rinsing the flask three times with HBSS. The cells were recovered from the HBSS by centrifugation. The pellet was washed with HBSS and then extracted in 3 ml of 1:2 chloroform:methanol ( 22, 30 ). The lipid fractions were recovered and dried to a volume of 100 µl. Methanol (10 µl) was added, and the samples were sonicated and then centrifuged. They were subsequently assayed for total cholesterol using a commercially available kit (Thermo DMA, Arlington, TX), as previously performed in this laboratory ( 22 ). Results were expressed as nmoles cholesterol per micromole of phospholipid P i in each sample (the latter determined by sample ashing and P i analysis; Ref. 22 ).7 R& Q( }/ m  C) U- e

To assess whether a second mitochondrial inhibitor might alter cell cholesterol levels, the above experiment was repeated in three control and three rotenone-treated flasks (10 or 50 µM; n = 2 and 1, respectively; in 0.05% ethanol). After 18 h, cell cholesterol levels for the control and rotenone-treated cells were compared.
6 h7 r. p, _8 ~1 z
Effect of HMGCR inhibition on AA-induced cholesterol increments. The above experiments demonstrated that mitochondrial blockade causes cholesterol loading of serum-exposed HK-2 cells (see RESULTS ). To ascertain whether HMGCR-mediated cholesterol synthesis was completely responsible for this result, the following experiment was undertaken. Eight T25 flasks of HK-2 cells were divided into two equal groups: 1 ) incubation with 2.5% serum 10 µM mevastatin (in 0.1% DMSO; Ref. 29 ); and 2 ) incubation with AA serum mevastatin. After the completion of an 18-h incubation, the cells were harvested and assayed for total cholesterol content, as noted above.+ Q# F- _) H6 \! m2 w
' \! n8 Z8 e2 I% {+ @
HK-2 cell cholesterol levels with the AA challenge in the absence of serum. As presented in RESULTS, statin therapy eliminated 50% of AA-induced cholesterol loading, suggesting that both synthesis and possible increased cell cholesterol uptake were involved. In that case, if the AA challenge were conducted in the absence of serum, increased cholesterol uptake would not be possible. Hence, this would be expected to decrease the AA-induced cholesterol loading by 50%. The following experiment tested this hypothesis. Eight flasks of HK-2 cells were cultured in the absence of serum, and four underwent the AA challenge. The remaining four served as controls. After an 18-h incubation, cell cholesterol levels were determined.) _. f5 N3 G0 e3 u
" X4 N8 g9 q" Q1 ?
ATP depletion from glycolysis inhibition: impact on cell cholesterol content. To contrast results obtained by mitochondrial inhibition, eight flasks of HK-2 cells were prepared and divided into two equal groups: 1 ) addition of 2.5% serum; and 2 ) serum addition 20 mM 2-deoxyglucose (to inhibit glycolysis; Ref. 6 ). Eighteen hours later, cell cholesterol levels were determined, as noted above.4 f. Y: C! N) H  `" \

Impact of AA and 2-deoxyglucose on HK-2 cell viability. The following experiments were conducted to ascertain the extent to which AA and 2-deoxyglucose caused lethal cell injury under conditions used in the above experiments. To this end, two 24-well Costar cluster plates were seeded with HK-2 cells. At near confluence, the 24 wells/plate were each divided into the following groups ( n = 6/ treatment): 1 ) control incubation with 2.5% serum; 2 ) incubation with serum 7.5 µM AA; 3 ) a second set of control incubation with serum; and 4 ) incubation with serum 20 mM 2-deoxyglucose. After an 18-h incubation was completed, lethal cell injury was assessed by %lactate dehydrogenase (LDH) release.5 g2 T7 F; f6 }. V/ t! m1 X

Impact of AA and 2-deoxyglucose on cellular ATP levels. The following experiment was undertaken to ascertain relative degrees of impairment of cellular energetics induced by AA vs. 2-deoxyglucose. To this end, 18 T75 flasks were seeded with HK-2 cells. At near confluence, they were divided into three groups: 1 ) addition of 2.5% serum, 2 ) addition of serum AA, and 3 ) addition of serum 2-deoxyglucose. After a 4-h incubation, adenine nucleotides were extracted in 6.66% trichloroacetic acid and then analyzed for ATP and ADP concentrations by HPLC as previously described ( 23 ). The degree of suppression of cellular energetics was assessed by calculating ATP/ADP ratios and the amount of ATP extracted per flask.

Effect of AA on HMGCR, ABCA1, SR-B1, and LDL-R mRNAs. Cellular cholesterol levels reflect a balance among 1 ) synthesis; 2 ) LDL-R-mediated cholesterol uptake; and 3 ) cell free cholesterol (FC)/cholesteryl ester (CE) efflux/influx (ABCA1:FC efflux; SR-B1:CE influx/FC efflux; Ref. 8 ). To gain insights into these pathways during AA-induced injury, the mRNAs for each were determined. To this end, HK-2 cells were seeded in 36 separate T25 flasks. On reaching near confluence, they were divided into four groups: 1 ) control incubation with 2.5% serum x 4 h ( n = 10); 2 ) incubation with 7.5 µM AA 2.5% serum x 4 h ( n = 10); 3 ) control incubation with 2.5% serum x 18 h ( n = 8); and 4 ) incubation with AA 2.5% serum for 18 h ( n = 8). At the completion of incubation, the cells were recovered as noted above and immediately placed into TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA). Total RNA was extracted as according to the manufacturer's instructions. The final RNA pellet was brought up in RNase-free water to an approximate concentration of 1 mg/ml. The samples were electrophoresed for 30 min through 1.2% agarose containing ethidium bromide (Sigma) to ensure a lack of degradation (preservation of 18S and 28S ribosomal RNA) before PCR analysis ( 30 ).+ X1 z3 n) ?7 F7 |: ]% f% u" _& ]
- U0 ?1 N6 [- l! n3 {
ABCA1, LDL-R, and SR-B1 (as well as GAPDH) mRNAs were analyzed simultaneously by multiplexing RT-PCR using human-specific primers, as previously described ( 28 ) and as shown in Table 1. The validity of this method has been previously confirmed by documenting that mevastatin therapy of HK-2 cells leads to appropriate responses of the LDL-R, ABCA1, and SR-B1 messages [a marked increase, a marked decrease, and minimal decrease, respectively ( 28 )]. LDL-R, ABCA1, and SR-B1 values were expressed by factoring them by the GAPDH reference product.& X) I( V9 p7 ~2 j! v+ P

Table 1. Primers and conditions used for PCR analyses0 B. y6 s6 W- O8 p- r+ P# B

HMGCR mRNA was analyzed separately using human-specific primers and conditions listed in Table 1. Otherwise, the conditions were identical to those previously described ( 28; Table 1 ). To confirm the biological validity of this assay, samples were obtained from four flasks of HK-2 cells treated with ( n = 2) or without ( n = 2) 10 µM mevastatin (to confirm a statin-induced increase in HMGCR mRNA).
* I1 S' z" Y1 b, [! q0 |. s3 \
Effects of AA on ABCA1, SR-B1, LDL-R, and HMGCR protein expression. Eight T25 flasks were seeded with HK-2 cells. On reaching near confluence, they were equally divided into two groups: 1 ) addition of 2.5% serum AA or 2 ) addition of 2.5% serum 0.1% ethanol (AA vehicle). Eighteen hours later, cell protein extracts from each flask were prepared as previously described and used for Western blotting for SR-B1, ABCA1, LDL-R, and HMGCR ( 8, 24, 28 ). In brief, for SR-B1 detection, 5 µg of protein extract were electrophoresed through a 12% Bis-Tris acrylamide Nupage gel (Invitrogen, Carlsbad, CA) and probed with rabbit anti-SR-B1 antibody (NB-400-104, Novus Biologicals, Littleton, CO). For ABCA1 detection, a 25-µg protein extract sample was electrophoresed through a 4-12% gradient Bis-Tris acrylamide Nupage gel. Rabbit anti-ABCA1 (NB-400-15, Novus Biologicals) was used as the primary antibody. Secondary detection for both anti-SR-B1 and anti-ABCA1 antibodies was performed using horseradish peroxidase-labeled donkey anti-rabbit IgG (NA 934, Amersham-Pharmacia, Piscataway, NJ) and enhanced chemiluminescence (ECL Kit, Amersham-Pharmacia; Ref. 10 ). For detection of LDL-R, 25-µg protein extract samples were electrophoresed through a 4-12% gradient Bis-Tris acrylamide Nupage gel. Mouse anti-LDL-R (LP02, Oncogene Research Products, Boston, MA) was used as the primary antibody. Secondary detection was performed with horseradish peroxidase-labeled sheep anti-mouse IgG (NA 931, Amersham-Pharmacia) and ECL ( 8 ). HMGCR was probed in 25-µg protein samples electrophoresed into the above 4-12% Nupage gel, using rabbit anti-HMGCR (a gift from P. A. Edwards, Los Angeles, CA). Detection was performed with donkey anti-rabbit IgG ( 24 ). Nonspecific secondary antibody staining in the above Western blots has been excluded by the fact that the secondary antibody, in the absence of the primary antibody, does not identify the relevant protein band(s). Equal protein loading/transfer with these protocols has been confirmed by India ink staining. Relevant protein band quantitation was performed by optical density scanning (ABCA1, 220 kDa; SR-B1, 82 kDa; nonglycosylated LDL-R, 120 kDa; glycosylated LDL-R, 160 kDa; HMGCR, 45 kDa; Ref. 24 ).
& g( }! s: a, m2 r% |" C* K8 v9 l
Effects of deoxyglucose on ABCA1, SR-B1, LDL-R, and HMGCR protein expression. To ascertain whether ATP depletion induced by 2-deoxyglucose causes a different pattern of cholesterol homeostatic protein expression than does AA, the above experiment was repeated, with the exception that the 2-deoxyglucose challenge was substituted for AA treatment.
6 Q0 L; a  N; \7 Q1 u* Q
Effect of AA-induced cholesterol loading on HK-2 cell Rho and Ras prenylation. To ascertain whether AA-induced changes in the mevalonate pathway impacts protein prenylation, four control samples and four samples obtained 18 h post-AA treatment were probed for Rho and Ras by Western blotting as previously described ( 30 ). In the case of Rho, prenylation was assessed by comparing the ratio between the prenylated ( bottom ) band and unprenylated ( top ) band on the gel after their quantitation by ECL. The relative position of these two bands has previously been determined by treating HK-2 cells with mevastatin, which causes a marked decrease in the bottom (prenylated), but not in the top (unprenylated) band ( 30 ). Thus the ratio of the bottom to the top (prenylated/unprenylated) band serves as an indication of Rho prenylation. In the case of Ras, a dominant unprenylated band appears and a second ( top ) band develops with prenylation ( 30 ). Hence, the latter was sought after AA treatment.3 H, G6 B- [+ b: z5 R

Calculations and statistics. All values are presented as means ± SE. Statistical comparisons were performed by unpaired Student's t -test. Significance was judged by a P value of7 F5 f: g. f" M  L- W
3 ^  J/ y+ g- L5 \; j
RESULTS: L; F! d1 ]5 `& H1 _" p- y! Z
/ t- W* \) C; K, v/ q/ m
HK-2 cell cholesterol levels in response to AA with and without mevastatin. As shown in Fig. 1, the AA challenge caused a marked increase in cell cholesterol content, rising from 255 ± 6 to 307 ± 13 nmol/µmol P i. When this same experiment was performed in the presence of mevastatin (in both the control and AA incubations), AA still induced a statistically significant cholesterol increase. To contrast the relative degree of AA-induced cholesterol increases in the presence vs. absence of mevastatin, the %increase in cholesterol between the two sets of flasks was compared. The %increase in cholesterol induced by AA in the absence of mevastatin was 25 ± 3% compared with 11 ± 1% in the presence of AA statin ( P0 G& ^  J& C' ]- N
/ }- p' C0 g9 R7 G% ?
Fig. 1. A : cholesterol levels in HK-2 cells after 18-h incubations under control (Cont) conditions (2.5% serum in keratinocyte serum-free medium) or under the same conditions antimycin A (AA) treatment. As seen in the pair of bars on the left, AA caused an 50 nmol/µmol P i increase in cholesterol values. When this same experiment was repeated in the presence of mevastatin (Statin; in both control and AA-treated cells; the pair of bars on the right ), a still significant but smaller AA-induced increase in cholesterol resulted. B 50% of the AA-induced cholesterol increase, a significant increase still resulted. This indicates that both synthetic and nonsynthetic pathways were involved in AA-induced cholesterol loading.
% E% M, l5 [) t/ |' s9 c4 _' `
AA-induced cholesterol accumulation in the absence of serum. The above results are consistent with the view that 50% of the AA-induced cholesterol accumulation was due to synthesis (i.e., statin inhibitable) and 50% was statin resistent (i.e, likely due to increased uptake ± decreased efflux). If so, then eliminating uptake by removing serum should reduce the AA-induced cholesterol loading by 50%. Indeed, this is precisely what was found with AA treatment of cells. In the absence of serum, AA induced a 12% cholesterol increase (control cells, 264 ± 2; AA, 295 ± 4; P 25% cholesterol increase in the presence of serum, the withholding of serum decreased cholesterol loading by 50% (i.e., from 25 to 12%).

Effects of rotenone on HK-2 cell cholesterol levels. To test whether a second model of mitochondrial inhibition would raise cholesterol levels in the presence of serum, the effects of rotenone were assessed. Rotenone caused a smaller ( 10%), but still significant ( P = 0.01), increase in total cellular cholesterol, compared with the increase observed with AA ( 25%). [Of note, however, these rotenone and AA results cannot be directly compared because both rotenone doses, unlike AA, caused 40% cell detachment, consistent with greater injury (e.g., more dead cells that would not synthesize or transport cholesterol). Because of this cell detachment, the model was not chosen for further experiments].
* _' h8 w+ M4 T5 i  P$ f
Effect of 2-deoxyglucose on HK-2 cell cholesterol levels. Unlike AA, treatment of cells with 2-deoxyglucose did not significantly alter cellular cholesterol levels. Total values for the control and 2-deoxyglucose-treated cells were 264 ± 11 and 262 ± 15 nmol/µmol P i, respectively.6 U2 M( D! T8 T; a/ L0 P
2 P0 c4 M' x& G
Extent of cell injury induced by AA and 2-deoxyglucose. AA treatment did not cause a significant increase in %LDH release [control 7 ± 1%; AA, 8 ± 1%; not significant (NS)]. However, total adherent cell LDH content (a reflection of total viable cell mass; Refs. 6, 7 ) of the AA-exposed cells was decreased by 12% (16.4 ± 2 units vs. 14.6 ± 0.1 units; P an antiproliferative response ± a small amount of apoptosis) did occur. In contrast to AA, 2-deoxglucose did cause a small but significant increase in %LDH release (6 ± 1 vs. 10 ± 1%; P slight, but significant, and comparable, injury could be detected. [Note, because 2.5% serum contains a small amount of LDH, this value was subtracted from the total supernatant values before determination of %LDH release].
6 O: r. D; R, i3 }% C
Cellular energetics after AA or 2-deoxyglucose treatment. Cellular energetics, as assessed by ATP/ADP ratios, were significantly depressed with both AA and 2-deoxyglucose treatment (control ATP/ADP ratios, 18.3 ± 0.6; with AA, 11.8 ± 1.6; with 2-deoxyglucose, 8.5 ± 0.4) ( P NS vs. each other). A comparison of absolute ATP concentrations per flask showed that 2-deoxyglucose caused a slightly greater depression in ATP than did AA treatment (64 ± 2 vs. 79 ± 5 µmol, respectively, P 90 ± 4 µmol). Thus, despite the fact that AA, but not 2-deoxyglucose, treatment raised cell cholesterol levels, the degree of ATP reduction was at least comparable, or slightly worse, with the latter agent." m3 Z9 q" ?9 O& {1 j6 @

HMGCR, LDL-R, ABCA1, and SR-B1 mRNA levels after the AA challenge. The results of RT-PCR analyses are presented in Table 2. In brief, HMGCR mRNA values were slightly, but statistically, higher at both 4 and 18 h after AA addition. As a physiological control for the HMGCR mRNA assay, levels were measured in control and statin-treated cells, as noted in METHODS. Statin therapy induced an approximate sevenfold increase in HMGCR values (6.5 ± 0.2 vs. 1.0 ± 0.2, statin vs. control values; P 0.005).2 O# W* L3 s  s7 a) S" V
" D5 f3 P$ D5 E, X* k; b7 Z; {1 c
Table 2. mRNA values after 4 or 18 h of control incubations or incubation with antimycin A

AA also induced modest, and statistically significant, elevations in LDL-R mRNA values at each of the two time points. Conversely, ABCA1 message was suppressed by 25% at 4 h post-AA addition. Although ABCA1 mRNA values returned to normal by 18 h of AA treatment, these values should be considered physiologically abnormal, given that cholesterol loading at 18 h would be expected to increase the ABCA1 message (and protein) levels to help restore normal cholesterol content. SR-B1 mRNA levels were trivially, but significantly, elevated at both the 4- and 18-h time points.

Effects of AA on HMGCR, LDL-R, SR-B1, and ABCA1 protein expression. Representative Western blots and statistical comparisons between control cells (open bars) and 18-h AA-treated cells (shaded bars) are presented in Fig. 2. As shown at the far left, HMGCR protein expression was significantly increased in the AA treatment group compared with controls ( P an 33% increase in unglycosylated LDL-R protein levels ( bottom band; P increase in the amount of the LDL-R glycosylated moiety ( top band; not depicted in bar graph). Unlike the results with HMGCR and LDL-R, AA had no significant effect on ABCA1 expression. However, AA induced a 35% decrease in SR-B1 ( P A corollary of the SR-B1 reductions was the appearance of a lower molecular weight protein band, consistent with an SR-B1 degradative product ( Fig. 2 ). This band was not seen in control HK-2 cell protein samples.
0 w: H$ J( V6 t, b
Fig. 2. Comparison of Western blot analysis results from control cells (open bars) and AA-treated cells (shaded bars) after 18-h incubation. Hydroxymethylglutaryl (HMG)-CoA reductase (HMGCR) appeared as a single band at 45 kDa and was significantly increased by AA treatment. LDL receptor (LDL-R) appears as a nonglycosylated and a glycosylated band at 120 and 160 kDa, respectively. AA induced a statistically significant increase in the nonglycosylated band (depicted by the bars). A trend, but not a statistically significant increase, in the glycosylated band existed (not depicted by bar graph). ABCA1 appears as a broad band at 220 kDa, the appearance of which was not significantly altered by AA treatment. Conversely, AA induced a significant decrease in SR-B1 (seen at 82 kDa) with the concomitant appearance of a lower molecular weight, possibly degradative, product that was not observed in the control cell protein extracts.& h6 X! v9 F7 _; C. g! _  J8 m, J+ O
( b1 f% K1 \- @" S, A+ r& t
Effects of 2-deoxyglucose on HMGCR, LDL-R, SR-B1, and ABCA1 protein expression. After 18 h of 2-deoxglucose exposure, a significant increase in HMGCR protein was noted (230 ± 24 vs. controls, 151 ± 29; P basis, this was highly comparable to the %increase observed with AA treatment. Also, like AA, 2-deoxyglucose caused a significant decrease in SR-B1 levels (126 ± 10 vs. controls, 205 ± 20; P with AA (although with 2-deoxyglucose, a second apparently degraded band did not appear). ABCA1 was not significantly affected by 2-deoxyglucose, again recapitulating the AA results. However, in striking contrast to AA treatment, 2-deoxyglucose failed to increase LDL-R expression. Rather, a nonsignificant trend toward decreased, rather than increased, LDL-R protein expression resulted ( top band: 96 ± 28, control; 69 ± 15, deoxyglucose; NS; bottom band: 183 ± 12, control; 168 ± 6, deoxyglucose; NS). Thus of the four proteins tested, only LDL-R expression differed between the AA and 2-deoxyglucose treatment. All values are presented as densitometry units.& g: b! n/ }3 V' r$ ~

Rho and Ras prenylation. Western blot probes of Rho and Ras are presented in Fig. 3. Two Rho bands were observed, with the top and bottom bands representing the unprenylated and prenylated moieties, respectively. No obvious differences in their expression were observed between the control (open bars) and AA-treated samples (shaded bars). Furthermore, the prenylated/unprenylated band ratios did not significantly differ for the control and AA treatment groups (as depicted by the bars). The Ras probe revealed only a single band that did not differ between the control (open bars) and AA-treated (shaded bars) groups (i.e., no evidence of a prenylated band was apparent in either group).$ H4 e! \4 @" F7 B0 R3 M

Fig. 3. Western blots of Rho and Ras from control (open bars) and AA-treated cells (shaded bars) after 18-h incubation. Prenylated and unprenylated Rho appear as the bottom and top band, respectively (at 20-24 kDa). The ratios of the 2 (prenylated/unprenylated) are depicted in the bars and did not statistically differ for the control and AA treatment groups. Ras appears as a single, nonprenylated band at 21 kDa, which was slightly, but not significantly, decreased by AA treatment.: W, j& u4 D  G, O1 [6 w; x) ?

DISCUSSION

All forms of in vivo renal injury tested to date ( 22, 24 - 30 ) have demonstrated a number of common features vis à vis the development of a cholesterol overload state. These include the following: 1 ) an 18-h lag time is required for cholesterol loading to develop; 2 ) both unesterified (free) and esterified cholesterol participate in this response; 3 ) these changes are fully expressed within proximal tubules; and 4 ) cholesterol loading confers cellular resistance to further attack. Despite this common postinjury cholesterol phenotype, less consistent information exists regarding the in vivo mechanisms that initiate and maintain this cholesterol overload state. For example, after glycerol- or heat shock-mediated injury, increased renal cortical HMGCR protein expression results ( 24 ). Conversely, although sepsis syndrome and renal ischemia induce marked cholesterol loading, renal cortical HMGCR protein levels remain at control levels ( 24, 26 ). Studies of corresponding HMGCR mRNA levels after renal injury further suggest different operant pathways. For example, glycerol-induced myohemoglobinuria induces stepwise increases in renal cortical HGMCR mRNA ( 30 ). However, after renal ischemia, renal cortical HMGCR mRNA decrements result ( 30 ). These discrepancies suggest that multiple, and potentially disease-specific, pathways mediate the injury-initiated cholesterol-loading state.# p5 g3 U( i! H7 U' ~" J! Q
; T% s! ~/ A) c* B' Z
A major caveat to assessments of cellular mechanisms performed using in vivo tissues is the heterogeneity of cell types within the renal cortex. This precludes evaluation of proximal tubule cell-specific events. To circumvent this problem, this laboratory has explored pathways of cholesterol accumulation in cultured HK-2 cells subjected to Fe-mediated oxidative attack ( 28, 30 ). These studies further suggest that multiple pathways may contribute to in vitro cholesterol accumulation, including increased synthesis and possibly decreased cholesterol efflux due to reductions in the ABCA1 and SR-B1 free cholesterol efflux proteins. However, HMGCR-driven synthesis appears to be primary in Fe-challenged cells, based on the fact that statin therapy completely blocks the cholesterol-loading state ( 30 ). However, two significant caveats to these prior HK-2 cell experiments exist ( 28, 30 ): first, they were conducted under serum-free conditions, thereby precluding the possibility of cholesterol uptake from the extracellular space (e.g., via the LDL receptor); and, second, the Fe challenge caused significant cell death ( 30-40% LDH release). Hence, the relative partitioning of cholesterol between viable vs. non viable cells could not be ascertained. Therefore, to explore mechanisms of injury-induced tubular cholesterol accumulation under conditions devoid of these two caveats, the present study employed a sublethal ATP depletion model conducted in the presence of serum. Because partial ATP depletion exists during the recovery phase of ischemic ARF (e.g., Ref. 17 ), it was hoped that the results obtained might provide insights into mechanisms of renal cholesterol accumulation with postischemic kidney damage.

The first notable result stemming from the present study is that sublethal ATP depletion, when induced by mitochondrial blockade, can initiate the cholesterol-loading state. Although this is the first demonstration of this phenomenon, it does not seem particularly surprising, given prior observations that multiple forms of injury can induce this result ( 19, 22, 24 - 30 ). However, what is surprising is that the pathway by which partial ATP depletion occurs appears to be critical. As detailed in RESULTS, AA caused an 25% cholesterol increase. Rotenone also induced a cholesterol overload state. However, when a comparable, if not greater, degree of ATP depletion was induced by glycolytic blockade with 2-deoxyglucose, absolutely no cholesterol loading resulted. These findings indicate that ATP decrements alone are insufficient to evoke cholesterol increments. Rather, some unidentified consequence(s) of mitochondrial dysfunction (e.g., potentially cytochrome c release, free radical generation, loss of Ca buffering capacity, etc.) appears to be required./ {" [, R9 H8 A$ i! w
9 _( c! a6 |3 d- c! J
A second notable result of the present investigation is the demonstration that increased HMGCR-mediated synthesis is not necessarily the sole mechanism by which injury evokes a cholesterol-loading state. In contrast to prior studies of in vitro Fe-mediated injury in which mevastatin completely blocked cholesterol accumulation ( 29, 30 ), in the present study statin treatment only blocked 50% of the AA-induced cholesterol overload state. Thus this finding indicates that although cholesterol synthesis was clearly operative, an alternative (statin-resistant) pathway(s) was also involved. There are at least three potential candidates in this regard. First, AA treatment significantly increased LDL-R expression. Given that serum was present, increased cholesterol uptake was a plausible result. That AA caused 50% less cholesterol uptake in the absence (vs. the presence) of serum is consistent with this concept. Second, SR-B1 was decreased with AA treatment. Although this transporter can mediate cholesteryl ester influx, as well as free cholesterol efflux, the latter function appears dominant in proximal tubular cells ( 8 ). Thus AA-induced SR-B1 reductions could have contributed to the cholesterol-loading state. Third, despite AA-induced cholesterol accumulation, a failure of a normal compensatory increase in the ABCA1 cholesterol transporter did not result. Thus the finding of normal ABCA1 levels in the setting of AA-induced cholesterol increases represents an aberrant response.

In sum, the above findings point to multiple AA-induced defects in cholesterol homeostasis: 1 ) increased HMGCR protein expression and cholesterol synthesis; 2 ) increased LDL-R expression/cholesterol uptake; 3 ) decreased SR-B1 levels; and 4 ) a failure to physiologically upregulate ABCA1. To ascertain whether these alterations could have stemmed, at least in part, from transcriptional events, the mRNAs for each of these proteins were assessed at two time points: 1 ) during the initiation phase of cholesterol accumulation; and 2 ) when cholesterol overload was fully developed (4 and 18 h post-AA addition, respectively). In the cases of HMGCR and the LDL-R, modest yet statistically significant mRNA increments were observed. In contrast, ABCA1 message at the 18-h time point remained at normal values. Thus in each case protein and message levels qualitatively paralleled each other, suggesting that altered transcriptional regulation of each was likely involved. In contrast, SR-B1 reductions appeared to be dissociated from its message levels, which were very slightly, but significantly, increased above control levels. This suggests that post-transcriptional events (e.g., increased SR-B1 catabolism), rather than reduced synthesis, were more likely involved.: T! S8 Q7 a. Q3 N2 ~" z) W
, ~3 g. J7 t  e" X; ]  c% O
Given that AA and 2-deoxyglucose each induced ATP depletion, but only the former evoked a cholesterol-loading state, potential differences in HMGCR, LDL-R, ABCA1, and SR-B1 protein expression with these two challenges were sought. By so doing, a distinction between ATP-dependent changes and changes evoked by mitochondrial inhibition per se could be assessed. Interestingly, 2-deoxyglucose induced comparable HMGCR protein increments to those that followed AA treatment. Deoxyglucose also decreased SR-B1 levels, while leaving ABCA1 intact. Again, these results are consistent with those that followed AA treatment. However, what did differ between the two forms of ATP depletion injury was that 2-deoxyglucose tended to decrease rather than increase the LDL-R. While tenable, it is premature to conclude that this difference in LDL-R expression fully explains why AA, but not 2-deoxyglucose, caused the cholesterol-loading state. However, this difference does underscore our basic hypothesis: that mitochondrial dysfunction per se, and not simply ATP reductions, is a critical determinant of postinjury cholesterol homeostasis.+ e/ G8 v" d' i0 w& ]6 W! a# X
- C8 I0 B6 ?" o4 |
The final goal of this study was to ascertain whether partial ATP depletion-mediated cholesterol enrichment is associated with changes in the prenylation of signaling molecules, notably Ras and Rho. The relevance of this issue is that prenylation allows for Ras/ Rho translocation to the plasma membrane, where they can contribute to cellular resistance to stress ( 2, 3, 13, 15 ). Because isoprenoids (e.g., farnesyl and geranylgeranylpyrophosphate) are intermediary products of the mevalonate pathway, it is conceivable that injury-induced alterations in cholesterol homeostasis could directly impact prenylation events. For example, an increase in HMGCR activity could increase isoprenoid production. Alternatively, increased cholesterol uptake via LDL-R-mediated transport, coupled with decreased ABCA1- or SR-B1-mediated free cholesterol efflux, should cause feedback inhibition of the HMGCR axis, potentially decreasing isoprenoid synthesis, and hence, prenylation. Therefore, Ras and Rho were probed by Western blot analysis in AA-treated and control HK-2 cells. Despite the above considerations, no changes in their expression were observed. Of note, in previous studies conducted in renal cortex after Fe-mediated oxidative stress ( 30 ), cholesterol accumulation was also dissociated from changes in Ras/Rho appearance. Thus when these past in vivo and present in vitro data are viewed together, they indicate that degrees of mevalonate pathway activity and protein prenylation are not necessarily linked. Because Ras and Rho activation may enhance cell survival, the finding that an increase in cell cholesterol can be dissociated from Ras/Rho prenylation helps to underscore that cholesterol per se is the dominant mevalonate pathway product that helps to confer the postinjury cytoresistant state.

In conclusion, the present study documents for the first time that partial ATP depletion increases and decreases the HMGCR axis and SR-B1 expression, respectively. That two disparate mechanisms for inducing ATP depletion, i.e., mitochondrial and glycolytic blockade, each evoke these two changes suggests that they represent a previously unrecognized ATP depletion-induced phenotype. However, despite these changes, cellular cholesterol accumulation need not result, presumably due to offsetting homeostatic mechanisms. However, with concomitant mitochondrial blockade, cholesterol loading does result, possibly due to an associated increase in the LDL-R. This suggests that multiple defects must exist for cholesterol accumulation to result. The specific mechanism(s) by which mitochondrial inhibition conditions cholesterol accumulation, beyond ATP depletion-related changes, remains unknown. However, a mitochondria-initiated increase in LDL-R expression appears to be one likely candidate in this regard.4 S4 @0 Z+ _# f0 K$ c: L

DISCLOSURES) O* T$ P# H1 u
6 V/ k7 @' v' M9 w+ s; ]$ C- W0 `
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants RO-1 DK-38432 and RO-1 DK-54200 and an unrestricted research grant from Abbott Laboratories, North Chicago, IL., a/ q8 Z& ]. [+ B$ D, [2 `
      【参考文献】/ O0 j4 ?$ U' o, T# j& Q+ n
Alam J, Killeen E, Gong P, Naquin R, Hu B, Stewart D, Ingelfinger JR, and Nath KA. Heme activates the heme oxygenase-1 gene in renal epithelial cells by stabilizing Nrf2. Am J Physiol Renal Physiol 284: F743-F752, 2003.

7 w: F/ a) n) m; f0 G
Danesh FR, Sadeghi MM, Amro N, Phillips C, Zeng L, Sahai A, and Kanwar YS. 3-Hydroxymethylglutaryl CoA reductase inhibitors prevent high glucose-induced proliferation of mesangial cells via modulation of Rho GTPase/p21 signaling pathway: implications for diabetic nephropathy. Proc Natl Acad Sci USA 99: 8301-8305, 2002.+ u. s# K* S+ s+ b7 {; ^7 S) G
' |! r4 U* [" Y- K. v. F

Edwards PA and Ericsson J. Sterols and isoprenoids: signaling molecules derived from the cholesterol biosynthetic pathway. Annu Rev Biochem 68: 157-185, 1999.
% |' L3 B% z# d3 R1 {8 p
- w' g3 a3 `1 _
: N6 h( |/ X/ T" A! D
Inguaggiato P, Gonzalez-Michaca L, Croatt AJ, Haggard JJ, Alam J, and Nath KA. Cellular overexpression of heme oxygenase-1 up-regulates p21, and confers resistance to apoptosis. Kidney Int 60: 2181-2191, 2001.

9 I) a; i0 a- z. B( g5 A7 q+ y

Iwata M, Herrington J, and Zager RA. Sphingosine: a mediator of acute renal tubular injury, and subsequent cytoresistance. Proc Nat Acad Sci USA 92: 8970-8974, 1995.7 x. K) n! v% @! D9 ?) a: D

Iwata M, Herrington JB, and Zager RA. Protein synthesis inhibition induces cytoresistance in cultured proximal tubular (HK-2) cells. Am J Physiol Renal Fluid Electrolyte Physiol 268: F1154-F1163, 1995.
1 y; F* s9 e- T" F& u

Iwata M and Zager RA. Myoglobin induces transient growth arrest in cultured human proximal tubular cells. Kidney Int 50: 796-804, 1996.

/ \' s  Q( Y& A6 Z# x+ F
Johnson ACM, Yabu J, Hanson S, Shah VO, and Zager RA. Experimental glomerulopathy alters renal cortical cholesterol, S.R-B1, ABCA1, and HMG CoA reductase expression. Am J Pathol 162: 273-281, 2003.7 D" M  _7 j( ?* q" x: s  u5 J
' ^0 ?& U, R8 \: K0 \0 I) E$ D
. b  [; m) }) t- Q: U

Kanakiriya SK, Croatt AJ, Haggard JJ, Ingelfinger JR, Tang SS, Alam J, and Nath KA. Heme: a novel inducer of MCP-1 through HO-dependent and HO-independent mechanisms. Am J Physiol Renal Physiol 284: F546-F554, 2003.5 ^; x; C/ y' [4 {* j

Levine JS, Koh JS, Triaca V, and Lieberthal W. Lysophosphatidic acid: a novel growth, and survival factor for renal proximal tubular cells. Am J Physiol Renal Physiol 273: F575-F585, 1997." V. Z9 Z: J, {# b9 f1 W

& N: S- n* U7 d# c& f  t

Li H, Appelbaum F, Zager RA, and Banker DE. Cholesterol modulating agents kill acute myeloid leukemia cells, and sensitize them to therapeutics by blocking adaptive cholesterol responses. Blood 101: 3628-3634, 2003.0 w6 Y/ u% @7 b4 h' v' n

" v/ B. p7 K) n9 e$ [/ i# j
5 Q- D- }" A- H: P% k: B
Li M, Mao HP, Ruchalski KL, Wang YH, Choy W, Schwartz JH, and Borkan SC. Heat stress prevents mitochondrial injury in ATP-depleted renal epithelial cells. Am J Physiol Cell Physiol 283: C917-C926, 2002.

1 l! ?3 q; ~# e( h' ^

McGuire TF and Sebti SM. Geranylgeroniol potentiates lovastatin inhibition of oncogenic H-Ras processing, and signaling while preventing cytotoxicity. Oncogene 14: 305-312, 1997.
5 y2 [4 e4 l4 `
# d+ ~# o, I) m! j- a8 E2 V

Ryan MJ, Johnson G, Kirk J, Fuerstenberg SM, Zager RA, and Torok-Storb B. HK-2: an immortalized proximal tubule epithelial cell line from normal adult human kidney. Kidney Int 45: 48-57, 1994.
; B/ c* e- ~+ j, c; C$ c# }

Stark WW Jr, Blaskovich MA, Johnson BA, Qian Y, Vasudevan A, Pitt B, Hamilton AD, Sebto SM, and Davies P. Inhibiting geranyl geranylation blocks growth, and promotes apoptosis in pulmonary vascular smooth muscle cells. Am J Physiol Lung Cell Mol Physiol 275: L55-L63, 1998.

# b9 H4 }# Z; c
Stirewalt DL, Appelbaum FR, Willman CL, Zager RA, and Banker DE. Mevastatin can increase toxicity in primary AMLs exposed to standard therapeutic agents, but statin efficacy is not simply associated with RAS hotspot mutations or overexpression. Leuk Res 1572: 1-13, 2002.0 R+ m) D1 ^" ?8 h  X

( H" F* _3 @) R3 |
Sutter PM, Thulin G, Stromski M, Ardito T, Gaudio KM, Kashgarian M, and Siegel NJ. Beneficial effect of thyroxin in the treatment of ischemic acute renal failure. Pediatr Nephrol 2: 1-7, 1988.

& o0 X. g8 a7 }) i- k( w
2 C! u% ^' ]- |8 e7 _
Wang YH, Knowlton AA, Li FH, and Borkan SC. Hsp72 expression enhances survival in adenosine triphosphate-depleted renal epithelial cells. Cell Stress Chaperones 7: 137-145, 2002.

( u0 {2 [, x- K8 ~+ P. R$ q
9 ?) J' A  s, d# Y: ]2 i) |* x
Zager RA. Plasma membrane cholesterol: a critical determinant of cellular energetics, and tubular resistance to attack. Kidney Int 58: 193-205, 2000.

% _, v+ ?$ D5 o: l1 M

Zager RA. P glycoprotein-mediated cholesterol cycling: a potentially important determinant of proximal tubular cell viability. Kidney Int 60: 944-956, 2001.

$ c% W2 `! \1 D7 w; I' m
Zager RA, Burkhart KM, Conrad DS, Gmur DJ, and Iwata M. Phospholipase A 2 induced cytoprotection of proximal tubules: potential determinants, and specificity for ATP depletion injury. J Am Soc Nephrol 7: 64-72, 1996.

0 H9 P8 f6 i& c7 O6 S* Q

Zager RA, Burkhart KM, Johnson AC, and Sacks BM. Increased proximal tubular cholesterol content: implications for cell injury, and acquired cytoresistance. Kidney Int 56: 1788-1797, 1999.: t* F* {* V# ^# v* ?7 o& b' G
7 A' [, v; W- l' N' l

* F4 S$ Z5 U% x8 m1 o
Zager RA, Gmur DJ, Bredl CR, and Eng MJ. Temperature effects on ischemic, and hypoxic renal proximal tubular injury. Lab Invest 64: 766-776, 1991.
$ {7 X1 U  O7 `
( W$ \1 K: c/ ?
+ I1 Q! S/ g2 O! U' [! w$ |8 Z# e
Zager RA and Johnson A. Renal cortical cholesterol accumulation is an integral component of the systemic stress response. Kidney Int 60: 2299-2310, 2001.
6 Q% \  g9 f3 A+ `9 ~3 c3 ~! X: c

Zager RA, Johnson A, Anderson K, and Wright S. Cholesterol ester accumulation: an immediate consequence of acute in vivo ischemic renal injury. Kidney Int 59: 1750-1761, 2001.
+ \- x. E  A. J: a4 [
; ^5 L" ?: s" J8 C3 q

Zager RA, Johnson A, and Hanson SY. Sepsis syndrome stimulates proximal tubule cholesterol synthesis, and suppresses the SR-B1 cholesterol transporter. Kidney Int 63: 123-133, 2003.

) ?* j8 i5 B) ]0 d# M, D- |, M
Zager RA, Johnson A, Hanson S, and dela Rosa V. Altered cholesterol localization, and caveolin expression during the evolution of acute renal failure. Kidney Int 61: 1674-1683, 2002.+ T3 f& F& ~4 ]- b4 b) W& ?
! {$ W: m- p; X/ ]$ W& a$ A6 Y
+ f0 }- Q) f8 h+ H! o9 M3 s
. |; A' a6 F9 |; d  o
Zager RA, Johnson ACM, and Hanson SY, and Shah VO. Acute tubular injury causes dysregulation of cellular cholesterol transport proteins. Am J Pathol 163: 313-320, 2003.

( F# L2 E' t( l7 |0 l9 l6 J' u
8 L* W! n9 A: ?+ W! G
Zager RA and Kalhorn TF. Changes in free and esterified cholesterol: hallmarks of acute renal tubular injury and acquired cytoresistance. Am J Pathol 157: 1007-1016, 2000.* D  @. n! Z( ^% P; B7 ^$ P
+ z1 r8 c9 \4 @2 o* w

Zager RA, Shah VO, Shah HV, Zager PA, Johnson AC, and Hanson S. The mevalonate pathway during acute tubular injury. Selected determinants and consequences. Am J Pathol 161: 681-692, 2002.

收藏0 分享0 顶0 踩0